1 2017-08-01 Jakub Jelinek <jakub@redhat.com>
3 PR tree-optimization/81588
4 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
5 ranges[i].in_p, invert comparison code ccode. For >/>=,
6 swap rhs1 and rhs2 and comparison code unconditionally,
7 for </<= don't do that. Don't swap rhs1/rhs2 again if
8 ranges[i].in_p, instead invert comparison code ccode if
9 opcode or oe->rank is BIT_IOR_EXPR.
12 * optabs.def (vec_extract_optab, vec_init_optab): Change from
13 a direct optab to conversion optab.
14 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
15 with GET_MODE_INNER as last argument instead of optab_handler.
16 * expmed.c (extract_bit_field_1): Likewise. Use vector from
17 vector extraction if possible and optab is available.
18 * expr.c (store_constructor): Use convert_optab_handler instead
19 of optab_handler. Use vector initialization from smaller
20 vectors if possible and optab is available.
21 * tree-vect-stmts.c (vectorizable_load): Likewise.
22 * doc/md.texi (vec_extract, vec_init): Document that the optabs
24 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
25 of vec_init from half-sized vectors with the same element mode.
26 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
27 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
28 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
29 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
30 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
31 after mode in gen_vec_extract* calls.
32 (vec_extract<mode>): Renamed to ...
33 (vec_extract<mode><ssescalarmodelower>): ... this.
34 (vec_extract<mode><ssehalfvecmodelower>): New expander.
35 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
36 element mode after mode in gen_vec_init* calls.
37 (VEC_INIT_HALF_MODE): New mode iterator.
38 (vec_init<mode>): Renamed to ...
39 (vec_init<mode><ssescalarmodelower>): ... this.
40 (vec_init<mode><ssehalfvecmodelower>): New expander.
41 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
42 (vec_extractv2sfsf): ... this.
43 (vec_initv2sf): Renamed to ...
44 (vec_initv2sfsf): ... this.
45 (vec_extractv2si): Renamed to ...
46 (vec_extractv2sisi): ... this.
47 (vec_initv2si): Renamed to ...
48 (vec_initv2sisi): ... this.
49 (vec_extractv4hi): Renamed to ...
50 (vec_extractv4hihi): ... this.
51 (vec_initv4hi): Renamed to ...
52 (vec_initv4hihi): ... this.
53 (vec_extractv8qi): Renamed to ...
54 (vec_extractv8qiqi): ... this.
55 (vec_initv8qi): Renamed to ...
56 (vec_initv8qiqi): ... this.
57 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
58 (vec_init<mode>): Renamed to ...
59 (vec_init<mode><VEC_base_l>): ... this.
60 (vec_extract<mode>): Renamed to ...
61 (vec_extract<mode><VEC_base_l>): ... this.
62 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
63 (vec_initv2sfsf): ... this.
64 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
65 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
66 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
67 element mode after mode in gen_vec_init* calls.
68 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
69 (vec_init<mode><Vel>): ... this.
70 (vec_extract<mode>): Renamed to ...
71 (vec_extract<mode><Vel>): ... this.
72 * config/aarch64/iterators.md (Vel): New mode attribute.
73 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
74 Add element mode after mode in gen_vec_extract* calls.
75 * config/s390/vector.md (non_vec_l): New mode attribute.
76 (vec_extract<mode>): Renamed to ...
77 (vec_extract<mode><non_vec_l>): ... this.
78 (vec_init<mode>): Renamed to ...
79 (vec_init<mode><non_vec_l>): ... this.
80 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
81 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
83 * config/arm/iterators.md (V_elem_l): New mode attribute.
84 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
85 (vec_extract<mode><V_elem_l>): ... this.
86 (vec_extractv2di): Renamed to ...
87 (vec_extractv2didi): ... this.
88 (vec_init<mode>): Renamed to ...
89 (vec_init<mode><V_elem_l>): ... this.
90 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
91 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
92 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
93 Add element mode after gen_vec_extract* calls.
94 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
95 (vec_init<mode><unitmode>): ... this.
96 (vec_extract<mode>): Renamed to ...
97 (vec_extract<mode><unitmode>): ... this.
98 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
99 (vec_init<mode><unitmode>): ... this.
100 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
101 (vec_initv2sfsf): ... this.
102 (vec_extractv2sf): Renamed to ...
103 (vec_extractv2sfsf): ... this.
104 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
105 Add element mode after gen_vec_extract* calls.
106 * config/mips/mips.md (unitmode): New mode iterator.
107 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
108 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
109 * config/spu/spu.md (inner_l): New mode attribute.
110 (vec_init<mode>): Renamed to ...
111 (vec_init<mode><inner_l>): ... this.
112 (vec_extract<mode>): Renamed to ...
113 (vec_extract<mode><inner_l>): ... this.
114 * config/sparc/sparc.md (veltmode): New mode iterator.
115 (vec_init<VMALL:mode>): Renamed to ...
116 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
117 * config/ia64/vect.md (vec_initv2si): Renamed to ...
118 (vec_initv2sisi): ... this.
119 (vec_initv2sf): Renamed to ...
120 (vec_initv2sfsf): ... this.
121 (vec_extractv2sf): Renamed to ...
122 (vec_extractv2sfsf): ... this.
123 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
124 (vec_init<mode>): Renamed to ...
125 (vec_init<mode><VEC_base_l>): ... this.
126 (vec_extract<mode>): Renamed to ...
127 (vec_extract<mode><VEC_base_l>): ... this.
128 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
129 (vec_initv2sfsf): ... this.
130 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
131 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
132 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
135 2017-08-01 Richard Biener <rguenther@suse.de>
137 PR tree-optimization/81297
138 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
139 TREE_OVERFLOW from INTEGER_CSTs.
141 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
143 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
145 2017-07-31 Carl Love <cel@us.ibm.com>
147 * config/rs6000/rs6000-c: Add support for built-in functions
148 vector signed char vec_xl_be (signed long long, signed char *);
149 vector unsigned char vec_xl_be (signed long long, unsigned char *);
150 vector signed int vec_xl_be (signed long long, signed int *);
151 vector unsigned int vec_xl_be (signed long long, unsigned int *);
152 vector signed long long vec_xl_be (signed long long, signed long long *);
153 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
154 vector signed short vec_xl_be (signed long long, signed short *);
155 vector unsigned short vec_xl_be (signed long long, unsigned short *);
156 vector double vec_xl_be (signed long long, double *);
157 vector float vec_xl_be (signed long long, float *);
158 * config/rs6000/altivec.h (vec_xl_be): Add #define.
159 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI,
160 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions for the builtins.
161 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
162 (altivec_expand_builtin): Add switch statement to call altivec_expand_xl_be
164 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
165 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di, __builtin_vsx_le_be_v4sf,
166 __builtin_vsx_le_be_v2df, __builtin_vsx_le_be_v16qi.
167 * doc/extend.texi: Update the built-in documentation file for the
168 new built-in functions.
170 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
173 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
175 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
177 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
179 * config.gcc: Add z14.
180 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
181 CPU model numbers for z13s and z14.
182 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
184 * config/s390/s390-opts.h (enum processor_type): Rename
185 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
186 * config/s390/s390.c (processor_table): Add field for CPU name to
187 be passed to Binutils.
188 (s390_asm_output_machine_for_arch): Use the new field in
189 processor_table for Binutils.
190 (s390_expand_builtin): Replace arch12 with z14.
191 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
192 (s390_get_sched_attrmask): Likewise.
193 (s390_get_unit_mask): Likewise.
194 * config/s390/s390.opt: Add z14 to processor_type enum.
196 2017-07-31 Martin Jambor <mjambor@suse.cz>
199 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
200 regardless of optimization level.
202 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
203 Martin Liska <mliska@suse.cz>
205 * predict.def: Remove old comment and adjust probability.
206 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
209 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
212 * config/i386/i386.c (ix86_function_naked): New function.
213 (ix86_can_use_return_insn_p): Return false for naked functions.
214 (ix86_expand_prologue): Skip prologue for naked functions.
215 (ix86_expand_epilogue): Skip epilogue for naked functions
216 and emit trap instruction.
217 (ix86_warn_func_return): New function.
218 (ix86_attribute_table): Add "naked" attribute specification.
219 (TARGET_WARN_FUNC_RETURN): Define.
220 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
222 2017-07-31 Martin Liska <mliska@suse.cz>
224 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
225 (dump_gimple_bb_header): Always dump BB info.
226 (pp_cfg_jump): Do not append info about BB when dumping a jump.
228 2017-07-31 Martin Liska <mliska@suse.cz>
231 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
232 also with current_function_decl non-null equality.
234 2017-07-31 Jakub Jelinek <jakub@redhat.com>
237 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
238 change type to the element type, instead add eltype variable and
239 use it where we are interested in the element type.
241 PR tree-optimization/81603
242 * ipa-polymorphic-call.c
243 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
244 offset arithmetic in offset_int, bail out if the resulting bit offset
245 doesn't fit into shwi.
247 2017-07-31 Martin Liska <mliska@suse.cz>
249 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
250 (gimplify_save_expr): Fix comment.
252 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
255 * config/i386/i386.c (ix86_function_arg): Update arguments for
257 (ix86_compute_frame_layout): Set the initial stack offset to
258 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
259 INCOMING_FRAME_SP_OFFSET.
260 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
261 stack before exception handler returns.
262 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
263 the 'ERROR_CODE' for exception handler.
265 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
267 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
268 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
269 (ASM_OUTPUT_REG_POP): Ditto.
270 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
271 instead of asm_fprintf to output pure string.
273 2017-07-29 Jakub Jelinek <jakub@redhat.com>
275 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
276 to imported_module_or_decl hook.
277 (debug_nothing_tree_tree_tree_bool): Remove.
278 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
279 * debug.c (do_nothing_debug_hooks): Use
280 debug_nothing_tree_tree_tree_bool_bool instead of
281 debug_nothing_tree_tree_tree_bool.
282 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
283 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
284 * sdbout.c (sdb_debug_hooks): Likewise.
285 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
286 (gen_namespace_die): Add DW_AT_export_symbols attribute if
288 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
289 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
290 attribute, don't add anything.
292 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
294 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
295 (fold_build2_stat_loc): Likewise.
296 (fold_build3_stat_loc): Likewise.
297 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
298 (fold_build1_loc): Remove macro.
299 (fold_build2_loc): Likewise.
300 (fold_build3_loc): Likewise.
302 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
304 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
305 (gimple_build_debug_bind_source_stat): Likewise.
306 * gimple.h (gimple_build_debug_bind): Remove macro.
307 (gimple_build_debug_bind_source): Likewise.
309 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
311 * bitmap.c (bitmap_alloc): Adjust.
312 (bitmap_gc_alloc): Likewise.
313 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
315 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
317 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
318 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
319 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
320 (bitmap_gc_alloc_stat): Likewise.
321 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
323 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
325 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
326 * rtl.h (shallow_copy_rtx): Remove macro.
328 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
330 * emit-rtl.c (gen_raw_REG): Adjust.
331 * gengenrtl.c (gendef): Likewise.
332 * rtl.c (rtx_alloc_stat): Remove _stat from name.
333 * rtl.h (rtx_alloc): Remove macro.
335 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
337 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
338 (build_tree_list_stat): Likewise.
339 * tree.h (build_tree_list): Remove macro.
340 (build_tree_list_vec): Likewise.
342 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
344 * tree.c (make_vector_stat): Remove _stat from name.
345 (build_vector_stat): Likewise.
346 * tree.h (make_vector_stat): Remove macro.
347 (build_vector_stat): Likewise.
349 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
351 * tree.h (build_var_debug_value): Remove prototype.
353 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
355 * tree.c (tree_cons_stat): Remove _stat from name.
356 * tree.h (tree_cons): Remove macro.
358 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
360 * tree.c (build_vl_exp_stat): Remove _stat from name.
361 * tree.h (build_vl_exp): Remove macro.
363 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
365 * tree.c (build_decl_stat): Remove _stat from name.
366 * tree.h (build_decl): Remove macro.
368 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
370 * gimple.c (gimple_build_with_ops_stat): Adjust.
371 (gimple_alloc_stat): Remove _stat from name.
372 * gimple.h (gimple_alloc): Remove macro.
374 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
376 * tree.c (make_tree_vec_stat): Remove _stat from name.
377 (grow_tree_vec_stat): Likewise.
378 * tree.h (make_tree_vec_stat): Adjust prototype.
379 (grow_tree_vec_stat): Likewise.
380 (make_tree_vec): Remove macro.
381 (grow_tree_vec): Likewise.
383 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
385 * fold-const.c (fold_build1_stat_loc): Adjust.
386 (fold_build2_stat_loc): Likewise.
387 (fold_build3_stat_loc): Likewise.
388 * tree.c (build0_stat): Remove _stat from name.
389 (build1_stat): Likewise.
390 (build2_stat): Likewise.
391 (build3_stat): Likewise.
392 (build4_stat): Likewise.
393 (build5_stat): Likewise.
394 * tree.h (build1_loc): Remove macro, and rename _stat function
396 (build2_loc): Likewise.
397 (build3_loc): Likewise.
398 (build4_loc): Likewise.
399 (build5_loc): Likewise.
401 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
403 * tree.c (make_int_cst_stat): Remove _stat from name.
404 * tree.h (make_int_cst_stat): Adjust prototype.
405 (make_int_cst): Remove macro.
407 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
409 * tree.c (make_tre_binfo_stat): Remove _stat from name.
410 * tree.h (make_tree_binfo_stat): Adjust prototype.
411 (make_tree_binfo): Remove.
413 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
415 * tree.c (copy_node_stat): Rename to copy_node.
416 (build_distinct_type_copy): Adjust.
417 * tree.h (copy_node_stat): Adjust prototype.
418 (copy_node): Remove macro.
420 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
422 * tree.c (make_node_stat): rename to make_node.
423 (build_tree_list_stat): Adjust.
424 (build0_stat): Likewise.
425 (build2_stat): Likewise.
426 (build3_stat): Likewise.
427 (build4_stat): Likewise.
428 (build5_stat): Likewise.
429 (build_decl_stat): Likewise.
430 * tree.h (make_node_stat): Adjust prototype.
431 (make_node): remove macro.
433 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
435 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
436 (PPC_FEATURE2_SCV): Likewise.
437 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
439 2017-07-28 Tamar Christina <tamar.christina@arm.com>
441 * config/aarch64/aarch64.c
442 (aarch64_internal_mov_immediate): Add new special pattern.
443 * config/aarch64/aarch64.md (*movdi_aarch64):
444 Add reg/32bit const mov case.
446 2017-07-28 Tamar Christina <tamar.christina@arm.com>
447 Richard Sandiford <richard.sandiford@linaro.org>
449 * config/aarch64/aarch64.md (mov<mode>): Generalize.
450 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
451 Add integer and movi cases.
452 (movi-split-hf-df-sf split, fp16): New.
453 (enabled): Added TARGET_FP_F16INST.
454 * config/aarch64/iterators.md (GPF_HF): New.
455 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
457 2017-07-28 Tamar Christina <tamar.christina@arm.com>
459 * config/aarch64/aarch64.c
460 (aarch64_simd_container_mode): Add prototype.
461 (aarch64_expand_mov_immediate): Add HI support.
462 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
463 (aarch64_can_const_movi_rtx_p): New.
464 (aarch64_preferred_reload_class):
465 Remove restrictions of using FP registers for certain SIMD operations.
466 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
467 (aarch64_valid_floating_const): Add integer move validation.
468 (aarch64_simd_imm_scalar_p): Remove.
469 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
470 (aarch64_legitimate_constant_p): Expand list of supported cases.
471 * config/aarch64/aarch64-protos.h
472 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
473 (aarch64_reinterpret_float_as_int): New.
474 (aarch64_simd_imm_scalar_p): Remove.
475 * config/aarch64/constraints.md (Uvi): New.
476 (Dd): Split into Ds and new Dd.
477 * config/aarch64/aarch64.md (*movsi_aarch64):
479 (*movdi_aarch64): Add SIMD mov case.
481 2017-07-28 Bin Cheng <bin.cheng@arm.com>
483 * tree-predcom.c: (struct chain): Handle store-store chain in which
484 stores for elimination only store loop invariant values.
485 (execute_pred_commoning_chain): Ditto.
486 (prepare_initializers_chain_store_elim): Ditto.
487 (prepare_finalizers): Ditto.
488 (is_inv_store_elimination_chain): New function.
489 (initialize_root_vars_store_elim_1): New function.
491 2017-07-28 Bin Cheng <bin.cheng@arm.com>
493 * tree-predcom.c: Revise general description of the pass.
494 (enum chain_type): New enum type for store elimination.
495 (struct chain): New field supporting store elimination.
496 (struct component): Ditto.
497 (dump_chain): Dump store-stores chain.
498 (release_chain): Release resources.
499 (split_data_refs_to_components): Compute and create component
500 contains only stores for elimination.
501 (get_chain_last_ref_at): New function.
502 (make_invariant_chain): Initialization.
503 (make_rooted_chain): Specify chain type in parameter and record it.
504 (add_looparound_copies): Skip for store-stores chain.
505 (determine_roots_comp): Compute type of chain and pass it to
507 (initialize_root_vars_store_elim_2): New function.
508 (finalize_eliminated_stores): New function.
509 (remove_stmt): Handle store for elimination.
510 (execute_pred_commoning_chain): Execute predictive commoning on
512 (determine_unroll_factor): Skip unroll for store-stores chain.
513 (prepare_initializers_chain_store_elim): New function.
514 (prepare_initializers_chain): Hanlde store-store chain.
515 (prepare_finalizers_chain, prepare_finalizers): New function.
516 (tree_predictive_commoning_loop): Return integer value indicating
517 if loop is unrolled or lcssa form is corrupted.
518 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
520 2017-07-28 Bin Cheng <bin.cheng@arm.com>
522 * tree-predcom.c (initialize_root): Delete.
523 (execute_pred_commoning_chain): Initialize root vars and replace
524 reference of non-combined chain directly, rather than call above
527 2017-07-28 Bin Cheng <bin.cheng@arm.com>
529 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
530 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
532 2017-07-28 Bin Cheng <bin.cheng@arm.com>
534 * tree-predcom.c (struct chain): New field init_seq.
535 (release_chain): Release init_seq.
536 (prepare_initializers_chain): Record intialization stmts in above
538 (insert_init_seqs): New function.
539 (tree_predictive_commoning_loop): Call insert_init_seqs.
541 2017-07-28 Bin Cheng <bin.cheng@arm.com>
543 * tree-predcom.c (determine_roots_comp): Skip trivial components.
545 2017-07-28 Richard Biener <rguenther@suse.de>
547 * match.pd: Remove superfluous :c.
548 * genmatch.c (simplify::id): Add member.
549 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
551 (current_id): New global.
552 (dt_node::parent): Move from ...
553 (dt_operand::parent): ... here. Add for_id member.
554 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
555 (decision_tree::find_node): Relax order requirement when
556 merging DT_TRUE nodes to ones inbetween the current simplify
557 and the one we try to merge with. Add diagnostic whenever
558 we need to enforce pattern order by not merging.
559 (decision_tree::insert): Set current_id.
560 (decision_tree::print_node): Dump parent node and for_id.
561 (parser::last_id): Add member.
562 (parser::push_simplify): Assign unique id.
563 (parser::parser): Initialize last_id.
565 2017-07-28 Martin Liska <mliska@suse.cz>
568 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
569 gimple_build_debug_bind.
571 2017-07-28 Richard Biener <rguenther@suse.de>
573 PR tree-optimization/81502
574 * match.pd: Add pattern combining BIT_INSERT_EXPR with
576 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
578 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
579 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
580 for BIT_FIELD_REF args.
581 * fold-const.c (make_bit_field_ref): Likewise.
582 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
584 2017-07-28 Jakub Jelinek <jakub@redhat.com>
587 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
588 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
589 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
590 Or it into SANITIZER_UNDEFINED.
591 * ubsan.c: Include gimple-fold.h and varasm.h.
592 (ubsan_expand_ptr_ifn): New function.
593 (instrument_pointer_overflow): New function.
594 (maybe_instrument_pointer_overflow): New function.
595 (instrument_object_size): Formatting fix.
596 (pass_ubsan::execute): Call instrument_pointer_overflow
597 and maybe_instrument_pointer_overflow.
598 * internal-fn.c (expand_UBSAN_PTR): New function.
599 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
600 * sanitizer.def (__ubsan_handle_pointer_overflow,
601 __ubsan_handle_pointer_overflow_abort): New builtins.
602 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
603 * internal-fn.def (UBSAN_PTR): New internal function.
604 * opts.c (sanitizer_opts): Add pointer-overflow.
605 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
606 * fold-const.c (build_range_check): Compute pointer range check in
607 integral type if pointer arithmetics would be needed. Formatting
610 2017-07-28 Martin Liska <mliska@suse.cz>
613 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
614 parameters that are of a variable-length.
616 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
618 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
620 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
621 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
622 (CRT_CALL_STATIC_FUNCTION): Likewise.
623 (ASM_DEFAULT_SPEC): New define.
624 (ASM_SPEC32): Likewise.
625 (ASM_SPEC64): Likewise.
626 (ASM_SPEC_COMMON): Likewise.
627 (ASM_SPEC): Likewise.
628 (INVALID_64BIT): Likewise.
629 (LINK_OS_DEFAULT_SPEC): Likewise.
630 (LINK_OS_SPEC32): Likewise.
631 (LINK_OS_SPEC64): Likewise.
632 (POWERPC_LINUX): Likewise.
633 (PTRDIFF_TYPE): Likewise.
634 (RESTORE_FP_PREFIX): Likewise.
635 (RESTORE_FP_SUFFIX): Likewise.
636 (SAVE_FP_PREFIX): Likewise.
637 (SAVE_FP_SUFFIX): Likewise.
638 (SIZE_TYPE): Likewise.
639 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
640 (TARGET_64BIT): Likewise.
641 (TARGET_64BIT): Likewise.
642 (TARGET_AIX): Likewise.
643 (WCHAR_TYPE_SIZE): Likewise.
644 (WCHAR_TYPE): Undefine.
645 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
646 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
647 (CPP_OS_RTEMS_SPEC): Delete.
648 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
649 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
651 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
653 2017-07-28 Jakub Jelinek <jakub@redhat.com>
655 PR tree-optimization/81578
656 * tree-parloops.c (build_new_reduction): Bail out if
657 reduction_code isn't one of the standard OpenMP reductions.
658 Move the details printing after that decision.
660 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
662 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
663 related to reload_in_progress.
664 (splat_input_operand): Likewise.
665 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
667 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
669 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
670 (TARGET_INSTANTIATE_DECLS): Likewise.
671 (legitimate_indexed_address_p): Delete reload_in_progress code.
672 (rs6000_debug_legitimate_address_p): Likewise.
673 (rs6000_eliminate_indexed_memrefs): Likewise.
674 (rs6000_emit_le_vsx_store): Likewise.
675 (rs6000_emit_move_si_sf_subreg): Likewise.
676 (rs6000_emit_move): Likewise.
677 (register_to_reg_type): Likewise.
678 (rs6000_pre_atomic_barrier): Likewise.
679 (rs6000_machopic_legitimize_pic_address): Likewise.
680 (rs6000_allocate_stack_temp): Likewise.
681 (rs6000_address_for_fpconvert): Likewise.
682 (rs6000_address_for_altivec): Likewise.
683 (rs6000_secondary_memory_needed_rtx): Delete function.
684 (rs6000_check_sdmode): Likewise.
685 (rs6000_alloc_sdmode_stack_slot): Likewise.
686 (rs6000_instantiate_decls): Likewise.
687 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
688 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
689 Delete reload_in_progress.
690 (*vec_reload_and_plus_<mptrsize>): Likewise.
691 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
692 (vsx_div_v2di): Likewise.
693 (vsx_udiv_v2di): Likewise.
695 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
697 * config/rs6000/rs6000.opt (mlra): Replace with stub.
698 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
699 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
700 (rs6000_debug_reg_global): Delete print of LRA status.
701 (rs6000_option_override_internal): Delete dead LRA related code.
702 (rs6000_lra_p): Delete function.
703 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
705 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
707 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
708 * config/riscv/rtems.h: New file.
710 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
711 Sudakshina Das <sudi.das@arm.com>
713 * config/aarch64/aarch64.md
714 (define_split for and<mode>3nr_compare): Move
715 non aarch64_logical_operand to a register.
716 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
717 register immediate operand to a register.
718 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
720 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
723 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
725 2017-07-27 Richard Biener <rguenther@suse.de>
727 PR tree-optimization/81573
728 PR tree-optimization/81494
729 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
730 multi defuse cycle case.
732 2017-07-27 Richard Biener <rguenther@suse.de>
734 PR tree-optimization/81571
735 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
738 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
740 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
741 earlier and only if MASK_FPU is set. Adjust formatting.
743 2017-07-27 Martin Liska <mliska@suse.cz>
745 * opt-functions.awk: Add validation of value of Init.
746 * optc-gen.awk: Pass new argument.
748 2017-07-27 Martin Liska <mliska@suse.cz>
750 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
753 2017-07-27 Martin Liska <mliska@suse.cz>
755 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
756 BBs and edges seen by autoFDO.
758 2017-07-27 Richard Biener <rguenther@suse.de>
760 PR tree-optimization/81502
761 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
762 with incompatible but same sized type.
763 (execute_update_addresses_taken): Likewise.
765 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
767 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
768 flag_tree_loop_vectorize rather than flag_tree_vectorize.
770 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
773 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
774 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
775 Change s_operand to memory_operand.
777 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
779 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
780 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
781 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
782 Emit instructions rather than returning an expression. Handle TFmode
783 and KFmode by casting to TImode.
784 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
785 (rs6000_emit_le_vsx_store): Likewise.
786 * config/rs6000/vsx.md (VSX_TI): New iterator.
787 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
788 (*vsx_le_undo_permute_<mode>): Likewise.
789 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
790 emit the split sequence.
791 (*vsx_le_perm_store_<mode>): Likewise.
793 2017-07-27 Jakub Jelinek <jakub@redhat.com>
795 PR tree-optimization/81555
796 PR tree-optimization/81556
797 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
798 if true, force CHANGED for the recursive invocation.
799 (reassociate_bb): Remember original length of ops array, pass
800 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
802 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
803 attributes for noipa attribute. For naked attribute use
804 lookup_attribute first before lookup_attribute_spec.
805 * final.c (rest_of_handle_final): Disable IPA RA for functions with
807 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
808 for functions with noipa attribute.
809 (cgraph_externally_visible_p): Return true for functions with noipa
811 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
812 for functions with noipa attribute.
813 * doc/extend.texi: Document noipa function attribute.
814 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
815 also for functions with noipa attribute.
816 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
818 2017-07-26 Andrew Pinski <apinski@cavium.com>
820 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
821 vec_unalign_load_cost and vec_unalign_store_cost.
823 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
825 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
826 -mvsx-small-integer option.
827 (ISA_3_0_MASKS_IEEE): Likewise.
828 (OTHER_VSX_VECTOR_MASKS): Likewise.
829 (POWERPC_MASKS): Likewise.
830 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
831 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
832 code, only testing for DImode being allowed in non-VSX floating
834 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
835 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
837 (rs6000_option_override_internal): Delete -mvsx-small-integer.
838 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
839 TARGET_P8_VECTOR test.
840 (rs6000_secondary_reload_simple_move): Likewise.
841 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
842 since TARGET_P9_VECTOR was already tested.
843 (rs6000_opt_masks): Remove -mvsx-small-integer.
844 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
845 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
847 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
848 test for TARGET_VEXTRACTUB was used, and that uses
850 (p9 extract splitter): Likewise.
851 (vsx_extract_<mode>_di_p9): Likewise.
852 (vsx_extract_<mode>_store_p9): Likewise.
853 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
854 for TARGET_P9_VECTOR was used. Delete code that is now dead with
855 the elimination of TARGET_VSX_SMALL_INTEGER.
856 (vsx_extract_<mode>_p8): Likewise.
857 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
858 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
859 (vsx_set_<mode>_p9): Likewise.
860 (vsx_set_v4sf_p9): Likewise.
861 (vsx_set_v4sf_p9_zero): Likewise.
862 (vsx_insert_extract_v4sf_p9): Likewise.
863 (vsx_insert_extract_v4sf_p9_2): Likewise.
864 * config/rs6000/rs6000.md (sign extend splitter): Change
865 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
866 (floatsi<mode>2_lfiwax_mem): Likewise.
867 (floatunssi<mode>2_lfiwzx_mem): Likewise.
868 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
869 since a test for TARGET_P9_VECTOR was used.
870 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
871 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
872 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
873 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
874 TARGET_P8_VECTOR test.
875 (fix_trunc<mode>si2_stfiwx): Likewise.
876 (fix_trunc<mode>si2_internal): Likewise.
877 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
878 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
880 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
881 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
882 TARGET_P8_VECTOR test.
883 (fixuns_trunc<mode>si2_stfiwx): Likewise.
884 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
885 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
887 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
888 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
889 since a test for TARGET_P9_VECTOR was used.
890 (splitter for loading small constants): Likewise.
892 2017-07-26 Andrew Pinski <apinski@cavium.com>
894 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
897 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
900 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
901 (fp_valid_at): Likewise.
903 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
905 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
906 (qdf24xx_addrcost_table): Likewise.
907 (cortexa57_tunings): Update to use generic_branch_cost.
908 (cortexa72_tunings): Likewise.
909 (cortexa73_tunings): Likewise.
910 (qdf24xx_tunings): Likewise.
912 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
914 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
915 (thunderx2t99_branch_cost): Likewise.
916 (cortexa35_tunings): Update to use generic_branch_cost.
917 (cortexa53_tunings): Likewise.
918 (cortexa57_tunings): Likewise.
919 (cortexa72_tunings): Likewise.
920 (cortexa73_tunings): Likewise.
921 (thunderx2t99_tunings): Likewise.
923 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
925 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
926 (sparc_option_override): Honour MASK_FSMULD.
927 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
928 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
929 * config/sparc/sparc.opt (mfsmuld): New option.
930 * doc/invoke.texi (mfsmuld): Document option.
932 2017-07-26 Marek Polacek <polacek@redhat.com>
935 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
937 2017-07-26 Richard Biener <rguenther@suse.de>
939 * gimple-match-head.c (do_valueize): Return OP if valueize
941 (get_def): New helper to get at the def stmt of a SSA name
943 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
944 do_valueize to get at the def stmt.
945 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
947 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
950 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
952 2017-07-26 Martin Liska <mliska@suse.cz>
955 * function.c (expand_function_start): Make expansion of
956 nonlocal_goto_save_area after parm_birth_insn.
958 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
960 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
961 from all CPU target flags enable members.
963 2017-07-26 Richard Biener <rguenther@suse.de>
965 * genmatch.c (dt_simplify::gen): Make iterator vars const.
966 (decision_tree::gen): Make 'type' const.
967 (write_predicate): Likewise.
969 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
971 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
972 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
973 (rs6000_option_override_internal): Likewise.
974 (rs6000_expand_vector_set): Likewise.
975 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
976 (TARGET_UPPER_REGS_SF): Likewise.
977 (TARGET_UPPER_REGS_DI): Likewise.
978 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
979 (TARGET_DIRECT_MOVE_64BIT): Likewise.
980 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
981 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
982 (Splitters for DI constants in Altivec registers): Likewise.
983 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
984 (vsx_set_v4sf_p9): Likewise.
985 (vsx_set_v4sf_p9_zero): Likewise.
986 (vsx_insert_extract_v4sf_p9): Likewise.
987 (vsx_insert_extract_v4sf_p9_2): Likewise.
989 2017-07-25 Carl Love <cel@us.ibm.com>
991 * doc/extend.texi: Update the built-in documentation file for the
992 existing built-in functions
993 vector signed char vec_cnttz (vector signed char);
994 vector unsigned char vec_cnttz (vector unsigned char);
995 vector signed short vec_cnttz (vector signed short);
996 vector unsigned short vec_cnttz (vector unsigned short);
997 vector signed int vec_cnttz (vector signed int);
998 vector unsigned int vec_cnttz (vector unsigned int);
999 vector signed long long vec_cnttz (vector signed long long);
1000 vector unsigned long long vec_cnttz (vector unsigned long long);
1002 2017-07-25 Andrew Pinski <apinski@cavium.com>
1004 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
1005 accesses where the use is for the first operand of a BIT_INSERT.
1007 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
1010 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
1011 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
1013 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
1015 * config/i386/gstabs.h: Delete.
1016 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
1018 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
1020 * config/i386/i386.c (ix86_decompose_address): Do not check for
1021 register RTX when looking at index_reg or base_reg.
1022 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
1024 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
1026 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
1027 to update EH info here.
1029 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
1031 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
1033 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
1035 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
1037 2017-07-25 Torsten Duwe <duwe@suse.de>
1039 * common.opt: Introduce -fpatchable-function-entry
1040 command line option, and its variables function_entry_patch_area_size
1041 and function_entry_patch_area_start.
1042 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
1043 including a two-value parser.
1044 * target.def (print_patchable_function_entry): New target hook.
1045 * targhooks.h (default_print_patchable_function_entry): New function.
1046 * targhooks.c (default_print_patchable_function_entry): Likewise.
1047 * toplev.c (process_options): Switch off IPA-RA if
1048 patchable function entries are being generated.
1049 * varasm.c (assemble_start_function): Look at the
1050 patchable-function-entry command line switch and current
1051 function attributes and maybe generate NOP instructions by
1052 calling the print_patchable_function_entry hook.
1053 * doc/extend.texi: Document patchable_function_entry attribute.
1054 * doc/invoke.texi: Document -fpatchable_function_entry
1055 command line option.
1056 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
1058 * doc/tm.texi: Re-generate.
1060 2017-07-25 Jakub Jelinek <jakub@redhat.com>
1063 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
1064 TARGET_AVX512DQ rather than TARGET_AVX512BW.
1066 2017-07-25 Tamar Christina <tamar.christina@arm.com>
1068 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
1070 2017-07-25 Richard Biener <rguenther@suse.de>
1072 PR tree-optimization/81455
1073 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
1074 not walk in cycles when looking for guards.
1076 2017-07-25 Richard Biener <rguenther@suse.de>
1078 PR tree-optimization/81529
1079 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
1080 when optimizing backedge uses.
1082 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
1084 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
1086 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
1087 to dl_section_ref. On AIX, append an expression to subtract
1088 the size of the section length to dl_section_ref.
1090 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
1092 * configure.ac: If any of the config.* scripts fail, exit 1.
1093 * configure: Regenerate.
1095 2017-07-25 Richard Biener <rguenther@suse.de>
1098 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
1099 of immediate uses, be more verbose on errors.
1101 2017-07-25 Richard Biener <rguenther@suse.de>
1103 PR tree-optimization/81510
1104 * tree-vect-loop.c (vect_is_simple_reduction): When the
1105 reduction stmt is not inside the loop bail out.
1107 2017-07-25 Richard Biener <rguenther@suse.de>
1109 PR tree-optimization/81303
1110 * tree-vect-loop-manip.c (vect_loop_versioning): Build
1111 profitability check against LOOP_VINFO_NITERSM1.
1113 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
1115 * domwalk.c (cmp_bb_postorder): Simplify.
1116 (sort_bbs_postorder): New function. Use it...
1117 (dom_walker::walk): ...here to optimize common cases.
1119 2017-07-25 Martin Liska <mliska@suse.cz>
1122 * ipa-visibility.c (function_and_variable_visibility): Make the
1123 redirection just on target that supports aliasing.
1124 Fix GNU coding style.
1126 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
1129 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
1131 * config/arm/rtems.h: Likewise.
1132 * config/bfin/rtems.h: Likewise.
1133 * config/i386/rtemself.h: Likewise.
1134 * config/lm32/rtems.h: Likewise.
1135 * config/m32c/rtems.h: Likewise.
1136 * config/m68k/rtemself.h: Likewise.
1137 * config/microblaze/rtems.h: Likewise.
1138 * config/mips/rtems.h: Likewise.
1139 * config/moxie/rtems.h: Likewise.
1140 * config/nios2/rtems.h: Likewise.
1141 * config/powerpcspe/rtems.h: Likewise.
1142 * config/rs6000/rtems.h: Likewise.
1143 * config/rtems.h: Likewise.
1144 * config/sh/rtems.h: Likewise.
1145 * config/sh/rtemself.h: Likewise.
1146 * config/sparc/rtemself.h: Likewise.
1148 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
1151 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
1152 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
1153 * tree-ssa-structalias.c (alias_get_name): Same.
1155 2017-07-25 Bin Cheng <bin.cheng@arm.com>
1158 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
1159 instructions if no du chain is found.
1161 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
1163 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
1165 2017-07-25 Richard Biener <rguenther@suse.de>
1168 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
1171 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
1173 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
1175 (ISA_2_7_MASKS_SERVER): Likewise.
1176 (ISA_3_0_MASKS_IEEE): Likewise.
1177 (OTHER_P8_VECTOR_MASKS): Likewise.
1178 (OTHER_VSX_VECTOR_MASKS): Likewise.
1179 (POWERPC_MASKS): Likewise.
1180 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
1181 duplicate list of options.
1182 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
1183 explicit -mupper-regs options.
1184 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
1185 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
1186 alias for -mupper-regs-df.
1187 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
1188 (rs6000_init_hard_regno_mode_ok): Likewise.
1189 (rs6000_option_override_internal): Likewise.
1190 (rs6000_opt_masks): Likewise.
1191 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
1192 options in terms of whether -mvsx or -mpower8-vector was used.
1193 (TARGET_UPPER_REGS_DI): Likewise.
1194 (TARGET_UPPER_REGS_SF): Likewise.
1195 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
1196 -mupper-regs-* options.
1198 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
1200 * passes.c (emergency_dump_function): Print some empty lines and a
1201 header before the RTL dump.
1203 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
1205 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
1207 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
1210 * config/aarch64/aarch64.c (aarch64_classify_symbol):
1211 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
1213 2017-07-24 Carl Love <cel@us.ibm.com>
1215 * config/rs6000/rs6000-c.c: Add support for built-in functions
1216 vector float vec_extract_fp32_from_shorth (vector unsigned short);
1217 vector float vec_extract_fp32_from_shortl (vector unsigned short);
1218 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
1219 vec_extract_fp_from_shortl): Add defines for the two builtins.
1220 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
1221 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
1223 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
1224 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
1225 * doc/extend.texi: Update the built-in documentation file for the
1226 new built-in function.
1228 2017-07-24 Jakub Jelinek <jakub@redhat.com>
1231 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
1233 * doc/generic.texi: Likewise.
1234 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
1235 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
1237 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
1239 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
1240 (aarch64_mls_elt_merge<mode>): Likewise.
1242 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
1244 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
1245 having __cxa_atexit.
1247 2017-07-23 Michael Collison <michael.collison@arm.com>
1249 * config/arm/arm.c (arm_option_override): Deprecate
1250 use of -mstructure-size-boundary.
1251 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
1252 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
1254 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1257 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
1258 Reduce cost estimate for direct moves.
1260 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
1263 * config/i386/i386.c (ix86_option_override_internal): Disable
1264 BMI, BMI2 and TBM instructions for -m16.
1266 2017-07-21 Carl Love <cel@us.ibm.com>
1268 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
1269 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
1270 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
1271 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
1272 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
1273 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
1274 VMULOSW): New enum "unspec" values.
1275 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
1276 altivec_vmulosw): New patterns.
1277 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
1278 VMULOSW): Add definitions.
1280 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
1282 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
1283 (qdf24xx): Likewise.
1284 * config/aarch64/aarch64-options-extensions.def (rdma); New.
1285 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
1286 (AARCH64_FL_V8_1): Renumber.
1287 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
1288 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
1289 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
1290 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
1291 rdma to feature modifiers list.
1293 2017-07-21 Yury Gribov <tetra2005@gmail.com>
1296 * ipa-visibility (function_and_variable_visibility): Convert
1297 recursive PLT call to direct call if appropriate.
1299 2017-07-21 Andrew Pinski <apinski@cavium.com>
1301 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
1302 operand 1 to see if the types precision matches.
1303 * fold-const.c (operand_equal_p): Likewise.
1305 2017-07-21 Richard Biener <rguenther@suse.de>
1307 PR tree-optimization/81303
1308 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
1309 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
1310 (vect_peeling_hash_get_lowest_cost): Adjust.
1311 (vect_enhance_data_refs_alignment): Likewise. Use
1312 vect_get_peeling_costs_all_drs to compute the penalty for no
1313 peeling to match up costs.
1315 2017-07-21 Richard Biener <rguenther@suse.de>
1317 PR tree-optimization/81500
1318 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
1319 we didn't identify a reduction path.
1321 2017-07-21 Tom de Vries <tom@codesourcery.com>
1322 Cesar Philippidis <cesar@codesourcery.com>
1324 PR gcov-profile/81442
1325 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
1328 2017-07-21 Tom de Vries <tom@codesourcery.com>
1331 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
1333 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
1334 nvptx_override_options_after_change.
1336 2017-07-21 Ulrich Drepper <drepper@redhat.com>
1338 * dwarf2out.c (output_file_names): Avoid double testing for
1341 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
1343 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
1345 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
1347 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
1349 (try_crossjump_to_edge): Do not punt on partitioned functions.
1351 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
1353 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
1354 Put all BBs reachable only via paths crossing cold region to cold
1356 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
1358 2016-07-21 Richard Biener <rguenther@suse.de>
1360 PR tree-optimization/81303
1361 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
1362 into account prologue and epilogue iterations when raising
1363 min_profitable_iters to sth at least covering one vector iteration.
1365 2017-07-21 Tamar Christina <tamar.christina@arm.com>
1367 * config/arm/arm.c (arm_test_cpu_arch_dat):
1370 2017-07-20 Nathan Sidwell <nathan@acm.org>
1372 Remove TYPE_METHODS.
1373 * tree.h (TYPE_METHODS): Delete.
1374 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
1375 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
1376 (dbxout_type_methods): Scan TYPE_FIELDS.
1377 (dbxout_type): Don't check TYPE_METHODS here.
1378 * function.c (use_register_for_decl): Always ignore register for
1379 class types when not optimizing.
1380 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
1381 * tree.c (free_lang_data_in_type): Stitch out member functions and
1382 templates from TYPE_FIELDS.
1383 (build_distinct_type_copy, verify_type_variant,
1384 verify_type): Member fns are on TYPE_FIELDS.
1385 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
1386 * tree-pretty-print.c (dump_generic_node): Likewise.
1388 2017-07-20 Jakub Jelinek <jakub@redhat.com>
1391 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
1392 V2TImode and V4TImode.
1393 (ix86_expand_vector_extract): Likewise.
1394 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
1395 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
1396 (ssescalarmode): Handle V4TImode and V2TImode.
1397 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
1398 (*vec_extractv2ti, *vec_extractv4ti): New insns.
1399 (VEXTRACTI128_MODE): New mode iterator.
1400 (splitter for *vec_extractv?ti first element): New.
1401 (VEC_INIT_MODE): New mode iterator.
1402 (vec_init<mode>): Consolidate 3 expanders into one using
1403 VEC_INIT_MODE mode iterator.
1405 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
1407 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
1408 non_spilled_static_chain_regno_p.
1410 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
1412 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
1414 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
1416 * bb-reorder.c (connect_traces): Allow copying of blocks within
1419 2017-07-20 Richard Biener <rguenther@suse.de>
1421 * gimple.h (gimple_phi_result): Add gphi * overload.
1422 (gimple_phi_result_ptr): Likewise.
1423 (gimple_phi_arg): Likewise. Adjust index assert to only
1424 allow actual argument accesses rather than all slots available
1426 (gimple_phi_arg_def): Add gphi * overload.
1427 * tree-phinodes.c (make_phi_node): Initialize only actual
1429 (resize_phi_node): Clear memory not covered by old node,
1430 do not initialize excess argument slots.
1431 (reserve_phi_args_for_new_edge): Initialize new argument slot
1434 2017-07-20 Bin Cheng <bin.cheng@arm.com>
1436 PR tree-optimization/81388
1438 2016-07-21 Bin Cheng <bin.cheng@arm.com>
1440 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
1441 by removing computation of may_be_zero.
1443 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
1444 Tom de Vries <tom@codesourcery.com>
1447 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
1448 when gimple level profile disagrees with what RTL expander did.
1450 2017-07-20 Richard Biener <rguenther@suse.de>
1452 PR tree-optimization/61171
1453 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
1454 (vect_analyze_stmt): Add slp instance parameter.
1455 (vectorizable_reduction): Likewise.
1456 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
1457 (vect_is_simple_reduction): Deal with chains not detected
1458 as SLP reduction chain, specifically not properly associated
1459 chains containing a mix of plus/minus.
1460 (get_reduction_op): Remove.
1461 (get_initial_defs_for_reduction): Simplify, pass in whether
1462 this is a reduction chain, pass in the SLP node for the PHIs.
1463 (vect_create_epilog_for_reduction): Get the SLP instance as
1465 (vectorizable_reduction): Get the SLP instance as arg.
1466 During analysis remember the SLP node with the PHIs in the
1467 instance. Simplify getting at the vectorized reduction PHIs.
1468 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
1469 through SLP instance.
1470 (vect_slp_analyze_operations): Likewise.
1471 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
1472 (vect_transform_stmt): Likewise.
1474 2017-07-20 Tom de Vries <tom@codesourcery.com>
1476 PR tree-optimization/81489
1477 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
1478 read of phi arg location to before loop that modifies phi.
1480 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
1482 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
1485 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
1488 * except.c (execute): Fix ordering issue.
1490 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
1492 PR rtl-optimization/81423
1493 * combine.c (make_compound_operation_int): Don't try to optimize
1494 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
1496 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
1498 PR rtl-optimization/81423
1499 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
1500 with a constant that is -1 in the truncated to mode.
1502 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
1504 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
1505 (determine_unlikely_bbs): ... here.
1506 * predict.h (propagate_unlikely_bbs_forward): Declare.
1507 * cfgexpand.c (pass_expand::execute): Use it.
1508 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
1510 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
1511 propagate_unlikely_bbs_forward.
1513 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
1516 * except.c (maybe_add_nop_after_section_switch): New function.
1519 2017-07-19 Tom de Vries <tom@codesourcery.com>
1521 * gimple.h (gimple_phi_set_arg): Make assert more strict.
1523 2017-07-19 Tom de Vries <tom@codesourcery.com>
1525 * gimple.h (gimple_phi_arg): Make assert more strict.
1527 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
1529 * config.gcc (powerpc*-*-*): Add mmintrin.h.
1530 * config/rs6000/mmintrin.h: New file.
1531 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
1533 2017-07-19 Jakub Jelinek <jakub@redhat.com>
1535 PR tree-optimization/81346
1536 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
1538 2017-07-19 Tom de Vries <tom@codesourcery.com>
1540 * config/nvptx/nvptx.md (VECIM): Add V2DI.
1542 2017-07-19 Tom de Vries <tom@codesourcery.com>
1544 * config/nvptx/nvptx-modes.def: Add V2DImode.
1545 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
1546 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
1547 (nvptx_output_mov_insn): Handle lack of mov.b128.
1548 (nvptx_print_operand): Handle 'H' and 'L' codes.
1549 (nvptx_vector_mode_supported): Allow V2DImode.
1550 (nvptx_preferred_simd_mode): New function.
1551 (nvptx_data_alignment): New function.
1552 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
1553 nvptx_preferred_simd_mode.
1554 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
1556 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
1558 2017-07-19 Tom de Vries <tom@codesourcery.com>
1560 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
1561 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
1562 (nvptx_vector_mode_supported): New function. Allow V2SImode.
1563 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
1564 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
1565 (mov<VECIM>_insn): New define_insn.
1566 (define_expand "mov<VECIM>): New define_expand.
1568 2017-07-19 Tom de Vries <tom@codesourcery.com>
1570 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
1572 2017-07-19 Jakub Jelinek <jakub@redhat.com>
1574 PR tree-optimization/81346
1575 * fold-const.h (fold_div_compare, range_check_type): Declare.
1576 * fold-const.c (range_check_type): New function.
1577 (build_range_check): Use range_check_type.
1578 (fold_div_compare): No longer static, rewritten into
1579 a match.pd helper function.
1580 (fold_comparison): Don't call fold_div_compare here.
1581 * match.pd (X / C1 op C2): New optimization using fold_div_compare
1584 2017-07-19 Nathan Sidwell <nathan@acm.org>
1586 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
1587 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
1588 * tree.c (find_decls_types_r, verify_type): Use
1589 TYPE_{MIN,MAX}_VALUE_RAW.
1590 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
1591 (hash_tree): Likewise.
1592 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
1594 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
1597 2017-07-18 Tom de Vries <tom@codesourcery.com>
1600 * omp-expand.c (expand_omp_for_static_chunk): Handle
1601 equal-argument loop exit phi.
1603 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
1606 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
1607 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
1608 operand 2 predicate.
1609 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
1610 operand 2 predicate.
1611 (ror,rol -> rorx splitters): Use const_int_operand as
1612 operand 2 predicate.
1614 2017-06-18 Richard Biener <rguenther@suse.de>
1616 PR tree-optimization/81410
1617 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
1618 the gap in the ! slp_perm SLP case after each group.
1620 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
1623 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
1626 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
1629 * predict.c (set_even_probabilities): Cleanup; do not affect
1630 probabilities that are already known.
1631 (combine_predictions_for_bb): Call even when count is set.
1633 2017-07-18 Nathan Sidwell <nathan@acm.org>
1635 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
1638 2017-07-18 Bin Cheng <bin.cheng@arm.com>
1641 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
1642 optimization for loop niter analysis.
1644 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
1647 * config/avr/avr.c (avr_optimize_casesi): Don't use
1648 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
1650 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
1652 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
1653 body_cost_vec from _vect_peel_extended_info.
1654 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
1655 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
1658 2017-07-18 Bin Cheng <bin.cheng@arm.com>
1660 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
1662 2017-07-18 Richard Biener <rguenther@suse.de>
1664 PR tree-optimization/80620
1665 PR tree-optimization/81403
1666 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
1667 info when re-using a VN table entry.
1669 2017-07-18 Richard Biener <rguenther@suse.de>
1671 PR tree-optimization/81418
1672 * tree-vect-loop.c (vectorizable_reduction): Properly compute
1673 vectype_in. Verify that with lane-reducing reduction operations
1674 we have a single def-use cycle.
1676 2017-07-17 Carl Love <cel@us.ibm.com>
1678 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
1680 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
1681 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
1682 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
1683 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
1684 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
1685 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
1686 VMULOSW): New enum "unspec" values.
1687 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
1688 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
1689 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
1690 altivec_vmulosw): New patterns.
1691 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
1692 VMULOSW): Add definitions.
1693 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
1695 * config/alpha/alpha.c: Include predict.h.
1697 2017-07-17 Yury Gribov <tetra2005@gmail.com>
1699 * tree-vrp.c (compare_assert_loc): Fix comparison function
1700 to return predictable results.
1702 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1704 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
1707 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
1708 * doc/invoke.texi (mexpand-adddi): Update text.
1710 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1712 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
1713 that also clobbers the CC register. The old expand code is moved
1715 (*arc_clzsi2): ... here.
1716 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
1717 the CC register. The old expand code is moved to ...
1718 (arc_ctzsi2): ... here.
1720 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1722 * config/arc/arc.opt (mindexed-loads): Use initial value
1723 TARGET_INDEXED_LOADS_DEFAULT.
1724 (mauto-modify-reg): Use initial value
1725 TARGET_AUTO_MODIFY_REG_DEFAULT.
1726 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
1727 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
1728 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
1729 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
1731 2017-07-17 Martin Liska <mliska@suse.cz>
1734 * opts.c (finish_options): Do not allow -fgnu-tm
1735 w/ -fsanitize={kernel-,}address. Say sorry.
1737 2017-07-17 Bin Cheng <bin.cheng@arm.com>
1740 * tree-loop-distribution.c (classify_partition): Only assert on
1741 numer of iterations.
1742 (merge_dep_scc_partitions): Delete prameter. Update function call.
1743 (distribute_loop): Remove code handling loop with unknown niters.
1744 (pass_loop_distribution::execute): Skip loop with unknown niters.
1746 2017-07-17 Bin Cheng <bin.cheng@arm.com>
1749 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
1750 function sort_partitions_by_post_order.
1752 2017-07-17 Bin Cheng <bin.cheng@arm.com>
1754 PR tree-optimization/81374
1755 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
1756 the max index of basic blocks, rather than number of basic blocks.
1758 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1760 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
1762 (arc_legitimate_pic_operand_p): Likewise.
1763 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
1765 (arc_needs_pcl_p): Likewise.
1766 (arc_legitimate_pc_offset_p): Likewise.
1767 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
1768 function is also used in constrains.md.
1769 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
1770 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
1771 PLUS. Only return true/false in known cases, otherwise assert.
1772 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
1773 is already called in arc_legitimate_constant_p.
1774 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
1776 (LEGITIMATE_PIC_OPERAND_P): Use
1777 arc_raw_symbolic_reference_mentioned_p function.
1778 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
1783 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
1784 Andrew Burgess <andrew.burgess@embecosm.com>
1786 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
1787 (arc_return_address_register): New function.
1788 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
1789 (arc_handle_fndecl_attribute): Add naked attribute.
1790 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
1791 (TARGET_WARN_FUNC_RETURN): Likewise.
1792 (arc_allocate_stack_slots_for_args): New function.
1793 (arc_warn_func_return): Likewise.
1794 (machine_function): Change type fn_type.
1795 (arc_compute_function_type): Consider new naked function type,
1796 change function return type.
1797 (arc_must_save_register): Adapt to handle new
1798 arc_compute_function_type's return type.
1799 (arc_expand_prologue): Likewise.
1800 (arc_expand_epilogue): Likewise.
1801 (arc_return_address_regs): Delete.
1802 (arc_return_address_register): New function.
1803 (arc_epilogue_uses): Use above function.
1804 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
1805 (arc_function_type): Change encoding, add naked type.
1806 (ARC_INTERRUPT_P): Change to handle the new encoding.
1807 (ARC_FAST_INTERRUPT_P): Likewise.
1808 (ARC_NORMAL_P): Define.
1809 (ARC_NAKED_P): Likewise.
1810 (arc_compute_function_type): Delete prototype.
1811 * config/arc/arc.md (in_ret_delay_slot): Use
1812 arc_return_address_register function.
1813 (simple_return): Likewise.
1814 (p_return_i): Likewise.
1816 2017-07-17 Jakub Jelinek <jakub@redhat.com>
1818 PR tree-optimization/81428
1819 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
1820 can't be built for those types.
1822 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
1824 Remove stuff dead since r239246.
1826 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
1827 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
1828 (avr_inform_devices): Remove dead stuff.
1830 2017-07-17 Tamar Christina <tamar.christina@arm.com>
1832 * config/arm/arm_neon.h: Fix softp typo.
1834 2017-07-17 Jakub Jelinek <jakub@redhat.com>
1836 PR tree-optimization/81365
1837 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
1838 aggregate moves onto bb predecessor edges, make sure there are no
1839 loads that could alias the lhs in between the start of bb and the
1842 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
1845 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
1846 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
1847 [LSHIFTRT, outer_code = TRUNCATE]: Same.
1849 2017-07-17 Jakub Jelinek <jakub@redhat.com>
1851 PR tree-optimization/81396
1852 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
1853 (init_symbolic_number): Initialize it to 1.
1854 (perform_symbolic_merge): Add n_ops from both operands into the new
1856 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
1857 without base_addr as useless if they need more than one operation.
1858 (bswap_replace): Handle !bswap case for NULL base_addr.
1860 2017-07-17 Tom de Vries <tom@codesourcery.com>
1863 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
1866 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
1868 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
1869 conditional builtin define __FIX_LEON3FT_B2BST.
1871 2017-07-17 Daniel Cederman <cederman@gaisler.com>
1873 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
1874 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
1877 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
1879 PR rtl-optimization/81424
1880 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
1881 to remove potential trapping from operands if -fnon-call-exceptions.
1883 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1885 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
1886 profile_proability for scalling.
1887 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
1889 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1891 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
1893 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1895 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
1896 fixpoint arithmetics.
1898 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1900 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
1901 fixpoint arithmetics.
1903 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1905 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
1906 fixpoint arithmetics.
1908 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
1910 * profile-count.h (profile_probability::from_reg_br_prob_note,
1911 profile_probability::to_reg_br_prob_note): New functions.
1912 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
1913 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
1914 * predict.c (probability_reliable_p): Update.
1915 (edge_probability_reliable_p): Update.
1916 (br_prob_note_reliable_p): Update.
1917 (invert_br_probabilities): Update.
1918 (add_reg_br_prob_note): New function.
1919 (combine_predictions_for_insn): Update.
1920 * asan.c (asan_clear_shadow): Update.
1921 * cfgbuild.c (compute_outgoing_frequencies): Update.
1922 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
1923 (update_br_prob_note): Update.
1924 (rtl_verify_edges): Update.
1925 (purge_dead_edges): Update.
1926 (fixup_reorder_chain): Update.
1927 * emit-rtl.c (try_split): Update.
1928 * ifcvt.c (cond_exec_process_insns): Update.
1929 (cond_exec_process_if_block): Update.
1930 (dead_or_predicable): Update.
1931 * internal-fn.c (expand_addsub_overflow): Update.
1932 (expand_neg_overflow): Update.
1933 (expand_mul_overflow): Update.
1934 * loop-doloop.c (doloop_modify): Update.
1935 * loop-unroll.c (compare_and_jump_seq): Update.
1936 * optabs.c (emit_cmp_and_jump_insn_1): Update.
1937 * predict.h: Update.
1938 * reorg.c (mostly_true_jump): Update.
1940 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
1941 * config/alpha/alpha.c (emit_unlikely_jump): Update.
1942 * config/arc/arc.c: (emit_unlikely_jump): Update.
1943 * config/arm/arm.c: (emit_unlikely_jump): Update.
1944 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
1945 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
1946 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
1947 (ix86_print_operand): Update.
1948 (ix86_split_fp_branch): Update.
1949 (predict_jump): Update.
1950 * config/ia64/ia64.c (ia64_print_operand): Update.
1951 * config/mmix/mmix.c (mmix_print_operand): Update.
1952 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
1953 (rs6000_expand_split_stack_prologue): Update.
1954 * config/rs6000/rs6000.c: Update.
1955 * config/s390/s390.c (s390_expand_vec_strlen): Update.
1956 (s390_expand_vec_movstr): Update.
1957 (s390_expand_cs_tdsi): Update.
1958 (s390_expand_split_stack_prologue): Update.
1959 * config/sh/sh.c (sh_print_operand): Update.
1960 (expand_cbranchsi4): Update.
1961 (expand_cbranchdi4): Update.
1962 * config/sparc/sparc.c (output_v9branch): Update.
1963 * config/spu/spu.c (get_branch_target): Update.
1964 (ea_load_store_inline): Update.
1965 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
1966 * config/tilepro/tilepro.c: Update.
1968 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
1970 * gimplify.c (mostly_copy_tree_r): Revert latest change.
1971 (gimplify_save_expr): Likewise.
1973 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1975 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
1977 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1979 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
1981 * timevar.def (TV_IPA_FNSUMMARY): Define.
1983 2017-07-16 Daniel Cederman <cederman@gaisler.com>
1985 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
1986 to back store errata sensitive sequence from being generated.
1987 (sqrtdf2_fix): Likewise.
1989 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
1991 * tree-ssa-threadupdate.c (compute_path_counts,
1992 update_joiner_offpath_counts): Use profile_probability.
1994 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
1997 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
1999 * config/arm/arm-c.c (arm_cpu_builtins): Define
2000 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
2002 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
2004 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
2005 array entries to represent __ieee128 versions of the
2006 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
2007 scalar_extract_sig, and scalar_insert_exp built-in functions.
2008 (altivec_resolve_overloaded_builtin): Add special case handling
2009 for the __builtin_scalar_insert_exp function, as represented by
2010 the P9V_BUILTIN_VEC_VSIEDP constant.
2011 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
2012 exponent support for __ieee128 argument.
2013 (VSESQP): Add scalar extract signature support for __ieee128
2015 (VSTDCNQP): Add scalar test negative support for __ieee128
2017 (VSIEQP): Add scalar insert exponent support for __int128 argument
2018 with __ieee128 result.
2019 (VSIEQPF): Add scalar insert exponent support for __ieee128
2020 argument with __ieee128 result.
2021 (VSTDCQP): Add scalar test data class support for __ieee128
2023 (VSTDCNQP): Add overload support for scalar test negative with
2025 (VSTDCQP): Add overload support for scalar test data class
2027 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
2029 (UNSPEC_VSX_SIEXPQP): New constant.
2030 (xsxexpqp): New insn for VSX scalar extract exponent quad
2032 (xsxsigqp): New insn for VSX scalar extract significand quad
2034 (xsiexpqpf): New insn for VSX scalar insert exponent quad
2035 precision with floating point argument.
2036 (xststdcqp): New expand for VSX scalar test data class quad
2038 (xststdcnegqp): New expand for VSX scalar test negative quad
2040 (xststdcqp): New insn to match expansions for VSX scalar test data
2041 class quad precision and VSX scalar test negative quad precision.
2042 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
2043 special case operand checking to enforce that second operand of
2044 VSX scalar test data class with quad precision argument is a 7-bit
2046 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
2047 prototypes and descriptions of __ieee128 versions of
2048 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
2049 scalar_test_data_class, and scalar_test_neg built-in functions.
2051 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2053 PR tree-optimization/81162
2054 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
2055 replace a negate with an add.
2057 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
2059 * doc/invoke.texi (arm/-mcpu): Document +crypto.
2061 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
2063 * config/arm/arm-c.c (arm_cpu_builtins): Define
2064 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
2066 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
2068 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
2069 (armv8-r): Set ARM Cortex-R52 as default CPU.
2070 * config/arm/arm-tables.opt: Regenerate.
2071 * config/arm/arm-tune.md: Regenerate.
2072 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
2074 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
2075 extension for -mcpu=cortex-r52.
2077 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
2079 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
2080 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
2081 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
2082 (fp-armv8): Define it as FP_ARMv8 only.
2083 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
2084 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
2086 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
2087 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
2088 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
2089 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
2090 than TARGET_FPU_ARMV8.
2091 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
2092 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
2093 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
2095 * config/arm/neon.md (neon_vrint): Likewise.
2096 (neon_vcvt): Likewise.
2097 (neon_<fmaxmin_op><mode>): Likewise.
2098 (<fmaxmin><mode>3): Likewise.
2099 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
2100 * config/arm/predicates.md (arm_cond_move_operator): Check against
2101 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
2103 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
2105 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
2108 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2110 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
2111 loop in comment with memset.
2113 2017-07-14 Martin Liska <mliska@suse.cz>
2115 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
2116 * dwarf2out.c (is_java): Remove the function.
2117 (output_pubname): Remove usage of the function.
2118 (lower_bound_default): Remove usage of DW_LANG_Java.
2119 (gen_compile_unit_die): Likewise.
2120 * gcc.c: Remove compiler defaults for .java and .zip files.
2121 * gimple-expr.c (remove_suffix): Change as there's no longer
2122 extension than 4-letter one.
2123 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
2124 (gimplify_save_expr): Likewise.
2125 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
2126 as it's possible even for other languages than Java.
2127 * langhooks.h (struct lang_hooks): Remove Java from a comment.
2128 * lto-opts.c (lto_write_options): Remove reference to Java.
2129 * opts.c (strip_off_ending): Update file extension handling.
2130 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
2131 * tree-eh.c (lower_resx): Likewise.
2132 * tree.c (free_lang_data_in_type): Remove dead code.
2133 (find_decls_types_r): Likewise.
2134 (build_common_builtin_nodes): Remove Java from a comment.
2135 (verify_type): Remove dead code.
2136 * varasm.c (assemble_external): Remove Java from a comment.
2138 2017-07-14 Martin Liska <mliska@suse.cz>
2140 * opts.c (finish_options): Add quotes.
2141 (common_handle_option): Likewise.
2143 2017-07-14 Martin Liska <mliska@suse.cz>
2145 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
2146 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
2148 * dwarf2out.c (lower_bound_default): Do not handle
2150 (gen_compile_unit_die): Likewise.
2151 * gcc.c: Remove default extension binding for GNU Pascal.
2152 * stmt.c: Remove Pascal language from a comment.
2153 * xcoffout.c: Likewise.
2155 2017-07-13 David Malcolm <dmalcolm@redhat.com>
2158 * diagnostic-show-locus.c (fixit_cmp): New function.
2159 (layout::layout): Sort m_fixit_hints.
2160 (column_range::column_range): Assert that the values are valid.
2161 (struct char_span): New struct.
2162 (correction::overwrite): New method.
2163 (struct source_line): New struct.
2164 (line_corrections::add_hint): Add assertions. Reimplement memcpy
2165 calls in terms of classes source_line and char_span, and
2166 correction::overwrite.
2167 (selftest::test_overlapped_fixit_printing_2): New function.
2168 (selftest::diagnostic_show_locus_c_tests): Call it.
2170 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
2172 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
2173 early if there is no lhs.
2175 2017-07-13 Martin Liska <mliska@suse.cz>
2177 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
2178 (gen_reference_type_die): Likewise.
2179 * stor-layout.c: Remove Pascal-related comment.
2181 2017-07-13 Martin Liska <mliska@suse.cz>
2183 * opts.c (finish_options): Add quotes to error messages.
2184 (parse_sanitizer_options): Likewise.
2186 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
2188 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
2190 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
2192 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
2194 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
2196 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
2198 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
2200 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
2203 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
2204 provides the hardware capability bits, define the macro
2205 __BUILTIN_CPU_SUPPORTS__.
2206 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
2207 if GLIBC does not provide the hardware capability bits. Add a
2208 gcc_unreachable call if the built-in cpu function is neither
2209 __builtin_cpu_is nor __builtin_cpu_supports.
2210 (rs6000_get_function_versions_dispatcher): Change the warning
2211 that an old GLIBC is used which does not export the capability
2212 bits to be an error.
2213 * doc/extend.texi (target_clones attribute): Document the
2214 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
2215 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
2216 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
2217 the macros defined by GCC if the newer GLIBC is available.
2219 2017-07-12 Jeff Law <law@redhat.com>
2221 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
2222 remaining includes slightly.
2223 * config/riscv/riscv-builtins.c: Include profile-count.h.
2225 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
2228 * config/avr/avr.c (avr_set_current_function): In diagnostic
2229 messages: Quote keywords and (parts of) identifiers.
2230 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
2233 2017-07-12 Carl Love <cel@us.ibm.com>
2235 * config/rs6000/rs6000-c.c: Add support for built-in functions
2236 vector bool char vec_revb (vector bool char);
2237 vector bool short vec_revb (vector short char);
2238 vector bool int vec_revb (vector bool int);
2239 vector bool long long vec_revb (vector bool long long);
2240 * doc/extend.texi: Update the built-in documentation file for the
2241 new built-in functions.
2243 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2245 * config/s390/s390.md: Remove movcc splitter.
2247 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2249 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
2250 load/store on condition.
2252 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
2255 * config/avr/avr.c (avr_encode_section_info)
2256 [progmem && !TREE_READONLY]: Error if progmem object needs
2259 2017-07-11 Michael Collison <michael.collison@arm.com>
2261 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
2264 2017-07-11 Carl Love <cel@us.ibm.com>
2266 * config/rs6000/rs6000-c.c: Add support for builtins
2267 vector unsigned int vec_parity_lsbb (vector signed int);
2268 vector unsigned int vec_parity_lsbb (vector unsigned int);
2269 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
2270 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
2271 vector unsigned long long vec_parity_lsbb (vector signed long long);
2272 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
2273 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
2274 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
2275 * doc/extend.texi: Update the built-in documentation file for the
2276 new built-in functions.
2278 2017-07-11 David Malcolm <dmalcolm@redhat.com>
2280 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
2281 (layout::m_primary_loc): New field.
2282 (layout::layout): Initialize new field. Move location filtering
2283 logic from here to...
2284 (layout::maybe_add_location_range): ...this new method. Add
2285 support for filtering to just the lines already specified by other
2287 (layout::will_show_line_p): New method.
2288 (gcc_rich_location::add_location_if_nearby): New method.
2289 (selftest::test_add_location_if_nearby): New test function.
2290 (selftest::diagnostic_show_locus_c_tests): Call it.
2291 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
2294 2017-07-11 Tom de Vries <tom@codesourcery.com>
2296 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
2297 (bb_first_real_insn): New function.
2298 (nvptx_single): Add extra initialization of broadcasted condition
2301 2017-07-11 Nathan Sidwell <nathan@acm.org>
2303 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
2305 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
2307 * doc/extend.texi (AVR Function Attributes): Remove weblink to
2308 Binutils doc as TEXI will mess them up.
2309 * doc/invoke.texi (AVR Options): Same here.
2311 2017-07-11 Daniel Cederman <cederman@gaisler.com>
2313 * config/sparc/sparc.opt (mfix-ut700): New option.
2314 (mfix-gr712rc): Likewise.
2315 (sparc_fix_b2bst): New variable.
2316 * doc/invoke.texi (SPARC options): Document them.
2317 (ARM options): Fix warnings.
2318 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
2319 instructions to prevent sequences that can trigger the store-store
2320 errata for certain LEON3FT processors.
2321 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
2322 (sparc_option_override): Set sparc_fix_b2bst appropriately.
2323 * config/sparc/sparc.md (fix_b2bst): New attribute.
2324 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
2326 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
2329 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
2331 (*rsqrtsf2_sse): Ditto.
2333 (div<mode>3): Macroize insn from divdf3 and divsf3
2334 using MODEF mode iterator.
2336 2017-07-10 Martin Sebor <msebor@redhat.com>
2338 PR tree-optimization/80397
2339 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
2340 instead of testing for equality to INTEGER_TYPE.
2342 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
2344 * config.gcc: Remove uclibc from arc target spec.
2346 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
2348 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
2350 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
2353 * lto-wrapper.c (remove_option): New function.
2354 (merge_and_complain): Merge PIC/PIE options more realistically.
2356 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
2358 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
2362 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
2363 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
2364 * config.in: Regenerate.
2365 * configure: Regenerate.
2366 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
2367 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
2368 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
2369 (TARGET_GASISR_PROLOGUES): ...target mask.
2370 * common/config/avr/avr-common.c
2371 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
2372 Set -mgas-isr-prologues.
2373 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
2374 INSERT_PASS_BEFORE for it.
2375 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
2376 * config/avr/avr.c (avr_option_override)
2377 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
2378 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
2379 (avr_attribute_table) <no_gccisr>: Add new function attribute.
2380 (avr_set_current_function) <is_no_gccisr>: Init machine field.
2381 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
2383 (make_avr_pass_pre_proep): New function.
2384 (emit_push_sfr) <treg>: Add argument to function and use it
2386 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
2387 and set machine->gasisr.yes.
2388 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
2389 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
2390 __gcc_isr.n_pushed to .L__stack_usage.
2391 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
2392 (avr_asm_final_postscan_insn): ...this new static function.
2393 * config/avr/avr.h (machine_function)
2394 <is_no_gccisr, use_L__stack_usage>: New fields.
2395 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
2396 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
2397 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
2398 (gasisr, *gasisr): New expander and insn.
2399 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
2400 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
2401 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
2403 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
2405 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
2408 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
2410 Move jump-tables out of .text again.
2413 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
2414 (ASM_OUTPUT_ADDR_VEC): New function.
2415 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
2416 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
2417 INSN_ADDRESSes as asm comment.
2418 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
2419 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
2420 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
2421 * config/avr/avr.md (*tablejump): Adjust comment.
2422 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
2423 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
2425 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
2426 (avr_output_addr_vec): New proto.
2427 (avr_log_t) <insn_addresses>: New field.
2429 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
2432 * config/i386/i386.c (ix86_function_arg_advance): Set
2433 outgoing_args_on_stack to true if there are outgoing arguments
2435 (ix86_function_arg): Likewise.
2436 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
2437 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
2438 * config/i386/i386.h (machine_function): Add
2439 outgoing_args_on_stack.
2441 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
2443 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
2445 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
2447 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
2449 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
2450 (REAL_H): Remove $(MACHMODE_H).
2451 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
2453 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
2454 $(MACHMODE_H) and double-int.h.
2455 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
2457 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
2460 2017-07-07 Andrew Pinski <apinski@cavium.com>
2462 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
2463 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
2465 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
2467 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
2468 Add warning if GCC was not configured to link against a GLIBC that
2469 exports the hardware capability bits.
2470 (make_resolver_func): Make resolver function private and not a
2471 COMDAT function. Create the name with clone_function_name instead
2472 of make_unique_name.
2475 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
2476 correct operand in doing the split.
2478 2017-07-07 Carl Love <cel@us.ibm.com>
2480 * config/rs6000/rs6000-c: Add support for built-in function
2481 vector unsigned short vec_pack_to_short_fp32 (vector float,
2483 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
2484 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
2485 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
2486 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
2487 (convert_4f32_8i16): Add define_expand.
2488 * doc/extend.texi: Update the built-in documentation file for the
2489 new built-in function.
2491 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2493 * config/sparc/m8.md: New file.
2494 * config/sparc/sparc.md: Include m8.md.
2496 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2498 * config/sparc/sparc.opt: New option -mvis4b.
2499 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
2500 (sparc_option_override): Handle VIS4B.
2501 (enum sparc_builtins): Define
2502 SPARC_BUILTIN_DICTUNPACK{8,16,32},
2503 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
2504 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
2505 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
2506 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
2507 (check_constant_argument): New function.
2508 (sparc_vis_init_builtins): Define builtins
2509 __builtin_vis_dictunpack{8,16,32},
2510 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
2511 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
2512 __builtin_vis_fpcmpde{8,16,32}shl and
2513 __builtin_vis_fpcmpur{8,16,32}shl.
2514 (sparc_expand_builtin): Check that the constant operands to
2515 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
2516 constant and in range.
2517 * config/sparc/sparc-c.c (sparc_target_macros): Handle
2519 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
2520 (SPARC_IMM5_P): Likewise.
2521 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
2522 (enabled): Handle vis4b.
2523 (UNSPEC_DICTUNPACK): New unspec.
2524 (UNSPEC_FPCMPSHL): Likewise.
2525 (UNSPEC_FPUCMPSHL): Likewise.
2526 (UNSPEC_FPCMPDESHL): Likewise.
2527 (UNSPEC_FPCMPURSHL): Likewise.
2528 (cpu_feature): New CPU feature `vis4b'.
2529 (dictunpack{8,16,32}): New insns.
2530 (FPCSMODE): New mode iterator.
2531 (fpcscond): New code iterator.
2532 (fpcsucond): Likewise.
2533 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
2534 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
2535 (fpcmpde{8,16,32}{si,di}shl): Likewise.
2536 (fpcmpur{8,16,32}{si,di}shl): Likewise.
2537 * config/sparc/constraints.md: Define constraints `q' for unsigned
2538 2-bit integer constants and `t' for unsigned 5-bit integer
2540 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
2542 (imm5_operand_dictunpack16): Likewise.
2543 (imm5_operand_dictunpack32): Likewise.
2544 (imm2_operand): Likewise.
2545 * doc/invoke.texi (SPARC Options): Document -mvis4b.
2546 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
2547 ditunpack* and fpcmp*shl builtins.
2549 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2551 * config.gcc: Handle m8 in --with-{cpu,tune} options.
2552 * config.in: Add HAVE_AS_SPARC6 define.
2553 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
2555 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
2557 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
2558 (CPP_CPU_SPEC): Handle m8.
2559 (ASM_CPU_SPEC): Likewise.
2560 * config/sparc/sparc-opts.h (enum processor_type): Add
2562 * config/sparc/sparc.c (m8_costs): New struct.
2563 (sparc_option_override): Handle TARGET_CPU_m8.
2564 (sparc32_initialize_trampoline): Likewise.
2565 (sparc64_initialize_trampoline): Likewise.
2566 (sparc_issue_rate): Likewise.
2567 (sparc_register_move_cost): Likewise.
2568 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
2569 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
2570 (ASM_CPU64_DEFAULT_SPEC): Likewise.
2571 (CPP_CPU_SPEC): Handle M8.
2572 (ASM_CPU_SPEC): Likewise.
2573 (AS_M8_FLAG): Define.
2574 * config/sparc/sparc.md: Add m8 to the cpu attribute.
2575 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
2576 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
2578 * configure: Regenerate.
2579 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
2582 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2584 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
2586 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
2587 ("*movdi_insn_sp32"): Do not set v3pipe.
2588 ("*movsi_insn"): Likewise.
2589 ("*movdi_insn_sp64"): Likewise.
2590 ("*movsf_insn"): Likewise.
2591 ("*movdf_insn_sp32"): Likewise.
2592 ("*movdf_insn_sp64"): Likewise.
2593 ("*zero_extendsidi2_insn_sp64"): Likewise.
2594 ("*sign_extendsidi2_insn"): Likewise.
2595 ("*mov<VM32:mode>_insn"): Likewise.
2596 ("*mov<VM64:mode>_insn_sp64"): Likewise.
2597 ("*mov<VM64:mode>_insn_sp32"): Likewise.
2598 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
2599 ("<vlop:code><VL:mode>3"): Likewise.
2600 ("*not_<vlop:code><VL:mode>3"): Likewise.
2601 ("*nand<VL:mode>_vis"): Likewise.
2602 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
2603 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
2604 ("one_cmpl<VL:mode>2"): Likewise.
2605 ("faligndata<VM64:mode>_vis"): Likewise.
2606 ("alignaddrsi_vis"): Likewise.
2607 ("alignaddrdi_vis"): Likweise.
2608 ("alignaddrlsi_vis"): Likewise.
2609 ("alignaddrldi_vis"): Likewise.
2610 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
2611 ("bmaskdi_vis"): Likewise.
2612 ("bmasksi_vis"): Likewise.
2613 ("bshuffle<VM64:mode>_vis"): Likewise.
2614 ("cmask8<P:mode>_vis"): Likewise.
2615 ("cmask16<P:mode>_vis"): Likewise.
2616 ("cmask32<P:mode>_vis"): Likewise.
2617 ("pdistn<P:mode>_vis"): Likewise.
2618 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
2620 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2622 * config/sparc/sparc.md ("subtype"): New insn attribute.
2623 ("*wrgsr_sp64"): Set insn subtype.
2624 ("*rdgsr_sp64"): Likewise.
2625 ("alignaddrsi_vis"): Likewise.
2626 ("alignaddrdi_vis"): Likewise.
2627 ("alignaddrlsi_vis"): Likewise.
2628 ("alignaddrldi_vis"): Likewise.
2629 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
2630 ("fexpand_vis"): Likewise.
2631 ("fpmerge_vis"): Likewise.
2632 ("faligndata<VM64:mode>_vis"): Likewise.
2633 ("bshuffle<VM64:mode>_vis"): Likewise.
2634 ("cmask8<P:mode>_vis"): Likewise.
2635 ("cmask16<P:mode>_vis"): Likewise.
2636 ("cmask32<P:mode>_vis"): Likewise.
2637 ("fchksm16_vis"): Likewise.
2638 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
2639 ("fmean16_vis"): Likewise.
2640 ("fp<plusminus_insn>64_vis"): Likewise.
2641 ("<plusminus_insn>v8qi3"): Likewise.
2642 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
2643 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
2644 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
2645 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
2646 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
2647 ("*movqi_insn"): Likewise.
2648 ("*movhi_insn"): Likewise.
2649 ("*movsi_insn"): Likewise.
2650 ("movsi_pic_gotdata_op"): Likewise.
2651 ("*movdi_insn_sp32"): Likewise.
2652 ("*movdi_insn_sp64"): Likewise.
2653 ("movdi_pic_gotdata_op"): Likewise.
2654 ("*movsf_insn"): Likewise.
2655 ("*movdf_insn_sp32"): Likewise.
2656 ("*movdf_insn_sp64"): Likewise.
2657 ("*zero_extendhisi2_insn"): Likewise.
2658 ("*zero_extendqihi2_insn"): Likewise.
2659 ("*zero_extendqisi2_insn"): Likewise.
2660 ("*zero_extendqidi2_insn"): Likewise.
2661 ("*zero_extendhidi2_insn"): Likewise.
2662 ("*zero_extendsidi2_insn_sp64"): Likewise.
2663 ("ldfsr"): Likewise.
2664 ("prefetch_64"): Likewise.
2665 ("prefetch_32"): Likewise.
2666 ("tie_ld32"): Likewise.
2667 ("tie_ld64"): Likewise.
2668 ("*tldo_ldub_sp32"): Likewise.
2669 ("*tldo_ldub1_sp32"): Likewise.
2670 ("*tldo_ldub2_sp32"): Likewise.
2671 ("*tldo_ldub_sp64"): Likewise.
2672 ("*tldo_ldub1_sp64"): Likewise.
2673 ("*tldo_ldub2_sp64"): Likewise.
2674 ("*tldo_ldub3_sp64"): Likewise.
2675 ("*tldo_lduh_sp32"): Likewise.
2676 ("*tldo_lduh1_sp32"): Likewise.
2677 ("*tldo_lduh_sp64"): Likewise.
2678 ("*tldo_lduh1_sp64"): Likewise.
2679 ("*tldo_lduh2_sp64"): Likewise.
2680 ("*tldo_lduw_sp32"): Likewise.
2681 ("*tldo_lduw_sp64"): Likewise.
2682 ("*tldo_lduw1_sp64"): Likewise.
2683 ("*tldo_ldx_sp64"): Likewise.
2684 ("*mov<VM32:mode>_insn"): Likewise.
2685 ("*mov<VM64:mode>_insn_sp64"): Likewise.
2686 ("*mov<VM64:mode>_insn_sp32"): Likewise.
2688 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2690 * config/sparc/sparc.md ("type"): New insn type viscmp.
2691 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
2693 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
2694 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
2695 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
2696 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
2698 ("n7_vis_logical_11cycle"): Likewise.
2699 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
2700 * config/sparc/niagara2.md ("niag3_vis": Likewise.
2701 * config/sparc/niagara.md ("niag_vis"): Likewise.
2702 * config/sparc/ultra3.md ("us3_fga"): Likewise.
2703 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
2705 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
2707 * config/sparc/sparc.md: New instruction type `bmask'.
2708 (bmaskdi_vis): Use the `bmask' type.
2709 (bmasksi_vis): Likewise.
2710 * config/sparc/ultra3.md (us3_array): Likewise.
2711 * config/sparc/niagara7.md (n7_array): Likewise.
2712 * config/sparc/niagara4.md (n4_array): Likewise.
2713 * config/sparc/niagara2.md (niag2_vis): Likewise.
2714 (niag3_vis): Likewise.
2715 * config/sparc/niagara.md (niag_vis): Likewise.
2717 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2719 * ipa-comdats.c: Remove optimize check from gate.
2720 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
2721 for functions not optimized.
2722 (ipa_fn_summary_read): Skip optimize check.
2723 (ipa_fn_summary_write): Likewise.
2724 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
2726 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
2728 (can_inline_edge_p): Check flag_pcc_struct_return for match.
2729 (check_callers): Give up on caller which is not optimized.
2730 (inline_small_functions): Likewise.
2731 (ipa_inline): Do not give up when not optimizing.
2732 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
2733 away unoptimizes cdtors.
2734 (whole_program_function_and_variable_visibility): Do
2735 ipa_discover_readonly_nonaddressable_vars in LTO mode.
2736 * ipa.c (process_references): Do not check optimize.
2737 (symbol_table::remove_unreachable_nodes): Update optimize check.
2738 (set_writeonly_bit): Update optimize check.
2739 (pass_ipa_cdtor_merge::gate): Do not check optimize.
2740 (pass_ipa_single_use::gate): Remove.
2742 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2744 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
2745 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
2746 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
2747 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
2748 permute_load, permute_store, adjust_extract, adjust_splat,
2749 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
2750 replace_swap_with_copy, dump_swap_insn_table,
2751 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
2752 recombine_lvx_pattern, recombine_stvx_pattern,
2753 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
2754 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
2755 to file rs6000-p8swap.c.
2756 * config/rs6000/rs6000-p8swap.c: New file.
2757 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
2758 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
2759 and rs6000*-*-* targets.
2761 2017-07-06 David Malcolm <dmalcolm@redhat.com>
2763 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
2765 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2767 * lto-wrapper.c (merge_and_complain): Do not merge
2768 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
2769 fsigned_zeros, ftrapping_math, fwrapv.
2770 (append_compiler_options): Do not track these options.
2771 (append_linker_options): Likewie
2773 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2775 * cgraphunit.c (cgraph_node::finalize_function): When
2776 !flag_toplevel_reorde set no_reorder flag.
2777 (varpool_node::finalize_decl): Likewise.
2778 (symbol_table::compile): Drop no toplevel reorder path.
2780 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2782 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
2783 edges; zero probability is not better than uninitialized.
2785 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
2787 * asan.h (asan_sanitize_allocas_p): Declare.
2788 * asan.c (asan_sanitize_allocas_p): New function.
2789 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
2790 (handle_builtin_alloca): Likewise.
2791 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
2792 if !asan_sanitize_allocas_p.
2793 * params.def (asan-instrument-allocas): Add new option.
2794 * params.h (ASAN_PROTECT_ALLOCAS): Define.
2795 * opts.c (common_handle_option): Disable allocas sanitization for
2798 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
2800 * asan.c: Include gimple-fold.h.
2801 (get_last_alloca_addr): New function.
2802 (handle_builtin_stackrestore): Likewise.
2803 (handle_builtin_alloca): Likewise.
2804 (asan_emit_allocas_unpoison): Likewise.
2805 (get_mem_refs_of_builtin_call): Add new parameter, remove const
2806 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
2807 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
2808 (instrument_builtin_call): Pass gimple iterator to
2809 get_mem_refs_of_builtin_call.
2810 (last_alloca_addr): New global.
2811 * asan.h (asan_emit_allocas_unpoison): Declare.
2812 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
2813 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
2814 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
2815 if function calls alloca.
2816 * gimple-fold.c (replace_call_with_value): Remove static keyword.
2817 * gimple-fold.h (replace_call_with_value): Declare.
2818 * internal-fn.c: Include asan.h.
2819 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
2820 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
2822 2017-07-06 David Malcolm <dmalcolm@redhat.com>
2824 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
2825 (C_SELFTEST_FLAGS): New.
2826 (CPP_SELFTEST_FLAGS): New.
2827 (SELFTEST_DEPS): New, from deps of s-selftest.
2828 (C_SELFTEST_DEPS): New, from deps of s-selftest.
2829 (CPP_SELFTEST_DEPS): New.
2830 (selftest): Add dependency on s-selftest-c++.
2831 (s-selftest): Rename to...
2832 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
2833 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
2834 than SELFTEST_FLAGS.
2835 (selftest-gdb): Rename to...
2836 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
2838 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
2839 (selftest-valgrind): Rename to...
2840 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
2842 (selftest-valgrind): Reintroduce as an alias for
2843 selftest-c-valgrind.
2844 (s-selftest-c++): New.
2845 (selftest-c++-gdb): New.
2846 (selftest-c++-valgrind): New.
2848 2017-07-06 Olivier Hainque <hainque@adacore.com>
2850 * gcc.c (process_command): When deciding if undefined variables
2851 should be ignored when processing specs, accept "gcc -v" as well.
2853 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
2855 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
2856 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
2858 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
2860 * config/arm/arm-cpus.in (armv8-r): Add new entry.
2861 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
2862 * config/arm/arm-tables.opt: Regenerate.
2863 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
2865 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
2867 2017-07-06 Carl Love <cel@us.ibm.com>
2869 * ChangeLog: Clean up from mid air collision
2871 2017-07-06 Carl Love <cel@us.ibm.com>
2873 * config/rs6000/rs6000-c.c: Add support for built-in functions
2874 vector signed int vec_subc (vector signed int, vector signed int);
2875 vector signed __int128 vec_subc (vector signed __int128,
2876 vector signed __int128);
2877 vector unsigned __int128 vec_subc (vector unsigned __int128,
2878 vector unsigned __int128);
2879 vector signed int vec_sube (vector signed int, vector signed int,
2881 vector unsigned int vec_sube (vector unsigned int,
2882 vector unsigned int,
2883 vector unsigned int);
2884 vector signed __int128 vec_sube (vector signed __int128,
2885 vector signed __int128,
2886 vector signed__int128);
2887 vector unsigned __int128 vec_sube (vector unsigned __int128,
2888 vector unsigned __int128,
2889 vector unsigned __int128);
2890 vector signed int vec_subec (vector signed int, vector signed int,
2892 vector unsigned int vec_subec (vector unsigned int,
2893 vector unsigned int,
2894 vector unsigned int);
2895 vector signed __int128 vec_subec (vector signed __int128,
2896 vector signed __int128,
2897 vector signed__int128);
2898 vector unsigned __int128 vec_subec (vector unsigned __int128,
2899 vector unsigned __int128,
2900 vector unsigned __int128);
2901 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
2902 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
2903 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
2904 BU_ALTIVEC_OVERLOAD_X definitions.
2905 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
2906 * doc/extend.texi: Update the built-in documentation file for the new
2909 2017-07-06 David Malcolm <dmalcolm@redhat.com>
2912 * diagnostic-show-locus.c (layout::layout): Use start and finish
2913 spelling location for the start and finish of each range.
2914 * genmatch.c (linemap_client_expand_location_to_spelling_point):
2915 Add unused aspect param.
2916 * input.c (expand_location_1): Add "aspect" param, and use it
2917 to access the correct part of the location.
2918 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
2920 (expand_location_to_spelling_point): Likewise.
2921 (linemap_client_expand_location_to_spelling_point): Add "aspect"
2922 param, and pass it to expand_location_1.
2924 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
2926 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
2927 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
2928 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
2929 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
2930 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
2931 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
2932 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
2933 _mm_maskz_getmant_ss): New intrinsics.
2934 (__builtin_ia32_getexpss128_mask): Changed to ...
2935 __builtin_ia32_getexpss128_round ... this.
2936 (__builtin_ia32_getexpsd128_mask): Changed to ...
2937 __builtin_ia32_getexpsd128_round ... this.
2938 * config/i386/i386-builtin-types.def
2939 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
2940 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
2941 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
2942 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
2943 __builtin_ia32_getmantss_mask_round): New builtins.
2944 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
2945 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
2946 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
2947 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
2948 * config/i386/sse.md
2949 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
2950 avx512f_sgetexp<mode><mask_scalar_name>
2951 <round_saeonly_scalar_name> ... this.
2952 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
2953 %0, %1, %2<round_saeonly_op3>}): Changed to ...
2954 vgetexp<ssescalarmodesuffix>
2955 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
2956 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
2957 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
2958 avx512f_vgetmant<mode><mask_scalar_name>
2959 <round_saeonly_scalar_name> ... this.
2960 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
2961 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
2962 vgetmant<ssescalarmodesuffix>
2963 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
2964 %0<mask_scalar_operand4>, %1, %2
2965 <round_saeonly_scalar_mask_op4>, %3} ... this.
2966 * config/i386/subst.md (mask_scalar_operand4,
2967 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
2968 round_saeonly_scalar_nimm_predicate): New subst attributes.
2970 2017-07-06 Julia Koval <julia.koval@intel.com>
2972 * config/i386/i386.c (ix86_erase_embedded_rounding):
2973 Remove code for old rounding pattern.
2975 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
2977 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
2979 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
2981 * doc/sourcebuild.texi (Test Directives, Variants of
2982 dg-require-support): Add documentation for dg-require-stack-check.
2984 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
2986 * config/i386/subst.md (mask_scalar, round_scalar,
2987 round_saeonly_scalar): New meta-templates.
2988 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
2989 round_scalar_mask_operand3, round_scalar_mask_op3,
2990 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
2991 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
2992 round_saeonly_scalar_constraint,
2993 round_saeonly_scalar_prefix): New subst attribute.
2994 * config/i386/sse.md
2995 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
2996 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
2997 <round_scalar_name> ... this.
2998 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
2999 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
3000 <round_scalar_name> ... this.
3001 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
3002 <sse>_vm<code><mode>3<mask_scalar_name>
3003 <round_saeonly_scalar_name> ... this.
3004 (v<plusminus_mnemonic><ssescalarmodesuffix>
3005 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
3006 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
3007 v<plusminus_mnemonic><ssescalarmodesuffix>
3008 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
3009 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
3010 (v<multdiv_mnemonic><ssescalarmodesuffix>
3011 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
3012 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
3013 v<multdiv_mnemonic><ssescalarmodesuffix>
3014 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
3015 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
3016 (v<maxmin_float><ssescalarmodesuffix>
3017 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
3018 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
3019 v<maxmin_float><ssescalarmodesuffix>
3020 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
3021 %0<mask_scalar_operand3>, %1, %<iptr>2
3022 <round_saeonly_scalar_mask_op3>} ... this.
3024 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
3026 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
3027 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
3029 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
3030 Alan Hayward <alan.hayward@arm.com>
3031 David Sherwood <david.sherwood@arm.com>
3033 * combine.c (simplify_if_then_else): Remove "enum" before
3035 * compare-elim.c (can_eliminate_compare): Likewise.
3036 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
3038 (aarch64_lookup_simd_builtin_type): Likewise.
3039 (aarch64_simd_builtin_type): Likewise.
3040 (aarch64_init_simd_builtin_types): Likewise.
3041 (aarch64_simd_expand_args): Likewise.
3042 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
3044 (aarch64_reverse_mask): Likewise.
3045 (aarch64_simd_emit_reg_reg_move): Likewise.
3046 (aarch64_gen_adjusted_ldpstp): Likewise.
3047 (aarch64_ccmp_mode_to_code): Likewise.
3048 (aarch64_operands_ok_for_ldpstp): Likewise.
3049 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
3050 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
3052 (aarch64_min_divisions_for_recip_mul): Likewise.
3053 (aarch64_reassociation_width): Likewise.
3054 (aarch64_get_condition_code_1): Likewise.
3055 (aarch64_simd_emit_reg_reg_move): Likewise.
3056 (aarch64_simd_attr_length_rglist): Likewise.
3057 (aarch64_reverse_mask): Likewise.
3058 (aarch64_operands_ok_for_ldpstp): Likewise.
3059 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
3060 (aarch64_gen_adjusted_ldpstp): Likewise.
3061 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
3063 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
3064 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
3065 (arm_lookup_simd_builtin_type): Likewise.
3066 (arm_simd_builtin_type): Likewise.
3067 (arm_init_simd_builtin_types): Likewise.
3068 (arm_expand_builtin_args): Likewise.
3069 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
3070 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
3071 (ft32_setup_incoming_varargs): Likewise.
3072 (ft32_function_arg): Likewise.
3073 (ft32_function_arg_advance): Likewise.
3074 (ft32_pass_by_reference): Likewise.
3075 (ft32_arg_partial_bytes): Likewise.
3076 (ft32_valid_pointer_mode): Likewise.
3077 (ft32_addr_space_pointer_mode): Likewise.
3078 (ft32_addr_space_legitimate_address_p): Likewise.
3079 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
3081 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
3082 (ix86_emit_outlined_ms2sysv_restore): Likewise.
3083 (iamcu_alignment): Likewise.
3084 (canonicalize_vector_int_perm): Likewise.
3085 (ix86_noce_conversion_profitable_p): Likewise.
3086 (ix86_mpx_bound_mode): Likewise.
3087 (ix86_operands_ok_for_move_multiple): Likewise.
3088 * config/microblaze/microblaze-protos.h
3089 (microblaze_expand_conditional_branch_reg): Likewise.
3090 * config/microblaze/microblaze.c
3091 (microblaze_expand_conditional_branch_reg): Likewise.
3092 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
3094 (rs6000_reassociation_width): Likewise.
3095 (rs6000_invalid_binary_op): Likewise.
3096 (fusion_p9_p): Likewise.
3097 (emit_fusion_p9_load): Likewise.
3098 (emit_fusion_p9_store): Likewise.
3099 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
3101 (riscv_hard_regno_mode_ok_p): Likewise.
3102 (riscv_address_insns): Likewise.
3103 (riscv_split_symbol): Likewise.
3104 (riscv_legitimize_move): Likewise.
3105 (riscv_function_value): Likewise.
3106 (riscv_hard_regno_nregs): Likewise.
3107 (riscv_expand_builtin): Likewise.
3108 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
3109 (riscv_build_integer): Likewise.
3110 (riscv_split_integer): Likewise.
3111 (riscv_legitimate_constant_p): Likewise.
3112 (riscv_cannot_force_const_mem): Likewise.
3113 (riscv_regno_mode_ok_for_base_p): Likewise.
3114 (riscv_valid_base_register_p): Likewise.
3115 (riscv_valid_offset_p): Likewise.
3116 (riscv_valid_lo_sum_p): Likewise.
3117 (riscv_classify_address): Likewise.
3118 (riscv_legitimate_address_p): Likewise.
3119 (riscv_address_insns): Likewise.
3120 (riscv_load_store_insns): Likewise.
3121 (riscv_force_binary): Likewise.
3122 (riscv_split_symbol): Likewise.
3123 (riscv_force_address): Likewise.
3124 (riscv_legitimize_address): Likewise.
3125 (riscv_move_integer): Likewise.
3126 (riscv_legitimize_const_move): Likewise.
3127 (riscv_legitimize_move): Likewise.
3128 (riscv_address_cost): Likewise.
3129 (riscv_subword): Likewise.
3130 (riscv_output_move): Likewise.
3131 (riscv_canonicalize_int_order_test): Likewise.
3132 (riscv_emit_int_order_test): Likewise.
3133 (riscv_function_arg_boundary): Likewise.
3134 (riscv_pass_mode_in_fpr_p): Likewise.
3135 (riscv_pass_fpr_single): Likewise.
3136 (riscv_pass_fpr_pair): Likewise.
3137 (riscv_get_arg_info): Likewise.
3138 (riscv_function_arg): Likewise.
3139 (riscv_function_arg_advance): Likewise.
3140 (riscv_arg_partial_bytes): Likewise.
3141 (riscv_function_value): Likewise.
3142 (riscv_pass_by_reference): Likewise.
3143 (riscv_setup_incoming_varargs): Likewise.
3144 (riscv_print_operand): Likewise.
3145 (riscv_elf_select_rtx_section): Likewise.
3146 (riscv_save_restore_reg): Likewise.
3147 (riscv_for_each_saved_reg): Likewise.
3148 (riscv_register_move_cost): Likewise.
3149 (riscv_hard_regno_mode_ok_p): Likewise.
3150 (riscv_hard_regno_nregs): Likewise.
3151 (riscv_class_max_nregs): Likewise.
3152 (riscv_memory_move_cost): Likewise.
3153 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
3154 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
3155 (rl78_addr_space_address_mode): Likewise.
3156 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3158 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
3159 (rs6000_reassociation_width): Likewise.
3160 (rs6000_invalid_binary_op): Likewise.
3161 (fusion_p9_p): Likewise.
3162 (emit_fusion_p9_load): Likewise.
3163 (emit_fusion_p9_store): Likewise.
3164 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
3165 (ok_for_simple_move_operands): Likewise.
3166 (ok_for_simple_move_strict_operands): Likewise.
3167 (ok_for_simple_arith_logic_operands): Likewise.
3168 (visium_legitimize_reload_address): Likewise.
3169 (visium_select_cc_mode): Likewise.
3170 (output_cbranch): Likewise.
3171 (visium_split_double_move): Likewise.
3172 (visium_expand_copysign): Likewise.
3173 (visium_expand_int_cstore): Likewise.
3174 (visium_expand_fp_cstore): Likewise.
3175 * config/visium/visium.c (visium_pass_by_reference): Likewise.
3176 (visium_function_arg): Likewise.
3177 (visium_function_arg_advance): Likewise.
3178 (visium_libcall_value): Likewise.
3179 (visium_setup_incoming_varargs): Likewise.
3180 (visium_legitimate_constant_p): Likewise.
3181 (visium_legitimate_address_p): Likewise.
3182 (visium_legitimize_address): Likewise.
3183 (visium_secondary_reload): Likewise.
3184 (visium_register_move_cost): Likewise.
3185 (visium_memory_move_cost): Likewise.
3186 (prepare_move_operands): Likewise.
3187 (ok_for_simple_move_operands): Likewise.
3188 (ok_for_simple_move_strict_operands): Likewise.
3189 (ok_for_simple_arith_logic_operands): Likewise.
3190 (visium_function_value_1): Likewise.
3191 (rtx_ok_for_offset_p): Likewise.
3192 (visium_legitimize_reload_address): Likewise.
3193 (visium_split_double_move): Likewise.
3194 (visium_expand_copysign): Likewise.
3195 (visium_expand_int_cstore): Likewise.
3196 (visium_expand_fp_cstore): Likewise.
3197 (visium_split_cstore): Likewise.
3198 (visium_select_cc_mode): Likewise.
3199 (visium_split_cbranch): Likewise.
3200 (output_cbranch): Likewise.
3201 (visium_print_operand_address): Likewise.
3202 * expmed.c (flip_storage_order): Likewise.
3203 * expmed.h (emit_cstore): Likewise.
3204 (flip_storage_order): Likewise.
3205 * genrecog.c (validate_pattern): Likewise.
3206 * hsa-gen.c (gen_hsa_addr): Likewise.
3207 * internal-fn.c (expand_arith_overflow): Likewise.
3208 * ira-color.c (allocno_copy_cost_saving): Likewise.
3209 * lra-assigns.c (find_hard_regno_for_1): Likewise.
3210 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
3211 (process_invariant_for_inheritance): Likewise.
3212 * lra-eliminations.c (move_plus_up): Likewise.
3213 * omp-low.c (lower_oacc_reductions): Likewise.
3214 * simplify-rtx.c (simplify_subreg): Likewise.
3215 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
3216 (TARGET_CHKP_BOUND_MODE): Likewise..
3217 * targhooks.c (default_chkp_bound_mode): Likewise.
3218 (default_setup_incoming_vararg_bounds): Likewise.
3219 * targhooks.h (default_chkp_bound_mode): Likewise.
3220 (default_setup_incoming_vararg_bounds): Likewise.
3221 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
3222 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
3223 (have_whole_vector_shift): Likewise.
3224 * tree-vect-stmts.c (vectorizable_load): Likewise.
3225 * doc/tm.texi: Regenerate.
3227 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
3229 Graceful degrade if Binutils PR21472 is not available.
3232 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
3233 .rodata in flash test fails.
3234 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
3235 * confgure: Regenerate.
3236 * config.in: Regenerate.
3237 * config/avr/avr.c (avr_asm_named_section)
3238 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
3239 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
3240 (avr_asm_init_sections): Same.
3242 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3244 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
3245 (fma<VH:mode>4_intrinsic): Likewise.
3246 (*fmsub<VCVTF:mode>4): Likewise.
3247 (*fmsub<VH:mode>4_intrinsic): Likewise.
3249 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
3252 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
3253 Don't depend on "optimize > 0".
3254 (out_movhi_r_mr, out_movqi_mr_r): Same.
3255 (out_movhi_mr_r, out_movqi_r_mr): Same.
3256 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
3257 io_address_operand on "optimize > 0".
3259 2017-07-05 Bin Cheng <bin.cheng@arm.com>
3261 * tree-loop-distribution.c: Add general explanantion on the pass.
3262 (generate_loops_for_partition): Mark distributed loop.
3263 (pg_add_dependence_edges): New parameter. Handle alias data
3264 dependence specially and record it in the parameter if asked.
3265 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
3266 (init_partition_graph_vertices, add_partition_graph_edge): New.
3267 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
3268 (free_partition_graph_vdata, build_partition_graph): New.
3269 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
3270 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
3271 (data_ref_segment_size, latch_dominated_by_data_ref): New.
3272 (compute_alias_check_pairs, version_loop_by_alias_check): New.
3273 (version_for_distribution_p, finalize_partitions): New.
3274 (distribute_loop): Handle alias data dependence specially. Factor
3275 out loop fusion code as functions and call these functions.
3277 2017-07-05 Bin Cheng <bin.cheng@arm.com>
3279 * tree-loop-distribution.c (classify_partition): New parameter and
3280 better handle reduction statement.
3281 (rdg_build_partitions): Revise comment.
3282 (distribute_loop): Compute statements in all partitions and pass it
3283 to classify_partition.
3285 2017-07-05 Bin Cheng <bin.cheng@arm.com>
3287 * tree-loop-distribution.c (enum partition_type): New.
3288 (struct partition): New field type.
3289 (partition_merge_into): Add parameter. Update partition type.
3290 (data_dep_in_cycle_p, update_type_for_merge): New functions.
3291 (build_rdg_partition_for_vertex): Compute partition type.
3292 (rdg_build_partitions): Dump partition type.
3293 (distribute_loop): Update calls to partition_merge_into.
3295 2017-07-05 Bin Cheng <bin.cheng@arm.com>
3297 * tree-loop-distribution.c (struct ddr_hasher): New.
3298 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
3300 (classify_partition): Call get_data_dependence.
3301 (pg_add_dependence_edges): Ditto.
3302 (distribute_loop): Release data dependence hash table.
3304 2017-07-05 Bin Cheng <bin.cheng@arm.com>
3306 * tree-loop-distribution.c (ref_base_address): Delete.
3307 (similar_memory_accesses): Rename ...
3308 (share_memory_accesses): ... to this. Check if partitions access
3309 the same memory reference.
3310 (distribute_loop): Call share_memory_accesses.
3312 2017-07-05 Bin Cheng <bin.cheng@arm.com>
3314 * tree-loop-distribution.c (struct partition): New field recording
3316 (partition_alloc, partition_free): Init and release data refs.
3317 (partition_merge_into): Merge data refs.
3318 (build_rdg_partition_for_vertex): Collect data refs for partition.
3319 (pg_add_dependence_edges): Change parameters from vector to bitmap.
3321 (distribute_loop): Remve data refs from vertice data of partition
3324 2017-07-05 Bin Cheng <bin.cheng@arm.com>
3326 * tree-loop-distribution.c (params.h): Include header file.
3327 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
3328 (datarefs_vec): New global var.
3329 (create_rdg_vertices): Use datarefs_vec directly.
3330 (free_rdg): Don't free data references.
3331 (build_rdg): Update use. Don't free data references.
3332 (distribute_loop): Compute global variable for data references.
3333 Bail out if there are too many data references.
3335 2017-07-05 Bin Cheng <bin.cheng@arm.com>
3337 * tree-loop-distribution.c (loop_nest): New global var.
3338 (build_rdg): Use loop directly, rather than loop nest.
3339 (pg_add_dependence_edges): Remove loop nest parameter. Use global
3341 (distribute_loop): Compute global variable loop nest. Update use.
3343 2017-07-05 Bin Cheng <bin.cheng@arm.com>
3345 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
3346 (partition_merge_into): New parameter. Dump reason for fusion.
3347 (distribute_loop): Update use of partition_merge_into.
3349 2017-07-05 Bin Cheng <bin.cheng@arm.com>
3351 * tree-loop-distribution.c (bb_top_order_index): New.
3352 (bb_top_order_index_size, bb_top_order_cmp): New.
3353 (stmts_from_loop): Use topological order.
3354 (pass_loop_distribution::execute): Compute and release topological
3355 order for basic blocks.
3357 2017-07-05 Bin Cheng <bin.cheng@arm.com>
3359 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
3362 2017-07-05 Bin Cheng <bin.cheng@arm.com>
3364 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
3365 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
3366 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
3367 * internal-fn.def (LOOP_DIST_ALIAS): New.
3368 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
3369 (fold_loop_internal_call): ... this.
3370 (vect_loop_dist_alias_call): New function.
3371 (set_uid_loop_bbs): Call fold_loop_internal_call.
3372 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
3375 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
3378 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
3379 Require dead FLAGS_REG at the beginning of a peephole.
3381 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
3384 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
3385 arguments in the call to __builtin_ia32_sbb_u32.
3386 (_subborrow_u64): Swap _X and _Y arguments in the call to
3387 __builtin_ia32_sbb_u64.
3389 2017-07-04 Jakub Jelinek <jakub@redhat.com>
3392 * tree-vrp.c (compare_assert_loc): Turn into a function template
3393 with stable template parameter. Only test if a->e is NULL,
3394 !a->e == !b->e has been verified already. Use e == NULL or
3395 e != NULL instead of e or ! e tests. If stable is true, don't use
3396 iterative_hash_expr, on the other side allow a or b or both NULL
3397 and sort the NULLs last.
3398 (process_assert_insertions): Sort using compare_assert_loc<false>
3399 instead of compare_assert_loc, later sort using
3400 compare_assert_loc<true> before calling process_assert_insertions_for
3401 in a loop. Use break instead of continue once seen NULL pointer.
3403 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
3405 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
3406 Cortex-R7 and Cortex-R8 processors.
3408 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
3410 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
3411 uninitialized while src is not.
3413 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
3415 * common/config/arm/arm-common.c: Adjust include path for
3417 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
3418 (arm-cpu.h): Create in build directory. Adjust dependency rules.
3419 (arm-cpu-data.h): Likewise.
3420 (arm-cpu-cdata.h): Likewise.
3421 * config/arm/arm-cpu.h: Delete.
3422 * config/arm/arm-cpu-cdata.h: Delete.
3423 * config/arm/arm-cpu-data.h: Delete.
3425 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
3427 * config/arm/arm-cpus.in (cortex-a55): New.
3428 (cortex-a75): Likewise.
3429 (cortex-a75.cortex-a55): Likewise.
3430 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
3432 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
3433 * config/arm/arm-cpu-cdata.h: Regenerate.
3434 * config/arm/arm-cpu-data.h: Regenerate.
3435 * config/arm/arm-cpu.h: Regenerate.
3436 * config/arm/arm-tables.opt: Regenerate.
3437 * config/arm/arm-tune.md: Regenerate.
3439 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
3441 * haifa-sched.c (sched_create_recovery_edges): Update profile.
3443 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
3445 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
3448 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
3450 PR tree-optimization/81292
3451 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
3452 full_string_p, also call adjust_related_strinfos if the adjustment
3453 is simple, otherwise invalidate related strinfos.
3455 2017-07-04 Martin Liska <mliska@suse.cz>
3458 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
3459 newly created variable as DECL_IGNORED_P.
3461 2017-07-04 Martin Liska <mliska@suse.cz>
3464 * ipa-inline.c (inline_small_functions):
3465 Use xstrdup_for_dump.
3467 2017-07-04 Tom de Vries <tom@codesourcery.com>
3469 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
3471 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
3474 * config/darwin.c (darwin_function_switched_text_sections):
3477 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
3479 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
3481 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
3483 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
3485 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3487 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
3488 min_profitable_iters, and th as inclusive lower bounds.
3489 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
3490 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
3491 for min_profitable_iters and min_profitable_estimate.
3492 (vect_transform_loop): Treat th as an inclusive lower bound.
3493 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
3495 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
3498 * config/darwin.c (darwin_function_switched_text_sections):
3499 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
3500 in two pieces, and suppress the use of buf.
3502 2017-07-03 Nathan Sidwell <nathan@acm.org>
3504 * hash-table.h (hash_table_mod1): Fix indentation.
3506 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3509 * predict.c (force_edge_cold): Be more careful about propagation
3511 * profile-count.h (profile_probability::guessed,
3512 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
3514 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
3516 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
3518 * doc/invoke.texi (rcpc architecture extension): Document it.
3520 2017-07-03 Richard Biener <rguenther@suse.de>
3522 PR tree-optimization/60510
3523 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
3524 the scalar reduction PHI and use it.
3525 (vectorizable_reduction): Properly guard the single_defuse_cycle
3526 path for non-SLP reduction chains where we cannot use it.
3527 Rework reduc_def/index and vector type deduction. Rework
3528 vector operand gathering during reduction op code-gen.
3529 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
3530 chains dissolve the chain and leave it to non-SLP reduction
3533 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3535 * tree-data-ref.h (dr_alignment): Declare.
3536 * tree-data-ref.c (dr_alignment): New function.
3537 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
3538 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
3540 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
3542 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3544 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
3545 and base_misalignment fields.
3546 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
3547 * tree-data-ref.c: Include builtins.h.
3548 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
3549 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
3550 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
3551 * tree-vect-data-refs.c: Include tree-cfg.h.
3552 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
3553 fields instead of calculating an alignment here.
3554 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
3555 innermost_loop_behavior fields.
3557 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3559 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
3561 (DR_STEP_ALIGNMENT): New macro.
3562 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
3563 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
3564 (create_data_ref): Print it.
3565 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
3566 to tell whether the step preserves vector (mis)alignment.
3567 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
3568 Move the check for an integer step and generalise to all INTEGER_CST.
3569 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
3570 Print the outer step alignment.
3572 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3574 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
3575 with offset_alignment.
3576 (DR_ALIGNED_TO): Delete.
3577 (DR_OFFSET_ALIGNMENT): New macro.
3578 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
3579 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
3580 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
3581 (create_data_ref): Likewise.
3582 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
3583 (vect_analyze_data_refs): Likewise.
3584 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
3585 creating dummy innermost behavior.
3587 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3589 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
3590 with a "innermost_loop_behavior *" and refeence tree.
3591 * tree-data-ref.c (dr_analyze_innermost): Likewise.
3592 (create_data_ref): Update call accordingly.
3593 * tree-predcom.c (find_looparound_phi): Likewise.
3595 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3597 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
3598 fields with dr_wrt_vec_loop.
3599 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
3600 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
3601 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
3602 (vect_dr_behavior): New function.
3603 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
3604 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
3605 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
3606 track whether the step preserves the misalignment.
3607 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
3608 Use vect_dr_behavior.
3609 (vect_setup_realignment): Update call accordingly.
3610 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
3611 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
3612 call to vect_create_addr_base_for_vector_ref.
3613 (vect_create_cond_for_align_checks): Likewise.
3614 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
3615 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
3616 (vect_recog_mask_conversion_pattern): Likewise.
3617 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
3618 (new_stmt_vec_info): Remove redundant zeroing.
3620 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
3622 * common/config/arm/arm-common.c (arm_be8_option): New function.
3623 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
3624 (ISA_ARMv6): Add isa_bit_be8.
3625 * config/arm/arm.h (arm_be8_option): Add prototype.
3626 (BE8_SPEC_FUNCTION): New define.
3627 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
3628 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
3629 (mlittle-endian): Similarly.
3630 (mbe8, mbe32): New options.
3631 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
3632 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
3634 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3636 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
3638 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3640 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
3641 (cleanup_tree_cfg_bb): Use it.
3642 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
3644 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
3646 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3649 * loop-doloop.c (add_test): Update profile.
3651 2017-07-03 Martin Liska <mliska@suse.cz>
3654 * sanopt.c (rewrite_usage_of_param): New function.
3655 (sanitize_rewrite_addressable_params): Likewise.
3656 (pass_sanopt::execute): Call rewrite_usage_of_param.
3658 2017-07-03 Richard Biener <rguenther@suse.de>
3660 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
3661 back to using VIEW_CONVERT_EXPR.
3663 2017-07-03 Martin Liska <mliska@suse.cz>
3666 * doc/extend.texi: Document when a resolver function is
3667 generated for target_clones.
3669 2017-07-03 Martin Liska <mliska@suse.cz>
3671 * asan.c (asan_emit_stack_protection): Unpoison just red zones
3672 and shadow memory of auto variables which are subject of
3673 use-after-scope sanitization.
3674 (asan_expand_mark_ifn): Add do set only when is_poison.
3676 2016-07-03 Richard Biener <rguenther@suse.de>
3678 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
3680 (vect_force_simple_reduction): Record reduction def -> phi mapping.
3681 (vectorizable_reduction): Perform reduction PHI creation when
3682 visiting a reduction PHI and adjust and simplify code generation
3683 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
3684 (vect_transform_loop): Visit reduction PHIs.
3685 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
3686 defs into the SLP tree.
3687 (vect_build_slp_tree): Reduction defs terminate the recursion.
3688 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
3690 (vect_get_vec_defs_for_stmt_copy): Export.
3691 (vect_get_vec_defs): Likewise.
3692 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
3694 (vect_get_vec_defs_for_stmt_copy): Declare.
3695 (vect_get_vec_defs): Likewise.
3697 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
3699 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
3700 parameter with a "loop" parameter and use it instead of the
3701 loop containing DR_STMT. Don't check simple_iv when doing
3702 BB analysis. Describe the two analysis modes in the comment.
3704 2017-07-03 Tom de Vries <tom@codesourcery.com>
3706 PR tree-optimization/69468
3707 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
3708 (find_same_succ_bb): Handle ignore_edge_flags.
3710 2017-07-03 Tom de Vries <tom@codesourcery.com>
3712 PR tree-optimization/81192
3713 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
3715 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
3717 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
3719 2017-07-03 Tom de Vries <tom@codesourcery.com>
3721 PR tree-optimization/81192
3722 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
3723 BB_SAME_SUCC (bb) == NULL.
3725 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3727 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
3730 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3732 * dumpfile.c: Include profile-count.h
3733 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
3735 (insert_cond_bb): Update profile.
3736 * tree-cfg.h (insert_cond_bb): Update prototype.
3737 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
3738 * tree-dump.c: Do not include tree-cfg.
3740 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3742 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
3744 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3746 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
3749 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
3751 * tree-complex.c (expand_complex_div_wide): update profile.
3753 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
3754 Alan Hayward <alan.hayward@arm.com>
3755 David Sherwood <david.sherwood@arm.com>
3757 * Makefile.in (MACHMODE_H): Remove insn-modes.h
3758 (CORETYPES_H): New define.
3759 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
3760 (insn-modes-inline.h, s-modes-inline-h): New rules.
3761 (generated_files): Add insn-modes-inline.h.
3762 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
3763 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
3764 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
3765 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
3766 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
3767 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
3768 (build/gencodes.o, build/genconditions.o): Likewise.
3769 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
3770 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
3771 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
3772 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
3773 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
3774 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
3775 * coretypes.h: Include everything up to real.h for generators.
3776 Include insn-modes.h first. Include wide-int-print.h after
3777 wide-int.h. Include insn-modes-inline.h and then machmode.h.
3778 * machmode.h: Don't include insn-modes.h here.
3779 * function-tests.c: Remove includes of signop.h, machmode.h,
3780 double-int.h and wide-int.h.
3782 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
3784 * optc-save-gen.awk: Likewise.
3785 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
3786 * godump.c: Remove include of wide-int-print.h.
3787 * pretty-print.h: Likewise.
3788 * wide-int-print.cc: Likewise.
3789 * wide-int.cc: Likewise.
3790 * hash-map-tests.c: Remove include of signop.h.
3791 * hash-set-tests.c: Likewise.
3792 * rtl-tests.c: Likewise.
3793 * mkconfig.sh: Remove include of machmode.h.
3794 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
3796 (emit_insn_modes_inline_h): ...this new function. Emit the code
3797 into an insn-modes-inline.h header file, adding appropriate
3798 include guards and end comments.
3799 (emit_insn_modes_c_header): Remove include of machmode.h.
3800 (emit_min_insn_modes_c_header): Include coretypes.h rather than
3802 (main): Handle -i flag and call emit_insn_modes_inline_h when
3805 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
3807 * tree-ssa-strlen.c (strinfo): Rename the length field to
3808 nonzero_chars. Add a full_string_p field.
3809 (compare_nonzero_chars, zero_length_string_p): New functions.
3810 (get_addr_stridx): Add an offset_out parameter.
3811 Use compare_nonzero_chars.
3812 (get_stridx): Update accordingly. Use compare_nonzero_chars.
3813 (new_strinfo): Update after above changes to strinfo.
3814 (set_endptr_and_length): Set full_string_p.
3815 (get_string_length): Update after above changes to strinfo.
3816 (unshare_strinfo): Update call to new_strinfo.
3817 (maybe_invalidate): Likewise.
3818 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
3819 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
3820 as a uhwi instead of an shwi. Update after above changes to
3821 strinfo and new_strinfo.
3822 (zero_length_string): Assert that chainsi contains full strings.
3823 Use zero_length_string_p. Update call to new_strinfo.
3824 (adjust_related_strinfos): Update after above changes to strinfo.
3825 Copy full_string_p from origsi.
3826 (adjust_last_stmt): Use zero_length_string_p.
3827 (handle_builtin_strlen): Update after above changes to strinfo and
3828 new_strinfo. Install the lhs as the string length if the previous
3829 entry didn't describe a full string.
3830 (handle_builtin_strchr): Update after above changes to strinfo
3832 (handle_builtin_strcpy): Likewise.
3833 (handle_builtin_strcat): Likewise.
3834 (handle_builtin_malloc): Likewise.
3835 (handle_pointer_plus): Likewise.
3836 (handle_builtin_memcpy): Likewise. Track nonzero characters
3837 that aren't necessarily followed by a nul terminator.
3838 (handle_char_store): Likewise.
3840 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
3842 PR tree-optimization/80769
3843 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
3844 for malloc and calloc. Document the new invariant that all related
3845 strinfos have delayed lengths or none do.
3846 (verify_related_strinfos): Move earlier in file.
3847 (set_endptr_and_length): New function, split out from...
3848 (get_string_length): ...here. Also set the lengths of related
3850 (zero_length_string): Assert that chainsi has known (rather than
3852 (adjust_related_strinfos): Likewise.
3854 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
3856 PR tree-optimization/81136
3857 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
3858 assert that two references with the same misalignment have the same
3859 compile-time misalignment if those compile-time misalignments
3862 2017-07-01 Andi Kleen <ak@linux.intel.com>
3864 * print-tree.c (print_node): Print all attributes.
3866 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3868 * cfg.c (scale_bbs_frequencies): New function.
3869 * cfg.h (scale_bbs_frequencies): Declare it.
3870 * cfgloopanal.c (single_likely_exit): Cleanup.
3871 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
3873 (scale_loop_profile): Likewise.
3874 (loop_version): Likewise.
3875 (create_empty_loop_on_edge): Update.
3876 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
3877 scale_loop_frequencies, scale_loop_profile, loopify,
3878 loop_version): Update prototypes.
3879 * modulo-sched.c (sms_schedule): Update.
3880 * predict.c (unlikely_executed_edge_p): Also check probability.
3881 (probably_never_executed_edge_p): Fix typo.
3882 * tree-if-conv.c (version_loop_for_if_conversion): Update.
3883 * tree-parloops.c (gen_parallel_loop): Update.
3884 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
3885 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
3886 * tree-ssa-loop-split.c (split_loop): Update.
3887 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
3888 * tree-vect-loop-manip.c (vect_do_peeling): Update.
3889 (vect_loop_versioning): Update.
3890 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
3892 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3894 * trans-mem.c (split_bb_make_tm_edge): Update profile.
3896 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3898 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
3899 to keep profile consistent.
3901 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3903 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
3904 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
3905 * profile-count.h (max_safe_multiplier): Make unsigned.
3906 (profile_count::guessed_zero): New.
3908 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3910 * bb-reorder.c (fix_up_crossing_landing_pad,
3911 fix_crossing_conditional_branches): Use make_single_succ_edge
3912 to keep profile consistent.
3914 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
3916 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
3919 2017-07-01 Jakub Jelinek <jakub@redhat.com>
3922 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
3923 the right scopes, make sure cond_jump isn't preserved between multiple
3924 iterations. Search for fallthru edge whenever there are 3+ edges and
3925 use find_fallthru_edge for it.
3927 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3929 Patch by Alexander Monakov <amonakov@ispras.ru>
3930 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
3931 probabilities consistently.
3933 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3935 * pa.c (pa_expand_compare_and_swap_loop): Update call of
3936 emit_cmp_and_jump_insns.
3938 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
3941 * tree-inline.c (expand_call_inline): Combine profile statuses.
3943 2017-06-30 Andrew Pinski <apinski@cavium.com>
3945 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
3946 fold_stmt returned true.
3948 2017-06-30 Nathan Sidwell <nathan@acm.org>
3950 * ggc.h (empty_string): Delete.
3951 * cfgexpand.c (expand_asm_stmt): Use plain "".
3952 * optabs.c (expand_asm_memory_barrier): Likewise.
3953 * stringpool.c (empty_string): Delete.
3954 (digit_vector, digit_string): Delete.
3955 (ggc_alloc_string): Use plain "", don't optimize single digit
3956 strings. Use ggc_alloc_atomic.
3958 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
3960 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
3961 comparison set and one other set, use the cost of the non-comparison
3964 2017-06-30 Nathan Sidwell <nathan@acm.org>
3966 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
3969 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
3971 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
3972 loops. Remove now unneeded calls to gimple_switch_set_label() that
3973 just set removed labels to NULL_TREE.
3975 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
3977 * tree-ssanames.c (set_range_info_raw): Abstract from ...
3978 (set_range_info): ...here. Only call set_range_info_raw if domain
3980 (set_nonzero_bits): Call set_range_info_raw.
3981 * tree-ssanames.h (set_range_info_raw): New.
3983 2017-06-30 Jakub Jelinek <jakub@redhat.com>
3986 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
3987 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
3988 of nonimmediate_operand and <store_mask_constraint> instead of m
3989 for the input operand. For V8FI iterator, always split if input
3990 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
3991 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
3992 <store_mask_predicate> instead of register_operand and
3993 <store_mask_constraint> instead of v for the input operand. Make
3994 sure both operands aren't MEMs for if not <mask_applied>.
3996 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
3998 * lto-wrapper.c (copy_file) Close both file descriptors before
4001 2017-06-30 Martin Liska <mliska@suse.cz>
4004 * multiple_target.c (create_dispatcher_calls): Make ifunc
4005 also for function that don't have calls or are not referenced.
4007 2017-06-30 Richard Biener <rguenther@suse.de>
4009 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
4010 analyze the first scalar stmt. Move vector type computation
4011 for the BB case here from ...
4012 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
4013 live operation processing in the SLP case properly.
4015 2017-06-30 Richard Biener <rguenther@suse.de>
4017 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
4019 2017-06-30 Martin Liska <mliska@suse.cz>
4022 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
4023 before BUILT_IN_UNWIND_RESUME when ASAN is used.
4025 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
4027 * doc/invoke.texi (AArch64): Add missing options and remove redundant
4030 2017-06-30 Richard Biener <rguenther@suse.de>
4032 PR tree-optimization/81249
4033 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
4034 condition reduction result to original scalar type.
4036 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4038 * profile-count.h (enum profile_quality): Fix typos and whitespace
4041 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4043 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
4044 type for branch probabilities.
4046 2017-06-29 Julian Brown <julian@codesourcery.com>
4047 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
4049 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
4050 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
4051 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
4052 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
4054 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
4056 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
4057 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
4058 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
4059 CC usage from generic code to here.
4060 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
4061 CC usage into the target macros.
4063 2017-06-29 Maya Rashish <coypu@sdf.org>
4065 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
4068 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
4070 * arm/arm-builtins.c: Include profile-count.h
4071 * except.c (sjlj_emit_function_enter): Use
4072 profile_probability::unlikely.
4074 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
4076 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
4077 and tocrel_offset be pointer args rather than implicitly using
4079 (legitimate_constant_pool_address_p, rs6000_emit_move,
4080 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
4081 tocrel_offset and use in toc_relative_expr_p call.
4082 (print_operand, print_operand_address): Use static tocrel_base_oac
4083 and tocrel_offset_oac.
4084 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
4087 2017-06-29 Maya Rashish <coypu@sdf.org>
4089 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
4091 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
4093 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
4094 objects, take into account only the alignment of 'op0' and 'mode1' if
4097 2017-06-29 Steve Ellcey <sellcey@cavium.com>
4099 * ccmp.c (ccmp_tree_comparison_p): New function.
4100 (ccmp_candidate_p): Update to use above function.
4101 (get_compare_parts): New function.
4102 (expand_ccmp_next): Update to use new functions.
4103 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
4105 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
4106 take mode as argument.
4107 * ccmp.h (expand_ccmp_expr): Add mode as argument.
4108 * expr.c (expand_expr_real_1): Pass mode as argument.
4110 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
4112 * combine.c (combine_instructions): Print insns to dump_file, together
4115 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
4117 * asan.c (asan_emit_stack_protection): Update.
4118 (create_cond_insert_point): Update.
4119 * auto-profile.c (afdo_propagate_circuit): Update.
4120 * basic-block.h (struct edge_def): Turn probability to
4121 profile_probability.
4122 (EDGE_FREQUENCY): Update.
4123 * bb-reorder.c (find_traces_1_round): Update.
4124 (better_edge_p): Update.
4125 (sanitize_hot_paths): Update.
4126 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
4127 (make_single_succ_edge): Update.
4128 (check_bb_profile): Update.
4129 (dump_edge_info): Update.
4130 (update_bb_profile_for_threading): Update.
4131 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
4132 probabilitycount to 0.
4133 * cfgbuild.c (compute_outgoing_frequencies): Update.
4134 * cfgcleanup.c (try_forward_edges): Update.
4135 (outgoing_edges_match): Update.
4136 (try_crossjump_to_edge): Update.
4137 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
4138 (expand_gimple_tailcall): Update.
4139 (construct_init_block): Use make_single_succ_edge.
4140 (construct_exit_block): Use make_single_succ_edge.
4141 * cfghooks.c (verify_flow_info): Update.
4142 (redirect_edge_succ_nodup): Update.
4143 (split_edge): Update.
4144 (account_profile_record): Update.
4145 * cfgloopanal.c (single_likely_exit): Update.
4146 * cfgloopmanip.c (scale_loop_profile): Update.
4147 (set_zero_probability): Remove.
4148 (duplicate_loop_to_header_edge): Update.
4149 * cfgloopmanip.h (loop_version): Update prototype.
4150 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
4151 (force_nonfallthru_and_redirect): Update.
4152 (update_br_prob_note): Update.
4153 (rtl_verify_edges): Update.
4154 (purge_dead_edges): Update.
4155 (rtl_lv_add_condition_to_bb): Update.
4156 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
4157 * cgraphunit.c (init_lowered_empty_function): Update.
4158 (cgraph_node::expand_thunk): Update.
4159 * cilk-common.c: Include profile-count.h
4160 * dojump.c (inv): Remove.
4161 (jumpifnot): Update.
4162 (jumpifnot_1): Update.
4163 (do_jump_1): Update.
4165 (do_jump_by_parts_greater_rtx): Update.
4166 (do_compare_rtx_and_jump): Update.
4167 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
4168 do_jump_1. do_compare_rtx_and_jump): Update prototype.
4169 * dwarf2cfi.c: Include profile-count.h
4170 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
4171 (sjlj_emit_dispatch_table): Likewise.
4172 * explow.c: Include profile-count.h
4173 * expmed.c (emit_store_flag_force): Update.
4174 (do_cmp_and_jump): Update.
4175 * expr.c (compare_by_pieces_d::generate): Update.
4176 (compare_by_pieces_d::finish_mode): Update.
4177 (emit_block_move_via_loop): Update.
4178 (store_expr_with_bounds): Update.
4179 (store_constructor): Update.
4180 (expand_expr_real_2): Update.
4181 (expand_expr_real_1): Update.
4182 * expr.h (try_casesi, try_tablejump): Update prototypes.
4183 * gimple-pretty-print.c (dump_probability): Update.
4184 (dump_profile): New.
4185 (dump_gimple_label): Update.
4186 (dump_gimple_bb_header): Update.
4187 * graph.c (draw_cfg_node_succ_edges): Update.
4188 * hsa-gen.c (convert_switch_statements): Update.
4189 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
4190 (find_if_case_1): Update.
4191 (find_if_case_2): Update.
4192 * internal-fn.c (expand_arith_overflow_result_store): Update.
4193 (expand_addsub_overflow): Update.
4194 (expand_neg_overflow): Update.
4195 (expand_mul_overflow): Update.
4196 (expand_vector_ubsan_overflow): Update.
4197 * ipa-cp.c (good_cloning_opportunity_p): Update.
4198 * ipa-split.c (split_function): Use make_single_succ_edge.
4199 * ipa-utils.c (ipa_merge_profiles): Update.
4200 * loop-doloop.c (add_test): Update.
4201 (doloop_modify): Update.
4202 * loop-unroll.c (compare_and_jump_seq): Update.
4203 (unroll_loop_runtime_iterations): Update.
4204 * lra-constraints.c (lra_inheritance): Update.
4205 * lto-streamer-in.c (input_cfg): Update.
4206 * lto-streamer-out.c (output_cfg): Update.
4207 * mcf.c (adjust_cfg_counts): Update.
4208 * modulo-sched.c (sms_schedule): Update.
4209 * omp-expand.c (expand_omp_for_init_counts): Update.
4210 (extract_omp_for_update_vars): Update.
4211 (expand_omp_ordered_sink): Update.
4212 (expand_omp_for_ordered_loops): Update.
4213 (expand_omp_for_generic): Update.
4214 (expand_omp_for_static_nochunk): Update.
4215 (expand_omp_for_static_chunk): Update.
4216 (expand_cilk_for): Update.
4217 (expand_omp_simd): Update.
4218 (expand_omp_taskloop_for_outer): Update.
4219 (expand_omp_taskloop_for_inner): Update.
4220 * omp-simd-clone.c (simd_clone_adjust): Update.
4221 * optabs.c (expand_doubleword_shift): Update.
4222 (expand_abs): Update.
4223 (emit_cmp_and_jump_insn_1): Update.
4224 (expand_compare_and_swap_loop): Update.
4225 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
4226 * predict.c (predictable_edge_p): Update.
4227 (edge_probability_reliable_p): Update.
4228 (set_even_probabilities): Update.
4229 (combine_predictions_for_insn): Update.
4230 (combine_predictions_for_bb): Update.
4231 (propagate_freq): Update.
4232 (estimate_bb_frequencies): Update.
4233 (force_edge_cold): Update.
4234 * profile-count.c (profile_count::dump): Add missing space into dump.
4235 (profile_count::debug): Add newline.
4236 (profile_count::differs_from_p): Explicitly convert to unsigned.
4237 (profile_count::stream_in): Update.
4238 (profile_probability::dump): New member function.
4239 (profile_probability::debug): New member function.
4240 (profile_probability::differs_from_p): New member function.
4241 (profile_probability::differs_lot_from_p): New member function.
4242 (profile_probability::stream_in): New member function.
4243 (profile_probability::stream_out): New member function.
4244 * profile-count.h (profile_count_quality): Rename to ...
4245 (profile_quality): ... this one.
4246 (profile_probability): New.
4247 (profile_count): Update.
4248 * profile.c (compute_branch_probabilities): Update.
4249 * recog.c (peep2_attempt): Update.
4250 * sched-ebb.c (schedule_ebbs): Update.
4251 * sched-rgn.c (find_single_block_region): Update.
4252 (compute_dom_prob_ps): Update.
4253 (schedule_region): Update.
4254 * sel-sched-ir.c (compute_succs_info): Update.
4255 * stmt.c (struct case_node): Update.
4256 (do_jump_if_equal): Update.
4257 (get_outgoing_edge_probs): Update.
4258 (conditional_probability): Update.
4259 (emit_case_dispatch_table): Update.
4260 (expand_case): Update.
4261 (expand_sjlj_dispatch_table): Update.
4262 (emit_case_nodes): Update.
4263 * targhooks.c: Update.
4264 * tracer.c (better_p): Update.
4265 (find_best_successor): Update.
4266 * trans-mem.c (expand_transaction): Update.
4267 * tree-call-cdce.c: Update.
4268 * tree-cfg.c (gimple_split_edge): Upate.
4269 (move_sese_region_to_fn): Upate.
4270 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
4271 * tree-eh.c (lower_resx): Upate.
4272 (cleanup_empty_eh_move_lp): Upate.
4273 * tree-if-conv.c (version_loop_for_if_conversion): Update.
4274 * tree-inline.c (copy_edges_for_bb): Update.
4275 (copy_cfg_body): Update.
4276 * tree-parloops.c (gen_parallel_loop): Update.
4277 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
4278 (gimple_gen_time_profiler): Update.
4279 * tree-ssa-dce.c (remove_dead_stmt): Update.
4280 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
4281 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
4282 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
4283 (unloop_loops): Update.
4284 (try_peel_loop): Update.
4285 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
4286 * tree-ssa-loop-split.c (connect_loops): Update.
4287 (split_loop): Update.
4288 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
4289 (hoist_guard): Update.
4290 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
4291 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
4292 (value_replacement): Update.
4293 * tree-ssa-reassoc.c (branch_fixup): Update.
4294 * tree-ssa-tail-merge.c (replace_block_by): Update.
4295 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
4296 (create_edge_and_update_destination_phis): Update.
4297 (compute_path_counts): Update.
4298 (recompute_probabilities): Update.
4299 (update_joiner_offpath_counts): Update.
4300 (freqs_to_counts_path): Update.
4301 (duplicate_thread_path): Update.
4302 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
4303 (struct switch_conv_info): Update.
4304 (gen_inbound_check): Update.
4305 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
4306 (vect_do_peeling): Update.
4307 (vect_loop_versioning): Update.
4308 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
4309 (optimize_mask_stores): Update.
4310 * ubsan.c (ubsan_expand_null_ifn): Update.
4311 * value-prof.c (gimple_divmod_fixed_value): Update.
4312 (gimple_divmod_fixed_value_transform): Update.
4313 (gimple_mod_pow2): Update.
4314 (gimple_mod_pow2_value_transform): Update.
4315 (gimple_mod_subtract): Update.
4316 (gimple_mod_subtract_transform): Update.
4317 (gimple_ic): Update.
4318 (gimple_stringop_fixed_value): Update.
4319 (gimple_stringops_transform): Update.
4320 * value-prof.h: Update.
4322 2017-06-29 Carl Love <cel@us.ibm.com>
4324 * config/rs6000/rs6000-c.c: Add support for built-in functions
4325 vector signed int vec_signed (vector float);
4326 vector signed long long vec_signed (vector double);
4327 vector signed int vec_signed2 (vector double, vector double);
4328 vector signed int vec_signede (vector double);
4329 vector signed int vec_signedo (vector double);
4330 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
4331 instruction generator.
4332 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
4333 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
4334 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
4336 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
4337 vunsignede_v2df): Add define_expands.
4338 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
4339 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
4340 VEC_UNSIGNEDO): Add definitions.
4341 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
4342 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
4343 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
4344 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
4345 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
4346 * config/rs6000/altivec.h (vec_signed, vec_signed2,
4347 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
4348 vec_unsignede, vec_unsignedo): Add builtin defines.
4349 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
4351 * doc/extend.texi: Update the built-in documentation file for the
4352 new built-in functions.
4354 2017-06-29 Richard Biener <rguenther@suse.de>
4356 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
4357 reduction chains to LOOP_VINFO_REDUCTIONS.
4358 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
4359 SLP reductions after processing reduction chains.
4361 2017-06-29 Nathan Sidwell <nathan@acm.org>
4363 * builtins.c (fold_builtin_FUNCTION): Use
4364 lang_hooks.decl_printable_name.
4366 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
4369 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
4370 with only one label.
4371 * stmt.c (expand_case): Assert NCASES is greater than one.
4373 2017-06-29 Richard Biener <rguenther@suse.de>
4375 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
4377 (group_case_labels): Likewise.
4378 (find_taken_edge): Push sanity checking on val to workers...
4379 (find_taken_edge_cond_expr): ... here
4380 (find_taken_edge_switch_expr): ... and here, handle cases
4381 with just a default label.
4382 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
4383 (group_case_labels): Likewise.
4384 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
4385 group_case_labels does anything cleanup the CFG again.
4387 2017-06-29 Bin Cheng <bin.cheng@arm.com>
4389 PR tree-optimization/81196
4390 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
4391 exit condition comparing two IVs.
4393 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
4395 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
4396 profile to the dummy entry at the end of the list of architectures.
4397 * config/arm/arm-cpu-cdata.h: Regenerated.
4399 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4400 Michael Collison <michael.collison@arm.com>
4403 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
4405 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
4406 (*aarch64_reg_<mode>3_minus_mask): New pattern.
4407 (*aarch64_<optab>_reg_di3_mask2): New pattern.
4408 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
4409 of shift when the shift amount is masked with constant equal to
4410 the size of the mode.
4411 * config/aarch64/predicates.md (subreg_lowpart_operator): New
4414 2017-06-29 Martin Liska <mliska@suse.cz>
4416 * config/i386/i386.opt: Change range from [1,5] to [0,5].
4418 2017-06-29 Yury Gribov <tetra2005@gmail.com>
4421 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
4424 (edge_growth_cache_entry::edge_growth_cache_entry): New
4426 (reset_edge_growth_cache): Update to use constructor.
4428 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4430 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
4431 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
4432 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
4434 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
4436 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
4437 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
4439 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
4441 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
4442 (*-linux-uclibc*): Add t-uclibc tmake_file.
4443 * config/t-musl: New.
4444 * config/t-uclibc: New.
4446 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
4448 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
4450 (gen_comm_data): Emit architectural setting of arch_prof.
4451 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
4453 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
4454 (armv8-m.base, armv8-m.main): Likewise.
4455 * arm-protos.h (arm_build_target): Add profile field.
4456 (arch_option): Likewise.
4457 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
4459 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
4460 arm_active_target.profile.
4462 2017-06-28 Richard Biener <rguenther@suse.de>
4465 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
4466 TYPE_OVERFLOW_WRAPS.
4467 * match.pd (negate_expr_p): Likewise.
4468 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
4469 fold_build2, not fold_binary.
4471 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
4473 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
4474 Convert memory address to Pmode.
4475 (aarch64_print_operand): Assert MEM operands are always Pmode.
4477 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
4480 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
4481 Remove redundant if.
4482 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
4483 * config/arm/aarch-common-protos.h
4484 (aarch_forward_to_shift_is_not_shifted_re): Remove.
4485 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
4487 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
4490 * multiple_target.c (create_dispatcher_calls): Set the default
4491 clone to be static, not public.
4493 2017-06-28 Richard Biener <rguenther@suse.de>
4495 * tree-vect-loop.c (vectorizable_reduction): Move special
4496 cond reduction IV var creation ...
4497 (vect_create_epilog_for_reduction): ... here. Remove induction_index
4498 parameter. Use STMT_VINFO_VECTYPE.
4499 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
4502 2017-06-28 Martin Liska <mliska@suse.cz>
4505 * ipa-visibility.c (non_local_p): Handle visibility.
4507 2017-06-28 Martin Liska <mliska@suse.cz>
4510 * common.opt: Add IntegerRange to various options.
4511 * opt-functions.awk (integer_range_info): New function.
4512 * optc-gen.awk: Add integer_range_info to cl_options struct.
4513 * opts-common.c (decode_cmdline_option): Handle
4514 CL_ERR_INT_RANGE_ARG.
4515 (cmdline_handle_error): Likewise.
4516 * opts.c (print_filtered_help): Show valid interval in
4517 when --help is provided.
4518 * opts.h (struct cl_option): Add range_min and range_max fields.
4519 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
4521 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
4523 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
4524 (x * C EQ/NE y * C): New transformation.
4526 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
4528 * genmultilib (combination_space): Accept '+' in option names.
4530 2017-06-28 Martin Liska <mliska@suse.cz>
4533 * asan.c (instrument_derefs): Bail out inner references
4534 that are hard register variables.
4536 2017-06-28 Jakub Jelinek <jakub@redhat.com>
4539 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
4540 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
4542 2017-06-28 Richard Biener <rguenther@suse.de>
4544 * tree-vectorizer.h (vect_get_vec_defs): Remove.
4545 (vect_get_slp_defs): Adjust.
4546 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
4548 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
4550 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
4551 get_initial_defs_for_reduction instead of vect_get_vec_defs.
4552 (vectorizable_reduction): Adjust.
4553 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
4555 (vect_get_slp_defs): Likewise.
4556 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
4557 (vectorizable_bswap): Adjust.
4558 (vectorizable_call): Likewise.
4559 (vectorizable_conversion): Likewise.
4560 (vectorizable_assignment): Likewise.
4561 (vectorizable_shift): Likewise.
4562 (vectorizable_operation): Likewise.
4563 (vectorizable_store): Likewise.
4564 (vectorizable_condition): Likewise.
4565 (vectorizable_comparison): Likewise.
4567 2017-06-28 Michael Collison <michael.collison@arm.com>
4570 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
4572 (arm_gen_movmemqi): Removed unused variable 'i'.
4573 Convert 'for' loop into 'while' loop.
4574 (arm_expand_prologue): Remove last unnecessary set of insn.
4575 (thumb_pop): Remove unused variable 'pushed_words'.
4576 (thumb_exit): Remove last unnecessary set of regs_to_pop.
4578 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4580 * config/s390/predicates.md: Use s390_rel_address_ok_p.
4581 * config/s390/s390-protos.h: Add prototype of
4582 s390_rel_address_ok_p.
4583 * config/s390/s390.c (s390_got_symbol): New function.
4584 (s390_rel_address_ok_p): New function.
4585 (legitimize_pic_address): Use s390_rel_address_ok_p.
4586 (s390_load_got): Use s390_got_symbol.
4587 (s390_option_override): Issue error if
4588 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
4589 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
4591 * config/s390/s390.opt: New option mpic-data-is-text-relative.
4593 2017-06-27 Andrew Pinski <apinski@cavium.com>
4595 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
4596 (X * copysign (1.0, X)): New pattern.
4597 (X * copysign (1.0, -X)): New pattern.
4598 (copysign (-1.0, CST)): New pattern.
4600 2017-06-27 Joseph Myers <joseph@codesourcery.com>
4602 * genmultilib (combination_space): Remove variable.
4603 Validate reuse rules against regular expression for any sequence
4604 of multilib options in any order.
4606 2017-06-27 Michael Collison <michael.collison@arm.com>
4608 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
4609 call aarch64_split_simd_combine.
4610 * (aarch64_combine_internal<mode>): Delete pattern.
4611 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
4612 Allow register and subreg operands.
4614 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
4616 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
4617 specific need, just fallback on defaults.
4618 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
4620 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
4621 Olivier Hainque <hainque@adacore.com>
4623 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
4625 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
4626 targets. Pick a default if no particular attempt applied.
4627 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
4630 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
4632 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
4633 (x86_64-wrs-vxworks7): Likewise.
4635 2017-06-27 Marek Polacek <polacek@redhat.com>
4638 * ubsan.c (instrument_null): Check get_base_address's result for null.
4640 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
4642 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
4644 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
4646 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
4647 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
4648 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
4649 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
4651 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
4652 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
4653 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
4654 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
4655 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
4656 BUILT_IN_FEUPDATEENV): New builtins.
4657 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
4658 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
4659 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
4660 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
4662 (builtin_structptr_types): Adjust size.
4663 * tree.c (builtin_structptr_types): Add four entries.
4665 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
4666 Olivier Hainque <hainque@adacore.com>
4668 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
4669 (TLS_SYM): New local macro, forcing reference to __tls__ on
4670 link command lines for VxWorks 7 RTPs, triggering initialization
4672 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
4673 OS features TLS support, true for RTPs on VxWorks 7.
4674 * config/vxworks.c (vxworks_override_options): Setup emutls
4677 2017-06-27 Jakub Jelinek <jakub@redhat.com>
4679 * predict.c (test_prediction_value_range): Use -1U instead of -1
4680 to avoid narrowing conversion warning.
4681 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
4682 to avoid narrowing conversion warning.
4683 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
4685 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
4687 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
4689 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
4690 64bit configurations.
4691 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
4692 (SIZE_TYPE): Likewise.
4693 * config/vxworks.c (vxworks_emutls_var_fields): Use
4694 long_unsigned_type_node instead of unsigned_type_node as the offset
4695 field type, which is "pointer" mode in emutls.c.
4697 2017-06-27 Jakub Jelinek <jakub@redhat.com>
4700 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
4703 * gimple-fold.c (replace_call_with_call_and_fold): Handle
4704 gimple_vuse copying separately from gimple_vdef copying.
4706 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4708 * value-prof.c (free_hist): Remove call to memset and the enclosing if
4711 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
4712 Olivier Hainque <hainque@adacore.com>
4714 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
4715 for all vxworks7 targets.
4716 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
4717 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
4718 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
4719 variations for VX6/VX7 and 32/64bits later on in ...
4720 (VXWORKS_LIB_SPEC): Leverage new macros.
4721 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
4722 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
4724 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
4726 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
4727 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
4729 2017-06-26 Carl Love <cel@us.ibm.com>
4731 * config/rs6000/rs6000-c.c: Add support for built-in functions
4732 vector bool char vec_reve (vector bool char);
4733 vector signed char vec_reve (vector signed char);
4734 vector unsigned char vec_reve (vector unsigned char);
4735 vector bool int vec_reve (vector bool int);
4736 vector signed int vec_reve (vector signed int);
4737 vector unsigned int vec_reve (vector unsigned int);
4738 vector bool long long vec_reve (vector bool long long);
4739 vector signed long long vec_reve (vector signed long long);
4740 vector unsigned long long vec_reve (vector unsigned long long);
4741 vector bool short vec_reve (vector bool short);
4742 vector signed short vec_reve (vector signed short);
4743 vector double vec_reve (vector double);
4744 vector float vec_reve (vector float);
4745 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
4746 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
4747 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
4748 (altivec_vreve): New pattern.
4749 * config/rs6000/altivec.h (vec_reve): New define.
4750 * doc/extend.texi (vec_rev): Update the built-in documentation file
4751 for the new built-in functions.
4753 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4755 PR tree-optimization/71815
4756 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
4758 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
4760 (slsr_process_phi): Likewise.
4761 (replace_uncond_cands_and_profitable_phis): Don't replace a
4762 multiply candidate with a stride of 1 (copy or cast).
4763 (phi_incr_cost): Call uses_consumed_by_stmt rather than
4765 (lowest_cost_path): Likewise.
4766 (total_savings): Likewise.
4768 2017-06-26 Richard Biener <rguenther@suse.de>
4771 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
4772 Use def_builtin_pure for all gather builtins.
4774 2017-06-26 Richard Biener <rguenther@suse.de>
4776 PR tree-optimization/81203
4777 * tree-tailcall.c (find_tail_calls): Do not move stmts into
4780 2017-06-26 Marek Polacek <polacek@redhat.com>
4783 * doc/invoke.texi: Document -Wmultistatement-macros.
4785 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
4787 * doc/sourcebuild.texi (ARM-specific attributes): Document new
4788 arm_neon_ok_no_float_abi effective target.
4790 2017-06-26 Richard Biener <rguenther@suse.de>
4792 PR tree-optimization/80928
4793 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
4794 (copy_bbs): Set BB_DUPLICATED flag early.
4795 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
4797 (execute_on_shrinking_pred): Likewise.
4798 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
4799 BB_DUPLICATED blocks.
4800 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
4801 iterate over all PHIs considering removal of *gsi.
4803 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
4805 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
4808 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
4810 * config/rs6000/rs6000-string.c: (expand_block_clear,
4811 do_load_for_compare, select_block_compare_mode,
4812 compute_current_alignment, expand_block_compare,
4813 expand_strncmp_align_check, expand_strn_compare,
4814 expand_block_move, rs6000_output_load_multiple)
4815 Move functions related to string/block move/compare
4817 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
4818 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
4819 for this function which is now used in two files.
4820 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
4821 * config.gcc: Add rs6000-string.o to extra_objs for
4822 targets powerpc*-*-* and rs6000*-*-*.
4824 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
4827 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
4828 32-bit, since indexed is not valid for DImode.
4829 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
4830 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
4831 (define_peephole2 for Altivec d-form load): Add 32-bit support.
4832 (define_peephole2 for Altivec d-form store): Likewise.
4835 * multiple_target.c (create_dispatcher_calls): Only create the
4836 dispatcher call if the function is the default clone of a
4839 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
4842 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
4843 a call, force the call to not be a tail call.
4845 2017-06-23 Jeff Law <law@redhat.com>
4847 * doc/contrib.texi: Add entry for Steven Pemberton's work on
4850 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
4852 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
4853 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
4854 handling for early expansion of vector shifts (sl,sr,sra,rl).
4855 (builtin_function_type): Add vector shift right instructions
4856 to the unsigned argument list.
4858 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
4860 rtl-optimizatoin/79286
4861 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
4862 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
4863 trap. PIC register plus a const unspec without offset can never trap.
4865 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
4867 * tree.h (builtin_structptr_type): New type.
4868 (builtin_structptr_types): Declare new array.
4869 * tree.c (builtin_structptr_types): New array.
4870 (free_lang_data, build_common_tree_nodes): Use it.
4872 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
4875 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
4878 2017-06-22 Matt Turner <mattst88@gmail.com>
4880 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
4881 Lake models to skylake case. Assume skylake for unknown
4882 models with clflushopt.
4884 2017-06-22 Jeff Law <law@redhat.com>
4886 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
4887 frame sizes that do not satisfy aarch64_uimm12_shift.
4889 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
4891 * profile-count.h (apply_probability,
4892 apply_scale, probability_in): Fix checks for zero.
4894 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4896 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
4897 * doc/cppdiropts.texi (-I @var{dir}): Document it.
4899 2016-06-22 Richard Biener <rguenther@suse.de>
4901 * tree-vect-loop.c (vect_model_reduction_cost): Handle
4902 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
4903 REDUC_MAX_EXPR support.
4904 (vectorizable_reduction): Likewise.
4905 (vect_create_epilog_for_reduction): Likewise.
4907 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
4909 * match.pd (A / (1 << B) -> A >> B): New.
4910 * generic-match-head.c: Include optabs-tree.h.
4911 * gimple-match-head.c: Likewise.
4912 * optabs-tree.h (target_supports_op_p): New.
4913 * optabs-tree.c (target_supports_op_p): New.
4915 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4917 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
4918 $gcc_cv_ld --help output.
4919 (gcc_cv_ld_demangle): Likewise.
4920 (gcc_cv_ld_eh_frame_hdr): Likewise.
4921 (gcc_cv_ld_pie): Likewise.
4922 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
4923 (gcc_cv_ld_buildid): Likewise.
4924 (gcc_cv_ld_sysroot): Likewise.
4925 (ld_bndplt_support): Likewise.
4926 (ld_pushpopstate_support): Likewise.
4927 * configure: Regenerate.
4928 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
4930 2017-06-21 Jakub Jelinek <jakub@redhat.com>
4933 * config/i386/sse.md (round<mode>2): Renumber match_dup and
4934 operands indexes to avoid gap between operands and match_dups.
4936 2017-06-21 Andrew Pinski <apinski@cavium.com>
4938 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
4939 Increment Arith_shift and Arith_shift_reg by 1.
4940 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
4942 * config/aarch64/aarch64.c (thunderx_tunings): Enable
4943 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
4944 (aarch64_strip_extend): Add new argument and test for it.
4945 (aarch64_cheap_mult_shift_p): New function.
4946 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
4947 add a cost if it is true.
4948 Update calls to aarch64_strip_extend.
4949 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
4951 2017-06-21 Andrew Pinski <apinski@cavium.com>
4953 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
4955 (thunderxt88): Likewise.
4956 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
4957 (thunderx_prefetch_tune): New variable.
4958 (thunderx2t99_prefetch_tune): Update for the correct values.
4959 (thunderxt88_tunings): New variable.
4960 (thunderx_tunings): Use thunderx_prefetch_tune instead of
4961 generic_prefetch_tune.
4962 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
4964 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4966 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
4967 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
4968 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
4969 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
4970 (aarch64_atomic_cas<mode>, GPI): Likewise.
4972 2017-06-21 Martin Liska <mliska@suse.cz>
4974 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
4975 statements on cold and hot labels.
4976 * predict.c (tree_estimate_probability_bb): Remove the
4977 prediction from this place.
4979 2017-06-21 Martin Liska <mliska@suse.cz>
4981 PR tree-optimization/79489
4982 * gimplify.c (maybe_add_early_return_predict_stmt): New
4984 (gimplify_return_expr): Call the function.
4985 * predict.c (tree_estimate_probability_bb): Remove handling
4987 * predict.def: Update comment about early return predictor.
4988 * gimple-predict.h (is_gimple_predict): New function.
4989 * predict.def: Change default value of early return to 66.
4990 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
4992 * passes.def: Put pass_strip_predict_hints to the beginning of
4995 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
4997 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
4998 FUNCTION_DECL declarations.
4999 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
5001 (dwaf2out_decl): Likewise.
5002 * godump.c (go_early_global_decl): Skip call to the real debug hook
5003 for FUNCTION_DECL declarations.
5004 * passes.c (rest_of_decl_compilation): Skip call to the
5005 early_global_decl debug hook for FUNCTION_DECL declarations, unless
5006 -fdump-go-spec is passed.
5008 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
5010 * config/i386/i386.c (struct builtin_isa): New field pure_p.
5011 Reorder for compactness.
5012 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
5013 (def_builtin_pure, def_builtin_pure2): New functions.
5014 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
5016 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
5018 * match.pd (nop_convert): New predicate.
5019 ((A +- CST1) +- CST2): Allow some NOP conversions.
5021 2017-06-21 Jakub Jelinek <jakub@redhat.com>
5024 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
5025 with ctors/dtors if GOVD_SHARED is set.
5027 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
5029 * config/aarch64/aarch64.md (movti_aarch64):
5030 Emit mov rather than orr.
5031 (movtf_aarch64): Likewise.
5032 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
5033 Emit mov rather than orr.
5035 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
5037 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
5038 Swap alternatives, make integer dup more expensive.
5040 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
5042 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
5043 Return true for non-tls symbols.
5045 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
5047 * config/aarch64/aarch64-cores.def (cortex-a55): New.
5048 (cortex-a75): Likewise.
5049 (cortex-a75.cortex-a55): Likewise.
5050 * config/aarch64/aarch64-tune.md: Regenerate.
5051 * doc/invoke.texi (-mtune): Document new values for -mtune.
5053 2017-06-21 Tom de Vries <tom@codesourcery.com>
5055 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
5057 (Effective-Target Keywords, Other attributes): Suggest using
5058 dg-add-options stack_size feature to get stack limit in stack_size
5059 effective target documentation.
5061 2017-06-21 Julian Brown <julian@codesourcery.com>
5062 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
5064 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
5065 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
5066 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
5068 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
5069 attribute type list for neon_multiply.
5070 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
5071 attribute type list for neon_multiply.
5072 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
5073 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
5074 attribute type list for neon_multiply.
5075 * config/arm/types.md (crypto_pmull): Add.
5076 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
5077 attribute type list.
5079 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
5081 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
5084 2017-06-20 Jakub Jelinek <jakub@redhat.com>
5086 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
5087 to make sure not to dereference a NULL cost_classes_ptr pointer.
5089 2017-06-20 Carl Love <cel@us.ibm.com>
5091 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5092 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
5093 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
5094 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
5095 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
5096 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
5097 VMULOSW): New enum "unspec" values.
5098 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
5099 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
5100 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
5101 altivec_vmulosw): New patterns.
5102 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
5103 VMULOSW): Add definitions.
5105 2017-06-20 Julia Koval <julia.koval@intel.com>
5107 * config/i386/i386.c: Fix rounding expand for new pattern.
5108 * config/i386/subst.md: Fix pattern (parallel -> unspec).
5110 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
5112 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
5113 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
5115 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
5117 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
5120 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
5122 * config/aarch64/aarch64-cores.def: Rearrange to sort by
5123 architecture, then by implementer ID.
5124 * config/aarch64/aarch64-tune.md: Regenerate.
5126 2017-06-20 Richard Biener <rguenther@suse.de>
5129 * fold-const.c (split_tree): Fold to type before negating.
5131 2017-06-20 David Malcolm <dmalcolm@redhat.com>
5133 * diagnostic-show-locus.c
5134 (selftest::test_fixit_deletion_affecting_newline): New function.
5135 (selftest::diagnostic_show_locus_c_tests): Call it.
5137 2017-06-20 Andreas Schwab <schwab@suse.de>
5140 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
5143 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
5145 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
5146 __ARM_FEATURE_COPROC according to support.
5148 2017-06-20 Jakub Jelinek <jakub@redhat.com>
5150 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
5151 Rewritten to avoid overflow for > 32-bit pointers.
5154 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
5155 by removing enum keyword.
5156 (ubsan_type_descriptor): Likewise. Formatting fix.
5159 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
5160 splitter): Require TARGET_SSE2 in the condition.
5162 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
5165 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
5166 for doing vector set of SFmode on ISA 3.0.
5167 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
5168 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
5170 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
5171 SFmode value into a V4SF variable that was extracted from another
5172 V4SF variable without converting the element to double precision
5173 and back to single precision vector format.
5174 (vsx_insert_extract_v4sf_p9_2): Likewise.
5176 2017-06-19 Jakub Jelinek <jakub@redhat.com>
5178 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
5179 in UWHI to avoid undefined overflow.
5182 * ubsan.h (enum ubsan_encode_value_phase): New.
5183 (ubsan_encode_value): Change second argument to
5184 enum ubsan_encode_value_phase with default value of
5185 UBSAN_ENCODE_VALUE_GENERIC.
5186 * ubsan.c (ubsan_encode_value): Change second argument to
5187 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
5188 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
5189 create_tmp_var_raw instead of create_tmp_var and use a
5191 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
5192 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
5193 ubsan_encode_value callers.
5196 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
5197 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
5198 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
5200 2017-06-19 Richard Biener <rguenther@suse.de>
5203 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
5204 estimates if we changed anything.
5206 2017-06-19 Richard Biener <rguenther@suse.de>
5208 PR tree-optimization/80887
5209 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
5210 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
5211 simplified lookups, then reset mprts_hook.
5212 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
5214 (try_to_simplify): Likewise.
5216 2017-06-19 Martin Liska <mliska@suse.cz>
5219 * gimplify.c (gimplify_switch_expr):
5220 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
5222 2017-06-19 Martin Liska <mliska@suse.cz>
5224 * doc/install.texi: Document that PGO runs in 4 stages.
5226 2017-06-19 Martin Liska <mliska@suse.cz>
5229 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
5230 to dispatcher function name.
5231 * multiple_target.c (replace_function_decl): New function.
5232 (create_dispatcher_calls): Redirect both edges and references.
5234 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
5236 * profile-count.c (profile_count::dump): Dump quality.
5237 (profile_count::differs_from_p): Update for unsigned val.
5238 * profile-count.h (profile_count_quality): New enum.
5239 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
5241 2017-06-19 Richard Biener <rguenther@suse.de>
5243 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
5244 struct function as arg.
5245 (estimate_numbers_of_iterations): Export overload with loop arg.
5246 (free_numbers_of_iterations_estimates_loop): Use an overload of
5247 free_numbers_of_iterations_estimates instead.
5248 * tree-cfg.c (remove_bb): Adjust.
5249 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
5250 * tree-parloops.c (gen_parallel_loop): Likewise.
5251 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
5253 (tree_unroll_loops_completely): Likewise.
5254 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
5255 Use an overload instead and export.
5256 (estimated_loop_iterations): Adjust.
5257 (max_loop_iterations): Likewise.
5258 (likely_max_loop_iterations): Likewise.
5259 (estimate_numbers_of_iterations): Take struct function as arg
5261 (loop_exits_before_overflow): Adjust.
5262 (free_numbers_of_iterations_estimates_loop): Use an overload.
5263 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
5264 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
5266 2017-06-19 Richard Biener <rguenther@suse.de>
5269 * ipa-prop.c (find_constructor_constant_at_offset): Handle
5270 RANGE_EXPR conservatively.
5272 2017-06-16 Carl Love <cel@us.ibm.com>
5274 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5275 definitions for vec_float, vec_float2, vec_floato,
5276 vec_floate built-ins.
5277 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
5278 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
5280 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
5281 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
5282 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
5283 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
5284 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
5285 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
5286 vec_floato): Add builtin defines.
5287 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
5288 Update the built-in documentation file for the new built-in
5291 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5293 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
5294 (mthumb): Mark as the negative of -marm.
5296 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5298 * doc/invoke.texi (ARM Options, -mcpu): Document supported
5300 (ARM Options, -mtune): Document that this accepts the same
5301 extension options as -mcpu.
5302 (ARM Options, -mfpu): Document addition of -mfpu=auto.
5304 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5306 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
5307 permitted extensions.
5309 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5311 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
5312 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
5313 (armv8-m.main): Add option +nodsp.
5314 * config/arm/arm-cpu-cdata.h: Regenerated.
5316 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5318 * config/arm/t-fuchsia: New file.
5319 * config.gcc (arm*-*-fuchsia*): Use it.
5321 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5323 * config/arm/t-symbian: Rewrite for new option infrastructure.
5325 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5327 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
5328 (MULTILIB_REQUIRED): Likewise.
5330 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5332 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
5333 (MULTILIB_RESUE): Likewise.
5334 (MULTILIB_MATCHES): Likewise.
5335 (MULTLIB_REQUIRED): Likewise.
5337 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5339 * config/arm/t-rtems: Rewrite for new option framework.
5341 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5343 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
5344 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
5345 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
5346 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
5347 * config/arm/t-multilib: ... here.
5348 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
5349 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
5350 armv7-a and armv8*-a when A-profile libraries have not been built.
5351 * config/arm/t-rmprofile: Rewrite.
5353 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5355 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
5356 with a backslash. Remove the backslash after substituting unescaped
5358 * doc/fragments.texi (MULTILIB_REUSE): Document it.
5360 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5362 * config.gcc: (arm*-*-*): When building a-profile libraries, force
5363 the driver to pass through the default setting of -mfloat-abi.
5364 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
5366 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
5367 (all_feat_combs): New rule.
5368 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
5370 * config/arm/t-aprofile: Rewrite.
5372 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5374 * config/arm/arm.h (FPUTYPE_AUTO): Define.
5375 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
5376 fpu is not specified by the user/command-line.
5377 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
5378 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
5379 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
5380 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
5381 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
5382 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
5384 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5386 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
5387 * config/arm/t-arm-elf: Rewritten.
5389 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5391 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
5392 have some floating-point instructions.
5393 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
5394 (TARGET_MAYBE_HARD_FLOAT): New macro.
5395 * config/arm/arm-builtins.c (arm_init_builtins): Use
5396 TARGET_MAYBE_HARD_FLOAT.
5397 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
5399 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5401 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
5402 (configargs.h): Include it.
5403 (arm_print_hint_for_fpu_option): New function.
5404 (arm_parse_fpu_option): New function.
5405 (candidate_extension): New class.
5406 (arm_canon_for_multilib): New function.
5407 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
5408 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
5409 (ARCH_CANONICAL_SPECS): New macro.
5410 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
5412 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5414 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
5415 are set after handling multilib fragments. Set target_cpu_default2
5418 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5420 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
5422 (arm*-*-*): Set target_cpu_default2 to a quoted string.
5423 * config/arm/parsecpu.awk (check_cpu): Validate any extension
5425 (check_arch): Likewise.
5426 * config/arm/arm.c (arm_configure_build_target): Handle
5427 TARGET_CPU_DEFAULT being a string constant. Scan any feature
5428 options in the default.
5430 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5432 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
5433 when an option is an alias of another.
5434 * config/arm/parsecpu.awk (optalias): New parser token.
5435 (gen_comm_data): Mark non-alias options as such. Emit entries
5436 for extension aliases.
5437 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
5438 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
5439 (armv6kz, armv6zk, armv6t2): Likewise.
5440 (armv7): Make vfpv3-d16 an alias.
5441 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
5443 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
5444 Sort in canonical order.
5445 (armv8-a): Sort in canonical order.
5446 (armv8.1-a, armv8.2-a): Likewise.
5447 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
5449 (cortex-a9): Sort in canonical order.
5450 * config/arm/arm.c (selftests.h): Include it.
5451 (arm_test_cpu_arch_data): New function.
5452 (arm_run_self_tests): New function.
5453 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
5454 (targetm): Move declaration to the end of the file.
5455 * arm-cpu-cdata.h: Regenerated.
5457 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5459 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
5460 call to target_mode_check describing the type of option passed.
5461 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
5462 (arm_target_thumb_only): Use arm_parse_arch_option_name or
5463 arm_parse_cpu_option_name to match parameters against list of
5465 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
5466 arm_arch_core_flags data structure.
5467 * config/arm/arm-cpu_cdata.h: Regenerated.
5469 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5471 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
5473 (arm_print_hint_for_cpu_option): Likewise.
5474 (arm_print_hint_for_arch_option): Likewise.
5475 (arm_parse_cpu_option_name): Likewise.
5476 (arm_parse_arch_option_name): Likewise.
5477 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
5478 of entries in the all_fpus list.
5479 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
5480 (arm_parse_cpu_option_name): Declare.
5481 (arm_parse_arch_option_name): Declare.
5482 (arm_parse_option_features): Declare.
5483 (arm_intialize_isa): Declare.
5484 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
5486 (gen_comm_data): ... here. Make definitions non-static.
5487 * config/arm/arm-cpu-data.h: Regenerated.
5488 * config/arm/arm-cpu-cdata.h: Regenerated.
5490 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5492 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
5493 (cpu_arch_extension): New structure.
5494 (cpu_arch_option, arch_option, cpu_option): New structures.
5495 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
5497 (gen_data): Generate new format data tables.
5498 * config/arm/arm.c (cpu_tune): New structure.
5499 (cpu_option, processors): Delete.
5500 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
5501 (arm_print_hint_for_cpu_option): ... this and ...
5502 (arm_print_hint_for_arch_option): ... this.
5503 (arm_parse_arch_cpu_name): Delete. Replace with ...
5504 (arm_parse_cpu_option_name): ... this and ...
5505 (arm_parse_arch_option_name): ... this.
5506 (arm_unrecognized_feature): Change type of target parameter to
5508 (arm_parse_arch_cpu_features): Delete. Replace with ...
5509 (arm_parse_option_features): ... this.
5510 (arm_configure_build_target): Rework to use new configuration data
5512 (arm_print_tune_info): Rework for new configuration data tables.
5513 * config/arm/arm-cpu-data.h: Regenerated.
5514 * config/arm/arm-cpu.h: Regenerated.
5516 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5518 * Makefile.in (OBJS): Move sbitmap.o from here ...
5519 (OBJS-libcommon): ... to here.
5521 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5523 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
5524 (ISA_ALL_CRYPTO): New macro.
5525 (ISA_ALL_SIMD): New macro
5526 (ISA_ALL_FP): New macro.
5527 * config/arm/arm.c (fpu_bitlist): Update initializer.
5528 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
5530 (arm9e): Add fpu. Add option for nofp
5531 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
5532 (arm926ej-s, arm1026ej-s): Likewise.
5533 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
5534 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
5535 neon-fp16, neon-vfpv4, nofp and nosimd.
5536 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
5537 (cortex-a8): Add fpu. Add option for nofp.
5538 (cortex-a9): Add fpu. Add options for nosimd and nofp.
5539 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
5540 (cortex-r4f): Add fpu.
5541 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
5542 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
5544 (cortex-r8): Likewise.
5545 (cortex-m4): Add fpu. Add option for nofp.
5546 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
5547 (cortex-a17.cortex-a7): Likewise.
5548 (cortex-a32): Add fpu. Add options for crypto and nofp.
5549 (cortex-a35, cortex-a53): Likewise.
5550 (cortex-a57): Add fpu. Add option for crypto.
5551 (cortex-a72, cortex-a73): Likewise.
5552 (exynos-m1): Likewise.
5553 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
5554 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
5555 (cortex-m33): Add fpu. Add option for nofp.
5556 * config/arm/arm-cpu-cdata.h: Regenerated
5557 * config/arm/arm-cpu-data.h: Regenerated.
5559 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5561 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
5562 (armv5te, armv5tej): Likewise.
5563 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
5564 (armv7): Add options fp and vfpv3-d16.
5565 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
5566 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
5568 (armv7ve): Likewise.
5569 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
5570 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
5571 (armv8-a): Add nocrypto option.
5572 (armv8.1-a, armv8.2-a): Likewise.
5573 (armv8-m.main): add options fp, fp.dp and nofp.
5575 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5577 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
5579 (armv8-a+crc): Delete.
5580 (armv8.1-a): Add options simd, crypto and nofp.
5581 (armv8.2-a): Add options fp16, simd, crypto and nofp.
5582 (armv8.2-a+fp16): Delete.
5583 (armv8-m.main): Add option dsp.
5584 (armv8-m.main+dsp): Delete.
5585 (cortex-a8): Add fpu. Add nofp option.
5586 (cortex-a9): Add fpu. Add nofp and nosimd options.
5587 * config/arm/parsecpu.awk (gen_data): Generate option tables and
5588 link to main cpu and architecture data structures.
5589 (gen_comm_data): Only put isa attributes from the main architecture
5591 (option): New statement for architecture and CPU entries.
5592 * arm.c (struct cpu_option): New structure.
5593 (struct processors): Add entry for options.
5594 (arm_unrecognized_feature): New function.
5595 (arm_parse_arch_cpu_name): Ignore any characters after the first
5597 (arm_parse_arch_cpu_feature): New function.
5598 (arm_configure_build_target): Separate out any CPU and architecture
5599 features and parse separately. Don't error out if -mfpu=auto is
5600 used with only an architecture string.
5601 (arm_print_asm_arch_directives): New function.
5602 (arm_file_start): Call it.
5603 * config/arm/arm-cpu-cdata.h: Regenerated.
5604 * config/arm/arm-cpu-data.h: Likewise.
5605 * config/arm/arm-tables.opt: Likewise.
5607 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5609 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
5610 assembler when it is not -mfpu=auto.
5612 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5614 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
5615 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
5616 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
5617 (ASM_CPU_SPEC): Rewrite.
5618 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
5619 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
5620 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
5621 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
5622 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
5623 copied string is NUL-terminated. Also strip any characters prefixed
5625 (arm_rewrite_selected_arch): New function.
5626 (arm_rewrite_march): New function.
5628 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
5630 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
5631 (x_arm_cpu_string, x_arm_tune_string): Likewise.
5632 (march, mcpu, mtune): Convert to string-based options.
5633 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
5634 (arm_parse_arch_cpu_name): New function.
5635 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
5636 identify selected architecture or CPU.
5637 (arm_option_save): New function.
5638 (TARGET_OPTION_SAVE): Redefine.
5639 (arm_option_restore): Restore string options.
5640 (arm_option_print): Print string options.
5642 2017-06-16 Martin Sebor <msebor@redhat.com>
5644 PR tree-optimization/80933
5645 PR tree-optimization/80934
5646 * builtins.c (fold_builtin_3): Do not handle bcmp here.
5647 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
5648 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
5649 (gimple_fold_builtin): Call them.
5651 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5653 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
5654 as unlikely; update profile.
5656 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5658 * predict.c (force_edge_cold): Handle declaring edges impossible
5661 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5663 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
5665 (try_unroll_loop_completely): Fix reporting.
5667 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5669 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
5671 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
5674 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
5675 if given a non-constant argument for an intrinsic which requires a
5678 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5680 * profile.c (compare_freqs): New function.
5681 (branch_prob): Sort edge list.
5682 (find_spanning_tree): Assume that the list is priority sorted.
5684 2017-06-16 Richard Biener <rguenther@suse.de>
5686 PR tree-optimization/81090
5687 * passes.def (pass_record_bounds): Remove.
5688 * tree-pass.h (make_pass_record_bounds): Likewise.
5689 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
5690 make_pass_record_bounds): Likewise.
5691 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
5692 not free niter estimates at the beginning but at the end.
5693 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
5695 2017-06-16 Richard Biener <rguenther@suse.de>
5697 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
5698 initializer to workaround ICE in host GCC 4.8.
5700 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
5702 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
5704 (clone_inlined_nodes): Update.
5706 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
5708 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
5709 prefetch settings, and enable prefetching by default at -O3.
5711 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
5713 * config/aarch64/aarch64.c (aarch64_override_options_internal):
5714 Set flag_prefetch_loop_arrays according to tuning data.
5716 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
5718 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
5720 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
5721 [Unrelated to main purpose of the patch] Place the pointer field last
5722 to enable type checking errors when tune structure are wrongly merged.
5723 * config/aarch64/aarch64.c (generic_prefetch_tune,)
5724 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
5725 (thunderx2t99_prefetch_tune): New tune constants.
5726 (tune_params *_tunings): Update all tunings (no functional change).
5727 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
5728 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
5729 from tunings structures.
5731 2017-06-16 Jakub Jelinek <jakub@redhat.com>
5734 * ubsan.c (instrument_null): Add T argument, use it instead
5735 of computing it based on IS_LHS.
5736 (instrument_object_size): Likewise.
5737 (pass_ubsan::execute): Adjust instrument_null and
5738 instrument_object_size callers to pass gimple_get_lhs or
5739 gimple_assign_rhs1 result to it. Use instrument_null instead of
5740 calling get_base_address and instrument_mem_ref. Handle
5741 aggregate call arguments for object-size sanitization.
5743 2017-06-16 Yury Gribov <tetra2005@gmail.com>
5745 PR tree-optimization/81089
5746 * tree-vrp.c (is_masked_range_test): Validate operands of
5749 2017-06-15 Martin Sebor <msebor@redhat.com>
5752 * dumpfile.c (dump_register): Avoid calling memset to initialize
5753 a class with a default ctor.
5754 * gcc.c (struct compiler): Remove const qualification.
5755 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
5756 * hash-table.h: Ditto.
5757 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
5759 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
5760 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
5762 * params.h (struct param_info): Make struct members non-const.
5763 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
5764 with default initialization.
5765 * vec.h (vec_copy_construct, vec_default_construct): New helper
5767 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
5768 with vec_copy_construct.
5769 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
5770 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
5771 * doc/invoke.texi (-Wclass-memaccess): Document.
5773 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
5775 * emit-rtl.h (is_leaf): Update comment about local
5778 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
5781 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
5782 for a variable to have a section before checking if the section has a
5784 Set section to.persistent if persistent attribute is set.
5785 Warn if .persistent attribute is used on an automatic variable.
5787 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
5789 PR rtl-optimization/80474
5790 * reorg.c (update_block): Do not ignore instructions in a delay slot.
5792 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
5794 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
5797 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
5799 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
5800 (casesi): Emit bounds checking as RTL.
5801 (casesi_internal_mips16_<mode>): Remove bounds checking.
5803 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
5805 * config/xtensa/xtensa.c (xtensa_option_override): Append
5806 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
5807 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
5808 xtensa_doloop_hooks): Define unconditionally.
5809 (xtensa_reorg_loops): Only call reorg_loops in the presence of
5811 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
5812 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
5813 for it in xtensa_option_override.
5814 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
5815 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
5817 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
5819 * doc/cppopts.texi: Document '-' special value to -MF.
5821 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
5823 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
5824 (cortex_a53_fconst): Likewise.
5825 (cortex_a53_fpmul): Likewise.
5826 (cortex_a53_f_load_64): Likewise.
5827 (cortex_a53_f_load_many): Likewise.
5828 (cortex_a53_advsimd_alu): Likewise.
5829 (cortex_a53_advsimd_alu_q): Likewise.
5830 (cortex_a53_advsimd_mul): Likewise.
5831 (cortex_a53_advsimd_mul_q): Likewise.
5832 (fpmac bypass): Add new bypass for fpmac-fpmac case.
5833 Add missing fmul, r2f_cvt and fconst cases.
5835 2017-06-14 Richard Biener <rguenther@suse.de>
5838 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
5840 (fold_binary_loc): When associating do not treat pre-existing
5841 TREE_OVERFLOW on literal constants as a reason to allow
5842 TREE_OVERFLOW on associated literal constants.
5844 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
5846 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
5847 (MASK_FEATURES): New macro.
5848 * config/sparc/sparc.c (sparc_option_override): Remove the special
5849 handling of -mfpu and generalize it to all MASK_FEATURES switches.
5851 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
5853 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
5854 a division of 0 if non-call exceptions are enabled.
5856 2017-06-14 Andrew Pinski <apinski@cavium.com>
5857 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
5860 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
5861 Improve vector initialization code gen for only variable case.
5863 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
5865 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
5867 2017-06-14 Richard Biener <rguenther@suse.de>
5869 PR tree-optimization/81083
5870 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
5873 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5875 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
5876 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
5877 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
5878 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
5879 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
5880 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
5882 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5884 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
5885 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
5887 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5889 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
5891 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5893 * config/rs6000/t-rtems: Don't handle SPE.
5895 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5897 * config/rs6000/t-linux: Don't handle SPE.
5899 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5901 * config/rs6000/eabispe.h: Delete file.
5903 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5905 * config/rs6000/t-spe: Delete file.
5907 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5909 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
5910 (rs6000_legitimate_offset_address_p): Return false for anything in
5911 V2SImode or V2SFmode.
5913 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
5915 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
5916 except V2SF and V2SI. Rearrange the vector modes, and add comments.
5917 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
5919 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
5920 (rs6000_legitimate_offset_address_p): Ditto.
5921 (rs6000_emit_move): Ditto.
5922 (rs6000_init_builtins): Remove V4HI_type_node.
5924 2017-06-13 Martin Liska <mliska@suse.cz>
5927 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
5928 (gate_asan): Likewise.
5929 * asan.h (asan_no_sanitize_address_p): Remove the function.
5930 (sanitize_flags_p): New function.
5931 * builtins.def: Fix coding style.
5932 * common.opt: Use renamed enum value.
5933 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
5934 * doc/extend.texi: Document no_sanitize attribute.
5935 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
5936 to SANITIZE_UNDEFINED_NONDEFAULT.
5937 * gcc.c (sanitize_spec_function): Use the renamed enum value.
5938 * gimple-fold.c (optimize_atomic_compare_exchange_p):
5939 Use sanitize_flags_p.
5940 * gimplify.c (gimplify_function_tree): Likewise.
5941 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
5942 * opts.c (parse_no_sanitize_attribute): New function.
5943 (common_handle_option): Use renamed enum value.
5944 * opts.h (parse_no_sanitize_attribute): Declare.
5945 * tree.c (sanitize_flags_p): New function.
5946 * tree.h: Declared here.
5947 * tsan.c: Use sanitize_flags_p.
5948 * ubsan.c (ubsan_expand_null_ifn): Likewise.
5949 (instrument_mem_ref): Likewise.
5950 (instrument_bool_enum_load): Likewise.
5951 (do_ubsan_in_current_function): Remove the function.
5952 (pass_ubsan::execute): Use sanitize_flags_p.
5953 * ubsan.h: Remove do_ubsan_in_current_function
5954 * tree-cfg.c (print_no_sanitize_attr_value): New function.
5955 (dump_function_to_file): Use it here.
5957 2017-06-13 Martin Jambor <mjambor@suse.cz>
5959 PR tree-optimization/80803
5960 PR tree-optimization/81063
5961 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
5962 (propagate_subaccesses_across_link): Enqueue subtree whenever
5963 necessary instead of relying on the caller.
5965 2017-06-13 Martin Jambor <mjambor@suse.cz>
5967 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
5968 that have a first_link.
5969 (sort_and_splice_var_accesses): Do not check first_link before
5971 (subtree_mark_written_and_enqueue): Likewise.
5972 (propagate_all_subaccesses): Likewise and do not stop at first
5973 parent with a first_link.
5975 2017-06-13 Martin Jambor <mjambor@suse.cz>
5977 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
5980 2017-06-13 Yury Gribov <tetra2005@gmail.com>
5982 * match.pd: New pattern.
5984 2017-06-13 Yury Gribov <tetra2005@gmail.com>
5986 * tree-vrp.c (is_masked_range_test): New function.
5987 (register_edge_assert_for): Determine ranges for
5990 2017-06-13 Yury Gribov <tetra2005@gmail.com>
5992 PR tree-optimization/67328
5993 * fold-const.c (maskable_range_p): New function.
5994 (build_range_check): Generate bittests if possible.
5996 2017-06-13 Martin Liska <mliska@suse.cz>
5998 * gimple-pretty-print.c (dump_probability): Add new argument.
5999 (dump_edge_probability): Dump both probability and count.
6000 (dump_gimple_label): Likewise.
6001 (dump_gimple_bb_header): Likewise.
6003 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
6006 * config/avr/avr-devices.c: Fix indentation.
6007 * config/avr/gen-avr-mmcu-specs.c: Dito.
6009 2017-06-13 Richard Biener <rguenther@suse.de>
6011 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
6012 instead get vector type from stmt_info.
6013 (vectorizable_reduction): Adjust. Remove dead code.
6015 2017-06-13 Richard Biener <rguenther@suse.de>
6018 * fold-const.c (extract_muldiv_1): Remove bogus distribution
6019 case of C * (x * C2 + C3).
6020 (fold_addr_of_array_ref_difference): Properly fold index difference.
6022 2017-06-12 David S. Miller <davem@davemloft.net>
6025 * config/sparc/sparc.md (return expander): Emit frame blockage if
6026 function uses alloca.
6028 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
6030 * combine.c (make_field_assignment): Check len rather than the mode
6031 precision when calling force_to_mode.
6033 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
6035 Support multilibs and devices that see flash in RAM address range.
6038 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
6039 (avr_mcu_t) <flash_pm_offset>: New field.
6040 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
6041 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
6042 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
6043 (AVR_TINY_PM_OFFSET): Remove macro.
6044 * config/avr/avr.opt (-mshort-calls): New option.
6045 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
6046 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
6047 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
6048 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
6049 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
6050 instead of avr_arch->have_jmp_call.
6051 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
6052 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
6053 avr_arch->flash_pm_offset to define.
6054 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
6055 new field flash_pm_offset. Add entry for avrxmega3.
6056 (avr_texinfo): Add entry for avrxmega3.
6057 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
6058 attiny212, attiny214,
6059 attiny412, attiny414, attiny416, attiny417,
6060 attiny814, attiny816, attiny817,
6061 attiny1614, attiny1616, attiny1617,
6062 attiny3214, attiny3216, attiny3217.
6063 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
6064 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
6065 (avr_print_operand_address) [AVR_TINY]: Same.
6066 (avr_asm_init_sections) <readonly_data_section>: Only patch
6067 callback if avr_arch->flash_pm_offset = 0.
6068 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
6069 for rodata if avr_arch->flash_pm_offset != 0.
6070 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
6071 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
6072 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
6073 (m_options): Append opt_rcall.
6074 (m_dirnames): Append dir_rcall.
6075 * config/avr/t-multilib: Regenerate.
6077 * configure.ac [target=avr]: Check whether avrxmega3 default
6078 linker description file works as needed.
6079 * configure: Regenerate.
6080 * doc/avr-mmcu.texi: Regenerate.
6081 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
6082 <__AVR_ARCH__>: Document avrxmega3 and 103.
6083 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
6084 <__AVR_SHORT_CALLS__>: Document it.
6085 <__AVR_PM_BASE_ADDRESS__>: Document it.
6086 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
6087 (AVR Variable Attributes) <progmem>: Document this is
6088 not needed for avrxmega3.
6089 (AVR Named Address Spaces) <__flash>: Dito.
6091 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
6093 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
6095 2017-06-12 Doug Rupp <rupp@adacore.com>
6097 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
6098 Append vxworks-stdint.h to the tm_file list.
6099 * config/vxworks-stdint.h: New file.
6101 2017-06-12 Martin Liska <mliska@suse.cz>
6103 PR tree-optimization/81041
6104 * tree-profile.c (gimple_gen_ic_func_profiler):
6105 Create an extra BB in profile-generate
6106 (gimple_gen_time_profiler): Likewise.
6108 2017-06-12 Jakub Jelinek <jakub@redhat.com>
6110 PR tree-optimization/81003
6111 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
6112 (update_range_test): Use it instead of force_gimple_operand_gsi.
6114 2017-06-12 Richard Biener <rguenther@suse.de>
6116 PR tree-optimization/81053
6117 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
6118 with backedge value not defined in loop. Simplify def stmt
6121 2017-06-11 Tom de Vries <tom@codesourcery.com>
6124 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
6126 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
6127 nvptx_cannot_force_const_mem.
6129 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
6131 * opts.c (finish_options): Move test for flag_split_stack after
6132 it has been initialized.
6134 2017-06-11 Jason Merrill <jason@redhat.com>
6136 * tree.h (id_equal): New.
6137 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
6138 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
6139 instead of strcmp of IDENTIFIER_POINTER.
6141 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
6143 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
6144 (mark_all_inlined_calls_cdtor): Fix formating.
6145 (inline_transform): Rescale profile before inlining.
6147 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
6149 * cgraph.h (cgraph_edge::clone): Update prototype.
6150 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
6151 (cgraph_node::create_clone): Update.
6152 (cgraph_node::create_version_clone): Update.
6153 * tree-inline.c (copy_bb): Update.
6154 (expand_call_inline): Update.
6156 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
6158 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
6159 factored out from ...
6160 (rs6000_emit_prologue): ... here.
6162 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
6164 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
6165 factored out from ...
6166 (rs6000_emit_prologue): ... here.
6168 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
6170 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
6172 (handle_missing_profiles): Fix computation of tp_first_run.
6173 (counts_to_freqs): Do not touch freqs when count is 0.
6175 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
6177 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
6180 2017-06-10 Tom de Vries <tom@codesourcery.com>
6182 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
6183 attributes): Document signal effective target.
6185 2017-06-10 Tom de Vries <tom@codesourcery.com>
6187 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
6188 Document effective target stack_size.
6190 2017-06-09 David Malcolm <dmalcolm@redhat.com>
6192 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
6193 to the edit_context if they can be auto-applied.
6195 2017-06-9 Ian Lance Taylor <iant@golang.org>
6197 * opts.c (finish_options): If -fsplit-stack, disable implicit
6198 -forder-blocks-and-partition.
6199 * doc/invoke.texi (Optimize Options): Document that when using
6200 -fsplit-stack -forder-blocks-and-partition is not implicitly
6203 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
6205 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
6206 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
6207 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
6208 * builtins.def (abort, trap, unreachable): Declare cold.
6209 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
6210 * tree-core.h (ECF_COLD): New.
6211 * tree.c (set_call_expr_flags): Handle ECF_COLD.
6212 (build_common_builtin_nodes): Mark unreachable and abort as cold.
6214 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
6216 * predict.c (unlikely_executed_stmt_p): Cleanup.
6218 2017-06-09 Richard Biener <rguenther@suse.de>
6220 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
6221 model if the ref is always written to.
6223 2017-06-09 Tamar Christina <tamar.christina@arm.com>
6225 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
6227 2017-06-09 Tamar Christina <tamar.christina@arm.com>
6229 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
6232 2017-06-09 Tom de Vries <tom@codesourcery.com>
6235 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
6236 "target cannot support label values" when encountering LABEL_REF.
6238 2017-06-09 Martin Liska <mliska@suse.cz>
6240 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
6241 (gimple_gen_ic_func_profiler): Emit direct comparison
6242 of __gcov_indirect_call_callee with NULL.
6243 (gimple_gen_time_profiler): Change probability from
6244 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
6246 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
6248 * profile.c (edge_gcov_counts): Turn to pointer.
6249 (compute_branch_probabilities, compute_branch_probabilities): Update.
6250 (branch_prob): Do not clear edge_gcov_count.
6251 * profile.h (edge_gcov_counts): Turn to pointer.
6252 (edge_gcov_count): Update.
6254 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
6256 * gimple.h (gimple_check_failed): Mark cold.
6258 2017-06-09 Richard Biener <rguenther@suse.de>
6260 PR tree-optimization/66623
6261 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
6262 refactor check_reduction into two parts, properly computing
6263 whether we have to check reduction validity for outer loop
6266 2017-06-09 Richard Biener <rguenther@suse.de>
6268 PR tree-optimization/79483
6269 * graphite-scop-detection.c (order): New global.
6270 (get_order): Compute bb to order mapping that satisfies code
6271 generation constraints.
6272 (cmp_pbbs): New helper.
6273 (build_scops): Start domwalk at entry block, sort generated
6276 2017-06-09 Richard Biener <rguenther@suse.de>
6279 * ipa-polymorphic-call.c
6280 (ipa_polymorphic_call_context::restrict_to_inner_class):
6281 Skip FIELD_DECLs with error_mark_node type.
6282 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
6285 2017-06-09 Martin Liska <mliska@suse.cz>
6287 * predict.c (struct branch_predictor): New struct.
6288 (test_prediction_value_range): New test.
6289 (predict_c_tests): New function.
6290 * selftest-run-tests.c (selftest::run_tests): Run the function.
6291 * selftest.h: Declare new tests.
6293 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
6296 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
6297 gen_add3_insn did not fail.
6298 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
6299 r0, construct that number in a temporary reg and add that reg to r0.
6300 If asked to put the result in r0 as well, fail.
6302 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
6304 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6305 for early expansion of vec_eqv.
6307 2017-06-08 Jakub Jelinek <jakub@redhat.com>
6310 * ubsan.c (instrument_null): Avoid pointless code temporary.
6311 (pass_ubsan::execute): Instrument aggregate arguments of calls.
6313 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
6317 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
6320 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
6323 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
6325 * predict.c (unlikely_executed_edge_p): Move ahead.
6326 (probably_never_executed_edge_p): Use it.
6328 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
6331 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
6332 gimple_call_builtin_p call.
6334 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
6336 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
6337 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
6338 rtl_check_failed_type2, rtl_check_failed_code1,
6339 rtl_check_failed_code2, rtl_check_failed_code_mode,
6340 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
6341 rtvec_check_failed_bounds, rtl_check_failed_flag,
6342 _fatal_insn_not_found, _fatal_insn): Likewise.
6343 * tree.h (tree_contains_struct_check_failed,
6344 tree_check_failed, tree_not_check_failed,
6345 tree_class_check_failed, tree_range_check_failed,
6346 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
6347 tree_vec_elt_check_failed, phi_node_elt_check_failed,
6348 tree_operand_check_failed, omp_clause_check_failed,
6349 omp_clause_operand_check_failed, omp_clause_range_check_failed):
6352 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
6354 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
6355 flag_branch_probabilities.
6356 * ipa-inline.c (edge_badness): Likewise.
6357 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
6358 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
6359 * predict.c (maybe_hot_frequency_p): Likewise.
6360 (probably_never_executed): Likewise.
6361 * sched-ebb.c (schedule_ebbs): Likewise.
6362 * sched-rgn.c (find_single_block_region): Likewise.
6363 * tracer.c (tail_duplicate): Likewise.
6365 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
6367 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
6368 longer requires x_flag_profile_use.
6370 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
6372 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
6373 instead of flag_reorder_blocks_and_partition.
6374 * dbxout.c (dbxout_function_end): Likewise.
6375 * dwarf2out.c (gen_subprogram_die): Likewise.
6376 * haifa-sched.c (sched_create_recovery_edges): Likewise.
6377 * hw-doloop.c (reorg_loops): Likewise.
6378 * varasm.c (assemble_start_function,
6379 assemble_end_function): Likewise.
6380 (decide_function_section): Do not check for
6381 flag_reorder_blocks_and_partition.
6383 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
6385 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
6387 (chkp_get_hard_register_fake_addr_expr): Ditto.
6388 (chkp_build_addr_expr): Add check for hard reg case.
6389 (chkp_parse_array_and_component_ref): Ditto.
6390 (chkp_find_bounds_1): Ditto.
6391 (chkp_process_stmt): Don't generate bounds store for
6394 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
6396 * predict.c (maybe_hot_bb_p): Do not check profile status.
6397 (maybe_hot_edge_p): Likewise.
6398 (probably_never_executed): Check for zero counts even if profile
6400 (unlikely_executed_edge_p): New function.
6401 (unlikely_executed_stmt_p): New function.
6402 (unlikely_executed_bb_p): New function.
6403 (set_even_probabilities): Use unlikely predicates.
6404 (combine_predictions_for_bb): Likewise.
6405 (predict_paths_for_bb): Likewise.
6406 (predict_paths_leading_to_edge): Likewise.
6407 (determine_unlikely_bbs): New function.
6408 (estimate_bb_frequencies): Use it.
6409 (compute_function_frequency): Use zero counts even if profile is
6411 * profile-count.h: Fix typo.
6413 2017-08-08 Julia Koval <julia.koval@intel.com>
6415 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
6416 _mm512_mask_cvtsepi16_storeu_epi8,
6417 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
6418 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
6419 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
6420 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
6421 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
6422 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
6423 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
6424 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
6425 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
6426 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
6427 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
6428 __builtin_ia32_pmovuswb256mem_mask,
6429 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
6430 __builtin_ia32_pmovwb512mem_mask): New builtins.
6432 2017-08-08 Julia Koval <julia.koval@intel.com>
6434 PR target/73350,80862
6435 * config/i386/subst.md (round): Fix round pattern.
6436 * config/i386/i386.c (ix86_erase_embedded_rounding):
6437 Fix erasing rounding for the fixed pattern.
6439 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
6441 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
6443 2017-06-08 Martin Liska <mliska@suse.cz>
6445 PR gcov-profile/80911
6446 * gcov.c (block_info::block_info): New constructor.
6448 2017-06-07 Carl Love <cel@us.ibm.com>
6450 * config/rs6000/rs6000-c: The return type of the following
6451 built-in functions was implemented as int not long long. Fix sign
6452 of return value for the unsigned version of vec_mulo and vec_mule.
6453 vector unsigned long long vec_bperm (vector unsigned long long,
6454 vector unsigned char)
6455 vector signed long long vec_mule (vector signed int,
6457 vector unsigned long long vec_mule (vector unsigned int,
6458 vector unsigned int)
6459 vector signed long long vec_mulo (vector signed int,
6461 vector unsigned long long vec_mulo (vector unsigned int,
6462 vector unsigned int)
6463 * doc/extend.texi: Fix the documentation for the built-in
6466 2017-06-07 Carl Love <cel@us.ibm.com>
6469 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
6472 2017-06-07 Carl Love <cel@us.ibm.com>
6474 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
6475 support, Generate doublehv for signed int/float for BE case only.
6477 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
6479 * doc/invoke.texi (mcx16): Rewrite.
6481 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6483 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
6484 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
6485 *mov<mode>_softfloat, and an anonymous splitter): Use
6486 nonimmediate_operand instead of rs6000_nonimmediate_operand.
6488 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6490 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
6492 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
6493 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
6494 (rs6000_debug_reg_global): Adjust.
6495 (rs6000_init_hard_regno_mode_ok): Adjust.
6496 (rs6000_dbx_register_number): Adjust.
6497 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
6498 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
6499 Remove SPE_ACC and SPEFSCR.
6500 (REG_ALLOC_ORDER): Ditto.
6501 (FRAME_POINTER_REGNUM): Change to 111.
6502 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
6503 (REG_CLASS_NAMES): Ditto.
6504 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
6505 (REGISTER_NAMES): Ditto.
6506 (ADDITIONAL_REG_NAMES): Ditto.
6507 (rs6000_reg_names): Ditto.
6508 * config/rs6000/rs6000.md: Renumber some register number
6511 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6513 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
6515 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
6516 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
6518 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
6519 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
6520 Delete the SPE high registers.
6521 (REG_ALLOC_ORDER): Ditto.
6522 (enum reg_class): Remove SPE_HIGH_REGS.
6523 (REG_CLASS_NAMES): Ditto.
6524 (REG_CLASS_CONTENTS): Delete the SPE high registers.
6525 (REGISTER_NAMES): Ditto.
6526 (rs6000_reg_names): Ditto.
6527 * doc/tm.texi.in: Remove SPE as example.
6528 * doc/tm.texi: Regenerate.
6530 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6532 * config/rs6000/8540.md (ppc8540_brinc): Delete.
6533 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
6534 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
6535 * config/rs6000/rs6000.md (type): Remove "brinc".
6537 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6539 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
6540 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
6541 * config/rs6000/linuxspe.h: Delete file.
6542 * config/rs6000/rs6000.md: Don't include spe.md.
6543 * config/rs6000/spe.h: Delete file.
6544 * config/rs6000/spe.md: Delete file.
6545 * config/rs6000/t-rs6000: Remove spe.md.
6547 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6549 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
6550 (reg_or_none500mem_operand): Delete.
6551 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
6552 instead of reg_or_none500mem_operand.
6554 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6556 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
6557 handling of SPE flags.
6558 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
6560 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6562 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
6564 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
6565 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
6566 paired_divv2sf3): Similar.
6567 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
6568 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
6569 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
6571 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
6572 Rename the paired_* instruction patterns.
6573 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
6575 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
6576 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
6577 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
6578 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
6579 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
6581 (struct machine_function): Delete field spe_insn_chain_scanned_p.
6582 (spe_func_has_64bit_regs_p): Delete.
6583 (spe_expand_predicate_builtin): Delete.
6584 (spe_expand_evsel_builtin): Delete.
6585 (TARGET_DWARF_REGISTER_SPAN): Do not define.
6586 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
6587 (invalid_e500_subreg): Delete.
6588 (rs6000_legitimize_address): Always force_reg op2 as well, for
6589 paired single memory accesses.
6590 (rs6000_member_type_forces_blk): Delete.
6591 (rs6000_spe_function_arg): Delete.
6592 (rs6000_expand_unop_builtin): Delete SPE handling.
6593 (rs6000_expand_binop_builtin): Ditto.
6594 (spe_expand_stv_builtin): Delete.
6595 (bdesc_2arg_spe): Delete.
6596 (spe_expand_builtin): Delete.
6597 (spe_expand_predicate_builtin): Delete.
6598 (spe_expand_evsel_builtin): Delete.
6599 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
6600 (spe_init_builtins): Delete.
6601 (spe_func_has_64bit_regs_p): Delete.
6602 (savres_routine_name): Delete "info" parameter. Adjust callers.
6603 (rs6000_emit_stack_reset): Ditto.
6604 (rs6000_dwarf_register_span): Delete.
6605 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
6606 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
6607 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
6609 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
6611 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
6612 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
6613 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
6614 mulv2sf3, divv2sf3): Delete expanders.
6616 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6618 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
6620 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6622 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
6623 * config/rs6000/rs6000.c: Ditto.
6625 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6627 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
6628 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
6629 comparison_operator.
6631 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6633 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
6634 * config/rs6000/rs6000.opt: Ditto.
6635 * config/rs6000/t-rtems: Ditto.
6637 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6639 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
6640 TARGET_E500_SINGLE by 0, simplify.
6641 * config/rs6000/rs6000.c: Ditto.
6642 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
6643 (spe_build_register_parallel): Delete.
6644 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
6645 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
6646 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
6647 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
6648 (E500_CONVERT): Delete.
6649 * config/rs6000/spe.md: Remove many patterns and all define_constants.
6651 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
6653 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
6654 * config/rs6000/dfp.md: Ditto.
6655 (negdd2, *negdd2_fpr): Merge.
6656 (absdd2, *absdd2_fpr): Merge.
6657 (negtd2, *negtd2_fpr): Merge.
6658 (abstd2, *abstd2_fpr): Merge.
6659 * config/rs6000/e500.h: Delete file.
6660 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
6661 TARGET_FPRS by 1 and simplify.
6662 * config/rs6000/rs6000-c.c: Ditto.
6663 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
6665 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
6667 * config/rs6000/rs6000.md: Ditto.
6668 (floatdidf2, *floatdidf2_fpr): Merge.
6669 (move_from_CR_gt_bit): Delete.
6670 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
6671 (E500_CR_IOR_COMPARE): Delete.
6672 (All patterns that require !TARGET_FPRS): Delete.
6673 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
6675 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6677 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
6679 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6681 * graphds.c (add_edge): Intitialize edge's attached data.
6682 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
6683 pointer parameter. Call pointed function on each edge during
6684 graph traversing. Skip traversing the edge when the function
6686 (graphds_dfs, graphds_scc): Ditto.
6687 (for_each_edge): New parameter. Pass the new parameter to callback
6689 * graphds.h (skip_edge_callback): New function pointer type.
6690 (graphds_dfs, graphds_scc): New function pointer parameter.
6691 (graphds_edge_callback, for_each_edge): New parameter.
6693 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6695 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
6696 out code checking if runtime alias check is possible to below ...
6697 Call the new function.
6698 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
6699 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
6701 2017-06-07 Marek Polacek <polacek@redhat.com>
6704 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
6705 TYPE_OVERFLOW_WRAPS check.
6707 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6709 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
6710 if versioning is required.
6711 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
6712 peeling with the check for versioning.
6714 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6716 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
6717 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
6718 Set true to new parameter if new ssa variable is defined.
6719 (vect_gen_vector_loop_niters): Refactor. Set range information
6720 for the new vector loop bound variable.
6721 (vect_do_peeling): Ditto.
6723 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6725 * tree-affine.c (ssa.h): Include header file.
6726 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
6727 has wrapping overflow behavior.
6729 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6731 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
6733 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6735 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
6736 (tree_to_aff_combination): ... here.
6738 2017-06-07 Bin Cheng <bin.cheng@arm.com>
6740 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
6741 reg_pressure model function.
6742 (ivopts_global_cost_for_size): Delete.
6743 (determine_set_costs, iv_ca_recount_cost): Call new model function
6744 ivopts_estimate_reg_pressure.
6746 2017-06-07 Tamar Christina <tamar.christina@arm.com>
6748 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
6749 expensive than udiv. Remove floating point cases from mod.
6751 2017-06-07 Tamar Christina <tamar.christina@arm.com>
6753 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
6756 2017-06-07 Tamar Christina <tamar.christina@arm.com>
6758 * config/aarch64/aarch64.md
6759 (copysignsf3): Fix mask generation.
6761 2017-06-07 Jakub Jelinek <jakub@redhat.com>
6763 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
6765 (class dump_manager): Add register_dumps method.
6766 * dumpfile.c: Include langhooks.h.
6767 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
6768 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
6769 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
6770 (dump_manager::dump_register): Start with 512 entries instead of 32.
6771 (dump_manager::register_dumps): New method.
6772 * toplev.c (general_init): Instead of invoking register_dumps
6773 langhook, invoke register_dumps method on the dump manager.
6774 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
6777 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
6779 * doc/md.texi: Clarify the restrictions on a define_insn condition.
6780 Say that # requires an associated define_split to exist, and that
6781 the define_split must be suitable for use after register allocation.
6783 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6785 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
6786 (compute_outgoing_frequencies): Also initialize zero counts.
6787 (find_many_sub_basic_blocks): Do not produce uninitialized profile
6788 around loops; preserve more of profile when nothing changes.
6790 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
6792 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
6794 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
6795 * config/arm/arm-cpu-cdata.h: Regenerate.
6796 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
6797 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
6798 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
6799 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
6800 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
6802 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
6804 * config/arm/t-rmprofile: Likewise.
6805 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
6807 2017-06-06 David S. Miller <davem@davemloft.net>
6810 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
6811 blockage if function uses alloca.
6813 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
6815 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
6816 New "uid" fields to hold pretty-print IDs of group and ref.
6817 Memory references are now identified as <group_id>:<ref_id>
6818 instead of using [random] addresses.
6819 (dump_mem_details): Simplify, no functional change.
6820 (dump_mem_ref): Simplify and make output more concise.
6821 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
6822 (find_or_create_group): Initialize group uid.
6823 (record_ref): Initialize ref uid. Improve debug output.
6824 (prune_group_by_reuse, should_issue_prefetch_p,)
6825 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
6826 (mark_nontemporal_store, determine_loop_nest_reuse):
6827 Improve debug output.
6829 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
6831 * dbgcnt.def (prefetch): New debug counter.
6832 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
6833 (schedule_prefetches): Stop issueing prefetches if debug counter
6836 2017-06-06 Tom de Vries <tom@codesourcery.com>
6838 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
6839 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
6841 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6843 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
6844 Use aarch64_reg_or_zero predicate for operand 4.
6845 (aarch64_compare_and_swap<mode> define_insn_and_split):
6846 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
6847 (aarch64_store_exclusive<mode>): Likewise for operand 2.
6849 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
6851 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
6852 (arm_compute_save_core_reg_mask): This.
6853 (thumb1_compute_save_reg_mask): Rename into ...
6854 (thumb1_compute_save_core_reg_mask): This.
6855 (arm_compute_save_reg0_reg12_mask): Adapt comment.
6856 (arm_compute_frame_layout): Likewise.
6858 2017-06-06 Richard Biener <rguenther@suse.de>
6860 PR tree-optimization/80974
6861 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
6862 keep or clear leaders SSA info.
6864 2017-06-06 Tom de Vries <tom@codesourcery.com>
6866 * config/nvptx/nvptx.c (split_mode_p): New function.
6867 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
6869 2017-06-06 Tom de Vries <tom@codesourcery.com>
6871 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
6873 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6876 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
6879 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6881 * shrink-wrap.c (handle_simple_exit): Update profile.
6882 (try_shrink_wrapping): Upate profile.
6884 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6886 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
6887 (tree_guess_outgoing_edge_probabilities): New.
6888 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
6889 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
6891 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6893 * ipa-split.c (split_function): Initialize return bb profile.
6895 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
6897 * profile.c (compute_branch_probabilities): Also initialize
6900 2017-06-06 Richard Biener <rguenther@suse.de>
6902 PR tree-optimization/80928
6903 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
6904 (vect_analyze_loop_operations): Properly guard analysis for
6906 (vect_transform_loop): Likewise.
6907 (vect_analyze_loop_2): Also reset SLP type on PHIs.
6908 (vect_model_induction_cost): Do not cost for pure SLP.
6909 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
6910 of induction in inner loop vectorization.
6911 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
6912 (vect_get_and_check_slp_defs): Handle vect_induction_def.
6913 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
6915 (vect_analyze_slp_cost_1): Cost induction.
6916 (vect_detect_hybrid_slp_stmts): Handle PHIs.
6917 (vect_get_slp_vect_defs): Likewise.
6918 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
6919 (vect_transform_stmt): Handle SLP reductions.
6920 * tree-vectorizer.h (vectorizable_induction): Adjust.
6922 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
6924 * config/rs6000/rs6000.c (make_resolver_func): Update
6925 init_lowered_empty_function call.
6927 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
6929 * doc/invoke.texi: Document the -fprofile-abs-path option.
6930 * common.opt (fprofile-abs-path): New option.
6931 * gcov-io.h (gcov_write_filename): Declare.
6932 * gcov-io.c (gcov_write_filename): New function.
6933 * coverage.c (coverage_begin_function): Use gcov_write_filename.
6934 * profile.c (output_location): Likewise.
6936 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6938 * shring-wrap.c: Revert accidental commit.
6940 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
6942 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
6944 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6946 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
6948 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
6949 profile in callgraph edge.
6950 * profile-count.h (apply_probability): If THIS is 0, then result is 0
6951 (apply_scale): Likewise.
6952 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
6953 Also scale profile when inlining function with zero profile.
6954 (initialize_cfun): Update exit block profile even when it is zero.
6955 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
6956 when profile is read.
6958 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
6960 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
6961 (CLONE_*): New constants to define the processors we can generate
6962 code for with the target_clone attribute.
6963 (rs6000_clone_map): New array to identify which clone processors
6964 the current program is running on.
6965 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
6966 target_clone attribute.
6967 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
6968 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
6969 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
6970 (cpu_expand_builtin): Add support for target_clone attribute.
6971 (rs6000_valid_attribute_p): Allow "default" attribute.
6972 (get_decl_name): New debug function to simplify printing the
6973 current function name in debugging statements.
6974 (rs6000_clone_priority): New functions to support the target_clone
6975 attribute, and be able to generate code to switch between ISA 2.05
6976 through ISA 3.0 (power6 through power9).
6977 (rs6000_compare_version_priority): Likewise.
6978 (rs6000_get_function_versions_dispatcher): Likewise.
6979 (make_resolver_func): Likewise.
6980 (add_condition_to_bb): Likewise.
6981 (dispatch_function_versions): Likewise.
6982 (rs6000_generate_version_dispatcher_body): Likewise.
6983 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
6984 (fusion_gpr_load_p): Fix a spacing issue.
6985 * doc/extend.texi (Common Function Attributes): Document that the
6986 PowerPC supports the target_clone attribute.
6988 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
6990 * config/arm/arm.h: explain F symbol found in description of ARM
6991 register allocation in its legend.
6993 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
6995 * config/mips/frame-header-opt.c: Include profile-count.h.
6996 * config/riscv/riscv.c: Include profile-count.h
6998 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
7000 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
7002 (sm_set_flag_if_changed): Add bbs field.
7003 (execute_sm_if_changed_flag_set): Pass BBS.
7004 (execute_sm): Update.
7006 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7008 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
7011 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7013 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
7014 (peephole2): New peephole2 to emit the above.
7015 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
7017 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7019 * config/aarch64/aarch64.c (define_peephole2 above
7020 *sub_<shift>_<mode>): New peephole.
7022 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
7024 * config/i386/i386.c (make_resolver_func): Update.
7025 * Makefile.in: Add profile-count.h and profile-count.o
7026 * auto-profile.c (afdo_indirect_call): Update to new API.
7027 (afdo_set_bb_count): Update.
7028 (afdo_propagate_edge): Update.
7029 (afdo_propagate_circuit): Update.
7030 (afdo_calculate_branch_prob): Update.
7031 (afdo_annotate_cfg): Update.
7032 * basic-block.h: Include profile-count.h
7033 (struct edge_def): Turn count to profile_count.
7034 (struct basic_block_def): Likewie.
7035 (REG_BR_PROB_BASE): Move to profile-count.h
7036 (RDIV): Move to profile-count.h
7037 * bb-reorder.c (max_entry_count): Turn to profile_count.
7038 (find_traces): Update.
7039 (rotate_loop):Update.
7040 (connect_traces):Update.
7041 (sanitize_hot_paths):Update.
7042 * bt-load.c (migrate_btr_defs): Update.
7043 * cfg.c (RDIV): Remove.
7044 (init_flow): Use alloc_block.
7045 (alloc_block): Uninitialize count.
7046 (unchecked_make_edge): Uninitialize count.
7047 (check_bb_profile): Update.
7048 (dump_edge_info): Update.
7049 (dump_bb_info): Update.
7050 (update_bb_profile_for_threading): Update.
7051 (scale_bbs_frequencies_int): Update.
7052 (scale_bbs_frequencies_gcov_type): Update.
7053 (scale_bbs_frequencies_profile_count): New.
7054 * cfg.h (update_bb_profile_for_threading): Update.
7055 (scale_bbs_frequencies_profile_count): Declare.
7056 * cfgbuild.c (compute_outgoing_frequencies): Update.
7057 (find_many_sub_basic_blocks): Update.
7058 * cfgcleanup.c (try_forward_edges): Update.
7059 (try_crossjump_to_edge): Update.
7060 * cfgexpand.c (expand_gimple_tailcall): Update.
7061 (construct_exit_block): Update.
7062 * cfghooks.c (verify_flow_info): Update.
7063 (dump_bb_for_graph): Update.
7064 (split_edge): Update.
7065 (make_forwarder_block): Update.
7066 (duplicate_block): Update.
7067 (account_profile_record): Update.
7068 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
7069 (get_estimated_loop_iterations): Update.
7070 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
7071 (single_likely_exit): Update.
7072 * cfgloopmanip.c (scale_loop_profile): Update.
7074 (set_zero_probability): Update.
7075 (lv_adjust_loop_entry_edge): Update.
7076 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
7077 (purge_dead_edges): Update.
7078 (rtl_account_profile_record): Update.
7079 * cgraph.c (cgraph_node::create): Uninitialize count.
7080 (symbol_table::create_edge): Uninitialize count.
7081 (cgraph_update_edges_for_call_stmt_node): Update.
7082 (cgraph_edge::dump_edge_flags): Update.
7083 (cgraph_node::dump): Update.
7084 (cgraph_edge::maybe_hot_p): Update.
7085 * cgraph.h: Include profile-count.h
7086 (create_clone), create_edge, create_indirect_edge): Update.
7087 (cgraph_node): Turn count to profile_count.
7088 (cgraph_edge0: Likewise.
7089 (make_speculative, clone): Update.
7090 (create_edge): Update.
7091 (init_lowered_empty_function): Update.
7092 * cgraphclones.c (cgraph_edge::clone): Update.
7093 (duplicate_thunk_for_node): Update.
7094 (cgraph_node::create_clone): Update.
7095 * cgraphunit.c (cgraph_node::analyze): Update.
7096 (cgraph_node::expand_thunk): Update.
7097 * final.c (dump_basic_block_info): Update.
7098 * gimple-streamer-in.c (input_bb): Update.
7099 * gimple-streamer-out.c (output_bb): Update.
7100 * graphite.c (print_global_statistics): Update.
7101 (print_graphite_scop_statistics): Update.
7102 * hsa-brig.c: Include basic-block.h.
7103 * hsa-dump.c: Include basic-block.h.
7104 * hsa-gen.c (T sum_slice): Update.
7105 (convert_switch_statements):Update.
7106 * hsa-regalloc.c: Include basic-block.h.
7107 * ipa-chkp.c (chkp_produce_thunks): Update.
7108 * ipa-cp.c (struct caller_statistics): Update.
7109 (init_caller_stats): Update.
7110 (gather_caller_stats): Update.
7111 (ipcp_cloning_candidate_p): Update.
7112 (good_cloning_opportunity_p): Update.
7113 (get_info_about_necessary_edges): Update.
7114 (dump_profile_updates): Update.
7115 (update_profiling_info): Update.
7116 (update_specialized_profile): Update.
7117 (perhaps_add_new_callers): Update.
7118 (decide_about_value): Update.
7119 (ipa_cp_c_finalize): Update.
7120 * ipa-devirt.c (struct odr_type_warn_count): Update.
7121 (struct decl_warn_count): Update.
7122 (struct final_warning_record): Update.
7123 (possible_polymorphic_call_targets): Update.
7124 (ipa_devirt): Update.
7125 * ipa-fnsummary.c (redirect_to_unreachable): Update.
7126 * ipa-icf.c (sem_function::merge): Update.
7127 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
7128 * ipa-inline.c (compute_uninlined_call_time): Update.
7129 (compute_inlined_call_time): Update.
7130 (want_inline_small_function_p): Update.
7131 (want_inline_self_recursive_call_p): Update.
7132 (edge_badness): Update.
7133 (lookup_recursive_calls): Update.
7134 (recursive_inlining): Update.
7135 (inline_small_functions): Update.
7136 (dump_overall_stats): Update.
7137 (dump_inline_stats): Update.
7138 * ipa-profile.c (ipa_profile_generate_summary): Update.
7139 (ipa_propagate_frequency): Update.
7140 (ipa_profile): Update.
7141 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
7142 * ipa-utils.c (ipa_merge_profiles): Update.
7143 * loop-doloop.c (doloop_modify): Update.
7144 * loop-unroll.c (report_unroll): Update.
7145 (unroll_loop_runtime_iterations): Update.
7146 * lto-cgraph.c (lto_output_edge): Update.
7147 (lto_output_node): Update.
7148 (input_node): Update.
7149 (input_edge): Update.
7150 (merge_profile_summaries): Update.
7151 * lto-streamer-in.c (input_cfg): Update.
7152 * lto-streamer-out.c (output_cfg): Update.
7153 * mcf.c (create_fixup_graph): Update.
7154 (adjust_cfg_counts): Update.
7155 (sum_edge_counts): Update.
7156 * modulo-sched.c (sms_schedule): Update.
7157 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
7158 * predict.c (maybe_hot_count_p): Update.
7159 (probably_never_executed): Update.
7160 (dump_prediction): Update.
7161 (combine_predictions_for_bb): Update.
7162 (propagate_freq): Update.
7163 (handle_missing_profiles): Update.
7164 (counts_to_freqs): Update.
7165 (rebuild_frequencies): Update.
7166 (force_edge_cold): Update.
7167 * predict.h: Include profile-count.h
7168 (maybe_hot_count_p, counts_to_freqs): UPdate.
7169 * print-rtl-function.c: Do not include cfg.h
7170 * print-rtl.c: Include basic-block.h
7171 * profile-count.c: New file.
7172 * profile-count.h: New file.
7173 * profile.c (is_edge_inconsistent): Update.
7174 (correct_negative_edge_counts): Update.
7175 (is_inconsistent): Update.
7176 (set_bb_counts): Update.
7177 (read_profile_edge_counts): Update.
7178 (compute_frequency_overlap): Update.
7179 (compute_branch_probabilities): Update; Initialize and deinitialize
7181 (branch_prob): Update.
7182 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
7183 (edge_gcov_count): New.
7184 (bb_gcov_count): New.
7185 * shrink-wrap.c (try_shrink_wrapping): Update.
7186 * tracer.c (better_p): Update.
7187 * trans-mem.c (expand_transaction): Update.
7188 (ipa_tm_insert_irr_call): Update.
7189 (ipa_tm_insert_gettmclone_call): Update.
7190 * tree-call-cdce.c: Update.
7191 * tree-cfg.c (gimple_duplicate_sese_region): Update.
7192 (gimple_duplicate_sese_tail): Update.
7193 (gimple_account_profile_record): Update.
7194 (execute_fixup_cfg): Update.
7195 * tree-inline.c (copy_bb): Update.
7196 (copy_edges_for_bb): Update.
7197 (initialize_cfun): Update.
7198 (freqs_to_counts): Update.
7199 (copy_cfg_body): Update.
7200 (expand_call_inline): Update.
7201 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
7202 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
7203 (try_unroll_loop_completely): Update.
7204 (try_peel_loop): Update.
7205 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
7206 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
7207 * tree-ssa-loop-split.c (connect_loops): Update.
7208 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
7209 * tree-ssa-reassoc.c (branch_fixup): Update.
7210 * tree-ssa-tail-merge.c (replace_block_by): Update.
7211 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
7212 (compute_path_counts): Update.
7213 (update_profile): Update.
7214 (recompute_probabilities): Update.
7215 (update_joiner_offpath_counts): Update.
7216 (estimated_freqs_path): Update.
7217 (freqs_to_counts_path): Update.
7218 (clear_counts_path): Update.
7219 (ssa_fix_duplicate_block_edges): Update.
7220 (duplicate_thread_path): Update.
7221 * tree-switch-conversion.c (case_bit_test_cmp): Update.
7222 (struct switch_conv_info): Update.
7223 * tree-tailcall.c (decrease_profile): Update.
7224 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
7225 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
7226 * value-prof.c (check_counter): Update.
7227 (gimple_divmod_fixed_value): Update.
7228 (gimple_mod_pow2): Update.
7229 (gimple_mod_subtract): Update.
7230 (gimple_ic_transform): Update.
7231 (gimple_stringop_fixed_value): Update.
7232 * value-prof.h (gimple_ic): Update.
7234 2017-06-02 Carl Love <cel@us.ibm.com>
7236 * config/rs6000/rs6000-c: Add support for built-in functions
7237 vector double vec_doublee (vector signed int);
7238 vector double vec_doublee (vector unsigned int);
7239 vector double vec_doublee (vector float);
7240 vector double vec_doubleh (vector signed int);
7241 vector double vec_doubleh (vector unsigned int);
7242 vector double vec_doubleh (vector float);
7243 vector double vec_doublel (vector signed int);
7244 vector double vec_doublel (vector unsigned int);
7245 vector double vec_doublel (vector float);
7246 vector double vec_doubleo (vector signed int);
7247 vector double vec_doubleo (vector unsigned int);
7248 vector double vec_doubleo (vector float);.
7249 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
7250 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
7252 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
7253 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
7254 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
7256 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
7257 vec_doublel, vec_doubleh.
7258 * doc/extend.texi: Update the built-in documentation file for the
7259 new built-in functions.
7261 2017-06-02 David Malcolm <dmalcolm@redhat.com>
7264 * ipa-inline-analysis.c (free_growth_caches): Set
7265 edge_removal_hook_holder to NULL after removing it.
7267 2017-06-02 Sudakshina Das <sudi.das@arm.com>
7269 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
7270 comparision with zero.
7272 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
7273 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
7274 for early expansion of vec_min and vec_max builtins.
7275 (builtin_function_type): Add min/max unsigned variants to those
7276 identified as having unsigned arguments.
7278 2017-06-02 Olivier Hainque <hainque@adacore.com>
7280 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
7282 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7284 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
7285 Use VALL_F16 iterator rather than VALL.
7287 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7289 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
7290 Emit CBNZ inside loop when doing a strong exchange and comparing
7291 against zero. Generate the CC flags after the loop.
7293 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
7295 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
7296 (dl_section_ref): New.
7297 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
7298 On AIX, append an expression to subtract the size of the
7299 section length to dl_section_ref.
7301 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
7303 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
7304 for early expansion of vector absolute builtins.
7306 2017-06-02 Richard Biener <rguenther@suse.de>
7308 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
7309 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
7311 2017-06-02 Richard Biener <rguenther@suse.de>
7313 PR tree-optimization/80948
7314 * tree-tailcall.c (find_tail_calls): Track stmts to move in
7317 2017-06-02 Richard Biener <rguenther@suse.de>
7319 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
7321 * tree-vect-stmts.c (process_use): Do not mark backedge defs
7322 for inductions as relevant.
7324 2017-06-02 Richard Biener <rguenther@suse.de>
7326 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
7327 (vectorizable_induction): ... this. Remove dead code.
7329 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
7331 * builtins. (expand_builtin_alloca): Remove second parameter and
7332 infer its value from the first parameter instead.
7333 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
7335 2017-06-02 Jakub Jelinek <jakub@redhat.com>
7337 PR rtl-optimization/80903
7338 * loop-doloop.c (add_test): Unshare sequence.
7340 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
7342 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
7344 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
7346 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
7348 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
7349 xlogue_layout::get_instance, logue_layout::xlogue_layout,
7350 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
7351 (xlogue_layout::get_stub_rtx): Make static.
7352 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
7353 (xlogue_layout::compute_stub_managed_regs): Rename to...
7354 (xlogue_layout::count_stub_managed_regs): ...this.
7355 (xlogue_layout::is_stub_managed_reg): New function.
7356 (xlogue_layout::m_stub_names): Rename to...
7357 (xlogue_layout::s_stub_names): ...this, make static.
7358 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
7359 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
7360 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
7361 xlogue_layout::s_stub_names): Instantiate statics.
7362 (stub_managed_regs): Remove.
7363 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
7364 (disable_call_ms2sysv_xlogues): Rename to...
7365 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
7366 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
7368 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
7369 change after reload_completed.
7370 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
7372 (ix86_expand_prologue): Likewise.
7373 (ix86_expand_epilogue): Likewise.
7374 (ix86_expand_split_stack_prologue): Likewise.
7375 (ix86_compute_frame_layout): Remove frame parameter ...
7376 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
7377 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
7379 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
7380 (ix86_frame): Move from here ...
7381 * config/i386/i386.h (ix86_frame): ... to here.
7382 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
7383 complete ix86_frame data structure instead. Remove some_ld_name.
7385 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
7387 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
7388 symbols that hold a DECL_VALUE_EXPR.
7390 2017-06-01 Martin Jambor <mjambor@suse.cz>
7392 PR tree-optimization/80898
7393 * tree-sra.c (process_subtree_disqualification): Removed.
7394 (disqualify_candidate): Do not acll
7395 process_subtree_disqualification.
7396 (subtree_mark_written_and_enqueue): New function.
7397 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
7398 RHS has been disqualified and re-queue LHS if necessary. Apart
7399 from that, ignore disqualified RHS.
7401 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7403 * config/s390/s390.c (s390_emit_epilogue): Disable early return
7404 address fetch for z10 or later.
7406 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7408 * config/arc/arc.md (tst_movb): Add guard when splitting.
7410 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7412 * config/arc/arc.c (arc_can_eliminate): Test against
7413 arc_frame_pointer_needed.
7415 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7417 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
7418 to prevent store reordering.
7419 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
7420 (type): Add block type.
7421 (stack_tie): Define special instruction to be used in
7424 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7426 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
7427 constraint. It is not valid for the pattern.
7428 (noncommutative_binary_comparison): Likewise.
7430 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7432 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
7435 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7437 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
7438 be used by the reg-alloc.
7440 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7442 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
7443 reg-alloc when having mul64 or mul32x16 instructions.
7444 (mulsidi3): Likewise.
7445 (umulsidi3): Likewise.
7446 (mulsi32x16): New pattern.
7447 (mulsi64): Likewise.
7448 (mulsidi64): Likewise.
7449 (umulsidi64): Likewise.
7450 (MUL32x16_REG): Define.
7451 (mul64_600): Use MUL32x16_REG.
7452 (mac64_600): Likewise.
7453 (umul64_600): Likewise.
7454 (umac64_600): Likewise.
7456 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
7458 * config/arc/arc.md (mulsi3_700): Make it commutative.
7460 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
7462 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
7464 (*sign_extendsidi2_insn): Likewise for movstosw.
7466 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
7468 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
7469 the type of the input discriminant value. Convert the
7470 discriminant value of signedness vary.
7472 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
7474 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
7475 Add to -Wall section.
7477 2017-06-01 Richard Biener <rguenther@suse.de>
7480 * fold-const.c (fold_plusminus_mult_expr): If the factored
7481 factor may be zero use a wrapping type for the inner operation.
7482 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
7483 and handle moved defs.
7484 (process_assignment): Properly guard the unary op case. Return a
7485 tri-state indicating that moving the stmt before the call may allow
7486 to continue. Pass through to_move.
7487 (find_tail_calls): Handle moving unrelated defs before
7490 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
7493 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
7494 splitter result in the canonical way.
7496 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
7498 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
7499 also for 32bit target. Update insn attributes.
7500 (zero-extendsidi2 splitter): Allow all registers for operand 1.
7502 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
7504 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
7505 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
7506 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
7507 (_mm_maskz_min_ss): New intrinsics.
7509 2017-05-31 Martin Liska <mliska@suse.cz>
7511 * tree-vect-loop.c (vect_create_epilog_for_reduction):
7512 Change comment style to one we normally use.
7513 (vectorizable_reduction): Likewise.
7514 (vectorizable_induction): Likewise.
7515 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
7516 (vectorizable_call): Likewise.
7517 (vectorizable_simd_clone_call): Likewise.
7518 (vectorizable_conversion): Likewise.
7519 (vectorizable_assignment): Likewise.
7520 (vectorizable_shift): Likewise.
7521 (vectorizable_operation): Likewise.
7522 (vectorizable_store): Likewise.
7523 (vectorizable_load): Likewise.
7524 * tree-vectorizer.h: Likewise.
7526 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
7528 * passes.c (emergency_dump_function): New.
7529 * tree-pass.h (emergency_dump_function): Declare.
7530 * plugin.c (plugins_internal_error_function): Remove.
7531 * plugin.h (plugins_internal_error_function): Remove declaration.
7532 * toplev.c (internal_error_function): New static function. Use it...
7533 (general_init): ...here.
7535 2017-05-31 Graham Markall <graham.markall@embecosm.com>
7537 * config/arc/arc.c (arc_print_operand): Handle constant operands.
7538 (arc_rtx_costs): Add costs for new patterns.
7539 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
7540 * config/arc/predicates.md: Add _1_2_3_operand predicate.
7542 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
7544 * tree-ssa-strlen.c (get_next_strinfo): New function.
7545 (get_stridx_plus_constant): Use it.
7546 (zero_length_string): Likewise.
7547 (adjust_related_strinfos): Likewise.
7548 (adjust_last_stmt): Likewise.
7550 2017-05-31 Richard Biener <rguenther@suse.de>
7553 * config/i386/i386.c (ix86_expand_builtin): Remove assert
7554 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
7556 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
7558 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
7559 loop_vinfo argument and use of dependence distance vectors.
7560 Check instead whether the two references differ only in their
7561 initial value and assume that they have the same alignment if the
7562 difference is a multiple of the vector alignment.
7563 (vect_analyze_data_refs_alignment): Update call accordingly.
7565 2017-05-31 Martin Liska <mliska@suse.cz>
7568 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
7570 2017-05-31 Bin Cheng <bin.cheng@arm.com>
7572 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
7573 (create_intersect_range_checks): Move from ...
7574 * tree-data-ref.c (create_intersect_range_checks_index)
7575 (create_intersect_range_checks): ... to here.
7576 (create_runtime_alias_checks): New function factored from ...
7577 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
7578 here. Call above function.
7579 * tree-data-ref.h (create_runtime_alias_checks): New function.
7581 2017-05-31 Bin Cheng <bin.cheng@arm.com>
7583 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
7584 segment length for dr_b and compute it in wide_int.
7586 2017-05-31 Richard Biener <rguenther@suse.de>
7588 PR tree-optimization/80906
7589 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
7590 and pass through iv_map.
7591 (copy_bb_and_scalar_dependences): Adjust.
7592 (translate_pending_phi_nodes): Likewise.
7593 (copy_loop_close_phi_args): Handle code-generating IVs instead
7596 2017-05-30 David Malcolm <dmalcolm@redhat.com>
7598 * diagnostic-color.c (color_dict): Add "type-diff".
7599 (parse_gcc_colors): Update comment.
7600 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
7601 -fdiagnostics-show-template-tree and -fno-elide-type.
7602 (GCC_COLORS): Add type-diff to example.
7604 (-fdiagnostics-show-template-tree): New.
7605 (-fno-elide-type): New.
7606 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
7607 the pp_format_decoder callback. Call any m_format_postprocessor's
7609 (pretty_printer::pretty_printer): Initialize
7610 m_format_postprocessor.
7611 (pretty_printer::~pretty_printer): Delete any
7612 m_format_postprocessor.
7613 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
7614 (class format_postprocessor): New class.
7615 (struct pretty_printer::format_decoder): Document the new parameters.
7616 (struct pretty_printer::m_format_postprocessor): New field.
7617 * tree-diagnostic.c (default_tree_printer): Update for new
7618 bool and const char ** params.
7619 * tree-diagnostic.h (default_tree_printer): Likewise.
7621 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
7623 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
7624 (lwa_operand): Delete rs6000_gen_cell_microcode test.
7625 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
7626 rs6000_gen_cell_microcode code.
7627 (rs6000_final_prescan_insn): Delete.
7628 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
7629 "warn-cell-microcode" entries.
7630 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
7631 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
7632 throughout. Change cc_reg_not_micro_cr0_operand to
7633 cc_reg_not_cr0_operand throughout.
7634 (*extendhi<mode>2_noload): Delete.
7635 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
7636 (mwarn-cell-microcode): Delete.
7637 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
7638 -mgen-cell-microcode and -mwarn-cell-microcode.
7640 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
7643 * config/i386/constraints.md (Yd): New constraint.
7645 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
7646 and (?Yd, r) alternatives. Update insn attributes.
7647 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
7648 and (?*Yd, r) alternatives. Update insn attributes.
7649 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
7651 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
7653 * gimplify.c (gimplify_modify_expr): Don't create a
7654 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
7657 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
7659 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
7661 2017-05-30 Richard Biener <rguenther@suse.de>
7663 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
7664 and reduc_def fields.
7665 (STMT_VINFO_REDUC_TYPE): New define.
7666 (STMT_VINFO_REDUC_DEF): Likewise.
7667 (vect_force_simple_reduction): Adjust prototype.
7668 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
7669 (vect_is_simple_reduction): Remove check_reduction argument.
7670 (vect_force_simple_reduction): Adjust and set
7671 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
7672 (vectorizable_reduction): Do not re-do reduction analysis
7673 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
7674 * tree-parloops.c (gather_scalar_reductions): Adjust.
7676 2017-05-30 Richard Biener <rguenther@suse.de>
7679 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
7682 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
7684 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
7685 Introduce unknown_misalignment parameter and remove vf.
7686 (vect_peeling_hash_get_lowest_cost):
7687 Pass unknown_misalignment parameter.
7688 (vect_enhance_data_refs_alignment):
7689 Fix unsupportable data ref treatment.
7691 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
7693 * tree-vect-data-refs.c (vect_get_data_access_cost):
7694 Workaround for SLP handling.
7695 (vect_enhance_data_refs_alignment):
7696 Compute costs for doing no peeling at all, compare to the best
7697 peeling costs so far and avoid peeling if cheaper.
7699 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
7701 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
7702 Return peeling info and set costs to zero for unlimited cost
7704 (vect_enhance_data_refs_alignment): Also inspect all datarefs
7705 with unknown misalignment. Compute and costs for unknown
7706 misalignment, compare them to the costs for known misalignment
7707 and choose the cheapest for peeling.
7709 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
7711 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
7712 (vect_get_peeling_costs_all_drs): Create function.
7713 (vect_peeling_hash_get_lowest_cost):
7714 Use vect_get_peeling_costs_all_drs.
7715 (vect_peeling_supportable): Create function.
7716 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
7718 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
7720 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
7721 DR_HAS_NEGATIVE_STEP.
7722 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
7723 (vect_enhance_data_refs_alignment): Use.
7724 (vect_duplicate_ssa_name_ptr_info): Use.
7725 * tree-vectorizer.h (dr_misalignment): Use.
7726 (known_alignment_for_access_p): Use.
7728 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
7731 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
7733 (has_section_name): New function.
7735 2017-05-30 Martin Liska <mliska@suse.cz>
7738 * auto-profile.c (get_function_decl_from_block): Fix
7741 2017-05-30 Richard Biener <rguenther@suse.de>
7744 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
7746 2017-05-30 Martin Liska <mliska@suse.cz>
7748 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
7749 * dumpfile.h (struct dump_file_info): Remove ctors.
7751 2017-05-30 Martin Liska <mliska@suse.cz>
7753 * predict.def: Fix GNU coding style.
7755 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
7757 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
7758 Mark 'to' argument with ATTRIBUTE_UNUSED.
7760 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
7762 * config/xtensa/xtensa.c (xtensa_emit_call): Use
7763 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
7764 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
7767 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
7769 * doc/install.texi (Options specification): Restore entry of
7770 --enable-sjlj-exceptions.
7772 2017-05-27 Michael Eager <eager@eagercon.com>
7775 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
7777 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
7779 * config/microblaze/microblaze.h
7780 (FIXED_REGISTERS): Update in macro.
7781 (CALL_USED_REGISTERS): Update in macro.
7783 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
7785 * doc/install.texi: Add links to macOS binary distributions.
7787 2017-05-27 Jakub Jelinek <jakub@redhat.com>
7791 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
7793 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
7795 2017-05-26 Martin Liska <mliska@suse.cz>
7797 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
7799 2017-05-26 Martin Liska <mliska@suse.cz>
7801 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
7802 always leading ';; '.
7803 (dump_bb_info): Likewise.
7804 (brief_dump_cfg): Likewise.
7805 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
7806 * dumpfile.c: Remove usage of TDF_VERBOSE.
7807 * dumpfile.h (enum dump_kind): Likewise.
7808 (dump_gimple_bb_header): Do not use TDF_COMMENT.
7809 * print-tree.c (debug_verbose): Remove.
7810 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
7811 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
7812 * tree-diagnostic.c (default_tree_printer): Replace
7813 TDF_DIAGNOSTIC with TDF_SLIM.
7815 2017-05-26 Bin Cheng <bin.cheng@arm.com>
7817 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
7818 in parameter loop, rather than loop_vinfo.
7819 (create_intersect_range_checks): Ditto.
7820 (vect_create_cond_for_alias_checks): Update call to above functions.
7822 2017-05-26 Bin Cheng <bin.cheng@arm.com>
7824 PR tree-optimization/80815
7825 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
7826 for merging runtime alias checks. Handle negative DR_STEPs.
7828 2017-05-26 Bin Cheng <bin.cheng@arm.com>
7830 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
7832 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
7833 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
7834 out code pruning runtime alias checks.
7835 * tree-data-ref.c (prune_runtime_alias_test_list): New function
7836 factored out from above.
7837 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
7839 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
7841 (prune_runtime_alias_test_list): New decalaration.
7843 2017-05-26 Bin Cheng <bin.cheng@arm.com>
7845 * tree-vect-data-refs.c (compare_tree): Rename and move ...
7846 * tree-data-ref.c (data_ref_compare_tree): ... to here.
7847 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
7848 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
7849 (operator==, comp_dr_with_seg_len_pair): Ditto.
7850 (vect_prune_runtime_alias_test_list): Ditto.
7852 2017-05-26 Martin Liska <mliska@suse.cz>
7855 * params.def: Bound partial-inlining-entry-probability param.
7857 2017-05-26 Marek Polacek <polacek@redhat.com>
7860 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
7863 2017-05-26 Richard Biener <rguenther@suse.de>
7865 PR tree-optimization/80842
7866 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
7869 2017-05-26 Richard Biener <rguenther@suse.de>
7871 PR tree-optimization/80844
7872 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
7874 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
7876 * doc/md.texi (Machine Constraints): Update x86 family
7877 machine constraints section to match 'config/i386/constraints.md'.
7879 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
7881 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
7883 2017-05-25 Nathan Sidwell <nathan@acm.org>
7885 * doc/invoke.texi (--enable-languages): Update documentation.
7887 2017-05-25 Martin Liska <mliska@suse.cz>
7889 * dumpfile.c: Add TDF_FOLDING.
7890 * dumpfile.h (enum dump_kind): Likewise.
7891 * genmatch.c (dt_simplify::gen_1): Use it.
7893 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
7895 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
7897 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
7899 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
7900 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
7902 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
7904 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
7905 * match.pd (X == C): Rewrite it here.
7906 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
7907 with_certain_nonzero_bits2): New predicates.
7908 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
7910 2017-05-24 Nathan Sidwell <nathan@acm.org>
7912 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
7915 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
7918 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
7920 * config/powerpcspe: New port. Files are copied from the rs6000
7921 port, with "rs6000" in filenames replaced by "powerpcspe".
7923 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
7925 PR rtl-optimization/80754
7926 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
7928 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
7930 * config/sparc/sparc.md (length): Return the correct value for -mflat
7931 sibcalls to match output_sibcall.
7933 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
7937 * config/rs6000/rs6000.c (struct machine_function): Add new field
7939 (rs6000_get_separate_components): Init that field, use it.
7940 (rs6000_components_for_bb): Use the field.
7942 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
7944 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
7946 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
7949 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
7951 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7954 * config/s390/s390.c (s390_check_qrst_address): Check incoming
7955 address against address_operand predicate.
7956 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
7958 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
7960 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
7961 parameters passed indirectly.
7963 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
7965 * config/i386/i386.md (*movdi_internal): Remove SSE4
7966 alternative 18 (?r, *v). Update insn attributes.
7967 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
7968 Update insn attributes.
7969 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
7970 Update insn attributes.
7971 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
7972 alternative 1 (r, v). Remove isa attribute.
7973 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
7974 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
7975 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
7977 2017-05-23 Tom de Vries <tom@codesourcery.com>
7979 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
7982 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
7984 * cgraphunit.c (symbol_table::process_new_functions): Update.
7985 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
7986 (inline_generate_summary): Rename to ...
7987 (ipa_fn_summary_generate): ... this one.
7988 (inline_read_summary): Rename to ...
7989 (ipa_fn_summary_read): ... this one.
7990 (inline_write_summary): Rename to ...
7991 (ipa_fn_summary_write): ... this one.
7992 (inline_free_summary): Rename to ...
7993 (ipa_free_fn_summary): ... this one.
7994 (pass_data_local_fn_summary, pass_local_fn_summary,
7995 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
7996 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
7997 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
7998 make_pass_ipa_fn_summary): New.
7999 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
8000 inline_write_summary, inline_free_summary): Remove.
8001 (ipa_free_fn_summary) : New.
8002 * ipa-inline.c (ipa_inline): Update.
8003 (pass_ipa_inline): Do not generate summaries.
8004 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
8006 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
8007 and add pass_ipa_fn_summary.
8008 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
8010 (make_pass_inline_parameters): Remove.
8012 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
8014 * omp-low.c (struct omp_context): Remove "default_kind" member.
8017 * omp-offload.c (execute_oacc_device_lower): Remove the
8018 parallelism dimensions function attributes for unparallelized
8019 OpenACC kernels constructs.
8021 2017-05-23 Martin Liska <mliska@suse.cz>
8023 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
8025 (cgraph_edge::make_speculative): Likewise.
8026 (cgraph_edge::resolve_speculation): Likewise.
8027 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
8028 (cgraph_node::dump): Likewise.
8029 * cgraph.h: Likewise.
8030 * cgraphunit.c (analyze_functions): Likewise.
8031 (symbol_table::compile): Likewise.
8032 * ipa-cp.c (print_all_lattices): Likewise.
8033 (determine_versionability): Likewise.
8034 (initialize_node_lattices): Likewise.
8035 (ipcp_verify_propagated_values): Likewise.
8036 (estimate_local_effects): Likewise.
8037 (update_profiling_info): Likewise.
8038 (create_specialized_node): Likewise.
8039 (perhaps_add_new_callers): Likewise.
8040 (decide_about_value): Likewise.
8041 (decide_whether_version_node): Likewise.
8042 (identify_dead_nodes): Likewise.
8043 (ipcp_store_bits_results): Likewise.
8044 * ipa-devirt.c (dump_targets): Likewise.
8045 (ipa_devirt): Likewise.
8046 * ipa-icf.c (sem_item::dump): Likewise.
8047 (sem_function::equals): Likewise.
8048 (sem_variable::equals): Likewise.
8049 (sem_item_optimizer::read_section): Likewise.
8050 (sem_item_optimizer::execute): Likewise.
8051 (congruence_class::dump): Likewise.
8052 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
8053 (dump_inline_summary): Likewise.
8054 (estimate_node_size_and_time): Likewise.
8055 (inline_analyze_function): Likewise.
8056 * ipa-inline-transform.c (inline_call): Likewise.
8057 * ipa-inline.c (report_inline_failed_reason): Likewise.
8058 (want_early_inline_function_p): Likewise.
8059 (edge_badness): Likewise.
8060 (update_edge_key): Likewise.
8061 (inline_small_functions): Likewise.
8062 * ipa-profile.c (ipa_profile): Likewise.
8063 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
8064 (ipa_make_edge_direct_to_target): Likewise.
8065 (remove_described_reference): Likewise.
8066 (ipa_impossible_devirt_target): Likewise.
8067 (propagate_controlled_uses): Likewise.
8068 (ipa_print_node_params): Likewise.
8069 (ipcp_transform_function): Likewise.
8070 * ipa-pure-const.c (pure_const_read_summary): Likewise.
8071 (propagate_pure_const): Likewise.
8072 * ipa-reference.c (generate_summary): Likewise.
8073 (read_write_all_from_decl): Likewise.
8074 (propagate): Likewise.
8075 (ipa_reference_read_optimization_summary): Likewise.
8076 * ipa-utils.c (ipa_merge_profiles): Likewise.
8077 * ipa.c (walk_polymorphic_call_targets): Likewise.
8078 (symbol_table::remove_unreachable_nodes): Likewise.
8079 (ipa_single_use): Likewise.
8080 * passes.c (execute_todo): Likewise.
8081 * predict.c (drop_profile): Likewise.
8082 * symtab.c (symtab_node::get_dump_name): New function.
8083 (symtab_node::dump_name): Likewise.
8084 (symtab_node::dump_asm_name): Likewise.
8085 (symtab_node::dump_references): Likewise.
8086 (symtab_node::dump_referring): Likewise.
8087 (symtab_node::dump_base): Likewise.
8088 (symtab_node::debug_symtab): Likewise.
8089 * tree-sra.c (convert_callers_for_node): Likewise.
8090 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
8091 * value-prof.c (init_node_map): Likewise.
8093 2017-05-23 Martin Liska <mliska@suse.cz>
8095 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
8096 and symtab_node::debug_symtab to symbol_table::debug.
8097 * cgraphunit.c (analyze_functions): Use the renamed function.
8098 (symbol_table::compile): Likewise.
8099 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
8100 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
8101 * passes.c (execute_todo): Likewise.
8102 * symtab.c (symbol_table::dump): New function.
8103 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
8105 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
8107 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
8108 that nonconst implies exec.
8110 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
8112 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
8113 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
8114 (inline_edge_summary_vec): Turn into ...
8115 (ipa_call_summaries): ... this one.
8116 (redirect_to_unreachable, edge_set_predicate,
8117 evaluate_properties_for_edge, inline_summary_alloc,
8118 reset_ipa_call_summary, reset_inline_summary,
8119 inline_summary_t::duplicate): Update.
8120 (inline_edge_duplication_hook): Turn to ...
8121 (ipa_call_summary_t::duplicate): ... this one.
8122 (inline_edge_removal_hook): Turn to ...
8123 (ipa_call_summary_t::remove): ... this one.
8124 (dump_inline_edge_summary): Turn to ...
8125 (dump_ipa_call_summary): ... this one.
8126 (estimate_function_body_sizes): Update.
8127 (inline_update_callee_summaries): Update.
8128 (remap_edge_change_prob): Update.
8129 (remap_edge_summaries): Update.
8130 (inline_merge_summary): Update.
8131 (do_estimate_edge_time): Update.
8132 (inline_generate_summary): Update.
8133 (inline_read_section): Update.
8134 (inline_read_summary): Update.
8135 (inline_free_summary): Update.
8136 * ipa-inline.c (can_inline_edge_p): Update.
8137 (compute_inlined_call_time): Update.
8138 (want_inline_small_function_p): Update.
8139 (edge_badness): Update.
8140 (early_inliner): Update.
8141 * ipa-inline.h (inline_edge_summary): Turn to ...
8142 (ipa_call_summary): ... this one.
8143 (ipa_call_summary_t): New class.
8144 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
8145 (ipa_call_summaries): New.
8146 (inline_edge_summary): Remove.
8147 (estimate_edge_growth): Update.
8148 * ipa-profile.c (ipa_propagate_frequency_1): Update.
8149 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
8150 * ipa-split.c (execute_split_functions): Update.
8151 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
8153 2017-05-23 Tom de Vries <tom@codesourcery.com>
8155 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
8156 attributes): Document rdrand effective target.
8158 2017-05-23 Tom de Vries <tom@codesourcery.com>
8160 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
8161 attributes): Sort alphabetically.
8163 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
8165 * config/avr/genmultilib.awk: Use gsub instead of gensub.
8167 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
8170 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
8171 V2DF/V2DI splat into two separate patterns, one that handles
8172 registers, and the other that only handles memory. Drop support
8173 for splatting from a GPR on ISA 2.07 and then splitting the
8174 splat into direct move and splat.
8175 (vsx_splat_<mode>_reg): Likewise.
8176 (vsx_splat_<mode>_mem): Likewise.
8178 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
8180 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
8182 2017-05-22 Jakub Jelinek <jakub@redhat.com>
8185 * omp-low.c (finish_taskreg_remap): New function.
8186 (finish_taskreg_scan): If unit size of ctx->record_type
8187 is non-constant, unshare the size expression and replace
8188 decls in it with possible outer var refs.
8191 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
8192 GOVD_SHARED rather than GOVD_PRIVATE with it.
8193 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
8194 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
8197 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
8198 as last argument to build_outer_var_ref for pointer bases of array
8201 2017-05-19 Martin Sebor <msebor@redhat.com>
8203 * print-tree.c (print_node): Print DECL_READ_P flag.
8205 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
8207 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
8208 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
8209 * cgraph.c: Likewise.
8210 * cgraphunit.c: Likewise.
8211 * gengtype.c: Likewise.
8212 * ipa-cp.c: Likewise.
8213 * ipa-devirt.c: Likewise.
8214 * ipa-icf.c: Likewise.
8215 * ipa-predicate.c: Likewise.
8216 * ipa-profile.c: Likewise.
8217 * ipa-prop.c: Likewise.
8218 * ipa-split.c: Likewise.
8220 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
8221 edge_predicate_pool, dump_inline_hints,
8222 inline_summary::account_size_time, redirect_to_unreachable,
8223 edge_set_predicate, set_hint_predicate,
8224 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
8225 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
8226 inline_summary_t::remove, remap_hint_predicate_after_duplication,
8227 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
8228 ipa_call_summary_t::remove, initialize_growth_caches,
8229 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
8230 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
8231 mark_modified, unmodified_parm_1, unmodified_parm,
8232 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
8233 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
8234 compute_bb_predicates, will_be_nonconstant_expr_predicate,
8235 will_be_nonconstant_predicate, record_modified_bb_info,
8236 get_minimal_bb, record_modified, param_change_prob,
8237 phi_result_unknown_predicate, predicate_for_phi_result,
8238 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
8239 estimate_function_body_sizes, compute_inline_parameters,
8240 compute_inline_parameters_for_curren, pass_data_inline_parameters,
8241 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
8242 inline_update_callee_summaries, remap_edge_change_prob,
8243 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
8244 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
8245 inline_analyze_function, inline_summary_t::insert,
8246 inline_generate_summary, read_ipa_call_summary, inline_read_section,
8247 inline_read_summary, write_ipa_call_summary, inline_write_summary,
8248 inline_free_summary): Move to ipa-fnsummary.h
8249 (predicate_t): Remove.
8250 * ipa-fnsummary.c: New file.
8251 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
8252 (enum inline_hints_vals, inline_hints, agg_position_info,
8253 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
8254 inline_summaries, ipa_call_summary, ipa_call_summary_t,
8255 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
8256 dump_inline_summary, dump_inline_hints, inline_generate_summary,
8257 inline_read_summary, inline_write_summary, inline_free_summary,
8258 inline_analyze_function, initialize_inline_failed,
8259 inline_merge_summary, inline_update_overall_summary,
8260 compute_inline_parameters): Move to ipa-fnsummary.h
8261 * ipa-fnsummary.h: New file.
8262 * ipa-inline-transform.h: Include ipa-inline.h.
8263 * ipa-inline.c: LIkewise.
8265 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
8267 * ipa-inline.c (edge_badness): Use inlined_time instead of
8268 inline_summaries->get.
8270 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
8272 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
8274 2017-05-22 Nathan Sidwell <nathan@acm.org>
8276 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
8277 (fdump-lang): Document 'raw' option.
8278 * dumpfile.h (TDI_tu): Delete.
8279 * dumpfile.c (dump_files): Remove translation-unit.
8280 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
8282 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
8284 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
8285 command option from $(AWK) call.
8286 * config/avr/genmultilib.awk: Simplify and rewrite so that it
8287 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
8288 [FORMAT]: Remove handling of variable.
8289 * config/avr/t-multilib: Regenerate.
8291 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
8293 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
8295 (dump_inline_summary): Do not print self_time.
8296 (estimate_function_body_sizes): Do not set self_time.
8297 (compute_inline_parameters): Likewise.
8298 (inline_read_section, inline_write_summary): Do not stream self_time.
8299 * ipa-inline.h (inline_summary): Drop self_time.
8301 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
8303 * ipa-inline-analysis.c (account_size_time): Rename to ...
8304 (inline_summary::account_size_time): ... this one.
8305 (reset_ipa_call_summary): Turn to ...
8306 (ipa_call_summary::reset): ... this one.
8307 (reset_inline_summary): Turn to ...
8308 (inline_summary::reset): ... this one.
8309 (inline_summary_t::remove): Update.
8310 (inline_summary_t::duplicate): Update.
8311 (ipa_call_summary_t::remove): Update.
8312 (dump_inline_summary): Update.
8313 (estimate_function_body_sizes): Update.
8314 (compute_inline_parameters): Update.
8315 (estimate_node_size_and_time): Update.
8316 (inline_merge_summary): Update.
8317 (inline_update_overall_summary): Update.
8318 (inline_read_section): Update.
8319 (inline_write_summary): Update.
8320 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
8321 add account_size_time and reset member functions.
8322 (ipa_call_summary): Add reset function.
8323 * ipa-predicate.h (predicate::operator &): Constify.
8325 2017-05-22 Richard Biener <rguenther@suse.de>
8327 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
8329 2017-05-19 Jason Merrill <jason@redhat.com>
8331 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
8333 2017-05-19 Marek Polacek <polacek@redhat.com>
8336 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
8337 TYPE_OVERFLOW_WRAPS checks.
8339 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
8341 * tree-core.h (enum omp_clause_default_kind): Add
8342 "OMP_CLAUSE_DEFAULT_PRESENT".
8343 * tree-pretty-print.c (dump_omp_clause): Handle it.
8344 * gimplify.c (enum gimplify_omp_var_data): Add
8345 "GOVD_MAP_FORCE_PRESENT".
8346 (gimplify_adjust_omp_clauses_1): Map it to
8347 "GOMP_MAP_FORCE_PRESENT".
8348 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
8350 * gimplify.c (oacc_default_clause): Clarify.
8352 2017-05-19 Nathan Sidwell <nathan@acm.org>
8354 LANG_HOOK_REGISTER_DUMPS
8355 * toplev.c (general_init): Call register dump lang hook.
8356 * doc/invoke.texi: Document -fdump-lang option family.
8357 * dumpfile.c (dump_files): Remove class dump here.
8358 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
8359 * dumpfile.h (tree_dump_index): Remove TDI_class.
8360 * langhooks-def.h (lhd_register_dumps): Declare.
8361 (LANG_HOOKS_REGISTER_DUMPS): Define.
8362 (LANG_HOOKS_INITIALIZER): Add it.
8363 * langhooks.c (lhd_register_dumps): Define.
8364 * langhooks.h (struct lang_hooks): Add register_dumps.
8366 2017-05-19 Nathan Sidwell <nathan@acm.org>
8368 * context.h (context::set_passes): New.
8369 * context.c (context::context): Do not create pass manager.
8370 * toplev.c (general_init): Create pass manager here.
8372 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
8374 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
8375 use this splitter if two add or or instructions would also work for
8376 the constant we want to generate.
8378 2017-05-19 Richard Biener <rguenther@suse.de>
8381 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
8382 predicate evaluation.
8384 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
8386 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
8388 * ipa-inline.c (want_inline_small_function_p): Do not cast to
8391 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
8393 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
8394 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
8395 (inline_edge_summary_vec): Turn into ...
8396 (ipa_call_summaries): ... this one.
8397 (redirect_to_unreachable, edge_set_predicate,
8398 evaluate_properties_for_edge, inline_summary_alloc,
8399 reset_ipa_call_summary, reset_inline_summary,
8400 inline_summary_t::duplicate): Update.
8401 (inline_edge_duplication_hook): Turn to ...
8402 (ipa_call_summary_t::duplicate): ... this one.
8403 (inline_edge_removal_hook): Turn to ...
8404 (ipa_call_summary_t::remove): ... this one.
8405 (dump_inline_edge_summary): Turn to ...
8406 (dump_ipa_call_summary): ... this one.
8407 (estimate_function_body_sizes): Update.
8408 (inline_update_callee_summaries): Update.
8409 (remap_edge_change_prob): Update.
8410 (remap_edge_summaries): Update.
8411 (inline_merge_summary): Update.
8412 (do_estimate_edge_time): Update.
8413 (inline_generate_summary): Update.
8414 (inline_read_section): Update.
8415 (inline_read_summary): Update.
8416 (inline_free_summary): Update.
8417 * ipa-inline.c (can_inline_edge_p): Update.
8418 (compute_inlined_call_time): Update.
8419 (want_inline_small_function_p): Update.
8420 (edge_badness): Update.
8421 (early_inliner): Update.
8422 * ipa-inline.h (inline_edge_summary): Turn to ...
8423 (ipa_call_summary): ... this one.
8424 (ipa_call_summary_t): New class.
8425 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
8426 (ipa_call_summaries): New.
8427 (inline_edge_summary): Remove.
8428 (estimate_edge_growth): Update.
8429 * ipa-profile.c (ipa_propagate_frequency_1): Update.
8430 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
8431 * ipa-split.c (execute_split_functions): Update.
8432 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
8434 2017-05-19 Richard Biener <rguenther@suse.de>
8437 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
8439 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
8441 * config/rs6000/rs6000.c (struct machine_function): Add field
8442 fpr_is_wrapped_separately.
8443 (rs6000_get_separate_components): Use 64 components. Handle the
8445 (rs6000_components_for_bb): Handle the FPR components.
8446 (rs6000_emit_prologue_components): Handle the FPR components.
8447 (rs6000_emit_epilogue_components): Handle the FPR components.
8448 (rs6000_set_handled_components): Handle the FPR components.
8449 (rs6000_emit_prologue): Don't output prologue code for those FPRs
8450 that are already separately shrink-wrapped.
8451 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
8452 that are already separately shrink-wrapped.
8454 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
8457 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
8460 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
8461 (define_peephole2 for Altivec d-form load): Add peepholes to catch
8462 cases where the register allocator uses a move and an offsettable
8463 memory operation to/from a FPR register on ISA 2.06/2.07.
8464 (define_peephole2 for Altivec d-form store): Likewise.
8466 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
8469 * config/i386/mmx.md (*mov<mode>_internal): Enable
8470 alternatives 11, 12, 13 and 14 also for 32bit targets.
8471 Remove alternatives 15, 16, 17 and 18.
8472 * config/i386/sse.md (vec_concatv2di): Change
8473 alternative (!x, *y) to (x, ?!*Yn).
8475 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
8477 * dumpfile.h (enum dump_kind): Remove stray comma.
8479 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
8481 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
8482 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
8483 predicate::num_conditions
8484 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
8485 (CHANGED): turn into predicate::changed.
8486 (agg_position_info): Move to ipa-predicate.h
8487 (add_condition, predicate::add_clause, predicate::operator &=,
8488 predicate::or_with, predicate::evaluate, predicate::probability,
8489 dump_condition, dump_clause, predicate::dump,
8490 predicate::remap_after_duplication, predicate::remap_after_inlining,
8491 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
8492 (evaluate_conditions_for_known_args): Update.
8493 (set_cond_stmt_execution_predicate): Update.
8494 * ipa-inline.h: Include ipa-predicate.h
8495 (condition, inline_param_summary, conditions, agg_position_info,
8496 predicate): Move to ipa-predicate.h
8497 * ipa-predicate.c: New file.
8498 * ipa-predicate.h: New file.
8500 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
8502 * final.c (leaf_function_p): Check we are not in a sequence.
8504 2017-05-18 Martin Liska <mliska@suse.cz>
8506 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
8507 * dumpfile.c (dump_register): Use new enum dump_kind.
8508 (get_dump_file_name): Likewise.
8509 (dump_enable_all): Likewise.
8510 (dump_switch_p_1): Likewise.
8511 (enable_rtl_dump_file): Remove usage of TDF_RTL.
8512 * dumpfile.h (enum dump_kind): New enum type.
8513 (struct dump_file_info): Create constructor and
8514 format fields and comments.
8515 * passes.c (pass_manager::register_one_dump_file):
8517 * statistics.c (statistics_early_init): Likewise.
8518 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
8519 TDF_TREE with TDF_SLIM.
8520 (gather_memory_references_ref): Likewise.
8522 2017-05-18 Martin Liska <mliska@suse.cz>
8524 * vec.h (struct vnull): Use it.
8526 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
8528 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
8529 (true_predicate, false_predicate, true_predicate_p,
8530 false_predicate_p): Remove.
8531 (single_cond_predicate, not_inlined_predicate): Turn to member function
8533 (add_condition): Update.
8534 (add_clause): Turn to...
8535 (predicate::add_clause): ... this one; update; allow passing NULL
8537 (and_predicates): Turn to ...
8538 (predicate::operator &=): ... this one.
8539 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
8540 (or_predicates): Turn to ...
8541 (predicate::or_with): ... this one.
8542 (evaluate_predicate): Turn to ...
8543 (predicate::evaluate): ... this one.
8544 (predicate_probability): Turn to ...
8545 (predicate::probability): ... this one.
8546 (dump_condition): Update.
8547 (dump_predicate): Turn to ...
8548 (predicate::dump): ... this one.
8549 (account_size_time): Update.
8550 (edge_set_predicate): Update.
8551 (set_hint_predicate): UPdate.
8552 (evaluate_conditions_for_known_args): Update.
8553 (evaluate_properties_for_edge): Update.
8554 (remap_predicate_after_duplication): Turn to...
8555 (predicate::remap_after_duplication): ... this one.
8556 (remap_hint_predicate_after_duplication): Update.
8557 (inline_summary_t::duplicate): UPdate.
8558 (dump_inline_edge_summary): Update.
8559 (dump_inline_summary): Update.
8560 (set_cond_stmt_execution_predicate): Update.
8561 (set_switch_stmt_execution_predicate): Update.
8562 (compute_bb_predicates): Update.
8563 (will_be_nonconstant_expr_predicate): Update.
8564 (will_be_nonconstant_predicate): Update.
8565 (phi_result_unknown_predicate): Update.
8566 (predicate_for_phi_result): Update.
8567 (array_index_predicate): Update.
8568 (estimate_function_body_sizes): Update.
8569 (estimate_node_size_and_time): Update.
8570 (estimate_ipcp_clone_size_and_time): Update.
8571 (remap_predicate): Rename to ...
8572 (predicate::remap_after_inlining): ... this one.
8573 (remap_hint_predicate): Update.
8574 (inline_merge_summary): Update.
8575 (inline_update_overall_summary): Update.
8576 (estimate_size_after_inlining): Update.
8577 (read_predicate): Rename to ...
8578 (predicate::stream_in): ... this one.
8579 (read_inline_edge_summary): Update.
8580 (write_predicate): Rename to ...
8581 (predicate::stream_out): ... this one.
8582 (write_inline_edge_summary): Update.
8583 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
8584 (clause_t): Turn to uint32_t
8585 (predicate): Turn to class; implement constructor and operators
8587 (size_time_entry): Update.
8588 (inline_summary): Update.
8589 (inline_edge_summary): Update.
8591 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
8593 * fold-const.c (fold_binary_loc): Move transformation...
8594 * match.pd (C - X CMP X): ... here.
8596 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
8598 * config/sparc/sparc.c (sparc_option_override): Set function
8599 alignment for -mcpu=niagara7 to 64 to match the I$ line.
8600 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
8602 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
8604 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
8606 2017-05-18 Marek Polacek <polacek@redhat.com>
8609 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
8610 (pass_ubsan::execute): Call gimple_assign_single_p instead of
8611 gimple_assign_load_p.
8613 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
8616 * real.c (do_compare): Give decimal_do_compare preference over
8617 comparing just the signs.
8619 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
8621 * doc/md.texi (Canonicalization of Instructions): Describe the
8622 canonical form of instructions that inherently set a condition
8625 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
8628 * tree-cfg.c: Move deletion of unreachable case statements to after
8629 the merging of consecutive case labels.
8631 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
8633 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
8634 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
8635 restoring of callee-saved registers.
8637 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
8639 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
8640 * config/visium/visium.c (single_set_and_flags): Likewise.
8641 * config/visium/visium.md (Substitutions): Likewise.
8643 2017-05-17 Martin Liska <mliska@suse.cz>
8645 * cfg.c: Introduce dump_flags_t type and
8646 use it instead of int type.
8648 * cfghooks.c: Likewise.
8649 * cfghooks.h (struct cfg_hooks): Likewise.
8650 * cfgrtl.c: Likewise.
8651 * cfgrtl.h: Likewise.
8652 * cgraph.c (cgraph_node::get_body): Likewise.
8653 * coretypes.h: Likewise.
8654 * domwalk.c: Likewise.
8655 * domwalk.h: Likewise.
8656 * dumpfile.c (struct dump_option_value_info): Likewise.
8657 (dump_enable_all): Likewise.
8658 (dump_switch_p_1): Likewise.
8659 (opt_info_switch_p): Likewise.
8660 * dumpfile.h (enum tree_dump_index): Likewise.
8661 (struct dump_file_info): Likewise.
8662 * genemit.c: Likewise.
8663 * generic-match-head.c: Likewise.
8664 * gengtype.c (open_base_files): Likewise.
8665 * gimple-pretty-print.c: Likewise.
8666 * gimple-pretty-print.h: Likewise.
8667 * graph.c (print_graph_cfg): Likewise.
8668 * graphite-scop-detection.c (dot_all_sese): Likewise.
8669 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
8670 * loop-unroll.c (report_unroll): Likewise.
8671 * passes.c (pass_manager::register_one_dump_file): Likewise.
8672 * print-tree.c: Likewise.
8673 * statistics.c: Likewise.
8674 * tree-cfg.c: Likewise.
8675 * tree-cfg.h: Likewise.
8676 * tree-dfa.c: Likewise.
8677 * tree-dfa.h: Likewise.
8678 * tree-dump.c (dump_function): Likewise.
8679 * tree-dump.h (struct dump_info): Likewise.
8680 * tree-pretty-print.c: Likewise.
8681 * tree-pretty-print.h: Likewise.
8682 * tree-ssa-live.c: Likewise.
8683 * tree-ssa-live.h: Likewise.
8684 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
8685 * tree-vect-loop.c: Likewise.
8686 * tree-vect-slp.c: Likewise.
8688 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
8689 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8691 PR tree-optimization/80457
8692 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
8693 of all arguments to a statement as scalar_to_vec operations.
8694 (vectorizable_call): Adjust call to vect_model_simple_cost for
8696 (vectorizable_conversion): Likewise.
8697 (vectorizable_assignment): Likewise.
8698 (vectorizable_shift): Likewise.
8699 (vectorizable_operation): Likewise.
8700 (vectorizable_comparison): Likewise.
8701 (vect_is_simple_cond): Record the def types for operands.
8702 (vectorizable_condition): Likewise, call vect_model_simple_cost.
8703 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
8704 for statement argument count.
8706 2017-05-16 Carl Love <cel@us.ibm.com>
8708 * config/rs6000/rs6000-c: Add support for built-in functions
8709 vector unsigned long long vec_bperm (vector unsigned long long,
8710 vector unsigned char)
8711 vector signed long long vec_mule (vector signed int,
8713 vector unsigned long long vec_mule (vector unsigned int,
8714 vector unsigned int)
8715 vector signed long long vec_mulo (vector signed int,
8717 vector unsigned long long vec_mulo (vector unsigned int,
8718 vector unsigned int)
8719 vector signed char vec_sldw (vector signed char,
8722 vector unsigned char vec_sldw (vector unsigned char,
8723 vector unsigned char,
8725 vector signed short vec_sldw (vector signed short,
8726 vector signed short,
8728 vector unsigned short vec_sldw (vector unsigned short,
8729 vector unsigned short,
8731 vector signed int vec_sldw (vector signed int,
8734 vector unsigned int vec_sldw (vector unsigned int,
8735 vector unsigned int,
8737 vector signed long long vec_sldw (vector signed long long,
8738 vector signed long long,
8740 vector unsigned long long vec_sldw (vector unsigned long long,
8741 vector unsigned long long,
8743 * config/rs6000/rs6000-c: Add support for built-in functions
8744 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
8745 * config/rs6000/altivec.h: Add defintion for vec_sldw.
8746 * doc/extend.texi: Update the built-in documentation for the
8747 new built-in functions.
8749 2017-05-16 Marek Polacek <polacek@redhat.com>
8753 * tree.c (save_expr): Don't fold the expression.
8755 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
8757 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
8758 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
8759 and (?*y,m). Update insn attributes.
8761 2017-05-16 Martin Liska <mliska@suse.cz>
8763 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
8764 flags argument of print_gimple_stmt, print_gimple_expr,
8765 print_generic_stmt and print_generic_expr.
8766 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
8767 * coretypes.h: Likewise.
8768 * except.c (dump_eh_tree): Likewise.
8769 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
8770 * gimple-pretty-print.h: Likewise.
8771 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
8772 (backprop::push_to_worklist): Likewise.
8773 (backprop::pop_from_worklist): Likewise.
8774 (backprop::process_use): Likewise.
8775 (backprop::intersect_uses): Likewise.
8776 (note_replacement): Likewise.
8777 * gimple-ssa-store-merging.c
8778 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
8779 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
8780 (pass_store_merging::execute): Likewise.
8781 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
8782 (ssa_base_cand_dump_callback): Likewise.
8783 (dump_incr_vec): Likewise.
8784 (replace_refs): Likewise.
8785 (replace_mult_candidate): Likewise.
8786 (create_add_on_incoming_edge): Likewise.
8787 (create_phi_basis): Likewise.
8788 (insert_initializers): Likewise.
8789 (all_phi_incrs_profitable): Likewise.
8790 (introduce_cast_before_cand): Likewise.
8791 (replace_one_candidate): Likewise.
8792 * gimplify.c (gimplify_expr): Likewise.
8793 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
8794 (set_rename): Likewise.
8795 (rename_uses): Likewise.
8796 (copy_loop_phi_nodes): Likewise.
8797 (add_close_phis_to_merge_points): Likewise.
8798 (copy_loop_close_phi_args): Likewise.
8799 (copy_cond_phi_args): Likewise.
8800 (graphite_copy_stmts_from_block): Likewise.
8801 (translate_pending_phi_nodes): Likewise.
8802 * graphite-poly.c (print_pdr): Likewise.
8803 (dump_gbb_cases): Likewise.
8804 (dump_gbb_conditions): Likewise.
8805 (print_scop_params): Likewise.
8806 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
8807 (build_cross_bb_scalars_use): Likewise.
8808 (gather_bbs::before_dom_children): Likewise.
8809 * hsa-dump.c (dump_hsa_immed): Likewise.
8810 * ipa-cp.c (print_ipcp_constant_value): Likewise.
8811 (get_replacement_map): Likewise.
8812 * ipa-inline-analysis.c (dump_condition): Likewise.
8813 (estimate_function_body_sizes): Likewise.
8814 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
8815 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
8816 * ipa-prop.c (ipa_dump_param): Likewise.
8817 (ipa_print_node_jump_functions_for_edge): Likewise.
8818 (ipa_modify_call_arguments): Likewise.
8819 (ipa_modify_expr): Likewise.
8820 (ipa_dump_param_adjustments): Likewise.
8821 (ipa_dump_agg_replacement_values): Likewise.
8822 (ipcp_modif_dom_walker::before_dom_children): Likewise.
8823 * ipa-pure-const.c (check_stmt): Likewise.
8824 (pass_nothrow::execute): Likewise.
8825 * ipa-split.c (execute_split_functions): Likewise.
8826 * omp-offload.c (dump_oacc_loop_part): Likewise.
8827 (dump_oacc_loop): Likewise.
8828 * trans-mem.c (tm_log_emit): Likewise.
8829 (tm_memopt_accumulate_memops): Likewise.
8830 (dump_tm_memopt_set): Likewise.
8831 (dump_tm_memopt_transform): Likewise.
8832 * tree-cfg.c (gimple_verify_flow_info): Likewise.
8833 (print_loop): Likewise.
8834 * tree-chkp-opt.c (chkp_print_addr): Likewise.
8835 (chkp_gather_checks_info): Likewise.
8836 (chkp_get_check_result): Likewise.
8837 (chkp_remove_check_if_pass): Likewise.
8838 (chkp_use_outer_bounds_if_possible): Likewise.
8839 (chkp_reduce_bounds_lifetime): Likewise.
8840 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
8841 (chkp_mark_completed_bounds): Likewise.
8842 (chkp_register_incomplete_bounds): Likewise.
8843 (chkp_mark_invalid_bounds): Likewise.
8844 (chkp_maybe_copy_and_register_bounds): Likewise.
8845 (chkp_build_returned_bound): Likewise.
8846 (chkp_get_bound_for_parm): Likewise.
8847 (chkp_build_bndldx): Likewise.
8848 (chkp_get_bounds_by_definition): Likewise.
8849 (chkp_generate_extern_var_bounds): Likewise.
8850 (chkp_get_bounds_for_decl_addr): Likewise.
8851 * tree-chrec.c (chrec_apply): Likewise.
8852 * tree-data-ref.c (dump_data_reference): Likewise.
8853 (dump_subscript): Likewise.
8854 (dump_data_dependence_relation): Likewise.
8855 (analyze_overlapping_iterations): Likewise.
8856 * tree-inline.c (expand_call_inline): Likewise.
8857 (tree_function_versioning): Likewise.
8858 * tree-into-ssa.c (dump_defs_stack): Likewise.
8859 (dump_currdefs): Likewise.
8860 (dump_names_replaced_by): Likewise.
8861 (dump_update_ssa): Likewise.
8862 (update_ssa): Likewise.
8863 * tree-object-size.c (pass_object_sizes::execute): Likewise.
8864 * tree-parloops.c (build_new_reduction): Likewise.
8865 (try_create_reduction_list): Likewise.
8866 (ref_conflicts_with_region): Likewise.
8867 (oacc_entry_exit_ok_1): Likewise.
8868 (oacc_entry_exit_single_gang): Likewise.
8869 * tree-pretty-print.h: Likewise.
8870 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
8871 (get_scalar_evolution): Likewise.
8872 (add_to_evolution): Likewise.
8873 (get_loop_exit_condition): Likewise.
8874 (analyze_evolution_in_loop): Likewise.
8875 (analyze_initial_condition): Likewise.
8876 (analyze_scalar_evolution): Likewise.
8877 (instantiate_scev): Likewise.
8878 (number_of_latch_executions): Likewise.
8879 (gather_chrec_stats): Likewise.
8880 (final_value_replacement_loop): Likewise.
8881 (scev_const_prop): Likewise.
8882 * tree-sra.c (dump_access): Likewise.
8883 (disqualify_candidate): Likewise.
8884 (create_access): Likewise.
8886 (maybe_add_sra_candidate): Likewise.
8887 (create_access_replacement): Likewise.
8888 (analyze_access_subtree): Likewise.
8889 (analyze_all_variable_accesses): Likewise.
8890 (sra_modify_assign): Likewise.
8891 (initialize_constant_pool_replacements): Likewise.
8892 (find_param_candidates): Likewise.
8893 (decide_one_param_reduction): Likewise.
8894 (replace_removed_params_ssa_names): Likewise.
8895 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
8896 * tree-ssa-copy.c (dump_copy_of): Likewise.
8897 (copy_prop_visit_cond_stmt): Likewise.
8898 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
8899 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
8900 (record_equivalences_from_stmt): Likewise.
8901 * tree-ssa-dse.c (compute_trims): Likewise.
8902 (delete_dead_call): Likewise.
8903 (delete_dead_assignment): Likewise.
8904 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
8905 (forward_propagate_into_cond): Likewise.
8906 (pass_forwprop::execute): Likewise.
8907 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
8908 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
8910 (move_computations_worker): Likewise.
8911 (execute_sm): Likewise.
8912 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
8913 (remove_exits_and_undefined_stmts): Likewise.
8914 (remove_redundant_iv_tests): Likewise.
8915 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
8916 (adjust_iv_update_pos): Likewise.
8917 * tree-ssa-math-opts.c (bswap_replace): Likewise.
8918 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
8919 (value_replacement): Likewise.
8920 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
8921 * tree-ssa-pre.c (print_pre_expr): Likewise.
8922 (get_representative_for): Likewise.
8923 (create_expression_by_pieces): Likewise.
8924 (insert_into_preds_of_block): Likewise.
8925 (eliminate_insert): Likewise.
8926 (eliminate_dom_walker::before_dom_children): Likewise.
8927 (eliminate): Likewise.
8928 (remove_dead_inserted_code): Likewise.
8929 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
8930 * tree-ssa-reassoc.c (get_rank): Likewise.
8931 (eliminate_duplicate_pair): Likewise.
8932 (eliminate_plus_minus_pair): Likewise.
8933 (eliminate_not_pairs): Likewise.
8934 (undistribute_ops_list): Likewise.
8935 (eliminate_redundant_comparison): Likewise.
8936 (update_range_test): Likewise.
8937 (optimize_range_tests_var_bound): Likewise.
8938 (optimize_vec_cond_expr): Likewise.
8939 (rewrite_expr_tree): Likewise.
8940 (rewrite_expr_tree_parallel): Likewise.
8941 (linearize_expr): Likewise.
8942 (break_up_subtract): Likewise.
8943 (linearize_expr_tree): Likewise.
8944 (attempt_builtin_powi): Likewise.
8945 (attempt_builtin_copysign): Likewise.
8946 (transform_stmt_to_copy): Likewise.
8947 (transform_stmt_to_multiply): Likewise.
8948 (dump_ops_vector): Likewise.
8949 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
8950 (print_scc): Likewise.
8951 (set_ssa_val_to): Likewise.
8952 (visit_reference_op_store): Likewise.
8953 (visit_use): Likewise.
8954 (sccvn_dom_walker::before_dom_children): Likewise.
8955 (run_scc_vn): Likewise.
8956 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
8958 (expr_hash_elt::print): Likewise.
8959 (const_and_copies::pop_to_marker): Likewise.
8960 (const_and_copies::record_const_or_copy_raw): Likewise.
8961 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
8962 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
8963 (dump_predicates): Likewise.
8964 (find_uninit_use): Likewise.
8965 (warn_uninitialized_phi): Likewise.
8966 (pass_late_warn_uninitialized::execute): Likewise.
8967 * tree-ssa.c (verify_vssa): Likewise.
8968 (verify_ssa): Likewise.
8969 (maybe_optimize_var): Likewise.
8970 * tree-vrp.c (dump_value_range): Likewise.
8971 (dump_all_value_ranges): Likewise.
8972 (dump_asserts_for): Likewise.
8973 (register_edge_assert_for_2): Likewise.
8974 (vrp_visit_cond_stmt): Likewise.
8975 (vrp_visit_switch_stmt): Likewise.
8976 (vrp_visit_stmt): Likewise.
8977 (vrp_visit_phi_node): Likewise.
8978 (simplify_cond_using_ranges_1): Likewise.
8979 (fold_predicate_in): Likewise.
8980 (evrp_dom_walker::before_dom_children): Likewise.
8981 (evrp_dom_walker::push_value_range): Likewise.
8982 (evrp_dom_walker::pop_value_range): Likewise.
8983 (execute_early_vrp): Likewise.
8985 2017-05-16 Richard Biener <rguenther@suse.de>
8987 * dwarf2out.c (loc_list_from_tree_1): Do not create
8988 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
8990 2017-05-16 Richard Biener <rguenther@suse.de>
8992 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
8994 (note_variable_value_in_expr): If we resolved the decl ref
8995 do not push to the stack.
8997 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
8999 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
9000 operations in fast-math mode.
9001 (vaddq_f16): Likewise.
9002 (vmul_f16): Likewise.
9003 (vmulq_f16): Likewise.
9004 (vsub_f16): Likewise.
9005 (vsubq_f16): Likewise.
9006 * config/arm/neon.md (add<mode>3): New.
9008 (fma:<VH:mode>3): New. Also remove outdated comment.
9011 2017-05-16 Martin Liska <mliska@suse.cz>
9015 * ipa-devirt.c (warn_types_mismatch): Fix typo.
9016 (odr_types_equivalent_p): Likewise.
9018 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
9020 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
9022 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
9025 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
9026 non-interunit SSE move alternatives with '?'.
9027 (zero-extendsidi peephole2): New peephole to skip intermediate
9028 general register in SSE zero-extend sequence.
9030 2017-05-15 Jeff Law <law@redhat.com>
9032 * reorg.c (relax_delay_slots): Create a new variable to hold
9033 the temporary target rather than clobbering TARGET_LABEL.
9035 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
9036 missing argument to extract_bit_field call.
9037 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
9039 2017-05-15 Martin Liska <mliska@suse.cz>
9042 * gcc.c (process_command): Do not allow empty argument of -o option.
9044 2017-05-15 Renlin Li <renlin.li@arm.com>
9046 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
9047 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
9048 * config/aarch64/constraints.md (Usf): Add long call check.
9049 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
9050 (call_value): Likewise.
9051 (sibcall): Likewise.
9052 (sibcall_value): Likewise.
9054 (call_value_insn): New.
9055 (sibcall_insn): Update rtx pattern.
9056 (sibcall_value_insn): Likewise.
9057 (call_internal): Remove.
9058 (call_value_internal): Likewise.
9059 (sibcall_internal): Likewise.
9060 (sibcall_value_internal): Likewise.
9061 (call_reg): Likewise.
9062 (call_symbol): Likewise.
9063 (call_value_reg): Likewise.
9064 (call_value_symbol): Likewise.
9066 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
9069 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
9071 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
9073 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
9074 compatible with CCGOCmode and with CCZmode.
9076 2017-05-14 Martin Sebor <msebor@redhat.com>
9079 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
9080 (gimple_fold_builtin_snprintf): Same.
9081 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
9082 (gimple_fold_builtin_snprintf): Same.
9083 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
9085 (is_call_safe): New function.
9086 (try_substitute_return_value): Call it.
9087 (try_simplify_call): New function.
9088 (pass_sprintf_length::handle_gimple_call): Call it.
9090 2017-05-14 Martin Sebor <msebor@redhat.com>
9093 * builtins.c (expand_builtin_stpncpy): Simplify.
9095 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
9097 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
9098 * config/i386/i386.h
9099 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
9100 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
9101 (struct machine_function): Add new members call_ms2sysv,
9102 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
9103 (struct machine_frame_state): New fields sp_realigned and
9104 sp_realigned_offset.
9105 * config/i386/i386.c
9106 (enum xlogue_stub): New enum.
9107 (enum xlogue_stub_sets): New enum.
9108 (class xlogue_layout): New class.
9109 (struct ix86_frame): New fields stack_realign_allocate_offset,
9110 stack_realign_offset and outlined_save_offset. Modify comments to
9111 detail stack layout when using out-of-line stubs.
9112 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
9113 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
9114 -mcall-ms2sysv-xlogues.
9115 (stub_managed_regs): New static variable.
9116 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
9117 registers managed by out-of-line stub.
9118 (disable_call_ms2sysv_xlogues): New function.
9119 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
9120 m->call_ms2sysv when appropriate and compute frame layout for
9122 (sp_valid_at, fp_valid_at): New inline functions.
9123 (choose_basereg): New function.
9124 (choose_baseaddr): Add align parameter, use choose_basereg and modify
9126 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
9127 Use align parameter of choose_baseaddr to generated aligned SSE movs
9129 (pro_epilogue_adjust_stack): Modify to track
9130 machine_frame_state::sp_realigned.
9131 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
9132 (ix86_nsaved_sseregs): Likewise.
9133 (ix86_emit_save_regs): Likewise.
9134 (ix86_emit_save_regs_using_mov): Likewise.
9135 (ix86_emit_save_sse_regs_using_mov): Likewise.
9136 (get_scratch_register_on_entry): Likewise.
9137 (gen_frame_set): New function.
9138 (gen_frame_load): Likewise.
9139 (gen_frame_store): Likewise.
9140 (emit_outlined_ms2sysv_save): Likewise.
9141 (emit_outlined_ms2sysv_restore): Likewise.
9142 (ix86_expand_prologue): Modify stack re-alignment code and call
9143 emit_outlined_ms2sysv_save when appropriate.
9144 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
9145 parameter rtx_insn *insn, which allows the function to be used to only
9147 (ix86_expand_epilogue): Modify validity checks of frame and stack
9148 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
9149 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
9150 * config/i386/predicates.md
9151 (save_multiple): New predicate.
9152 (restore_multiple): Likewise.
9153 * config/i386/sse.md
9154 (save_multiple<mode>): New pattern.
9155 (save_multiple_realign<mode>): Likewise.
9156 (restore_multiple<mode>): Likewise.
9157 (restore_multiple_and_return<mode>): Likewise.
9158 (restore_multiple_leave_return<mode>): Likewise.
9159 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
9161 2017-05-14 Julia Koval <julia.koval@intel.com>
9163 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
9164 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
9165 (__builtin_ia32_xsetbv): New builtins.
9166 * config/i386/i386.c (ix86_expand_special_args_builtin):
9168 (ix86_expand_builtin): Special expand for new intrinsics.
9169 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
9170 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
9171 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
9173 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9175 * cfganal.c (inverted_post_order_compute): Change argument type
9177 * cfganal.h (inverted_post_order_compute): Adjust prototype.
9178 * df-core.c (rest_of_handle_df_initialize): Adjust.
9179 (rest_of_handle_df_finish): Likewise.
9180 (df_analyze_1): Likewise.
9181 (df_analyze): Likewise.
9182 (loop_inverted_post_order_compute): Change argument to be a vec *.
9183 (df_analyze_loop): Adjust.
9184 (df_get_n_blocks): Likewise.
9185 (df_get_postorder): Likewise.
9186 * df.h (struct df_d): Change field to be a vec.
9187 * lcm.c (compute_laterin): Adjust.
9188 (compute_available): Likewise.
9189 * lra-lives.c (lra_create_live_ranges_1): Likewise.
9190 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
9191 * tree-ssa-pre.c (compute_antic): Likewise.
9193 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9195 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
9196 (depth_first_search::depth_first_search): Change structure init
9197 function to this constructor.
9198 (depth_first_search::add_bb): Rename function to this member.
9199 (depth_first_search::execute): Likewise.
9200 (flow_dfs_compute_reverse_finish): Adjust.
9202 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9204 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
9205 (longest_simple_path): Likewise.
9206 * shrink-wrap.c (spread_components): Likewise.
9207 (disqualify_problematic_components): Likewise.
9208 (emit_common_heads_for_components): Likewise.
9209 (emit_common_tails_for_components): Likewise.
9210 (insert_prologue_epilogue_for_components): Likewise.
9212 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9214 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
9217 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9219 * df-core.c (df_set_blocks): Start using auto_bitmap.
9220 (df_compact_blocks): Likewise.
9221 * df-problems.c (df_rd_confluence_n): Likewise.
9222 * df-scan.c (df_insn_rescan_all): Likewise.
9223 (df_process_deferred_rescans): Likewise.
9224 (df_update_entry_block_defs): Likewise.
9225 (df_update_exit_block_uses): Likewise.
9226 (df_entry_block_bitmap_verify): Likewise.
9227 (df_exit_block_bitmap_verify): Likewise.
9228 (df_scan_verify): Likewise.
9229 * lra-constraints.c (lra_constraints): Likewise.
9230 (undo_optional_reloads): Likewise.
9231 (lra_undo_inheritance): Likewise.
9232 * lra-remat.c (calculate_gen_cands): Likewise.
9233 (do_remat): Likewise.
9234 * lra-spills.c (assign_spill_hard_regs): Likewise.
9235 (spill_pseudos): Likewise.
9236 * tree-ssa-pre.c (bitmap_set_and): Likewise.
9237 (bitmap_set_subtract_values): Likewise.
9239 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9241 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
9242 management with auto_bitmap.
9243 (fix_inter_tick): Likewise.
9244 (fix_recovery_deps): Likewise.
9245 * ira.c (add_store_equivs): Likewise.
9246 (find_moveable_pseudos): Likewise.
9247 (split_live_ranges_for_shrink_wrap): Likewise.
9248 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
9249 (rtx_reuse_manager::seen_def_p): Likewise.
9250 (rtx_reuse_manager::set_seen_def): Likewise.
9251 * print-rtl.h (class rtx_reuse_manager): Likewise.
9253 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9255 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
9257 (migrate_btr_def): Likewise.
9258 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
9259 * df-core.c (loop_post_order_compute): Likewise.
9260 (loop_inverted_post_order_compute): Likewise.
9261 * hsa-common.h: Likewise.
9262 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
9263 * init-regs.c (initialize_uninitialized_regs): Likewise.
9264 * ipa-inline.c (resolve_noninline_speculation): Likewise.
9265 (inline_small_functions): Likewise.
9266 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
9267 * ira.c (combine_and_move_insns): Likewise.
9268 (build_insn_chain): Likewise.
9269 * loop-invariant.c (find_invariants): Likewise.
9270 * lower-subreg.c (propagate_pseudo_copies): Likewise.
9271 * predict.c (tree_predict_by_opcode): Likewise.
9272 (predict_paths_leading_to): Likewise.
9273 (predict_paths_leading_to_edge): Likewise.
9274 (estimate_loops_at_level): Likewise.
9275 (estimate_loops): Likewise.
9276 * shrink-wrap.c (try_shrink_wrapping): Likewise.
9277 (spread_components): Likewise.
9278 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
9279 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
9280 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
9281 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
9282 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
9283 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
9284 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
9285 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
9286 (mark_threaded_blocks): Likewise.
9287 (thread_through_all_blocks): Likewise.
9288 * tree-ssa.c (verify_ssa): Likewise.
9289 (execute_update_addresses_taken): Likewise.
9290 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
9292 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9294 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
9296 (post_order_compute): Likewise.
9297 (inverted_post_order_compute): Likewise.
9298 (pre_and_rev_post_order_compute_fn): Likewise.
9300 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9302 * genrecog.c (int_set::int_set): Explicitly construct our
9303 auto_vec base class.
9304 * vec.h (auto_vec::auto_vec): New constructor.
9306 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9308 * bitmap.h (class auto_bitmap): New constructor taking
9309 bitmap_obstack * argument.
9311 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9313 * bitmap.h (class auto_bitmap): Change type of m_bits to
9314 bitmap_head, and adjust ctor / dtor and member operators.
9316 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
9318 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
9319 when returned register mode doesn't match original mode.
9321 2017-05-12 Jeff Law <law@redhat.com>
9322 Jakub Jelinek <jakub@redhat.com>
9324 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
9325 we look for cc setter after the compare-elim changes.
9326 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
9327 within the vector to match what compare-elim now expects.
9328 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
9329 (xorsi3_flags, one_cmplsi2_flags): Likewise.
9331 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
9332 after the compare-elim changes.
9333 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
9334 the vector to match what compare-elim now expects.
9335 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
9336 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
9337 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
9338 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
9339 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
9341 * config/visium/visium.c (single_set_and_flags): Fix where
9342 we look for cc setter after the compare-elim changes.
9343 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
9344 with the vector to match what compare-elim now expects.
9345 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
9346 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
9347 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
9348 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
9349 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
9350 (neg<mode>2_insn_set_overflow): Likewise.
9352 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
9355 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
9356 maybe_expand_insn call, set ops[0].target. If still set after call,
9357 set alt_rtl. Add extra arg to recursive calls.
9358 (extract_bit_field): Add alt_rtl argument. Pass to
9360 * expmed.h (extract_bit_field): Fix prototype.
9361 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
9362 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
9363 to extract_bit_field_calls.
9364 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
9365 Pass alt_rtl to extract_bit_field calls.
9366 * calls.c (store_unaligned_arguments_into_psuedos)
9367 load_register_parameters): Pass extra NULL to extract_bit_field calls.
9368 * optabs.c (maybe_legitimize_operand): Clear op->target when call
9370 * optabs.h (struct expand_operand): Add target bitfield.
9372 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
9374 * compare-elim.c (try_eliminate_compare): Canonicalize
9375 operation with embedded compare to
9376 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
9377 (set (reg) (operation)].
9379 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
9381 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
9384 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
9385 cost of adding a carry flag for ADC instruction.
9386 [case MINUS]: Ignore the cost of subtracting a carry flag
9387 for SBB instruction.
9389 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
9391 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
9393 * config/rs6000/bmiintrin.h: New file.
9394 * config/rs6000/bmi2intrin.h: New file.
9395 * config/rs6000/x86intrin.h: New file.
9397 2017-05-12 Jeff Law <law@redhat.com>
9399 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
9402 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
9405 * tree-cfg.c: Remove cfg edges of unreachable case statements.
9407 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
9409 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
9410 early expansion of vector divide builtins.
9411 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
9412 builtins identified as having unsigned arguments.
9414 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
9416 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
9417 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
9418 expansion of vector logical operations (and, andc, or, xor,
9421 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
9423 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
9424 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
9426 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
9428 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
9429 early GIMPLE expansion of vector multiplies.
9431 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
9433 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
9434 TARGET_HAVE_MOVT conditional.
9435 (movt splitter): Likewise.
9437 2017-05-12 Richard Biener <rguenther@suse.de>
9439 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
9440 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
9441 Fold all stmts not inplace.
9443 2017-05-12 Richard Biener <rguenther@suse.de>
9445 PR tree-optimization/80713
9446 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
9447 inserted_exprs bit for not removed stmts.
9449 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
9452 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
9453 parallelized" attribute for parallelized OpenACC kernels.
9454 * omp-offload.c (execute_oacc_device_lower): Use it.
9456 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
9457 Set "oacc kernels" attribute.
9458 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
9459 parameter. Adjust all users.
9460 (oacc_fn_attrib_kernels_p): Remove function.
9461 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
9462 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
9463 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
9464 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
9465 assert "oacc kernels" attribute is set.
9467 2017-05-11 Carl Love <cel@us.ibm.com>
9469 * config/rs6000/rs6000-c: Add support for built-in functions
9470 vector unsigned char vec_popcnt (vector signed char)
9471 vector unsigned char vec_popcnt (vector unsigned char)
9472 vector unsigned short vec_popcnt (vector signed short)
9473 vector unsigned short vec_popcnt (vector unsigned short)
9474 vector unsigned int vec_popcnt (vector signed int)
9475 vector unsigned int vec_popcnt (vector unsigned int)
9476 vector unsigned long long vec_popcnt (vector signed long long)
9477 vector unsigned long long vec_popcnt (vector unsigned long long)
9478 vector signed long long vec_slo (vector signed long long,
9480 vector signed long long vec_slo (vector signed long long,
9481 vector unsigned char)
9482 vector unsigned long long vec_slo (vector unsigned long long,
9484 vector unsigned long long vec_slo (vector unsigned long long,
9485 vector unsigned char)
9486 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
9487 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
9488 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
9489 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
9490 * doc/extend.texi: Update the built-in documentation file for the
9491 new built-in functions.
9493 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
9495 * attribs.h (sorted_attr_string): Move machine independent
9496 functions for target clone support from the i386 port to common
9497 code. Rename ix86_function_versions to common_function_versions.
9498 Rename make_name to make_unique_name.
9499 (common_function_versions): Likewise.
9500 (make_unique_name): Likewise.
9501 (make_dispatcher_decl): Likewise.
9502 (is_function_default_version): Likewise.
9503 * attribs.c (attr_strcmp): Likewise.
9504 (sorted_attr_string): Likewise.
9505 (common_function_versions): Likewise.
9506 (make_unique_name): Likewise.
9507 (make_dispatcher_decl): Likewise.
9508 (is_function_default_version): Likewise.
9509 * config/i386/i386.c (attr_strcmp): Likewise.
9510 (sorted_attr_string): Likewise.
9511 (ix86_function_versions): Likewise.
9512 (make_name): Likewise.
9513 (make_dispatcher_decl): Likewise.
9514 (is_function_default_version): Likewise.
9515 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
9517 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9520 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
9521 Account for direct move costs for vec_construct of integer
9524 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
9527 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
9528 (UNSPEC_STX_ATOMIC): Ditto.
9529 (loaddi_via_sse): New insn.
9530 (storedi_via_sse): Ditto.
9531 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
9532 Update corresponding peephole2 patterns.
9533 (atomic_storedi_fpu): Ditto.
9535 2017-05-11 Julia Koval <julia.koval@intel.com>
9537 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
9538 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
9540 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
9541 (__builtin_ia32_rsqrt14ss_mask): New builtins.
9542 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
9544 2017-05-11 Nathan Sidwell <nathan@acm.org>
9546 * graphite-poly.c: Include dumpfile.h.
9548 * dumpfle.h (dump_function): Declare here ...
9549 * tree-dump.h (dump_function): ... not here.
9550 * dumpfile.c: #include tree-cfg.h.
9551 (dump_function): Move here from ...
9552 * tree-dump.c (dump_function): ... here.
9553 * gimplify.c: #include splay-tree.h, not tree-dump.h.
9554 * graphite-poly.c: Don't include tree-dump.h.
9555 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
9556 * print-tree.c: Likewise.
9557 * stor-layout.c: Likewise.
9558 * tree-nested.c: Likewise.
9560 * dumpfile.c (dump_start): Use TDF_FLAGS.
9561 (dump_enable_all): Fix TDF_KIND check thinko.
9563 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
9565 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9566 array entries to represent two legal parameterizations of the
9567 overloaded __builtin_cmpb function, as represented by the
9568 P6_OV_BUILTIN_CMPB constant.
9569 (altivec_resolve_overloaded_builtin): Add special case handling
9570 for the __builtin_cmpb function, as represented by the
9571 P6_OV_BUILTIN_CMPB constant.
9572 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
9573 (BU_P6_64BIT_2): New macro.
9574 (BU_P6_OVERLOAD_2): New macro
9575 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
9576 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
9577 (CMPB): Add overload support to represent both 32-bit and 64-bit
9578 compare-bytes function.
9579 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
9580 support for TARGET_CMPB.
9581 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
9582 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
9583 documentation of the __builtin_cmpb overloaded built-in function.
9585 2017-05-11 Richard Biener <rguenther@suse.de>
9587 PR tree-optimization/80705
9588 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
9589 bases are not vectorizable.
9591 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9593 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
9594 when counting register pressure.
9596 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9598 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
9599 (iv_ca_more_deps): Renamed to ...
9600 (iv_ca_compare_deps): ... this.
9601 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
9603 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9605 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
9607 (determine_group_iv_costs): ... here.
9608 (find_inv_vars_cb): Record inv var if it's not recorded before.
9610 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9612 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
9613 (get_shiftadd_cost): Ditto.
9615 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9617 * tree-ssa-address.c: Include header file.
9618 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
9620 (add_to_parts): Refactor.
9621 (addr_to_parts): New parameter. Update use of move_hint_to_base.
9622 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
9625 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9627 PR tree-optimization/53090
9628 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
9630 (extract_cond_operands): Detect condition with IV on both sides
9631 and return COMP_IV_EXPR_2.
9632 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
9633 (rewrite_use_compare): Simplify by removing call to function
9634 extract_cond_operands.
9636 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9638 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
9639 (extract_cond_operands): Detect condition comparing against non-
9640 invariant bound and return appropriate enum value.
9641 (find_interesting_uses_cond): Update use of extract_cond_operands.
9642 Handle its return value accordingly.
9643 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
9645 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9647 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
9648 nonlinear iv_use computation in loop invariant sensitive way.
9650 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9652 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
9653 (find_iv_candidates): Call relate_compare_use_with_all_cands.
9655 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9657 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
9658 (dump_cand): Support iv_cand.inv_exprs.
9659 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
9661 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
9664 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9666 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
9668 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
9669 as local function. Include necessary header files.
9670 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
9672 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9674 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
9676 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9678 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
9679 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
9680 RSHIFT_EXPR and BIT_NOT_EXPR.
9682 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9684 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
9685 (adjust_setup_cost): New parameter supporting round up adjustment.
9686 (struct address_cost_data): Delete.
9687 (force_expr_to_var_cost): Don't bound cost with spill_cost.
9688 (split_address_cost, ptr_difference_cost): Delete.
9689 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
9690 (struct ainc_cost_data): New struct.
9691 (get_address_cost_ainc): New function.
9692 (get_address_cost, get_computation_cost): Reimplement.
9693 (determine_group_iv_cost_address): Record inv_expr for all uses of
9695 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
9696 (iv_ca_has_deps): Reimplemented to ...
9697 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
9699 (iv_ca_extend): Call iv_ca_more_deps.
9701 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9703 * tree-ssa-address.c (struct mem_address): Move to header file.
9704 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
9705 * tree-ssa-address.h (struct mem_address): Move from C file.
9706 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
9708 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9710 * tree-affine.h (aff_combination_type): New interface.
9711 (aff_combination_zero_p): Remove static.
9712 (aff_combination_const_p): New interface.
9713 (aff_combination_singleton_var_p): New interfaces.
9715 2017-05-11 Richard Biener <rguenther@suse.de>
9717 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
9718 Skip unreachable blocks and destinations.
9719 (eliminate): Move stmt removal and fixup ...
9720 (fini_eliminate): ... here. Skip inserted exprs.
9721 (pass_pre::execute): Move fini_pre after fini_eliminate.
9722 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
9723 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
9724 PRE to get rid of dead code that has invalid SSA form and
9725 split critical edges again.
9727 2017-05-11 Bin Cheng <bin.cheng@arm.com>
9729 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
9731 2017-05-11 Richard Biener <rguenther@suse.de>
9733 * passes.c (execute_function_todo): Verify loops if they are
9734 said to be up-to-date.
9735 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
9736 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
9738 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
9741 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
9742 handle calling assemble_external ourself.
9745 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
9746 modes with zero size. Enhance comment.
9748 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9750 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
9751 built-ins for vec_xl and vec_xst with short and char pointer
9754 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
9756 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
9757 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
9758 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
9759 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
9760 (_mm_maskz_min_round_ss): New intrinsics.
9761 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
9762 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
9763 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
9764 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
9765 (__builtin_ia32_minss_mask_round): New builtins.
9766 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
9767 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
9768 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
9770 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
9771 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
9773 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
9776 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
9778 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
9779 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
9780 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
9781 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
9782 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
9783 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
9784 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
9785 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
9786 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
9787 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
9788 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
9789 (__builtin_ia32_mulss_mask_round): New builtins.
9790 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
9791 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
9792 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
9794 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
9795 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
9797 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
9800 2017-05-10 Julia Koval <julia.koval@intel.com>
9802 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
9803 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
9804 (_mm256_setr_m128i): New intrinsics.
9806 2017-05-10 Julia Koval <julia.koval@intel.com>
9808 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
9809 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
9810 (_mm_maskz_rcp14_ss): New intrinsics.
9811 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
9812 (__builtin_ia32_rcp14ss_mask): New builtins.
9813 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
9815 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
9817 PR tree-optimization/51513
9818 * tree-cfg.c (gimple_seq_unreachable_p): New function.
9819 (assert_unreachable_fallthru_edge_p): Use it.
9820 (group_case_labels_stmt): Likewise.
9821 * tree-cfg.h: Prototype it.
9822 * stmt.c: Include cfghooks.h and tree-cfg.h.
9823 (emit_case_dispatch_table) <gap_label>: New local variable.
9824 Use it to fill dispatch table gaps.
9825 Test for default_label before updating probabilities.
9826 (expand_case) <default_label>: Remove unneeded initialization.
9827 Test for unreachable default case statement and remove its edge.
9828 Set default_label accordingly.
9829 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
9831 2017-05-10 Carl Love <cel@us.ibm.com>
9833 * config/rs6000/rs6000-c: Add support for built-in functions
9834 vector signed char vec_neg (vector signed char)
9835 vector signed short int vec_neg (vector short int)
9836 vector signed int vec_neg (vector signed int)
9837 vector signed long long vec_neg (vector signed long long)
9838 vector float vec_neg (vector float)
9839 vector double vec_neg (vector double)
9840 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
9842 * config/rs6000/altivec.h: Add define for vec_neg
9843 * doc/extend.texi: Update the built-in documentation for the
9844 new built-in functions.
9846 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9848 PR tree-optimization/77644
9849 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
9851 2017-05-10 Nathan Sidwell <nathan@acm.org>
9853 * dumpfile.h (TDI_lang_all): New.
9854 (TDF_KIND): New. Renumber others
9855 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
9857 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
9859 (get_dump_file_name): Adjust suffix generation.
9860 (dump_enable_all): Use TDF_KIND.
9861 * doc/invoke.texi (-fdump-lang-all): Document.
9863 * dumpfile.h: Tabify.
9865 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
9868 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
9869 Move member access before delete.
9871 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
9873 * tree-inline.c (expand_call_inline): Split block at stmt
9876 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
9879 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
9880 are now unused after splitting mov{sf,sd}_hardfloat.
9881 (f32_lr2): Likewise.
9883 (f32_lm2): Likewise.
9885 (f32_li2): Likewise.
9888 (f32_sr2): Likewise.
9890 (f32_sm2): Likewise.
9892 (f32_si2): Likewise.
9895 (f32_vsx): Likewise.
9897 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
9898 For movsf, order stores so the VSX stores occur before the GPR
9899 store which encourages the register allocator to use a traditional
9900 FPR instead of a GPR. For movsd, order the stores so that the GPR
9901 store comes before the VSX stores to allow the power6 to work.
9902 This is due to the power6 not having a 32-bit integer store
9903 instruction from a FPR.
9904 (movsf_hardfloat): Likewise.
9905 (movsd_hardfloat): Likewise.
9907 2017-05-09 Martin Sebor <msebor@redhat.com>
9909 PR translation/80280
9910 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
9913 PR translation/80280
9914 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
9915 data member added in r247778.
9916 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
9918 2017-05-09 Nathan Sidwell <nathan@acm.org>
9920 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
9922 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
9925 2017-05-09 Marek Polacek <polacek@redhat.com>
9927 * doc/invoke.texi: Fix typo.
9929 2017-05-09 Richard Biener <rguenther@suse.de>
9931 * tree-vrp.c (vrp_val_is_max): Adjust comment.
9932 (vrp_val_is_min): Likewise.
9933 (set_value_range_to_value): Likewise.
9934 (set_value_range_to_nonnegative): Likewise.
9935 (gimple_assign_nonzero_p): Likewise.
9936 (gimple_stmt_nonzero_p): Likewise.
9937 (vrp_int_const_binop): Likewise. Remove unreachable case.
9938 (adjust_range_with_scev): Adjust comments.
9939 (compare_range_with_value): Likewise.
9940 (extract_range_from_phi_node): Likewise.
9941 (test_for_singularity): Likewise.
9943 2017-05-09 Richard Biener <rguenther@suse.de>
9945 * tree-vrp.c (get_single_symbol): Add assert that we don't
9946 get overflowed constants as invariant part.
9947 (compare_values_warnv): Add comment before the TREE_NO_WARNING
9948 checks. Use wi::cmp instead of recursing for integer constants.
9949 (compare_values): Just ignore whether we assumed undefined
9950 overflow instead of failing the compare.
9951 (extract_range_for_var_from_comparison_expr): Add comment before the
9952 TREE_NO_WARNING sets.
9953 (test_for_singularity): Likewise.
9954 (extract_range_from_comparison): Do not disable optimization
9955 when we assumed undefined overflow.
9956 (extract_range_basic): Remove init of unused var.
9958 2017-05-09 Richard Biener <rguenther@suse.de>
9960 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
9961 (extract_range_from_multiplicative_op_1): Adjust.
9962 (extract_range_from_binary_expr_1): Use int_const_binop.
9964 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
9967 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
9968 rs6000_store_data_bypass_p in seven define_bypass directives and
9969 in several comments.
9970 * config/rs6000/rs6000-protos.h: Add prototype for
9971 rs6000_store_data_bypass_p function.
9972 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
9973 function implements slightly different (rs6000-specific) semantics
9974 than store_data_bypass_p, returning false rather than aborting
9975 with assertion error when arguments do not satisfy the
9976 requirements of store data bypass.
9977 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
9978 rs6000_store_data_bypass_p.
9980 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
9982 * config/xtensa/xtensa-protos.h
9983 (xtensa_initial_elimination_offset): New declaration.
9984 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
9985 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
9986 macro definition, add case for FRAME_POINTER_REGNUM when
9987 FRAME_GROWS_DOWNWARD.
9988 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
9989 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
9990 xtensa_initial_elimination_offset.
9992 2017-05-08 Nathan Sidwell <nathan@acm.org>
9994 * doc/invoke.texi: Alphabetize -fdump options.
9996 2017-05-08 Martin Sebor <msebor@redhat.com>
9998 PR translation/80280
9999 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
10001 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
10003 * target.def (compute_frame_layout): New optional target hook.
10004 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
10005 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
10006 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
10008 * reload1.c (verify_initial_elim_offsets): Likewise.
10009 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
10010 (use_simple_return_p): Call arm_compute_frame_layout if needed.
10011 (arm_get_frame_offsets): Split up into this ...
10012 (arm_compute_frame_layout): ... and this function.
10014 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
10016 * config/aarch64/constraints.md (Usa): New constraint.
10017 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
10019 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
10021 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
10022 with_multilib_list after it has been checked.
10024 2017-05-08 Richard Biener <rguenther@suse.de>
10026 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
10027 (bitmap_set_subtract_values): Likewise.
10029 2017-05-08 Richard Biener <rguenther@suse.de>
10031 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
10032 (gimple_assign_nonzero): ... this and remove strict_overflow_p
10034 (gimple_stmt_nonzero_warnv_p): Rename to ...
10035 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
10037 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
10038 (extract_range_basic): Adjust, do not disable propagation on
10039 strict overflow sensitive simplification.
10040 (vrp_visit_cond_stmt): Likewise.
10042 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
10044 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
10045 body size unconditionally.
10047 2017-05-07 Jeff Law <law@redhat.com>
10050 2017-05-06 Jeff Law <law@redhat.com>
10051 PR tree-optimization/78496
10052 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
10055 PR tree-optimization/78496
10056 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
10057 (simplify_stmt_using_ranges): Call it.
10058 (vrp_dom_walker::before_dom_children): Extract equivalences
10059 from an ASSERT_EXPR with an equality comparison against a
10062 2017-05-06 Jeff Law <law@redhat.com>
10064 PR tree-optimization/78496
10065 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
10068 PR tree-optimization/78496
10069 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
10070 (simplify_stmt_using_ranges): Call it.
10071 (vrp_dom_walker::before_dom_children): Extract equivalences
10072 from an ASSERT_EXPR with an equality comparison against a
10075 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
10077 * lra-constraints.c (lra_copy_reg_equiv): New function.
10078 (split_reg): Use it to copy equivalence information from the
10079 original register to the spill register.
10081 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
10083 PR rtl-optimization/75964
10084 * simplify-rtx.c (simplify_const_relational_operation): Remove
10085 invalid handling of comparisons of integer ABS.
10087 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
10089 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
10090 initialize to zero.
10091 (init_regs): Remove declaration.
10092 (function_arg_advance_32): Initialize error_p as boolean variable.
10094 2017-05-05 Nathan Sidwell <nathan@acm.org>
10096 * store-motion.c (remove_reachable_equiv_notes): Reformat long
10097 lines. Use for (;;).
10099 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10101 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
10102 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
10103 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
10104 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
10105 VF=2 that require versioning.
10107 2017-05-05 David Malcolm <dmalcolm@redhat.com>
10109 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
10112 2017-05-05 David Malcolm <dmalcolm@redhat.com>
10114 * diagnostic.h (diagnostic_override_option_index): Convert from
10115 macro to inline function.
10117 2017-05-05 David Malcolm <dmalcolm@redhat.com>
10119 * diagnostic.c (last_module_changed_p): New function.
10120 (set_last_module): New function.
10121 (diagnostic_report_current_module): Convert macro usage to
10122 the above functions.
10123 * diagnostic.h (diagnostic_context::last_module): Strengthen
10124 from const line_map * to const line_map_ordinary *.
10125 (diagnostic_last_module_changed): Delete macro.
10126 (diagnostic_set_last_module): Delete macro.
10128 2017-05-05 David Malcolm <dmalcolm@redhat.com>
10130 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
10131 with diagnostic_report_diagnostic.
10132 (diagnostic_n_impl_richloc): Likewise.
10133 * diagnostic.h (report_diagnostic): Delete macro.
10134 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
10135 with diagnostic_report_diagnostic.
10136 * substring-locations.c (format_warning_va): Likewise.
10138 2017-05-05 David Malcolm <dmalcolm@redhat.com>
10140 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
10141 save/restor of format_spec. Move option-printing code to...
10142 (print_option_information): ...this new function, and
10143 reimplement by simply printing to the pretty_printer,
10144 rather than appending to the format string.
10146 2017-05-05 David Malcolm <dmalcolm@redhat.com>
10148 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
10149 handling logic into...
10150 (update_effective_level_from_pragmas): ...this new function.
10152 2017-05-04 Andrew Waterman <andrew@sifive.com>
10154 * config/riscv/riscv.opt (mstrict-align): New option.
10155 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
10156 (SLOW_UNALIGNED_ACCESS): Define.
10157 (riscv_slow_unaligned_access): Declare.
10158 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
10160 (riscv_slow_unaligned_access): New variable.
10161 (rocket_tune_info): Set slow_unaligned_access to true.
10162 (optimize_size_tune_info): Set slow_unaligned_access to false.
10163 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
10164 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
10165 (riscv_option_override): Set riscv_slow_unaligned_access.
10166 * doc/invoke.texi: Add -mstrict-align to RISC-V.
10168 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
10170 * config/riscv/riscv.md: Unify indentation.
10172 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
10177 * config/rs6000/rs6000.md (u code attribute): Add FIX and
10179 (extendsi<mode>2): Add support for doing sign extension via
10180 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
10181 don't have ISA 3.0 instructions.
10182 (extendsi<mode>2 splitter): Likewise.
10183 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
10184 generate the normal insns since SImode can now go in vector
10185 registers. Disallow the special UNSPECs needed for previous
10186 machines to hide SImode being used. Add new insns
10187 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
10188 (fix_trunc<mode>si2_stfiwx): Likewise.
10189 (fix_trunc<mode>si2_internal): Likewise.
10190 (fixuns_trunc<mode>si2): Likewise.
10191 (fixuns_trunc<mode>si2_stfiwx): Likewise.
10192 (fctiw<u>z_<mode>_smallint): Likewise.
10193 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
10194 of floating point to 32-bit integer from doing a direct move to
10195 the GPR registers to do a store.
10196 (fctiwz_<mode>): Break long line.
10198 2017-05-05 Bin Cheng <bin.cheng@arm.com>
10200 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
10201 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
10202 (addr_list, addr_offset_valid_p): New.
10203 (split_address_groups): Check offset validity with above function.
10204 (gt-tree-ssa-loop-ivopts.h): Include header file.
10206 2017-05-05 Nathan Sidwell <nathan@acm.org>
10208 * config.gcc (arm*-*-*): Add missing 'fi'.
10210 2017-05-05 Steve Ellcey <sellcey@cavium.com>
10212 * doc/invoke.texi (-fopt-info): Explicitly say order of options
10213 included in -fopt-info does not matter.
10214 * doc/optinfo.texi (-fopt-info): Fix description of default
10215 behavour. Explicitly say order of options included in -fopt-info
10218 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
10220 * config.gcc: Allow combinations of aprofile and rmprofile values for
10221 --with-multilib-list.
10222 * config/arm/t-multilib: New file.
10223 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
10224 variables. Remove setting of ISA and floating-point ABI in
10225 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
10226 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
10227 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
10228 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
10230 * config/arm/t-rmprofile: Likewise except for the matches changes.
10231 * doc/install.texi (--with-multilib-list): Document the combination of
10232 aprofile and rmprofile values and warn about pitfalls in doing that.
10234 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
10236 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
10237 (movdi_aarch64): Likewise.
10239 2017-05-05 Jakub Jelinek <jakub@redhat.com>
10241 PR tree-optimization/80632
10242 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
10244 (build_arrays): Initialize it for virtual phis.
10245 (fix_phi_nodes): Use it for virtual phis.
10247 PR tree-optimization/80558
10248 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
10249 [x, y] op z into [x op, y op z] for op & or | if conditions
10252 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
10253 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
10256 * config/arm/arm.md (use_literal_pool): Remove.
10257 (64-bit immediate split): No longer takes cost into consideration
10258 if arm_disable_literal_pool is enabled.
10259 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
10260 used when arm_disable_literal_pool is enabled.
10261 (arm_max_const_double_inline_cost): Remove use of
10262 arm_disable_literal_pool.
10263 (push_minipool_fix): Add assert.
10264 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
10265 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
10266 (no_literal_pool_sf_immediate): New.
10268 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10270 PR tree-optimization/80613
10271 * tree-ssa-dce.c (propagate_necessity): Remove cases for
10272 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
10274 2017-05-05 Richard Biener <rguenther@suse.de>
10276 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
10278 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
10280 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
10281 of this flag from insn conditions due to removal from r247495.
10283 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
10285 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
10287 (arm_early_store_addr_dep_ptr): Likewise.
10288 * config/arm/aarch-common-protos.h
10289 (arm_early_load_addr_dep_ptr): Add prototype.
10290 (arm_early_store_addr_dep_ptr): Likewise.
10291 * config/arm/cortex-a53.md: Add new bypasses.
10293 2017-05-05 Jakub Jelinek <jakub@redhat.com>
10295 * tree.c (next_type_uid): Change type to unsigned.
10296 (type_hash_canon): Decrement back next_type_uid if
10297 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
10298 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
10301 2017-05-04 Martin Sebor <msebor@redhat.com>
10303 * builtins.c: Fix a trivial typo in a comment.
10305 PR middle-end/79234
10306 * builtins.c (check_sizes): Adjust to handle reading past the end.
10307 Avoid printing excessive upper bound of ranges. Use %E to print
10308 tree nodes instead of converting them to %wu.
10309 (expand_builtin_memchr): New function.
10310 (compute_dest_size): Rename...
10311 (compute_objsize): ...to this.
10312 (expand_builtin_memcpy): Adjust.
10313 (expand_builtin_mempcpy): Adjust.
10314 (expand_builtin_strcat): Adjust.
10315 (expand_builtin_strcpy): Adjust.
10316 (check_strncat_sizes): Adjust.
10317 (expand_builtin_strncat): Adjust.
10318 (expand_builtin_strncpy): Adjust and simplify.
10319 (expand_builtin_memset): Adjust.
10320 (expand_builtin_bzero): Adjust.
10321 (expand_builtin_memcmp): Adjust.
10322 (expand_builtin): Handle memcmp.
10323 (maybe_emit_chk_warning): Check strncat just once.
10325 2017-05-04 Martin Sebor <msebor@redhat.com>
10327 PR preprocessor/79214
10328 PR middle-end/79222
10329 PR middle-end/79223
10330 * builtins.c (check_sizes): Add inlining context and issue
10331 warnings even when -Wno-system-headers is set.
10332 (check_strncat_sizes): Same.
10333 (expand_builtin_strncat): Same.
10334 (expand_builtin_memmove): New function.
10335 (expand_builtin_stpncpy): Same.
10336 (expand_builtin): Handle memmove and stpncpy.
10338 2017-05-04 Bin Cheng <bin.cheng@arm.com>
10340 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
10341 which is not used any more.
10343 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
10345 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
10347 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
10349 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
10350 (cortexa53_tunings): Likewise.
10351 (cortexa57_tunings): Likewise.
10352 (cortexa72_tunings): Likewise.
10353 (cortexa73_tunings): Likewise.
10355 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
10357 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
10358 Set loop alignment to 8.
10360 2017-05-04 Martin Sebor <msebor@redhat.com>
10362 PR translation/80280
10363 * builtins.c (expand_builtin_object_size): Add missing quoting to
10364 %D and like directives.
10365 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
10366 (hsa_type_for_tree_type): Same.
10367 (verify_function_arguments): Same.
10368 * symtab.c (symbol_table::change_decl_assembler_name): Same.
10369 * varasm.c (get_section): Same.
10372 2017-05-04 Martin Sebor <msebor@redhat.com>
10374 PR translation/80280
10375 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
10377 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
10379 * config/aarch64/aarch64.c (generic_addrcost_table):
10380 Change HI/TI mode setting.
10382 2017-05-04 Martin Jambor <mjambor@suse.cz>
10384 PR tree-optimization/80622
10385 * tree-sra.c (comes_initialized_p): New function.
10386 (build_accesses_from_assign): Only set write lazily when
10387 comes_initialized_p is false.
10388 (analyze_access_subtree): Use comes_initialized_p.
10389 (propagate_subaccesses_across_link): Assert !comes_initialized_p
10390 instead of testing for PARM_DECL.
10392 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10394 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
10395 constraint on operand 0 to allow more general addressing modes.
10396 Adjust output template.
10397 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
10399 * config/aarch64/aarch64-protos.h
10400 (aarch64_address_valid_for_prefetch_p): Declare prototype.
10401 * config/aarch64/constraints.md (Dp): New address constraint.
10402 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
10405 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
10407 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
10408 update use of estimate_ipcp_clone_size_and_time.
10409 (estimate_local_effects): Update use of
10410 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
10411 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
10412 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
10413 Return nonspecialized time.
10415 2017-05-04 Richard Biener <rguenther@suse.de>
10417 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
10418 for the last VUSE which def dominates the PHI. Directly call
10420 (get_continuation_for_phi_1): Remove.
10422 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
10424 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
10425 to explain the use of truncating division. Cap the number of
10426 iterations to the maximum given by nb_iterations_upper_bound,
10429 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
10431 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
10432 * configure: Regenerate.
10433 * config.in: Regenerate.
10434 * config/i386/driver-mingw32.c: new file.
10435 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
10436 * config.host: Link driver-mingw32.o on MinGW host.
10437 * doc/install.texi: Document new --enable-mingw-wildcard configure
10440 2017-05-04 Marek Polacek <polacek@redhat.com>
10442 PR tree-optimization/80612
10443 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
10445 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
10446 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
10448 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
10449 (movt splitter): Likewise.
10450 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
10451 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
10452 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
10453 block for Thumb-1 with MOVT.
10454 (thumb2_legitimate_address_p): Move code block ...
10455 (can_avoid_literal_pool_for_label_p): ... into this new function.
10456 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
10458 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
10459 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
10460 "M-profile targets with the MOVT instruction".
10462 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
10464 * config/arm/arm-builtins.c (arm_init_builtins): Rename
10465 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
10466 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
10468 2017-05-04 Martin Liska <mliska@suse.cz>
10470 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
10471 variable cond_code.
10473 2017-05-04 Richard Biener <rguenther@suse.de>
10475 * tree.c (array_at_struct_end_p): Handle arrays at struct
10476 end with flexarrays more conservatively. Refactor and treat
10477 arrays of arrays or aggregates more strict. Fix
10478 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
10479 * tree.c (array_at_struct_end_p): Adjust prototype.
10480 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
10481 * gimple-fold.c (get_range_strlen): Likewise.
10482 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
10484 2017-05-04 Richard Biener <rguenther@suse.de>
10486 PR tree-optimization/31130
10487 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
10489 (supports_overflow_infinity): Likewise.
10490 (is_negative_overflow_infinity): Likewise.
10491 (is_positive_overflow_infinity): Likewise.
10492 (is_overflow_infinity): Likewise.
10493 (stmt_overflow_infinity): Likewise.
10494 (overflow_infinity_range_p): Likewise.
10495 (usable_range_p): Remove as always returning true.
10496 (make_overflow_infinity): Remove.
10497 (negative_overflow_infinity): Likewise.
10498 (positive_overflow_infinity): Likewise.
10499 (avoid_overflow_infinity): Likewise.
10500 (set_value_range): Adjust accordingly.
10501 (set_value_range_to_nonnegative): Likewise, remove now unused
10502 overflow_infinity arg.
10503 (vrp_operand_equal_p): Adjust.
10504 (update_value_range): Likewise.
10505 (range_int_cst_singleton_p): Likewise.
10506 (operand_less_p): Likewise.
10507 (compare_values_warnv): Likewise.
10508 (extract_range_for_var_from_comparison_expr): Likewise.
10509 (vrp_int_const_binop): Likewise.
10510 (zero_nonzero_bits_from_vr): Likewise.
10511 (extract_range_from_multiplicative_op_1): Likewise.
10512 (extract_range_from_binary_expr_1): Likewise.
10513 (extract_range_from_unary_expr): Likewise.
10514 (extract_range_from_comparison): Likewise.
10515 (extract_range_basic): Likewise.
10516 (adjust_range_with_scev): Likewise.
10517 (compare_ranges): Likewise.
10518 (compare_range_with_value): Likewise.
10519 (dump_value_range): Likewise.
10520 (test_for_singularity): Likewise, remove strict_overflow_p parameter
10522 (simplify_cond_using_ranges): Adjust.
10524 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
10526 * brig-builtins.def: Added a builtin for class_f64.
10527 * builtin-types.def: Added a builtin type needed by class_f64.
10529 2017-05-03 Jason Merrill <jason@redhat.com>
10531 * timevar.def: Add TV_CONSTEXPR.
10533 2017-05-03 David Malcolm <dmalcolm@redhat.com>
10535 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
10537 2017-05-03 Martin Jambor <mjambor@suse.cz>
10539 * ipa-prop.c (ipa_update_after_lto_read): Removed.
10540 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
10541 * ipa-cp.c (ipcp_propagate_stage): Do not call
10542 ipa_update_after_lto_read.
10543 * ipa-inline.c (ipa_inline): Likewise.
10545 2017-05-03 Martin Jambor <mjambor@suse.cz>
10547 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
10548 tag. Added a default constructor and a destructor.
10549 (ipa_edge_args_sum_t): New class;
10550 (ipa_edge_args_sum): Declare.
10551 (ipa_edge_args_vector): Remove declaration.
10552 (IPA_EDGE_REF): Use ipa_edge_args_sum.
10553 (ipa_free_edge_args_substructures): Remove declaration.
10554 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
10555 (ipa_edge_args_info_available_for_edge_p): Likewise.
10556 * ipa-prop.c (ipa_edge_args_vector): Removed.
10557 (edge_removal_hook_holder): Likewise.
10558 (edge_duplication_hook_holder): Likewise.
10559 (ipa_edge_args_sum): New variable.
10560 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
10561 ipa_edge_args_vector.
10562 (ipa_free_edge_args_substructures): Likewise.
10563 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
10564 ipa_edge_args_vector.
10565 (ipa_edge_removal_hook): Turned into method
10566 ipa_edge_args_sum_t::remove.
10567 (ipa_edge_duplication_hook): Turned into method
10568 ipa_edge_args_sum_t::duplicate.
10569 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
10570 registering edge hooks.
10571 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
10572 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
10573 ipa_edge_args_sum instead of ipa_edge_args_vector.
10574 * ipa-profile.c (ipa_profile): Likewise.
10576 2017-05-03 Martin Jambor <mjambor@suse.cz>
10578 * symbol-summary.h (function_summary): New method exists.
10579 (function_summary::symtab_removal): Deallocate through release.
10580 (call_summary): New class.
10581 (gt_ggc_mx): New overload.
10582 (gt_pch_nx): Likewise.
10583 (gt_pch_nx): Likewise.
10585 2017-05-03 Jeff Law <law@redhat.com>
10587 PR tree-optimization/78496
10588 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
10589 from simplify_cond_using_ranges. Split off code to walk
10590 backwards through casts into ...
10591 (simplify_cond_using_ranges_2): New function.
10592 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
10593 (execute_vrp): After identifying jump threads, call
10594 simplify_cond_using_ranges_2.
10596 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
10599 * ipa-inline.h (inline_summary): Add ctor.
10600 (create_ggc): Do not use ggc_cleared_alloc.
10602 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
10603 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10605 * gcc.c (handle_braces): Support escaping in switch matching
10607 * doc/invoke.texi (Spec Files): Document it.
10608 Remove superfluous @code markup in items.
10610 2017-05-03 David Malcolm <dmalcolm@redhat.com>
10612 * diagnostic-show-locus.c (struct column_range): New struct.
10613 (get_affected_columns): New function.
10614 (get_printed_columns): New function.
10615 (struct correction): New struct.
10616 (correction::ensure_capacity): New function.
10617 (correction::ensure_terminated): New function.
10618 (struct line_corrections): New struct.
10619 (line_corrections::~line_corrections): New dtor.
10620 (line_corrections::add_hint): New function.
10621 (layout::print_trailing_fixits): Reimplement in terms of the new
10623 (selftest::test_overlapped_fixit_printing): New function.
10624 (selftest::diagnostic_show_locus_c_tests): Call it.
10626 2017-05-03 Nathan Sidwell <nathan@acm.org>
10628 Canonicalize canonical type hashing
10629 * tree.h (type_hash_canon_hash): Declare.
10630 * tree.c (type_hash_list, attribute_hash_list): Move into
10631 type_hash_canon_hash.
10632 (build_type_attribute_qual_variant): Break out hash code calc into
10633 type_hash_canon_hash.
10634 (type_hash_canon_hash): New. Generic type hash computation.
10635 (build_range_type_1, build_array_type_1, build_function_type,
10636 build_method_type_directly, build_offset_type, build_complex_type,
10637 make_vector_type): Call it.
10639 2017-05-03 Richard Biener <rguenther@suse.de>
10641 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
10642 When all DRs have unknown misaligned do not always peel
10643 when there is a store but apply the same costing model as if
10644 there were only loads.
10646 2017-05-03 Richard Biener <rguenther@suse.de>
10649 PR tree-optimization/80492
10650 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
10651 compare_base_decls returning dont-know properly.
10653 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
10655 * config/arm/iterators.md (CCSI): New mode iterator.
10656 (arch): New mode attribute.
10657 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
10658 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
10659 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
10660 code iterator for success result mode.
10661 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
10662 the corresponding new insn generators.
10664 2017-05-03 Bin Cheng <bin.cheng@arm.com>
10667 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10668 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
10670 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
10672 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
10673 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
10674 (DDR_A): Wrap DDR argument in brackets.
10675 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
10676 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
10677 (DDR_REVERSED_P): Likewise.
10679 2017-05-03 Jakub Jelinek <jakub@redhat.com>
10681 PR tree-optimization/79472
10682 * tree-switch-conversion.c (struct switch_conv_info): Add
10683 contiguous_range and default_case_nonstandard fields.
10684 (collect_switch_conv_info): Compute contiguous_range and
10685 default_case_nonstandard fields, don't clear final_bb if
10686 contiguous_range and only the default case doesn't have the required
10688 (check_all_empty_except_final): Set default_case_nonstandard instead
10689 of failing if contiguous_range and the default case doesn't have empty
10691 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
10692 and only the default case doesn't have the required constants. Skip
10694 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
10695 if default_case_nonstandard.
10696 (build_constructors): Build constant 1 just once. Assert that default
10697 values aren't inserted in between cases if contiguous_range. Skip
10699 (build_arrays): Skip virtual phis.
10700 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
10701 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
10702 Handle virtual phis.
10703 (gen_inbound_check): Handle default_case_nonstandard case.
10704 (process_switch): Adjust check_final_bb caller. Call
10705 gather_default_values with the first non-default case instead of
10706 default case if default_case_nonstandard.
10708 2017-05-02 Nathan Sidwell <nathan@acm.org>
10710 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
10711 check. Fix formatting.
10713 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
10715 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
10716 errors when comparing specialized and unspecialized times.
10718 2017-05-02 David Malcolm <dmalcolm@redhat.com>
10720 * diagnostic-show-locus.c
10721 (layout::should_print_annotation_line_p): Make private.
10722 (layout::print_annotation_line): Make private.
10723 (layout::annotation_line_showed_range_p): Make private.
10724 (layout::show_ruler): Make private.
10725 (layout::print_source_line): Make private. Pass in line and
10726 line_width, rather than calling location_get_source_line. Drop
10728 (layout::print_leading_fixits): New method.
10729 (layout::print_any_fixits): Rename to...
10730 (layout::print_trailing_fixits): ...this, and make private.
10731 Don't print newline fixits.
10732 (diagnostic_show_locus): Move logic for printing one row into...
10733 (layout::print_line): ...this new function. Move the
10734 location_get_source_line call and error-handling from
10735 print_source_line to here. Call print_leading_fixits, and rename
10736 print_any_fixits to print_trailing_fixits.
10737 (selftest::test_fixit_insert_containing_newline): Update now that
10738 newlines are partially supported.
10739 (selftest::test_fixit_insert_containing_newline_2): New test.
10740 (selftest::test_fixit_replace_containing_newline): Update comments.
10741 (selftest::diagnostic_show_locus_c_tests): Call the new test.
10742 * edit-context.c (class added_line): New class.
10743 (class edited_line): Describe newline handling in comment.
10744 (edited_line::actually_edited_p): New method.
10745 (edited_line::print_content): Delete redundant decl.
10746 (edited_line::m_predecessors): New field.
10747 (edited_file::print_content): Call edited_line::print_content.
10748 (edited_file::print_diff): Update to support newlines.
10749 (edited_file::print_diff_hunk): Likewise.
10750 (edited_file::print_run_of_changed_lines): New function.
10751 (edited_file::print_diff_line): Convert to...
10752 (print_diff_line): ...this.
10753 (edited_file::get_effective_line_count): New function.
10754 (edited_line::edited_line): Initialize new field m_predecessors.
10755 (edited_line::~edited_line): Clean up m_predecessors.
10756 (edited_line::apply_fixit): Handle newlines.
10757 (edited_line::get_effective_line_count): New function.
10758 (edited_line::print_content): New function.
10759 (edited_line::print_diff_lines): New function.
10760 (selftest::test_applying_fixits_insert_containing_newline): New
10762 (selftest::test_applying_fixits_replace_containing_newline): New
10764 (selftest::insert_line): New function.
10765 (selftest::test_applying_fixits_multiple_lines): Add example of
10767 (selftest::edit_context_c_tests): Call the new tests.
10769 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10771 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
10772 parameter cand. Update dump information.
10773 (get_computation_cost): Update uses.
10775 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10777 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
10778 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
10779 (get_computation_at, rewrite_use_address): Update use of
10780 get_computation_aff.
10782 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10784 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
10785 (get_computation): Delete.
10786 (get_computation_cost): Implement like get_computation_cost_at.
10787 Use get_computation_at.
10788 (get_computation_cost_at): Delete.
10789 (rewrite_use_nonlinear_expr): Use get_computation_at.
10790 (rewrite_use_compare, remove_unused_ivs): Ditto.
10792 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10794 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
10796 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10798 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
10799 (ivopts_global_cost_for_size): Rename parameter and update uses.
10800 (iv_ca_recount_cost): Update uses.
10801 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
10802 candidates seperately in n_invs and n_cands.
10803 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
10805 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10807 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
10808 (find_inv_vars_cb): New.
10809 (find_depends): Renamed to ...
10810 (find_inv_vars): ... this.
10811 (add_candidate_1, force_var_cost): Call find_inv_vars.
10812 (split_address_cost, determine_group_iv_cost_cond): Ditto.
10814 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10816 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
10817 inv_vars. Add inv_exprs.
10818 (struct iv_cand): Rename depends_on to inv_vars.
10819 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
10820 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
10821 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
10822 (dump_cand): Dump inv_vars.
10823 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
10824 (record_invariant, find_depends, add_candidate_1): Ditto.
10825 (set_group_iv_cost, force_var_cost): Ditto.
10826 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
10827 (get_computation_cost_at, get_computation_cost): Ditto.
10828 (determine_group_iv_cost_generic): Ditto.
10829 (determine_group_iv_cost_address): Ditto.
10830 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
10831 (determine_group_iv_costs): Ditto.
10832 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
10833 (iv_ca_set_remove_invariants): Renamed to ...
10834 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
10835 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
10836 (iv_ca_set_add_invariants): Renamed to ...
10837 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
10838 (iv_ca_set_cp): Use iv_ca_set_add_invs.
10839 (iv_ca_has_deps): Support inv_vars and inv_exprs.
10840 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
10841 (create_new_ivs): Remove useless dump.
10843 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10845 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
10847 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
10848 (iv_ca_set_no_cp, create_new_iv): Ditto.
10850 2017-05-02 Bin Cheng <bin.cheng@arm.com>
10852 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
10854 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
10856 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
10857 function tree_check2.
10859 2017-05-02 Martin Liska <mliska@suse.cz>
10861 * doc/gcov.texi: Add missing preposition.
10862 * gcov.c (function_info::function_info): Properly fill up
10863 all member variables.
10865 2017-05-02 Tamar Christina <tamar.christina@arm.com>
10867 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
10869 2017-05-02 Tamar Christina <tamar.christina@arm.com>
10871 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
10873 2017-05-02 Martin Liska <mliska@suse.cz>
10876 * lto-streamer-in.c (lto_read_tree_1): Remove
10877 LTO_STREAMER_DEBUG.
10878 * lto-streamer.c (struct tree_hash_entry): Likewise.
10879 (struct tree_entry_hasher): Likewise.
10880 (tree_entry_hasher::hash): Likewise.
10881 (tree_entry_hasher::equal): Likewise.
10882 (lto_streamer_init): Likewise.
10883 (lto_orig_address_map): Likewise.
10884 (lto_orig_address_get): Likewise.
10885 (lto_orig_address_remove): Likewise.
10886 * lto-streamer.h: Likewise.
10887 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
10888 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
10890 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
10892 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
10893 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
10894 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
10895 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
10896 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
10897 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
10898 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
10899 (mm_maskz_sub_ss): New intrinsics.
10900 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
10901 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
10902 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
10903 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
10904 (__builtin_ia32_subss_mask_round): New builtins.
10905 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
10906 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
10907 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
10909 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
10910 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
10912 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
10915 2017-05-02 Martin Jambor <mjambor@suse.cz>
10917 PR tree-optimization/78687
10918 * tree-sra.c (access): New field parent.
10919 (process_subtree_disqualification): New function.
10920 (disqualify_candidate): Call it.
10921 (build_accesses_from_assign): Reset write flag if creating an
10923 (build_access_subtree): Fill in parent field and also prpagate
10924 down grp_write flag.
10925 (create_artificial_child_access): New parameter set_grp_write, set
10926 grp_write to its value.
10927 (propagate_subaccesses_across_link): Also propagate grp_write flag
10929 (propagate_all_subaccesses): Push the closest parent back to work
10930 queue if add_access_to_work_queue returned true.
10932 2017-05-02 Richard Biener <rguenther@suse.de>
10934 * common.opt (fstrict-overflow): Alias negative to fwrapv.
10935 * doc/invoke.texi (fstrict-overflow): Remove all traces of
10936 -fstrict-overflow documentation.
10937 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
10938 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
10939 flag_strict_overflow.
10940 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
10941 * lto-opts.c (lto_write_options): Do not stream it.
10942 * lto-wrapper.c (merge_and_complain): Do not handle it.
10943 * opts.c (default_options_table): Do not set -fstrict-overflow.
10944 (finish_options): Likewise do not clear it when sanitizing.
10945 * simplify-rtx.c (simplify_const_relational_operation): Do not
10946 test flag_strict_overflow.
10948 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
10950 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
10951 using enabled attribute.
10952 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
10953 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
10954 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
10955 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
10956 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
10957 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
10958 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
10959 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
10960 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
10961 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
10963 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
10965 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
10967 2017-05-02 Richard Biener <rguenther@suse.de>
10969 PR tree-optimization/80591
10971 2017-04-10 Richard Biener <rguenther@suse.de>
10973 * tree-ssa-structalias.c (find_func_aliases): Properly handle
10976 2017-05-02 Richard Biener <rguenther@suse.de>
10978 PR tree-optimization/80549
10979 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
10980 (cleanup_tree_cfg_noloop): Create forwarders to known loop
10981 headers if they do not have a preheader.
10983 2017-05-02 Martin Liska <mliska@suse.cz>
10986 * common.opt: Fix typo.
10987 * doc/invoke.texi: Likewise.
10989 2017-05-01 Jan Beulich <jbeulich@suse.com>
10991 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
10992 swapping, add (x,x,m,x,n) alternative.
10994 2017-05-01 Nathan Sidwell <nathan@acm.org>
10996 * calls.c (combine_pending_stack_adjustment_and_call): Remove
10997 unnecessary unadjusted_alignment check.
10999 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
11002 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
11004 * gimplify.c (gimplify_cilk_detach): New function.
11005 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
11006 * tree-core.h: Document EXPR_CILK_SPAWN.
11007 * tree.h (EXPR_CILK_SPAWN): Define.
11009 2017-05-01 David Malcolm <dmalcolm@redhat.com>
11011 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
11012 to use new fixit_hint representation, using the "replace" logic.
11013 (get_line_span_for_fixit_hint): Likewise.
11014 (layout::print_any_fixits): Likewise.
11015 (selftest::test_one_liner_many_fixits): Rename to...
11016 (selftest::test_one_liner_many_fixits_1): ...this, and update
11017 comment and expected output to reflect that the multiple fix-it
11018 hints are now consolidated into one insertion.
11019 (selftest::test_one_liner_many_fixits_2): New test.
11020 (selftest::test_diagnostic_show_locus_one_liner): Update for
11022 (selftest::test_fixit_consolidation): Update for fix-it API
11024 * diagnostic.c (print_parseable_fixits): Likewise.
11025 * edit-context.c (edited_line::m_line_events): Convert from
11026 auto_vec <line_event *> to auto_vec <line_event>.
11027 (class line_event): Convert from abstract base class to a concrete
11028 class, taking over the role of replace_event.
11029 (class insert_event): Delete.
11030 (class replace_event): Rename to class line_event. Convert to
11032 (edit_context::add_fixits): Reimplement.
11033 (edit_context::apply_insert): Delete.
11034 (edit_context::apply_replace): Rename to...
11035 (edit_context::apply_fixit): ...this. Convert to half-open range.
11036 (edited_file::apply_insert): Delete.
11037 (edited_file::apply_replace): Rename to...
11038 (edited_file::apply_fixit): ...this.
11039 (edited_line::~edited_line): Drop deletion of events.
11040 (edited_line::apply_insert): Delete.
11041 (edited_line::apply_replace): Rename to...
11042 (edited_line::apply_fixit): ...this. Convert to half-open range.
11043 Update for change to type of m_line_events.
11044 * edit-context.h (edit_context::apply_insert): Delete.
11045 (edit_context::apply_replace): Rename to...
11046 (edit_context::apply_fixit): ...this.
11048 2017-05-01 Martin Sebor <msebor@redhat.com>
11050 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
11053 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
11056 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
11057 __get_cpuid_max returns 0.
11058 (__get_cpuid_count): Ditto.
11060 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
11062 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
11063 replacement expression is another instance of one of its arguments.
11065 2017-05-01 Jakub Jelinek <jakub@redhat.com>
11068 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
11069 check for stack push/pop autoinc.
11070 * config/i386/i386.c (ix86_agi_dependent): Return false
11071 if the only reason why modified_in_p returned true is that
11072 addr is SP based and set_insn is a push or pop.
11074 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
11076 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
11079 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
11082 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
11083 (account_size_time): Use two predicates - exec_pred and
11085 (evaluate_conditions_for_known_args): Compute both clause and
11087 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
11088 (inline_summary_t::duplicate): Update.
11089 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
11091 (compute_inline_parameters): Likewise.
11092 (estimate_edge_size_and_time): Update caluclation of time.
11093 (estimate_node_size_and_time): Compute both time and nonspecialized
11095 (estimate_ipcp_clone_size_and_time): Update.
11096 (inline_merge_summary): Update.
11097 (do_estimate_edge_time): Update.
11098 (do_estimate_edge_size): Update.
11099 (do_estimate_edge_hints): Update.
11100 (inline_read_section, inline_write_summary): Stream both new predicates.
11101 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
11103 (compute_inlined_call_time): Cleanup.
11104 (big_speedup_p): Update.
11105 (edge_badness): Update.
11106 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
11107 (size_time_entry): Replace predicate by exec_predicate and
11108 nonconst_predicate.
11109 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
11110 (estimate_edge_time): Return also nonspec_time.
11111 (reset_edge_growth_cache): Update.
11113 2017-04-29 Jakub Jelinek <jakub@redhat.com>
11115 PR rtl-optimization/80491
11116 * ifcvt.c (noce_process_if_block): When looking for x setter
11117 with missing else_bb, don't check only the insn right before
11118 cond_earliest, but look for the last insn that x is modified in
11119 within the same bb.
11121 PR rtl-optimization/80491
11122 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
11124 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
11126 PR tree-optimization/80487
11127 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
11129 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
11131 PR tree-optimization/79697
11132 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
11133 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
11134 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
11136 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
11137 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
11139 2017-04-28 Martin Sebor <msebor@redhat.com>
11141 PR tree-optimization/80523
11142 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
11143 (init_target_to_host_charmap, target_to_host, target_strtol10): New
11145 (maybe_warn, format_directive, parse_directive): Use new functions.
11146 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
11148 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
11150 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
11152 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
11154 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
11155 target_header_dir): Set correctly.
11156 * configure: Regenerated.
11157 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
11158 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
11159 instead of SYSTEM_HEADER_DIR.
11161 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
11163 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
11164 (estimate_local_effects): Likewise.
11165 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
11166 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
11167 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
11168 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
11169 do_estimate_edge_time, estimate_edge_time): Likewise.
11170 * ipa-inline-analysis.c (estimate_node_size_and_time,
11171 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
11172 (estimate_time_after_inlining): Remove.
11174 2017-04-28 Martin Liska <mliska@suse.cz>
11176 * doc/gcov.texi: Enhance documentation of gcov.
11178 2017-04-28 Martin Liska <mliska@suse.cz>
11180 * doc/gcov.texi: Sort options in alphabetic order.
11181 * doc/gcov-dump.texi: Likewise.
11182 * doc/gcov-tool.texi: Likewise.
11183 * gcov.c (print_usage): Likewise.
11184 * gcov-dump.c (print_usage): Likewise.
11185 * gcov-tool.c (print_merge_usage_message): Likewise.
11186 (print_rewrite_usage_message): Likewise.
11187 (print_overlap_usage_message): Likewise.
11189 2017-04-28 Martin Liska <mliska@suse.cz>
11191 PR gcov-profile/53915
11192 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
11194 2017-04-28 Martin Liska <mliska@suse.cz>
11196 PR gcov-profile/79891
11197 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
11198 is marked by compiler as living on a line.
11199 (get_cycles_count): Remove usage of the union.
11200 (output_intermediate_file): Likewise.
11201 (find_source): Fix GNU coding style.
11202 (accumulate_line_counts): Remove old non-all block mode.
11203 (output_lines): Remove usage of the union.
11204 * profile.c (output_location): Include all BBs, even if
11205 belonging to a same line (and file) as a previous BB.
11207 2017-04-28 Martin Liska <mliska@suse.cz>
11209 * gcov.c (process_args): Handle new argument 'w'.
11210 (read_graph_file): Assign ID to BBs.
11211 (output_branch_count): Display BB # if verbose flag is set.
11212 (output_lines): Likewise for arcs.
11213 (print_usage): Add '--verbose' option help.
11214 * doc/gcov.texi: Document --verbose (-w) option.
11216 2017-04-28 Martin Liska <mliska@suse.cz>
11218 * gcov.c (struct block_location_info): New struct.
11219 (process_file): Fill up the new structure.
11220 (read_graph_file): Replace usage of encoding by the newly added
11222 (add_line_counts): Likewise.
11223 (accumulate_line_counts): Remove usage of the union.
11224 (function_info::function_info): New function.
11225 (function_info::~function_info): Likewise.
11226 (process_file): Call delete instead of release_function.
11227 (release_function): Release the function.
11228 (release_structures): Call delete instead of release_function.
11229 (solve_flow_graph): Replace usage of num_blocks.
11230 (find_exception_blocks): Likewise.
11231 (output_lines): Fix GNU coding style.
11233 2017-04-28 Martin Liska <mliska@suse.cz>
11236 * coverage.c (coverage_remove_note_file): New function.
11237 * coverage.h: Declare the function.
11238 * toplev.c (finalize): Clean if an error has been seen.
11240 2017-04-28 Martin Liska <mliska@suse.cz>
11242 PR gcov-profile/80031
11243 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
11244 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
11245 * gcov.c (read_graph_file): Read just number of blocks.
11246 * profile.c (branch_prob): Do not stream 0 flags per a basic
11249 2017-04-28 Martin Liska <mliska@suse.cz>
11251 * gcov-dump.c (tag_*): Add new argument to declarations.
11252 (dump_gcov_file): Likewise.
11253 (tag_blocks): Add and use new argument depth.
11254 (tag_arcs): Likewise.
11255 (tag_lines): Likewise.
11256 (tag_counters): Likewise.
11257 (tag_summary): Likewise.
11258 (dump_working_sets): Use depth to do a proper indentation.
11260 2017-04-28 Jakub Jelinek <jakub@redhat.com>
11263 * cgraph.h (symtab_node::debug_symtab): No longer inline.
11264 * symtab.c (symtab_node::debug_symtab): Move definition here.
11266 2017-04-28 Richard Biener <rguenther@suse.de>
11268 * lto-streamer.h (LTO_major_version): Bump to 7.
11270 2017-04-28 Richard Biener <rguenther@suse.de>
11272 * tree-vrp.c (assert_info): New struct.
11273 (add_assert_info): New helper.
11274 (register_edge_assert_for_2): Refactor to add asserts to a vector
11276 (register_edge_assert_for_1): Likewise.
11277 (register_edge_assert_for): Likewise.
11278 (finish_register_edge_assert_for): New helper actually registering
11279 asserts where live on edge.
11280 (find_conditional_asserts): Adjust.
11281 (find_switch_asserts): Likewise.
11282 (evrp_dom_walker::try_find_new_range): Generalize.
11283 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
11285 2017-04-27 Marek Polacek <polacek@redhat.com>
11288 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
11289 arg10 and arg11 to itype.
11291 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
11293 * doc/extend.texi (Object Size Checking): Improve grammar.
11295 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
11298 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
11299 that the logic for permitting reciprocal estimates matches that
11302 2017-04-27 Jakub Jelinek <jakub@redhat.com>
11305 * tree.c (type_cache_hasher::equal): Only compare
11306 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
11307 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
11308 non-aggregate element types.
11309 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
11310 about the flag on ARRAY_TYPEs in the comment, formatting fix.
11312 2017-04-27 Richard Biener <rguenther@suse.de>
11314 PR middle-end/80533
11315 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
11316 stripping ARRAY_REFs from MEM_EXPR make sure we're not
11317 keeping a reference to a trailing array.
11319 2017-04-27 Richard Biener <rguenther@suse.de>
11321 PR middle-end/80539
11322 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
11323 being in loop-closed SSA form conservatively.
11324 (chrec_fold_multiply_poly_poly): Likewise.
11326 2017-04-27 Tamar Christina <tamar.christina@arm.com>
11328 PR middle-end/79665
11329 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
11330 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
11332 2017-04-27 Jakub Jelinek <jakub@redhat.com>
11335 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
11336 (aarch64_function_arg_alignment): Return unsigned int again, but still
11337 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
11338 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
11339 Don't emit -Wpsabi note.
11340 (aarch64_function_arg_boundary): Likewise.
11341 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
11344 2017-04-26 Nathan Sidwell <nathan@acm.org>
11346 * tree.h (crc32_unsigned_n): Declare.
11347 (crc32_unsigned, crc32_unsigned): Make inline.
11348 * tree.c (crc32_unsigned_bits): Replace with ...
11349 (crc32_unsigned_n): ... this.
11350 (crc32_unsigned, crc32_byte): Remove.
11351 (crc32_string): Remove unnecessary braces.
11353 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
11355 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
11356 * ipa-inline-analysis.c (MAX_TIME): Remove.
11357 (account_size_time): Use sreal for time.
11358 (dump_inline_summary): Update.
11359 (estimate_function_body_sizes): Update.
11360 (estimate_edge_size_and_time): Update.
11361 (estimate_calls_size_and_time): Update.
11362 (estimate_node_size_and_time): Update.
11363 (inline_merge_summary): Update.
11364 (inline_update_overall_summary): Update.
11365 (estimate_time_after_inlining): Update.
11366 (inline_read_section): Update.
11367 (inline_write_summary): Update.
11368 * ipa-inline.c (compute_uninlined_call_time): Update.
11369 (compute_inlined_call_time): Update.
11370 (recursive_inlining): Update.
11371 (inline_small_functions): Update.
11372 (dump_overall_stats): Update.
11373 * ipa-inline.h: Include sreal.h.
11374 (size_time_entry): Turn time to sreal.
11375 (inline_summary): Turn self_time nad time to sreal.
11377 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
11379 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
11381 (sreal::stream_out, sreal::stream_in): New.
11382 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
11384 2017-04-25 Jakub Jelinek <jakub@redhat.com>
11386 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
11389 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
11392 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
11394 (dimode_scalar_chain::compute_convert_gain): Ditto.
11395 (dimode_scalar_chain::make_vector_copies): Ditto.
11396 (dimode_scalar_chain::convert_reg): Ditto.
11397 (dimode_scalar_chain::convert_insn): Ditto.
11398 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
11399 (VI248_AVX512BW_1): New mode iterator.
11400 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
11401 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
11404 2017-04-25 Martin Sebor <msebor@redhat.com>
11406 PR tree-optimization/80497
11407 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
11408 constants are representable in HOST_WIDE_INT.
11409 (parse_directive): Ditto.
11411 2017-04-25 Martin Sebor <msebor@redhat.com>
11414 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
11415 (new_zero_array): Adjust signature.
11416 (dom_info::dom_init): Used unsigned rather that size_t.
11417 (dom_info::dom_info): Same.
11419 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11420 Jakub Jelinek <jakub@redhat.com>
11423 * config/arm/arm.c: Include gimple.h.
11424 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
11425 returns negative, increment ncrn only if it returned positive.
11426 (arm_needs_doubleword_align): Return int instead of bool,
11427 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
11428 members, but if there is any such non-FIELD_DECL
11429 > PARM_BOUNDARY aligned decl, return -1 instead of false.
11430 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
11431 returns negative, increment nregs only if it returned positive.
11432 (arm_setup_incoming_varargs): Likewise.
11433 (arm_function_arg_boundary): Emit -Wpsabi note if
11434 arm_needs_doubleword_align returns negative, return
11435 DOUBLEWORD_ALIGNMENT only if it returned positive.
11437 2017-04-25 Marek Polacek <polacek@redhat.com>
11440 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
11441 first argument to type.
11443 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
11446 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
11447 type checks to test for compatibility instead of equality.
11449 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11450 Jakub Jelinek <jakub@redhat.com>
11453 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
11455 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
11456 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
11457 the alignment computation, but return their maximum in warn_alignment.
11458 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
11459 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
11461 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
11462 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
11465 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11467 * config/arc/simdext.md (dmpyh): Fix typo.
11469 2017-04-25 Richard Biener <rguenther@suse.de>
11471 PR tree-optimization/80492
11472 * alias.c (compare_base_decls): Handle registers with asm
11473 specification conservatively.
11474 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
11475 compare_base_decls returning dont-know properly.
11477 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11479 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
11480 (legitimate_offset_address_p): New function.
11481 (arc_legitimate_address_p): Use above function.
11483 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11485 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
11487 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11489 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
11490 ACCH registers whenever they are available.
11492 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11494 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
11495 double regs fix when not used.
11497 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11499 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
11501 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
11502 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
11504 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11506 * config/arc/arc.c (arc_output_addsi): Check for h-register class
11507 when emitting short ADD instructions.
11509 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
11511 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
11513 (cmpsi_cc_c_insn): Likewise.
11514 (cbranchsi4_scratch): Compute proper instruction length using
11515 compact_hreg_operand.
11516 * config/arc/predicates.md (compact_hreg_operand): New predicate.
11518 2017-04-25 Richard Biener <rguenther@suse.de>
11520 PR middle-end/80509
11521 * passes.c (pass_manager::pass_manager): Initialize
11522 m_name_to_pass_map.
11524 2017-04-25 Richard Biener <rguenther@suse.de>
11526 PR tree-optimization/79201
11527 * tree-ssa-sink.c (statement_sink_location): Handle calls.
11529 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11532 * config/s390/vector.md: Split MEM->GPR vector moves for
11533 non-s_operand addresses.
11535 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11538 * config/s390/predicates.md (reload_const_wide_int_operand): New
11540 * config/s390/s390.md ("movti"): Remove d/P alternative.
11541 ("movti_bigconst"): New pattern definition.
11543 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
11546 * s390-protos.h (s390_expand_cs_hqi): Removed.
11547 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
11548 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
11549 modes as well as CCZ1mode and CCZmode.
11550 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
11551 signature of s390_emit_compare_and_swap.
11552 (s390_expand_cs_hqi): Likewise, make static.
11553 (s390_expand_cs_tdsi): Generate an explicit compare before trying
11554 compare-and-swap, in some cases.
11555 (s390_expand_cs): Wrapper function.
11556 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
11558 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
11559 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
11560 patterns for small and large integers. Forbid symref memory operands.
11561 Move expander to s390.c. Require cc register.
11562 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
11563 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
11564 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
11565 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
11566 symref memory operands. Remove CC mode and call s390_match_ccmode
11568 ("atomic_exchange<mode>"): Allow and implement all integer modes.
11570 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
11572 * config/s390/s390.md (define_peephole2): New peephole to help
11573 combining the load-and-test pattern with volatile memory.
11575 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
11577 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
11578 with CCZmode for TARGET_Z196.
11580 2017-04-25 Jakub Jelinek <jakub@redhat.com>
11582 PR rtl-optimization/80501
11583 * combine.c (make_compound_operation_int): Set subreg_code to SET
11584 even for AND with mask of the sign bit of mode.
11586 PR rtl-optimization/80500
11587 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
11588 sum's initial value.
11590 2017-04-25 Julian Brown <julian@codesourcery.com>
11591 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11593 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
11595 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
11597 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
11599 2017-04-25 Julian Brown <julian@codesourcery.com>
11600 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11602 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
11603 (thunderx2t99_sha): New Reservation.
11605 2017-04-25 Julian Brown <julian@codesourcery.com>
11606 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11608 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
11609 type for 1-element load.
11611 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
11613 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
11615 2017-04-24 Martin Jambor <mjambor@suse.cz>
11617 PR tree-optimization/80293
11618 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
11619 char arrays not totally scalarizable if it is false.
11620 (analyze_all_variable_accesses): Pass correct value in the new
11621 parameter. Add a statistics counter.
11623 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
11625 PR middle-end/79931
11626 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
11628 2017-04-24 Richard Biener <rguenther@suse.de>
11630 PR tree-optimization/80494
11631 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
11632 out for complex types.
11634 2017-04-24 Richard Biener <rguenther@suse.de>
11636 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
11637 * tree-ssa-sccvn.c (print_scc): Print SCC size.
11638 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
11639 (DFS): Adjust and never fail.
11640 (sccvn_dom_walker::fail): Remove.
11641 (sccvn_dom_walker::before_dom_children): Adjust.
11642 (run_scc_vn): Likewise and never fail.
11643 * tree-ssa-pre.c (pass_pre::execute): Adjust.
11644 (pass_fre::execute): Likewise.
11646 2017-04-24 Richard Biener <rguenther@suse.de>
11648 PR tree-optimization/79725
11649 * tree-ssa-sink.c (statement_sink_location): Return whether
11650 failure reason was zero uses. Move that check later.
11651 (sink_code_in_bb): Deal with zero uses by removing the stmt
11654 2017-04-24 Richard Biener <rguenther@suse.de>
11657 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
11658 pointer-based references.
11660 2017-04-24 Richard Biener <rguenther@suse.de>
11663 * pass_manager.h (pass_manager::operator new): Remove.
11664 (pass_manager::operator delete): Likewise.
11665 * passes.c (pass_manager::operator new): Remove.
11666 (pass_manager::operator delete): Likewise.
11667 (pass_manager::pass_manager): Zero individual pass members.
11669 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
11672 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
11673 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
11674 Check "XEXP (src, 1)" operand here.
11675 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
11676 Check "XEXP (src, 1)" operand here.
11677 (dimode_scalar_chain::make_vector_copies): Detect count register
11678 of a shift instruction. Zero extend count register from QImode
11679 to DImode to satisfy vector shift pattern count operand predicate.
11680 Substitute vector shift count operand with a DImode copy.
11681 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
11684 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
11686 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
11687 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
11688 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
11689 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
11690 (UNSPEC_NOREX_MEM): Remove definition.
11692 2017-04-21 Richard Biener <rguenther@suse.de>
11694 PR tree-optimization/79547
11695 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
11696 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
11697 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
11698 without any constraints.
11700 2017-04-21 Richard Biener <rguenther@suse.de>
11702 PR tree-optimization/78847
11703 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
11705 2017-04-21 Richard Biener <rguenther@suse.de>
11707 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
11708 (build_distinct_type_copy): Likewise.
11709 (build_variant_type_copy): Likewise.
11710 * tree.c (build_qualified_type): Pass down mem-stat info.
11711 (build_distinct_type_copy): Likewise.
11712 (build_variant_type_copy): Likewise.
11714 2017-04-21 Richard Biener <rguenther@suse.de>
11716 PR tree-optimization/80237
11717 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
11719 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
11720 for a simplified result.
11722 2016-04-21 Richard Biener <rguenther@suse.de>
11724 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
11725 sth as strict as a simple_iv but a chrec without symbols and an
11726 operand defined in the loop we are peeling (and not some subloop).
11727 (propagate_constants_for_unrolling): Propagate all constants.
11729 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
11732 * config/i386/i386.c (print_reg): Remove assert for disalowed
11733 regno values, call output_operand_lossage instead.
11735 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
11738 * config/i386/constraints.md (Yc): New register constraint.
11739 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
11740 Use Yc constraint for alternative 2 of operand 0. Remove
11741 preferred_for_speed attribute.
11743 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
11745 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
11746 lastprivate clauses in SIMT case.
11748 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
11750 * doc/invoke.texi (-Wextra-semi): Document new warning option.
11752 2017-04-20 Richard Biener <rguenther@suse.de>
11754 PR tree-optimization/57796
11755 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
11756 as N scalar stores.
11757 (vect_model_load_cost): Cost gathers as N scalar loads.
11759 2017-04-20 Richard Biener <rguenther@suse.de>
11761 * ggc-page.c (ggc_allocated_p): Rename to ...
11762 (safe_lookup_page_table_entry): ... this and return the lookup
11764 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
11766 2017-04-20 Richard Biener <rguenther@suse.de>
11768 PR tree-optimization/80453
11769 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
11770 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
11771 from the conditions.
11772 (vn_phi_eq): Pass them down.
11773 (vn_phi_lookup): Record them.
11774 (vn_phi_insert): Likewise.
11776 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
11778 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
11779 uninitialized variable warning to avoid buffer overrun.
11781 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
11784 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
11785 is suppressed for '{ 0 }' in C.
11787 2017-04-20 Jakub Jelinek <jakub@redhat.com>
11789 * BASE-VER: Set to 8.0.0.
11791 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
11793 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
11794 priority .init_array and .fini_array section with SECTION_NOTYPE
11797 2017-04-20 Jakub Jelinek <jakub@redhat.com>
11799 PR middle-end/80423
11800 * tree.h (build_array_type): Add typeless_storage default argument.
11801 * tree.c (type_cache_hasher::equal): Also compare
11802 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
11803 (build_array_type): Add typeless_storage argument, set
11804 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
11806 (build_nonshared_array_type): Adjust build_array_type_1 caller.
11807 (build_array_type): Likewise. Add typeless_storage argument.
11809 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
11810 Jakub Jelinek <jakub@redhat.com>
11812 PR tree-optimization/80426
11813 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
11814 operation on symbolic operands, also compute the overflow for the
11815 invariant part when the operation degenerates into a negation.
11817 2017-04-19 Jakub Jelinek <jakub@redhat.com>
11820 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
11821 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
11824 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
11826 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
11829 * config/avr/avr.c (tree.h): Include it.
11830 (cgraph.h): Include it.
11831 (avr_encode_section_info): Don't warn for uninitialized progmem
11832 variable if it's just an alias.
11834 2017-04-19 Richard Biener <rguenther@suse.de>
11837 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
11838 when needed by AutoPGO.
11840 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
11843 * doc/lto.texi: Remove an extra 'that'.
11845 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
11847 PR rtl-optimization/80429
11848 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
11849 are only used in debug insns.
11851 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
11852 Vladimir Makarov <vmakarov@redhat.com>
11854 * config/sparc/predicates.md (input_operand): Add comment. Return
11855 true for any memory operand when LRA is in progress.
11856 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
11858 2017-04-18 Jeff Law <law@redhat.com>
11861 * mips.md ({return,simple_return}_internal): Do not overwrite
11864 2017-04-18 Jakub Jelinek <jakub@redhat.com>
11866 PR tree-optimization/80443
11867 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
11868 instead of adding 1, subtract -1 and similarly instead of subtracting
11871 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
11873 PR rtl-optimization/80357
11874 * haifa-sched.c (tmp_bitmap): New variable.
11875 (model_recompute): Handle duplicate use records.
11876 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
11877 (free_global_sched_pressure_data): Free it.
11879 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
11882 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
11883 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
11884 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
11885 instead of SYSTEM_HEADER_DIR.
11887 2017-04-18 Jeff Law <law@redhat.com>
11889 PR middle-end/80422
11890 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
11891 predecessors after walking up the insn chain.
11893 2017-04-18 Jakub Jelinek <jakub@redhat.com>
11896 * dwarf2out.c (modified_type_die): Try harder not to emit internal
11897 sizetype type into debug info.
11899 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
11902 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
11903 unneeded test for TARGET_UPPER_REGS_SF.
11904 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
11906 2017-04-18 Jakub Jelinek <jakub@redhat.com>
11909 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
11910 instead of gsi_after_labels.
11912 2017-04-18 Jeff Law <law@redhat.com>
11914 * regcprop.c (maybe_mode_change): Avoid creating copies of the
11918 2017-04-13 Jeff Law <law@redhat.com>
11919 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
11920 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
11922 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
11925 * config/avr/avr.c (intl.h): Include it.
11926 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
11928 2017-04-18 Martin Liska <mliska@suse.cz>
11930 PR gcov-profile/78783
11931 * gcov-tool.c (gcov_output_files): Validate that destination
11932 file is either removed by the tool or by a user.
11934 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
11935 Guy Benyei <guybe@mellanox.com>
11937 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
11938 block, and do not negate it, the stored id is already negative.
11940 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
11942 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
11944 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
11947 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
11948 masks of options that should be turned off if the VSX vector
11949 options are turned off.
11950 (OTHER_P8_VECTOR_MASKS): Likewise.
11951 (OTHER_VSX_VECTOR_MASKS): Likewise.
11952 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
11953 rs6000_disable_incompatible_switches to validate no type switches
11955 (rs6000_incompatible_switch): New function to disallow turning on
11956 other vector options if -mno-vsx, -mno-power8-vector, or
11957 -mno-power9-vector are specified.
11959 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
11961 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
11963 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
11965 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
11966 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
11967 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
11968 (ARG_POINTER_CFA_OFFSET): Likewise.
11970 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
11972 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
11973 conditions to take advantage of various optimizations.
11975 2017-04-13 Jeff Law <law@redhat.com>
11977 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
11978 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
11979 (zero_extendsidi2_dext): Likewise.
11981 2017-04-13 Jakub Jelinek <jakub@redhat.com>
11984 * fold-const.c (fold_ternary_loc): Revert
11985 use op0 instead of fold_convert_loc (loc, type, arg0) part of
11988 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
11990 PR rtl-optimization/80343
11991 * lra-remat.c (update_scratch_ops): Assign original hard reg to
11992 new scratch pseudo.
11994 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
11997 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
11998 to ubsan_encode_value.
12000 2017-04-13 Jeff Law <law@redhat.com>
12002 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
12003 appearing in DEBUG_INSNs.
12005 2017-04-13 Martin Liska <mliska@suse.cz>
12007 PR gcov-profile/80413
12008 * gcov-io.c (gcov_write_string): Copy to buffer just when
12009 allocated size is greater than zero.
12011 2017-04-13 Jakub Jelinek <jakub@redhat.com>
12014 * dwarf2out.c (decls_for_scope): Ignore declarations of
12015 current_function_decl in BLOCK_NONLOCALIZED_VARS.
12017 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
12020 * ipa-visibility.c (non_local_p): Fix typos.
12021 (localize_node): When localizing symbol in same comdat group,
12022 dissolve the group only when we know external symbols are going
12024 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
12026 2017-04-12 Jakub Jelinek <jakub@redhat.com>
12028 PR tree-optimization/79390
12029 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
12030 order does not result in usable sequence, retry with reversed operand
12036 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
12037 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
12038 op0 instead of fold_convert_loc (loc, type, arg0).
12040 2017-04-12 Jeff Law <law@redhat.com>
12042 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
12043 has a delay slot in the generated code.
12045 * config/cris/cris.md (cris_preferred_reload_class): Return
12046 GENNONACR_REGS rather than GENERAL_REGS.
12048 2017-04-12 Jakub Jelinek <jakub@redhat.com>
12051 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
12052 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
12053 signedness of the result type.
12055 2017-04-12 Richard Biener <rguenther@suse.de>
12056 Jeff Law <law@redhat.com>
12058 PR tree-optimization/80359
12059 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
12060 trim stores to TARGET_MEM_REFs.
12062 2017-04-12 Richard Biener <rguenther@suse.de>
12064 PR tree-optimization/79390
12065 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
12066 threading case even more.
12068 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
12071 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
12072 for quad_address_p for TImode, instead of just not indexed_address.
12074 2017-04-12 Richard Biener <rguenther@suse.de>
12075 Bernd Edlinger <bernd.edlinger@hotmail.de>
12077 PR middle-end/79671
12078 * alias.c (component_uses_parent_alias_set_from): Handle
12079 TYPE_TYPELESS_STORAGE.
12080 (get_alias_set): Likewise.
12081 * tree-core.h (tree_type_common): Add typeless_storage flag.
12082 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
12083 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
12084 for types containing members with TYPE_TYPELESS_STORAGE.
12085 (place_field): Likewise.
12086 (layout_type): Likewise for ARRAY_TYPE.
12087 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
12088 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
12089 TYPE_TYPELESS_STORAGE.
12090 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
12092 2017-04-12 Jakub Jelinek <jakub@redhat.com>
12095 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
12096 first argument to type.
12098 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12102 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
12103 CONST0_RTX (mode) rather than const0_rtx where appropriate.
12104 (rs6000_expand_binop_builtin): Likewise.
12105 (rs6000_expand_ternop_builtin): Likewise; also add missing
12106 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
12107 vshasigma built-ins.
12108 * doc/extend.texi: Document that vec_xxpermdi's third argument
12109 must be a constant.
12111 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
12113 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
12114 Use shift_const cost parameter when calculating gain of STV shifts.
12116 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
12118 PR rtl-optimization/70478
12119 * lra-constraints.c (process_alt_operands): Check memory for
12120 disfavoring memory insn operand.
12122 2017-04-11 Jakub Jelinek <jakub@redhat.com>
12124 PR middle-end/80100
12125 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
12126 left shift in unsigned HOST_WIDE_INT type.
12128 PR rtl-optimization/80385
12129 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
12130 (not (neg X)) into (plus X -1) for complex or non-integral modes.
12133 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
12134 if they have any depend clauses.
12136 2017-04-11 Martin Liska <mliska@suse.cz>
12139 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
12140 * ipa-split.c (split_function): Create a local comdat symbol
12141 if caller is in a comdat group.
12143 2017-04-11 Martin Liska <mliska@suse.cz>
12146 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
12149 2017-04-11 Martin Sebor <msebor@redhat.com>
12151 PR middle-end/80364
12152 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
12153 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
12155 (directive::set_width, directive::set_precision, format_character):
12157 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
12160 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
12163 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
12164 conflict, set target->arch_name instead of target->cpu_name.
12166 2017-04-11 Richard Biener <rguenther@suse.de>
12168 PR tree-optimization/80374
12169 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
12170 build_zero_cst, remove fold_convertible_p check again.
12172 2017-04-11 Martin Liska <mliska@suse.cz>
12175 * ubsan.c (instrument_object_size): Do not instrument register
12178 2017-04-11 Jakub Jelinek <jakub@redhat.com>
12181 * config/i386/i386-builtin-types.def
12182 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
12183 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
12184 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
12185 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
12186 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
12187 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
12188 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
12189 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
12190 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
12191 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
12192 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
12193 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
12194 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
12195 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
12196 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
12197 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
12198 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
12199 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
12200 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
12201 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
12202 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
12203 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
12204 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
12205 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
12206 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
12207 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
12208 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
12209 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
12210 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
12211 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
12212 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
12213 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
12214 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
12215 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
12216 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
12217 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
12218 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
12219 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
12220 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
12221 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
12222 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
12223 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
12224 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
12225 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
12226 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
12227 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
12229 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
12230 flag to second_arg_count, handle 4 argument function type _COUNT
12231 aliases, handle second_arg_count on second argument rather than last.
12233 2017-04-10 Jeff Law <law@redhat.com>
12235 PR tree-optimization/80374
12236 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
12237 record anything if we can not convert integer_zero_node to the
12240 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
12243 * config/rs6000/rs6000.c (rs6000_option_override_internal):
12244 Enhance special handling given to the TARGET_P9_MINMAX option in
12245 relation to certain other options.
12247 2017-04-10 Bin Cheng <bin.cheng@arm.com>
12249 PR tree-optimization/80153
12250 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
12251 remove POINTER_PLUS_EXPR's base part directly, rather than through
12254 2017-04-10 Richard Biener <rguenther@suse.de>
12255 Bin Cheng <bin.cheng@arm.com>
12257 PR tree-optimization/80153
12258 * tree-affine.c (aff_combination_to_tree): Get base pointer from
12259 the first element of pointer type aff_tree. Build result expr in
12261 (add_elt_to_tree): Convert to type unconditionally. Remove other
12262 fold_convert calls.
12263 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
12264 (rewrite_use_nonlinear_expr): Check invariant using iv information.
12266 2017-04-10 Richard Biener <rguenther@suse.de>
12268 * tree-ssa-structalias.c (find_func_aliases): Properly handle
12271 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
12273 PR rtl-optimization/70478
12274 * lra-constraints.c (curr_small_class_check): New.
12275 (update_and_check_small_class_inputs): New.
12276 (process_alt_operands): Update curr_small_class_check. Disfavor
12277 alternative insn memory operands. Check available regs for small
12280 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
12283 * config/mips/mips.opt (-mvirt): Update description.
12284 * doc/invoke.texi (-mvirt): Likewise.
12286 2017-04-10 Richard Biener <rguenther@suse.de>
12288 PR middle-end/80362
12289 * fold-const.c (fold_binary_loc): Look at unstripped ops when
12290 looking for NEGATE_EXPR in -A / -B to A / B folding.
12292 2017-04-10 Martin Liska <mliska@suse.cz>
12294 PR gcov-profile/80224
12295 * gcov.c (print_usage): Fix usage string.
12296 (get_gcov_intermediate_filename): Remove.
12297 (output_gcov_file): Use both for normal and intermediate format.
12298 (generate_results): Do not initialize special file for
12299 intermediate format.
12301 2017-04-10 Richard Biener <rguenther@suse.de>
12303 PR tree-optimization/80304
12304 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
12307 2017-04-10 Nathan Sidwell <nathan@acm.org>
12310 * config/rs6000/rs6000.c (rs6000_vector_type): New.
12311 (rs6000_init_builtins): Use it.
12313 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12315 * config/arm/arm.md (<mrc>): Add mode to SET source.
12316 (<mrrc>): Likewise.
12318 2017-04-10 Richard Biener <rguenther@suse.de>
12320 PR middle-end/80344
12321 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
12323 2017-04-10 Jakub Jelinek <jakub@redhat.com>
12326 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
12327 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
12328 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
12329 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
12330 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
12331 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
12332 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
12333 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
12334 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
12335 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
12336 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
12337 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
12338 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
12339 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
12340 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
12341 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
12342 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
12343 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
12344 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
12345 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
12346 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
12347 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
12348 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
12350 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
12352 PR rtl-optimization/70478
12353 * lra-constraints.c: Reverse the last patch.
12355 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
12357 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
12358 Add comment for WCHAR_T.
12360 2017-04-08 Martin Liska <mliska@suse.cz>
12363 2017-04-07 Martin Liska <mliska@suse.cz>
12366 * ipa-split.c (split_function): Add function part to a same comdat
12369 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12372 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
12374 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
12376 * rs6000/rs6000.c (vec_load_pendulum): Rename...
12377 (vec_pairing): ...to this.
12378 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
12379 (rs6000_sched_init): Adjust for name change.
12380 (struct rs6000_sched_context): Likewise.
12381 (rs6000_init_sched_context): Likewise.
12382 (rs6000_set_sched_context): Likewise.
12384 2017-04-07 Jakub Jelinek <jakub@redhat.com>
12390 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
12392 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
12393 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
12394 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
12396 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
12398 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
12400 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
12402 PR rtl-optimization/70703
12403 * ira-color.c (update_conflict_hard_regno_costs): Use
12404 int64_t instead of HOST_WIDE_INT.
12406 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
12408 PR rtl-optimization/70478
12409 * lra-constraints.c (process_alt_operands): Disfavor alternative
12410 insn memory operands.
12412 2017-04-07 Jeff Law <law@redhat.com>
12414 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
12415 CALL and NOTE_INSN_CALL_ARG_LOCATION.
12417 2017-04-07 Martin Liska <mliska@suse.cz>
12420 * config/aarch64/aarch64.c (aarch64_process_target_attr):
12421 Show error message instead of an ICE.
12423 2017-04-07 Martin Liska <mliska@suse.cz>
12426 * ipa-split.c (split_function): Add function part to a same comdat
12429 2017-04-07 Richard Biener <rguenther@suse.de>
12431 PR middle-end/80341
12432 * tree.c (get_unwidened): Also handle ! for_type case for
12434 * convert.c (do_narrow): Split out from ...
12435 (convert_to_integer_1): ... here. Do not pass final truncation
12436 type to get_unwidened for TRUNC_DIV_EXPR.
12438 2017-04-07 Richard Biener <rguenther@suse.de>
12440 * tree-affine.c (wide_int_ext_for_comb): Take type rather
12442 (aff_combination_const): Adjust.
12443 (aff_combination_scale): Likewise.
12444 (aff_combination_add_elt): Likewise.
12445 (aff_combination_add_cst): Likewise.
12446 (aff_combination_convert): Likewise.
12447 (add_elt_to_tree): Likewise. Remove unused argument.
12448 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
12450 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
12452 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
12454 * config/arm/arm.c (arm_default_short_enums): Use
12455 ARM_DEFAULT_SHORT_ENUMS.
12456 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
12458 2017-04-06 Jakub Jelinek <jakub@redhat.com>
12461 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
12462 members with redundant out-of-class redeclaration.
12464 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
12467 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
12468 * config/i386/i386.md (*zero_extendsidi2):
12469 Add (?*x,*x) and (?*v,*v) alternatives.
12471 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
12474 * config/i386/i386.c (ix86_expand_builtin)
12475 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
12476 mode from insn data. Convert operands to insn operand mode.
12477 Copy operands that don't satisfy insn predicate to a register.
12479 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
12481 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
12484 2017-04-06 Richard Biener <rguenther@suse.de>
12486 PR tree-optimization/80334
12487 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
12488 preserve alignment of accesses.
12490 2017-04-06 Richard Biener <rguenther@suse.de>
12492 PR tree-optimization/80262
12493 * tree-sra.c (build_ref_for_offset): Preserve address-space
12495 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
12496 Drop useless address-space information on MEM_REF offsets.
12498 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
12500 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
12502 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
12504 PR rtl-optimization/70703
12505 * ira-color.c (update_conflict_hard_regno_costs): Use
12506 HOST_WIDE_INT instead of long.
12508 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
12511 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
12512 not defined for x86_64 target. Add -mmmx target option when __SSE2__
12514 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
12515 for x86_64 target. Handle -m3dnowa option.
12517 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
12519 PR rtl-optimization/70703
12520 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
12521 (update_conflict_hard_regno_costs): Use long instead of unsigned
12522 arithmetic for cost calculation.
12524 2017-04-05 Jakub Jelinek <jakub@redhat.com>
12525 Bernd Edlinger <bernd.edlinger@hotmail.de>
12528 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
12531 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
12534 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
12535 ptr_mode with Pmode throughout.
12536 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
12537 into probe_stack_range and use DImode.
12539 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
12542 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
12543 call_eh_return is true.
12545 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12547 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
12548 Initialize last_match_fntype_index.
12550 2017-04-05 Jakub Jelinek <jakub@redhat.com>
12553 * tree-nvr.c: Include internal-fn.h.
12554 (pass_return_slot::execute): Ignore internal calls without
12557 2017-04-04 Jakub Jelinek <jakub@redhat.com>
12558 Richard Biener <rguenther@suse.de>
12561 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
12562 captures used multiple times, except for the last use.
12563 * generic-match-head.c: Include gimplify.h.
12565 2017-04-04 Jakub Jelinek <jakub@redhat.com>
12567 PR tree-optimization/79390
12568 * target.h (struct noce_if_info): Declare.
12569 * targhooks.h (default_noce_conversion_profitable_p): Declare.
12570 * target.def (noce_conversion_profitable_p): New target hook.
12571 * ifcvt.h (struct noce_if_info): New type, moved from ...
12572 * ifcvt.c (struct noce_if_info): ... here.
12573 (noce_conversion_profitable_p): Renamed to ...
12574 (default_noce_conversion_profitable_p): ... this. No longer
12576 (noce_try_store_flag_constants, noce_try_addcc,
12577 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
12578 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
12579 instead of noce_conversion_profitable_p.
12580 * config/i386/i386.c: Include ifcvt.h.
12581 (ix86_option_override_internal): Don't override
12582 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
12583 (ix86_noce_conversion_profitable_p): New function.
12584 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
12585 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
12586 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
12587 * doc/tm.texi: Regenerated.
12589 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12591 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
12594 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
12597 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
12598 instructions for small multiply cores.
12600 2017-04-04 Jeff Law <law@redhat.com>
12602 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
12604 (mips_expand_vec_perm_const): Initialize elements in orig_perm
12605 that are not set by the loop over the elements.
12607 2017-04-04 Jakub Jelinek <jakub@redhat.com>
12610 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
12611 int mode, convert_modes it to mode as unsigned, otherwise use
12612 lowpart_subreg to mode rather than SImode.
12613 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
12614 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
12615 Use DImode instead of SImode for the shift count operand.
12616 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
12619 2017-04-04 Richard Biener <rguenther@suse.de>
12621 PR middle-end/80281
12622 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
12623 arithmetic done for the negate or the plus. Simplify.
12624 (A - (-B) -> A + B): Likewise.
12625 * fold-const.c (split_tree): Make sure to not negate pointers.
12627 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
12629 PR rtl-optimization/60818
12630 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
12631 a compare of comparisons with the thing compared if this results
12632 in a different machine mode.
12634 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
12636 * alias.c (base_alias_check): Fix typo in comment.
12637 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
12638 * cgraphunit.c (symbol_table::compile): Likewise.
12639 * collect2.c (maybe_run_lto_and_relink): Likewise.
12640 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
12641 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
12642 * config/avr/avr.c (avr_map_op_t): Likewise.
12643 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
12644 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
12645 * config/epiphany/epiphany.md (movcc): Likewise.
12646 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
12647 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
12649 * config/mips/mips.c (mips_save_restore_reg): Likewise.
12650 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
12651 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
12652 * config/sh/sh.c (sh_rtx_costs): Likewise.
12653 * fold-const.c (fold_truth_andor): Likewise.
12654 * genautomata.c (collapse_flag): Likewise.
12655 * gengtype.h (struct type::u::s): Likewise.
12656 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
12657 * input.c (FORMAT_AMOUNT): Likewise.
12658 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
12659 (known_aggs_to_agg_replacement_list): Likewise.
12660 * ipa-inline-analysis.c: Likewise.
12661 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
12662 * ipa-polymorphic-call.c
12663 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
12664 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
12665 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
12667 * modulo-sched.c (apply_reg_moves): Likewise.
12668 * omp-expand.c (build_omp_regions_1): Likewise.
12669 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
12670 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
12671 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
12672 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
12673 * value-prof.c: Likewise.
12674 * var-tracking.c (val_reset): Likewise.
12676 2017-04-03 Richard Biener <rguenther@suse.de>
12678 PR tree-optimization/80275
12679 * fold-const.c (split_address_to_core_and_offset): Handle
12682 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
12684 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
12685 descriptors is at least equal to that of functions.
12687 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
12689 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
12691 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
12694 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
12695 (mov<IMOD4:mode>): New expander.
12696 (*mov<IMOD4:mode>_internal): New insn and split pattern.
12698 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
12700 PR rtl-optimization/79405
12701 * fwprop.c (propagations_left): New variable.
12702 (forward_propagate_into): Decrement it.
12703 (fwprop_init): Initialize it.
12704 (fw_prop): If the variable has reached zero, stop propagating.
12705 (fwprop_addr): Ditto.
12707 2017-03-31 Jakub Jelinek <jakub@redhat.com>
12710 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
12711 a FUNCTION_DECL, pass it as decl instead of origin to
12714 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
12716 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
12719 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
12722 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
12723 TARGET_VSX_SMALL_INTEGER.
12725 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12727 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
12728 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
12730 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
12732 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
12733 extraction from odd-numbered MSA register.
12735 2017-03-31 Jakub Jelinek <jakub@redhat.com>
12737 PR middle-end/80173
12738 * expmed.c (store_bit_field_1): Don't attempt to create
12739 a word subreg out of hard registers wider than word if they
12740 have HARD_REGNO_NREGS of 1 for their mode.
12742 PR middle-end/80163
12743 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
12744 conversions to integer types wider than word and pointer.
12747 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
12748 (rtx_equal_for_cselib_p): Pass 0 to it.
12749 * cselib.c (cselib_hasher::equal): Likewise.
12750 (rtx_equal_for_cselib_1): Add depth argument. If depth
12751 is 128, don't look up VALUE locs and punt. Increment
12752 depth in recursive calls when walking VALUE locs.
12754 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
12756 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
12757 (make_gcov_file_name): Use the canonical path name for generating
12759 (read_line): Fix handling of files with ascii null bytes.
12761 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
12763 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
12764 to initialise a vector register instead
12765 of using a const_int.
12767 2017-03-30 Jakub Jelinek <jakub@redhat.com>
12769 PR translation/80189
12770 * gimplify.c (omp_default_clause): Use %qs instead of %s in
12771 diagnostic messages.
12773 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
12776 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
12777 (dfp_diex_<mode>): Update mode of operand 1.
12778 * doc/extend.texi (dxex, dxexq): Document change to return type.
12779 (diex, diexq): Document change to argument type.
12781 2017-03-30 Martin Jambor <mjambor@suse.cz>
12784 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
12785 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
12786 it reflects the signature changes performed at the callee side.
12787 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
12788 to cgraph_build_function_type_skip_args.
12789 (build_function_decl_skip_args): Adjust call to the above function.
12791 2017-03-30 Jakub Jelinek <jakub@redhat.com>
12794 * config/i386/sse.md
12795 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
12796 register as dest whenever it is a MEM not rtx_equal_p to the
12797 corresponding dup operand, and when forcing into reg move the
12798 reg into the memory afterwards.
12799 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
12800 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
12801 for the force_reg mode.
12802 (avx512vl_vextractf128<mode>): Use register as dest either
12803 always when a MEM, or when it is a MEM not rtx_equal_p to the
12804 corresponding dup operand, or even not when it is a CONST_VECTOR
12805 depending on the mode and lo vs. hi.
12806 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
12808 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
12809 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
12810 Likewise. Require that operands[2] is even.
12811 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
12812 Remove extraneous parens. Require that operands[2] is a multiple
12814 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
12815 operands[0] is a MEM if <mask_applied>, the predicates/constraints
12816 disallow memory then.
12818 2017-03-30 Richard Biener <rguenther@suse.de>
12820 PR tree-optimization/77498
12821 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
12822 to non-constants over backedges.
12824 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
12826 PR rtl-optimization/80233
12827 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
12828 as last_combined_insn. Do not test for BARRIER_P separately.
12830 2017-03-29 Andreas Schwab <schwab@suse.de>
12833 * calls.c (prepare_call_address): Convert funexp to Pmode before
12834 copying to temp reg.
12836 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12838 PR tree-optimization/80158
12839 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
12840 Handle possible future case of more than one alternate
12842 (replace_rhs_if_not_dup): Likewise.
12843 (replace_one_candidate): Likewise.
12845 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
12847 PR rtl-optimization/80193
12848 * ira.c (ira): Do not check allocation for LRA.
12850 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
12852 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
12853 (nvptx_output_simt_exit): Declare.
12854 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
12855 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
12856 (init_softstack_frame): Move initialization of crtl->is_leaf to...
12857 (nvptx_declare_function_name): ...here. Emit declaration of local
12858 memory space buffer for omp_simt_enter insn.
12859 (nvptx_output_unisimt_switch): New.
12860 (nvptx_output_softstack_switch): New.
12861 (nvptx_output_simt_enter): New.
12862 (nvptx_output_simt_exit): New.
12863 * config/nvptx/nvptx.h (struct machine_function): New fields
12864 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
12865 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
12866 (UNSPECV_SIMT_EXIT): Ditto.
12867 (omp_simt_enter_insn): New insn.
12868 (omp_simt_enter): New expansion.
12869 (omp_simt_exit): New insn.
12870 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
12872 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
12873 (expand_GOMP_SIMT_ENTER_ALLOC): New.
12874 (expand_GOMP_SIMT_EXIT): New.
12875 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
12876 (GOMP_SIMT_ENTER_ALLOC): Ditto.
12877 (GOMP_SIMT_EXIT): Ditto.
12878 * target-insns.def (omp_simt_enter): New insn.
12879 (omp_simt_exit): Ditto.
12880 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
12882 (lower_rec_simd_input_clauses): Implement SIMT privatization.
12883 (lower_rec_input_clauses): Likewise.
12884 (lower_lastprivate_clauses): Handle SIMT privatization.
12886 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
12887 (ompdevlow_adjust_simt_enter): New.
12888 (find_simtpriv_var_op): New.
12889 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
12890 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
12892 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
12893 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
12894 (copy_decl_for_dup_finish): Ditto.
12896 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
12898 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
12901 * config/i386/i386.c (ix86_option_override_internal): Always
12902 allow -mpreferred-stack-boundary=3 for 64-bit targets.
12904 2017-03-28 Bin Cheng <bin.cheng@arm.com>
12906 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
12908 2017-03-28 Bin Cheng <bin.cheng@arm.com>
12910 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
12911 mark new edge's irreducible flag accordign to it.
12912 (vect_do_peeling): Check loop preheader edge's irreducible flag
12913 and pass it to function slpeel_add_loop_guard.
12915 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
12917 PR tree-optimization/80218
12918 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
12919 Update block frequencies and counts.
12921 2017-03-28 Richard Biener <rguenther@suse.de>
12923 PR tree-optimization/78644
12924 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
12925 of a simplification result we may not use it at all.
12927 2017-03-28 Richard Biener <rguenther@suse.de>
12930 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
12931 without arguments, generate default definition of a SSA name.
12933 2017-03-28 Richard Biener <rguenther@suse.de>
12935 PR middle-end/80222
12936 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
12937 TYPE_REF_CAN_ALIAS_ALL references.
12938 * fold-const.c (fold_indirect_ref_1): Likewise.
12940 2017-03-28 Martin Liska <mliska@suse.cz>
12943 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
12944 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
12946 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
12947 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
12949 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
12950 (EXTRA_SPECS): Define.
12951 (SUBTARGET_EXTRA_SPECS): Likewise.
12952 (SUBTARGET_CPP_SPEC): Likewise.
12953 * config/arc/elf.h (EXTRA_SPECS): Renamed to
12954 SUBTARGET_EXTRA_SPECS.
12955 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
12957 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
12959 * config/arc/simdext.md (vst64_insn): Update pattern.
12960 (vld32wh_insn): Likewise.
12961 (vld32wl_insn): Likewise.
12962 (vld64_insn): Likewise.
12963 (vld32_insn): Likewise.
12965 2017-03-28 Marek Polacek <polacek@redhat.com>
12968 * fold-const.c (fold_comparison): Use protected_set_expr_location
12969 instead of SET_EXPR_LOCATION.
12971 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
12973 * tree.c (add_expr): Avoid name lookup warning.
12975 2017-03-27 Jeff Law <law@redhat.com>
12977 PR tree-optimization/80216
12978 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
12979 function name. Limit recursion depth.
12980 (record_temporary_equivalences): Corresponding changes.
12982 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
12984 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
12987 2017-03-27 Jakub Jelinek <jakub@redhat.com>
12990 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
12992 * cfgcleanup.c (reg_note_cfa_p): New array.
12993 (insns_have_identical_cfa_notes): New function.
12994 (old_insns_match_p): Don't cross-jump in between /f
12995 and non-/f instructions. If both i1 and i2 are frame related,
12996 verify all CFA notes, their order and content.
12998 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
13001 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
13002 HImode and SImode with zero extend to DImode to one insn.
13003 (bswap<mode>2_extenddi): Likewise.
13004 (bswapsi2_extenddi): Likewise.
13005 (bswaphi2_extendsi): Likewise.
13006 (bswaphi2): Combine bswap HImode and SImode into one insn.
13007 Separate memory insns from swapping register.
13008 (bswapsi2): Likewise.
13009 (bswap<mode>2): Likewise.
13010 (bswaphi2_internal): Delete, no longer used.
13011 (bswapsi2_internal): Likewise.
13012 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
13013 store, and gpr<-gpr swap insns.
13014 (bswap<mode>2_store): Likewise.
13015 (bswaphi2_reg): Register only splitter, combine with the splitter.
13016 (bswaphi2 splitter): Likewise.
13017 (bswapsi2_reg): Likewise.
13018 (bswapsi2 splitter): Likewise.
13019 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
13020 the insns into load, store, and register/register insns.
13021 (bswapdi2_ldbrx): Likewise.
13022 (bswapdi2_load): Likewise.
13023 (bswapdi2_store): Likewise.
13024 (bswapdi2_reg): Likewise.
13026 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
13028 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
13029 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
13031 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
13034 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
13036 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
13037 special handling for target option conflicts between dform
13038 options (-mpower9-dform, -mpower9-dform-vector,
13039 -mpower9-dform-scalar) and -mno-direct-move.
13041 2017-03-27 Richard Biener <rguenther@suse.de>
13043 PR tree-optimization/80181
13044 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
13046 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
13048 * config/arc/predicates.md (move_double_src_operand): Replace the
13049 call to move_double_src_operand with a call to address_operand.
13051 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
13053 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
13054 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
13055 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
13057 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
13059 * config/arc/predicates.md (long_immediate_loadstore_operand):
13060 Consider scaled addresses cases.
13062 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
13064 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
13065 restored when in interrupt.
13066 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
13067 doesn't have delay slot.
13069 2017-03-27 Richard Biener <rguenther@suse.de>
13072 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
13073 inlined thunk clones.
13075 2017-03-27 Jakub Jelinek <jakub@redhat.com>
13078 * asan.c (instrument_derefs): Copy over last operand from
13079 original COMPONENT_REF to the new COMPONENT_REF with
13080 DECL_BIT_FIELD_REPRESENTATIVE.
13081 * ubsan.c (instrument_object_size): Likewise.
13083 2017-03-27 Richard Biener <rguenther@suse.de>
13085 PR tree-optimization/80170
13086 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
13087 sure DR/SCEV didnt fold in constants we do not see when looking
13088 at the reference base alignment.
13090 2017-03-27 Richard Biener <rguenther@suse.de>
13092 PR middle-end/80171
13093 * gimple-fold.c (fold_ctor_reference): Properly guard against
13094 NULL return value from canonicalize_constructor_val.
13096 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
13099 * config/i386/i386.c (ix86_expand_builtin)
13100 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
13101 flags reg setting and flags reg using instructions.
13102 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
13103 clobbering instructions to zero extend op2.
13105 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
13107 * doc/install.texi (Configuration) <--with-aix-soname>:
13108 Update link to AIX ld.
13110 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
13112 PR rtl-optimization/80160
13113 PR rtl-optimization/80159
13114 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
13115 reg_alternate_class into account.
13117 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
13120 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
13121 to consider in curr_insn_transform.
13123 2017-03-24 Jakub Jelinek <jakub@redhat.com>
13125 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
13126 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
13127 and emit_mode_inner.
13129 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13131 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
13132 argument to the overloaded builtin variants. Use the new flag to
13133 deprecate certain builtin variants.
13134 * config/s390/s390-builtin-types.def: Add new builtin types.
13135 * config/s390/s390-builtins.h: Support new flags field for
13136 overloaded builtins.
13137 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
13138 (s390_macro_to_expand): Enable vector float data type.
13139 (s390_cpu_cpp_builtins_internal): Indicate support of the new
13140 builtins by incrementing the __VEC__ version number.
13141 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
13143 (s390_resolve_overloaded_builtin): Emit error messages depending
13144 on the builtin flags.
13145 * config/s390/s390.c (s390_expand_builtin): Support additional
13146 flags argument. Change error message to match the messages
13147 emitted in s390-c.c.
13148 * config/s390/s390.md: New UNSPEC_* constants.
13149 (op_type): Add new instruction types.
13150 * config/s390/vecintrin.h: Add new builtins and test data class
13152 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
13153 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
13154 (VEC_INEXACT, VEC_NOINEXACT): New constants.
13155 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
13156 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
13157 ("vec_mergel<mode>"): V_HW -> VEC_HW.
13159 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
13160 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
13161 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
13162 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
13164 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
13165 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
13166 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
13167 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
13169 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
13170 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
13171 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
13172 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
13173 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
13174 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
13175 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
13177 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
13178 ("vec_scatter_element<V_HW_4:mode>_DI")
13179 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
13180 ("vec_fpint<mode>", "vflls")
13181 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
13182 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
13183 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
13184 ("*vec_cmphe<mode>_cc"): ... these.
13186 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
13187 mode constant instead of magic value.
13189 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13191 * config/s390/s390.c (s390_expand_vec_compare): Support other
13192 vector floating point modes than just V2DF.
13193 (s390_expand_vcond): Likewise.
13194 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
13195 (s390_cannot_change_mode_class): Prevent mode changes between TF
13196 and V1TF in vector registers.
13197 * config/s390/s390.md (DF, SF): New mode attributes.
13198 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
13199 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
13200 SFmode support for VRs.
13201 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
13203 (VFT, VF_HW): New mode iterators.
13204 (vw, sdx): New mode attributes.
13205 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
13206 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
13207 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
13208 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
13209 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
13210 also the new vector floating point modes. Renaming to ...
13212 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
13213 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
13214 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
13215 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
13216 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
13217 ("vec_unordered<mode>"): ... these.
13219 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
13220 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
13221 ("*vec_extendv2df"): New insn definitions.
13223 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13225 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
13226 ("mulditi3_2", "*muldi3_sign"): New patterns.
13227 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
13228 rename the pattern definition.
13230 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13232 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
13234 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
13236 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13238 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
13239 instruction if possible.
13240 * config/s390/vector.md (vec_halfnumelts): New mode
13242 ("*vec_vllezlf<mode>"): New pattern.
13244 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13246 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
13247 ("popcountv4si2", "popcountv2di2"): Rename to ...
13248 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
13249 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
13251 ("popcount<mode>2_vxe"): New pattern.
13253 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13255 * common/config/s390/s390-common.c (processor_flags_table): Add
13257 * config.gcc: Add arch12.
13258 * config/s390/driver-native.c (s390_host_detect_local_cpu):
13259 Default to arch12 for unknown CPU model numbers.
13260 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
13261 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
13262 PROCESSOR_max sanity check.
13263 * config/s390/s390-opts.h (enum processor_type): Add
13265 * config/s390/s390.c (processor_table): Add arch12.
13266 (s390_expand_builtin): Add check for B_VXE flag.
13267 (s390_issue_rate): Add PROCESSOR_ARCH12.
13268 (s390_get_sched_attrmask): Likewise.
13269 (s390_get_unit_mask): Likewise.
13270 (s390_sched_score): Enable z13 scheduling for arch12.
13271 (s390_sched_reorder): Likewise.
13272 (s390_sched_variable_issue): Likewise.
13273 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
13275 (s390_tune_attr): Use z13 scheduling also for arch12.
13276 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
13277 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
13278 (TARGET_VXE_P): New macros.
13279 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
13280 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
13281 * config/s390/s390.opt: Add arch12 as processor_type.
13283 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13285 * config/s390/s390.md
13286 ("fixuns_truncdddi2", "fixuns_trunctddi2")
13287 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
13288 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
13290 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
13291 Rename expanders to ...
13293 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
13294 ("fixuns_truncdddi2_emu"): ... these.
13296 ("fixuns_trunc<mode>si2_emu"): New expander.
13298 ("*fixuns_truncdfdi2_z13"): Rename to ...
13299 ("*fixuns_truncdfdi2_vx"): ... this.
13301 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13303 * config/s390/2964.md: Remove the single element vector compare
13304 instructions which are no longer used.
13305 * config/s390/s390.c (s390_select_ccmode): Remove handling of
13307 (s390_canonicalize_comparison): Remove handling of DFmode
13309 (s390_expand_vec_compare_scalar): Remove function.
13310 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
13311 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
13313 ("*cmp<mode>_ccs"): Add wfcdb instruction.
13315 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13317 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
13319 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
13320 will anyway by matched by mov<mode>_64dfp.
13322 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13324 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
13325 vlef/vstef. Add missing operand to vleif.
13327 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13329 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
13330 pair for all vector types with 64 bit elements.
13331 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
13332 * config/s390/vector.md (V_HW_64): ... here.
13333 (V_128_NOSINGLE): New mode iterator.
13334 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
13335 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
13336 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
13337 ("*vec_load_pairv2di"): Change to ...
13338 ("*vec_load_pair<mode>"): ... this one.
13340 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13342 * config/s390/constraints.md: Add comments.
13343 (jKK): Reject element sizes > 8 bytes.
13344 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
13346 * config/s390/s390.md: Add the s_operand checks formerly in
13347 s390_split_ok_p to various splitters where they are still
13349 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
13350 for 128 bit vectors. Plus two splitters.
13352 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13354 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
13357 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13360 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
13361 error if the boundary argument is not constant.
13363 2017-03-24 Jakub Jelinek <jakub@redhat.com>
13365 PR rtl-optimization/80112
13366 * loop-doloop.c (doloop_condition_get): Don't check condition
13367 if cmp isn't SET with IF_THEN_ELSE src.
13369 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13371 PR tree-optimization/80158
13372 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
13373 replacing a candidate statement, also replace it for the
13374 candidate's alternate interpretation.
13375 (replace_rhs_if_not_dup): Likewise.
13376 (replace_one_candidate): Likewise.
13378 2017-03-24 Richard Biener <rguenther@suse.de>
13380 PR tree-optimization/80167
13381 * graphite-isl-ast-to-gimple.c
13382 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
13384 (translate_isl_ast_to_gimple::get_rename): Likewise.
13386 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
13388 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
13389 handling of certain combinations of target options, including the
13390 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
13391 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
13393 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13396 * config/arm/arm.md (*load_multiple): Add reload_completed to
13397 matching condition.
13399 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13400 Richard Biener <rguenth@suse.de>
13402 PR tree-optimization/79908
13403 PR tree-optimization/80136
13404 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
13405 been cast away, gimplify_and_add suffices.
13407 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
13409 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
13411 2017-03-23 Richard Biener <rguenther@suse.de>
13413 PR tree-optimization/80032
13414 * gimplify.c (gimple_push_cleanup): Forced unconditional
13415 cleanups still have to go to the conditional_cleanups
13418 2017-03-22 Jakub Jelinek <jakub@redhat.com>
13420 PR tree-optimization/80072
13421 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
13423 (next_operand_entry_id): Change type to unsigned int.
13424 (sort_by_operand_rank): Make sure to return the right return value
13425 even if unsigned fields are bigger than INT_MAX.
13426 (struct oecount): Change cnt and id type to unsigned int.
13427 (oecount_hasher::equal): Formatting fix.
13428 (oecount_cmp): Make sure to return the right return value
13429 even if unsigned fields are bigger than INT_MAX.
13430 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
13433 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
13434 TREE_READONLY on result if writing it more than once.
13437 * doc/invoke.texi (-fsanitize=thread): Document that with
13438 -fnon-call-exceptions atomics are not able to throw
13442 * tsan.c: Include tree-eh.h.
13443 (instrument_builtin_call): Call maybe_clean_eh_stmt or
13444 maybe_clean_or_replace_eh_stmt where needed.
13445 (instrument_memory_accesses): Add cfg_changed argument.
13446 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
13447 if it returned true.
13448 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
13450 PR rtl-optimization/63191
13451 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
13452 wrapper function, moved the whole old content into ...
13453 (ix86_delegitimize_address_1): ... this. New inline function.
13454 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
13455 true as last argument instead of ix86_delegitimize_address.
13457 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
13459 * config/aarch64/aarch64.c (generic_branch_cost): Copy
13460 cortexa57_branch_cost.
13462 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
13464 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
13466 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13469 * doc/md.texi (Constraints): Document wA constraint.
13470 * config/rs6000/constraints.md (wA): New.
13471 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
13472 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
13473 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
13474 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
13476 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
13479 * gimplify.c (is_oacc_declared): New function.
13480 (oacc_default_clause): Use it to set default flags for acc declared
13481 variables inside parallel regions.
13482 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
13483 declared variables.
13484 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
13485 declare attribute to any decl as necessary.
13487 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
13490 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
13491 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
13492 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
13493 (arm_arch_lpae): This.
13494 * config/arm/arm.c (arm_arch7ve): Rename into ...
13495 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
13496 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
13499 2017-03-22 Martin Liska <mliska@suse.cz>
13502 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
13503 error message instead of an ICE.
13505 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13507 * doc/extend.texi (6.11 Additional Floating Types): Revise.
13509 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
13511 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
13513 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
13516 2017-03-21 Martin Sebor <msebor@redhat.com>
13518 * doc/extend.texi: Use "cannot" instead of "can't."
13519 * doc/hostconfig.texi: Same.
13520 * doc/install.texi: Same.
13521 * doc/invoke.texi: Same.
13522 * doc/loop.texi: Same.
13523 * doc/md.texi: Same.
13524 * doc/objc.texi: Same.
13525 * doc/rtl.texi: Same.
13526 * doc/tm.texi: Same.
13527 * doc/tm.texi.in: Same.
13528 * doc/trouble.texi: Same.
13530 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
13533 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
13534 (collect_checksum_attributes): Set it.
13535 (die_checksum_ordered): Use it.
13537 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13539 PR tree-optimization/79908
13540 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
13541 change: For a VA_ARG whose LHS has been cast away, use
13542 force_gimple_operand to construct the side effects.
13544 2017-03-21 David Malcolm <dmalcolm@redhat.com>
13546 PR translation/80001
13547 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
13548 more amenable to translation.
13549 (oacc_loop_auto_partitions): Likewise.
13551 2017-03-21 Marek Polacek <polacek@redhat.com>
13552 Martin Sebor <msebor@redhat.com>
13554 PR tree-optimization/80109
13555 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
13556 on INTEGRAL_TYPE_P.
13558 2017-03-21 Jakub Jelinek <jakub@redhat.com>
13559 Segher Boessenkool <segher@kernel.crashing.org>
13562 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
13563 check reg_used_between_p between insn and one of succ or succ2
13564 depending on if succ is artificial insn not inserted into insn
13567 2017-03-21 Martin Liska <mliska@suse.cz>
13569 PR gcov-profile/80081
13570 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
13571 * doc/gcc.texi: Include gcov-dump stuff.
13572 * doc/gcov-dump.texi: New file.
13574 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
13576 PR rtl-optimization/79150
13577 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
13578 conditional jump, if the jump is the last insn of the loop.
13580 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13581 Richard Biener <rguenth@suse.de>
13583 PR tree-optimization/79908
13584 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
13585 been cast away, use force_gimple_operand to construct the side
13588 2017-03-21 Martin Liska <mliska@suse.cz>
13590 PR libfortran/79956
13591 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
13594 2017-03-21 Brad Spengler <spender@grsecurity.net>
13597 * plugin.c (htab_hash_plugin): New function.
13598 (add_new_plugin): Use it and adjust.
13599 (parse_plugin_arg_opt): Adjust.
13600 (init_one_plugin): Likewise.
13602 2017-03-21 Richard Biener <rguenther@suse.de>
13604 PR tree-optimization/80032
13605 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
13606 if set force the cleanup to happen unconditionally.
13607 (gimplify_target_expr): Push inserted clobbers with force_uncond
13608 to avoid them being removed by control-dependent DCE.
13610 2017-03-21 Richard Biener <rguenther@suse.de>
13612 PR tree-optimization/80122
13613 * tree-inline.c (copy_bb): Do not expans va-arg packs or
13614 va_arg_pack_len when the inlined call stmt requires pack
13616 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
13618 2017-03-21 Jakub Jelinek <jakub@redhat.com>
13621 * tsan.c (instrument_builtin_call): If the memory model argument
13622 is not a constant, assume it is valid.
13625 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
13628 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
13630 PR rtl-optimization/79910
13631 * combine.c (can_combine_p): Do not allow combining an I0 or I1
13632 if its dest is used by an insn before I2 (other than the combined
13633 insns themselves, which are properly handled already).
13635 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
13638 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
13640 * combine.c (record_used_regs): New static function.
13641 (try_combine): Handle situations where there is an additional
13642 instruction between I2 and I3 which needs to have a LOG_LINK
13646 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
13648 * combine.c (try_combine): Delete redundant i1 test. Call
13649 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
13651 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13654 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
13655 alternatives 13/14.
13657 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13659 PR tree-optimization/80054
13660 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
13661 the optimization if a PHI or any of its arguments is not dominated
13662 by the candidate's basis. Use gphi* rather than gimple* as
13664 (replace_profitable_candidates): Clean up a gimple* variable that
13665 should be a gphi* variable.
13667 2017-03-20 Martin Sebor <msebor@redhat.com>
13670 * doc/extend.texi (attribute constructor): Document present limitation.
13672 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
13675 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
13676 __POWER9_VECTOR__ #ifdef control, change template definition to
13677 use Power9-specific built-in function.
13678 (vec_any_eq): Likewise.
13679 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
13680 to control outcomes from this test.
13681 (vector_ae_<mode>p): For VEC_F modes, likewise.
13683 2017-03-20 Ian Lance Taylor <iant@google.com>
13685 * config/i386/i386.c (ix86_function_regparm): Save an extra
13686 register for -fsplit-stack with DECL_STATIC_CHAIN.
13688 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
13691 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
13692 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
13694 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
13696 * config/riscv/riscv.c (riscv_print_operand): Use "fence
13698 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
13701 2017-03-20 Marek Polacek <polacek@redhat.com>
13704 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
13706 2017-03-20 Richard Biener <rguenther@suse.de>
13708 PR tree-optimization/80113
13709 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
13710 allocate extra SSA name for PHI def.
13711 (add_close_phis_to_outer_loops): Likewise.
13712 (add_close_phis_to_merge_points): Likewise.
13713 (copy_loop_close_phi_args): Likewise.
13714 (copy_cond_phi_nodes): Likewise.
13716 2017-03-20 Martin Liska <mliska@suse.cz>
13718 PR middle-end/79753
13719 * tree-chkp.c (chkp_build_returned_bound): Do not build
13720 returned bounds for a LHS that's not a BOUNDED_P type.
13722 2017-03-20 Martin Liska <mliska@suse.cz>
13726 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
13727 COMPLEX_CST and VECTOR_CST.
13729 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13732 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
13733 target operand. A new splitter adds the clobber statement in case
13734 the target operand is dead anyway.
13736 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
13738 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
13739 to age-old versions of binutils and glibc.
13741 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
13743 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
13745 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
13747 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
13749 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
13751 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
13752 requirement for binutils 2.13.
13754 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
13756 * combine.c (try_combine): Delete redundant i1 test. Call
13757 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
13759 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
13761 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
13762 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
13764 <riscv64-*-elf>: Re-arrange section
13765 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
13766 <riscv32-*-linux>: Likewise.
13767 <riscv64-*-elf>: Likewise
13768 <riscv64-*-linux>: Likewise.
13770 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
13773 * aarch64.opt(verbose-cost-dump): Fix typo.
13775 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
13778 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
13779 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
13781 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
13783 * reload.c (find_reloads): When reloading a nonoffsettable address,
13784 use RELOAD_OTHER for it and its address reloads.
13786 PR rtl-optimization/79910
13787 * combine.c (record_used_regs): New static function.
13788 (try_combine): Handle situations where there is an additional
13789 instruction between I2 and I3 which needs to have a LOG_LINK
13792 2017-03-17 Jeff Law <law@redhat.com>
13794 PR tree-optimization/71437
13795 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
13796 conditional in the hash table first.
13797 (vrp_dom_walker::before_dom_children): Extract condition from
13798 ASSERT_EXPR. Record condition, its inverion and any implied
13799 conditions as well.
13801 2017-03-17 Marek Polacek <polacek@redhat.com>
13802 Markus Trippelsdorf <markus@trippelsdorf.de>
13804 PR tree-optimization/80079
13805 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
13808 2017-03-17 Richard Biener <rguenther@suse.de>
13810 PR middle-end/80075
13811 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
13812 Properly verify the LHS before the RHS possibly claims to be
13814 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
13817 2017-03-17 Martin Jambor <mjambor@suse.cz>
13819 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
13820 (List of -O2 options): Likewise.
13821 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
13824 2017-03-17 Tom de Vries <tom@codesourcery.com>
13826 * gcov-dump.c (print_usage): Print bug_report_url.
13828 2017-03-17 Richard Biener <rguenther@suse.de>
13830 PR middle-end/80050
13831 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
13832 (parser::peek): Likewise.
13834 2017-03-17 Richard Biener <rguenther@suse.de>
13836 PR tree-optimization/80048
13837 * sese.c (free_sese_info): Properly release rename_map and
13838 copied_bb_map elements.
13840 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
13842 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
13843 Add linked-list forward and backlinks. Insert on
13844 construction, remove on destruction.
13845 (class pass_store_merging): Add m_stores_head field.
13846 (pass_store_merging::terminate_and_process_all_chains):
13847 Iterate over m_stores_head list.
13848 (pass_store_merging::terminate_all_aliasing_chains):
13850 (pass_store_merging::execute): Check for debug stmts first.
13851 Push new chains onto the m_stores_head stack.
13853 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
13856 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
13857 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
13858 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
13860 2017-03-16 Jeff Law <law@redhat.com>
13862 PR tree-optimization/71437
13863 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
13864 member function. Implementation moved into after_dom_children
13865 member function and into the threader's thread_outgoing_edges
13867 (dom_opt_dom_walker::after_dom_children): Simplify by moving
13868 some code into new thread_outgoing_edges.
13869 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
13870 definition. Simplify marker handling (do it here). Assume we always
13871 have the available expression and the const/copies tables.
13872 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
13874 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
13875 * tree-vrp.c (equiv_stack): No longer file scoped.
13876 (vrp_dom_walker): New class.
13877 (vrp_dom_walker::before_dom_children): New member function.
13878 (vrp_dom_walker::after_dom_children): Likewise.
13879 (identify_jump_threads): Setup domwalker. Use it rather than
13880 walking edges in a random order by hand. Simplify setup/finalization.
13881 (finalize_jump_threads): Remove.
13882 (vrp_finalize): Do not call identify_jump_threads here.
13883 (execute_vrp): Do it here instead and call thread_through_all_blocks
13886 PR tree-optimization/71437
13887 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
13889 (simplify_stmt_for_jump_threading): Add basic_block argument. All
13891 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
13892 (dom_opt_dom_walker::thread_across_edge): Remove
13893 handle_dominating_asserts argument. All callers changed.
13894 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
13895 changes. Remove calls to lhs_of_dominating_assert. Other
13896 uses of handle_dominating_asserts turn into unconditional code
13897 (simplify_control_stmt_condition_1): Likewise.
13898 (simplify_control_stmt_condition): Likewise.
13899 (thread_through_normal_block, thread_across_edge): Likewise.
13900 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
13901 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
13902 object if it is not an SSA_NAME.
13903 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
13904 before calling into the VRP specific simplifiers.
13905 (identify_jump_threads): Remove handle_dominating_asserts
13908 2017-03-16 Jakub Jelinek <jakub@redhat.com>
13911 * tree-diagnostic.c (default_tree_printer): No longer static.
13912 * tree-diagnostic.h (default_tree_printer): New prototype.
13914 2017-03-16 Tamar Christina <tamar.christina@arm.com>
13916 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
13917 Change ins into fmov.
13919 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13921 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
13922 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
13923 Use h_con constraint for operand 1.
13924 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
13925 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
13927 2017-03-15 Jeff Law <law@redhat.com>
13929 PR tree-optimization/71437
13930 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
13931 (record_temporary_equivalences): Use it.
13933 PR tree-optimization/71437
13934 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
13935 tree-ssa-scopedtables.
13936 (lookup_avail_expr, build_and_record_new_cond): Likewise.
13937 (record_conditions, record_cond, vuse_eq): Likewise.
13938 (record_edge_info): Adjust to API tweak of record_conditions.
13939 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
13940 (record_temporary_equivalences, optimize_stmt): Likewise.
13941 (eliminate_redundant_computations): Likewise.
13942 (record_equivalences_from_stmt): Likewise.
13943 * tree-ssa-scopedtables.c: Include options.h and params.h.
13944 (vuse_eq): New function, moved from tree-ssa-dom.c
13945 (build_and_record_new_cond): Likewise.
13946 (record_conditions): Likewise. Accept vector of conditions rather
13947 than edge_equivalence structure for first argument.
13948 for the first argument.
13949 (avail_exprs_stack::lookup_avail_expr): New member function, moved
13950 from tree-ssa-dom.c.
13951 (avail_exprs_stack::record_cond): Likewise.
13952 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
13953 from tree-ssa-dom.c.
13954 (avail_exprs_stack): Add new member functions lookup_avail_expr
13956 (record_conditions): Declare.
13958 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
13961 * lra-constraints.c (process_alt_operands): Increase reject for
13962 reloading an input/output operand.
13964 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
13967 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
13968 insns to convert from signed/unsigned char/short to IEEE 128-bit
13970 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
13972 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
13975 * config/i386/i386.c (ix86_vector_duplicate_value): Create
13976 subreg of inner mode for values already in registers.
13978 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
13980 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
13981 iteration reg is used after the loop.
13983 2017-03-14 Martin Sebor <msebor@redhat.com>
13985 PR tree-optimization/79800
13986 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
13987 precision in negative-positive range.
13988 (format_floating): Call non-const overload with adjusted precision.
13990 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
13993 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
13996 2017-03-14 Martin Sebor <msebor@redhat.com>
13998 PR middle-end/80020
13999 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
14000 * builtins.def (aligned_alloc): Use it.
14003 * Makefile.in (GTFILES): Add calls.c.
14004 * calls.c: Include "gt-calls.h".
14006 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
14008 PR rtl-optimization/79728
14009 * regs.h (struct target_regs): New field
14010 x_contains_allocatable_regs_of_mode.
14011 (contains_allocatable_regs_of_mode): New macro.
14012 * reginfo.c (init_reg_sets_1): Initialize it, and change
14013 contains_reg_of_mode so it includes global regs as well.
14014 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
14015 rather than contains_regs_of_mode.
14017 2017-03-14 Martin Liska <mliska@suse.cz>
14019 * doc/invoke.texi: Document options that can't be combined with
14020 -fcheck-pointer-bounds.
14022 2017-03-14 Martin Liska <mliska@suse.cz>
14024 PR middle-end/79831
14025 * doc/invoke.texi (-Wchkp): Document the option.
14027 2017-03-14 Martin Liska <mliska@suse.cz>
14029 * Makefile.in: Install gcov-dump.
14031 2017-03-14 Martin Liska <mliska@suse.cz>
14033 * multiple_target.c (expand_target_clones): Bail out for
14034 an invalid attribute.
14036 2017-03-14 Richard Biener <rguenther@suse.de>
14038 * alias.c (struct alias_set_entry): Pack properly.
14039 * cfgloop.h (struct loop): Likewise.
14040 * cse.c (struct set): Likewise.
14041 * ipa-utils.c (struct searchc_env): Likewise.
14042 * loop-invariant.c (struct invariant): Likewise.
14043 * lra-remat.c (struct cand): Likewise.
14044 * recog.c (struct change_t): Likewise.
14045 * rtl.h (struct address_info): Likewise.
14046 * symbol-summary.h (function_summary): Likewise.
14047 * tree-loop-distribution.c (struct partition): Likewise.
14048 * tree-object-size.c (struct object_size_info): Likewise.
14049 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
14050 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
14051 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
14052 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
14053 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
14054 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
14055 (struct _stmt_vec_info): Likewise.
14057 2017-03-14 Martin Liska <mliska@suse.cz>
14060 * multiple_target.c (create_dispatcher_calls): Check that
14061 a target can create a function dispatcher.
14063 2017-03-14 Martin Liska <mliska@suse.cz>
14066 * multiple_target.c (expand_target_clones): Drop local.local
14067 flag for default implementation.
14069 2017-03-14 Richard Biener <rguenther@suse.de>
14071 PR tree-optimization/80030
14072 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
14074 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
14076 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
14077 gcc_fallthrough() instead of __attribute__((fallthrough));
14079 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
14081 * doc/gcc.texi: Remove "up" link to (DIR).
14082 * doc/gccint.texi: Ditto.
14084 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
14086 * doc/install.texi (Specific) <avr>: Remove reference to
14089 2017-03-13 Jeff Law <law@redhat.com>
14091 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
14092 attribute rather than comments.
14094 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
14095 match_scratch operand is highest.
14097 2017-03-13 Martin Liska <mliska@suse.cz>
14099 PR middle-end/78339
14100 * ipa-pure-const.c (warn_function_noreturn): If the declarations
14101 is a CHKP clone, use original declaration.
14103 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
14105 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
14106 (arc_conditional_register_usage): Use a different allocation order
14107 when optimizing for size.
14108 * common/config/arc/arc-common.c (arc_option_optimization_table):
14109 Section anchors default on when optimizing for size.
14111 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
14113 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
14115 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
14117 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
14118 * config/arc/arc.md (cpu_facility): Add cd variant.
14119 (*movqi_insn): Add code density variant.
14120 (*movhi_insn): Likewise.
14121 (*movqi_insn): Likewise.
14122 (*addsi3_mixed): Likewise.
14123 (subsi3_insn): Likewise.
14125 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
14127 * config/arc/arc.md (movsi_cond_exec): Update constraint.
14129 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
14131 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
14132 expressions with MINUS and UNARY ops.
14134 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14137 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
14139 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
14140 between vec_select and vector argument.
14141 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
14142 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
14143 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
14144 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
14145 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
14146 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
14148 2017-03-13 Richard Biener <rguenther@suse.de>
14151 * params.def (vect-max-peeling-for-alignment): Fix typo.
14153 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
14155 * doc/install.texi (Specific) <mips-*-*>: Remove description of
14156 issue that only occurred with binutils below 2.18.
14158 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
14160 * doc/install.texi (Specific) <cris-axis-elf>: No longer
14161 refer to binutils 2.11/2.12 minimum.
14163 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
14165 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
14166 ftp.kernel.org and simplify binutils requirement.
14168 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
14170 * doc/invoke.texi (Warning Options): Fix spelling of link-time
14172 (Optimize Options): Ditto. Also remove redundancy.
14174 2017-03-10 David Malcolm <dmalcolm@redhat.com>
14176 PR translation/79848
14177 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
14179 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
14180 to G_ to avoid double translation.
14182 2017-03-10 David Malcolm <dmalcolm@redhat.com>
14184 PR translation/79923
14185 * auto-profile.c (get_combined_location): Convert leading
14186 character of diagnostics to lower case and remove trailing period.
14187 (read_profile): Likewise for various diagnostics.
14188 * config/arm/arm.c (arm_option_override): Remove trailing period
14189 from various diagnostics.
14190 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
14191 (msp430_expand_delay_cycles): Likewise.
14193 2017-03-10 David Malcolm <dmalcolm@redhat.com>
14196 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
14197 full command-line argument, rather than just "str".
14198 (aarch64_validate_march): Likewise.
14199 (aarch64_validate_mtune): Likewise.
14201 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
14203 PR rtl-optimization/78911
14204 * lra-assigns.c (must_not_spill_p): New function.
14205 (spill_for): Use it.
14207 2017-03-10 Jakub Jelinek <jakub@redhat.com>
14209 PR tree-optimization/79981
14210 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
14211 ATOMIC_COMPARE_EXCHANGE ifn result.
14212 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
14213 IFN_ATOMIC_COMPARE_EXCHANGE.
14215 2017-03-10 David Malcolm <dmalcolm@redhat.com>
14218 * opts.c (parse_sanitizer_options): Add missing question mark to
14219 "did you mean" message.
14221 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14223 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
14225 (VMULEUH_UNS): Likewise.
14226 (VMULOUB_UNS): Likewise.
14227 (VMULOUH_UNS): Likewise.
14228 * config/rs6000/rs6000.c (builtin_function_type): Remove
14229 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
14231 2017-03-10 David Malcolm <dmalcolm@redhat.com>
14234 * read-rtl-function.c (function_reader::read_rtx_operand): Update
14235 x with result of extra_parsing_for_operand_code_0.
14236 (function_reader::extra_parsing_for_operand_code_0): Convert
14237 return type from void to rtx, returning x. When reading
14238 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
14239 larger size containing struct block_symbol.
14241 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
14243 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
14244 -mfloat128-hardware without -m64.
14246 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
14249 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
14250 entries to the case statement that marks unsigned arguments to
14251 overloaded functions.
14253 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
14255 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
14256 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
14258 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
14261 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
14262 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
14264 2017-03-10 Martin Liska <mliska@suse.cz>
14268 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
14269 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
14272 2017-03-10 Olivier Hainque <hainque@adacore.com>
14274 * tree-switch-conversion (array_value_type): Start by resetting
14275 candidate type to it's main variant.
14277 2017-03-10 Jakub Jelinek <jakub@redhat.com>
14279 PR rtl-optimization/79909
14280 * combine.c (try_combine): Use simplify_replace_rtx on individual
14281 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
14282 of the whole CALL_INSN_FUNCTION_USAGE.
14284 PR tree-optimization/79972
14285 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
14286 get_range_info on SSA_NAMEs. Formatting fixes.
14288 2017-03-10 Richard Biener <rguenther@suse.de>
14289 Jakub Jelinek <jakub@redhat.com>
14291 PR tree-optimization/77975
14292 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
14293 edge to be constant.
14294 (get_val_for): For constant x return it. Formatting fix.
14295 (loop_niter_by_eval): Avoid pointless looping if the next iteration
14296 would use the same bases as the current one.
14298 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14300 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
14301 instead of vec_select for V1TImode.
14302 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
14304 (VSX_LE_128): Add V1TI to this mode iterator.
14305 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
14306 (*vsx_le_perm_store_<mode>): Likewise.
14307 (pre-reload splitter for VSX stores): Likewise.
14308 (post-reload splitter for VSX stores): Likewise.
14309 (*vsx_xxpermdi2_le_<mode>): Likewise.
14310 (*vsx_lxvd2x2_le_<mode>): Likewise.
14311 (*vsx_stxvd2x2_le_<mode>): Likewise.
14313 2017-03-09 Michael Eager <eager@eagercon.com>
14315 Correct failures with --enable-checking=yes,rtl.
14317 * config/microblaze/microblaze.c (microblaze_expand_shift):
14318 Replace GET_CODE test with CONST_INT_P and INTVAL test with
14319 test for const0_rtx.
14320 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
14321 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
14323 2017-03-09 Richard Biener <rguenther@suse.de>
14325 PR tree-optimization/79977
14326 * graphite-scop-detection.c (scop_detection::merge_sese):
14327 Handle the case of extra exits to blocks dominating the entry.
14329 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
14331 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
14334 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
14336 PR rtl-optimization/79949
14337 * lra-constraints.c (process_alt_operands): Check memory when
14338 trying to predict a cycle. Print about the overall increase.
14340 2017-03-09 Richard Biener <rguenther@suse.de>
14342 PR middle-end/79971
14343 * gimple-expr.c (useless_type_conversion_p): Preserve
14344 TYPE_SATURATING for fixed-point types.
14346 2017-03-09 Richard Biener <rguenther@suse.de>
14349 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
14350 alignment of BLKmode params.
14352 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14355 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
14356 (VALL_NO_V2Q): Likewise.
14358 * config/aarch64/aarch64-simd.md
14359 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
14361 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
14362 VALL_NO_V2Q mode iterator.
14363 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
14365 2017-03-09 Martin Liska <mliska@suse.cz>
14367 PR tree-optimization/79631
14368 * tree-chkp-opt.c (chkp_is_constant_addr): Call
14369 tree_int_cst_sign_bit just for INTEGER constants.
14371 2017-03-09 Martin Liska <mliska@suse.cz>
14375 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
14378 2017-03-09 Marek Polacek <polacek@redhat.com>
14381 * tree.c (inchash::add_expr): Handle TREE_VEC.
14383 2017-03-09 Martin Liska <mliska@suse.cz>
14386 (chkp_narrow_size_and_offset): New function.
14387 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
14388 (void chkp_parse_bit_field_ref): New function.
14389 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
14390 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
14392 2017-03-09 Martin Liska <mliska@suse.cz>
14395 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
14396 (chkp_find_bounds_1): Remove gcc_unreachable.
14398 2017-03-09 Jakub Jelinek <jakub@redhat.com>
14401 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
14402 BUILT_IN_SYNC*, determine the access type from the size suffix and
14403 always build a MEM_REF with that type. Handle forgotten
14404 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
14407 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
14408 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
14409 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
14410 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
14411 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
14412 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
14413 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
14414 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
14415 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
14416 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
14417 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
14418 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
14419 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
14420 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
14421 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
14422 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
14423 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
14424 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
14425 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
14426 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
14427 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
14428 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
14429 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
14430 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
14431 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
14432 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
14433 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
14434 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
14435 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
14436 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
14437 definitions outside of __OPTIMIZE__ guarded section.
14440 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
14441 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
14442 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
14443 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
14446 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14448 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
14449 ("vfenez<mode>"): Add missing constraints.
14451 2017-03-08 Martin Sebor <msebor@redhat.com>
14454 * config/nds32/nds32.c (nds32_option_override):
14455 Fix misspelled diagnostic.
14457 2017-03-08 Jakub Jelinek <jakub@redhat.com>
14460 * gimplify.c (gimplify_omp_for): Replace index var in outer
14461 taskloop statement with an artificial variable and add
14462 OMP_CLAUSE_PRIVATE clause for it.
14464 2017-03-08 Richard Biener <rguenther@suse.de>
14466 PR tree-optimization/79955
14467 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
14468 for accesses that are completely outside of the variable.
14470 2017-03-08 Andrew Haley <aph@redhat.com>
14472 PR tree-optimization/79943
14473 * tree-ssa-loop-split.c (compute_new_first_bound): When
14474 calculating the new upper bound, (END-BEG) should be added, not
14477 2017-03-08 Jakub Jelinek <jakub@redhat.com>
14479 * config/avr/avr.md (setmemhi): Make sure match_dup
14480 operand number comes before match_scratch.
14482 2017-03-08 Richard Biener <rguenther@suse.de>
14484 PR tree-optimization/79920
14485 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
14486 with ncopies == 1 to ...
14487 (vect_transform_slp_perm_load): ... here. Properly compute
14488 all element loads by iterating VF times over the group. Do
14489 not handle ncopies (computed in a broken way) in
14490 vect_create_mask_and_perm.
14492 2017-03-08 Jakub Jelinek <jakub@redhat.com>
14495 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
14496 is a uniform vector, use uniform_vector_p return value instead of
14497 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
14499 2017-03-07 Marek Polacek <polacek@redhat.com>
14501 PR middle-end/79809
14502 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
14503 (alloca_call_type): Likewise.
14505 2017-03-07 Martin Liska <mliska@suse.cz>
14507 * gcov.c (process_args): Put comment to correct location.
14509 2017-03-07 Martin Liska <mliska@suse.cz>
14511 PR middle-end/68270
14512 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
14513 Use array_at_struct_end_p instead of DECL_CHAIN (field).
14514 (chkp_narrow_bounds_for_field): Likewise.
14515 (chkp_parse_array_and_component_ref): Pass one more argument to
14518 2017-03-07 Richard Biener <rguenther@suse.de>
14520 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
14523 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
14525 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
14526 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
14528 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14531 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
14532 to end of description.
14533 (PARAM_MAX_STORES_TO_MERGE): Likewise.
14535 2017-03-07 Jakub Jelinek <jakub@redhat.com>
14537 PR rtl-optimization/79901
14538 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
14540 (*avx512f_<code><mode>3<mask_name>): ... this.
14541 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
14542 iterator instead of VI8_AVX2_AVX512BW.
14544 PR rtl-optimization/79901
14545 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
14546 min/max expander, expand it using expand_vec_cond_expr.
14549 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
14552 2017-03-06 Jakub Jelinek <jakub@redhat.com>
14555 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
14556 to void * for PCH reasons.
14557 * dwarf2out.c (output_loc_operands, output_die): Cast
14558 v.val_vec.array to unsigned char *.
14560 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
14563 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
14566 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
14568 PR rtl-optimization/79571
14569 * lra-constraints.c (process_alt_operands): Calculate static
14570 reject and subtract it from overall when only addresses will be
14573 2017-03-06 Julia Koval <julia.koval@intel.com>
14576 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
14577 incoming stack boundary to 128 for 64-bit targets.
14579 2017-03-06 Richard Biener <rguenther@suse.de>
14581 PR tree-optimization/79894
14582 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
14583 to NULL after folding it.
14585 2017-03-06 Richard Biener <rguenther@suse.de>
14587 PR tree-optimization/79824
14588 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
14589 check disabling peeling for gaps.
14591 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
14593 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
14594 attributes): Document gettimeofday.
14596 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
14598 * config/s390/s390.c (s390_option_override_internal): Set
14599 PARAM_MIN_VECT_LOOP_BOUND
14601 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
14603 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
14604 * config/s390/s390.md: Likewise.
14606 2017-03-06 Jakub Jelinek <jakub@redhat.com>
14609 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
14610 (<avx2_avx512>_perm<mode>): Rename to ...
14611 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
14613 (<avx512>_perm<mode>_mask): Rename to ...
14614 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
14616 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
14617 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
14618 instead of VI8F_256_512.
14619 (avx512f_perm<mode>): New define_expand.
14620 (avx512f_perm<mode>_mask): Likewise.
14621 (avx512f_perm<mode>_1<mask_name>): New define_insn.
14622 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
14624 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
14626 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
14627 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
14629 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
14631 2017-03-06 Martin Liska <mliska@suse.cz>
14634 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
14635 when having a SSA NAME w/o VAR_DECL assigned to it.
14637 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
14639 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
14640 msa_dpsub_<su>_d): Fix MODE for vec_select.
14642 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
14644 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
14646 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
14648 2017-03-06 Richard Biener <rguenther@suse.de>
14650 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
14651 * plugin.c (register_plugin_info): Likewise.
14652 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
14654 2017-03-05 Jakub Jelinek <jakub@redhat.com>
14656 * config/i386/sse.md (sse_storehps, sse_storelps,
14657 avx_<castmode><avxsizesuffix>_<castmode>,
14658 avx512f_<castmode><avxsizesuffix>_<castmode>,
14659 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
14660 in condition that at least one operand is not a MEM.
14662 2017-03-03 Jakub Jelinek <jakub@redhat.com>
14664 PR middle-end/79805
14665 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
14666 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
14668 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
14669 gimple_call_nothrow_p flag based on whether original builtin can throw.
14670 If it can, emit following stmts on the fallthrough edge.
14671 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
14672 don't create new bb if inserting just debug stmts on the edge, try to
14673 insert them on the fallthru bb or just reset debug stmts.
14675 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
14678 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
14679 restore recog_data (including the operand rtxes inside it) around
14680 the call to get_insn_template.
14682 2017-03-03 Martin Sebor <msebor@redhat.com>
14684 PR tree-optimization/79699
14685 * context.c (context::~context): Free MPFR caches to avoid
14686 a memory leak on program exit.
14688 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14690 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
14691 Use wide_int::ulow () instead of .elt (0).
14693 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
14695 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
14696 (*pushxf): Limit oF constraint to 32bit targets and add oC
14697 constraint for 64bit targets.
14698 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
14699 (*pushdf): Change rmF constraint to rmC.
14701 2017-03-03 Martin Liska <mliska@suse.cz>
14703 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
14704 Remove unused variable.
14706 2017-03-03 Jakub Jelinek <jakub@redhat.com>
14709 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
14710 is a memory operand, increase num_memory.
14711 (ix86_expand_args_builtin): Likewise.
14713 2017-03-03 Jan Hubicka <jh@suse.cz>
14716 * ipa-devirt.c (maybe_record_node): Properly handle
14717 __cxa_pure_virtual visibility.
14719 2017-03-03 Martin Liska <mliska@suse.cz>
14721 PR tree-optimization/79803
14722 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
14724 (pass_loop_prefetch::execute): Disabled optimization if an
14725 assumption about L1 cache size is not met.
14727 2017-03-03 Martin Liska <mliska@suse.cz>
14729 PR rtl-optimization/79574
14730 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
14731 (hash_scan_set): Likewise.
14732 (dump_hash_table): Likewise.
14733 (hoist_code): Likewise.
14735 2017-03-03 Richard Biener <rguenther@suse.de>
14737 * fixed-value.c (fixed_from_string): Restore use of elt (1)
14738 in place of uhigh ().
14739 (fixed_convert_from_real): Likewise.
14741 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
14744 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
14746 2017-03-03 Richard Biener <rguenther@suse.de>
14748 PR middle-end/79818
14749 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
14750 TYPE_OVERFLOW_UNDEFINED check.
14752 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14754 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
14756 (vector_ae_<mode>_p): Likewise.
14757 (vector_nez_<mode>_p): Likewise.
14758 (vector_ne_v2di_p): Likewise.
14759 (vector_ae_v2di_p): Likewise.
14760 (vector_ne_<mode>_p): Likewise.
14761 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
14763 (vsx_tsqrt<mode>2_fe): Likewise.
14765 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
14768 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
14770 2017-03-02 Jakub Jelinek <jakub@redhat.com>
14772 PR rtl-optimization/79780
14773 * cprop.c (one_cprop_pass): When second and further conditional trap
14774 in a single basic block is turned into an unconditional trap, turn it
14775 into a deleted note to avoid RTL verification failures.
14777 2017-03-02 Richard Biener <rguenther@suse.de>
14779 * fold-const.c (const_binop): Use ulow () instead of elt (0).
14781 2017-03-02 Richard Biener <rguenther@suse.de>
14783 PR tree-optimization/79345
14785 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
14786 param and abort the walk, returning -1 if it is hit.
14787 (walk_aliased_vdefs): Take a limit param and pass it on.
14788 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
14789 defaulting to 0 and return a signed int.
14790 * tree-ssa-uninit.c (struct check_defs_data): New struct.
14791 (check_defs): New helper.
14792 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
14793 about uninitialized memory.
14794 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
14795 bogus uninitialized warning.
14796 (fixed_convert_from_real): Likewise.
14798 2017-03-02 Bin Cheng <bin.cheng@arm.com>
14800 PR tree-optimization/66768
14801 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
14802 iv_use if base object can't be determined.
14804 2017-03-02 Jakub Jelinek <jakub@redhat.com>
14806 PR tree-optimization/79345
14807 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
14808 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
14809 (get_pattern_stats): Initialize it.
14810 * genemit.c (gen_expand): Verify match_scratch numbers come after
14811 match_operand/match_dup numbers.
14812 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
14813 match_scratch numbers.
14814 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
14816 * config/s390/s390.md (trunctdsd2): Likewise.
14818 2017-03-02 Richard Biener <rguenther@suse.de>
14820 * wide-int.h (wide_int_storage::operator=): Implement in terms
14823 2017-03-02 Richard Biener <rguenther@suse.de>
14825 PR tree-optimization/79777
14826 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
14827 the to insert expression to sth existing.
14829 2017-03-01 Martin Sebor <msebor@redhat.com>
14831 PR middle-end/79692
14832 * gimple-ssa-sprintf.c
14833 (directive::known_width_and_precision): New function.
14834 (format_integer): Use it.
14835 (get_mpfr_format_length): Consider the full range of precision
14836 when computing %g output with the # flag. Set the likely byte
14837 count to 3 rather than 1 when precision is indeterminate.
14838 (format_floating): Correct the lower bound of precision.
14840 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14842 * doc/invoke.texi: Document default code model for 64-bit Linux.
14844 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
14847 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
14848 udiv rather than div since input pattern is unsigned.
14850 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
14852 * config/i386/i386.c (print_reg): Warn for values of
14853 unsupported size in integer register.
14855 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
14858 * config/rs6000/predicates.md (current_file_function_operand): Do
14859 not allow self calls to be local if the function is replaceable.
14861 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
14864 * config/rs6000/altivec.h (vec_ctz and others): Change the
14865 preprocessor macro that controls conditional compilation from
14866 _ARCH_PWR9 to __POWER9_VECTOR__.
14867 (vec_all_ne): Change parameterization of __altivec_scalar_pred
14868 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
14869 control (instead of _ARCH_PWR9 control) so that template
14870 definition uses power9-specific function.
14871 (vec_any_eq): Likewise.
14872 (vec_all_ne): Change macro definition to use a power9-specific
14873 expansion under #ifdef __POWER9_VECTOR__ control (instead of
14874 _ARCH_PWR9 control).
14875 (vec_any_eq) Likewise.
14876 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
14877 expansion for CMPNEF to remove support for xvcmpnesp instruction.
14878 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
14879 support for xvcmpnedp instruction.
14880 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
14881 macro expansion so that Power9 implementation of vec_all_ne does
14882 not use the AltiVec predicate framework.
14883 (VCMPNEH_P): Likewise.
14884 (VCMPNEW_P): Likewise.
14885 (VCMPNED_P): Likewise.
14886 (VCMPNEFP_P): Likewise.
14887 (VCMPNEDP_P): Likewise.
14888 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
14889 implementation of vec_any_eq to not use AltiVec predicate
14891 (VCMPAEH_P): Likewise.
14892 (VCMPAEW_P): Likewise.
14893 (VCMPAED_P): Likewise.
14894 (VCMPAEFP_P): Likewise.
14895 (VCMPAEDP_P): Likewise.
14896 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
14897 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
14898 not use the AltiVec predicate framework.
14899 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
14900 of vec_any_eq to not use AltiVec predicate framework.
14901 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
14902 support for predefined __POWER9_VECTOR__ macro to indicate that
14903 Power9 instruction selection is enabled.
14904 (altivec_overloaded_builtins): Remove extraneous
14905 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
14906 function argument types RS6000_BTI_bool_V16QI and
14907 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
14908 entry for overloaded function argument types RS6000_BTI_bool_V4SI
14909 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
14910 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
14911 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
14912 Power9 for implementations of vec_cmpne. Change the signature for
14913 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
14914 (representing vec_all_ne) to remove the previously described first
14915 argument of type RS6000_BTI_INTSI, as this was an artifact of
14916 reliance on the AltiVec predicate framework, which is no longer
14917 used in the implementation of these functions. Add
14918 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
14919 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
14920 since, unlike the AltiVec predicate framework implementation, we
14921 do not share function descriptors between vec_alle and vec_anyeq.
14922 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
14923 set of modes that receive special treatment even when
14924 TARGET_P9_VECTOR is true. The special treatment emits code that
14925 does not depend on Power9 instructions.
14926 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
14927 define_expand to not rely on AltiVec predicate framework.
14928 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
14930 (vector_ne_v2di_p): Change this define_expand to not rely on
14931 AltiVec predicate framework.
14932 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
14934 (vector_ne_<mode>_p): Change this define_expand to not rely on
14935 AltiVec predicate framework.
14936 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
14938 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
14939 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
14940 define_insn pattern.
14941 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
14942 define_insn pattern because the xvcmpne<VSs>. instruction is not
14944 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
14945 instruction is not supported.
14947 2017-03-01 Jakub Jelinek <jakub@redhat.com>
14949 * config/nvptx/nvptx.c: Include intl.h.
14951 2017-03-01 Martin Jambor <mjambor@suse.cz>
14954 * ipa-prop.h (ipa_bits): Removed field known.
14955 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
14956 to pointers. Adjusted their comments to warn about their sharing.
14957 (ipcp_transformation_summary): Change bits to a vector of pointers.
14958 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
14959 (ipa_get_ipa_bits_for_value): Declare.
14960 * tree-vrp.h (value_range): Mark as GTY((for_user)).
14961 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
14962 (ipa_bits_hash_table): Likewise.
14963 (ipa_vr_ggc_hash_traits): Likewise.
14964 (ipa_vr_hash_table): Likewise.
14965 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
14966 being pointers and vr_known being removed.
14967 (ipa_set_jf_unknown): Likewise.
14968 (ipa_get_ipa_bits_for_value): New function.
14969 (ipa_set_jfunc_bits): Likewise.
14970 (ipa_get_value_range): New overloaded functions.
14971 (ipa_set_jfunc_vr): Likewise.
14972 (ipa_compute_jump_functions_for_edge): Use the above functions to
14973 construct bits and vr parts of jump functions.
14974 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
14975 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
14977 (ipcp_grow_transformations_if_necessary): Also allocate
14978 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
14980 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
14981 them. Fix too long lines.
14982 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
14983 vr_known being removed.
14984 (ipa_read_jump_function): Use new setter functions to construct bits
14985 and vr parts of jump functions or set them to NULL.
14986 (write_ipcp_transformation_info): Adjust for bits being pointers.
14987 (read_ipcp_transformation_info): Likewise.
14988 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
14990 Include gt-ipa-prop.h.
14991 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
14993 (ipcp_store_bits_results): Likewise.
14994 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
14995 Do not write to existing jump functions but use a temporary instead.
14997 2017-03-01 Jakub Jelinek <jakub@redhat.com>
15000 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
15001 attempt to use its first operand as BIT_FIELD_REF base.
15003 2017-03-01 Richard Biener <rguenther@suse.de>
15005 PR middle-end/79721
15006 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
15007 interpolating formula in wrapping arithmetic.
15008 (chrec_apply): Convert chrec_evaluate return value to wanted type.
15010 2017-03-01 Jakub Jelinek <jakub@redhat.com>
15012 PR tree-optimization/79734
15013 * tree-vect-generic.c (expand_vector_condition): Optimize
15014 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
15015 Handle VEC_COND_EXPR where comparison has different inner width from
15016 type's inner width.
15018 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
15020 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
15021 markup, and similar issues. Remove @opindex entries for things
15022 that aren't options. Add missing -mmpy-option entries.
15024 2017-02-28 Jakub Jelinek <jakub@redhat.com>
15026 PR tree-optimization/79737
15027 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
15028 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
15029 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
15030 instead of byte_size. Formatting fix.
15031 (shift_bytes_in_array_right): Formatting fix.
15033 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
15036 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
15037 condition on optimize for the leaf function test.
15039 2017-02-28 Martin Liska <mliska@suse.cz>
15042 * read-rtl-function.c (function_reader::handle_unknown_directive):
15043 Bail out when one uses -flto.
15045 2017-02-28 Martin Liska <mliska@suse.cz>
15047 * common.opt: Replace space with tabular for options of <number>
15049 * config/i386/i386.opt: Show <number> value for
15050 -mlarge-data-threshold.
15051 * opts.c (print_filtered_help): Do not display number in hexadecimal
15054 2017-02-28 Martin Liska <mliska@suse.cz>
15056 * common.opt: Fix --help=option -Q for options which are of
15059 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
15061 * config/i386/i386.c (print_reg): Error out for values
15062 of 8-bit size in invalid integer register.
15064 2017-02-28 Martin Sebor <msebor@redhat.com>
15066 PR tree-optimization/79691
15067 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
15069 2017-02-28 Jakub Jelinek <jakub@redhat.com>
15072 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
15073 gcc_unreachable with output_operand_lossage.
15075 2017-02-28 Richard Biener <rguenther@suse.de>
15077 PR tree-optimization/79740
15078 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
15080 (visit_nary_op): Insert the nary into the hashtable if we
15081 pattern-matched sth.
15082 * tree-ssa-pre.c (eliminate_insert): Robustify.
15084 2017-02-28 Richard Biener <rguenther@suse.de>
15086 PR middle-end/79731
15087 * fold-const.c (decode_field_reference): Reject out-of-bound
15090 2017-02-28 Jakub Jelinek <jakub@redhat.com>
15092 * config/i386/i386.c: Include intl.h.
15093 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
15094 instead of just cond ? "..." : "...".
15095 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
15096 * coverage.c (read_counts_file): Likewise.
15097 * omp-offload.c: Include intl.h.
15098 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
15099 of just cond ? "..." : "...".
15100 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
15101 of just cond ? "..." : "...".
15103 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
15106 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
15108 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
15109 'tune for' CPU name.
15110 * config/arm/arm-cpu-data.h: Regenerated.
15112 2017-02-28 Richard Biener <rguenther@suse.de>
15114 PR tree-optimization/79732
15115 * tree-inline.c (expand_call_inline): Do not shadow var.
15117 2017-02-28 Richard Biener <rguenther@suse.de>
15119 PR tree-optimization/79723
15120 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
15121 address-space properly.
15123 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
15125 * doc/optinfo.texi (Optimization groups): Fix option used for
15127 * doc/invoke.texi (-fopt-info): Document "omp".
15128 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
15129 (OPTGROUP_ALL): Add OPTGROUP_OMP.
15130 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
15131 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
15132 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
15134 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
15136 * dumpfile.c (optgroup_options): Instead of "openmp", associate
15137 OPTGROUP_OMP with "omp".
15139 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
15142 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
15143 for arithmetic shift of unsigned V2DI.
15145 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
15147 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
15148 arc/linux.h headers.
15149 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
15150 (LINK_SPEC): Likewise.
15151 (ARC_TLS_EXTRA_START_SPEC): Likewise.
15152 (EXTRA_SPECS): Likewise.
15153 (STARTFILE_SPEC): Likewise.
15154 (ENDFILE_SPEC): Likewise.
15155 (LIB_SPEC): Likewise.
15156 (TARGET_SDATA_DEFAULT): Likewise.
15157 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
15158 (MULTILIB_DEFAULTS): Likewise.
15159 (DWARF2_UNWIND_INFO): Likewise.
15160 * config/arc/big.h: New file.
15161 * config/arc/elf.h: Likewise.
15162 * config/arc/linux.h: Likewise.
15163 * config/arc/t-uClibc: Remove.
15165 2017-02-27 Bin Cheng <bin.cheng@arm.com>
15167 PR tree-optimization/77536
15168 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
15169 (tree_transform_and_unroll_loop): Use above function to compute the
15170 estimated niter of unrolled loop and use it when scaling profile.
15171 Also use count info rather than frequency if it's non-zero.
15172 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
15173 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
15174 (vect_transform_loop): Call above function.
15176 2017-02-27 Richard Biener <rguenther@suse.de>
15178 PR tree-optimization/45397
15179 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
15180 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
15181 (visit_nary_op): Add pattern matching for CSEing sign-changed
15182 or truncated operations with wider ones.
15184 2017-02-27 Richard Biener <rguenther@suse.de>
15186 PR tree-optimization/79690
15187 * tree-vect-stmts.c (vectorizable_store): Use vector type
15188 built from the DR with address-space.
15190 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
15192 * doc/invoke.texi (Optimize Options): Refine the description
15193 of asan-use-after-return.
15195 2017-02-25 Alan Modra <amodra@gmail.com>
15197 PR rtl-optimization/79584
15198 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
15199 base, not ad->base_term, the reg within base. Remove assertion
15200 that ad->base == ad->base_term. Replace gen_int_mode using
15201 bogus mode with const0_rtx.
15203 2017-02-25 Jakub Jelinek <jakub@redhat.com>
15205 PR middle-end/79396
15206 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
15207 FMA_EXPR like tcc_binary or tcc_unary.
15209 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
15212 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
15214 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
15215 (output_loc_operands): Handle DW_OP_call_ref and
15216 DW_OP_GNU_variable_value.
15217 (struct variable_value_struct): New type.
15218 (struct variable_value_hasher): Likewise.
15219 (variable_value_hash): New variable.
15220 (string_types): Remove.
15221 (copy_loc_descr): New function.
15222 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
15223 (prepend_loc_descr_to_each): New function.
15224 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
15225 instead of add_loc_descr_to_each if the first argument is single
15226 location list and the second has multiple.
15227 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
15228 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
15229 when looking for variable value which doesn't have other location info.
15230 (loc_list_from_tree): Formatting fix.
15231 (gen_array_type_die): Simplify DW_AT_string_length handling.
15232 (adjust_string_types): Remove.
15233 (gen_subprogram_die): Don't call adjust_string_types nor test/set
15234 string_types. Call resolve_variable_values.
15235 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
15236 (resolve_addr_in_expr): Likewise. Add A argument.
15237 (copy_deref_exprloc): Remove deref argument. Adjust for the
15238 original expression being DW_OP_GNU_variable_value with optionally
15239 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
15240 optionally after it.
15241 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
15242 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
15243 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
15244 (variable_value_hasher::hash, variable_value_hasher::equal): New
15246 (resolve_variable_value_in_expr, resolve_variable_value,
15247 resolve_variable_values, note_variable_value_in_expr,
15248 note_variable_value): New functions.
15249 (dwarf2out_early_finish): Call note_variable_value on all toplevel
15252 2017-02-24 Jakub Jelinek <jakub@redhat.com>
15255 * opts.h (handle_generated_option): Add GENERATED_P argument.
15256 * opts-common.c (handle_option): Adjust function comment.
15257 (handle_generated_option): Add GENERATED_P argument, pass it to
15259 (control_warning_option): Pass false to handle_generated_option
15261 * opts.c (maybe_default_option): Pass true to handle_generated_option
15263 * optc-gen.awk: Likewise.
15265 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
15267 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
15268 a REG, look at the REG it is a SUBREG of.
15269 (splitter for cmpeqsi_t): Ditto.
15271 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
15273 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
15274 the special USEs with the pattern of the insn, not the insn itself.
15276 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
15279 * doc/invoke.texi: Document -mload-store-pairs.
15281 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
15282 Sandra Loosemore <sandra@codesourcery.com>
15284 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
15285 argument isn't a CONST_INT.
15286 (nios2_alternate_compare_const): Assert op is a CONST_INT.
15287 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
15288 (nios2_validate_compare): Bypass alternate compare logic if *op2
15289 is not a CONST_INT.
15290 (ldstwm_operation_p): Return false if first_base is not a REG or
15291 if first_offset is not a CONST_INT.
15293 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
15295 * config/cris/cris.md: Use correct operand in a define_peephole2.
15297 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
15299 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
15301 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
15303 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
15304 this_insn if it is an INSN or JUMP_INSN.
15305 (force_offsettable): Look at base, not at addr.
15306 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
15307 on things that aren't necessarily CONST_INTs.
15309 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
15311 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
15312 -mfpmath=sse is the default also for x86-32 targets with SSE2
15313 instruction set when @option{-ffast-math} is enabled
15315 2017-02-24 Jeff Law <law@redhat.com>
15317 PR rtl-optimizatoin/79286
15318 * ira.c (update_equiv_regs): Drop may_trap_p exception to
15321 2017-02-24 Richard Biener <rguenther@suse.de>
15323 PR tree-optimization/79389
15324 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
15327 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
15329 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
15330 function comment to reflect reality.
15331 (loop_exits_before_overflow): Fix typo in function description.
15333 2017-02-24 Richard Biener <rguenther@suse.de>
15335 PR tree-optimization/79389
15336 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
15337 properly that a threading opportunity exists. Detect conditional
15338 copy/constant propagation opportunities.
15340 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
15342 * config/visium/visium.md (type): Add trap.
15343 (b): New mode attribute.
15344 (*btst): Rename into...
15345 (*btst<mode>): ...this and adjust.
15346 (*cbranchsi4_btst_insn): Rename into...
15347 (*cbranch<mode>4_btst_insn): ...this and adjust.
15348 (trap): New define_insn.
15350 2017-02-23 Jakub Jelinek <jakub@redhat.com>
15352 PR tree-optimization/79389
15353 * ifcvt.c (struct noce_if_info): Add rev_cond field.
15354 (noce_reversed_cond_code): New function.
15355 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
15356 reversed_comparison_code. Formatting fix.
15357 (noce_try_store_flag): Test rev_cond != NULL in addition to
15358 reversed_comparison_code.
15359 (noce_try_store_flag_constants): Likewise.
15360 (noce_try_store_flag_mask): Likewise.
15361 (noce_try_addcc): Use rev_cond if non-NULL instead of
15362 reversed_comparison_code.
15363 (noce_try_cmove_arith): Likewise. Formatting fixes.
15364 (noce_try_minmax, noce_try_abs): Clear rev_cond.
15365 (noce_find_if_block): Initialize rev_cond.
15366 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
15367 instead of false as last argument never attempt to reverse it
15370 2017-02-23 Bin Cheng <bin.cheng@arm.com>
15372 PR tree-optimization/79663
15373 * tree-predcom.c (combine_chains): Process refs in reverse order
15374 only for ZERO length chains, and add explaining comment.
15376 2017-02-23 Jeff Law <law@redhat.com>
15378 PR tree-optimization/79578
15379 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
15380 in call to operand_equal_p.
15382 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
15385 * config/i386/cpuid.h: Fix another undefined behavior.
15387 2017-02-23 Richard Biener <rguenther@suse.de>
15389 PR tree-optimization/79683
15390 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
15391 vector types for data-refs.
15393 2017-02-23 Martin Liska <mliska@suse.cz>
15395 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
15397 2017-02-23 Jakub Jelinek <jakub@redhat.com>
15399 PR middle-end/79665
15400 * internal-fn.c (get_range_pos_neg): Moved to ...
15401 * tree.c (get_range_pos_neg): ... here. No longer static.
15402 * tree.h (get_range_pos_neg): New prototype.
15403 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
15404 are known to be in between 0 and signed maximum inclusive, try to
15405 expand both unsigned and signed divmod and use the cheaper one from
15408 2017-02-22 Jeff Law <law@redhat.com>
15410 PR tree-optimization/79578
15411 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
15412 to compare base operands.
15414 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
15417 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
15418 gpc_reg_operand instead of fpr_reg_operand.
15420 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
15422 * config/mips/mips.c (mips_return_in_memory): Force FP
15423 vector types to be returned in memory for o32 ABI.
15425 2017-02-22 Jakub Jelinek <jakub@redhat.com>
15427 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
15428 instead of DW_TAG_member for static data member declarations and don't
15429 set no_linkage_name for static inline data members.
15430 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
15433 2017-02-22 Martin Liska <mliska@suse.cz>
15435 * doc/invoke.texi: Replace inequality signs with square brackets
15438 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15440 PR tree-optimization/68644
15441 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
15443 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
15446 * lra-constraints.c (simplify_operand_subreg): Handle
15447 WORD_REGISTER_OPERATIONS targets.
15449 2017-02-22 Jakub Jelinek <jakub@redhat.com>
15452 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
15453 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
15454 elimination by swapping fld*.
15456 2017-02-22 Richard Biener <rguenther@suse.de>
15458 PR tree-optimization/79673
15459 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
15460 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
15461 irrelevant address-space qualifiers and avoiding a
15462 ADDR_SPACE_CONVERT_EXPR from fold_convert.
15464 2017-02-22 Richard Biener <rguenther@suse.de>
15466 PR tree-optimization/79666
15467 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
15468 to not symbolically negate if that may introduce undefined
15471 2017-02-22 Martin Liska <mliska@suse.cz>
15474 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
15475 * data-streamer-out.c (streamer_write_gcov_count_stream):
15477 * value-prof.c (stream_out_histogram_value): Make assert more
15478 precise based on type of counter.
15480 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
15483 * config/i386/i386.md (standard_x87sse_constant_load splitter):
15484 Use nonimmediate_operand instead of memory_operand for operand 1.
15485 (float-extend standard_x87sse_constant_load splitter): Ditto.
15487 2017-02-21 Jeff Law <law@redhat.com>
15489 PR tree-optimization/79621
15490 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
15491 blocks with edges to themselves.
15493 2017-02-21 Jakub Jelinek <jakub@redhat.com>
15496 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
15497 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
15498 Use gimple_call_builtin_p.
15501 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
15502 on temporarily removed DEBUG_INSNs.
15504 PR tree-optimization/79649
15505 * tree-loop-distribution.c (classify_partition): Give up on
15506 non-generic address space loads/stores.
15508 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
15510 * doc/loop.texi (Loop manipulation): Remove nonexistent
15511 tree_ssa_loop_version from the documentation.
15512 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
15514 2017-02-21 Jakub Jelinek <jakub@redhat.com>
15517 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
15518 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
15519 * config/rs6000/rs6000.c: Include except.h.
15520 (rs6000_expand_split_stack_prologue): Call
15521 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
15523 2017-02-21 Martin Jambor <mjambor@suse.cz>
15526 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
15527 have been analyzed.
15529 2017-02-21 Martin Jambor <mjambor@suse.cz>
15531 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
15532 for backward compatibility only.
15533 * doc/invoke.texi (Option Summary): Remove all references to
15534 -fipa-cp-alignment.
15536 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
15540 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
15542 * lra-constraints.c (curr_insn_transform): Handle
15543 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
15545 2017-02-21 Martin Liska <mliska@suse.cz>
15547 * config/i386/i386.opt: Replace -masm-dialect with -masm.
15549 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
15551 PR translation/79638
15552 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
15554 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
15557 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
15558 (arm_function_ok_for_sibcall): Return false for an indirect call by
15559 descriptor if all the argument registers are used.
15560 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
15561 alignment of the function.
15563 2017-02-21 Jakub Jelinek <jakub@redhat.com>
15565 PR tree-optimization/61441
15566 * simplify-rtx.c (simplify_const_unary_operation): For
15567 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
15568 the sNaN unmodified.
15570 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
15572 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
15573 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
15574 instead of SYSTEM_HEADER_DIR.
15576 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
15577 Martin Liška <mliska@suse.cz>
15579 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
15580 Fix typos and grammar, use active voice, and clarify.
15582 2017-02-20 Marek Polacek <polacek@redhat.com>
15584 PR middle-end/79537
15585 * gimplify.c (gimplify_expr): Handle unused *&&L;.
15588 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
15590 2017-02-20 Jakub Jelinek <jakub@redhat.com>
15593 * config/i386/i386.c (ix86_expand_builtin): Handle
15594 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
15595 ix86_builtins_isa[fcode].isa as a requirement of those
15596 flags and any other flag in the bitmask.
15597 (ix86_init_mmx_sse_builtins): Use 0 instead of
15598 ~OPTION_MASK_ISA_64BIT as mask.
15599 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
15600 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
15601 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
15602 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
15604 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
15607 * lra-constraints.c (split_reg): Check requested split mode
15608 is supported by the register.
15610 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
15612 * lra-constraints.c (simplify_operand_subreg): Remove early
15615 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
15618 * lra-constraints.c (curr_insn_transform): Tighten condition
15619 for converting SUBREG reloads from OP_OUT to OP_INOUT.
15621 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
15624 * lra-constraints.c (curr_insn_transform): Handle
15625 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
15627 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
15630 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
15632 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
15634 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
15637 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
15640 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15642 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
15645 2017-02-18 Jakub Jelinek <jakub@redhat.com>
15647 * final.c (last_columnnum, override_columnnum): New variables.
15648 (final_start_function): Set last_columnnum, pass it to begin_prologue
15649 hook and pass 0 to dwarf2out_begin_prologue.
15650 (final_scan_insn): Update override_columnnum. Pass last_columnnum
15651 to source_line debug hook.
15652 (notice_source_line): Compute last_columnnum and for debug_column_info
15653 return true on column changes.
15654 * debug.h (struct gcc_debug_hooks): Add column argument to
15655 source_line and begin_prologue hooks.
15656 (debug_nothing_int_charstar_int_bool): Remove prototype.
15657 (debug_nothing_int_int_charstar,
15658 debug_nothing_int_int_charstar_int_bool): New prototypes.
15659 (dwarf2out_begin_prologue): Add column argument.
15660 * debug.c (do_nothing_debug_hooks): Adjust source_line and
15661 begin_prologue hooks.
15662 (debug_nothing_int_charstar_int_bool): Remove.
15663 (debug_nothing_int_int_charstar,
15664 debug_nothing_int_int_charstar_int_bool): New functions.
15665 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
15666 through to dwarf2out_source_line.
15667 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
15668 (dwarf2out_source_line): Add column argument, emit it if requested.
15669 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
15671 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
15672 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
15673 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
15674 through to dwarf2out_begin_prologue.
15675 (vmsdbgout_source_line): Add column argument, pass it through to
15676 dwarf2out_source_line.
15677 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
15678 dbxout_source_line caller.
15679 (dbxout_source_line): Add column argument.
15681 * common.opt (gno-column-info, gcolumn-info): New options.
15682 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
15683 (check_die): Also test for multiple DW_AT_decl_column attributes.
15684 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
15685 DW_AT_decl_column if requested.
15686 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
15688 (gen_variable_die): Likewise.
15689 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
15690 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
15693 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
15694 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
15695 (ix86_handle_option): Handle OPT_m3dnowa.
15696 * doc/invoke.texi (-m3dnowa): Document.
15697 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
15698 -m3dnowa instead of -m3dnow -march=athlon.
15701 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
15702 instead of gcc_assert for K, r and R code checks. Formatting fixes.
15704 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15707 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
15708 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
15709 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
15710 generator for vsx_xxpermdi_<mode>_be.
15711 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
15712 force big-endian semantics.
15713 (vsx_xxpermdi_<mode>_be): New define_expand with same
15714 implementation as previous version of vsx_xxpermdi_<mode>.
15716 2017-02-17 Jakub Jelinek <jakub@redhat.com>
15718 PR tree-optimization/79327
15719 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
15720 variable, its initialization and use.
15722 2017-02-17 Julia Koval <julia.koval@intel.com>
15724 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
15725 (OPTION_MASK_ISA_PKU_UNSET): New.
15726 (ix86_handle_option): Handle -mrdpid.
15727 * config/i386/cpuid.h (bit_RDPID): New.
15728 * config/i386/driver-i386.c (host_detect_local_cpu):
15729 Detect RDPID feature.
15730 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
15731 * config/i386/i386-c.c (ix86_target_macros_internal):
15733 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
15734 (ix86_valid_target_attribute_inner_p): Add "rdpid".
15735 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
15736 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
15737 * config/i386/i386.md (define_insn "rdpid"): New.
15738 * config/i386/i386.opt Add -mrdpid.
15739 * config/i386/immintrin.h (_rdpid_u32): New.
15741 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
15743 PR rtl-optimization/79541
15744 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
15745 instead of transforming it into USE.
15747 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
15749 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
15750 If HONOR_SNANS (SFmode) force the input to a register.
15751 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
15752 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
15753 an frsp or similar insn.
15755 2017-02-17 Martin Liska <mliska@suse.cz>
15757 PR rtl-optimization/79577
15758 * params.def (selsched-max-sched-times): Increase minimum to 1.
15760 2017-02-17 Martin Liska <mliska@suse.cz>
15762 PR rtl-optimization/79574
15763 * gcse.c (want_to_gcse_p): Prevent integer overflow.
15765 2017-02-17 Martin Liska <mliska@suse.cz>
15767 PR tree-optimization/79529
15768 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
15769 ssa_defined_default_def_p to handle cases which are implicitly
15771 * tree-ssa.c (ssa_defined_default_def_p): New function.
15772 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
15773 which are implicitly defined.
15774 * tree-ssa.h (ssa_defined_default_def_p): Declare.
15776 2017-02-17 Richard Biener <rguenther@suse.de>
15778 PR middle-end/79576
15779 * params.def (max-ssa-name-query-depth): Limit to 10.
15781 2017-02-17 Richard Biener <rguenther@suse.de>
15783 PR tree-optimization/79552
15784 * tree-ssa-structalias.c (visit_loadstore): Properly verify
15787 2017-02-17 Richard Biener <rguenther@suse.de>
15790 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
15792 2017-02-17 Marek Polacek <polacek@redhat.com>
15794 PR middle-end/79536
15795 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
15796 (fold_negate_expr): New wrapper.
15798 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
15800 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
15801 Correct terminology and de-emphasize pre-standard behavior.
15803 2017-02-16 Alan Modra <amodra@gmail.com>
15805 PR rtl-optimization/79286
15806 * ira.c (def_dominates_uses): New function.
15807 (update_equiv_regs): Don't create an equivalence for insns that
15808 may trap where the register def does not dominate the use.
15810 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
15812 PR rtl-optimization/78127
15813 * lra.c (lra): Call lra_eliminate before finish the loop after
15816 2017-02-16 Richard Biener <rguenther@suse.de>
15818 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
15820 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
15821 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
15822 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
15823 (isl_val_int_from_wi): New function.
15824 (extract_affine_gmp): Rename to ...
15825 (extract_affine_wi): ... this, take a widest_int.
15826 (extract_affine_int): Just wrap extract_affine_wi.
15827 (add_param_constraints): Use isl_val_int_from_wi.
15828 (add_loop_constraints): Likewise, and extract_affine_wi.
15830 2017-02-15 Jeff Law <law@redhat.com>
15832 PR middle-end/79521
15833 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
15834 ira_init_register_move_cost_if_necessary.
15836 2017-02-15 Martin Sebor <msebor@redhat.com>
15838 PR middle-end/32003
15839 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
15840 removed in a prior commit.
15842 2017-02-15 Bin Cheng <bin.cheng@arm.com>
15844 PR tree-optimization/79347
15845 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
15846 counters during peeling.
15848 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
15850 * Makefile.in (site.exp): Remove "set ISLVER".
15852 2017-02-15 Jakub Jelinek <jakub@redhat.com>
15855 * real.c (real_from_integer): Call real_convert even for decimal.
15857 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
15860 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
15862 2017-02-14 Andrew Pinski <apinski@cavium.com>
15864 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
15865 cores and change the partno/implementer to be correct.
15866 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
15867 the 'B" as the implementer.
15868 * config/aarch64/aarch64-tune.md: Regenerate.
15870 2017-02-14 Carl Love <cel@us.ibm.com>
15872 * config/rs6000/rs6000.c: Add case statement entry to make the
15873 xvcvuxdsp built-in argument unsigned.
15874 * config/rs6000/vsx.md: Fix the source and return operand types so they
15875 match the instruction definitions from the ISA document. Fix typo
15876 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
15879 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
15882 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
15883 member early_clobber_alts.
15884 * lra-lives.c (reg_early_clobber_p): New.
15885 (process_bb_lives): Use it.
15886 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
15887 (debug_operand_data): Initialize early_clobber_alts.
15888 (setup_operand_alternative): Set up early_clobber_alts.
15889 (collect_non_operand_hard_regs): Ditto. Pass early clobber
15890 alternatives to new_insn_reg.
15891 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
15893 (lra_update_insn_regno_info): Pass the new arg.
15895 2017-02-14 Jakub Jelinek <jakub@redhat.com>
15897 PR middle-end/79505
15898 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
15899 (new_oacc_loop_raw): Don't clear already cleared fields.
15902 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
15903 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
15904 _mm512_prefetch_i64gather_ps): New inline functions and macros.
15906 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
15909 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
15911 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
15914 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
15915 the extra instruction to the right place to store 128-bit constant
15918 2017-02-14 Martin Sebor <msebor@redhat.com>
15920 PR middle-end/79448
15921 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
15922 warning for strings of unknown length.
15924 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
15926 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
15928 2017-02-14 Jeff Law <law@redhat.com>
15931 * ira-costs.c (scan_one_insn): Initialize register move costs
15932 for pseudos seen in USE/CLOBBER insns.
15934 PR tree-optimization/79095
15935 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
15936 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
15937 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
15938 if the operands are known to be not equal, then the resulting range
15940 (intersect_ranges): If the new range is ~[0,0] and the old range is
15941 wide, then prefer ~[0,0].
15942 * tree-vrp.c (overflow_comparison_p_1): New function.
15943 (overflow_comparison_p): New function.
15944 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
15945 if NAME is used in an overflow test.
15946 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
15947 overflow check that can be expressed as an equality test, then adjust
15948 ops to be that equality test.
15950 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15952 * config/s390/s390-builtin-types.def: Remove flags argument.
15953 * config/s390/s390.c (s390_init_builtins): Likewise.
15955 2017-02-14 Martin Liska <mliska@suse.cz>
15957 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
15958 vector. Fix trailing white spaces.
15960 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
15962 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
15965 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15967 PR rtl-optimization/68664
15968 * config/arm/arm.c (arm_sched_can_speculate_insn):
15969 New function. Declare prototype.
15970 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
15972 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15974 PR rtl-optimization/68664
15975 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
15977 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
15979 2017-02-14 Amit Pawar <amit.pawar@amd.com>
15981 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
15982 max skip bytes for function, loop and jump.
15984 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15986 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
15987 ABS_EXPR for gimple dump.
15989 2017-02-14 Jakub Jelinek <jakub@redhat.com>
15992 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
15994 PR tree-optimization/79408
15995 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
15996 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
15997 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
15998 also if rhs1 is INTEGER_CST.
16000 2017-02-14 Richard Biener <rguenther@suse.de>
16002 PR middle-end/79432
16003 * tree-into-ssa.c (insert_phi_nodes): When the function can
16004 have abnormal edges rewrite SSA names with broken use-def
16005 dominance out of SSA and register them for PHI insertion.
16007 2017-02-13 Martin Sebor <msebor@redhat.com>
16009 PR middle-end/79496
16010 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
16011 clearing info.nowrite flag when snprintf size argument is a range.
16013 2017-02-13 Jakub Jelinek <jakub@redhat.com>
16015 * cprop.c (cprop_jump): Add missing space in string literal.
16016 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
16017 (get_constraint_for_component_ref): Likewise.
16018 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
16019 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
16020 * lra-constraints.c (process_alt_operands): Likewise.
16021 * ipa-inline.c (inline_small_functions): Likewise.
16022 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
16023 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
16024 * trans-mem.c (diagnose_tm_1_op): Likewise.
16025 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
16026 (grid_parallel_clauses_gridifiable): Likewise.
16028 * config/nvptx/mkoffload.c (process): Add space in between
16031 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
16032 "MOD4_SSE_REGS" and "ALL_REGS".
16034 * spellcheck.c (test_data): Add , in between "foo" and "food".
16036 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16039 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
16040 boundary crossing check and subsequent code generation agree.
16042 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16044 * config/aarch64/aarch64.c (has_memory_op): Delete.
16045 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
16048 2017-02-13 Jakub Jelinek <jakub@redhat.com>
16050 PR rtl-optimization/79388
16051 PR rtl-optimization/79450
16052 * combine.c (distribute_notes): When removing TEM_INSN for which
16053 corresponding dest has last value recorded, invalidate that last
16056 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16058 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
16059 of explicit '@'. Add missing assembly comment marker on branch costs
16062 2017-02-13 Nathan Sidwell <nathan@acm.org>
16064 * gengtype-lex.l (<in_struct>): Add '/'.
16066 2017-02-13 Martin Liska <mliska@suse.cz>
16069 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
16071 2017-02-13 Richard Biener <rguenther@suse.de>
16073 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
16075 * configure: Re-generate.
16076 * config.in: Likewise.
16077 * graphite-dependences.c: Simplify as if
16078 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
16079 * graphite-isl-ast-to-gimple.c: Likewise.
16080 * graphite-optimize-isl.c: Likewise.
16081 * graphite-poly.c: Likewise.
16082 * graphite-sese-to-poly.c: Likewise.
16083 * graphite.h: Likewise.
16084 * toplev.c: Include isl/version.h and use isl_version () for
16085 printing the ISL version.
16086 * doc/install.texi: Update ISL requirement.
16088 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
16090 * doc/standards.texi (Standards): Update reference to
16093 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
16095 * doc/extend.texi (Named Address Spaces): sourceware.org now
16097 * doc/install.texi (Binaries): Ditto.
16100 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
16102 * doc/cpp.texi: Replace "stringify"/"stringification" with C
16103 standard terminology "stringize"/"stringizing" throughout.
16104 * doc/cppinternals.texi: Likewise.
16106 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
16108 * doc/extend.texi: Fix some spelling mistakes and typos.
16109 * doc/invoke.texi: Likewise.
16111 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
16114 * params.def (inline-min-speedup) Change from 10 to 8.
16116 2017-02-11 Jakub Jelinek <jakub@redhat.com>
16118 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
16121 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
16124 * ipa-inline-analysis.c (get_minimal_bb): New function.
16125 (record_modified): Use it.
16126 (remap_edge_change_prob): Handle also ancestor functions.
16128 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
16130 * doc/contrib.texi (Contributors): Remove broken link into
16131 the Mauve CVS repository.
16133 2017-02-11 Jakub Jelinek <jakub@redhat.com>
16135 PR middle-end/79454
16136 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
16137 result computation whenever lhs doesn't have vector mode, not
16138 just when it has BLKmode.
16140 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
16142 * doc/makefile.texi (profiledbootstrap): Refer to the
16143 installation instructions only in textual form.
16145 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16148 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
16150 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
16152 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
16153 (Specific): Update mingw-w64 reference.
16155 (Specific): Remove broken link to Renesas RX processor.
16157 2017-02-10 Richard Biener <rguenther@suse.de>
16159 * toplev.c (process_options): Do not mention obsolete graphite
16160 options when printing sorry message about missing graphite support.
16161 Mention -floop-nest-optimize.
16163 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
16165 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
16166 (vtst_p16): Likewise.
16167 (vtstq_p8): Likewise.
16168 (vtstq_p16): Likewise.
16170 (vtstq_p64): Likewise.
16171 * config/arm/arm_neon.h (vgetq_lane_p64): New.
16172 (vset_lane_p64): New.
16173 (vsetq_lane_p64): New.
16175 2017-02-10 Jakub Jelinek <jakub@redhat.com>
16177 PR tree-optimization/79411
16178 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
16179 stmt operands are SSA_NAMEs used in abnormal phis.
16180 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
16183 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
16186 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
16189 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
16191 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
16193 2017-02-09 Jakub Jelinek <jakub@redhat.com>
16195 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
16199 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
16200 not arbitrary TREE_CONSTANT.
16203 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
16204 "omp declare target link" attribute unless is_global_var.
16205 * omp-offload.c (find_link_var_op): Likewise.
16207 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
16208 Chung-Lin Tang <cltang@codesourcery.com>
16210 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
16212 (gimplify_adjust_omp_clauses): Don't delete TILE.
16213 (gimplify_omp_for): Deal with TILE.
16214 * internal-fn.c (expand_GOACC_TILE): New function.
16215 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
16217 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
16218 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
16220 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
16221 avoid DIV for outermost collapse var.
16222 (expand_oacc_for): Insert tile element loop as needed. Adjust.
16223 Remove out of date comments, fix whitespace.
16224 * omp-general.c (omp_extract_for_data): Deal with tiling.
16225 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
16226 adjust OLF_DIM_BASE value.
16227 (struct omp_for_data): Add tiling field.
16228 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
16229 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
16230 for auto loops. Remove default auto determining, moved to
16231 oacc_loop_fixed_partitions.
16232 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
16233 stmts, add e_mask field.
16234 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
16235 (oacc_thread_numbers): Use oacc_dim_call.
16236 (oacc_xform_tile): New.
16237 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
16238 (finish_oacc_loop): Adjust for ifns vector.
16239 (oacc_loop_discover_walk): Append loop abstraction sites to list,
16240 add case for GOACC_TILE fns.
16241 (oacc_loop_xform_loop): Delete.
16242 (oacc_loop_process): Iterate over call list directly, and add
16243 handling for GOACC_TILE fns.
16244 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
16246 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
16247 vector partitioning to outer loops. Assign 2 partitions to loops
16248 when available. Add TILE handling.
16249 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
16250 (execite_oacc_device_lower): Process GOACC_TILE fns,
16251 ignore unknown specs.
16252 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
16253 * tree.c (omp_clause_num_ops): Adjust TILE ops.
16254 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
16256 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
16258 * configure.ac (ACX_BUGURL): Update.
16259 * configure: Regenerate.
16261 2017-02-09 Richard Biener <rguenther@suse.de>
16263 PR tree-optimization/69823
16264 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
16265 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
16267 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
16269 * config/arc/arc-c.def: Add __NPS400__ definition.
16270 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
16271 (TARGET_NPS400): Define.
16273 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
16275 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
16277 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
16278 pointer, arch_info.
16279 (arc_cpu_types): Fill the arch_info field with a pointer into the
16280 arc_arch_types table.
16281 (arc_selected_cpu): Declare.
16282 * config/arc/arc.c (arc_selected_cpu): Make global.
16283 (arc_selected_arch): Delete.
16284 (arc_base_cpu): Delete.
16285 (arc_override_options): Remove references to deleted variables,
16286 update access to arch information.
16287 (ARC_OPT): Update access to arch information.
16288 (ARC_OPTX): Likewise.
16289 * config/arc/arc.h (arc_base_cpu): Remove declaration.
16290 (TARGET_ARC600): Update access to arch information.
16291 (TARGET_ARC601): Likewise.
16292 (TARGET_ARC700): Likewise.
16293 (TARGET_EM): Likewise.
16294 (TARGET_HS): Likewise.
16295 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
16298 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
16301 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
16302 condition/operands for integer GE/LE/GEU/LEU operations.
16304 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
16306 PR translation/79397
16307 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
16310 2017-02-08 Martin Jambor <mjambor@suse.cz>
16313 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
16314 whether allocation happened.
16315 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
16316 nothing was allocated.
16318 2017-02-08 Jakub Jelinek <jakub@redhat.com>
16320 PR tree-optimization/79408
16321 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
16322 constant, but SSA_NAME with a known integer range, use the minimum
16323 of that range instead of op1 to determine if modulo can be replaced
16324 with its first operand.
16326 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16328 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
16330 2017-02-08 Richard Biener <rguenther@suse.de>
16332 PR tree-optimization/71824
16333 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
16334 Check all loops contained in the merged region.
16336 2017-02-07 Andrew Pinski <apinski@cavium.com>
16338 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
16340 2017-02-07 Andrew Pinski <apinski@cavium.com>
16342 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
16343 (thunderxt88): Likewise.
16344 (thunderxt81): Disable LSE and change v8.1 to v8.
16345 (thunderxt83): Likewise.
16347 2017-02-07 Jakub Jelinek <jakub@redhat.com>
16348 Richard Biener <rguenther@suse.de>
16350 PR middle-end/79399
16351 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
16352 type from int to size_t.
16353 * ira-costs.c (struct_costs_size): Change type from int to size_t.
16355 2017-02-07 Jakub Jelinek <jakub@redhat.com>
16357 PR rtl-optimization/79386
16358 * cprop.c (bypass_conditional_jumps): Initialize
16359 bypass_last_basic_block already before splitting bbs after
16360 unconditional traps...
16361 (bypass_conditional_jumps): ... rather than here.
16364 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
16365 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
16366 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
16367 fix -masm=intel patterns.
16369 2017-02-07 Richard Biener <rguenther@suse.de>
16371 PR tree-optimization/79256
16372 PR middle-end/79278
16373 * builtins.c (get_object_alignment_2): Use min_align_of_type
16374 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
16375 and ADJUST_FIELD_ALIGN.
16377 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
16379 * doc/tm.texi: Regenerate.
16380 * stor-layout.c (layout_decl): Adjust.
16381 (update_alignment_for_field): Likewise.
16382 (place_field): Likewise.
16383 (min_align_of_type): Likewise.
16384 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
16385 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
16386 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
16387 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
16388 * config/frv/frv.c (frv_adjust_field_align): Likewise.
16389 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
16390 * config/i386/i386.c (x86_field_alignment): Likewise.
16391 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
16392 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
16393 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
16394 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
16395 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
16396 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
16400 2017-01-30 Richard Biener <rguenther@suse.de>
16402 PR tree-optimization/79256
16403 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
16404 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
16407 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
16409 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
16410 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
16411 builtins to SImode and emit a zero-extend, if necessary.
16413 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
16415 * docs/invoke.texi (RISC-V Options): Alphabetize.
16417 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
16419 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
16422 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
16424 * config/riscv/riscv.c: New file.
16425 * common/config/riscv/riscv-common.c: Likewise.
16426 * config.gcc: Likewise.
16427 * config/riscv/constraints.md: Likewise.
16428 * config/riscv/elf.h: Likewise.
16429 * config/riscv/generic.md: Likewise.
16430 * config/riscv/linux.h: Likewise.
16431 * config/riscv/multilib-generator: Likewise.
16432 * config/riscv/peephole.md: Likewise.
16433 * config/riscv/pic.md: Likewise.
16434 * config/riscv/predicates.md: Likewise.
16435 * config/riscv/riscv-builtins.c: Likewise.
16436 * config/riscv/riscv-c.c: Likewise.
16437 * config/riscv/riscv-ftypes.def: Likewise.
16438 * config/riscv/riscv-modes.def: Likewise.
16439 * config/riscv/riscv-opts.h: Likewise.
16440 * config/riscv/riscv-protos.h: Likewise.
16441 * config/riscv/riscv.h: Likewise.
16442 * config/riscv/riscv.md: Likewise.
16443 * config/riscv/riscv.opt: Likewise.
16444 * config/riscv/sync.md: Likewise.
16445 * config/riscv/t-elf-multilib: Likewise.
16446 * config/riscv/t-linux: Likewise.
16447 * config/riscv/t-linux-multilib: Likewise.
16448 * config/riscv/t-riscv: Likewise.
16449 * configure.ac: Likewise.
16450 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
16451 Waterman as RISC-V maintainers.
16452 * doc/install.texi: Add RISC-V entries.
16453 * doc/invoke.texi: Add RISC-V options section.
16454 * doc/md.texi: Add RISC-V constraints section.
16455 * configure: Regenerated.
16457 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
16460 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
16461 false values to be constant vectors with all 0 or all 1 bits set.
16462 (vcondu<mode><mode>): Likewise.
16463 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
16465 (fpmask_comparison_operator): Update comment.
16466 (vecint_comparison_operator): New predicate.
16467 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
16468 vector conditionals when the true and false values are constant
16469 vectors with all 0 bits or all 1 bits set.
16471 2017-02-06 Martin Sebor <msebor@redhat.com>
16473 PR tree-optimization/79376
16474 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
16476 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
16478 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
16479 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
16480 to simplify split condition.
16482 2017-02-06 Jakub Jelinek <jakub@redhat.com>
16484 * omp-expand.c (oxpand_omp_atomic_fetch_op,
16485 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
16488 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
16490 PR rtl-optimization/68664
16491 * target.def (can_speculate_insn): New hook.
16492 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
16493 * doc/tm.texi: Regenerate.
16494 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
16495 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
16496 (rs6000_sched_can_speculate_insn): New function.
16498 2017-02-06 Jakub Jelinek <jakub@redhat.com>
16500 PR tree-optimization/79284
16501 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
16502 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
16503 vectorizable_mask_load_store, vectorizable_operation,
16504 vect_is_simple_cond, get_same_sized_vectype): Use it instead
16505 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
16506 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
16507 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
16508 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
16509 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
16510 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
16511 is_gimple_assign (stmt). Replace another such test with
16512 is_gimple_assign (stmt).
16514 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
16517 * config/avr/avr.c (rtl-iter.h): Include it.
16518 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
16519 (avr_legitimate_combined_insn): ...and implementation.
16521 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
16523 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
16524 * config/s390/s390.c (s390_const_operand_ok)
16525 (s390_canonicalize_comparison, s390_extract_part)
16526 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
16527 (s390_contiguous_bitmask_p, s390_rtx_costs)
16528 (legitimize_pic_address): Likewise.
16529 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
16530 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
16531 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
16532 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
16533 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
16535 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
16537 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
16538 REGNO($0) == REGNO($1).
16540 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16542 * config/s390/linux.h(SIZE_TYPE): Add comment.
16544 2017-02-06 Julian Brown <julian@codesourcery.com>
16545 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16546 Virendra Pathak <virendra.pathak@broadcom.com>
16548 * config/aarch64/aarch64-cores.def: Change the scheduler
16550 * config/aarch64/aarch64.md: Include thunderx2t99.md.
16551 * config/aarch64/thunderx2t99.md: New file.
16553 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
16555 * doc/standards.texi (Go Language): Update link to language
16558 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
16560 * tree-eh.c (lower_resx): Sanitize profile.
16561 (cleanup_empty_eh_move_lp): Likewise.
16563 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
16566 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
16568 * cfgloopmanip.h (loop_version): Update prototype.
16569 * modulo-sched.c (sms_schedule): Update call of loop_version.
16570 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
16571 * tree-parloops.c (gen_parallel_loop): Likewise.
16572 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
16573 * tree-ssa-loop-split.c (split_loop): Likewise.
16574 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
16575 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
16577 2017-02-05 Martin Liska <mliska@suse.cz>
16580 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
16582 (print_operand_address): Initialize a struct to zero.
16584 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
16586 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
16587 garbage collector only in textual form.
16589 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
16591 * doc/extend.texi (x86 specific memory model extensions for
16592 transactional memory): Simplify a phrase.
16594 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
16597 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
16598 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
16599 (atomic_storedi_1): Likewise.
16601 2017-02-04 Jakub Jelinek <jakub@redhat.com>
16603 PR tree-optimization/79338
16604 * tree-parloops.c (gather_scalar_reductions): Don't call
16605 vect_analyze_loop_form for loop->inner before destroying loop's
16608 2017-02-03 Martin Sebor <msebor@redhat.com>
16610 PR tree-optimization/79327
16611 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
16612 when precision has resulted in leading zeros.
16613 (format_integer): Adjust the likely counter to assume an unknown
16614 argument that may be zero is non-zero.
16616 2017-02-03 Jason Merrill <jason@redhat.com>
16619 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
16620 avoid copying non-taken branch.
16622 2017-02-03 Jakub Jelinek <jakub@redhat.com>
16624 PR tree-optimization/79340
16625 * tree-vect-loop.c (vectorizable_reduction): Release
16626 vec_defs elements after safe_splicing them into other vectors.
16629 PR tree-optimization/79327
16630 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
16631 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
16633 (format_integer): Use wide_int_to_tree instead of build_int_cst
16634 + to_?hwi. If argmin is NULL, just set argmin and argmax to
16635 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
16636 of shortest and longest sequence.
16638 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
16640 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
16641 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
16643 2017-02-03 Walter Lee <walt@tilera.com>
16646 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
16647 after initial stackframe link reg save.
16648 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
16650 2017-02-03 Jakub Jelinek <jakub@redhat.com>
16653 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
16654 wu for stxssp alternative.
16656 2017-02-03 Martin Sebor <msebor@redhat.com>
16658 PR tree-optimization/79352
16659 * gimple-fold.c (get_range_strlen): Add argument.
16660 (get_range_strlen): Change return type to bool.
16661 (get_maxval_strlen): Pass in a dummy argument.
16662 * gimple-fold.h (get_range_strlen): Change return type to bool.
16663 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
16664 * tree.h (array_at_struct_end_p): Add argument.
16665 * tree.c (array_at_struct_end_p): Handle it.
16667 2017-02-03 Martin Liska <mliska@suse.cz>
16670 * multiple_target.c (create_dispatcher_calls): Redirect edge
16671 from a caller of a dispatcher.
16672 (expand_target_clones): Make the clones local.
16673 (ipa_target_clone): Do both target clones and resolvers.
16674 (ipa_dispatcher_calls): Remove the pass.
16675 (pass_dispatcher_calls::gate): Likewise.
16676 (make_pass_dispatcher_calls): Likewise.
16677 * passes.def (pass_target_clone): Put as very first IPA early
16680 2017-02-03 Martin Liska <mliska@suse.cz>
16682 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
16683 in case of a function with ifunc attribute.
16685 2017-02-03 Martin Liska <mliska@suse.cz>
16687 * cgraph.c (cgraph_node::dump): Dump function version info.
16688 * symtab.c (symtab_node::dump_base): Add missing new line.
16690 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
16692 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
16693 (ifcombine_ifandif): Use it.
16695 2017-02-03 Martin Liska <mliska@suse.cz>
16697 * doc/invoke.texi: Document default value for
16698 use-after-scope-direct-emission-threshold.
16700 2017-02-03 Martin Liska <mliska@suse.cz>
16702 PR tree-optimization/79339
16703 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
16704 (format_floating): Likewise.
16706 2017-02-03 Martin Liska <mliska@suse.cz>
16709 * ipa-prop.c (ipa_node_params_t::insert): Remove current
16711 (ipa_node_params_t::remove): Likewise.
16712 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
16713 initialization from removed ipa_node_params_t::insert.
16714 (ipa_node_params::~ipa_node_params): Move from removed
16715 ipa_node_params_t::release.
16716 * symbol-summary.h (symbol_summary::m_released): New member.
16717 Do not release a summary twice. Do not allow to call finalizer
16718 for types of a summary that live in GGC memory.
16720 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16722 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
16725 2017-02-02 Martin Sebor <msebor@redhat.com>
16727 PR middle-end/79275
16728 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
16729 (format_string): Tighten up the range of output for non-constant
16730 strings and correct the expected range for wide non-constant strings.
16732 2017-02-02 Martin Sebor <msebor@redhat.com>
16734 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
16736 PR middle-end/32003
16737 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
16739 (-fdump-tree-@var): Add to index and document how to come up
16740 with pass-specific option and dump file names.
16741 (-fdump-passes): Clarify where to look for output.
16743 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
16745 PR middle-end/77445
16746 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
16747 statistics of the analyzed path; allow threading for speed when
16748 any of BBs along the path are optimized for speed.
16750 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
16752 PR middle-end/78468
16753 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
16754 settings of the virtual registers.
16757 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
16759 * explow.c (get_dynamic_stack_size): Take known alignment of stack
16760 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
16763 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16765 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
16766 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
16768 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16770 * config/s390/s390.md: Add missing comments with the expanded
16772 * config/s390/vector.md: Likewise.
16773 * config/s390/vx-builtins.md: Likewise.
16775 2017-02-02 Jakub Jelinek <jakub@redhat.com>
16778 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
16779 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
16780 conditions on a single line.
16782 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16784 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
16785 __S390_VX__ to __VX__.
16787 2017-02-01 Andrew Pinski <apinski@cavium.com>
16789 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
16790 stmt_info to record_stmt_cost.
16791 (vect_get_known_peeling_cost): Pass stmt_info if known to
16793 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
16794 cpu_vector_cost field into
16795 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
16796 field into vec_int_stmt_cost and vec_fp_stmt_cost.
16797 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
16798 splitting of scalar_stmt_cost and vec_stmt_cost.
16799 (thunderx_vector_cost): Likewise.
16800 (cortexa57_vector_cost): LIkewise.
16801 (exynosm1_vector_cost): Likewise.
16802 (xgene1_vector_cost): Likewise.
16803 (thunderx2t99_vector_cost): Improve after the splitting of the two
16805 (aarch64_builtin_vectorization_cost): Update for the splitting of
16806 scalar_stmt_cost and vec_stmt_cost.
16808 2017-02-01 Torvald Riegel <triegel@redhat.com>
16809 Richard Henderson <rth@redhat.com>
16811 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
16812 conditional on existance of a fast atomic load.
16813 * optabs-query.c (can_atomic_load_p): New function.
16814 * optabs-query.h (can_atomic_load_p): Declare it.
16815 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
16816 no fast atomic load is available for the particular size of access.
16817 (expand_atomic_compare_and_swap): Likewise.
16818 (expand_atomic_load): Likewise.
16819 (expand_atomic_store): Likewise.
16820 (expand_atomic_fetch_op): Likewise.
16821 * testsuite/lib/target-supports.exp
16822 (check_effective_target_sync_int_128): Remove x86 because it provides
16823 no fast atomic load.
16824 (check_effective_target_sync_int_128_runtime): Likewise.
16826 2017-02-01 Richard Biener <rguenther@suse.de>
16828 * graphite.c: Include tree-vectorizer.h for find_loop_location.
16829 (graphite_transform_loops): Provide opt-info for optimized nests.
16830 * tree-parloop.c (parallelize_loops): Provide opt-info for
16831 parallelized loops.
16833 2017-02-01 Richard Biener <rguenther@suse.de>
16835 PR middle-end/79315
16836 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
16837 was not set before.
16839 2017-02-01 Richard Biener <rguenther@suse.de>
16841 PR tree-optimization/71824
16842 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
16843 Verify the loops are valid in the merged SESE region.
16844 (scop_detection::can_represent_loop_1): Check analyzing the
16845 evolution of the number of iterations in the region succeeds.
16847 2017-01-31 Ian Lance Taylor <iant@golang.org>
16849 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
16850 REG_ARGS_SIZE note to 32-bit push insns and call insn.
16852 2017-01-31 David Malcolm <dmalcolm@redhat.com>
16854 PR preprocessor/79210
16855 * input.c (get_substring_ranges_for_loc): Replace line_width
16856 assertion with error-handling.
16858 2017-01-31 Richard Biener <rguenther@suse.de>
16860 PR tree-optimization/77318
16861 * graphite-sese-to-poly.c (extract_affine): Fix assert.
16862 (create_pw_aff_from_tree): Take loop parameter.
16863 (add_condition_to_pbb): Pass loop of the condition to
16864 create_pw_aff_from_tree.
16866 2017-01-31 Jakub Jelinek <jakub@redhat.com>
16868 * config/s390/s390.c (s390_asan_shadow_offset): New function.
16869 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
16871 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
16875 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
16877 (convert_int_to_float128): Likewise.
16878 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
16879 (convert_int_to_float128): Likewise.
16880 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
16881 (UNSPEC_IEEE128_CONVERT): Likewise.
16882 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
16883 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
16884 Use local variables for IBM extended format.
16885 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
16886 (fix_trunc<mode>si2_fprs): Likewise.
16887 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
16888 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
16889 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
16890 to know that we can now have integers of all sizes in vector
16892 (fix<uns>_<mode>di2_hw): Likewise.
16893 (float<uns>_<mode>si2_hw): Likewise.
16894 (fix_<mode>si2_hw): Likewise.
16895 (fixuns_<mode>si2_hw): Likewise.
16896 (float<uns>_<mode>di2_hw): Likewise.
16897 (float_<mode>di2_hw): Likewise.
16898 (float_<mode>si2_hw): Likewise.
16899 (floatuns_<mode>di2_hw): Likewise.
16900 (floatuns_<mode>si2_hw): Likewise.
16901 (xscvqp<su>wz_<mode>): Delete, no longer used.
16902 (xscvqp<su>dz_<mode>): Likewise.
16903 (xscv<su>dqp_<mode>): Likewise.
16904 (ieee128_mfvsrd_64bit): Likewise.
16905 (ieee128_mfvsrd_32bit): Likewise.
16906 (ieee128_mfvsrwz): Likewise.
16907 (ieee128_mtvsrw): Likewise.
16908 (ieee128_mtvsrd_64bit): Likewise.
16909 (ieee128_mtvsrd_32bit): Likewise.
16911 2017-01-31 Martin Liska <mliska@suse.cz>
16914 * ipa-prop.c (ipa_free_all_node_params): Call release method
16915 instead of ~sumbol_summary to not to trigger double times
16918 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
16920 PR tree-optimization/71691
16921 * bitmap.h (class auto_bitmap): New.
16922 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
16923 is_maybe_undefined instead of ssa_undefined_value_p.
16925 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16927 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
16928 __S390_ARCH_LEVEL__ to __ARCH__.
16930 2017-01-31 Jakub Jelinek <jakub@redhat.com>
16932 PR tree-optimization/79267
16933 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
16934 if should_remove_lhs_p is true.
16936 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
16939 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
16940 (add_alignment_attribute): New.
16941 (base_type_die): Add alignment attribute.
16942 (subrange_type_die): Likewise.
16943 (modified_type_die): Likewise.
16944 (gen_array_type_die): Likewise.
16945 (gen_descr_array_type_die: Likewise.
16946 (gen_enumeration_type_die): Likewise.
16947 (gen_subprogram_die): Likewise.
16948 (gen_variable_die): Likewise.
16949 (gen_field_die): Likewise.
16950 (gen_ptr_to_mbr_type_die): Likewise.
16951 (gen_struct_or_union_type_die): Likewise.
16952 (gen_subroutine_type_die): Likewise.
16953 (gen_typedef_die): Likewise.
16954 (base_type_cmp): Compare alignment attribute.
16956 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16959 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
16960 (setb_unsigned) New pattern for setb with CCUNS.
16961 * config/rs6000/rs6000.c (expand_block_compare): Use a different
16962 subfc./subfe sequence to avoid overflow problems. Generate a
16963 shorter sequence with cmpld/setb for power9.
16964 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
16965 for generating subfc. instruction.
16966 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
16967 now uses this instruction.
16969 2017-01-30 Ian Lance Taylor <iant@google.com>
16972 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
16973 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
16975 2017-01-30 Martin Sebor <msebor@redhat.com>
16977 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
16978 Move constant to the right of a relational operator.
16979 (get_mpfr_format_length, format_character, format_string): Ditto.
16980 (should_warn_p, maybe_warn): Same.
16982 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
16984 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
16987 * asan.c (get_translation_unit_decl): Remove function.
16988 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
16990 2017-01-30 Martin Liska <mliska@suse.cz>
16992 PR gcov-profile/79259
16993 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
16994 -fprofile-generate.
16996 2017-01-30 Martin Liska <mliska@suse.cz>
16999 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
17000 Initialize variables with NULL value.
17002 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
17005 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
17007 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
17009 2017-01-30 Richard Biener <rguenther@suse.de>
17011 PR tree-optimization/79276
17012 * tree-vrp.c (process_assert_insertions): Properly adjust common
17013 when removing a duplicate.
17015 * gcc.dg/torture/pr79276.c: New testcase.
17017 2017-01-30 Richard Biener <rguenther@suse.de>
17019 PR tree-optimization/79256
17020 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
17021 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
17023 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
17025 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
17028 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
17029 ("*r<noxa>sbg_<mode>_sll_bitmask")
17030 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
17031 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
17032 Use contiguous_bitmask_nowrap_operand.
17034 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17037 * config/rs6000/altivec.h (vec_xl): Revise #define.
17038 (vec_xst): Likewise.
17040 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
17042 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
17044 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
17046 PR rtl-optimization/79194
17047 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
17048 traps before call to bypass_conditional_jumps.
17050 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
17052 PR tree-optimization/71374
17053 * lra-constraints.c (check_conflict_input_operands): New.
17054 (match_reload): Use it.
17056 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
17059 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
17060 account to calculate conflict_set.
17062 2017-01-27 Bin Cheng <bin.cheng@arm.com>
17064 PR rtl-optimization/78559
17065 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
17066 other_insn in combine.
17068 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
17070 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
17071 uint16_type_node for BT_UINT16.
17073 2017-01-27 David Malcolm <dmalcolm@redhat.com>
17075 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
17076 "RTL Tests" to menu.
17077 (GIMPLE Tests): New node.
17078 (RTL Tests): New node.
17080 2017-01-27 Richard Biener <rguenther@suse.de>
17082 PR tree-optimization/79245
17083 * tree-loop-distribution.c (distribute_loop): Apply cost
17084 modeling also to detected patterns.
17086 2017-01-27 Richard Biener <rguenther@suse.de>
17088 PR tree-optimization/71433
17089 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
17090 (compare_assert_loc): New function.
17091 (process_assert_insertions): Sort and optimize assert locations
17092 to remove duplicates and push down identical assertions on
17093 edges to their destination block.
17095 2017-01-27 Richard Biener <rguenther@suse.de>
17097 PR tree-optimization/79244
17098 * tree-vrp.c (remove_range_assertions): Forcefully propagate
17099 out SSA names even if abnormal.
17101 2017-01-27 Jakub Jelinek <jakub@redhat.com>
17103 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
17104 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
17105 instead of MPFR_RNDN.
17107 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
17110 * arm.c (arm_option_override): Don't call build_target_option_node
17111 until after doing all option overrides.
17112 (arm_valid_target_attribute_tree): Likewise.
17114 2017-01-27 Martin Liska <mliska@suse.cz>
17116 * doc/invoke.texi (-fprofile-arcs): Document profiling support
17117 for {cd}tors and C++ {cd}tors.
17119 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
17121 * config/s390/s390.md ("*setmem_long_and")
17122 ("*setmem_long_and_31z"): Use zero_extend instead of and.
17124 2017-01-26 Martin Sebor <msebor@redhat.com>
17126 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
17129 2017-01-26 Martin Sebor <msebor@redhat.com>
17131 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
17132 HAVE_DFmode before using XFmode or DFmode.
17133 (parse_directive): Avoid using the z length modifier to avoid
17134 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
17136 PR middle-end/78703
17137 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
17138 to accept adjustment as an array.
17139 (get_int_range): New function.
17140 (struct directive): Make width and prec arrays.
17141 (directive::set_width, directive::set_precision): Call get_int_range.
17142 (format_integer, format_floating): Handle width and precision ranges.
17143 (format_string, parse_directive): Same.
17145 2017-01-26 Jakub Jelinek <jakub@redhat.com>
17148 * dwarf2out.c (generate_skeleton_bottom_up): For children with
17149 comdat_type_p set, just clone them, but keep the children in the
17153 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
17154 which have direct callers with -fvar-tracking-assignments enabled
17156 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
17157 inside of type units.
17159 2017-01-26 Martin Sebor <msebor@redhat.com>
17161 PR middle-end/78703
17162 * gimple-ssa-sprintf.c (struct result_range): Add likely and
17164 (struct format_result): Replace number_chars, number_chars_min,
17165 and number_chars_max with a single member of struct result_range.
17167 (format_result::operator+=): Adjust.
17168 (struct fmtresult): Remove bounded. Handle likely and unlikely
17170 (fmtresult::adjust_for_width_or_precision): New function.
17171 (fmtresult:type_max_digits): New function.
17172 (bytes_remaining): Handle likely and unlikely counters.
17173 (min_bytes_remaining): Remove.
17174 (format_percent): Simplify.
17175 (format_integer, format_floating): Set likely and unlikely counters.
17176 (get_string_length, format_character, format_string): Same.
17177 (format_plain, should_warn_p): New function.
17178 (maybe_warn): Call should_warn_p. Update diagnostic messages
17179 and handle those for all directives, including plain strings.
17180 (format_directive): Handle likely and unlikely counters.
17181 Remove unnecessary quoting from diagnostics. Add an informational
17183 (add_bytes): Remove.
17184 (pass_sprintf_length::compute_format_length): Simplify.
17185 (try_substitute_return_value): Handle likely and unlikely counters.
17187 2017-01-26 Carl Love <cel@us.ibm.com>
17189 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
17190 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
17192 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
17195 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
17196 endianess for subregs into account.
17197 * lra-constraints.c (lra_constraints): Do risky transformations
17198 always on the first iteration.
17199 * lra-lives.c (check_pseudos_live_through_calls): Add arg
17200 last_call_used_reg_set.
17201 (process_bb_lives): Define and use last_call_used_reg_set.
17202 * lra.c (lra): Always continue after lra_constraints on the first
17205 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
17207 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
17209 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
17211 2017-01-26 Jakub Jelinek <jakub@redhat.com>
17213 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
17214 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
17215 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
17216 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
17217 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
17218 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
17219 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
17220 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
17221 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
17223 2017-01-26 Marek Polacek <polacek@redhat.com>
17226 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
17227 for the third operand.
17229 2017-01-26 Jakub Jelinek <jakub@redhat.com>
17231 PR middle-end/79236
17232 * omp-low.c (struct omp_context): Add simt_stmt field.
17233 (scan_omp_for): Return omp_context *.
17234 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
17235 context to the _simt_ SIMD stmt.
17236 (lower_omp_for): For combined SIMD with sibling _simt_
17237 SIMD, make sure to use the same decls in _looptemp_
17238 clauses as in the sibling.
17240 2017-01-26 David Sherwood <david.sherwood@arm.com>
17242 PR middle-end/79212
17243 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
17246 2017-01-26 Jakub Jelinek <jakub@redhat.com>
17249 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
17250 emit fld b; fld a; if possible.
17252 * brig-builtins.def: Update copyright years.
17253 * config/arm/arm_acle_builtins.def: Update copyright years.
17255 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
17258 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
17259 constraint instead of o for the stxsd instruction.
17261 2017-01-25 Carl Love <cel@us.ibm.com>
17263 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
17264 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
17266 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
17268 * doc/invoke.texi (C++ Dialect Options): Fix typo.
17270 2017-01-25 Richard Biener <rguenther@suse.de>
17272 PR tree-optimization/69264
17273 * target.def (vector_alignment_reachable): Improve documentation.
17274 * doc/tm.texi: Regenerate.
17275 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
17277 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
17278 earlier changes with respect to TYPE_USER_ALIGN.
17279 (vector_alignment_reachable_p): Likewise. Improve dumping.
17281 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17284 * config/arm/arm.md (xordi3): Force constant operand into a register
17287 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17289 * doc/invoke.texi (-fstore-merging): Correct default optimization
17290 levels at which it is enabled.
17291 (-O): Move -fstore-merging from list to...
17294 2017-01-25 Richard Biener <rguenther@suse.de>
17297 * omp-expand.c: Include debug.h.
17298 (expand_omp_taskreg): Make sure to generate early debug before
17299 outlining anything from a function.
17300 (expand_omp_target): Likewise.
17301 (grid_expand_target_grid_body): Likewise.
17303 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
17306 * asan.c (get_translation_unit_decl): New function.
17307 (asan_add_global): Extract modules file name from globals
17308 TRANSLATION_UNIT_DECL name.
17310 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
17313 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
17314 for long calls with APCS frame and VFP.
17316 2017-01-24 David Malcolm <dmalcolm@redhat.com>
17318 * cfg.c (original_copy_tables_initialized_p): New function.
17319 * cfg.h (original_copy_tables_initialized_p): New decl.
17320 * cfgrtl.c (relink_block_chain): Guard the call to
17321 free_original_copy_tables with a call to
17322 original_copy_tables_initialized_p.
17323 * cgraph.h (symtab_node::native_rtl_p): New decl.
17324 * cgraphunit.c (symtab_node::native_rtl_p): New function.
17325 (symtab_node::needed_p): Don't assert for early assembly output
17326 for __RTL functions.
17327 (cgraph_node::finalize_function): Set "force_output" for __RTL
17329 (cgraph_node::analyze): Bail out early for __RTL functions.
17330 (analyze_functions): Update assertion to support __RTL functions.
17331 (cgraph_node::expand): Bail out early for __RTL functions.
17332 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
17334 * function.h (struct function): Update comment for field
17336 * gimple-expr.c: Include "tree-pass.h".
17337 (gimple_has_body_p): Return false for __RTL functions.
17338 * Makefile.in (OBJS): Add run-rtl-passes.o.
17339 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
17341 (gcc::pass_manager::get_clean_slate): New accessor.
17342 * passes.c: Include "insn-addr.h".
17343 (should_skip_pass_p): Add logging. Update logic for running
17344 "expand" to be compatible with both __GIMPLE and __RTL. Guard
17345 property-provider override so it is only done for gimple passes.
17347 (skip_pass): New function.
17348 (execute_one_pass): Call skip_pass when skipping passes.
17349 * read-md.c (md_reader::read_char): Support filtering
17350 the input to a subset of line numbers.
17351 (md_reader::md_reader): Initialize fields
17352 m_first_line and m_last_line.
17353 (md_reader::read_file_fragment): New function.
17354 * read-md.h (md_reader::read_file_fragment): New decl.
17355 (md_reader::m_first_line): New field.
17356 (md_reader::m_last_line): New field.
17357 * read-rtl-function.c (function_reader::create_function): Only
17358 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
17359 curr_properties. Set DECL_INITIAL to a dummy block.
17360 (read_rtl_function_body_from_file_range): New function.
17361 * read-rtl-function.h (read_rtl_function_body_from_file_range):
17363 * run-rtl-passes.c: New file.
17364 * run-rtl-passes.h: New file.
17366 2017-01-24 Jeff Law <law@redhat.com>
17368 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
17371 2017-01-24 Bin Cheng <bin.cheng@arm.com>
17373 PR tree-optimization/79159
17374 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
17375 (record_nonwrapping_iv): Improve boundary using above function if no
17376 value range information.
17378 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
17379 Martin Jambor <mjambor@suse.cz>
17381 * brig-builtins.def: New file.
17382 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
17383 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
17384 (DEF_HSAIL_SAT_BUILTIN): Likewise.
17385 (DEF_HSAIL_INTR_BUILTIN): Likewise.
17386 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
17387 * builtin-types.def (BT_INT8): New.
17388 (BT_INT16): Likewise.
17389 (BT_UINT8): Likewise.
17390 (BT_UINT16): Likewise.
17391 (BT_FN_ULONG): Likewise.
17392 (BT_FN_UINT_INT): Likewise.
17393 (BT_FN_UINT_ULONG): Likewise.
17394 (BT_FN_UINT_LONG): Likewise.
17395 (BT_FN_UINT_PTR): Likewise.
17396 (BT_FN_ULONG_PTR): Likewise.
17397 (BT_FN_INT8_FLOAT): Likewise.
17398 (BT_FN_INT16_FLOAT): Likewise.
17399 (BT_FN_UINT32_FLOAT): Likewise.
17400 (BT_FN_UINT16_FLOAT): Likewise.
17401 (BT_FN_UINT8_FLOAT): Likewise.
17402 (BT_FN_UINT64_FLOAT): Likewise.
17403 (BT_FN_UINT16_UINT32): Likewise.
17404 (BT_FN_UINT32_UINT16): Likewise.
17405 (BT_FN_UINT16_UINT16_UINT16): Likewise.
17406 (BT_FN_INT_PTR_INT): Likewise.
17407 (BT_FN_UINT_PTR_UINT): Likewise.
17408 (BT_FN_LONG_PTR_LONG): Likewise.
17409 (BT_FN_ULONG_PTR_ULONG): Likewise.
17410 (BT_FN_VOID_UINT64_UINT64): Likewise.
17411 (BT_FN_UINT8_UINT8_UINT8): Likewise.
17412 (BT_FN_INT8_INT8_INT8): Likewise.
17413 (BT_FN_INT16_INT16_INT16): Likewise.
17414 (BT_FN_INT_INT_INT): Likewise.
17415 (BT_FN_UINT_FLOAT_UINT): Likewise.
17416 (BT_FN_FLOAT_UINT_UINT): Likewise.
17417 (BT_FN_ULONG_UINT_UINT): Likewise.
17418 (BT_FN_ULONG_UINT_PTR): Likewise.
17419 (BT_FN_ULONG_ULONG_ULONG): Likewise.
17420 (BT_FN_UINT_UINT_UINT): Likewise.
17421 (BT_FN_VOID_UINT_PTR): Likewise.
17422 (BT_FN_UINT_UINT_PTR: Likewise.
17423 (BT_FN_UINT32_UINT64_PTR): Likewise.
17424 (BT_FN_INT_INT_UINT_UINT): Likewise.
17425 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
17426 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
17427 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
17428 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
17429 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
17430 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
17431 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
17432 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
17433 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
17434 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
17435 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
17436 * doc/frontends.texi: List BRIG FE.
17437 * doc/install.texi (Testing): Add BRIG tesring requirements.
17438 * doc/invoke.texi (Overall Options): Mention BRIG.
17439 * doc/standards.texi (Standards): Doucment BRIG HSA version.
17441 2017-01-24 Richard Biener <rguenther@suse.de>
17443 PR translation/79208
17444 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
17446 2017-01-24 Martin Jambor <mjambor@suse.cz>
17449 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
17450 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
17451 and known_contexts.
17453 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
17455 PR middle-end/79123
17456 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
17457 casts from signed to unsigned really don't have a range.
17459 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
17461 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
17462 GMP_RNDx for compatiblity.
17464 2017-01-24 Martin Liska <mliska@suse.cz>
17467 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
17468 that would prevent us to call alloca with -1 as argument.
17470 2017-01-24 Jakub Jelinek <jakub@redhat.com>
17472 * dwarf2out.c (output_compilation_unit_header, output_file_names):
17473 Avoid -Wformat-security warning.
17475 2017-01-23 Andrew Pinski <apinski@cavium.com>
17477 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
17480 2017-01-23 Martin Sebor <msebor@redhat.com>
17482 PR middle-end/78703
17483 * gimple-ssa-sprintf.c (warn_level): New global.
17484 (format_integer): Use it here and throughout the rest of the file.
17485 Use the same switch to compute sign as base.
17486 (maybe_warn): New function.
17487 (format_directive): Factor out warnings into maybe_warn.
17488 Add debugging output. Use warn_level.
17489 (add_bytes): Use warn_level.
17490 (pass_sprintf_length::compute_format_length): Add debugging output.
17491 (try_substitute_return_value): Same.
17492 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
17494 PR middle-end/78703
17495 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
17496 (struct fmtresult, format_integer, format_floating): Adjust.
17497 (fmtresult::fmtresult): Set max correctly in two argument ctor.
17498 (get_string_length, format_string,format_directive): Same.
17499 (pass_sprintf_length::compute_format_length): Same.
17500 (try_substitute_return_value): Simplify slightly.
17502 PR middle-end/78703
17503 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
17504 (fmtresult::operator+=): Outlined.
17505 (struct fmtresult): Add ctors.
17506 (struct conversion_spec): Rename...
17507 (struct directive): ...to this. Add and remove data members.
17508 (directive::set_width, directive::set_precision): New functions.
17509 (format_percent): Use fmtresult ctor.
17510 (get_width_and_precision): Remove.
17511 (format_integer): Make naming changes. Avoid computing width and
17513 (format_floating): Same. Adjust indentation.
17514 (format_character, format_none): New functions.
17515 (format_string): Moved character handling to format_character.
17516 (format_directive): Remove arguments, change return type.
17517 (parse_directive): New function.
17518 (pass_sprintf_length::compute_format_length): Move directive
17519 parsing to parse_directive.
17521 2017-01-23 Jakub Jelinek <jakub@redhat.com>
17523 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
17524 (assign_assembler_name_if_needed): ... this.
17525 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
17526 (assign_assembler_name_if_needed): ... this.
17527 (free_lang_data_in_cgraph): Adjust callers.
17528 * cgraphunit.c (cgraph_node::analyze): Likewise.
17529 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
17532 2017-01-23 Richard Biener <rguenther@suse.de>
17534 PR tree-optimization/79088
17535 PR tree-optimization/79188
17536 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
17537 resetting loop bounds after last path deletion. Reset loop
17538 bounds of the target loop, make code match the comments.
17539 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
17540 Make sure loops need no fixups.
17542 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
17544 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
17545 exponent support with double type for first argument.
17546 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
17547 type returned by __builtin_vec_extract_sig,
17548 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
17549 functions from "vector int" to "vector unsigned int" or from
17550 "vector long long int" to "vector unsigned long long int".
17551 Changed type returned by __builtin_vec_extract_exp,
17552 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
17553 functions from "vector int" to "vector unsigned int" or from
17554 "vector long long int" to "vector unsigned long long int".
17555 Changed return type of __builtin_vec_test_data_class,
17556 __builtin_vec_test_data_class_sp, and
17557 __builtin_vec_test_data_class_dp from "vector int" to
17558 "vector bool int" or from "vector long long int" to "vector bool
17559 long long int" and changed second argument type from "unsigned
17560 int" to "int". Added new overloaded function forms "vector float
17561 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
17562 "vector float __builtin_vec_insert_exp_sp (vector float, vector
17563 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
17564 double, vector unsigned long long int)" and "vector double
17565 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
17566 long int)". Changed return type of
17567 __builtin_scalar_test_data_class and
17568 __builtin_scalar_test_data_class_sp and
17569 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
17570 int" and changed second argument from "unsigned int" to "int".
17571 Changed type returned by __builtin_scalar_test_neg,
17572 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
17573 from "int" to "bool int". Added new overloaded function form
17574 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
17575 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
17576 exponent double-precision with floating point first argument.
17577 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
17578 documentation of scalar_test_data_class, scalar_test_neg,
17579 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
17580 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
17581 vec_test_data_class built-in functions to reflect refinements in
17582 their type signatures.
17584 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
17586 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
17588 (aarch64_elf_asm_destructor): Likewise.
17590 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
17592 PR rtl-optimization/78634
17593 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
17594 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
17595 * ifcvt.c (noce_try_cmove): Add missing cost check.
17597 PR rtl-optimization/71724
17598 * combine.c (if_then_else_cond): Look for situations where it is
17599 beneficial to undo the work of one of the recursive calls.
17601 2017-01-23 Bin Cheng <bin.cheng@arm.com>
17603 PR tree-optimization/70754
17604 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
17605 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
17606 combined stmt before it if not NULL.
17607 (combine_chains): Process refs reversely and compute dominance point
17610 2017-01-23 Martin Liska <mliska@suse.cz>
17612 PR tree-optimization/79196
17613 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
17614 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
17616 (strlen_optimize_stmt): Call the renamed function.
17618 2017-01-23 Michael Matz <matz@suse.de>
17620 PR tree-optimization/78384
17621 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
17623 2017-01-23 Richard Biener <rguenther@suse.de>
17625 PR tree-optimization/79186
17626 * tree-vrp.c (register_new_assert_for): Make sure we've seen
17627 both incoming edges before moving an assert.
17629 2017-01-23 Martin Jambor <mjambor@suse.cz>
17631 * ipa-prop.c (load_from_param_1): Removed.
17632 (load_from_unmodified_param): Bits from load_from_param_1 put back
17634 (load_from_param): Removed.
17635 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
17636 with stmt. Reverted back to use of load_from_unmodified_param.
17638 2017-01-23 Martin Jambor <mjambor@suse.cz>
17641 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
17642 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
17643 field a pointer to garbage collected vector, mark lattices and
17644 ipcp_orig_node with GTY((skip)).
17645 (ipa_get_param_count): Adjust to descriptors being a pointer.
17646 (ipa_get_param): Likewise.
17647 (ipa_get_type): Likewise.
17648 (ipa_get_param_move_cost): Likewise.
17649 (ipa_set_param_used): Likewise.
17650 (ipa_get_controlled_uses): Likewise.
17651 (ipa_set_controlled_uses): Likewise.
17652 (ipa_is_param_used): Likewise.
17653 (ipa_node_params_t): Move into garbage collector. New methods insert
17655 (ipa_node_params_sum): Annotate wth GTY(()).
17656 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
17658 (ipa_load_from_parm_agg): Adjust declaration.
17659 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
17660 * ipa-profile.c (ipa_profile): Likewise.
17661 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
17662 (ipa_populate_param_decls): Make descriptors parameter garbage
17664 (ipa_dump_param): Adjust to descriptors being a pointer.
17665 (ipa_alloc_node_params): Likewise.
17666 (ipa_initialize_node_params): Likewise.
17667 (load_from_param_1): Make descriptors parameter garbage collected.
17668 (load_from_unmodified_param): Likewise.
17669 (load_from_param): Likewise.
17670 (ipa_load_from_parm_agg): Likewise.
17671 (ipa_node_params::~ipa_node_params): Removed.
17672 (ipa_free_all_node_params): Remove call to delete operator.
17673 (ipa_node_params_t::insert): New.
17674 (ipa_node_params_t::remove): Likewise.
17675 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
17676 copy known_csts and known_contexts vectors.
17677 (ipa_read_node_info): Adjust to descriptors being a pointer.
17678 (ipcp_modif_dom_walker): Make m_descriptors field garbage
17680 (ipcp_transform_function): Make descriptors variable garbage
17683 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
17685 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
17686 * config/i386/avx512dqintrin.h: Ditto.
17687 * config/i386/avx512fintrin.h: Ditto.
17688 * config/i386/i386.c: Handle new builtins.
17689 * config/i386/i386-builtin.def: Add new builtins.
17690 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
17691 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
17693 2017-01-23 Jakub Jelinek <jakub@redhat.com>
17694 Martin Liska <mliska@suse.cz>
17696 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
17697 * asan.c (asan_expand_poison_ifn): Support stores and use
17698 appropriate ASAN report function.
17699 * internal-fn.c (expand_ASAN_POISON_USE): New function.
17700 * internal-fn.def (ASAN_POISON_USE): Declare.
17701 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
17702 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
17703 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
17704 ASAN_POISON calls w/o LHS.
17705 * tree-ssa.c (execute_update_addresses_taken): Create clobber
17706 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
17707 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
17708 * gimplify.c (asan_poison_variables): Add attribute
17709 use_after_scope_memory to variables that really needs to live
17711 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
17712 having the attribute.
17714 2017-01-23 Martin Liska <mliska@suse.cz>
17716 * asan.c (create_asan_shadow_var): New function.
17717 (asan_expand_poison_ifn): Likewise.
17718 * asan.h (asan_expand_poison_ifn): New declaration.
17719 * internal-fn.c (expand_ASAN_POISON): Likewise.
17720 * internal-fn.def (ASAN_POISON): New builtin.
17721 * sanopt.c (pass_sanopt::execute): Expand
17722 asan_expand_poison_ifn.
17723 * tree-inline.c (copy_decl_for_dup_finish): Make function
17725 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
17726 * tree-ssa.c (is_asan_mark_p): New function.
17727 (execute_update_addresses_taken): Rewrite local variables
17728 (identified just by use-after-scope as addressable) into SSA.
17730 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
17732 * doc/install.texi (Specific): opensource.apple.com uses https
17733 now. Remove trailing slash.
17735 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
17737 * README.Portability: Remove note on an Irix compatibility issue.
17739 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
17741 * gcov.c (INCLUDE_ALGORITHM): Define.
17742 (INCLUDE_VECTOR): Define.
17743 No longer include <vector> and <algorithm> directly.
17745 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
17747 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
17749 * doc/invoke.texi (Code Gen Options): Ditto.
17751 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
17754 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
17756 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
17758 rtl-optimization/79125
17759 * cprop.c (local_cprop_pass): Handle cases where we make an
17760 unconditional trap.
17762 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
17766 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
17767 read from, for big endian.
17769 2017-01-20 Jiong Wang <jiong.wang@arm.com>
17771 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
17772 register pauth builtins for LP64 only.
17774 2017-01-20 Marek Polacek <polacek@redhat.com>
17777 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
17780 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
17782 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
17784 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
17785 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
17786 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
17788 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17791 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
17792 in big-endian mode when they are not a single duplicated value.
17794 2017-01-20 Richard Biener <rguenther@suse.de>
17796 * BASE-VER: Bump to 7.0.1.
17798 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
17800 * omp-low.c (omplow_simd_context): New struct. Use it...
17801 (lower_rec_simd_input_clauses): ...here and...
17802 (lower_rec_input_clauses): ...here to hold common data. Adjust all
17803 references to idx, lane, max_vf, is_simt.
17805 2017-01-20 Graham Markall <graham.markall@embecosm.com>
17807 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
17810 2017-01-20 Martin Jambor <mjambor@suse.cz>
17812 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
17813 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
17815 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
17816 (GTFILES): Rename hsa.c to hsa-common.c.
17817 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
17818 * hsa-dump.c: Likewise.
17819 * hsa-gen.c: Likewise.
17820 * hsa-regalloc.c: Likewise.
17821 * ipa-hsa.c: Likewise.
17822 * omp-expand.c: Likewise.
17823 * omp-low.c: Likewise.
17824 * toplev.c: Likewise.
17826 2017-01-20 Marek Polacek <polacek@redhat.com>
17829 * doc/invoke.texi: Document -Wduplicated-branches.
17830 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
17831 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
17832 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
17833 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
17834 return 0 only when not OEP_LEXICOGRAPHIC.
17835 (fold_build_cleanup_point_expr): Use the expression
17836 location when building CLEANUP_POINT_EXPR.
17837 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
17838 * tree.c (add_expr): Handle error_mark_node.
17840 2017-01-20 Martin Liska <mliska@suse.cz>
17843 * tree-profile.c (init_ic_make_global_vars): Do not call
17845 (gimple_init_gcov_profiler): Likewise.
17847 2017-01-20 Martin Liska <mliska@suse.cz>
17850 * cgraph.h (maybe_create_reference): Remove argument and
17852 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
17854 * ipa-cp.c (create_specialized_node): Likewise.
17855 * symtab.c (symtab_node::maybe_create_reference): Handle
17856 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
17858 2017-01-20 Martin Liska <mliska@suse.cz>
17860 * read-rtl-function.c (function_reader::create_function): Use
17861 build_decl instread of build_decl_stat.
17863 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
17865 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
17866 * config/i386/avx512dqintrin.h: Ditto.
17867 * config/i386/avx512fintrin.h: Ditto.
17868 * config/i386/i386-builtin-types.def: Add new types.
17869 * config/i386/i386.c: Handle new types.
17870 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
17871 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
17872 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
17873 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
17874 (__builtin_ia32_kshiftridi): New.
17875 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
17877 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
17881 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
17882 define to rs6000_init_stack_protect_guard.
17883 (rs6000_init_stack_protect_guard): New function.
17885 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
17886 Yunqiang Su <yunqiang.su@imgtec.com>
17888 * config.gcc (supported_defaults): Add madd4.
17889 (with_madd4): Add validation.
17890 (all_defaults): Add madd4.
17891 * config/mips/mips.opt (mmadd4): New option.
17892 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
17894 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
17896 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
17897 (ISA_HAS_FUSED_MADD4): Likewise.
17898 * doc/invoke.texi (-mmadd4): Document the new option.
17899 * doc/install.texi (--with-madd4): Document the new option.
17901 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17903 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
17904 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
17905 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
17906 (aarch64_init_pauth_hint_builtins): New.
17907 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
17908 (aarch64_expand_builtin): Expand new builtins.
17910 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17912 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
17913 * combine-stack-adj.c (no_unhandled_cfa): Handle
17914 REG_CFA_TOGGLE_RA_MANGLE.
17915 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
17916 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
17917 info for return address signing.
17918 (aarch64_expand_epilogue): Likewise.
17920 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17922 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
17923 * config/aarch64/aarch64-protos.h
17924 (aarch64_return_address_signing_enabled): New declaration.
17925 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
17927 (aarch64_expand_prologue): Sign return address before it's pushed onto
17929 (aarch64_expand_epilogue): Authenticate return address fetched from
17931 (aarch64_override_options): Sanity check for ILP32 and ISA level.
17932 (aarch64_attributes): New function attributes for "sign-return-address".
17933 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
17934 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
17935 ("*do_return"): Generate combined instructions according to key index.
17936 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
17937 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
17939 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
17940 * config/aarch64/aarch64.opt (msign-return-address=): New.
17941 * doc/extend.texi (AArch64 Function Attributes): Documents
17942 "sign-return-address=".
17943 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
17945 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
17947 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
17948 overall option summary.
17950 2017-01-19 Jiong Wang <jiong.wang@arm.com>
17952 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
17953 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
17954 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
17955 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
17957 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
17959 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
17960 -mpower9-minmax by default for -mcpu=power9.
17961 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
17962 128-bit floating point.
17964 2017-01-20 Alan Modra <amodra@gmail.com>
17966 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
17967 optimizing for size.
17969 2017-01-20 Alan Modra <amodra@gmail.com>
17972 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
17973 for strcmp and strncmp from corresponding builtin decl.
17975 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
17977 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
17978 instead of i386/rtems-64.h.
17979 * config/i386/rtems-64.h: Remove.
17981 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
17985 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
17987 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
17989 2017-01-19 Tamar Christina <tamar.christina@arm.com>
17991 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
17992 Change int to HOST_WIDE_INT.
17993 * config/aarch64/aarch64-protos.h
17994 (aarch64_simd_gen_const_vector_dup): Likewise.
17995 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
17997 2017-01-19 David Malcolm <dmalcolm@redhat.com>
17999 * langhooks-def.h (lhd_type_for_size): New decl.
18000 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
18001 * langhooks.c (lhd_type_for_size): New function, taken from
18004 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
18006 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
18007 define_bypass for CR latency.
18008 (power9-cracked-alu): Update bypass latency and remove power9-branch.
18009 (power9-alu2): Add define_bypass for CR latency.
18011 (power9-mul): Update insn latency.
18012 (power9-mul-compare): Update insn latency, bypass latency and remove
18015 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18017 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
18019 * config/aarch64/aarch64.md
18020 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
18021 aarch64_nopcrelative_literal_loads.
18022 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
18024 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
18026 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
18027 TARGET_LOONGSON_3A.
18028 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
18030 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
18033 * config.gcc (supported_defaults): Add lxc1-sxc1.
18034 (with_lxc1_sxc1): Add validation.
18035 (all_defaults): Add lxc1-sxc1.
18036 * config/mips/mips.opt (mlxc1-sxc1): New option.
18037 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
18039 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
18040 __mips_no_lxc1_sxc1.
18041 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
18042 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
18043 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
18045 2017-01-19 Richard Biener <rguenther@suse.de>
18047 PR tree-optimization/72488
18048 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
18049 sure to restore SSA info.
18050 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
18052 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
18054 PR rtl-optimization/79121
18055 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
18056 of the inner type when shifting an extended value.
18058 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
18061 * symtab.c (symtab_node::equal_address_to): Fix comparing of
18062 interposable aliases.
18064 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
18067 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
18068 Use the evmergelohi instruction.
18069 (mov_si<mode>_e500_subreg4_2_le): Likewise.
18070 (mov_sitf_e500_subreg8_2_be): Likewise.
18071 (mov_sitf_e500_subreg12_2_le): Likewise.
18072 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
18073 (mov_si<mode>_e500_subreg4_2_be): Likewise.
18074 (mov_sitf_e500_subreg8_2_le): Likewise.
18075 (mov_sitf_e500_subreg12_2_be): Likewise.
18077 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18079 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
18080 attribute from vecsimple to vecperm.
18081 (altivec_vbpermq2): Likewise.
18083 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18086 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
18088 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18089 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
18090 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
18091 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
18092 case where N arg is SIZE_MAX.
18093 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
18094 (cmpstrsi): Add pattern.
18096 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
18098 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18099 __builtin_vec_revb builtins.
18100 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
18101 built-in functions to support generation of the ISA 3.0 XXBR<x>
18102 vector byte reverse instructions.
18103 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
18104 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
18105 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
18106 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
18107 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
18108 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
18109 (P9V_BUILTIN_VEC_REVB): Likewise.
18110 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
18111 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
18112 (p9_xxbrq_v16qi): Likewise.
18113 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
18114 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
18115 (p9_xxbrh_v8hi): Likewise.
18116 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
18117 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
18118 vec_revb built-in functions.
18120 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
18122 PR rtl-optimization/78952
18123 * config/i386/i386.md (any_extract): New code iterator.
18124 (*insvqi_2): Use any_extract for source operand.
18125 (*insvqi_3): Use any_shiftrt for source operand.
18127 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
18129 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
18131 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
18133 2017-01-18 Matthias Klose <doko@ubuntu.com>
18135 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
18137 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18139 * config/rs6000/altivec.h (vec_bperm): Change #define.
18140 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
18141 (altivec_vbpermq2): New define_insn.
18142 (altivec_vbpermd): Likewise.
18143 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
18144 function interface.
18145 (VBPERMD): Likewise.
18146 (VBPERM): New polymorphic function interface.
18147 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
18148 Add entries for P9V_BUILTIN_VEC_VBPERM.
18149 * doc/extend.texi: Add interfaces for vec_bperm.
18151 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18153 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
18154 first letter of error messages.
18155 (s390_resolve_overloaded_builtin): Likewise.
18156 * config/s390/s390.c (s390_expand_builtin): Likewise.
18157 (s390_invalid_arg_for_unprototyped_fn): Likewise.
18158 (s390_valid_target_attribute_inner_p): Likewise.
18159 * config/s390/s390.md ("tabort"): Likewise.
18161 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
18163 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
18164 (ISA_AVOID_DIV_HILO): New macro.
18165 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
18166 (ISA_HAS_DDIV): Likewise.
18168 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
18170 * doc/invoke.texi (fabi-version): Correct number of occurrences.
18172 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
18174 * doc/invoke.texi (fabi-version): Spelling fix.
18176 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
18179 * doc/invoke.texi (fabi-version): Mention mangling fix for
18182 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
18185 * doc/invoke.texi (fabi-version): Document discriminator mangling.
18187 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
18190 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
18191 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
18193 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
18195 (stack_protect_test): Ditto.
18196 * config/rs6000/rs6000.opt (mstack-protector-guard=,
18197 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
18199 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
18200 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
18201 -mstack-protector-guard-offset=.
18202 (RS/6000 and PowerPC Options): Ditto.
18204 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
18206 * config/i386/i386.h (MASK_CLASS_P): New define.
18207 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
18208 there are no registers from different register sets also when
18209 mask registers are used. Update function comment.
18210 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
18211 to (*k/*r) and (*k/*km) alternatives.
18213 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
18215 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
18216 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
18217 (EH_RETURN_HANDLER_RTX): New define.
18218 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
18219 Force frame pointer in EH return functions.
18220 (aarch64_expand_epilogue): Add barrier for eh_return.
18221 (aarch64_final_eh_return_addr): Remove.
18222 (aarch64_eh_return_handler_rtx): New function.
18223 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
18225 (aarch64_eh_return_handler_rtx): New prototype.
18227 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18229 * config/rs6000/altivec.h (vec_rlmi): New #define.
18230 (vec_vrlnm): Likewise.
18231 (vec_rlnm): Likewise.
18232 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
18233 (UNSPEC_VRLNM): Likewise.
18234 (VIlong): New mode iterator.
18235 (altivec_vrl<VI_char>mi): New define_insn.
18236 (altivec_vrl<VI_char>nm): Likewise.
18237 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
18239 (VRLDNM): Likewise.
18240 (RLNM): New polymorphic function entry.
18241 (VRLWMI): New monomorphic function entry.
18242 (VRLDMI): Likewise.
18243 (RLMI): New polymorphic function entry.
18244 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
18245 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
18246 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
18249 2017-01-17 Jakub Jelinek <jakub@redhat.com>
18252 * dwarf2out.c (field_byte_offset): Restore the
18253 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
18254 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
18255 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
18258 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
18261 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
18263 2017-01-17 Jakub Jelinek <jakub@redhat.com>
18266 * dwarf2out.c (add_data_member_location_attribute): For constant
18267 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
18268 instead of DW_AT_data_member_location, DW_AT_bit_offset and
18269 DW_AT_byte_size attributes.
18271 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
18273 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
18274 after forcing to constant memory when the code model is medium.
18276 2017-01-17 Julia Koval <julia.koval@intel.com>
18279 * config/i386/avx512fintrin.h
18280 (_mm512_i32gather_ps): Change __addr type to void const*.
18281 (_mm512_mask_i32gather_ps): Ditto.
18282 (_mm512_i32gather_pd): Ditto.
18283 (_mm512_mask_i32gather_pd): Ditto.
18284 (_mm512_i64gather_ps): Ditto.
18285 (_mm512_mask_i64gather_ps): Ditto.
18286 (_mm512_i64gather_pd): Ditto.
18287 (_mm512_mask_i64gather_pd): Ditto.
18288 (_mm512_i32gather_epi32): Ditto.
18289 (_mm512_mask_i32gather_epi32): Ditto.
18290 (_mm512_i32gather_epi64): Ditto.
18291 (_mm512_mask_i32gather_epi64): Ditto.
18292 (_mm512_i64gather_epi32): Ditto.
18293 (_mm512_mask_i64gather_epi32): Ditto.
18294 (_mm512_i64gather_epi64): Ditto.
18295 (_mm512_mask_i64gather_epi64): Ditto.
18296 (_mm512_i32scatter_ps): Change __addr type to void*.
18297 (_mm512_mask_i32scatter_ps): Ditto.
18298 (_mm512_i32scatter_pd): Ditto.
18299 (_mm512_mask_i32scatter_pd): Ditto.
18300 (_mm512_i64scatter_ps): Ditto.
18301 (_mm512_mask_i64scatter_ps): Ditto.
18302 (_mm512_i64scatter_pd): Ditto.
18303 (_mm512_mask_i64scatter_pd): Ditto.
18304 (_mm512_i32scatter_epi32): Ditto.
18305 (_mm512_mask_i32scatter_epi32): Ditto.
18306 (_mm512_i32scatter_epi64): Ditto.
18307 (_mm512_mask_i32scatter_epi64): Ditto.
18308 (_mm512_i64scatter_epi32): Ditto.
18309 (_mm512_mask_i64scatter_epi32): Ditto.
18310 (_mm512_i64scatter_epi64): Ditto.
18311 (_mm512_mask_i64scatter_epi64): Ditto.
18312 * config/i386/avx512pfintrin.h
18313 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
18314 (_mm512_mask_prefetch_i32gather_ps): Ditto.
18315 (_mm512_mask_prefetch_i64gather_pd): Ditto.
18316 (_mm512_mask_prefetch_i64gather_ps): Ditto.
18317 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
18318 (_mm512_prefetch_i32scatter_ps): Ditto.
18319 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
18320 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
18321 (_mm512_prefetch_i64scatter_pd): Ditto.
18322 (_mm512_prefetch_i64scatter_ps): Ditto.
18323 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
18324 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
18325 * config/i386/avx512vlintrin.h
18326 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
18327 (_mm_mmask_i32gather_ps): Ditto.
18328 (_mm256_mmask_i32gather_pd): Ditto.
18329 (_mm_mmask_i32gather_pd): Ditto.
18330 (_mm256_mmask_i64gather_ps): Ditto.
18331 (_mm_mmask_i64gather_ps): Ditto.
18332 (_mm256_mmask_i64gather_pd): Ditto.
18333 (_mm_mmask_i64gather_pd): Ditto.
18334 (_mm256_mmask_i32gather_epi32): Ditto.
18335 (_mm_mmask_i32gather_epi32): Ditto.
18336 (_mm256_mmask_i32gather_epi64): Ditto.
18337 (_mm_mmask_i32gather_epi64): Ditto.
18338 (_mm256_mmask_i64gather_epi32): Ditto.
18339 (_mm_mmask_i64gather_epi32): Ditto.
18340 (_mm256_mmask_i64gather_epi64): Ditto.
18341 (_mm_mmask_i64gather_epi64): Ditto.
18342 (_mm256_i32scatter_ps): Change __addr type to void*.
18343 (_mm256_mask_i32scatter_ps): Ditto.
18344 (_mm_i32scatter_ps): Ditto.
18345 (_mm_mask_i32scatter_ps): Ditto.
18346 (_mm256_i32scatter_pd): Ditto.
18347 (_mm256_mask_i32scatter_pd): Ditto.
18348 (_mm_i32scatter_pd): Ditto.
18349 (_mm_mask_i32scatter_pd): Ditto.
18350 (_mm256_i64scatter_ps): Ditto.
18351 (_mm256_mask_i64scatter_ps): Ditto.
18352 (_mm_i64scatter_ps): Ditto.
18353 (_mm_mask_i64scatter_ps): Ditto.
18354 (_mm256_i64scatter_pd): Ditto.
18355 (_mm256_mask_i64scatter_pd): Ditto.
18356 (_mm_i64scatter_pd): Ditto.
18357 (_mm_mask_i64scatter_pd): Ditto.
18358 (_mm256_i32scatter_epi32): Ditto.
18359 (_mm256_mask_i32scatter_epi32): Ditto.
18360 (_mm_i32scatter_epi32): Ditto.
18361 (_mm_mask_i32scatter_epi32): Ditto.
18362 (_mm256_i32scatter_epi64): Ditto.
18363 (_mm256_mask_i32scatter_epi64): Ditto.
18364 (_mm_i32scatter_epi64): Ditto.
18365 (_mm_mask_i32scatter_epi64): Ditto.
18366 (_mm256_i64scatter_epi32): Ditto.
18367 (_mm256_mask_i64scatter_epi32): Ditto.
18368 (_mm_i64scatter_epi32): Ditto.
18369 (_mm_mask_i64scatter_epi32): Ditto.
18370 (_mm256_i64scatter_epi64): Ditto.
18371 (_mm256_mask_i64scatter_epi64): Ditto.
18372 (_mm_i64scatter_epi64): Ditto.
18373 (_mm_mask_i64scatter_epi64): Ditto.
18374 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
18375 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
18376 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
18377 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
18378 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
18379 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
18380 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
18381 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
18382 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
18383 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
18384 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
18385 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
18386 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
18387 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
18388 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
18389 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
18390 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
18391 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
18392 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
18393 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
18394 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
18395 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
18396 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
18397 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
18398 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
18399 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
18400 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
18401 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
18402 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
18403 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
18404 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
18405 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
18406 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
18407 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
18408 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
18409 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
18410 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
18411 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
18412 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
18413 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
18414 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
18415 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
18416 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
18417 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
18418 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
18419 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
18420 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
18421 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
18422 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
18423 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
18424 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
18425 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
18426 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
18427 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
18428 definitions accordingly.
18430 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
18431 Kuan-Lin Chen <kuanlinchentw@gmail.com>
18434 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
18437 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
18440 * ira-conflicts.c (ira_build_conflicts): Update total conflict
18441 hard regs for inner regno.
18443 2017-01-17 Martin Liska <mliska@suse.cz>
18446 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
18447 assumption and add comment.
18449 2017-01-17 Nathan Sidwell <nathan@acm.org>
18451 * ipa-visibility.c (localize_node): New function, broken out of ...
18452 (function_and_variable_visibility): ... here. Call it.
18454 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
18456 PR middle-end/77445
18457 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
18458 correctly set frequency of oudgoing edge.
18459 (duplicate_thread_path): Fix profile updating.
18461 2017-01-17 Jakub Jelinek <jakub@redhat.com>
18464 * configure.ac: Add GCC_BASE_VER.
18465 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
18466 version from BASE-VER file.
18467 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
18468 (gcc.o): Depend on $(BASEVER).
18469 * common.opt (dumpfullversion): New option.
18470 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
18471 * doc/invoke.texi: Document -dumpfullversion.
18472 * doc/install.texi: Document --with-gcc-major-version-only.
18473 * configure: Regenerated.
18475 2017-01-17 Richard Biener <rguenther@suse.de>
18477 PR tree-optimization/71433
18478 * tree-vrp.c (register_new_assert_for): Merge same asserts
18479 on all incoming edges.
18480 (process_assert_insertions_for): Handle insertions at the
18483 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
18485 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
18486 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
18488 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
18491 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
18494 2017-01-17 Alan Modra <amodra@gmail.com>
18497 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
18498 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
18499 symbolic stack limit when pic.
18501 2017-01-16 Martin Sebor <msebor@redhat.com>
18503 PR tree-optimization/78608
18504 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
18506 2017-01-16 Jeff Law <law@redhat.com>
18509 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
18510 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
18511 for several include directories that may be relative to sysroot.
18512 * config/i386/x-mingw32 (gplus_includedir): Define.
18513 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
18514 (native_system_includedir): Likewise.
18515 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
18516 override if TARGET_SYSTEM_ROOT is defined.
18517 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
18519 PR tree-optimization/79090
18520 PR tree-optimization/33562
18521 PR tree-optimization/61912
18522 PR tree-optimization/77485
18523 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
18524 and computed trims into the dump file.
18526 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
18528 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
18530 2017-01-16 Jakub Jelinek <jakub@redhat.com>
18533 * gimplify.c (gimplify_init_constructor): If want_value and
18534 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
18538 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
18539 sequence. Formatting fixes.
18540 (doloop_optimize): Formatting fixes.
18543 * gcc.c (debug_level_greater_than_spec_func): New function.
18544 (static_spec_functions): Add debug-level-gt spec function.
18545 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
18547 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
18548 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
18549 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
18550 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
18551 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
18552 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
18554 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
18556 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
18557 QImode fixups to general and mask registers only.
18559 2017-01-16 Carl Love <cel@us.ibm.com>
18561 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
18562 for built-in functions
18563 vector signed char vec_nabs (vector signed char)
18564 vector signed short vec_nabs (vector signed short)
18565 vector signed int vec_nabs (vector signed int)
18566 vector signed long long vec_nabs (vector signed long long)
18567 vector float vec_nabs (vector float)
18568 vector double vec_nabs (vector double)
18569 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
18571 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
18572 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
18573 * doc/extend.texi: Update the documentation file for the new built-in
18576 2017-01-16 Martin Sebor <msebor@redhat.com>
18578 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
18581 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18583 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
18584 UNSPEC_VSX__XXSPLTD to require special splat handling.
18586 2017-01-16 David Malcolm <dmalcolm@redhat.com>
18589 * system.h: Poison strndup.
18591 2017-01-16 Alan Modra <amodra@gmail.com>
18594 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
18597 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
18599 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
18601 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
18603 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
18604 call recog here. Assert that INSN_CODE (insn) is non-negative.
18606 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
18609 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
18611 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
18612 in the currently scheduled RTL fragment.
18614 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
18616 PR rtl-optimization/78751
18617 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
18620 2017-01-14 Jeff Law <law@redhat.com>
18622 PR tree-optimization/79090
18623 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
18624 variable length stores.
18625 (compute_trims): Delete dead assignment to *trim_tail.
18626 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
18629 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
18631 PR rtl-optimization/78626
18632 PR rtl-optimization/78727
18633 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
18634 of a block, and split such blocks after everything else is finished.
18636 2017-01-14 Alan Modra <amodra@gmail.com>
18639 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
18640 target legitimate_combined_insn.
18641 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
18642 (rs6000_legitimate_combined_insn): New function.
18643 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
18645 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
18646 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
18647 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
18649 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
18651 * doc/frontends.texi (G++ and GCC): Remove references to Java.
18653 2017-01-13 Jeff Law <law@redhat.com>
18655 PR tree-optimization/33562
18656 PR tree-optimization/61912
18657 PR tree-optimization/77485
18658 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
18660 (delete_dead_assignment): Likewise.
18661 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
18662 statement to delete_dead_call and delete_dead_assignment.
18664 2017-01-13 David Malcolm <dmalcolm@redhat.com>
18667 * substring-locations.c (format_warning_va): Strengthen case 1 so
18668 that both endpoints of the substring must be within the format
18669 range for just the substring to be printed.
18671 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
18673 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
18674 * config/i386/i386.c (ix86_target_string): Add missing options
18675 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
18676 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
18677 flags_other and ix86_target_other to flags2_other. Display unknown
18679 (ix86_valid_target_attribute_inner_p): Add missing options and
18680 reorder options by implied ISAs, as in ix86_target_string.
18682 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
18684 * hash-table.h (hash_table::too_empty_p): New function.
18685 (hash_table::expand): Use it.
18686 (hash_table::traverse): Likewise.
18687 (hash_table::empty_slot): Use sizeof (value_type) instead of
18688 sizeof (PTR) to convert bytes to elements. Shrink the table
18689 if the current size is excessive for the current number of
18692 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
18694 * ira-costs.c (record_reg_classes): Break from the inner loop
18695 early once alt_fail is known to be true. Update outer loop
18696 handling accordingly.
18698 2017-01-13 Jeff Law <law@redhat.com>
18700 * tree-ssa-dse.c (decrement_count): New function.
18701 (increment_start_addr, maybe_trim_memstar_call): Likewise.
18702 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
18703 when we know the partially dead statement is a mem* function.
18705 PR tree-optimization/61912
18706 PR tree-optimization/77485
18707 * tree-ssa-dse.c: Include expr.h.
18708 (maybe_trim_constructor_store): New function.
18709 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
18711 PR tree-optimization/33562
18712 PR tree-optimization/61912
18713 PR tree-optimization/77485
18714 * doc/invoke.texi: Document new dse-max-object-size param.
18715 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
18716 * tree-ssa-dse.c: Include params.h.
18717 (dse_store_status): New enum.
18718 (initialize_ao_ref_for_dse): New, partially extracted from
18720 (valid_ao_ref_for_dse, normalize_ref): New.
18721 (setup_live_bytes_from_ref, compute_trims): Likewise.
18722 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
18723 (maybe_trim_partially_dead_store): Likewise.
18724 (maybe_trim_complex_store): Likewise.
18725 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
18726 Track what bytes live from the original store. Return tri-state
18727 for dead, partially dead or live.
18728 (dse_dom_walker): Add constructor, destructor and new private members.
18729 (delete_dead_call, delete_dead_assignment): New extracted from
18731 (dse_optimize_stmt): Make a member of dse_dom_walker.
18732 Use initialize_ao_ref_for_dse.
18734 PR tree-optimization/33562
18735 PR tree-optimization/61912
18736 PR tree-optimization/77485
18737 * sbitmap.h (bitmap_count_bits): Prototype.
18738 (bitmap_clear_range, bitmap_set_range): Likewise.
18739 * sbitmap.c (bitmap_clear_range): New function.
18740 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
18742 2017-01-13 Martin Liska <mliska@suse.cz>
18745 * function.c (set_cfun): Add new argument force.
18746 * function.h (set_cfun): Likewise.
18747 * ipa-inline-transform.c (inline_call): Use the function when
18748 strict alising from is dropped for function we inline to.
18750 2017-01-13 Richard Biener <rguenther@suse.de>
18752 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
18753 for dumping GIMPLE INTEGER_CSTs.
18755 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18757 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
18758 to 201112L since C++17.
18760 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
18763 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
18764 if -fsanitize=kernel-address is present.
18766 2017-01-13 Richard Biener <rguenther@suse.de>
18768 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
18769 as _Literal ( type ) number in case usual suffixes do not
18770 preserve all information.
18772 2017-01-13 Richard Biener <rguenther@suse.de>
18774 PR tree-optimization/77283
18775 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
18776 and ssa-iterators.h.
18777 (is_feasible_trace): Implement a cost model based on joiner
18780 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
18783 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
18784 char or short to __float128/_Float128 directly.
18786 2017-01-12 Martin Sebor <msebor@redhat.com>
18788 to -Wformat-overflow.
18789 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
18790 (min_bytes_remaining): Same.
18791 (get_string_length): Same.
18792 (format_string): Same.
18793 (format_directive): Same.
18795 (pass_sprintf_length::handle_gimple_call): Same.
18797 2017-01-12 Jakub Jelinek <jakub@redhat.com>
18799 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
18800 info.nowrite calls with no lhs that can't throw. Return bool
18801 whether gsi_remove has been called or not.
18802 (pass_sprintf_length::handle_gimple_call): Return bool whether
18803 try_substitute_return_value called gsi_remove. Formatting fix.
18804 (pass_sprintf_length::execute): Don't use gsi_remove if
18805 handle_gimple_call returned true.
18808 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
18809 be removed due to side-effects, don't remove following barrier nor
18810 turn the successor edge into fallthru edge.
18812 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18815 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
18816 element-reversing loads and stores as not swappable.
18818 2017-01-12 Nathan Sidwell <nathan@acm.org>
18819 Nicolai Stange <nicstange@gmail.com>
18821 * combine.c (try_combine): Don't ignore result of overlap checking
18822 loop. Combine overlap & asm check into single loop.
18824 2017-01-12 Richard Biener <rguenther@suse.de>
18826 * tree-pretty-print.c (dump_generic_node): Provide -gimple
18827 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
18829 2017-01-12 Richard Biener <rguenther@suse.de>
18831 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
18832 and TS_TARGET_OPTION directly derive from TS_BASE.
18833 * tree-core.h (tree_optimization_option): Derive from tree_base.
18834 (tree_target_option): Likewise.
18836 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
18838 * config/i386/i386.c (memory_address_length): Increase len
18839 only when rip_relative_addr_p returns false.
18841 2017-01-11 Julia Koval <julia.koval@intel.com>
18843 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
18844 (OPTION_MASK_ISA_SGX_SET): New.
18845 (ix86_handle_option): Handle OPT_msgx.
18846 * config.gcc: Added sgxintrin.h.
18847 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
18848 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
18849 * config/i386/i386.c (ix86_target_string): Add -msgx.
18851 (ix86_option_override_internal): Handle new options.
18852 (ix86_valid_target_attribute_inner_p): Add sgx.
18853 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
18854 * config/i386/i386.opt: Add msgx.
18855 * config/i386/sgxintrin.h: New file.
18856 * config/i386/x86intrin.h: Add sgxintrin.h.
18858 2017-01-11 Jakub Jelinek <jakub@redhat.com>
18861 * fold-const.c (maybe_nonzero_address): Return 1 for function
18863 (tree_single_nonzero_warnv_p): Don't handle function local objects
18867 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
18870 2017-01-11 David Malcolm <dmalcolm@redhat.com>
18873 * opts.c: Include "spellcheck.h"
18874 (struct string_fragment): New struct.
18875 (struct edit_distance_traits<const string_fragment &>): New
18877 (get_closest_sanitizer_option): New function.
18878 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
18880 2017-01-11 Jakub Jelinek <jakub@redhat.com>
18882 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
18884 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
18885 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
18886 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
18887 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
18888 for initial die_offset if dwarf_split_debug_info.
18889 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
18890 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
18892 (output_skeleton_debug_sections): Formatting fix. Use
18893 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
18894 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
18896 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
18898 * config/arm/cortex-a53.md: Add bypasses for
18899 cortex_a53_r2f_cvt.
18900 (cortex_a53_r2f): Only use for transfers.
18901 (cortex_a53_f2r): Likewise.
18902 (cortex_a53_r2f_cvt): Add reservation for conversions.
18903 (cortex_a53_f2r_cvt): Likewise.
18905 2017-01-11 Tamar Christina <tamar.christina@arm.com>
18907 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
18908 to all inlined functions, change static to extern.
18910 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
18913 * config/arm/arm.c (legitimize_pic_address): Handle reference to
18915 (arm_assemble_integer): Likewise.
18917 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
18919 * config.gcc: Use new awk script to check CPU, FPU and architecture
18920 parameters for --with-... options.
18921 * config/arm/parsecpu.awk: New file
18922 * config/arm/arm-cpus.in: New file.
18923 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
18925 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
18927 * config/arm/t-arm: Update dependency rules.
18928 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
18929 of processing .def files.
18930 * config/arm/genopt.sh: Deleted.
18931 * config/arm/gentune.sh: Deleted.
18932 * config/arm/arm-cores.def: Deleted.
18933 * config/arm/arm-arches.def: Deleted.
18934 * config/arm/arm-fpus.def: Deleted.
18935 * config/arm/arm-tune.md: Regenerated.
18936 * config/arm/arm-tables.opt: Regenerated.
18937 * config/arm/arm-cpu.h: New generated file.
18938 * config/arm/arm-cpu-data.h: New generated file.
18939 * config/arm/arm-cpu-cdata.h: New generated file.
18941 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
18944 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
18946 (input_varpool_node): Unpack dynamically_initialized bit.
18948 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
18950 PR rtl-optimization/79032
18951 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
18952 the alignment of the adjusted memory reference against that of MODE,
18953 instead of the alignment of the original memory reference.
18955 2017-01-11 Martin Jambor <mjambor@suse.cz>
18957 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
18959 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
18960 decorated functions.
18962 2017-01-11 Richard Biener <rguenther@suse.de>
18964 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
18965 set range/nonnull info for PHI results. Do not set it on
18966 stmts marked for removal.
18968 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
18970 * expr.c (store_field): In the bitfield case, fetch the return value
18971 from the registers before applying a single big-endian adjustment.
18972 Always do a final load for a BLKmode value not larger than a word.
18974 2017-01-10 David Malcolm <dmalcolm@redhat.com>
18977 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
18978 that we correctly handle column numbers greater than
18979 LINE_MAP_MAX_COLUMN_NUMBER.
18981 2017-01-10 Martin Sebor <msebor@redhat.com>
18983 PR middle-end/78245
18984 * gimple-ssa-sprintf.c (get_destination_size): Call
18985 {init,fini}object_sizes.
18986 * tree-object-size.c (addr_object_size): Adjust.
18987 (pass_through_call): Adjust.
18988 (pass_object_sizes::execute): Adjust.
18989 * tree-object-size.h (fini_object_sizes): Declare.
18991 2017-01-10 Martin Sebor <msebor@redhat.com>
18993 PR tree-optimization/78775
18994 * builtins.c (get_size_range): Move...
18995 * calls.c: ...to here.
18996 (alloc_max_size): Accept zero argument.
18997 (operand_signed_p): Remove.
18998 (maybe_warn_alloc_args_overflow): Call get_size_range.
18999 * calls.h (get_size_range): Declare.
19001 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
19003 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
19004 from TI's devices.csv file as of September 2016.
19005 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
19007 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
19009 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
19010 * doc/invoke.texi: Likewise.
19011 * doc/md.texi: Likewise.
19012 * doc/objc.texi: Likewise.
19014 2017-01-10 Joshua Conner <joshconner@google.com>
19016 * config/arm/fuchsia-elf.h: New file.
19017 * config/fuchsia.h: New file.
19018 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
19019 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
19021 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
19023 2016-01-10 Richard Biener <rguenther@suse.de>
19025 PR tree-optimization/79034
19026 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
19027 Propagate out degenerate PHIs in the joiner.
19029 2017-01-10 Martin Liska <mliska@suse.cz>
19031 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
19032 (sort_congruence_classes_by_decl_uid): Likewise.
19033 (sort_congruence_class_groups_by_decl_uid): Likewise.
19034 (sem_item_optimizer::merge_classes): Sort class, groups in these
19035 classes and members in the groups by DECL_UID of declarations.
19036 This would make merge operations stable.
19038 2017-01-10 Martin Liska <mliska@suse.cz>
19040 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
19041 usage of m_classes_vec.
19042 (sem_item_optimizer::~sem_item_optimizer): Likewise.
19043 (sem_item_optimizer::get_group_by_hash): Likewise.
19044 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
19045 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
19046 (sem_item_optimizer::verify_classes): Likewise.
19047 (sem_item_optimizer::process_cong_reduction): Likewise.
19048 (sem_item_optimizer::dump_cong_classes): Likewise.
19049 (sem_item_optimizer::merge_classes): Likewise.
19050 * ipa-icf.h (congruence_class_hash): Rename from
19051 congruence_class_group_hash. Remove declaration of m_classes_vec.
19053 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
19055 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
19056 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
19057 * config.gcc: Add avx512vpopcntdqintrin.h.
19058 * config/i386/avx512vpopcntdqintrin.h: New.
19059 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
19060 * config/i386/i386-builtin-types.def: Add new types.
19061 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
19062 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
19063 __builtin_ia32_vpopcountq_v8di_mask): New.
19064 * config/i386/i386-c.c (ix86_target_macros_internal): Define
19065 __AVX512VPOPCNTDQ__.
19066 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
19067 (PTA_AVX512VPOPCNTDQ): Define.
19068 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
19069 TARGET_AVX512VPOPCNTDQ_P): Define.
19070 * config/i386/i386.opt: Add mavx512vpopcntdq.
19071 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
19072 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
19074 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
19076 PR middle-end/77484
19077 * predict.def (PRED_CALL): Set to 67.
19079 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
19081 * expr.c (store_field): In the bitfield case, if the value comes from
19082 a function call and is of an aggregate type returned in registers, do
19083 not modify the field mode; extract the value in all cases if the mode
19084 is BLKmode and the size is not larger than a word.
19086 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
19089 * config/i386/cpuid.h: Fix undefined behavior.
19091 2017-01-04 Jeff Law <law@redhat.com>
19093 PR tree-optimization/79007
19094 PR tree-optimization/67955
19095 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
19096 conservative for pt.null when flag_non_call_exceptions is on.
19098 2017-01-09 Jakub Jelinek <jakub@redhat.com>
19100 PR translation/79019
19101 PR translation/79020
19102 * params.def (PARAM_INLINE_MIN_SPEEDUP,
19103 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
19104 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
19106 * config/avr/avr.opt (maccumulate-args): Likewise.
19107 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
19108 * common.opt (freport-bug): Likewise.
19109 * cif-code.def (CIF_FINAL_ERROR): Likewise.
19110 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
19111 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
19112 translatable string.
19113 * config/i386/i386.c (function_value_32): Likewise.
19114 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
19115 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
19117 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
19118 * common/config/msp430/msp430-common.c (msp430_handle_option):
19120 * symtab.c (symtab_node::verify_base): Likewise.
19121 * opts.c (set_debug_level): Likewise.
19122 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
19123 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
19124 missing whitespace to translatable strings.
19125 * config/avr/avr.md (bswapsi2): Fix typo in comment.
19126 * config/sh/superh.h: Likewise.
19127 * config/i386/xopintrin.h: Likewise.
19128 * config/i386/znver1.md: Likewise.
19129 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
19130 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
19131 * double-int.h (struct double_int): Likewise.
19132 * double-int.c (div_and_round_double): Likewise.
19133 * wide-int.cc: Likewise.
19134 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
19135 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
19136 * cfgcleanup.c (crossjumps_occured): Renamed to ...
19137 (crossjumps_occurred): ... this.
19138 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
19141 PR tree-optimization/78899
19142 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
19143 returning bool return struct loop *, NULL for failure and the new
19145 (versionable_outer_loop_p): Don't version outer loop if it has
19146 dont_vectorized bit set.
19147 (tree_if_conversion): When versioning outer loop, ensure
19148 tree_if_conversion is performed also on the inner loop of the
19149 non-vectorizable outer loop copy.
19150 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
19151 LOOP_VECTORIZED in inner loop of the scalar outer loop and
19152 prevent vectorization of it.
19153 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
19154 the outer loop vectorization of the non-scalar version is attempted
19155 before vectorization of the inner loop in scalar version. If
19156 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
19157 vectorization of its inner loop.
19158 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
19159 has 2 inner loops, rename also on edges from bb whose single pred
19160 is outer_loop->header. Fix typo in function comment.
19162 2017-01-09 Martin Sebor <msebor@redhat.com>
19165 * asan.c (asan_emit_stack_protection): Increase local buffer size
19166 to avoid snprintf truncation warning.
19168 2017-01-09 Andrew Pinski <apinski@cavium.com>
19170 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
19171 to reference thunderx2t99 for the tuning structure
19172 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
19174 (thunderx2t99_extra_costs): This.
19175 * config/aarch64/aarch64-tune.md: Regenerate.
19176 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
19177 (vulcan_addrcost_table): This.
19178 (vulcan_regmove_cost): Rename to ...
19179 (thunderx2t99_regmove_cost): This.
19180 (vulcan_vector_cost): Rename to ...
19181 (thunderx2t99_vector_cost): this.
19182 (vulcan_branch_cost): Rename to ...
19183 (thunderx2t99_branch_cost): This.
19184 (vulcan_tunings): Rename to ...
19185 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
19186 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
19188 2017-01-09 Martin Jambor <mjambor@suse.cz>
19192 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
19193 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
19194 (propagate_vr_accross_jump_function): Use the above function for all
19195 value range computations for pass-through jump functions and type
19196 converasion from explicit value range values.
19197 (ipcp_propagate_stage): Do not attempt to deduce types of formal
19198 parameters from TYPE_ARG_TYPES.
19199 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
19200 (ipa_write_node_info): Stream type of the actual argument.
19201 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
19203 2017-01-09 Martin Liska <mliska@suse.cz>
19206 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
19207 (lookup_compiler): Do not show error message with have_E.
19209 2017-01-09 Jakub Jelinek <jakub@redhat.com>
19211 PR tree-optimization/78938
19212 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
19213 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
19214 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
19215 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
19218 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19220 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
19223 2017-01-09 Richard Biener <rguenther@suse.de>
19225 PR tree-optimization/78997
19226 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
19227 name condition properly.
19229 2017-01-09 Richard Biener <rguenther@suse.de>
19232 * dwarf2out.c (is_cxx): New overload with context.
19233 (is_naming_typedef_decl): Use it.
19235 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
19237 * invoke.texi (Option Summary): Correct spacing in option lists
19238 and add line breaks to fix over-long lines.
19240 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
19242 PR middle-end/17660
19244 * extend.texi (Common Variable Attributes): Add xref to GCC
19245 Internals manual to explain mode attribute keywords.
19247 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
19250 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
19251 and Preprocessor Options.
19252 (Options for Linking): Document -pthread here....
19253 (RS/6000 and PowerPC Options): ...not here.
19254 (Solaris 2 Options): ...or here.
19255 * doc/cppopts.texi: Document -pthread.
19257 2017-01-08 Martin Sebor <msebor@redhat.com>
19259 PR middle-end/77708
19260 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
19261 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
19262 New member functions.
19263 (format_directive): Used them.
19265 (pass_sprintf_length::handle_gimple_call): Same.
19266 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
19267 to avoid truncation for any argument.
19268 (extract_affine_mul): Same.
19269 * tree.c (get_file_function_name): Same.
19271 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
19273 PR middle-end/77484
19274 * predict.def (PRED_INDIR_CALL): Set to 86.
19276 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
19278 PR preprocessor/54124
19279 * doc/cppopts.texi: Reformat -d subtable to list the full name
19280 of the options. Add cross-reference to the docs for the general
19281 compiler -d options.
19282 * doc/invoke.texi (Developer Options): Add cross-reference to the
19283 preprocessor-specific -d option documentation.
19285 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
19287 PR preprocessor/13498
19288 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
19289 redudant material, and reflect new command-line options.
19290 (System Headers): Likewise.
19292 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
19294 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
19295 -isystem, and -idirafter. Copy-edit.
19296 * doc/cppopts.texi: Copy-edit. Remove contradiction about
19297 default for -ftrack-macro-expansion. Delete obsolete and
19298 badly-formatted implementation details about -fdebug-cpp output.
19299 * doc/cppwarnopts.texi: Copy-edit.
19301 2017-01-07 David Malcolm <dmalcolm@redhat.com>
19304 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
19305 that the transition from a max line width >= 1<<10 to narrower
19306 lines works correctly.
19308 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
19310 * doc/options.texi (PerFunction): New.
19311 * opt-functions.awk (switch_flags): Map both Optimization and
19312 PerFunction to CL_OPTIMIZATION.
19313 * opth-gen.awk: Test for PerFunction flag along with
19315 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
19316 it only when the latter is present. Skip those that don't in
19317 the hash function generator.
19318 * common.opt (fvar-tracking): Mark as PerFunction instead of
19320 (fvar-tracking-assignments): Likewise.
19321 (fvar-tracking-assignments-toggle): Likewise.
19322 (fvar-tracking-uninit): Likewise.
19324 2017-01-07 Jakub Jelinek <jakub@redhat.com>
19326 PR translation/79018
19327 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
19330 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
19333 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
19334 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
19335 TARGET_LONG_JUMP_TABLE_OFFSETS.
19336 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
19337 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
19338 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
19339 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
19340 * config/m68k/m68k.md (tablejump expander): Likewise.
19341 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
19342 TARGET_LONG_JUMP_TABLE_OFFSETS.
19343 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
19344 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
19346 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
19347 David Holsgrove <david.holsgrove@xilinx.com>
19349 * common/config/microblaze/microblaze-common.c
19350 (TARGET_EXCEPT_UNWIND_INFO): Remove.
19351 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
19353 * config/microblaze/microblaze.c (microblaze_must_save_register)
19354 (microblaze_expand_epilogue, microblaze_return_addr): Handle
19356 (microblaze_eh_return): New function.
19357 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
19358 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
19359 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
19360 * config/microblaze/microblaze.md (eh_return): New pattern.
19362 2017-01-06 Jakub Jelinek <jakub@redhat.com>
19364 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
19365 GCC_DIAGNOSTIC_STRINGIFY): Define.
19367 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
19369 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
19371 * config/arm/arm.md (<mcrr>): New.
19373 * config/arm/arm.c (arm_arch5te): New.
19374 (arm_option_override): Set arm_arch5te.
19375 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
19377 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
19378 (arm_mcrr_qualifiers): ... this. New.
19379 (MRRC_QUALIFIERS): Define to...
19380 (arm_mrrc_qualifiers): ... this. New.
19381 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
19383 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
19384 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
19385 (MRRCI, mrrc, MRRC): New.
19386 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
19387 VUNSPEC_MRRC2): New.
19389 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
19391 * config/arm/arm.md (<mcr>): New.
19393 * config/arm/arm.c (arm_coproc_builtin_available): Add
19394 support for mcr, mrc, mcr2 and mrc2.
19395 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
19396 (arm_mcr_qualifiers): ... this. New.
19397 (MRC_QUALIFIERS): Define to ...
19398 (arm_mrc_qualifiers): ... this. New.
19399 (MCR_QUALIFIERS): Define to ...
19400 (arm_mcr_qualifiers): ... this. New.
19401 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
19403 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
19404 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
19405 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
19406 VUNSPEC_MRC2): New.
19408 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
19410 * config/arm/arm.md (*ldc): New.
19414 * config/arm/arm.c (arm_coproc_builtin_available): Add
19415 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
19416 (arm_coproc_ldc_stc_legitimate_address): New.
19417 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
19418 'qualifier_const_pointer'.
19419 (LDC_QUALIFIERS): Define to...
19420 (arm_ldc_qualifiers): ... this. New.
19421 (STC_QUALIFIERS): Define to...
19422 (arm_stc_qualifiers): ... this. New.
19423 * config/arm/arm-protos.h
19424 (arm_coproc_ldc_stc_legitimate_address): New.
19425 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
19426 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
19427 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
19428 stc2, stcl, stc2l): New.
19429 * config/arm/constraints.md (Uz): New.
19430 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
19431 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
19432 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
19433 VUNSPEC_STC2L): New.
19435 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
19437 * config/arm/arm.md (<cdp>): New.
19438 * config/arm/arm.c (neon_const_bounds): Rename this ...
19439 (arm_const_bounds): ... this.
19440 (arm_coproc_builtin_available): New.
19441 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
19442 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
19443 (CDP_QUALIFIERS): Define to...
19444 (arm_cdp_qualifiers): ... this. New.
19446 (arm_expand_builtin_args): Add case for 6 arguments.
19447 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
19448 (arm_const_bounds): ... this.
19449 (arm_coproc_builtin_available): New.
19450 * config/arm/arm_acle.h (__arm_cdp): New.
19452 * config/arm/arm_acle_builtins.def (cdp): New.
19454 * config/arm/iterators.md (CDPI,CDP,cdp): New.
19455 * config/arm/neon.md: Rename all 'neon_const_bounds' to
19456 'arm_const_bounds'.
19457 * config/arm/types.md (coproc): New.
19458 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
19459 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
19460 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
19461 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
19463 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
19465 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
19466 (UBINOP_QUALIFIERS): New.
19468 (acle_builtin_data): New. Change comment.
19469 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
19470 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
19471 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
19472 arm_acle_builtins.def.
19473 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
19474 (arm_init_acle_builtins): New.
19475 (CRC32_BUILTIN): Remove.
19476 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
19477 crc32cb, crc32ch and crc32cw.
19478 (arm_init_crc32_builtins): Remove.
19479 (arm_init_builtins): Use arm_init_acle_builtins rather
19480 than arm_init_crc32_builtins.
19481 (arm_expand_acle_builtin): New.
19482 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
19483 * config/arm/arm_acle_builtins.def: New.
19485 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
19487 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
19488 (arm_builtin_datum): ... this.
19489 (arm_init_neon_builtin): Rename to ...
19490 (arm_init_builtin): ... this. Add a new parameters PREFIX
19491 and USE_SIG_IN_NAME.
19492 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
19493 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
19494 'arm_builtin_datum'.
19495 (arm_init_vfp_builtins): Likewise.
19496 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
19497 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
19498 (arm_expand_neon_args): Rename to ...
19499 (arm_expand_builtin_args): ... this. Rename builtin_arg
19500 enum values and differentiate between ARG_BUILTIN_MEMORY
19501 and ARG_BUILTIN_NEON_MEMORY.
19502 (arm_expand_neon_builtin_1): Rename to ...
19503 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
19504 values, arm_expand_builtin_args and add bool parameter NEON.
19505 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
19506 (arm_expand_vfp_builtin): Likewise.
19507 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
19509 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
19511 PR middle-end/77484
19512 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
19513 * predict.c (tree_estimate_probability_bb): Reverse direction of
19514 polymorphic call predictor.
19516 2017-01-06 David Malcolm <dmalcolm@redhat.com>
19518 * passes.c (execute_one_pass): Split out pass-skipping logic into...
19519 (determine_pass_name_match): ...this new function and...
19520 (should_skip_pass_p): ...this new function.
19522 2017-01-06 Nathan Sidwell <nathan@acm.org>
19524 * ipa-visibility.c (function_and_variable_visibility): Reformat
19525 comments and long lines. Remove extrneous if.
19526 * symtab.c (symtab_node::make_decl_local): Fix code format.
19527 (symtab_node::set_section_for_node): Fix comment typo.
19529 2017-01-06 Martin Liska <mliska@suse.cz>
19532 * lra-constraints.c: Rename invariant to lra_invariant.
19533 * predict.c (set_even_probabilities): Initialize e to NULL.
19535 2017-01-05 Martin Sebor <msebor@redhat.com>
19537 PR tree-optimization/78910
19538 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
19539 (format_integer): Correct off-by-one error in the handling
19540 of precision with negative numbers in signed conversions..
19542 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
19544 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
19546 2017-01-05 Jakub Jelinek <jakub@redhat.com>
19548 PR tree-optimization/71016
19549 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
19550 factor_out_conditional_conversion. Formatting fix.
19551 (factor_out_conditional_conversion): Add cond_stmt argument.
19552 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
19553 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
19556 2017-01-05 David Malcolm <dmalcolm@redhat.com>
19558 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
19559 read-rtl-function.o, and selftest-rtl.o.
19560 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
19561 (selftest::aarch64_test_loading_full_dump): New function.
19562 (selftest::aarch64_run_selftests): New function.
19563 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
19564 selftest::aarch64_run_selftests.
19565 * config/i386/i386.c
19566 (selftest::ix86_test_loading_dump_fragment_1): New function.
19567 (selftest::ix86_test_loading_call_insn): New function.
19568 (selftest::ix86_test_loading_full_dump): New function.
19569 (selftest::ix86_test_loading_unspec): New function.
19570 (selftest::ix86_run_selftests): Call the new functions.
19571 * emit-rtl.c (maybe_set_max_label_num): New function.
19572 * emit-rtl.h (maybe_set_max_label_num): New decl.
19573 * function.c (instantiate_decls): Guard call to
19574 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
19575 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
19577 * gensupport.c (gen_reader::gen_reader): Pass "false"
19578 for new "compact" param of rtx_reader.
19579 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
19580 rather than an empty string for NULL strings.
19581 * read-md.c: Potentially include config.h rather than bconfig.h.
19582 Wrap include of errors.h with #ifdef GENERATOR_FILE.
19583 (have_error): New global, copied from errors.c.
19584 (md_reader::read_name): Rename to...
19585 (md_reader::read_name_1): ...this, adding "out_loc" param,
19586 and converting "missing name or number" to returning false, rather
19588 (md_reader::read_name): Reimplement in terms of read_name_1.
19589 (md_reader::read_name_or_nil): New function.
19590 (md_reader::read_string): Handle "(nil)" by returning NULL.
19591 (md_reader::md_reader): Add new param "compact".
19592 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
19593 (md_reader::read_file): New method.
19594 * read-md.h (md_reader::md_reader): Add new param "compact".
19595 (md_reader::read_file): New method.
19596 (md_reader::is_compact): New accessor.
19597 (md_reader::read_name): Convert return type from void to file_location.
19598 (md_reader::read_name_or_nil): New decl.
19599 (md_reader::read_name_1): New decl.
19600 (md_reader::m_compact): New field.
19601 (noop_reader::noop_reader): Pass "false" for new "compact" param
19603 (rtx_reader::rtx_reader): Add new "compact" param.
19604 (rtx_reader::read_rtx_operand): Make virtual and convert return
19605 type from void to rtx.
19606 (rtx_reader::read_until): New decl.
19607 (rtx_reader::handle_any_trailing_information): New virtual function.
19608 (rtx_reader::postprocess): New virtual function.
19609 (rtx_reader::finalize_string): New virtual function.
19610 (rtx_reader::m_in_call_function_usage): New field.
19611 (rtx_reader::m_reuse_rtx_by_id): New field.
19612 * read-rtl-function.c: New file.
19613 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
19614 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
19615 (selftest::verify_three_block_rtl_cfg): New decl.
19616 * read-rtl-function.h: New file.
19617 * read-rtl.c: Potentially include config.h rather than bconfig.h.
19618 For host, include function.h, memmodel.h, and emit-rtl.h.
19619 (one_time_initialization): New function.
19620 (struct compact_insn_name): New struct.
19621 (compact_insn_names): New array.
19622 (find_code): Handle insn codes in compact dumps.
19623 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
19624 (bind_subst_iter_and_attr): Likewise.
19625 (add_condition_to_string): Likewise.
19626 (add_condition_to_rtx): Likewise.
19627 (apply_attribute_uses): Likewise.
19628 (add_current_iterators): Likewise.
19629 (apply_iterators): Likewise.
19630 (initialize_iterators): Guard usage of apply_subst_iterator with
19631 #ifdef GENERATOR_FILE.
19632 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
19633 (md_reader::read_mapping): Likewise.
19634 (add_define_attr_for_define_subst): Likewise.
19635 (add_define_subst_attr): Likewise.
19636 (read_subst_mapping): Likewise.
19637 (check_code_iterator): Likewise.
19638 (rtx_reader::read_rtx): Likewise. Move one-time initialization
19640 (one_time_initialization): New function.
19641 (rtx_reader::read_until): New method.
19642 (read_flags): New function.
19643 (parse_reg_note_name): New function.
19644 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
19645 Handle reuse_rtx ids.
19646 Wrap iterator lookup within #ifdef GENERATOR_FILE.
19647 Add parsing support for RTL dumps, mirroring the special-cases in
19648 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
19649 values, and calling handle_any_trailing_information.
19650 (rtx_reader::read_rtx_operand): Convert return type from void
19651 to rtx, returning return_rtx. Handle case 'e'. Call
19652 finalize_string on XSTR and XTMPL fields.
19653 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
19654 "(nil)" values were omitted. Call the postprocess vfunc on the
19656 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
19657 class ctor. Initialize m_in_call_function_usage. Call
19658 one_time_initialization.
19659 * rtl-tests.c (selftest::test_uncond_jump): Call
19660 set_new_first_and_last_insn.
19661 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
19662 * selftest-rtl.c: New file.
19663 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
19664 (selftest::get_insn_by_uid): New decl.
19665 * selftest-run-tests.c (selftest::run_tests): Call
19666 read_rtl_function_c_tests.
19667 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
19668 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
19671 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
19673 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
19674 operands in a special way. Assert that pos+len <= mode precision.
19676 2017-01-05 Jakub Jelinek <jakub@redhat.com>
19678 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
19679 3 argument Alias with unlimited for the negative form.
19680 (fno-vect-cost-model): Removed.
19682 2017-01-05 Martin Liska <mliska@suse.cz>
19684 * hsa-gen.c (gen_hsa_divmod): New function.
19685 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
19687 2017-01-05 Martin Liska <mliska@suse.cz>
19690 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
19693 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19695 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
19696 small constant length operands.
19698 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19700 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
19701 between loop iterations.
19703 2017-01-05 Martin Liska <mliska@suse.cz>
19706 * gimplify.c (gimplify_decl_expr): Compare to
19707 asan_poisoned_variables instread of checking flags.
19708 (gimplify_target_expr): Likewise.
19709 (gimplify_expr): Likewise.
19710 (gimplify_function_tree): Conditionally initialize
19711 asan_poisoned_variables.
19713 2017-01-04 Jeff Law <law@redhat.com>
19715 PR tree-optimizatin/78812
19716 * rtl.h (contains_mem_rtx_p): Prototype.
19717 * ifcvt.c (containts_mem_rtx_p): Move from here to...
19718 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
19719 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
19720 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
19721 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
19723 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
19725 * input.c (assert_char_at_range): Default-initialize actual_range.
19727 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
19729 * df-scan.c (df_ref_create_structure): Make regno unsigned,
19730 to match the caller.
19732 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
19734 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
19735 insns after final jump in test to emit dummy move.
19737 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
19739 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
19740 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
19742 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
19744 * multiple_target.c (create_dispatcher_calls): Init e_next.
19745 * tree-ssa-loop-split.c (split_loop): Init border.
19746 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
19749 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
19754 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
19755 (altivec_register_operand): Do not return true if the operand
19756 contains a SUBREG mixing SImode and SFmode.
19757 (vsx_register_operand): Likewise.
19758 (vsx_reg_sfsubreg_ok): New predicate.
19759 (vfloat_operand): Do not return true if the operand contains a
19760 SUBREG mixing SImode and SFmode.
19761 (vint_operand): Likewise.
19762 (vlogical_operand): Likewise.
19763 (gpc_reg_operand): Likewise.
19764 (int_reg_operand): Likewise.
19765 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
19766 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
19767 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
19769 (rs6000_emit_move_si_sf_subreg): New helper function.
19770 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
19771 fixup SUBREGs involving SImode and SFmode.
19772 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
19773 numbers for the new peephole2 optimization.
19774 (peephole2 for SFmode unions): New peephole2 to optimize cases in
19775 the GLIBC math library that do AND/IOR/XOR operations on single
19776 precision floating point.
19777 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
19778 target macros to say whether we need to avoid SUBREGs mixing
19780 (TARGET_ALLOW_SF_SUBREG): Likewise.
19781 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
19782 (UNSPEC_SI_FROM_SF): Likewise.
19783 (iorxor): Change spacing.
19784 (and_ior_xor): New iterator for AND, IOR, and XOR.
19785 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
19786 (movdi_from_sf_zero_ext): Likewise.
19787 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
19788 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
19789 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
19790 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
19791 (fms<mode>4): Likewise.
19792 (fnma<mode>4): Likewise.
19793 (fnms<mode>4): Likewise.
19794 (nfma<mode>4): Likewise.
19795 (nfms<mode>4): Likewise.
19797 2017-01-04 Marek Polacek <polacek@redhat.com>
19800 * doc/invoke.texi: Document -Wpointer-compare.
19802 2017-01-04 Jakub Jelinek <jakub@redhat.com>
19804 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
19807 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
19808 descriptions for -gdwarf-5 and emit them as uleb128 instead of
19811 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
19814 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
19815 documentation of the powerpc_popcntb_ok attribute.
19816 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
19817 code to issue warning messages if a requested CPU configuration is
19818 not supported by the binary (assembler and loader) toolchain.
19819 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
19820 made to define a built-in function that has been disabled.
19821 (paired_init_builtins): Add assertion to prevent ICE if attempt is
19822 made to define a built-in function that has been disabled.
19823 (altivec_init_builtins): Add comment explaining why definition
19824 of the DST built-in functions is not preceded by an assertion
19825 check. Add assertions to prevent ICE if attempts are made to
19826 define an altivec predicate or an abs* built-in function that has
19828 (htm_init_builtins): Add comment explaining why definition of the
19829 htm built-in functions is not preceded by an assertion check.
19831 2017-01-04 Jeff Law <law@redhat.com>
19833 PR tree-optimizatin/67955
19834 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
19835 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
19836 the points-to solution does not include pt_null. Use DECL_PT_UID
19839 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
19841 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
19842 Use gen_int_mode instead of gen_lopwart for const_int operands.
19844 2017-01-04 Jakub Jelinek <jakub@redhat.com>
19846 PR tree-optimization/71563
19847 * match.pd: Simplify X << Y into X if Y is known to be 0 or
19848 out of range value - has low bits known to be zero.
19850 2017-01-04 Alan Modra <amodra@gmail.com>
19852 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
19853 * configure: Regenerate.
19854 * config.in: Regenerate.
19856 2017-01-04 Jakub Jelinek <jakub@redhat.com>
19859 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
19860 a substring of the message, but strcmp with the whole message. Ifdef
19861 ENABLE_NLS, translate the message first using dgettext.
19863 2017-01-03 Jeff Law <law@redhat.com>
19865 PR tree-optimizatin/78856
19866 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
19867 (mark_threaded_blocks): Remove code to truncate thread paths that
19868 cross multiple loop headers. Instead invalidate the cached loop
19869 iteration information and handle case of a thread path walking
19870 into an irreducible region.
19872 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
19875 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
19876 assertions. Add support for doing the signbit if the IEEE 128-bit
19877 floating point value is in a GPR.
19878 * config/rs6000/rs6000.md (Fsignbit): Delete.
19879 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
19880 Update the length attribute if the value is in a GPR.
19881 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
19882 the sign or zero extension instruction, since the value is always 0/1.
19883 (signbit<mode>2_dm2): Delete using <Fsignbit>.
19886 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
19887 extracting SImode to a GPR register so that we can generate a
19888 store, limit the vector to be in a traditional Altivec register
19889 for the vextuwrx instruction.
19891 2017-01-03 Ian Lance Taylor <iant@google.com>
19893 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
19895 2017-01-03 Martin Sebor <msebor@redhat.com>
19897 PR tree-optimization/78696
19898 * gimple-ssa-sprintf.c (format_floating): Correct handling of
19899 precision. Use MPFR for %f for greater fidelity. Correct handling
19901 (pass_sprintf_length::compute_format_length): Set width and precision
19902 specified by asrerisk to void_node for vararg functions.
19903 (try_substitute_return_value): Adjust dump output.
19905 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
19907 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
19909 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
19911 * doc/invoke.texi (SPARC options): Document -mlra as the default.
19912 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
19913 -mlra/-mno-lra was passed to the compiler.
19915 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
19917 PR rtl-optimization/65618
19918 * emit-rtl.c (try_split): Move initialization of "before" and
19919 "after" to just before the call to emit_insn_after_setloc.
19921 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
19923 * doc/md.texi (Standard Names): Remove reference to Java frontend.
19925 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
19927 * dwarf2out.c (gen_enumeration_type_die): When
19928 -gno-strict-dwarf, add a DW_AT_encoding attribute.
19930 2017-01-03 Jakub Jelinek <jakub@redhat.com>
19932 PR tree-optimization/78965
19933 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
19934 Change first argument from const call_info & to call_info &. For %n
19935 set info.nowrite to false.
19937 PR middle-end/78901
19938 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
19939 possibly throwing calls.
19941 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
19942 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
19943 and fns handling, rather than in a separate case SSA_NAME.
19945 2017-01-02 Jeff Law <law@redhat.com>
19947 * config/darwin-driver.c (darwin_driver_init): Const-correctness
19948 fixes for first_period and second_period variables.
19950 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
19953 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
19954 (*insvqi_1): New insn pattern.
19955 (*insvqi_1_mem_rex64): Ditto.
19956 (*insvqi_2): Ditto.
19957 (*insvqi_3): Rename from *insvqi.
19959 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
19961 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
19963 * doc/cfg.texi (Edges): Remove reference to Java.
19964 (Maintaining the CFG): Ditto.
19966 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
19968 PR middle-end/77674
19969 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
19970 transparent aliases.
19972 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
19974 PR middle-end/77484
19975 * predict.def (PRED_CALL): Update hitrate.
19976 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
19977 * predict.c (tree_estimate_probability_bb): Split CALL predictor
19978 into direct/indirect/polymorphic variants.
19980 2017-01-01 Jakub Jelinek <jakub@redhat.com>
19982 Update copyright years.
19984 * gcc.c (process_command): Update copyright notice dates.
19985 * gcov-dump.c (print_version): Ditto.
19986 * gcov.c (print_version): Ditto.
19987 * gcov-tool.c (print_version): Ditto.
19988 * gengtype.c (create_file): Ditto.
19989 * doc/cpp.texi: Bump @copying's copyright year.
19990 * doc/cppinternals.texi: Ditto.
19991 * doc/gcc.texi: Ditto.
19992 * doc/gccint.texi: Ditto.
19993 * doc/gcov.texi: Ditto.
19994 * doc/install.texi: Ditto.
19995 * doc/invoke.texi: Ditto.
19997 Copyright (C) 2017 Free Software Foundation, Inc.
19999 Copying and distribution of this file, with or without modification,
20000 are permitted in any medium without royalty provided the copyright
20001 notice and this notice are preserved.