1 2017-10-23 Jakub Jelinek <jakub@redhat.com>
3 * common.opt (gcolumn-info): Enable by default.
4 * doc/invoke.texi (gcolumn-info): Document new default.
6 2017-10-23 Richard Biener <rguenther@suse.de>
8 PR tree-optimization/82672
9 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
10 Fold the stmt if we propagated into it.
12 2017-10-23 Richard Biener <rguenther@suse.de>
14 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
15 (bitmap_remove_expr_from_set): ... this. All callers call this
16 for non-constant values.
17 (bitmap_set_subtract): Rename to...
18 (bitmap_set_subtract_expressions): ... this. Adjust and
20 (bitmap_set_contains_value): Remove superfluous check.
21 (bitmap_set_replace_value): Inline into single caller ...
22 (bitmap_value_replace_in_set): ... here and simplify.
23 (dependent_clean): Merge into ...
24 (clean): ... this using an overload. Adjust.
25 (prune_clobbered_mems): Adjust.
26 (compute_antic_aux): Likewise.
27 (compute_partial_antic_aux): Likewise.
29 2017-10-23 Richard Biener <rguenther@suse.de>
31 PR tree-optimization/82129
33 2017-08-01 Richard Biener <rguenther@suse.de>
35 PR tree-optimization/81181
36 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
37 (compute_antic): ... end of iteration here.
39 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
41 * target.def (starting_frame_offset): New hook.
42 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
43 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
44 * doc/tm.texi.in: Regenerate.
45 * hooks.h (hook_hwi_void_0): Declare.
46 * hooks.c (hook_hwi_void_0): New function.
47 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
48 STARTING_FRAME_OFFSET.
49 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
50 * reload1.c (reload): Likewise.
51 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
52 instead of STARTING_FRAME_OFFSET.
53 * function.c (try_fit_stack_local): Likewise.
54 (assign_stack_local_1): Likewise
55 (instantiate_virtual_regs): Likewise.
56 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
57 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
58 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
59 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
60 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
61 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
62 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
63 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
64 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
65 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
66 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
67 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
68 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
69 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
70 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
71 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
72 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
73 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
74 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
75 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
76 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
77 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
78 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
79 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
80 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
81 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
82 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
83 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
84 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
85 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
86 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
87 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
88 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
89 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
90 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
91 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
92 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
93 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
94 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
95 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
96 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
97 * config/avr/avr.c (avr_starting_frame_offset): Make static and
98 return a HOST_WIDE_INT.
99 (avr_builtin_setjmp_frame_value): Use it instead of
100 STARTING_FRAME_OFFSET.
101 (TARGET_STARTING_FRAME_OFFSET): Redefine.
102 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
103 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
105 (TARGET_STARTING_FRAME_OFFSET): Redefine.
106 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
107 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
108 (TARGET_CONSTANT_ALIGNMENT): Redefine.
109 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
110 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
111 (TARGET_STARTING_FRAME_OFFSET): Redefine.
112 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
113 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
114 (TARGET_STARTING_FRAME_OFFSET): Redefine.
115 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
116 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
118 (TARGET_STARTING_FRAME_OFFSET): Redefine.
119 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
120 * config/mips/mips.c (mips_compute_frame_info): Refer to
121 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
122 (mips_starting_frame_offset): New function.
123 (TARGET_STARTING_FRAME_OFFSET): Redefine.
124 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
125 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
126 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
127 and return a HOST_WIDE_INT.
128 (TARGET_STARTING_FRAME_OFFSET): Redefine.
129 (mmix_initial_elimination_offset): Refer to
130 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
131 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
132 * config/pa/pa.c (pa_starting_frame_offset): New function.
133 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
134 (pa_expand_prologue): Likewise.
135 (TARGET_STARTING_FRAME_OFFSET): Redefine.
136 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
137 !FRAME_GROWS_DOWNWARD handling to...
138 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
139 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
140 !FRAME_GROWS_DOWNWARD handling to...
141 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
142 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
143 !FRAME_GROWS_DOWNWARD handling to...
144 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
145 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
147 (rs6000_starting_frame_offset): New function.
148 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
149 !FRAME_GROWS_DOWNWARD handling to...
150 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
151 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
152 !FRAME_GROWS_DOWNWARD handling to...
153 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
154 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
155 !FRAME_GROWS_DOWNWARD handling to...
156 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
157 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
158 (rs6000_starting_frame_offset): New function.
159 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
160 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
161 * config/vax/vax.c (vax_starting_frame_offset): New function.
162 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
163 (TARGET_STARTING_FRAME_OFFSET): Redefine.
164 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
165 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
166 (TARGET_STARTING_FRAME_OFFSET): Redefine.
167 * system.h (STARTING_FRAME_OFFSET): Poison.
169 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
171 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
172 SCALAR_TYPE_MODE instead of TYPE_MODE.
174 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
175 Alan Hayward <alan.hayward@arm.com>
176 David Sherwood <david.sherwood@arm.com>
178 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
180 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
181 Alan Hayward <alan.hayward@arm.com>
182 David Sherwood <david.sherwood@arm.com>
184 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
186 2017-10-23 Richard Biener <rguenther@suse.de>
188 PR tree-optimization/82129
189 * tree-ssa-pre.c (bitmap_set_and): Remove.
190 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
191 canonicalizing expressions in the set to those with lowest
192 ID rather than taking that from the first edge.
194 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
196 * combine.c (rtx_equal_for_field_assignment_p): Use
199 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
200 Alan Hayward <alan.hayward@arm.com>
201 David Sherwood <david.sherwood@arm.com>
203 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
204 to a SUBREG_PROMOTED_VAR.
206 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
207 Alan Hayward <alan.hayward@arm.com>
208 David Sherwood <david.sherwood@arm.com>
210 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
211 (expand_debug_source_expr): Likewise.
212 * combine.c (combine_simplify_rtx): Likewise.
213 * cse.c (fold_rtx): Likewise.
214 * optabs.c (expand_float): Likewise.
215 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
216 (simplify_binary_operation_1): Likewise.
218 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
219 Alan Hayward <alan.hayward@arm.com>
220 David Sherwood <david.sherwood@arm.com>
222 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
223 (record_promoted_value): Likewise.
224 * expr.c (expand_expr_real_2): Likewise.
225 * ree.c (update_reg_equal_equiv_notes): Likewise.
226 (combine_set_extension): Likewise.
227 * rtlanal.c (low_bitmask_len): Likewise.
228 * simplify-rtx.c (neg_const_int): Likewise.
229 (simplify_binary_operation_1): Likewise.
231 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
232 Alan Hayward <alan.hayward@arm.com>
233 David Sherwood <david.sherwood@arm.com>
235 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
236 * regcprop.c (maybe_mode_change): Likewise.
237 * reload1.c (alter_reg): Likewise.
239 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
241 * inchash.h (inchash::hash::add_wide_int): New function.
242 * lto-streamer-out.c (hash_tree): Use it.
244 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
246 * inchash.h (inchash::hash::add_wide_int): Rename to...
247 (inchash::hash::add_hwi): ...this.
248 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
249 (polymorphic_call_target_hasher::hash): Likewise.
250 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
251 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
252 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
253 * lto-streamer-out.c (hash_tree): Likewise.
254 * optc-save-gen.awk: Likewise.
255 * tree.c (add_expr): Likewise.
257 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
260 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
261 for ordered inequality comparisons even with TARGET_IEEE_FP.
263 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
266 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
267 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
268 Expand with cmp<dwi>_doubleword.
270 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
272 * extend.texi: Add x86 specific to 'nocf_check' attribute.
274 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
275 specific to -fcf-protection option.
277 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
279 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
280 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
281 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
282 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
283 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
284 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
285 (extra_objs): Add cet.o for Linux/x86 targets.
286 (tmake_file): Add i386/t-cet for Linux/x86 targets.
287 * config/i386/cet.c: New file.
288 * config/i386/cetintrin.h: Likewise.
289 * config/i386/t-cet: Likewise.
290 * config/i386/cpuid.h (bit_SHSTK): New.
292 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
293 pass IBT and SHSTK bits.
294 * config/i386/i386-builtin-types.def
295 (VOID_FTYPE_UNSIGNED_PVOID): New.
296 (VOID_FTYPE_UINT64_PVOID): Likewise.
297 * config/i386/i386-builtin.def: Add CET intrinsics.
298 * config/i386/i386-c.c (ix86_target_macros_internal): Add
299 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
300 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
301 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
303 * config/i386/i386.c (rest_of_insert_endbranch): New.
304 (pass_data_insert_endbranch): Likewise.
305 (pass_insert_endbranch): Likewise.
306 (make_pass_insert_endbranch): Likewise.
307 (ix86_notrack_prefixed_insn_p): Likewise.
308 (ix86_target_string): Add -mibt, -mshstk flags.
309 (ix86_option_override_internal): Add flag_cf_protection
311 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
312 (ix86_print_operand): Add 'notrack' prefix output.
313 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
314 (ix86_expand_builtin): Expand CET intrinsics.
315 (x86_output_mi_thunk): Add 'endbranch' instruction.
316 * config/i386/i386.h (TARGET_IBT): New.
317 (TARGET_IBT_P): Likewise.
318 (TARGET_SHSTK): Likewise.
319 (TARGET_SHSTK_P): Likewise.
320 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
321 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
322 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
323 (builtin_setjmp_setup): New pattern.
324 (builtin_longjmp): Likewise.
325 (rdssp<mode>): Likewise.
326 (incssp<mode>): Likewise.
327 (saveprevssp): Likewise.
328 (rstorssp): Likewise.
329 (wrss<mode>): Likewise.
330 (wruss<mode>): Likewise.
331 (setssbsy): Likewise.
332 (clrssbsy): Likewise.
333 (nop_endbr): Likewise.
334 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
336 * config/i386/immintrin.h: Include <cetintrin.h>.
337 * config/i386/linux-common.h
338 (file_end_indicate_exec_stack_and_cet): New prototype.
339 (TARGET_ASM_FILE_END): New.
341 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
343 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
344 latencies instead of having separate table; make difference between
345 integer and float costs.
346 * i386.h (processor_costs): Remove scalar_stmt_cost,
347 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
348 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
350 * x86-tune-costs.h: Remove entries which has been removed in
351 procesor_costs from all tables; make cond_taken_branch_cost
352 and cond_not_taken_branch_cost COST_N_INSNS based.
354 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
356 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
358 2017-10-20 Jakub Jelinek <jakub@redhat.com>
360 * config/i386/i386.md (isa): Remove fma_avx512f.
361 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
362 <avx512>_fmadd_<mode>_mask3<round_name>,
363 <avx512>_fmsub_<mode>_mask<round_name>,
364 <avx512>_fmsub_<mode>_mask3<round_name>,
365 <avx512>_fnmadd_<mode>_mask<round_name>,
366 <avx512>_fnmadd_<mode>_mask3<round_name>,
367 <avx512>_fnmsub_<mode>_mask<round_name>,
368 <avx512>_fnmsub_<mode>_mask3<round_name>,
369 <avx512>_fmaddsub_<mode>_mask<round_name>,
370 <avx512>_fmaddsub_<mode>_mask3<round_name>,
371 <avx512>_fmsubadd_<mode>_mask<round_name>,
372 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
373 (*vec_widen_umult_even_v16si<mask_name>,
374 *vec_widen_smult_even_v16si<mask_name>): Likewise.
375 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
377 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
379 * extend.texi: Add 'nocf_check' documentation.
380 * gimple.texi: Add second parameter to
381 gimple_build_call_from_tree.
382 * invoke.texi: Add -fcf-protection documentation.
383 * rtl.texi: Add REG_CALL_NOTRACK documenation.
385 2017-10-20 Richard Biener <rguenther@suse.de>
387 PR tree-optimization/82473
388 * tree-vect-loop.c (vectorizable_reduction): Properly get at
389 the largest input type.
391 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
393 * c-attribs.c (handle_nocf_check_attribute): New function.
394 (c_common_attribute_table): Add 'nocf_check' handling.
395 * gimple-parser.c: Add second argument NULL to
396 gimple_build_call_from_tree.
397 * attrib.c (comp_type_attributes): Check nocf_check attribute.
398 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
400 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
401 * common.opt: Add fcf-protection flag.
402 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
403 * flag-types.h: Add enum cf_protection_level.
404 * gimple.c (gimple_build_call_from_tree): Add second parameter.
405 Add 'nocf_check' attribute propagation to gimple call.
406 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
407 (gimple_build_call_from_tree): Update prototype.
408 (gimple_call_nocf_check_p): New function.
409 (gimple_call_set_nocf_check): Likewise.
410 * gimplify.c: Add second argument to gimple_build_call_from_tree.
411 * ipa-icf.c: Add nocf_check attribute in statement hash.
412 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
413 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
414 * toplev.c (process_options): Add flag_cf_protection handling.
416 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
418 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
420 2017-10-20 Richard Biener <rguenther@suse.de>
422 PR tree-optimization/82603
423 * tree-if-conv.c (predicate_mem_writes): Make sure to only
424 remove false predicated stores.
426 2017-10-20 Richard Biener <rguenther@suse.de>
428 * graphite-isl-ast-to-gimple.c
429 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
430 Remove return value and simplify, dump copied stmt after lhs
432 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
433 Reduce dump verbosity.
434 (gsi_insert_earliest): Likewise.
435 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
436 * graphite.c (print_global_statistics): Adjust dumping.
437 (print_graphite_scop_statistics): Likewise.
438 (print_graphite_statistics): Do not dump loops here.
439 (graphite_transform_loops): But here.
441 2017-10-20 Nicolas Roche <roche@adacore.com>
443 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
444 * configure: Regenerate.
446 2017-10-20 Jakub Jelinek <jakub@redhat.com>
449 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
450 functions when optimizing replace GIMPLE_RETURN stmts with
451 calls to __builtin_unreachable ().
454 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
455 for -fsanitize=thread link of executables.
456 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
460 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
462 (<shift_insn><mode>3<mask_name>): Change the last of the 3
463 define_insns for logical vector shifts to use VI248_AVX512BW
464 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
465 condition, useless isa and prefix attributes. Change the first
466 2 of these define_insns to ...
467 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
468 define_insn for avx512vl.
469 (<shift_insn><mode>3): ... and this, new define_insn without
470 masking for non-avx512vl.
473 * config/i386/sse.md (*andnot<mode>3,
474 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
475 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
476 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
477 not applied use empty suffix even for TARGET_AVX512VL.
478 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
479 is applied, supply evex,evex or evex,evex,evex instead of just
482 2017-10-20 Julia Koval <julia.koval@intel.com>
484 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
485 (OPTION_MASK_ISA_GFNI_UNSET): New.
486 (ix86_handle_option): Handle OPT_mgfni.
487 * config/i386/cpuid.h (bit_GFNI): New.
488 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
489 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
490 * config/i386/i386.c (ix86_target_string): Add -mgfni.
491 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
492 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
493 * config/i386/i386.opt: Add mgfni.
495 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
497 * config/msp430/msp430.c (msp430_option_override): Disable
498 -fdelete-null-pointer-checks.
499 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
501 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
503 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
504 of x87 and SSE instructions.
506 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
508 * asan.c (create_cond_insert_point): Do not update edge count.
509 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
510 (afdo_propagate_circuit): Likewise.
511 (afdo_calculate_branch_prob): Likewise.
512 (afdo_annotate_cfg): Likewise.
513 * basic-block.h (struct edge_def): Remove count.
514 (edge_def::count): New accessor.
515 * bb-reorder.c (rotate_loop): Update.
516 (find_traces_1_round): Update.
517 (connect_traces): Update.
518 (sanitize_hot_paths): Update.
519 * cfg.c (unchecked_make_edge): Update.
520 (make_single_succ_edge): Update.
521 (check_bb_profile): Update.
522 (dump_edge_info): Update.
523 (update_bb_profile_for_threading): Update.
524 (scale_bbs_frequencies_int): Update.
525 (scale_bbs_frequencies_gcov_type): Update.
526 (scale_bbs_frequencies_profile_count): Update.
527 (scale_bbs_frequencies): Update.
528 * cfganal.c (connect_infinite_loops_to_exit): Update.
529 * cfgbuild.c (compute_outgoing_frequencies): Update.
530 (find_many_sub_basic_blocks): Update.
531 * cfgcleanup.c (try_forward_edges): Update.
532 (try_crossjump_to_edge): Update
533 * cfgexpand.c (expand_gimple_cond): Update
534 (expand_gimple_tailcall): Update
535 (construct_exit_block): Update
536 * cfghooks.c (verify_flow_info): Update
537 (redirect_edge_succ_nodup): Update
539 (make_forwarder_block): Update
540 (duplicate_block): Update
541 (account_profile_record): Update
542 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
543 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
544 * cfgloopmanip.c (scale_loop_profile): Update.
546 (lv_adjust_loop_entry_edge): Update.
547 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
548 (force_nonfallthru_and_redirect): Update.
549 (purge_dead_edges): Update.
550 (rtl_flow_call_edges_add): Update.
551 * cgraphunit.c (init_lowered_empty_function): Update.
552 (cgraph_node::expand_thunk): Update.
553 * gimple-pretty-print.c (dump_probability): Update.
554 (dump_edge_probability): Update.
555 * gimple-ssa-isolate-paths.c (isolate_path): Update.
556 * haifa-sched.c (sched_create_recovery_edges): Update.
557 * hsa-gen.c (convert_switch_statements): Update.
558 * ifcvt.c (dead_or_predicable): Update.
559 * ipa-inline-transform.c (inline_transform): Update.
560 * ipa-split.c (split_function): Update.
561 * ipa-utils.c (ipa_merge_profiles): Update.
562 * loop-doloop.c (add_test): Update.
563 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
564 * lto-streamer-in.c (input_cfg): Update.
565 (input_function): Update.
566 * lto-streamer-out.c (output_cfg): Update.
567 * modulo-sched.c (sms_schedule): Update.
568 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
569 * predict.c (maybe_hot_edge_p): Update.
570 (unlikely_executed_edge_p): Update.
571 (probably_never_executed_edge_p): Update.
572 (dump_prediction): Update.
573 (drop_profile): Update.
574 (propagate_unlikely_bbs_forward): Update.
575 (determine_unlikely_bbs): Update.
576 (force_edge_cold): Update.
577 * profile.c (compute_branch_probabilities): Update.
578 * reg-stack.c (better_edge): Update.
579 * shrink-wrap.c (handle_simple_exit): Update.
580 * tracer.c (better_p): Update.
581 * trans-mem.c (expand_transaction): Update.
582 (split_bb_make_tm_edge): Update.
583 * tree-call-cdce.c: Update.
584 * tree-cfg.c (gimple_find_sub_bbs): Update.
585 (gimple_split_edge): Update.
586 (gimple_duplicate_sese_region): Update.
587 (gimple_duplicate_sese_tail): Update.
588 (gimple_flow_call_edges_add): Update.
589 (insert_cond_bb): Update.
590 (execute_fixup_cfg): Update.
591 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
592 * tree-complex.c (expand_complex_div_wide): Update.
593 * tree-eh.c (lower_resx): Update.
594 (unsplit_eh): Update.
595 (cleanup_empty_eh_move_lp): Update.
596 * tree-inline.c (copy_edges_for_bb): Update.
597 (freqs_to_counts): Update.
598 (copy_cfg_body): Update.
599 * tree-ssa-dce.c (remove_dead_stmt): Update.
600 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
601 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
602 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
603 (unloop_loops): Update.
604 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
605 * tree-ssa-loop-split.c (connect_loops): Update.
606 (split_loop): Update.
607 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
608 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
609 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
610 * tree-ssa-reassoc.c (branch_fixup): Update.
611 * tree-ssa-tail-merge.c (replace_block_by): Update.
612 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
613 (compute_path_counts): Update.
614 (update_profile): Update.
615 (recompute_probabilities): Update.
616 (update_joiner_offpath_counts): Update.
617 (estimated_freqs_path): Update.
618 (freqs_to_counts_path): Update.
619 (clear_counts_path): Update.
620 (ssa_fix_duplicate_block_edges): Update.
621 (duplicate_thread_path): Update.
622 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
623 (case_bit_test_cmp): Update.
624 (collect_switch_conv_info): Update.
625 (gen_inbound_check): Update.
626 (do_jump_if_equal): Update.
627 (emit_cmp_and_jump_insns): Update.
628 * tree-tailcall.c (decrease_profile): Update.
629 (eliminate_tail_call): Update.
630 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
631 (vect_do_peeling): Update.
632 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
633 * ubsan.c (ubsan_expand_null_ifn): Update.
634 (ubsan_expand_ptr_ifn): Update.
635 * value-prof.c (gimple_divmod_fixed_value): Update.
636 (gimple_mod_pow2): Update.
637 (gimple_mod_subtract): Update.
639 (gimple_stringop_fixed_value): Update.
641 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
644 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
646 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
648 PR rtl-optimization/82395
649 * ira-color.c (allocno_priority_compare_func): Fix comparison step
650 based on non_spilled_static_chain_regno_p.
652 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
654 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
655 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
656 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
658 2017-10-19 Martin Sebor <msebor@redhat.com>
660 PR tree-optimization/82596
661 * tree.c (array_at_struct_end_p): Handle STRING_CST.
663 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
665 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
666 (get_mem_refs_of_builtin_call): Likewise.
667 * builtins.c (expand_builtin_apply): Adjust call to
668 allocate_dynamic_stack_space.
669 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
670 the third argument to allocate_dynamic_stack_space, otherwise -1.
671 (expand_builtin): Deal with all alloca variants.
672 (is_inexpensive_builtin): Likewise.
673 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
674 * calls.c (special_function_p): Deal with all alloca variants.
675 (initialize_argument_information): Adjust call to
676 allocate_dynamic_stack_space.
677 (expand_call): Likewise.
678 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
679 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
680 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
681 use it for the stack usage computation.
682 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
683 * function.c (gimplify_parameters): Call build_alloca_call_expr.
684 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
685 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
686 (in_loop_p): Remove first argument and useless check.
687 (pass_walloca::execute): Remove useless test and adjust call to above.
688 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
689 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
690 (gimplify_call_expr): Deal with all alloca variants.
691 * hsa-gen.c (gen_hsa_alloca): Likewise.
692 (gen_hsa_insns_for_call): Likewise.
693 * ipa-pure-const.c (special_builtin_state): Likewise.
694 * tree-chkp.c (chkp_build_returned_bound): Likewise.
695 * tree-object-size.c (alloc_object_size): Likewise.
696 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
697 (call_may_clobber_ref_p_1): Likewise.
698 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
699 (ccp_fold_stmt): Likewise.
700 (optimize_stack_restore): Likewise.
701 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
702 (mark_all_reaching_defs_necessary_1): Likewise.
703 (propagate_necessity): Likewise.
704 (eliminate_unnecessary_stmts): Likewise.
705 * tree.c (build_common_builtin_nodes): Build
706 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
707 (build_alloca_call_expr): New function.
708 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
709 (CASE_BUILT_IN_ALLOCA): Likewise.
710 (build_alloca_call_expr): Declare.
711 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
713 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
716 * dwarf2out.c (new_die_raw): New static inline function.
717 (new_die): Use it to create the DIE.
718 (add_AT_external_die_ref): Likewise.
719 (clone_die): Likewise.
720 (clone_as_declaration): Likewise.
721 (dwarf2out_vms_debug_main_pointer): Likewise.
722 (base_type_die): Likewise. Remove early return for corner cases.
723 Do not call add_pubtype on the DIE here.
724 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
725 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
726 typedefs for base types with DW_AT_endianity. Make sure a DIE with
727 native order exists for base types, attach the DIE manually and call
728 add_pubtype on it. Do not equate a reverse order DIE to the type.
730 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
732 * config/arm/arm.c (align_ok_ldrd_strd): New function.
733 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
735 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
737 2017-10-19 Jakub Jelinek <jakub@redhat.com>
739 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
740 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
741 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
742 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
743 * opts.c (sanitizer_opts): Add builtin.
744 * ubsan.c (instrument_builtin): New function.
745 (pass_ubsan::execute): Call it.
746 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
747 * doc/invoke.texi: Document -fsanitize=builtin.
749 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
750 builtins, store max (log2 (align), 0) into uchar field instead of
751 align into uptr field.
752 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
753 store uchar 0 field instead of uptr 0 field.
754 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
755 instead of passing one address of struct with 2 locations pass
756 two addresses of structs with 1 location each.
757 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
758 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
759 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
760 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
761 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
762 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
763 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
764 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
766 2017-10-19 Martin Liska <mliska@suse.cz>
769 * file-find.c (remove_prefix): Remove.
770 * file-find.h (remove_prefix): Likewise.
771 * gcc-ar.c: Remove smartness of lookup.
773 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
775 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
776 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
777 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
779 2017-10-19 Jakub Jelinek <jakub@redhat.com>
782 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
783 (setcc + and to xor + setcc): New peephole2.
785 2017-10-19 Tom de Vries <tom@codesourcery.com>
787 * doc/sourcebuild.texi (Test Directives, Variants of
788 dg-require-support): Add dg-require-stack-size.
790 2017-10-19 Martin Liska <mliska@suse.cz>
793 * gimplify.c (gimplify_decl_expr): Do not instrument variables
794 that have a large alignment.
795 (gimplify_target_expr): Likewise.
797 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
799 PR rtl-optimization/82602
800 * ira.c (rtx_moveable_p): Return false for volatile asm.
802 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
805 * config/i386/i386-modes.def (CCGZ): New CC mode.
806 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
807 * config/i386/predicates.md (ix86_comparison_operator):
809 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
810 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
811 with double-word subtraction.
812 (put_condition_code): Handle CCGZmode.
814 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
816 * wide-int.cc (debug (const wide_int &)): New.
817 (debug (const wide_int *)): New.
818 (debug (const widest_int &)): New.
819 (debug (const widest_int *)): New.
821 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
824 * lra-constraints.c (curr_insn_transform): Use non-input operand
825 instead of output one for matched reload.
827 2017-10-18 Bin Cheng <bin.cheng@arm.com>
829 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
830 (tree-ssa-loop-ivopts.h): New header file.
831 (struct builtin_info): New fields.
832 (classify_builtin_1): Compute and record base and offset parts for
833 memset builtin partition by calling strip_offset.
834 (offset_cmp, fuse_memset_builtins): New functions.
835 (finalize_partitions): Fuse adjacent memset partitions by calling
837 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
838 Expose the interface.
839 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
841 2017-10-18 Bin Cheng <bin.cheng@arm.com>
843 PR tree-optimization/82574
844 * tree-loop-distribution.c (find_single_drs): New parameter. Check
845 that data reference must be executed exactly once per iteration
846 against the outermost loop in nest.
847 (classify_partition): Update call to above function.
849 2017-10-18 Richard Biener <rguenther@suse.de>
851 PR tree-optimization/82591
852 * graphite.c (graphite_transform_loops): Move code gen message
854 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
855 Here. Handle scop_to_isl_ast failing.
856 (scop_to_isl_ast): Limit the number of ISL operations.
858 2017-10-18 Richard Biener <rguenther@suse.de>
860 * graphite-isl-ast-to-gimple.c
861 (translate_isl_ast_to_gimple::set_rename): Simplify.
862 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
863 (graphite_copy_stmts_from_block): ... here.
864 (copy_bb_and_scalar_dependences): Simplify.
865 (add_parameters_to_ivs_params): Canonicalize.
866 (generate_entry_out_of_ssa_copies): Simplify.
867 * graphite-sese-to-poly.c (extract_affine_name): Simplify
868 by passing in ISL dimension.
869 (parameter_index_in_region_1): Rename to ...
870 (parameter_index_in_region): ... this.
871 (extract_affine): Adjust assert, pass down parameter index.
872 (add_param_constraints): Use range-info when available.
873 (build_scop_context): Adjust.
874 * sese.c (new_sese_info): Adjust.
875 (free_sese_info): Likewise.
876 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
877 Remove unused typedefs.
878 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
880 2017-10-18 Martin Liska <mliska@suse.cz>
882 * combine.c (simplify_compare_const): Add gcc_fallthrough.
884 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
886 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
887 (s390_sched_init): Do not reset s390_sched_state if we entered the
888 current basic block via a fallthru edge and all others are unlikely.
890 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
892 * config/s390/s390.c (NUM_SIDES): New variable.
893 (LONGRUNNING_THRESHOLD): New variable.
894 (LATENCY_FACTOR): New variable.
895 (s390_sched_score): Decrease score for long-running instructions on
897 (s390_sched_variable_issue): Perform bookkeeping for long-running
900 2017-10-18 Richard Biener <rguenther@suse.de>
902 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
903 Simplify with removal of the parameter rename map.
904 (set_rename): Likewise.
905 (should_copy_to_new_region): Likewise.
906 (graphite_copy_stmts_from_block): Likewise.
907 (copy_bb_and_scalar_dependences): Remove initialization of
908 unused copied_bb_map.
910 (copy_internal_parameters): Likewise.
911 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
912 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
914 (parameter_index_in_region_1): Rename to ...
915 (assign_parameter_index_in_region): ... this. Assert we have
916 a parameter we handle.
917 (scan_tree_for_params): Adjust.
918 * sese.h (parameter_rename_map_t): Remove.
919 (struct sese_info_t): Remove unused parameter_rename_map and
920 copied_bb_map members.
921 * sese.c (new_sese_info): Adjust.
922 (free_sese_info): Likewise.
924 2017-10-18 Martin Liska <mliska@suse.cz>
927 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
930 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
932 * doc/invoke.texi (ffunction-sections and fdata-sections):
935 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
937 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
938 the use statement can throw internally.
940 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
942 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
943 any RTX present on the RHS of a SET.
944 * compare-elim.c (try_eliminate_compare): Restore comment.
946 2017-10-17 Jakub Jelinek <jakub@redhat.com>
948 * langhooks.h (struct lang_hooks): Document that tree_size langhook
949 may be also called on tcc_type nodes.
950 * langhooks.c (lhd_tree_size): Likewise.
952 2017-10-17 David Malcolm <dmalcolm@redhat.com>
954 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
955 format_warning_at_substring.
956 (maybe_warn): Convert source_range * param to a location_t. Pass
957 UNKNOWN_LOCATION rather than NULL to fmtwarn.
958 (format_directive): Remove code to extract source_ranges and
959 source_range * in favor of just a location_t.
960 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
962 * substring-locations.c (format_warning_va): Convert
963 source_range * param to a location_t.
964 (format_warning_at_substring): Likewise.
965 * substring-locations.h (format_warning_va): Likewise.
966 (format_warning_at_substring): Likewise.
968 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
970 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
972 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
973 and scatter/gather ops.
975 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
976 vec_gather_load and vec_scatter_store.
977 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
978 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
979 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
980 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
981 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
982 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
984 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
986 * reg-stack.c (compare_for_stack_reg): Add bool argument.
987 Detect FTST instruction and handle its register pops. Only pop
988 second operand if can_pop_second_op is true.
989 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
990 set can_pop_second_op to false in the compare_for_stack_reg call.
992 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
993 output_fp_compare for stack register operands.
994 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
995 instructions here. Do not emit stack register pops here. Assert
996 that FCOMPP pops next to top stack register. Rewrite function.
998 2017-10-17 Nathan Sidwell <nathan@acm.org>
1001 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
1002 use DECL_ASSEMBLER_NAME_RAW.
1005 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
1008 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
1009 Wilco Dijkstra <wilco.dijkstra@arm.com>
1011 * builtins.c (expand_builtin_update_setjmp_buf): Add a
1012 converstion to Pmode from the buf_addr.
1014 2017-10-17 Richard Biener <rguenther@suse.de>
1016 * graphite-dependences.c (scop_get_reads_and_writes): Change
1017 output parameters to references.
1019 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
1021 PR 71026/tree-optimization
1022 * fold-const.c (distribute_real_division): Removed.
1023 (fold_binary_loc): Remove calls to distribute_real_divison.
1025 2017-10-17 Richard Biener <rguenther@suse.de>
1027 * graphite-scop-detection.c
1028 (scop_detection::stmt_has_simple_data_refs_p): Always use
1029 the full nest as region.
1030 (try_generate_gimple_bb): Likewise.
1031 * sese.c (scalar_evolution_in_region): Simplify now that
1032 SCEV can handle instantiation in regions.
1033 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
1034 in the non-loop part of a function if requested.
1036 2017-10-17 Richard Biener <rguenther@suse.de>
1038 PR tree-optimization/82563
1039 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
1041 (graphite_regenerate_ast_isl): Call it.
1042 * graphite-scop-detection.c (build_scops): Remove entry edge split.
1044 2017-10-17 Jakub Jelinek <jakub@redhat.com>
1046 PR tree-optimization/82549
1047 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
1048 Formatting fixes. Instead of calling make_bit_field_ref with negative
1051 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
1053 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
1054 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
1055 _mm_maskz_reduce_ss): New.
1056 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
1057 __builtin_ia32_reducess_mask): Ditto..
1058 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
1059 * config/i386/sse.md (reduces<mode>): Renamed to ...
1060 (reduces<mode><mask_scalar_name>): ... this.
1061 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}): Changed
1063 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
1064 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
1066 2017-10-16 David Malcolm <dmalcolm@redhat.com>
1068 * Makefile.in (OBJS): Add unique-ptr-tests.o.
1069 * selftest-run-tests.c (selftest::run_tests): Call
1070 selftest::unique_ptr_tests_cc_tests.
1071 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
1072 * unique-ptr-tests.cc: New file.
1074 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
1077 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
1079 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
1080 (make_hard_regno_born, make_hard_regno_dead): Update
1081 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
1083 2017-10-16 Jeff Law <law@redhat.com>
1085 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
1087 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
1089 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
1091 2017-10-16 Olivier Hainque <hainque@adacore.com>
1093 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
1094 with_cpu if we were configured for an e500v2 target cpu name.
1096 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
1098 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
1099 * doc/invoke.texi: Document +nodsp as a valid extension for
1102 2017-10-16 Martin Liska <mliska@suse.cz>
1104 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
1105 (test_set_range): Likewise.
1106 (test_range_functions): Rename to ...
1107 (test_bit_in_range): ... this.
1108 (sbitmap_c_tests): Add new test.
1110 2017-10-16 Tamar Christina <tamar.christina@arm.com>
1112 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32): New.
1113 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
1114 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
1116 2017-10-16 Tamar Christina <tamar.christina@arm.com>
1118 * config/aarch64/aarch64-builtins.c
1119 (aarch64_types_quadopu_lane_qualifiers): New.
1120 (TYPES_QUADOPU_LANE): New.
1121 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
1122 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
1123 (aarch64_<sur>dot_laneq<vsi2qi>): New.
1124 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
1125 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
1126 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
1127 (Vdottype, DOTPROD): New.
1128 (sur): Add SDOT and UDOT.
1130 2017-10-16 Tamar Christina <tamar.christina@arm.com>
1132 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
1133 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
1134 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Add TARGET_DOTPROD.
1135 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
1136 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75): Enable TARGET_DOTPROD.
1137 (cortex-a75.cortex-a55): Likewise.
1138 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
1140 2017-10-16 Tamar Christina <tamar.christina@arm.com>
1142 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
1143 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS): New.
1144 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane): new.
1145 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
1146 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
1147 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
1148 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
1149 * config/arm/types.md (neon_dot, neon_dot_q): New.
1150 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
1152 2017-10-16 Tamar Christina <tamar.christina@arm.com>
1154 * config/arm/arm.h (TARGET_DOTPROD): New.
1155 * config/arm/arm.c (arm_arch_dotprod): New.
1156 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
1157 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
1158 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
1159 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
1160 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
1161 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
1162 * doc/invoke.texi (armv8.2-a): Document dotprod
1164 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
1166 * i386.c (ix86_vec_cost): New function.
1167 (ix86_rtx_costs): Handle vector operations better.
1168 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
1169 * x86-tune-costs.h: Add new costs to all tables.
1171 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
1173 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
1175 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
1176 divsd, sqrtss and sqrtsd
1177 * x86-tune-costs.h: Add new entries to all costs.
1178 (znver1_cost): Fix to match real instruction latencies.
1180 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1181 Michael Collison <michael.collison@arm.com>
1183 * compare-elim.c: Include emit-rtl.h.
1184 (can_merge_compare_into_arith): New function.
1185 (try_validate_parallel): Likewise.
1186 (try_merge_compare): Likewise.
1187 (try_eliminate_compare): Call the above when no previous clobber
1189 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
1192 2017-10-14 Jakub Jelinek <jakub@redhat.com>
1196 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
1197 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
1201 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
1202 to be any operand_equal_p operands. For & (B - 1) require
1203 B to be power of 2. Recognize
1204 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
1206 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
1209 * optabs.c (expand_memory_blockage): Fix call of
1210 targetm.have_memory_blockage.
1212 2017-10-14 Jakub Jelinek <jakub@redhat.com>
1215 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
1216 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
1217 objects to extra_objs instead of overwriting it.
1219 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
1221 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
1222 Use any_fp_register_operand as operand[3] predicate. Simplify
1223 equality test for operands[2] and operands[4] memory location.
1224 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
1225 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
1226 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
1227 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
1228 any_fp_register_operand as operand[1] predicate. Simplify
1229 equality test for operands[0] and operands[3] memory location.
1230 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
1231 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
1232 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
1234 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
1236 * target-insns.def: Add memory_blockage.
1237 * optabs.c (expand_memory_blockage): New function.
1238 (expand_asm_memory_barrier): Rename ...
1239 (expand_asm_memory_blockage): ... to this.
1240 (expand_mem_thread_fence): Call expand_memory_blockage
1241 instead of expand_asm_memory_barrier.
1242 (expand_mem_singnal_fence): Ditto.
1243 (expand_atomic_load): Ditto.
1244 (expand_atomic_store): Ditto.
1245 * doc/md.texi (Standard Pattern Names For Generation):
1246 Document memory_blockage instruction pattern.
1248 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
1250 * config/rl78/rl78.c (rl78_emit_libcall): New function.
1251 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
1252 * config/rl78/rl78.md: New define_expand "adddi3".
1254 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
1256 * cfghooks.c (verify_flow_info): Disable check that all probabilities
1259 2017-10-13 Jeff Law <law@redhat.com>
1261 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
1263 2017-10-13 Jakub Jelinek <jakub@redhat.com>
1266 * internal-fn.c (expand_mul_overflow): If both operands have
1267 the same highpart of -1 or 0 and the topmost bit of lowpart
1268 is different, overflow is if res <= 0 rather than res < 0.
1270 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
1272 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
1273 TARGET_P9_VECTOR code for unaligned_load case.
1275 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
1277 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
1279 2017-10-13 Nathan Sidwell <nathan@acm.org>
1281 * tree-core.h (tree_contains_struct): Make bool.
1282 * tree.c (tree_contains_struct): Likewise.
1283 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
1284 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
1285 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
1286 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
1287 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
1289 2017-10-13 Richard Biener <rguenther@suse.de>
1291 * graphite-isl-ast-to-gimple.c
1292 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
1293 parameters and dominance check.
1294 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
1295 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
1296 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
1297 Do not update SSA form here or do intermediate IL verification.
1298 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
1299 (graphite_initialize): Remove check on the number of loops in
1300 the function and inline into graphite_transform_loops.
1301 (graphite_finalize): Inline into graphite_transform_loops.
1302 (graphite_transform_loops): Perform SSA update and IL verification
1304 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
1306 2017-10-13 Richard Biener <rguenther@suse.de>
1308 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
1309 graphite_expression_type_precision): Avoid global constructor
1311 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
1312 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
1313 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
1314 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
1315 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
1316 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
1318 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
1321 * config/i386/i386.h (ix86_red_zone_size): New.
1322 * config/i386/i386.md (push peephole2s): Replace
1323 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
1325 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
1326 Alan Hayward <alan.hayward@arm.com>
1327 David Sherwood <david.sherwood@arm.com>
1329 * combine.c (can_change_dest_mode): Reject changes in
1330 REGMODE_NATURAL_SIZE.
1332 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
1333 Alan Hayward <alan.hayward@arm.com>
1334 David Sherwood <david.sherwood@arm.com>
1336 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
1337 (expand_debug_source_expr): Likewise.
1338 * combine.c (combine_simplify_rtx): Likewise.
1339 * cse.c (fold_rtx): Likewise.
1340 * fwprop.c (canonicalize_address): Likewise.
1341 * targhooks.c (default_shift_truncation_mask): Likewise.
1343 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
1344 Alan Hayward <alan.hayward@arm.com>
1345 David Sherwood <david.sherwood@arm.com>
1347 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
1348 (widened_mode): Likewise.
1349 (expand_unop): Likewise.
1350 * ree.c (transform_ifelse): Likewise.
1351 (merge_def_and_ext): Likewise.
1352 (combine_reaching_defs): Likewise.
1353 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1355 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
1356 Alan Hayward <alan.hayward@arm.com>
1357 David Sherwood <david.sherwood@arm.com>
1359 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
1360 * combine.c (gen_lowpart_for_combine): Likewise.
1361 * dwarf2out.c (rtl_for_decl_location): Likewise.
1362 * final.c (alter_subreg): Likewise.
1363 * rtlhooks.c (gen_lowpart_general): Likewise.
1364 (gen_lowpart_if_possible): Likewise.
1366 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
1367 Alan Hayward <alan.hayward@arm.com>
1368 David Sherwood <david.sherwood@arm.com>
1370 * calls.c (expand_call): Use subreg_lowpart_offset.
1371 * cse.c (cse_insn): Likewise.
1372 * regcprop.c (copy_value): Likewise.
1373 (copyprop_hardreg_forward_1): Likewise.
1375 2017-10-13 Jakub Jelinek <jakub@redhat.com>
1378 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
1379 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
1380 =Q constraints to +Q and into insn condition add check
1381 that operands[0] and operands[1] are equal.
1382 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
1383 =Q constraints to +Q and into insn condition add check
1384 that operands[0] is equal to either operands[1] or operands[2].
1387 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
1388 instead of handling MINUS_EXPR twice (once for each argument),
1389 canonicalize operand order and handle just once, use rtype where
1390 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
1393 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
1394 any values of __C while still being pattern recognizable as a simple
1397 2017-10-13 Richard Biener <rguenther@suse.de>
1399 PR tree-optimization/82451
1401 2017-10-02 Richard Biener <rguenther@suse.de>
1403 PR tree-optimization/82355
1404 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
1405 a mapping for the enclosing loop but avoid generating one for
1407 (copy_bb_and_scalar_dependences): Remove premature codegen
1408 error on PHIs in blocks duplicated into multiple places.
1409 * graphite-scop-detection.c
1410 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
1411 in the region use it as loop and nest to analyze the DR in.
1412 (try_generate_gimple_bb): Likewise.
1413 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
1414 (add_loop_constraints): For blocks in a loop not in the region
1415 create a dimension with a single iteration.
1416 * sese.h (gbb_loop_at_index): Remove assert.
1418 * cfgloop.c (loop_preheader_edge): For the loop tree root
1419 return the single successor of the entry block.
1420 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
1421 Reset the SCEV hashtable and niters.
1422 * graphite-scop-detection.c
1423 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
1424 assert that we only have POLYNOMIAL_CHREC that vary in loops
1425 contained in the region.
1426 (scop_detection::graphite_can_represent_expr): Adjust.
1427 (scop_detection::stmt_has_simple_data_refs_p): For loops
1428 not in the region set loop to NULL. The nest is now the
1429 entry edge to the region.
1430 (try_generate_gimple_bb): Likewise.
1431 * sese.c (scalar_evolution_in_region): Adjust for
1432 instantiate_scev change.
1433 * tree-data-ref.h (graphite_find_data_references_in_stmt):
1434 Make nest parameter the edge into the region.
1435 (create_data_ref): Likewise.
1436 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
1437 entry edge into a region and adjust instantiate_scev calls.
1438 (create_data_ref): Likewise.
1439 (graphite_find_data_references_in_stmt): Likewise.
1440 (find_data_references_in_stmt): Pass the loop preheader edge
1441 from the nest argument.
1442 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
1443 parameter the edge into the region.
1444 (instantiate_parameters): Use the loop preheader edge as entry.
1445 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
1447 (get_instantiated_value_entry): Make instantiate_below parameter
1448 the edge into the region.
1449 (instantiate_scev_name): Likewise. Adjust dominance checks,
1450 when we cannot use loop-based instantiation instantiate by
1451 walking use-def chains.
1452 (instantiate_scev_poly): Adjust.
1453 (instantiate_scev_binary): Likewise.
1454 (instantiate_scev_convert): Likewise.
1455 (instantiate_scev_not): Likewise.
1456 (instantiate_array_ref): Remove.
1457 (instantiate_scev_3): Likewise.
1458 (instantiate_scev_2): Likewise.
1459 (instantiate_scev_1): Likewise.
1460 (instantiate_scev_r): Do not blindly handle N-operand trees.
1461 Do not instantiate array-refs. Handle all constants and invariants.
1462 (instantiate_scev): Make instantiate_below parameter
1463 the edge into the region.
1464 (resolve_mixers): Use the loop preheader edge for the region
1465 parameter to instantiate_scev_r.
1466 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
1468 2017-10-13 Richard Biener <rguenther@suse.de>
1470 PR tree-optimization/82525
1471 * graphite-isl-ast-to-gimple.c
1472 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
1474 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
1475 Fail code generation when we cannot represent the isl integer.
1476 (binary_op_to_tree): Elide modulo operations that are no-ops
1477 in the type we code generate. Remove now superfluous code
1480 2017-10-13 Richard Biener <rguenther@suse.de>
1482 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
1483 (scop_detection::harmful_loop_in_region): Remove premature
1484 IV type restriction.
1485 (scop_detection::graphite_can_represent_scev): We can handle
1486 pointer IVs just fine.
1488 2017-10-13 Alan Modra <amodra@gmail.com>
1490 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
1491 "Clobbers and Scratch Registers". Add paragraph on
1492 alternative to clobbers for scratch registers and OpenBLAS
1495 2017-10-13 Alan Modra <amodra@gmail.com>
1497 * doc/extend.texi (Clobbers): Correct vax example. Delete old
1498 example of a memory input for a string of known length. Move
1499 commentary out of table. Add a number of new examples
1500 covering array memory inputs.
1502 2017-10-12 Martin Liska <mliska@suse.cz>
1504 PR tree-optimization/82493
1505 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
1506 (test_range_functions): New function.
1507 (sbitmap_c_tests): Likewise.
1508 * selftest-run-tests.c (selftest::run_tests): Run new tests.
1509 * selftest.h (sbitmap_c_tests): New function.
1511 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
1513 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
1515 * config/rs6000/amo.h: Fix spacing issue.
1517 2017-10-12 Jakub Jelinek <jakub@redhat.com>
1520 * config/i386/i386.md (*ashl<mode>3_mask_1,
1521 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
1522 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
1525 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
1527 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
1528 (profile_probability): Set max_probability
1529 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
1531 * profile-count.c (profile_probability::differs_from_p): Do not
1532 rely on max_probaiblity == 10000
1534 2017-10-12 Jeff Law <law@redhat.com>
1536 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
1539 2017-10-12 Martin Sebor <msebor@redhat.com>
1543 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
1544 (handle_alias_pairs): Call it.
1545 * common.opt (-Wattribute-alias): New option.
1546 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
1547 * doc/invoke.texi (-Wattribute-alias): Document.
1549 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
1552 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
1554 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
1556 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
1557 (make_hard_regno_born, make_hard_regno_dead): Update
1558 bb_killed_pseudos and bb_gen_pseudos.
1560 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
1562 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
1564 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
1566 * config/alpha/alpha.c (alpha_split_conditional_move):
1567 Use std::swap instead of manually swapping.
1568 (alpha_stdarg_optimize_hook): Ditto.
1569 (alpha_canonicalize_comparison): Ditto.
1571 2017-10-12 Bin Cheng <bin.cheng@arm.com>
1573 * tree-loop-distribution.c (struct builtin_info): New struct.
1574 (struct partition): Refactor fields into struct builtin_info.
1575 (partition_free): Free struct builtin_info.
1576 (build_size_arg_loc, build_addr_arg_loc): Delete.
1577 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
1578 information from struct builtin_info.
1579 (find_single_drs): New function refactored from classify_partition.
1580 Also moved builtin validity checks to this function.
1581 (compute_access_range, alloc_builtin): New functions.
1582 (classify_builtin_st, classify_builtin_ldst): New functions.
1583 (classify_partition): Refactor code into functions find_single_drs,
1584 classify_builtin_st and classify_builtin_ldst.
1585 (distribute_loop): Don't do runtime alias check when distributing
1587 (find_seed_stmts_for_distribution): New function.
1588 (pass_loop_distribution::execute): Refactor code finding seed
1589 stmts into above function. Support distribution for the innermost
1590 two-level loop nest. Adjust dump information.
1592 2017-10-12 Bin Cheng <bin.cheng@arm.com>
1594 * tree-loop-distribution.c: Adjust the general comment.
1595 (NUM_PARTITION_THRESHOLD): New macro.
1596 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
1597 (classify_partition): Skip builtin pattern of loop nest's inner loop.
1598 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
1599 in call to build_partition_graph.
1600 (finalize_partitions): New parameter. Make loop distribution more
1601 conservative by fusing more partitions.
1602 (distribute_loop): Don't do runtime alias check in case of loop nest
1604 (find_seed_stmts_for_distribution): New function.
1605 (prepare_perfect_loop_nest): New function.
1606 (pass_loop_distribution::execute): Refactor code finding seed stmts
1607 and loop nest into above functions. Support loop nest distribution.
1608 Adjust dump information accordingly.
1610 2017-10-12 Bin Cheng <bin.cheng@arm.com>
1612 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
1613 and set PTYPE_SEQUENTIAL for merged partition.
1615 2017-10-12 Richard Biener <rguenther@suse.de>
1617 PR tree-optimization/69728
1619 2017-09-19 Richard Biener <rguenther@suse.de>
1621 PR tree-optimization/69728
1622 * graphite-sese-to-poly.c (schedule_error): New global.
1623 (add_loop_schedule): Handle empty domain by failing the
1625 (build_original_schedule): Handle schedule_error.
1627 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
1628 domain by returning an unchanged schedule.
1630 2017-10-12 Jakub Jelinek <jakub@redhat.com>
1632 * genrecog.c (validate_pattern): For VEC_SELECT verify that
1633 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
1635 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
1637 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
1640 2017-10-12 Jakub Jelinek <jakub@redhat.com>
1643 * expr.c (store_field): Don't optimize away bitsize == 0 store
1644 from CALL_EXPR with addressable return type.
1646 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
1648 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
1649 * config/rs6000/rs6000.md (sel): Delete mode attribute.
1650 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
1651 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
1652 TARGET_ISEL instead of TARGET_ISEL<sel>.
1654 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
1656 * config/rs6000/rs6000.c
1657 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
1659 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
1661 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
1663 (reg_or_cint_operand): Fix comment.
1664 (reg_or_zero_operand): New predicate.
1665 * config/rs6000/rs6000-protos.h (output_isel): Delete.
1666 * config/rs6000/rs6000.c (output_isel): Delete.
1667 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
1668 instead of reg_or_cint_operand. Output instruction directly (not via
1670 (isel_unsigned_<mode>): Ditto.
1671 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
1672 gpc_reg_operand. Add an instruction alternative for this. Output
1673 instruction directly.
1674 (*isel_reversed_unsigned_<mode>): Ditto.
1676 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
1678 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
1679 (TARGET_CANONICALIZE_COMPARISON): Define.
1681 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
1684 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
1685 Check whether the dest is REG before adding REG_EQUIV note.
1687 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
1690 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
1692 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
1693 (make_hard_regno_born, make_hard_regno_dead): Update
1694 bb_killed_pseudos and bb_gen_pseudos.
1696 2017-10-11 Nathan Sidwell <nathan@acm.org>
1698 * incpath.h (enum incpath_kind): Name enum, prefix values.
1699 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
1700 * incpath.c (heads, tails): Use INC_MAX.
1701 (add_env_var_paths, add_standard_paths): Use incpath_kind.
1702 (merge_include_chains, split_quote_chain,
1703 register_include_chains): Update incpath_kind names.
1704 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
1705 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
1707 (add_framework_path, darwin_register_objc_includes): Likewise.
1708 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
1710 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
1712 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
1713 Do not use float_operator operator predicate.
1714 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
1715 * config/i386/predicates.md (float_operator): Remove predicate.
1717 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
1719 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
1720 (*jccxf_i387): Ditto.
1721 (*jcc<mode>_i387): Ditto.
1722 (*jccu<mode>_i387): Ditto.
1723 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
1724 (*jcc_*_i387 splitters): Remove.
1725 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
1726 * config/i386/i386.c (ix86_split_fp_branch): Remove.
1727 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
1730 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
1732 * profile-count.h (slow_safe_scale_64bit): New function.
1733 (safe_scale_64bit): New inline.
1734 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
1735 * profile-count.c: Include wide-int.h
1736 (slow_safe_scale_64bit): New.
1738 2017-10-11 Nathan Sidwell <nathan@acm.org>
1740 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
1741 HAS_DECL_ASSEMBLER_NAME_P.
1742 * gimple-expr.c (gimple_decl_printable_name: Check
1743 HAS_DECL_ASSEMBLER_NAME_P too.
1744 * ipa-utils.h (type_in_anonymous_namespace_p): Check
1745 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
1746 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
1747 * passes.c (rest_of_decl_compilation): Check
1748 HAS_DECL_ASSEMBLER_NAME_P too.
1749 * recog.c (verify_changes): Likewise.
1750 * tree-pretty-print.c (dump_decl_name): Likewise.
1751 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
1753 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
1754 (SET_DECL_ASSEMBLER_NAME): Use it.
1755 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
1756 (COPY_DECL_ASSEMBLER_NAME): Likewise.
1757 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
1759 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
1761 * config.gcc (i386, x86_64): Add extra objects.
1762 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
1763 (ix86_min_insn_size): Declare.
1764 (ix86_issue_rate): Declare.
1765 (ix86_adjust_cost): Declare.
1766 (ia32_multipass_dfa_lookahead): Declare.
1767 (ix86_macro_fusion_p): Declare.
1768 (ix86_macro_fusion_pair_p): Declare.
1769 (ix86_bd_has_dispatch): Declare.
1770 (ix86_bd_do_dispatch): Declare.
1771 (ix86_core2i7_init_hooks): Declare.
1772 (ix86_atom_sched_reorder): Declare.
1773 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
1774 (COSTS_N_BYTES): Move to x86-tune-costs.h.
1775 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
1776 (rip_relative_addr_p): Rename to ...
1777 (ix86_rip_relative_addr_p): ... this one; export.
1778 (memory_address_length): Update.
1779 (ix86_issue_rate): Move to x86-tune-sched.c.
1780 (ix86_flags_dependent): Move to x86-tune-sched.c.
1781 (ix86_agi_dependent): Move to x86-tune-sched.c.
1782 (exact_dependency_1): Move to x86-tune-sched.c.
1783 (exact_store_load_dependency): Move to x86-tune-sched.c.
1784 (ix86_adjust_cost): Move to x86-tune-sched.c.
1785 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
1786 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
1787 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
1788 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
1789 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
1790 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
1791 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
1792 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
1793 (min_insn_size): Rename to ...
1794 (ix86_min_insn_size): ... this one; export.
1795 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
1796 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
1797 (core2i7_first_cycle_multipass_backtrack): Move to
1798 x86-tune-sched-core.c.
1799 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
1800 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
1801 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
1802 (ix86_avoid_jump_mispredicts): Update.
1803 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
1804 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
1805 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
1806 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
1807 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
1808 (MAX_INSN): Move to ix86-tune-sched-bd.c.
1809 (MAX_IMM): Move to ix86-tune-sched-bd.c.
1810 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
1811 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
1812 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
1813 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
1814 (MAX_STORE): Move to ix86-tune-sched-bd.c.
1815 (BIG): Move to ix86-tune-sched-bd.c.
1816 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
1817 (enum insn_path): Move to ix86-tune-sched-bd.c.
1818 (get_mem_group): Move to ix86-tune-sched-bd.c.
1819 (is_cmp): Move to ix86-tune-sched-bd.c.
1820 (dispatch_violation): Move to ix86-tune-sched-bd.c.
1821 (is_branch): Move to ix86-tune-sched-bd.c.
1822 (is_prefetch): Move to ix86-tune-sched-bd.c.
1823 (init_window): Move to ix86-tune-sched-bd.c.
1824 (allocate_window): Move to ix86-tune-sched-bd.c.
1825 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
1826 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
1827 (process_end_window): Move to ix86-tune-sched-bd.c.
1828 (allocate_next_window): Move to ix86-tune-sched-bd.c.
1829 (find_constant): Move to ix86-tune-sched-bd.c.
1830 (get_num_immediates): Move to ix86-tune-sched-bd.c.
1831 (has_immediate): Move to ix86-tune-sched-bd.c.
1832 (get_insn_path): Move to ix86-tune-sched-bd.c.
1833 (get_insn_group): Move to ix86-tune-sched-bd.c.
1834 (count_num_restricted): Move to ix86-tune-sched-bd.c.
1835 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
1836 (add_insn_window): Move to ix86-tune-sched-bd.c.
1837 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
1838 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
1839 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
1840 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
1841 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
1842 (do_dispatch): Move to ix86-tune-sched-bd.c.
1843 (has_dispatch): Move to ix86-tune-sched-bd.c.
1844 * config/i386/t-i386: Add new object files.
1845 * config/i386/x86-tune-costs.h: New file.
1846 * config/i386/x86-tune-sched-atom.c: New file.
1847 * config/i386/x86-tune-sched-bd.c: New file.
1848 * config/i386/x86-tune-sched-core.c: New file.
1849 * config/i386/x86-tune-sched.c: New file.
1851 2017-10-11 Liu Hao <lh_mouse@126.com>
1853 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
1854 the generic version below instead.
1855 (should_colorize): Recognize Windows consoles as terminals
1857 * pretty-print.c [__MINGW32__] (write_all): New function.
1858 [__MINGW32__] (find_esc_head): Likewise.
1859 [__MINGW32__] (find_esc_terminator): Likewise.
1860 [__MINGW32__] (eat_esc_sequence): Likewise.
1861 [__MINGW32__] (mingw_ansi_fputs): New function that handles
1863 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
1866 2017-10-11 Richard Biener <rguenther@suse.de>
1868 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
1869 Properly call analyze_scalar_evolution with the loop of the stmt.
1871 2017-10-11 Richard Biener <rguenther@suse.de>
1873 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
1874 * tree-core.h (tree_base): Add chrec_var union member.
1875 * tree.h (CHREC_VAR): Remove.
1876 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
1877 * tree-chrec.h (build_polynomial_chrec): Adjust.
1878 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
1879 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
1881 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
1883 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
1884 * match.pd: ... here.
1885 ((T) X == (T) Y): Relax condition.
1887 2017-10-11 Bin Cheng <bin.cheng@arm.com>
1889 PR tree-optimization/82472
1890 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
1892 (break_alias_scc_partitions): Update postorder number.
1894 2017-10-11 Martin Liska <mliska@suse.cz>
1897 * opts.c (parse_no_sanitize_attribute): Do not use error_value
1899 * opts.h (parse_no_sanitize_attribute): Remove last argument.
1901 2017-10-11 Martin Liska <mliska@suse.cz>
1903 * print-rtl.c (print_insn): Move declaration of idbuf
1904 to same scope as name.
1906 2017-10-11 Martin Liska <mliska@suse.cz>
1911 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
1914 2017-10-11 Martin Liska <mliska@suse.cz>
1917 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
1920 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
1922 * config.gcc (default_gnu_indirect_function): Default to yes for
1923 arm*-*-linux* with glibc.
1925 2017-10-11 Richard Biener <rguenther@suse.de>
1927 * tree-scalar-evolution.c (get_scalar_evolution): Handle
1928 default-defs and types we do not want to analyze.
1929 (interpret_loop_phi): Replace unreachable code with an assert.
1930 (compute_scalar_evolution_in_loop): Remove and inline ...
1931 (analyze_scalar_evolution_1): ... here, replacing condition with
1932 what makes the intent clearer. Remove handling of cases
1933 get_scalar_evolution now handles.
1935 2017-10-10 Jim Wilson <wilson@tuliptree.org>
1937 PR rtl-optimization/81434
1938 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
1939 comment for main loop. In sched_group_found if, also add checks for
1940 pass and min_cost_group.
1942 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
1944 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
1945 (rs6000_insn_cost): New function.
1946 * config/rs6000/rs6000.md (cost): New attribute.
1948 2017-10-10 Jakub Jelinek <jakub@redhat.com>
1949 H.J. Lu <hongjiu.lu@intel.com>
1953 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
1954 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
1955 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
1956 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
1957 to OPTION_MASK_ISA_AVX512VL - builtins that have both
1958 OPTION_MASK_ISA_MMX and some other bit set require both
1959 mmx and the ISAs without the mmx bit.
1960 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
1961 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
1962 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
1963 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
1964 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
1965 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
1966 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
1967 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
1968 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
1969 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
1970 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
1971 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
1972 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
1973 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
1974 Add OPTION_MASK_ISA_MMX.
1976 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
1978 * config.gcc (armv7*-*-freebsd*): New target.
1979 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
1981 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
1983 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
1984 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
1987 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
1989 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
1990 a template parameter.
1991 (WIDE_INT_REF_FOR): Update accordingly.
1992 * tree.h (wi::int_traits <const_tree>): Delete.
1993 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
1994 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
1995 (wi::tree_to_wide_ref): New typedef.
1996 (wi::to_wide): New function.
1997 * calls.c (get_size_range): Use wi::to_wide when operating on
1999 * cgraph.c (cgraph_node::create_thunk): Likewise.
2000 * config/i386/i386.c (ix86_data_alignment): Likewise.
2001 (ix86_local_alignment): Likewise.
2002 * dbxout.c (stabstr_O): Likewise.
2003 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
2004 * expr.c (const_vector_from_tree): Likewise.
2005 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
2006 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
2007 (fold_negate_expr_1, int_const_binop_1, const_binop)
2008 (fold_convert_const_int_from_real, optimize_bit_field_compare)
2009 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
2010 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
2011 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
2012 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
2013 (fold_not_const, round_up_loc): Likewise.
2014 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
2015 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
2016 (alloca_call_type): Likewise.
2017 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
2018 * godump.c (go_output_typedef): Likewise.
2019 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
2020 * internal-fn.c (get_min_precision): Likewise.
2021 * ipa-cp.c (ipcp_store_vr_results): Likewise.
2022 * ipa-polymorphic-call.c
2023 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
2024 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
2025 (ipa_modify_call_arguments): Likewise.
2026 * match.pd: Likewise.
2027 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
2028 * print-tree.c (print_node_brief, print_node): Likewise.
2029 * stmt.c (expand_case): Likewise.
2030 * stor-layout.c (layout_type): Likewise.
2031 * tree-affine.c (tree_to_aff_combination): Likewise.
2032 * tree-cfg.c (group_case_labels_stmt): Likewise.
2033 * tree-data-ref.c (dr_analyze_indices): Likewise.
2034 (prune_runtime_alias_test_list): Likewise.
2035 * tree-dump.c (dequeue_and_dump): Likewise.
2036 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
2037 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
2038 * tree-pretty-print.c (dump_generic_node): Likewise.
2039 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
2040 (simple_iv_with_niters): Likewise.
2041 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
2042 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
2043 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
2044 * tree-ssa-loop-niter.c (split_to_var_and_offset)
2045 (refine_value_range_using_guard, number_of_iterations_ne_max)
2046 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
2047 (get_cst_init_from_scev, record_nonwrapping_iv)
2048 (scev_var_range_cant_overflow): Likewise.
2049 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
2050 * tree-ssa-pre.c (compute_avail): Likewise.
2051 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
2052 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
2053 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
2054 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
2055 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
2056 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
2057 (dump_case_nodes, try_switch_expansion): Likewise.
2058 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
2059 (vect_do_peeling): Likewise.
2060 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
2061 * tree-vect-stmts.c (vectorizable_load): Likewise.
2062 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
2063 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
2064 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
2065 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
2066 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
2067 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
2068 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
2069 (evrp_dom_walker::before_dom_children): Likewise.
2070 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
2071 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
2072 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
2073 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
2074 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
2075 (get_range_pos_neg): Likewise.
2076 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
2077 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
2078 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
2079 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
2080 * config/avr/avr.c (avr_fold_builtin): Likewise.
2081 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
2082 * config/msp430/msp430.c (msp430_attr): Likewise.
2083 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
2084 * config/powerpcspe/powerpcspe-c.c
2085 (altivec_resolve_overloaded_builtin): Likewise.
2086 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
2087 (rs6000_expand_ternop_builtin): Likewise.
2088 * config/rs6000/rs6000-c.c
2089 (altivec_resolve_overloaded_builtin): Likewise.
2090 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
2091 (rs6000_expand_ternop_builtin): Likewise.
2092 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
2094 2017-10-10 Bin Cheng <bin.cheng@arm.com>
2096 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
2097 when copying loop nest with only one inner loop.
2099 2017-10-10 Richard Biener <rguenther@suse.de>
2101 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
2102 blocks if SCEV is active.
2103 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
2105 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
2106 (scev_initialize): Assert we are not yet initialized.
2108 2017-10-10 Bin Cheng <bin.cheng@arm.com>
2110 * tree-loop-distribution.c (generate_loops_for_partition): Remove
2111 inner loop's exit stmt by making it always exit the loop, otherwise
2112 we would generate an infinite empty loop.
2114 2017-10-10 Bin Cheng <bin.cheng@arm.com>
2116 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
2117 renaming variables in new preheader if it's deleted.
2119 2017-10-10 Bin Cheng <bin.cheng@arm.com>
2121 * tree-loop-distribution.c (struct partition): Remove unused field
2122 loops of the structure.
2123 (partition_alloc, partition_free): Ditto.
2124 (build_rdg_partition_for_vertex): Ditto.
2126 2017-10-09 Jeff Law <law@redhat.com>
2128 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
2129 return type to match prototype and documentation.
2131 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
2133 * config/rs6000/rs6000.c (processor_costs): Move to ...
2134 * config/rs6000/rs6000.h: ... here.
2135 (rs6000_cost): Declare.
2137 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
2139 * except.c (setjmp_fn): New global variable.
2140 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
2141 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
2142 if DONT_USE_BUILTIN_SETJMP is defined.
2144 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
2146 * target.def (insn_cost): New hook.
2147 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
2148 * doc/tm.texi: Regenerate.
2149 * rtlanal.c (insn_cost): Use the new hook.
2151 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
2153 * combine.c (combine_validate_cost): Compute the new insn_cost,
2154 not just pattern_cost.
2155 (try_combine): Adjust comment.
2157 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
2159 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
2161 * combine.c (uid_insn_cost): Adjust comment.
2162 (combine_validate_cost): Adjust comment. Use pattern_cost instead
2164 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
2165 * dse.c (find_shift_sequence): Ditto.
2166 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
2167 (bb_valid_for_noce_process_p): Use pattern_cost.
2168 * rtl.h (insn_rtx_cost): Delete.
2169 (pattern_cost): New prototype.
2170 (insn_cost): New prototype.
2171 * rtlanal.c (insn_rtx_cost): Rename to...
2172 (pattern_cost): ... this.
2175 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
2177 * config/i386/i386.md (*jcc_2): Remove insn pattern.
2178 (*jcc<mode>_0_r_i387): Ditto.
2179 (*jccxf_r_i387): Ditto.
2180 (*jcc<mode>_r_i387): Ditto.
2181 (*jccu<mode>_r_i387): Ditto.
2182 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
2183 (*jcc): Rename from *jcc_1.
2185 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2187 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
2188 deferred rescans after the lvx/stvx recombination pre-pass.
2190 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
2192 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
2193 memory operation instruction support.
2194 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
2195 (rs6000-ibm-aix[789]*): Likewise.
2196 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
2197 Document new functions.
2199 2017-10-09 Richard Biener <rguenther@suse.de>
2201 PR tree-optimization/82397
2202 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
2203 equality only for semantically equal trees.
2205 2017-10-09 Richard Biener <rguenther@suse.de>
2207 PR tree-optimization/82449
2208 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
2209 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
2210 allow constant addresses.
2211 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
2214 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2216 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
2219 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2222 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
2225 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2228 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
2230 2017-10-09 Jakub Jelinek <jakub@redhat.com>
2233 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
2234 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
2236 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
2238 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
2239 (WI_BINARY_PREDICATE_RESULT): Likewise.
2240 (wi::binary_traits::operator_result): New type.
2241 (wi::binary_traits::predicate_result): Likewise.
2242 (generic_wide_int::operator~, unary generic_wide_int::operator-)
2243 (generic_wide_int::operator==, generic_wide_int::operator!=)
2244 (generic_wide_int::operator&, generic_wide_int::and_not)
2245 (generic_wide_int::operator|, generic_wide_int::or_not)
2246 (generic_wide_int::operator^, generic_wide_int::operator+
2247 (binary generic_wide_int::operator-, generic_wide_int::operator*):
2249 (operator~, unary operator-, operator==, operator!=, operator&)
2250 (operator|, operator^, operator+, binary operator-, operator*): New
2252 * expr.c (get_inner_reference): Use wi::bit_and_not.
2253 * fold-const.c (fold_binary_loc): Likewise.
2254 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
2255 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
2256 (bit_value_binop): Likewise.
2257 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
2258 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
2259 (extract_range_from_binary_expr_1): Likewise.
2260 (masked_increment): Likewise.
2261 (simplify_bit_ops_using_ranges): Likewise.
2263 2017-10-09 Martin Jambor <mjambor@suse.cz>
2266 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
2267 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
2268 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
2270 (hsa_fixup_mov_insn_type): New function.
2271 (hsa_op_with_type::get_in_type): Use it.
2272 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
2273 immediates in an assert.
2274 (hsa_op_with_type::extend_int_to_32bit): New method.
2275 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
2276 types. Convert to dest type if necessary.
2277 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
2278 (reg_for_gimple_ssa): Pass false as min32int to
2279 hsa_type_for_scalar_tree_type.
2280 (gen_hsa_addr): Fixup type when creating addresable temporary.
2281 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
2282 (gen_hsa_unary_operation): Extend operands and convert to dest type if
2283 necessary. Call hsa_fixup_mov_insn_type.
2284 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
2285 extend operands and convert to dest type if necessary.
2286 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
2287 to dest type if necessary.
2288 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
2289 if conversion nt necessary and size matches.
2290 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
2291 to dest type if necessary.
2292 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
2293 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
2295 (gen_hsa_clrsb): Likewise.
2296 (gen_hsa_ffs): Likewise.
2297 (gen_hsa_divmod): Extend operands and convert to dest type if
2299 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
2301 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
2303 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
2304 Remove empty default arguments. Use a brace block as output
2306 (conditional return): Ditto.
2308 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
2309 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
2311 (group_ending_nop): Ditto.
2312 (doloop_end): Ditto.
2313 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
2314 (splitters for those): Ditto.
2316 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
2318 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
2319 a conditional jump (and the compare for it) so that pc_rtx is the
2321 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
2322 for the deleted and renamed ctr<mode>_internal[234] patterns.
2323 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
2324 Delete second conditional return pattern.
2325 (ctr<mode>_internal2): Delete this second bdnz pattern.
2326 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
2327 (ctr<mode>_internal4): Delete this second bdz pattern.
2329 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
2331 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
2332 (always_initialized_rtx_for_ssa_name_p): New predicate.
2333 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
2334 (finish_out_of_ssa): Free new field of SA.
2335 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
2336 * tree-ssa-coalesce.c: Include tree-ssa.h.
2337 (get_parm_default_def_partitions): Remove extern keyword.
2338 (get_undefined_value_partitions): New function.
2339 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
2340 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
2342 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
2344 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
2346 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
2348 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
2350 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
2351 for targets that preffer 128bit.
2353 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
2355 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
2357 2017-10-08 Olivier Hainque <hainque@adacore.com>
2359 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
2360 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
2362 (thumb_set_return_address): Likewise.
2364 2017-10-08 Olivier Hainque <hainque@adacore.com>
2366 * common/config/arm/arm-common.c (arm_except_unwind_info):
2367 Handle DWARF2_UNWIND_INFO.
2369 2017-10-07 Michael Collison <michael.collison@arm.com>
2371 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
2374 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
2376 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
2377 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
2378 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
2379 defined, force the creation of a new block for a dispatch label.
2381 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
2383 * invoke.texi (Wsuggest-attribute=cold): Document.
2384 * common.opt (Wsuggest-attribute=cold): New
2385 * ipa-pure-const.c (warn_function_cold): New function.
2386 * predict.c (compute_function_frequency): Use it.
2387 * predict.h (warn_function_cold): Declare.
2389 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
2391 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
2394 2017-10-06 Martin Liska <mliska@suse.cz>
2396 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
2397 keyword for member functions.
2398 (struct sanopt_tree_couple): New struct.
2399 (struct sanopt_tree_couple_hash): New function.
2400 (struct sanopt_ctx): Add new hash_map.
2401 (has_dominating_ubsan_ptr_check): New function.
2402 (record_ubsan_ptr_check_stmt): Likewise.
2403 (maybe_optimize_ubsan_ptr_ifn): Likewise.
2404 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
2405 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
2407 2017-10-06 Sudakshina Das <sudi.das@arm.com>
2410 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
2411 aarch64_simd_valid_immediate on CONST_VECTORs.
2412 (aarch64_reg_or_bic_imm): Likewise.
2414 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
2416 PR rtl-optimization/82396
2417 * haifa-sched.c (ready_sort_real): Disable qsort checking.
2419 2017-10-06 Sebastian Pop <sebpop@gmail.com>
2421 * graphite-dependences.c (scop_get_reads): Move code to...
2422 (scop_get_must_writes): Move code to...
2423 (scop_get_may_writes): Move code to...
2424 (scop_get_reads_and_writes): ... here.
2425 (scop_get_dependences): Call scop_get_reads_and_writes.
2427 2017-10-06 Jakub Jelinek <jakub@redhat.com>
2429 PR tree-optimization/82434
2430 * fold-const.h (can_native_encode_type_p,
2431 can_native_encode_string_p): Remove.
2432 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
2433 don't encode anything, just return what would be otherwise returned.
2434 (native_encode_fixed, native_encode_complex, native_encode_vector):
2436 (native_encode_string): Likewise. Inline by hand
2437 can_native_encode_string_p.
2438 (can_native_encode_type_p): Remove.
2439 (can_native_encode_string_p): Remove.
2440 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
2441 STRING_CSTs using can_native_encode_string_p, test all
2442 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
2443 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
2444 argument from native_encode_expr.
2445 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
2446 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
2449 2017-10-06 Richard Biener <rguenther@suse.de>
2451 PR tree-optimization/82397
2452 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
2453 operand_equal_p but rely on data_ref_compare_tree for detecting
2455 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
2456 to match up with dr_group_sort_cmp.
2458 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2461 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
2463 * config/s390/s390-builtin-types.def: Regenerate.
2465 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2468 * config/s390/s390-builtin-types.def: Regenerate.
2469 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
2470 Change flag from B_VXE to B_VX.
2471 (s390_vec_min_dbl): Remove B_VXE flag.
2473 2017-10-06 Richard Biener <rguenther@suse.de>
2475 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
2476 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
2477 translate_isl_ast_to_gimple::is_valid_rename,
2478 translate_isl_ast_to_gimple::get_rename,
2479 translate_isl_ast_to_gimple::get_def_bb_for_const,
2480 translate_isl_ast_to_gimple::get_new_name,
2481 translate_isl_ast_to_gimple::collect_all_ssa_names,
2482 translate_isl_ast_to_gimple::copy_loop_phi_args,
2483 translate_isl_ast_to_gimple::collect_all_ssa_names,
2484 translate_isl_ast_to_gimple::copy_loop_phi_args,
2485 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
2486 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
2487 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
2488 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
2489 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
2490 translate_isl_ast_to_gimple::copy_cond_phi_args,
2491 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
2492 translate_isl_ast_to_gimple::edge_for_new_close_phis,
2493 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
2494 translate_isl_ast_to_gimple::rename_uses,
2495 translate_isl_ast_to_gimple::rename_all_uses): Remove.
2496 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
2497 (set_rename_for_each_def): Likewise.
2498 (graphite_copy_stmts_from_block): Handle debug stmt resetting
2499 here. Handle rewriting SCEV analyzable uses here.
2500 (copy_bb_and_scalar_dependences): Generate code for PHI
2502 (graphite_regenerate_ast_isl): Adjust.
2503 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
2504 (add_write, add_read): New functions.
2505 (build_cross_bb_scalars_def): Use it and simplify.
2506 (build_cross_bb_scalars_use): Likewise.
2507 (graphite_find_cross_bb_scalar_vars): Inline into...
2508 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
2509 simulating out-of-SSA. Compute liveout and add dependencies.
2510 (build_scops): Force an empty entry block.
2511 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
2513 (sese_build_liveouts): Declare.
2514 (sese_trivially_empty_bb_p): Likewise.
2515 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
2516 compute liveout and debug_liveout.
2517 (sese_bad_liveouts_use): Remove.
2518 (sese_reset_debug_liveouts_bb): Likewise.
2519 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
2520 (sese_build_liveouts): Build liveout and debug_liveout and store
2522 (new_sese_info): Adjust.
2523 (free_sese_info): Likewise.
2524 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
2525 do not build liveout here.
2526 (move_sese_in_condition): Adjust region entry.
2527 (scev_analyzable_p): Match up with chrec_apply requirements.
2528 (sese_trivially_empty_bb_p): New.
2529 * tree-into-ssa.c (get_reaching_def): Properly support generating
2530 default-defs for incremental rewrite of anonymous names.
2532 2017-10-06 Richard Biener <rguenther@suse.de>
2534 * graphite-sese-to-poly.c (extract_affine): For casts increasing
2535 precision do not perform modulo reduction.
2537 2017-10-06 Richard Biener <rguenther@suse.de>
2539 PR tree-optimization/82436
2540 * tree-vect-slp.c (vect_supported_load_permutation_p): More
2541 conservatively choose the vectorization factor when checking
2542 whether we can perform the required load permutation.
2543 (vect_transform_slp_perm_load): Assert when we may not fail.
2545 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
2547 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
2548 message for incompatible -msdata=* and -mcall-* options.
2550 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
2552 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
2553 rate for post-reload scheduling.
2555 2017-10-05 Tamar Christina <tamar.christina@arm.com>
2557 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
2559 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
2561 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
2562 to improve monte carlo in scimark.
2564 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
2566 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
2567 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
2568 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
2569 pentium4_cost, nocona_cost): Set reassociation width to 1.
2570 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
2571 width to 2 for fp operations and 1 otherwise.
2572 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
2574 (atom_cost): Set reassociation width to 2.
2575 (slm_cost, generic_cost): Set fp reassociation width
2576 to 2 and 1 otherwise.
2577 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
2578 (core_cost): Set fp reassociation width to 4 and vector to 2.
2579 (ix86_reassociation_width): Rewrite using cost table; special case
2580 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
2581 and TARGET_AVX128_OPTIMAL.
2582 * config/i386/i386.h (processor_costs): Add
2583 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
2584 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
2585 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
2586 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
2587 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
2588 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
2590 2017-10-05 Nathan Sidwell <nathan@acm.org>
2592 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
2594 2017-10-05 Tamar Christina <tamar.christina@arm.com>
2596 * config/arm/arm.c (arm_test_fpu_data): New.
2597 (arm_run_selftests): Call arm_test_fpu_data.
2599 2017-10-04 Nathan Sidwell <nathan@acm.org>
2601 * toplev.c (toplev::main): Remove excess parens on pretty_printer
2603 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
2605 2017-10-04 Sudakshina Das <sudi.das@arm.com>
2607 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
2608 check type for aarch64_simd_valid_immediate.
2609 (aarch64_output_simd_mov_immediate): Update prototype.
2610 (aarch64_simd_valid_immediate): Update prototype.
2611 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
2612 support for ORR-immediate.
2613 (and<mode>3): modified pattern to add support for BIC-immediate.
2614 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
2615 now checks for valid immediate for BIC and ORR based on new enum
2617 (aarch64_output_simd_mov_immediate): Function now used to output
2618 BIC/ORR imm as well based on new enum argument.
2619 * config/aarch64/constraints.md (Do): New vector immediate constraint.
2621 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
2622 (aarch64_reg_or_bic_imm): Likewise.
2624 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2626 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
2627 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
2629 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
2633 PR rtl-optimization/82396
2634 * haifa-sched.c (autopref_multipass_init): Simplify
2636 (autopref_rank_data): Simplify sort order.
2637 * sched-int.h (autopref_multipass_data_): Remove
2638 multi_mem_insn_p, min_offset and max_offset.
2640 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2642 * doc/sourcebuild.texi: Document vect_peeling_profitable.
2644 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2646 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
2649 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2651 * doc/sourcebuild.texi: Document vect_long_mult.
2653 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
2655 PR tree-optimization/82413
2656 * fold-const.c (build_range_check): Use widest_int when comparing
2657 the maximum ETYPE value with HIGH.
2659 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
2661 PR rtl-optimization/82396
2662 * haifa-sched.c (autopref_multipass_init): Simplify
2664 (autopref_rank_data): Simplify sort order.
2665 * sched-int.h (autopref_multipass_data_): Remove
2666 multi_mem_insn_p, min_offset and max_offset.
2668 2017-10-04 Jakub Jelinek <jakub@redhat.com>
2670 PR tree-optimization/82381
2671 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
2672 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
2675 PR tree-optimization/82374
2676 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
2677 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
2678 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
2679 current_function_decl to the new decl.
2681 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
2683 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
2684 helper macro for IEEE float128 hardware built-in functions.
2685 (SQRTF128_ODD): Add built-in functions with the round-to-odd
2687 (TRUNCF128_ODD): Likewise.
2688 (ADDF128_ODD): Likewise.
2689 (SUBF128_ODD): Likewise.
2690 (MULF128_ODD): Likewise.
2691 (DIVF128_ODD): Likewise.
2692 (FMAF128_ODD): Likewise.
2693 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
2694 UNSPEC_TRUNC_ROUND_TO_ODD.
2695 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
2696 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
2697 floating point round to odd instructions.
2698 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
2699 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
2700 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
2701 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
2702 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
2703 (trunc<mode>sf2_hw): Change the truncate with round to odd
2704 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
2705 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
2706 to odd hardware instructions.
2707 (sub<mode>3_odd): Likewise.
2708 (mul<mode>3_odd): Likewise.
2709 (div<mode>3_odd): Likewise.
2710 (sqrt<mode>2_odd): Likewise.
2711 (fma<mode>4_odd): Likewise.
2712 (fms<mode>4_odd): Likewise.
2713 (nfma<mode>4_odd): Likewise.
2714 (nfms<mode>4_odd): Likewise.
2715 (trunc<mode>df2_odd): Change the truncate with round to odd
2716 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
2718 * doc/extend.texi (PowerPC built-in functions): Update documentation
2719 for existing IEEE float128-bit built-in functions. Add built-in
2720 functions that generate the IEEE 128-bit floating point round to
2723 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
2725 PR rtl-optimization/77729
2726 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
2727 to (X&(C1&~C2))|C2 transformations.
2729 2017-10-03 Martin Jambor <mjambor@suse.cz>
2731 PR tree-optimization/82363
2732 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
2733 mismatch, mark lacc written regardless of racc.
2735 2017-10-03 Jakub Jelinek <jakub@redhat.com>
2737 PR tree-optimization/82381
2738 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
2739 stmt_to_insert nor wheather SSA_NAMEs are default defs.
2740 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
2741 fallthrough into reassoc_stmt_dominates_stmt_p.
2744 * combine.c (combine_instructions): Don't combine in unreachable
2747 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
2750 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
2751 function to not use the have_cpu variable. Do not set cpu_index,
2752 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
2754 (rs6000_valid_attribute_p): Remove duplicate initializations of
2755 old_optimize and func_optimize.
2756 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
2757 (rs6000_activate_target_options): Make global.
2758 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
2761 2017-10-02 Jakub Jelinek <jakub@redhat.com>
2763 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
2764 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
2765 Set *poffset to 0 and *psize and *pmax_size to -1 if
2766 *poffset + *psize overflows in HOST_WIDE_INT.
2768 PR tree-optimization/82387
2769 PR tree-optimization/82388
2770 PR tree-optimization/82389
2771 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
2772 instead of live_bytes non-NULL.
2774 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
2777 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
2780 2017-10-02 Richard Biener <rguenther@suse.de>
2782 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
2783 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
2784 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
2786 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
2788 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
2789 requested precision matches the type's.
2790 * calls.c (alloc_max_size): Calculate the new candidate size as
2791 a widest_int and use wi::to_widest when comparing it with the
2792 current candidate size.
2793 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
2794 zero rather than integer_zero_node.
2795 * match.pd: Check for a no-op conversion before using wi::add
2796 rather than after. Use tree_to_uhwi when summing small shift
2797 counts into an unsigned int.
2799 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
2800 Alan Hayward <alan.hayward@arm.com>
2801 David Sherwood <david.sherwood@arm.com>
2804 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
2805 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
2806 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
2807 POINTER_AND_FP_REGS.
2809 2017-10-02 Richard Biener <rguenther@suse.de>
2811 PR tree-optimization/82355
2812 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
2813 a mapping for the enclosing loop but avoid generating one for
2815 (copy_bb_and_scalar_dependences): Remove premature codegen
2816 error on PHIs in blocks duplicated into multiple places.
2817 * graphite-scop-detection.c
2818 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
2819 in the region use it as loop and nest to analyze the DR in.
2820 (try_generate_gimple_bb): Likewise.
2821 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
2822 (add_loop_constraints): For blocks in a loop not in the region
2823 create a dimension with a single iteration.
2824 * sese.h (gbb_loop_at_index): Remove assert.
2826 2017-10-01 Kevin Buettner <kevinb@redhat.com>
2828 * omp-expand.c (adjust_context_scope): New function.
2829 (expand_parallel_call): Call adjust_context_scope.
2831 2017-10-01 Jeff Law <law@redhat.com>
2833 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
2834 dom_opt_dom_walker class with direct access to private members.
2835 Add comments. Call test_for_singularity.
2836 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
2837 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
2838 m_dummy_cond anymore.
2839 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
2841 (pass_dominator:execute): Build the dummy_cond here and pass it
2842 to the dom_opt_dom_walker ctor.
2843 (test_for_singularity): New function.
2845 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
2846 Maya Rashish <coypu@sdf.org>
2848 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
2849 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
2850 (alpha*-*-netbsd*) Use nbsd_tm_file.
2851 (arm*-*-netbsdelf*) Likewise.
2852 (i[34567]86-*-netbsdelf*) Likewise.
2853 (x86_64-*-netbsd*) Likewise.
2854 (mips*-*-netbsd*) Likewise.
2855 (powerpc-*-netbsd*) Likewise.
2856 (sh*-*-netbsd*) Likewise.
2857 (sparc-*-netbsdelf*) Likewise.
2858 (sparc64-*-netbsd*) Likewise.
2859 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
2861 (vax-*-netbsdelf*) Likewise.
2862 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
2863 (UINT_FAST8_TYPE) Likewise.
2864 (INT_FAST16_TYPE) Check CHAR_FAST16.
2865 (UINT_FAST16_TYPE) Likewise.
2867 2017-09-30 Jakub Jelinek <jakub@redhat.com>
2870 * config/i386/i386.md
2871 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
2872 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
2873 *divmodsi4_zext_2): New define_insn_and_split.
2874 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
2875 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
2876 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
2877 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
2878 New define_insn_and_split.
2879 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
2880 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
2881 operands[1] having DImode when mode is SImode.
2883 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
2884 always SImode for DIV and MOD in REG_EQUAL notes.
2886 2017-09-29 Yury Gribov <tetra2005@gmail.com>
2889 * match.pd: Fix handling of NaNs in pattern.
2891 2017-09-29 Jeff Law <law@redhat.com>
2893 * sbitmap.c (bitmap_bit_in_range_p): New function.
2894 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
2895 * tree-ssa-dse.c (live_bytes_read): New function.
2896 (dse_classify_store): Ignore reads of dead bytes.
2898 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
2899 typos and whitespace errors.
2900 * config/i386/predicates.md (address_no_seg_operand): Likewise.
2901 * config/s390/s390.c (s390_emit_prologue): Likewise.
2903 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
2906 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
2907 with a symbol for LRA.
2909 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
2911 PR rtl-optimization/82338
2912 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
2914 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
2916 * genmodes.c (calc_wider_mode): Suppress qsort macro.
2917 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
2918 (qsort_chk): Declare.
2919 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
2920 (qsort_chk): New function.
2922 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2924 PR tree-optimization/82337
2925 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
2926 phi definition if the PHI result appears in an abnormal PHI.
2927 (find_basis_for_base_expr): Don't record a basis if the LHS of the
2928 basis appears in an abnormal PHI.
2930 2017-09-29 Richard Biener <rguenther@suse.de>
2932 * graphite-isl-ast-to-gimple.c
2933 (translate_isl_ast_to_gimple::set_codegen_error): New function.
2934 (binary_op_to_tree): Use it.
2935 (get_rename_from_scev): Likewise.
2936 (copy_loop_phi_nodes): Likewise.
2937 (copy_bb_and_scalar_dependences): Likewise.
2938 (translate_pending_phi_nodes): Likewise.
2940 2017-09-29 Jakub Jelinek <jakub@redhat.com>
2943 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
2944 for movabsq $(i32 << shift), r64.
2946 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
2948 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
2949 index when encoding %esp as %rsp to avoid 0x67 prefix.
2951 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
2953 * config/i386/i386.md (*movsf_internal, *movdf_internal):
2954 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
2956 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
2958 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
2959 Extensions with more than 16 double VFP registers.
2960 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
2961 to_clear_mask and all code related to it. Replace the remaining
2962 entry by a sbitmap and adapt code accordingly.
2964 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
2966 * brig-builtins.def: Change pure attributes to const.
2968 2017-09-28 Joseph Myers <joseph@codesourcery.com>
2970 * config.gcc (default_gnu_indirect_function): Default to yes for
2971 sparc*-*-linux* with glibc.
2973 2017-09-28 Joseph Myers <joseph@codesourcery.com>
2975 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
2976 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
2977 when creating .init_array and .fini_array sections with priority
2980 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
2983 * config/aarch64/aarch64.c
2984 (aarch64_builtin_support_vector_misalignment): Always return false
2985 when misalignment is unknown.
2987 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
2989 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
2990 this function to return false if the definition used by the swap
2991 instruction is artificial, or if the memory address from which the
2992 constant value is loaded is not represented by a base address held
2993 in a register or if the base address register is a frame or stack
2994 pointer. Additionally, return false if the base address of the
2995 loaded constant is a SYMBOL_REF but is not considered to be a
2997 (replace_swapped_load_constant): New function.
2998 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
2999 loaded constant vector with a load of a swapped constant vector.
3001 2017-09-27 Carl Love <cel@us.ibm.com>
3003 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
3004 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
3005 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
3008 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
3010 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
3011 first, always call autopref_rank_data otherwise.
3013 2017-09-27 Richard Biener <rguenther@suse.de>
3015 * graphite-scop-detection.c (find_scop_parameters): Move
3016 loop bound handling ...
3017 (gather_bbs::before_dom_children): ... here, avoiding the need
3018 to build scop_info->loop_nest.
3019 (record_loop_in_sese): Remove.
3020 * sese.h (sese_info_t::loop_nest): Remove.
3021 * sese.c (new_sese_info): Do not allocate loop_nest.
3022 (free_sese_info): Do not free loop_nest.
3024 2017-09-27 Jakub Jelinek <jakub@redhat.com>
3027 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
3028 lhs from calls if the lhs has addressable type.
3030 2017-09-27 Richard Biener <rguenther@suse.de>
3032 * graphite.h (scop::max_alias_set): New member.
3033 * graphite-scop-detection.c: Remove references to non-existing
3034 --param in comments.
3035 (build_alias_sets): Record the maximum alias set used for drs.
3036 (build_scops): Support zero as unlimited for
3037 --param graphite-max-arrays-per-scop.
3038 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
3040 (build_poly_sr_1): ... here. Compute alias set based on the
3041 maximum alias set used for drs rather than
3042 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
3044 2017-09-27 Richard Biener <rguenther@suse.de>
3046 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
3047 --param loop-block-tile-size=0 to disable tiling.
3049 2017-09-27 Richard Biener <rguenther@suse.de>
3051 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
3052 (graphite-max-nb-scop-params): Document special value zero.
3053 * domwalk.h (dom_walker::STOP): New symbolical constant.
3054 (dom_walker::dom_walker): Add optional parameter for bb to
3056 (dom_walker::~dom_walker): Declare.
3057 (dom_walker::before_dom_children): Document STOP return value.
3058 (dom_walker::m_user_bb_to_rpo): New member.
3059 (dom_walker::m_bb_to_rpo): Likewise.
3060 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
3061 mapping here if not provided by the user.
3062 (dom_walker::~dom_walker): Free bb to RPO mapping if not
3063 provided by the user.
3064 (dom_walker::STOP): Define.
3065 (dom_walker::walk): Do not compute bb to RPO mapping here.
3066 Support STOP return value from before_dom_children to stop
3068 * graphite-optimize-isl.c (optimize_isl): If the schedule
3069 is the same still generate code if -fgraphite-identity
3070 or -floop-parallelize-all are given.
3071 * graphite-scop-detection.c: Include cfganal.h.
3072 (gather_bbs::gather_bbs): Get and pass through bb to RPO
3074 (gather_bbs::before_dom_children): Return STOP for BBs
3076 (build_scops): Compute bb to RPO mapping and pass it to
3077 the domwalk. Treat --param graphite-max-nb-scop-params=0
3078 as not limiting the number of params.
3079 * graphite.c (graphite_initialize): Remove limit on the
3080 number of basic-blocks in a function.
3081 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
3082 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
3083 default value of 10.
3085 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
3087 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
3088 Adjust code to eliminate needing to do the shift right 32-bits
3089 operation after XSCVDPSPN.
3091 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3093 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
3094 ((X / Y) != 0 -> X >= Y): Likewise.
3096 2017-09-26 Carl Love <cel@us.ibm.com>
3098 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
3099 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
3100 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
3101 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
3102 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
3103 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
3104 definitions and overloading.
3105 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
3106 statement for P9V_BUILTIN_XST_LEN_R.
3107 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
3108 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
3109 define_expand and define_insn for the instructions and builtins.
3110 * doc/extend.texi: Update the built-in documentation file for the new
3112 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
3113 define_insn for the instructions
3115 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
3118 * gcc/config/netbsd-protos.h: New file.
3119 * gcc/config/netbsd.c: New file.
3120 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
3121 * gcc/config/t-netbsd: New file.
3122 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
3123 (tmake_file) Add t-netbsd.
3124 (extra_objs) Add netbsd.o.
3126 2017-09-26 Janus Weil <janus@gcc.gnu.org>
3130 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
3132 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
3134 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
3135 sign extension from a vector register to a GPR by doing a 32-bit
3136 direct move and then an EXTSW.
3137 (extendsi<mode>2 splitter): Likewise.
3138 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
3139 right or vector extract after doing XSCVDPSPN. Use
3140 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
3142 (movdi_from_sf_zero_ext): Likewise.
3143 (reload_gpr_from_vsxsf): Likewise.
3144 (p8_mfvsrd_4_disf): Delete, no longer used.
3145 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
3146 then needing to move the SFmode to a GPR to use the XSCVDPSP
3147 instruction instead of FRSP and XSCVDPSPN.
3148 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
3149 it is adjacent to the other XSCVSPDP insns.
3150 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
3151 SFmode to be in traditional Altivec registers.
3152 (vsx_xscvdpspn): Eliminate useless alternative constraint.
3153 (vsx_xscvspdpn): Likewise.
3154 (vsx_xscvspdpn_scalar): Likewise.
3156 2017-09-26 Martin Jambor <mjambor@suse.cz>
3158 * tree-sra.c (compare_access_positions): Put integral types first,
3159 stabilize sorting of integral types, remove conditions putting
3160 non-full-precision integers last.
3161 (sort_and_splice_var_accesses): Disable scalarization if a
3162 non-integert would be represented by a non-full-precision integer.
3164 2017-09-26 Joseph Myers <joseph@codesourcery.com>
3166 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
3167 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
3168 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
3169 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
3170 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
3171 conditionals inside the function instead of around it. Call
3172 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
3173 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
3175 2017-09-26 Richard Biener <rguenther@suse.de>
3177 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
3179 (scop_detection::build_scop_breadth): ... this. Removed.
3180 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
3181 (scop_detection::harmful_stmt_in_bb): Likewise.
3182 (scop_detection::graphite_can_represent_stmt): Likewise.
3183 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
3184 (scop_detection::can_represent_loop): Remove recursion, fold in ...
3185 (scop_detection::can_represent_loop_1): ... this. Removed.
3186 (scop_detection::harmful_loop_in_region): Simplify after inlining
3187 the above and remove more quadraticness.
3188 (build_scops): Adjust.
3189 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
3192 2017-09-26 Jakub Jelinek <jakub@redhat.com>
3195 * config/i386/i386.c (ix86_print_operand_address_as): Only test
3196 REGNO (base) == SP_REG if base is a REG.
3199 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
3200 if it is different SSA_NAME.
3201 (optimize_range_tests_cmp_bitwise): New function.
3202 (optimize_range_tests): Call it.
3204 2017-09-26 Richard Biener <rguenther@suse.de>
3206 PR tree-optimization/82321
3207 * graphite.c (canonicalize_loop_closed_ssa): Properly check
3208 for the def being inside the loop.
3210 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3212 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
3214 * config/s390/s390-builtins.def: Fix constraint on op4.
3216 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3218 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
3219 independent expanders.
3220 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
3221 ("vec_ordered", "vec_unordered"): New expanders.
3223 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3225 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
3228 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3230 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
3231 vec_unpacks_lo_v16qi.
3232 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
3234 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3236 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
3237 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
3238 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
3240 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3242 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
3244 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
3246 ("*vec_slb<mode>"): New insn pattern.
3247 ("vec_shr_<mode>"): New expander.
3248 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
3249 and force the shift count operand to V16QImode.
3250 ("vec_srb<mode>"): Set shift count mode to V16QI.
3252 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3254 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
3255 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
3256 ("vec_widen_smult_hi_<mode>"): New expander definitions.
3258 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
3261 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
3263 2017-09-26 Richard Biener <rguenther@suse.de>
3265 PR tree-optimization/82320
3266 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
3269 2017-09-25 Jeff Law <law@redhat.com>
3271 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
3272 prototype for new argument.
3273 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
3274 mostly extracted from rs6000_emit_allocate_stack.
3275 (rs6000_emit_probe_stack_range_stack_clash): New function.
3276 (rs6000_emit_allocate_stack): Call
3277 rs6000_emit_probe_stack_range_stack_clash as needed.
3278 (rs6000_emit_probe_stack_range): Add additional argument
3279 to call to gen_probe_stack_range{si,di}.
3280 (output_probe_stack_range): New.
3281 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
3282 (output_probe_stack_range_stack_clash): New.
3283 (rs6000_emit_prologue): Emit notes into dump file as requested.
3284 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
3285 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
3286 Add additional operand and pass it to output_probe_stack_range.
3288 2017-09-25 Bin Cheng <bin.cheng@arm.com>
3290 PR tree-optimization/82163
3291 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
3292 (checking_verify_loop_closed_ssa): New parameter.
3293 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
3294 (check_loop_closed_ssa_stmt): Delete.
3295 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
3296 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
3297 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
3300 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
3302 * brig-builtins.def: Treat HSAIL barrier builtins as
3303 setjmp/longjump style functions.
3305 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
3307 * target.def (constant_alignment): New hook.
3308 * defaults.h (CONSTANT_ALIGNMENT): Delete.
3309 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
3310 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
3311 * doc/tm.texi: Regenerate.
3312 * targhooks.h (default_constant_alignment): Declare.
3313 (constant_alignment_word_strings): Likewise.
3314 * targhooks.c (default_constant_alignment): New function.
3315 (constant_alignment_word_strings): Likewise.
3316 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
3317 instead of CONSTANT_ALIGNMENT.
3318 * varasm.c (align_variable, get_variable_align, build_constant_desc)
3319 (force_const_mem): Likewise.
3320 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
3321 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
3322 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
3323 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3324 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
3326 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
3327 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3328 constant_alignment_word_strings.
3329 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
3330 (CONSTANT_ALIGNMENT): Likewise.
3331 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3332 (arm_constant_alignment): New function.
3333 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
3334 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3335 constant_alignment_word_strings.
3336 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
3337 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3338 constant_alignment_word_strings.
3339 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
3340 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3341 (cris_constant_alignment): New function.
3342 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
3343 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3344 (epiphany_constant_alignment): New function.
3345 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
3346 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3347 constant_alignment_word_strings.
3348 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
3349 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3350 constant_alignment_word_strings.
3351 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
3352 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3353 constant_alignment_word_strings.
3354 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
3355 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
3356 * config/i386/i386.c (ix86_constant_alignment): Make static.
3357 Use the same interface as the target hook.
3358 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3359 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
3360 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3361 constant_alignment_word_strings.
3362 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
3363 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
3364 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3365 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
3366 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3367 constant_alignment_word_strings.
3368 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
3369 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3370 constant_alignment_word_strings.
3371 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
3372 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3373 constant_alignment_word_strings.
3374 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
3375 * config/microblaze/microblaze.c (microblaze_constant_alignment):
3377 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3378 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
3379 * config/mips/mips.c (mips_constant_alignment): New function.
3380 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3381 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
3382 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
3383 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3384 (mmix_constant_alignment): Make static. Use the same interface
3386 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
3387 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3388 constant_alignment_word_strings.
3389 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
3390 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3391 constant_alignment_word_strings.
3392 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
3393 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3394 constant_alignment_word_strings.
3395 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
3396 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3397 (rs6000_constant_alignment): New function.
3398 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
3399 * config/riscv/riscv.c (riscv_constant_alignment): New function.
3400 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3401 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
3402 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3403 (rs6000_constant_alignment): New function.
3404 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
3405 * config/s390/s390.c (s390_constant_alignment): New function.
3406 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3407 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
3408 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3409 constant_alignment_word_strings.
3410 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
3411 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3412 (sparc_constant_alignment): New function.
3413 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
3414 * config/spu/spu.c (spu_constant_alignment): New function.
3415 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3416 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
3417 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3418 constant_alignment_word_strings.
3419 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
3420 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3421 constant_alignment_word_strings.
3422 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
3423 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3424 constant_alignment_word_strings.
3425 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
3426 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3427 (visium_constant_alignment): New function.
3428 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
3429 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3430 (xtensa_constant_alignment): New function.
3431 * system.h (CONSTANT_ALIGNMENT): Poison.
3433 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
3435 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
3436 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
3437 (rs6000_builtin_valid_without_lhs): New helper function.
3438 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3439 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
3441 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
3443 * target.h (vec_perm_indices): Use unsigned short rather than
3445 (auto_vec_perm_indices): Likewise.
3446 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
3447 Use unsigned int rather than unsigned char.
3448 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
3450 2017-09-25 Richard Biener <rguenther@suse.de>
3452 * cfgloop.h (sort_sibling_loops): Declare.
3453 * cfgloop.c (sort_sibling_loops_cmp): New helper.
3454 (sort_sibling_loops): New function sorting the sibling loop list
3456 * graphite.c (graphite_transform_loops): Sort sibling loops.
3458 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
3460 * target.def (vec_perm_const_ok): Change sel parameter to
3462 * optabs-query.c (can_vec_perm_p): Update accordingly.
3463 * doc/tm.texi: Regenerate.
3464 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
3465 auto_vec_perm_indices and remove separate nelt field.
3466 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
3467 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
3468 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
3469 (aarch64_expand_vec_perm_const): Update accordingly.
3470 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
3471 to vec_perm_indices.
3472 * config/arm/arm.c (expand_vec_perm_d): Change perm to
3473 auto_vec_perm_indices and remove separate nelt field.
3474 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
3475 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
3476 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
3478 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
3479 to vec_perm_indices.
3480 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
3481 sel to vec_perm_indices.
3482 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
3483 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
3484 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
3486 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
3489 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
3492 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
3493 on the FUNCTION_DECL function context if it has a DIE that is a
3496 2017-09-25 Richard Biener <rguenther@suse.de>
3498 PR tree-optimization/82285
3499 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
3502 2017-09-25 Tom de Vries <tom@codesourcery.com>
3506 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
3509 2017-09-25 Richard Biener <rguenther@suse.de>
3511 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
3512 ISL errors other than isl_error_quota happen. Dump if the
3513 schedule is the same.
3514 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
3515 errors instead of aborting inside ISL.
3517 2017-09-25 Iain Sandoe <iain@codesourcery.com>
3520 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
3521 of libgcc_eh for m64.
3522 * config/i386/darwin64.h: Likewise.
3524 2017-09-25 Richard Biener <rguenther@suse.de>
3527 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
3528 attribute for incomplete types nor twice for complete ones.
3530 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
3533 * config/i386/i386.c (ix86_print_operand_address_as): Encode
3534 %esp as %rsp to avoid 0x67 prefix if there is no index or base
3537 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
3540 * config/i386/i386.opt (mprefer-avx256): Use
3541 ix86_target_flags variable.
3542 * config/i386/i386.c (ix86_target_string): Move
3543 -mprefer-avx256 to flag2_opts.
3545 2017-09-22 Jakub Jelinek <jakub@redhat.com>
3548 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
3549 and x != -1 | y != -1 into (x & y) != -1.
3551 2017-09-22 Steve Ellcey <sellcey@cavium.com>
3553 * config.gcc: Add new case statement to set
3554 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
3555 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
3556 s390x-*-linux* case statements. Added aarch64 to the list of
3557 supported architectures.
3559 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
3561 PR tree-optimization/82289
3562 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
3563 STMT_VINFO_RELEVANT_P.
3565 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
3566 Alan Hayward <alan.hayward@arm.com>
3567 David Sherwood <david.sherwood@arm.com>
3569 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
3570 for VR_RANGE only; don't allow VR_ANTI_RANGE.
3571 (extract_range_from_binary_expr_1): Don't call
3572 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
3574 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
3575 Alan Hayward <alan.hayward@arm.com>
3576 David Sherwood <david.sherwood@arm.com>
3578 * target.def (preferred_vector_alignment): New hook.
3579 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
3581 * doc/tm.texi: Regenerate.
3582 * targhooks.h (default_preferred_vector_alignment): Declare.
3583 * targhooks.c (default_preferred_vector_alignment): New function.
3584 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
3586 (DR_TARGET_ALIGNMENT): New macro.
3587 (aligned_access_p): Update commentary.
3588 (vect_known_alignment_in_bytes): New function.
3589 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
3591 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
3592 Calculate the misalignment based on the target alignment rather than
3594 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
3595 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
3596 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
3597 the target alignment, rather than masking the element misalignment
3598 with the number of elements in a vector. Also use the target
3599 alignment when calculating the maximum number of peels.
3600 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
3601 instead of TYPE_ALIGN_UNIT.
3602 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
3603 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
3604 (vect_create_addr_base_for_vector_ref): Update call accordingly.
3605 (vect_create_data_ref_ptr): Likewise.
3606 (vect_setup_realignment): Realign by ANDing with
3607 -DR_TARGET_MISALIGNMENT.
3608 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
3609 the number of peels based on DR_TARGET_ALIGNMENT.
3610 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
3611 with the guaranteed alignment boundary when deciding whether
3613 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
3614 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
3615 (ensure_base_align): Remove stmt_info parameter. Get the
3616 target base alignment from DR_TARGET_ALIGNMENT.
3617 (vectorizable_store): Update call accordingly. Interpret
3618 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
3620 (vectorizable_load): Likewise.
3622 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
3623 Alan Hayward <alan.hayward@arm.com>
3624 David Sherwood <david.sherwood@arm.com>
3626 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
3627 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
3628 (vect_enhance_data_refs_alignment): Likewise.
3630 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
3632 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
3633 error. Only quit immediately if parsing is complete.
3634 (BEGIN): Initialize fatal_err and parse_done.
3635 (begin fpu, end fpu): Check number of arguments.
3636 (begin arch, end arch): Likewise.
3637 (begin cpu, end cpu): Likewise.
3638 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
3639 (optalias): Likewise.
3641 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
3643 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
3644 * config/arm/arm-isa.h: Delete. Move definitions to ...
3645 * arm-cpus.in: ... here. Use new feature and fgroup values.
3646 * config/arm/arm.c (arm_option_override): Use lower case for feature
3648 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
3649 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
3650 * config/arm/parsecpu.awk (END): Add new command 'isa'.
3652 (print_isa_bits_for): New function.
3653 (gen_isa): New function.
3654 (gen_comm_data): Use print_isa_bits_for.
3655 (define feature): New keyword.
3656 (define fgroup): New keyword.
3657 * config/arm/t-arm (TM_H): Remove.
3658 (GTM_H): Add arm-isa.h.
3659 (arm-isa.h): Add rule to generate file.
3660 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
3661 case for feature bit names.
3663 2017-09-22 Richard Biener <rguenther@suse.de>
3665 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
3667 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
3668 print of no dependency loops ...
3669 * graphite.c (graphite_transform_loops): ... here.
3670 (canonicalize_loop_closed_ssa_form): Work from inner to outer
3672 (same_close_phi_node, remove_duplicate_close_phi,
3673 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
3674 (canonicalize_loop_closed_ssa): ... here and simplify.
3675 * graphite-optimize-isl.c: Include tree-vectorizer.h.
3676 (optimize_isl): Use dump_printf_loc to tell when we stopped
3677 optimizing because of an ISL timeout.
3679 2017-09-22 Richard Biener <rguenther@suse.de>
3681 PR tree-optimization/82291
3682 * tree-if-conv.c (predicate_mem_writes): Make sure to
3683 remove writes in blocks predicated with false.
3685 2017-09-22 Richard Biener <rguenther@suse.de>
3687 * sese.c: Include cfganal.h.
3688 (if_region_set_false_region): Remove.
3689 (create_if_region_on_edge): Likewise.
3690 (move_sese_in_condition): Re-implement without destroying
3693 2017-09-22 Richard Biener <rguenther@suse.de>
3695 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
3696 Verify both BBs contain loop PHI nodes before dispatching to
3698 (graphite_regenerate_ast_isl): Do not recompute dominators,
3699 do not verify three times. Restructure for clarity.
3700 * graphite-scop-detection.c (same_close_phi_node,
3701 remove_duplicate_close_phi, make_close_phi_nodes_unique,
3702 defined_in_loop_p, canonicalize_loop_closed_ssa,
3703 canonicalize_loop_closed_ssa_form): Simplify, remove excess
3704 checking and SSA rewrite, move to ...
3705 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
3706 (graphite_initialize): Do not pass in ctx, do not reset the
3707 SCEV cache, compute only dominators.
3708 (graphite_transform_loops): Allocate ISL ctx after
3709 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
3710 Maintain post-dominators only around build_scops.
3711 * sese.c (if_region_set_false_region): Make static. Free
3712 and recompute dominators.
3713 (move_sese_in_condition): Assert we don't get called with
3714 post-dominators computed.
3715 * sese.h (if_region_set_false_region): Remove.
3717 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
3719 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
3720 mode attribute for TARGET_AVX512VL.
3722 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
3724 * config/i386/i386.opt (mprefer-avx256): New option.
3725 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
3727 (ix86_preferred_simd_mode): Return 256-bit AVX modes
3728 for TARGET_PREFER_AVX256.
3729 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
3731 2017-09-21 Jeff Law <law@redhat.com>
3733 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
3734 Fix dump output if the only stack space is for pushed registers.
3736 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3738 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
3741 2017-09-21 Martin Sebor <msebor@redhat.com>
3744 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
3745 code (in C++) or code that triggers warnings.
3747 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
3749 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
3751 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
3753 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
3754 * sched-rgn.c: Ditto.
3755 * sel-sched-ir.c: Ditto.
3757 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
3759 * toplev.h (set_random_seed): Adjust return type.
3760 * toplev.c (init_local_tick): Move eager initialization of random_seed
3761 to get_random_seed. Adjust comment.
3762 (init_random_seed): Inline to get_random_seed, delete.
3763 (get_random_seed): Initialize random_seed lazily.
3764 (set_random_seed): Do not return previous value.
3765 (print_switch_value): Do not call get_random_seed.
3767 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
3769 * cgraph.c (delete_function_version): New, broken out from...
3770 (cgraph_node::delete_function_version): ...here. Rename to
3771 cgraph_node::delete_function_version_by_decl. Update all uses.
3772 (cgraph_node::remove): Call delete_function_version.
3774 2017-09-21 Jakub Jelinek <jakub@redhat.com>
3777 * tree-inline.c (expand_call_inline): Emit clobber stmts for
3778 VAR_DECLs to which addressable non-volatile parameters are mapped
3779 and for id->retvar after the return value assignment. Clear
3780 id->retval and id->retbnd after inlining.
3782 2017-09-21 Richard Biener <rguenther@suse.de>
3784 PR tree-optimization/82276
3785 PR tree-optimization/82244
3786 * tree-vrp.c (build_assert_expr_for): Set
3787 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
3789 (remove_range_assertions): Revert earlier change.
3791 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
3794 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
3796 2017-09-21 Richard Biener <rguenther@suse.de>
3798 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
3799 Restore valid IL after code generation errors.
3800 * graphite.c (graphite_transform_loops): Diagnose code
3801 generation issues as MSG_MISSED_OPTIMIZATION and continue
3802 with processing SCOPs.
3804 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3805 Alan Hayward <alan.hayward@arm.com>
3806 David Sherwood <david.sherwood@arm.com>
3808 * calls.c (compute_argument_addresses): Use simplify_gen_binary
3809 rather than choosing between plus_constant and gen_rtx_<CODE>.
3810 * expr.c (emit_push_insn): Likewise.
3811 (expand_expr_real_2): Likewise.
3813 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3814 Alan Hayward <alan.hayward@arm.com>
3815 David Sherwood <david.sherwood@arm.com>
3817 * loop-unroll.c (split_iv): Call copy_rtx on the step.
3819 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3820 Alan Hayward <alan.hayward@arm.com>
3821 David Sherwood <david.sherwood@arm.com>
3823 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
3824 calling tree_to_uhwi.
3826 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3827 Alan Hayward <alan.hayward@arm.com>
3828 David Sherwood <david.sherwood@arm.com>
3830 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
3831 INTEGER_CST rather than a negative test for ADDR_EXPR.
3833 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
3834 Alan Hayward <alan.hayward@arm.com>
3835 David Sherwood <david.sherwood@arm.com>
3837 * tree-vrp.c (extract_range_from_binary_expr_1): Check
3838 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
3840 2017-09-21 Richard Biener <rguenther@suse.de>
3842 PR tree-optimization/71351
3843 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
3844 graphite_create_new_loop_guard): Remove, fold remaining parts
3846 (translate_isl_ast_node_for): ... here and simplify.
3848 2017-09-21 Jakub Jelinek <jakub@redhat.com>
3851 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
3852 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
3853 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
3854 alternative always use QI mode, for -Os imov (=R,R) alternative
3855 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
3858 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3859 Jeff Law <law@redhat.com>
3861 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
3862 (allocate_stack_space): New function, partially extracted from
3864 (s390_emit_prologue): Track offset to most recent stack probe.
3865 Code to allocate space moved into allocate_stack_space.
3866 Dump actions when no stack is allocated.
3867 (s390_prologue_plus_offset): New function.
3868 (s390_emit_stack_probe): Likewise.
3870 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
3872 * common.opt (Wa, Wl, Wp, g, gz=): Add
3874 (gno-column-info): Remove.
3875 (gcolumn-info): Drop RejectNegative.
3877 (gno-record-gcc-switches): Remove.
3878 (grecord-gcc-switches): Drop RejectNegative.
3879 (gno-split-dwarf): Remove.
3880 (gsplit-dwarf): Drop RejectNegative.
3881 (gno-strict-dwarf): Remove.
3882 (gstrict-dwarf): Drop RejectNegative.
3883 * config/darwin.opt (gfull, gused): Add RejectNegative.
3884 * dwarf2out.c (gen_producer_string): Drop
3885 gno-record-gcc-switches handler.
3886 * optc-gen.awk: Add g to prefixes with negative forms.
3887 * opts-common.c (remapping_prefix_p): New.
3888 (find_opt): Check it.
3889 (generate_canonical_option): Test g prefix.
3890 (option_map): Add -gno- mapping.
3891 (add_misspelling_candidates): Check remapping_prefix_p.
3893 2017-09-20 Jeff Law <law@redhat.com>
3895 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
3896 thinko in stack clash protection support.
3898 * explow.c (compute_stack_clash_protection_loop_data): Use
3899 CONST_INT_P instead of explicit test. Verify object is a
3900 CONST_INT_P before looking at INTVAL.
3901 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
3902 instead of explicit test.
3904 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
3907 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
3908 address instead of to r1 and r11.
3910 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
3912 * config.gcc: Support "knm".
3913 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
3914 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
3916 * config/i386/i386.c (m_KNM): Define.
3917 (processor_target_table): Add "knm".
3919 (ix86_option_override_internal): Add "knm".
3920 (ix86_issue_rate): Add PROCESSOR_KNM.
3921 (ix86_adjust_cost): Ditto.
3922 (ia32_multipass_dfa_lookahead): Ditto.
3923 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
3924 (fold_builtin_cpu): Add M_INTEL_KNM.
3925 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
3926 (processor_type): Add PROCESSOR_KNM.
3927 * config/i386/x86-tune.def: Add m_KNM.
3928 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
3930 2017-09-20 Richard Biener <rguenther@suse.de>
3932 PR tree-optimization/80213
3933 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
3934 are allowed in empty BBs as well.
3935 (canonicalize_loop_closed_ssa): Also look for other complex
3937 (scop_detection::get_sese): Include the loop-closed PHI block
3939 (scop_detection::merge_sese): Remove code adding extra blocks.
3940 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
3941 (build_scops): Assert the final returned scop is invalid.
3943 2017-09-20 Richard Biener <rguenther@suse.de>
3945 PR tree-optimization/82264
3946 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
3948 (vn_phi_lookup): Likewise.
3949 (vn_phi_insert): Likewise.
3951 2017-09-20 Jakub Jelinek <jakub@redhat.com>
3953 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
3954 that fits into uhwi or shwi, add DW_AT_const_value regardless
3955 of early_dwarf without going through RTL, using add_AT_unsigned
3958 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
3959 (DEBUG_LTO_ABBREV_SECTION): Likewise.
3960 (DEBUG_LTO_MACINFO_SECTION): Likewise.
3961 (DEBUG_MACRO_SECTION): Likewise.
3962 (DEBUG_LTO_MACRO_SECTION): Likewise.
3963 (DEBUG_STR_DWO_SECTION): Likewise.
3964 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
3965 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
3966 (DEBUG_LTO_DWO_LINE_SECTION): Define.
3967 (DEBUG_LTO_LINE_STR_SECTION): Define.
3968 (init_sections_and_labels): Initialize debug_line_str_section
3969 variable. Initialize debug_loc_section for -gdwarf-5 to
3970 DEBUG_LOCLISTS_SECTION. Formatting fixes.
3972 2017-09-20 Richard Biener <rguenther@suse.de>
3974 * graphite-sese-to-poly.c (extract_affine): Properly handle
3975 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
3977 2017-09-20 Richard Biener <rguenther@suse.de>
3979 PR tree-optimization/81373
3980 * graphite-scop-detection.c (build_cross_bb_scalars_def):
3981 Force SESE live-out defs to be handled even if they are
3984 2017-09-19 Jeff Law <law@redhat.com>
3986 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
3987 nothing for stack adjustments with REG_STACK_CHECK.
3988 * sched-deps.c (parse_add_or_inc): Reject insns with
3989 REG_STACK_CHECK from dependency breaking.
3990 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
3991 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
3992 * reg-notes.def (STACK_CHECK): New note.
3994 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
3995 (ix86_expand_prologue): Dump stack clash info as needed.
3996 Call ix86_adjust_stack_and_probe_stack_clash as needed.
3998 * function.c (dump_stack_clash_frame_info): New function.
3999 * function.h (dump_stack_clash_frame_info): Prototype.
4000 (enum stack_clash_probes): New enum.
4002 * config/alpha/alpha.c (alpha_expand_prologue): Also check
4003 flag_stack_clash_protection.
4004 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
4005 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
4006 (arm_frame_pointer_required): Likewise.
4007 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
4008 (ia64_expand_prologue): Likewise.
4009 * config/mips/mips.c (mips_expand_prologue): Likewise.
4010 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
4011 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
4012 (sparc_flat_expand_prologue): Likewise.
4013 * config/spu/spu.c (spu_expand_prologue): Likewise.
4015 * explow.c: Include "params.h".
4016 (anti_adjust_stack_and_probe_stack_clash): New function.
4017 (get_stack_check_protect): Likewise.
4018 (compute_stack_clash_protection_loop_data): Likewise.
4019 (emit_stack_clash_protection_loop_start): Likewise.
4020 (emit_stack_clash_protection_loop_end): Likewise.
4021 (allocate_dynamic_stack_space): Use get_stack_check_protect.
4022 Use anti_adjust_stack_and_probe_stack_clash.
4023 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
4024 (emit_stack_clash_protection_loop_start): Likewise.
4025 (emit_stack_clash_protection_loop_end): Likewise.
4026 * rtl.h (get_stack_check_protect): Prototype.
4027 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
4028 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
4029 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
4031 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
4033 * doc/tm.texi: Rebuilt.
4034 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
4035 get_stack_check_protect.
4036 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
4037 * config/arm/arm.c (arm_expand_prologue): Likewise.
4038 (arm_frame_pointer_required): Likewise.
4039 * config/i386/i386.c (ix86_expand_prologue): Likewise.
4040 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
4041 * config/mips/mips.c (mips_expand_prologue): Likewise.
4042 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
4043 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
4044 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
4045 (sparc_flat_expand_prologue): Likewise.
4047 * common.opt (-fstack-clash-protection): New option.
4048 * flag-types.h (enum stack_check_type): Note difference between
4049 -fstack-check= and -fstack-clash-protection.
4050 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
4051 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
4052 * toplev.c (process_options): Issue warnings/errors for cases
4053 not handled with -fstack-clash-protection.
4054 * doc/invoke.texi (-fstack-clash-protection): Document new option.
4055 (-fstack-check): Note additional problem with -fstack-check=generic.
4056 Note that -fstack-check is primarily for Ada and refer users
4057 to -fstack-clash-protection for stack-clash-protection.
4058 Document new params for stack clash protection.
4060 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
4062 * config/i386/i386.c (ix86_split_long_move): Do not handle
4063 address used for LEA in a special way.
4065 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
4067 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
4069 2017-09-19 Martin Sebor <msebor@redhat.com>
4072 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
4073 of incompatible types.
4075 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
4077 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
4078 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
4079 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4080 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
4082 2017-09-19 Richard Biener <rguenther@suse.de>
4084 PR tree-optimization/82244
4085 * tree-vrp.c (remove_range_assertions): Do not propagate
4086 a constant to abnormals but replace the assert with a copy.
4088 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
4090 PR rtl-optimization/57878
4091 PR rtl-optimization/68988
4092 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
4093 avoidance test involving non_reload_pseudos. Move frequency test
4094 below the general fragmentation avoidance test.
4096 2017-09-19 Richard Biener <rguenther@suse.de>
4098 PR tree-optimization/69728
4099 * graphite-sese-to-poly.c (schedule_error): New global.
4100 (add_loop_schedule): Handle empty domain by failing the
4102 (build_original_schedule): Handle schedule_error.
4104 2017-09-19 Richard Biener <rguenther@suse.de>
4106 * graphite-scop-detection.c (scop_detection::can_represent_loop):
4107 Do not iterate to sibling loops but only to siblings of inner
4110 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
4113 * config/m68k/m68k.md (moveq feeding equality comparison): Check
4114 that the registers are different.
4116 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
4118 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
4119 to processor_model and "amdfam17h" to arch_names_table.
4120 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
4122 2017-09-18 Jakub Jelinek <jakub@redhat.com>
4125 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
4127 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4128 Alan Hayward <alan.hayward@arm.com>
4129 David Sherwood <david.sherwood@arm.com>
4131 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
4133 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
4135 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
4136 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
4137 vect_schedule_slp_instance.
4138 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
4139 Update call to vect_slp_analyze_node_operations. Simplify return
4141 (vect_slp_analyze_bb_1): Update call accordingly.
4142 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
4143 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
4144 (vect_schedule_slp): Update call accordingly.
4146 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4147 Alan Hayward <alan.hayward@arm.com>
4148 David Sherwood <david.sherwood@arm.com>
4150 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
4151 with types that aren't in fact scalar.
4153 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4155 * tree-vect-slp.c (vect_record_max_nunits): New function,
4157 (vect_build_slp_tree_1): ...here.
4158 (vect_build_slp_tree_2): Call it for phis too.
4160 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4162 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
4163 to vect_get_vec_def_for_operand when getting the mask operand.
4165 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4166 Alan Hayward <alan.hayward@arm.com>
4167 David Sherwood <david.sherwood@arm.com>
4169 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
4172 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4173 Alan Hayward <alan.hayward@arm.com>
4174 David Sherwood <david.sherwood@arm.com>
4176 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
4177 calculation for vector booleans.
4179 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4180 Alan Hayward <alan.hayward@arm.com>
4181 David Sherwood <david.sherwood@arm.com>
4183 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
4185 (vect_transform_stmt): ...here.
4186 (vect_analyze_stmt): Use it instead of calling
4187 vectorizable_live_operation directly.
4189 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
4191 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
4192 non-SIMT targets in acc vector loops.
4194 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
4196 * configure.ac: Add arc and check if assembler supports gdwarf2.
4197 * configure: Regenerate.
4199 2017-09-18 Richard Biener <rguenther@suse.de>
4201 PR tree-optimization/82220
4202 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
4203 epilogue niters from the min_profitable_iters compute.
4205 2017-09-18 Jakub Jelinek <jakub@redhat.com>
4208 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
4209 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
4210 (ix86_init_pic_reg): Revert 2017-09-01 changes.
4212 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
4215 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
4216 switching to a new text section.
4218 2017-09-18 Richard Biener <rguenther@suse.de>
4220 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
4222 (build_alias_set): Reject aliases with no access function.
4224 2017-09-18 Richard Biener <rguenther@suse.de>
4226 PR tree-optimization/79622
4227 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
4229 (build_cross_bb_scalars_use): Likewise.
4231 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
4233 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
4235 2017-09-18 Alan Modra <amodra@gmail.com>
4238 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
4239 stack_pointer_rtx for count 0. Update comments. Break up
4240 large rtl expression.
4242 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
4244 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
4245 Increase to 20 bytes.
4246 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
4247 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
4248 or avx version of the stub.
4250 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
4253 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
4254 compute the minimum stack alignment. Also update preferred stack
4255 boundary for leaf functions.
4257 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
4259 PR tree-optimization/82228
4260 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
4263 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
4265 * common/config/nds32/nds32-common.c
4266 (nds32_option_optimization_table): Refine formatting.
4267 (nds32_option_optimization_table): Use -fsched-pressure and
4268 -fomit-frame-pointer for specific optimization level.
4270 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
4272 * config/nds32/nds32.c: Refine formatting and comments.
4273 * config/nds32/nds32.h: Likewise.
4274 * config/nds32/nds32.md: Likewise.
4275 * config/nds32/nds32-cost.c: Likewise.
4276 * config/nds32/nds32-isr.c: Likewise.
4277 * config/nds32/nds32-md-auxiliary.c: Likewise.
4278 * config/nds32/nds32-multiple.md: Likewise.
4279 * config/nds32/nds32-predicates.c: Likewise.
4281 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
4282 Jakub Jelinek <jakub@redhat.com>
4284 Add support for -std=c++2a.
4285 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
4287 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
4289 2017-09-15 Steve Ellcey <sellcey@cavium.com>
4292 * doc/extend.texi (Common Function Attributes): Add
4293 references to ARM, AArch64, and S/390 specific attributes.
4294 (Function Specific Option Pragmas): Add AArch64 and S/390
4295 to list of back ends that support the target pragma.
4297 2017-09-15 Nathan Sidwell <nathan@acm.org>
4299 * doc/standards.texi: Fix C++17 description. Update URLs for
4302 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
4304 * common.opt (Wcast-align=strict): New warning option.
4305 * doc/invoke.texi: Document -Wcast-align=strict.
4307 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
4309 * cgraph.h (cgraph_thunk_info): Add comments.
4310 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
4311 assert for VIRTUAL_* arguments stricter.
4313 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
4315 PR tree-optimization/71026
4316 * match.pd: Move RDIV patterns from fold-const.c
4317 * fold-const.c (distribute_real_division): Removed.
4318 (fold_binary_loc): Remove calls to distribute_real_divison.
4320 2017-09-15 Jakub Jelinek <jakub@redhat.com>
4322 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
4323 c++1z and gnu++1z as deprecated. Change other references to
4324 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
4325 Change -Wc++1z-compat to -Wc++17-compat.
4326 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
4327 * dwarf2out.c (highest_c_language): Handle C++17.
4328 (gen_compile_unit_die): Likewise.
4330 2017-09-15 Jakub Jelinek <jakub@redhat.com>
4332 PR rtl-optimization/82192
4333 * combine.c (make_extraction): Don't look through non-paradoxical
4334 SUBREGs or TRUNCATE if pos + len is or might be bigger than
4337 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
4338 Alan Hayward <alan.hayward@arm.com>
4339 David Sherwood <david.sherwood@arm.com>
4341 * target.def (function_arg_offset): New hook.
4342 * targhooks.h (default_function_arg_offset): Declare.
4343 * targhooks.c (default_function_arg_offset): New function.
4344 * function.c (locate_and_pad_parm): Use
4345 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
4346 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
4347 (TARGET_FUNCTION_ARG_OFFSET): ...this.
4348 * doc/tm.texi: Regenerate.
4349 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
4350 * config/spu/spu.c (spu_function_arg_offset): New function.
4351 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
4352 * system.h (FUNCTION_ARG_OFFSET): Poison.
4354 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
4355 Alan Hayard <alan.hayward@arm.com>
4356 David Sherwood <david.sherwood@arm.com>
4358 * target.def (truly_noop_truncation): New hook.
4359 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
4360 than TRULY_NOOP_TRUNCATION.
4361 * hooks.h (hook_bool_uint_uint_true): Declare.
4362 * hooks.c (hook_bool_uint_uint_true): New function.
4363 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
4364 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
4365 * doc/tm.texi: Regenerate.
4366 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
4367 rather than TRULY_NOOP_TRUNCATION in comments.
4368 (simplify_comparison): Likewise.
4369 (record_truncated_value): Likewise.
4370 * expmed.c (extract_bit_field_1): Likewise.
4371 (extract_split_bit_field): Likewise.
4372 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
4373 instead of TRULY_NOOP_TRUNCATION.
4374 * function.c (assign_parm_setup_block): Likewise.
4375 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
4376 * rtlhooks.c: Include target.h.
4377 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
4378 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
4379 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
4380 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
4381 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
4382 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
4383 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
4384 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
4385 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
4386 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
4387 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
4388 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
4389 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
4390 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
4391 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
4392 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
4393 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
4394 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
4395 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
4396 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
4397 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
4398 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
4399 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
4400 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
4401 * config/mips/mips.c (mips_truly_noop_truncation): New function.
4402 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
4403 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
4404 rather than TRULY_NOOP_TRUNCATION in comments.
4405 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
4406 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
4407 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
4408 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
4409 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
4410 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
4411 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
4412 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
4413 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
4414 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
4415 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
4416 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
4417 rather than TRULY_NOOP_TRUNCATION in comments.
4418 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
4419 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
4420 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
4421 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
4422 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
4423 TRULY_NOOP_TRUNCATION condition.
4424 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
4425 (TRULY_NOOP_TRUNCATION): Delete.
4426 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
4427 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
4428 * config/spu/spu.c (spu_truly_noop_truncation): New function.
4429 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
4430 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
4431 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
4432 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
4433 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
4434 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
4435 rather than TRULY_NOOP_TRUNCATION in comments.
4436 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
4437 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
4438 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
4439 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
4440 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
4441 * system.h (TRULY_NOOP_TRUNCATION): Poison.
4443 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
4446 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
4447 (*cmp_ior): Likewise.
4448 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
4449 (*ior_scc_scc_cmp): Likewise.
4450 (*and_scc_scc): Likewise.
4451 (*and_scc_scc_cmp): Likewise.
4453 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
4454 Alan Hayard <alan.hayward@arm.com>
4455 David Sherwood <david.sherwood@arm.com>
4457 * target.def (can_change_mode_class): New hook.
4458 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
4459 (hard_regno_nregs): Likewise.
4460 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
4461 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
4462 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
4463 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
4464 (LOAD_EXTEND_OP): Update accordingly.
4465 * doc/tm.texi: Regenerate.
4466 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
4467 CANNOT_CHANGE_MODE_CLASS.
4468 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
4469 (REG_CAN_CHANGE_MODE_P): ...this new macro.
4470 * combine.c (simplify_set): Update accordingly.
4471 * emit-rtl.c (validate_subreg): Likewise.
4472 * recog.c (general_operand): Likewise.
4473 * regcprop.c (mode_change_ok): Likewise.
4474 * reload1.c (choose_reload_regs): Likewise.
4475 (inherit_piecemeal_p): Likewise.
4476 * rtlanal.c (simplify_subreg_regno): Likewise.
4477 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
4478 instead of CANNOT_CHANGE_MODE_CLASS.
4479 (reload_cse_simplify_operands): Likewise.
4480 * reload.c (push_reload): Use targetm.can_change_mode_class
4481 instead of CANNOT_CHANGE_MODE_CLASS.
4482 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
4483 REG_CANNOT_CHANGE_MODE_P.
4484 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4485 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
4486 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4487 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4488 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4489 (arm_can_change_mode_class): New function.
4490 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
4491 than CANNOT_CHANGE_MODE_CLASS in comments.
4492 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4493 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
4494 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
4495 (ix86_can_change_mode_class): ...this new function, inverting the
4496 sense of the return value.
4497 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4498 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4499 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4500 (ia64_can_change_mode_class): New function.
4501 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4502 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
4503 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
4504 (m32c_can_change_mode_class): ...this new function, inverting the
4505 sense of the return value.
4506 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4507 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4508 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
4509 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
4510 (mips_can_change_mode_class): ...this new function, inverting the
4511 sense of the return value.
4512 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4513 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4514 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4515 (msp430_can_change_mode_class): New function.
4516 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4517 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
4518 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4519 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4520 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4521 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
4522 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4523 (pa_cannot_change_mode_class): Replace with...
4524 (pa_can_change_mode_class): ...this new function, inverting the
4525 sense of the return value.
4526 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
4527 than CANNOT_CHANGE_MODE_CLASS in comments.
4528 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4529 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
4530 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4531 (pdp11_cannot_change_mode_class): Replace with...
4532 (pdp11_can_change_mode_class): ...this new function, inverting the
4533 sense of the return value.
4534 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4535 * config/powerpcspe/powerpcspe-protos.h
4536 (rs6000_cannot_change_mode_class_ptr): Delete.
4537 * config/powerpcspe/powerpcspe.c
4538 (rs6000_cannot_change_mode_class_ptr): Delete.
4539 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4540 (rs6000_option_override_internal): Assign to
4541 targetm.can_change_mode_class instead of
4542 rs6000_cannot_change_mode_class_ptr.
4543 (rs6000_cannot_change_mode_class): Replace with...
4544 (rs6000_can_change_mode_class): ...this new function, inverting the
4545 sense of the return value.
4546 (rs6000_debug_cannot_change_mode_class): Replace with...
4547 (rs6000_debug_can_change_mode_class): ...this new function.
4548 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4549 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
4550 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4551 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4552 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
4554 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
4555 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4556 (rs6000_option_override_internal): Assign to
4557 targetm.can_change_mode_class instead of
4558 rs6000_cannot_change_mode_class_ptr.
4559 (rs6000_cannot_change_mode_class): Replace with...
4560 (rs6000_can_change_mode_class): ...this new function, inverting the
4561 sense of the return value.
4562 (rs6000_debug_cannot_change_mode_class): Replace with...
4563 (rs6000_debug_can_change_mode_class): ...this new function.
4564 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4565 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
4566 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
4567 (s390_can_change_mode_class): ...this new function, inverting the
4568 sense of the return value.
4569 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4570 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4571 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
4572 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4573 (sh_cannot_change_mode_class): Replace with...
4574 (sh_can_change_mode_class): ...this new function, inverting the
4575 sense of the return value.
4576 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4577 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4578 (sparc_can_change_mode_class): New function.
4579 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4580 * config/spu/spu.c (spu_can_change_mode_class): New function.
4581 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4582 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4583 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4584 (visium_can_change_mode_class): New function.
4585 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
4587 2017-09-15 Richard Biener <rguenther@suse.de>
4589 PR tree-optimization/82217
4590 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
4591 but not undefined case.
4593 2017-09-15 Jakub Jelinek <jakub@redhat.com>
4596 * postreload.c (reload_cse_simplify_operands): Skip
4597 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
4599 2017-09-15 Richard Biener <rguenther@suse.de>
4601 PR tree-optimization/68823
4602 * graphite-scop-detection.c (build_alias_set): If we have a
4603 possible dependence check whether we can handle them by just
4604 looking at the DRs DR_ACCESS_FNs.
4605 (build_scops): If build_alias_set fails, fail the SCOP.
4607 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
4609 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
4610 to support float128 built-in functions that require the ISA 3.0
4612 (BU_FLOAT128_3_HW): Likewise.
4613 (SQRTF128): Add support for the IEEE 128-bit square root and fma
4615 (FMAF128): Likewise.
4617 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
4618 support for built-in functions that need the ISA 3.0 IEEE 128-bit
4619 floating point instructions.
4620 (rs6000_invalid_builtin): Likewise.
4621 (rs6000_builtin_mask_names): Likewise.
4622 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
4623 (RS6000_BTM_FLOAT128_HW): Likewise.
4624 (RS6000_BTM_COMMON): Likewise.
4625 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
4627 * doc/extend.texi (RS/6000 built-in functions): Document the
4628 IEEE 128-bit floating point square root and fused multiply-add
4631 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
4633 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
4634 reg (r2) isn't in the set of registers defined in the prologue.
4636 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4637 Alan Hayward <alan.hayward@arm.com>
4638 David Sherwood <david.sherwood@arm.com>
4640 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
4641 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
4642 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
4643 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
4644 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
4646 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
4647 max_vectorization_factor.
4648 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
4650 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4651 Alan Hayward <alan.hayward@arm.com>
4652 David Sherwood <david.sherwood@arm.com>
4654 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
4655 (vect_worthwhile_without_simd_p): Declare.
4656 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
4657 (vectorizable_reduction): Use it.
4658 * tree-vect-stmts.c (vectorizable_shift): Likewise.
4659 (vectorizable_operation): Likewise.
4661 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4662 Alan Hayward <alan.hayward@arm.com>
4663 David Sherwood <david.sherwood@arm.com>
4665 * tree-vectorizer.h (vect_get_num_copies): New function.
4666 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
4667 * tree-vect-loop.c (vectorizable_reduction): Likewise.
4668 (vectorizable_induction): Likewise.
4669 (vectorizable_live_operation): Likewise.
4670 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
4671 (vectorizable_bswap): Likewise.
4672 (vectorizable_call): Likewise.
4673 (vectorizable_conversion): Likewise.
4674 (vectorizable_assignment): Likewise.
4675 (vectorizable_shift): Likewise.
4676 (vectorizable_operation): Likewise.
4677 (vectorizable_store): Likewise.
4678 (vectorizable_load): Likewise.
4679 (vectorizable_condition): Likewise.
4680 (vectorizable_comparison): Likewise.
4681 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
4683 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4684 Alan Hayward <alan.hayward@arm.com>
4685 David Sherwood <david.sherwood@arm.com>
4687 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
4688 of vect_init_vector.
4690 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4691 Alan Hayward <alan.hayward@arm.com>
4692 David Sherwood <david.sherwood@arm.com>
4694 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
4695 an inline wrapper that provides a location.
4696 (gimple_build_vector): Likewise.
4697 * gimple-fold.c (gimple_build_vector_from_val): New function.
4698 (gimple_build_vector): Likewise.
4699 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
4700 functions to build the initial value. Always return a gimple value.
4701 (get_initial_defs_for_reduction): Likewise. Only compute
4703 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
4704 vect_init_vector on the results from get_initial_def(s)_for_reduction.
4705 (vectorizable_induction): Use gimple_build_vector rather than
4708 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4709 Alan Hayward <alan.hayward@arm.com>
4710 David Sherwood <david.sherwood@arm.com>
4712 * target.h (vec_perm_indices): New typedef.
4713 (auto_vec_perm_indices): Likewise.
4714 * optabs-query.h: Include target.h
4715 (can_vec_perm_p): Take a vec_perm_indices *.
4716 * optabs-query.c (can_vec_perm_p): Likewise.
4717 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
4718 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
4719 * tree-vect-generic.c (lower_vec_perm): Likewise.
4720 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
4721 (vect_grouped_load_supported): Likewise.
4722 (vect_shift_permute_load_chain): Likewise.
4723 (vect_permute_store_chain): Use auto_vec_perm_indices.
4724 (vect_permute_load_chain): Likewise.
4725 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
4726 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
4727 Update uses of can_vec_perm_p.
4728 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
4729 mode with a number of elements. Take a vec_perm_indices *.
4730 (vect_create_epilog_for_reduction): Update accordingly.
4731 Use auto_vec_perm_indices.
4732 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
4733 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
4734 (vect_transform_slp_perm_load): Likewise.
4735 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
4736 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
4737 (vect_gen_perm_mask_checked): Likewise.
4738 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
4739 (vect_gen_perm_mask_checked): Likewise.
4740 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
4741 (vectorizable_store): Likewise.
4742 (vectorizable_load): Likewise.
4743 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
4744 (vectorizable_bswap): Likewise.
4746 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4747 Alan Hayward <alan.hayward@arm.com>
4748 David Sherwood <david.sherwood@arm.com>
4750 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
4751 * tree.c (build_vector): Likewise.
4752 (build_vector_from_ctor): Update accordingly.
4753 (build_vector_from_val): Likewise.
4754 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
4755 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
4756 * tree-vect-generic.c (add_rshift): Likewise.
4757 (expand_vector_divmod): Likewise.
4758 (optimize_vector_constructor): Likewise.
4759 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
4760 (vect_transform_slp_perm_load): Likewise.
4761 (vect_schedule_slp_instance): Likewise.
4762 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
4763 (vectorizable_call): Likewise.
4764 (vect_gen_perm_mask_any): Likewise. Add elements in order.
4765 * expmed.c (make_tree): Likewise.
4766 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
4767 a vector passed to build_vector.
4768 (fold_convert_const): Likewise.
4769 (exact_inverse): Likewise.
4770 (fold_ternary_loc): Likewise.
4771 (fold_relational_const): Likewise.
4772 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
4773 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
4774 (const_unop): Likewise. Store the reduction accumulator in a
4775 variable rather than an array.
4776 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
4777 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
4778 the new vector, rather than constructing it after the input arrays.
4779 (native_interpret_vector): Use auto_vec<tree> when building
4780 a vector passed to build_vector. Add elements in order.
4781 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
4782 auto_vec<tree> when building a vector passed to build_vector.
4783 (vect_create_epilog_for_reduction): Likewise.
4784 (vectorizable_induction): Likewise.
4785 (get_initial_def_for_reduction): Likewise. Fix indentation of
4787 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
4789 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
4790 passed to build_vector.
4792 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
4793 Alan Hayward <alan.hayward@arm.com>
4794 David Sherwood <david.sherwood@arm.com>
4796 * tree-core.h (tree_base::u): Add an "nelts" field.
4797 (tree_vector): Use VECTOR_CST_NELTS as the length.
4798 * tree.c (tree_size): Likewise.
4799 (make_vector): Initialize VECTOR_CST_NELTS.
4800 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
4801 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
4802 TYPE_VECTOR_SUBPARTS.
4803 * expr.c (const_vector_mask_from_tree): Consistently use "units"
4804 as the number of units, setting it from VECTOR_CST_NELTS.
4805 (const_vector_from_tree): Likewise.
4806 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
4807 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
4808 (fold_negate_expr_1): Likewise.
4809 (fold_convert_const): Likewise.
4810 (const_binop): Likewise. Differentiate the number of output and
4812 (const_unop): Likewise.
4813 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
4814 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
4815 in cases that did the opposite.
4817 2017-09-14 Richard Biener <rguenther@suse.de>
4819 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
4822 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
4824 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
4826 2017-09-14 Jakub Jelinek <jakub@redhat.com>
4829 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
4830 if and where to split a bb, except for splitting before debug insn
4831 sequences followed by non-label real insn. Delete debug insns
4832 in between basic blocks.
4834 * combine.c (make_compound_operation_int): Formatting fixes.
4836 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
4837 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4838 * config/netbsd.h (LINK_EH_SPEC): Likewise.
4839 * config/sol2.h (LINK_EH_SPEC): Likewise.
4840 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4841 * config/s390/linux.h (LINK_SPEC): Likewise.
4842 * config/freebsd.h (LINK_EH_SPEC): Likewise.
4843 * config/openbsd.h (LINK_EH_SPEC): Likewise.
4844 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4845 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
4846 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
4847 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
4848 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
4849 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
4851 2017-09-13 Jakub Jelinek <jakub@redhat.com>
4853 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
4855 (ENDFILE_LINUX_SPEC): Likewise.
4856 (LINK_EH_SPEC): Likewise.
4857 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
4858 (LINK_OS_LINUX_SPEC32): Likewise.
4859 (LINK_OS_LINUX_SPEC64): Likewise.
4860 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
4861 (LINK_OS_LINUX_SPEC): Likewise.
4863 2017-09-13 Martin Liska <mliska@suse.cz>
4866 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
4867 CASE_HIGH is NULL_TREE.
4869 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
4870 Alan Hayward <alan.hayward@arm.com>
4871 David Sherwood <david.sherwood@arm.com>
4873 * target.def (secondary_memory_needed): New hook.
4874 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
4875 instead of SECONDARY_MEMORY_NEEDED.
4876 (secondary_memory_needed_mode): Likewise.
4877 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
4878 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
4879 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
4880 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
4881 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
4882 * doc/tm.texi: Regenerate.
4883 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
4884 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
4885 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4886 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
4887 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
4888 * config/i386/i386.c (inline_secondary_memory_needed): Put the
4889 mode argument first and change the reg_class arguments to reg_class_t.
4890 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
4891 Make static. Update the call to inline_secondary_memory_needed.
4892 (ix86_register_move_cost): Update the call to
4893 inline_secondary_memory_needed.
4894 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4895 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
4897 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
4898 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
4900 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
4901 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
4902 * config/mips/mips.c (mips_secondary_memory_needed): Make static
4903 and match hook interface. Add comment from mips.h.
4904 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4905 * config/mmix/mmix.md (truncdfsf2): Refer to
4906 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
4908 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
4909 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
4910 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4911 (pa_secondary_memory_needed): New function.
4912 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
4913 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
4914 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4915 (pdp11_secondary_memory_needed): Make static and match hook interface.
4916 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
4917 * config/powerpcspe/powerpcspe-protos.h
4918 (rs6000_secondary_memory_needed_ptr): Delete.
4919 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
4921 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4922 (rs6000_option_override_internal): Assign to
4923 targetm.secondary_memory_needed rather than
4924 rs6000_secondary_memory_needed_ptr.
4925 (rs6000_secondary_memory_needed): Match hook interface.
4926 (rs6000_debug_secondary_memory_needed): Likewise.
4927 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
4928 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
4929 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
4930 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4931 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
4932 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
4934 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
4935 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4936 (rs6000_option_override_internal): Assign to
4937 targetm.secondary_memory_needed rather than
4938 rs6000_secondary_memory_needed_ptr.
4939 (rs6000_secondary_memory_needed): Match hook interface.
4940 (rs6000_debug_secondary_memory_needed): Likewise.
4941 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
4942 * config/s390/s390.c (s390_secondary_memory_needed): New function.
4943 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4944 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
4945 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
4946 (sparc_secondary_memory_needed): New function.
4947 * lra-constraints.c (check_and_process_move): Refer to
4948 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
4950 (curr_insn_transform): Likewise.
4951 (process_alt_operands): Use targetm.secondary_memory_needed
4952 instead of TARGET_SECONDARY_MEMORY_NEEDED.
4953 (check_secondary_memory_needed_p): Likewise.
4954 (choose_split_class): Likewise.
4955 * reload.c: Unconditionally include code that was previously
4956 conditional on SECONDARY_MEMORY_NEEDED.
4957 (push_secondary_reload): Use targetm.secondary_memory_needed
4958 instead of TARGET_SECONDARY_MEMORY_NEEDED.
4959 (push_reload): Likewise.
4960 * reload1.c: Unconditionally include code that was previously
4961 conditional on SECONDARY_MEMORY_NEEDED.
4962 (choose_reload_regs): Use targetm.secondary_memory_needed
4963 instead of TARGET_SECONDARY_MEMORY_NEEDED.
4964 (gen_reload): Likewise.
4965 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
4967 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
4968 Alan Hayward <alan.hayward@arm.com>
4969 David Sherwood <david.sherwood@arm.com>
4971 * target.def (secondary_memory_needed_mode): New hook:
4972 * targhooks.c (default_secondary_memory_needed_mode): Declare.
4973 * targhooks.h (default_secondary_memory_needed_mode): New function.
4974 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
4975 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
4976 * doc/tm.texi: Regenerate.
4977 * lra-constraints.c (check_and_process_move): Use
4978 targetm.secondary_memory_needed_mode instead of
4979 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
4980 (curr_insn_transform): Likewise.
4981 * reload.c (get_secondary_mem): Likewise.
4982 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4983 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
4985 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
4986 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4987 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
4988 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
4989 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
4991 * config/powerpcspe/powerpcspe-protos.h
4992 (rs6000_secondary_memory_needed_mode): Delete.
4993 * config/powerpcspe/powerpcspe.c
4994 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
4995 (rs6000_secondary_memory_needed_mode): Make static.
4996 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
4997 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
4999 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
5001 (rs6000_secondary_memory_needed_mode): Make static.
5002 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
5003 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
5004 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
5005 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
5006 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
5008 (sparc_secondary_memory_needed_mode): New function.
5009 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
5011 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
5013 * config/aarch64/constraints.md (Umq): New constraint.
5014 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
5016 (mov<mode>): Update condition.
5018 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5020 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
5021 when bitposition is the same.
5023 2017-09-13 Richard Biener <rguenther@suse.de>
5025 * dwarf2out.c (output_die_symbol): Remove.
5026 (output_die): Do not output a DIEs symbol.
5028 2017-09-13 Richard Biener <rguenther@suse.de>
5031 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
5032 default-def to avoid breaking iterator update with the weird
5033 interaction with cgraph_update_edges_for_call_stmt_node.
5035 2017-09-13 Richard Biener <rguenther@suse.de>
5037 * tree-cfg.c (verify_gimple_assign_binary): Add verification
5038 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
5039 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
5040 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
5042 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
5044 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
5045 Disable pc relative literal load irrespective of
5046 TARGET_FIX_ERR_A53_84341 for default.
5048 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
5050 * config/sparc/sparc.c (output_return): Output the source location of
5051 the insn in the delay slot, if any.
5052 (output_sibcall): Likewise.
5054 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
5057 * common.opt (-static-pie): New alias.
5058 (shared): Negate static-pie.
5059 (-no-pie): Update help text.
5061 (static-pie): New option.
5062 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
5063 -static-pie support.
5064 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
5065 (LINK_EH_SPEC): Likewise.
5066 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5067 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
5068 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
5069 * gcc.c (LINK_COMMAND_SPEC): Likewise.
5070 (init_gcc_specs): Likewise.
5071 (init_spec): Likewise.
5072 (display_help): Update help message for -pie.
5073 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
5076 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
5078 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
5079 (movdi_aarch64): Likewise.
5080 (movti_aarch64): Likewise.
5082 2017-09-12 Simon Wright <simon@pushface.org>
5085 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
5086 calculation of the minor version, always output as 0.
5088 2017-09-12 Jakub Jelinek <jakub@redhat.com>
5091 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
5092 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
5093 on it early, rather than manual conversion late. For
5094 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
5095 instead of performing manual conversion.
5097 2017-09-12 Carl Love <cel@us.ibm.com>
5099 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
5100 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
5102 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
5103 VMULOSW): Add definitions.
5104 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5105 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
5106 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
5107 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
5108 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
5110 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
5112 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
5114 (movti_aarch64): Likewise.
5115 (movdf_aarch64): Likewise.
5116 (movtf_aarch64): Likewise.
5117 (load_pairdi): Likewise.
5118 (store_pairdi): Likewise.
5119 (load_pairdf): Likewise.
5120 (store_pairdf): Likewise.
5121 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
5122 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
5123 (ldr_got_small_<mode>): Likewise.
5124 (ldr_got_small_28k_<mode>): Likewise.
5125 (ldr_got_tiny): Likewise.
5126 * config/aarch64/iterators.md (ldst_sz): New.
5127 (ldpstp_sz): Likewise.
5128 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
5130 (thunderx_load): Split load_8 to load_16.
5131 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
5133 (thunderx2t99_storepair_basic): Split store_8 to store_16.
5134 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
5135 (xgene1_store_pair): Split store_8 to store_16.
5136 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
5137 (falkor_st_0_st_sd): Split store_8 to store_16.
5139 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
5141 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
5142 and store1/2/3/4 to store_4/8/12/16.
5143 * config/aarch64/aarch64.md: Update for rename.
5144 * config/arm/arm.md: Likewise.: Likewise.
5145 * config/arm/arm.c: Likewise.
5146 * config/arm/thumb1.md: Likewise.
5147 * config/arm/thumb2.md: Likewise.
5148 * config/arm/vfp.md: Likewise.
5149 * config/arm/arm-generic.md: Likewise.
5150 * config/arm/arm1020e.md: Likewise.
5151 * config/arm/arm1026ejs.md: Likewise.
5152 * config/arm/arm1136jfs.md: Likewise.
5153 * config/arm/arm926ejs.md: Likewise.
5154 * config/arm/cortex-a15.md: Likewise.
5155 * config/arm/cortex-a17.md: Likewise.
5156 * config/arm/cortex-a5.md: Likewise.
5157 * config/arm/cortex-a53.md: Likewise.
5158 * config/arm/cortex-a57.md: Likewise.
5159 * config/arm/cortex-a7.md: Likewise.
5160 * config/arm/cortex-a8.md: Likewise.
5161 * config/arm/cortex-a9.md: Likewise.
5162 * config/arm/cortex-m4.md: Likewise.
5163 * config/arm/cortex-m7.md: Likewise.
5164 * config/arm/cortex-r4.md: Likewise.
5165 * config/arm/exynos-m1.md: Likewise.
5166 * config/arm/fa526.md: Likewise.
5167 * config/arm/fa606te.md: Likewise.
5168 * config/arm/fa626te.md: Likewise.
5169 * config/arm/fa726te.md: Likewise.
5170 * config/arm/fmp626.md: Likewise.
5171 * config/arm/iwmmxt.md: Likewise.
5172 * config/arm/ldmstm.md: Likewise.
5173 * config/arm/marvell-pj4.md: Likewise.
5174 * config/arm/xgene1.md: Likewise.
5175 * config/aarch64/thunderx.md: Likewise.
5176 * config/aarch64/thunderx2t99.md: Likewise.
5177 * config/aarch64/falkor.md: Likewise.
5179 2017-09-12 Martin Liska <mliska@suse.cz>
5181 * attribs.c (private_lookup_attribute): New function.
5182 * attribs.h (private_lookup_attribute): Declared here.
5183 (lookup_attribute): Called from this place.
5185 2017-09-12 Richard Biener <rguenther@suse.de>
5187 PR tree-optimization/82157
5188 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
5189 stmts with side-effects.
5191 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5192 Alan Hayward <alan.hayward@arm.com>
5193 David Sherwood <david.sherwood@arm.com>
5195 * target.def (hard_regno_nregs): New hook.
5196 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
5197 * targhooks.h (default_hard_regno_nregs): Declare.
5198 * targhooks.c (default_hard_regno_nregs): New function.
5199 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
5200 (TARGET_HARD_REGNO_NREGS): ...this hook.
5201 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
5202 (CLASS_MAX_NREGS): Likewise.
5203 * doc/tm.texi: Regenerate.
5204 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
5205 instead of HARD_REGNO_NREGS.
5206 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
5207 HARD_REGNO_NREGS in the comment.
5208 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
5209 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
5210 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
5211 Return an unsigned int.
5212 (TARGET_HARD_REGNO_NREGS): Redefine.
5213 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
5214 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
5215 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
5216 (arc_hard_regno_nregs): New function.
5217 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
5218 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
5219 (arm_hard_regno_nregs): New function.
5220 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
5221 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
5222 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
5223 (TARGET_HARD_REGNO_NREGS): Redefine.
5224 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
5225 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
5226 (HARD_REGNO_NREGS): Delete.
5227 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
5228 (cr16_hard_regno_nregs): New function.
5229 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
5230 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
5231 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
5232 (cris_hard_regno_nregs): New function.
5233 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
5234 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
5235 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
5236 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
5237 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
5238 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
5239 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
5240 (frv_hard_regno_nregs): Make static. Take and return an
5242 (frv_class_max_nregs): Remove outdated copy of documentation.
5243 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
5244 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
5245 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
5246 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
5247 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
5248 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
5249 (TARGET_HARD_REGNO_NREGS): Redefine.
5250 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
5251 (CLASS_MAX_NREGS): Update comment.
5252 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
5253 (ia64_hard_regno_nregs): New function.
5254 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
5255 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
5256 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
5257 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
5258 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
5260 (m32c_hard_regno_nregs): Likewise. Make static.
5261 (TARGET_HARD_REGNO_NREGS): Redefine.
5262 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
5263 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
5264 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
5265 (m68k_hard_regno_nregs): New function.
5266 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
5267 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
5268 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
5269 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
5270 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
5271 Take and return an unsigned int.
5272 (TARGET_HARD_REGNO_NREGS): Redefine.
5273 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
5274 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
5275 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
5276 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
5277 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
5278 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
5279 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
5280 (msp430_hard_regno_nregs): Make static. Take and return an
5282 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
5283 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
5284 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
5285 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
5286 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
5287 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
5288 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
5289 (TARGET_HARD_REGNO_NREGS): Redefine.
5290 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
5291 (PA_HARD_REGNO_NREGS): ...this.
5292 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
5293 (PA_HARD_REGNO_NREGS): ...this.
5294 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
5295 (pa_hard_regno_nregs): New function.
5296 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
5297 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
5298 (pdp11_hard_regno_nregs): New function.
5299 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
5300 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
5301 (rs6000_hard_regno_nregs_hook): New function.
5302 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
5303 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
5304 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
5305 Take and return an unsigned int. Move earlier in file.
5306 (TARGET_HARD_REGNO_NREGS): Redefine.
5307 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
5308 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
5309 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
5310 (rl78_hard_regno_nregs): Make static. Take and return an
5312 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
5313 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
5314 (rs6000_hard_regno_nregs_hook): New function.
5315 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
5316 * config/rx/rx.c (rx_hard_regno_nregs): New function.
5317 (TARGET_HARD_REGNO_NREGS): Redefine.
5318 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
5319 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
5320 instead of HARD_REGNO_NREGS.
5321 (s390_hard_regno_nregs): New function.
5322 (s390_hard_regno_mode_ok): Add comment from s390.h.
5323 (TARGET_HARD_REGNO_NREGS): Redefine.
5324 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
5325 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
5326 (sh_hard_regno_nregs): New function.
5327 (sh_pass_in_reg_p): Use it.
5328 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
5329 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
5330 (sparc_hard_regno_nregs): New function.
5331 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
5332 * config/spu/spu.c (spu_hard_regno_nregs): New function.
5333 (spu_function_arg_advance): Use it, supplying a valid register number.
5334 (TARGET_HARD_REGNO_NREGS): Redefine.
5335 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
5336 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
5337 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
5338 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
5339 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
5340 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
5341 (CLASS_MAX_NREGS): Remove copy of old documentation.
5342 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
5343 (visium_hard_regno_nregs): New function.
5344 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
5345 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
5346 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
5347 xtensa_hard_regno_nregs): New function.
5348 * system.h (HARD_REGNO_NREGS): Poison.
5350 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5352 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
5353 hard_regno_nregs instead of HARD_REGNO_NREGS.
5354 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
5355 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
5356 (c6x_expand_epilogue): Likewise.
5357 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
5358 (frv_read_iacc_argument): Likewise.
5359 * config/sh/sh.c: Include regs.h.
5360 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
5361 (regs_used): Likewise.
5362 (output_stack_adjust): Likewise.
5363 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
5364 * expmed.c: Include regs.h.
5365 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
5366 * ree.c: Include regs.h.
5367 (combine_reaching_defs): Use hard_regno_nregs instead of
5369 (add_removable_extension): Likewise.
5371 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5373 * regs.h (hard_regno_nregs): Turn into a function.
5374 (end_hard_regno): Update accordingly.
5375 * caller-save.c (setup_save_areas): Likewise.
5376 (save_call_clobbered_regs): Likewise.
5377 (replace_reg_with_saved_mem): Likewise.
5378 (insert_restore): Likewise.
5379 (insert_save): Likewise.
5380 * combine.c (can_change_dest_mode): Likewise.
5381 (move_deaths): Likewise.
5382 (distribute_notes): Likewise.
5383 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
5384 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
5385 (rs6000_split_multireg_move): Likewise.
5386 (rs6000_register_move_cost): Likewise.
5387 (rs6000_memory_move_cost): Likewise.
5388 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
5389 (rs6000_split_multireg_move): Likewise.
5390 (rs6000_register_move_cost): Likewise.
5391 (rs6000_memory_move_cost): Likewise.
5392 * cselib.c (cselib_reset_table): Likewise.
5393 (cselib_lookup_1): Likewise.
5394 * emit-rtl.c (set_mode_and_regno): Likewise.
5395 * function.c (aggregate_value_p): Likewise.
5396 * ira-color.c (setup_profitable_hard_regs): Likewise.
5397 (check_hard_reg_p): Likewise.
5398 (calculate_saved_nregs): Likewise.
5399 (assign_hard_reg): Likewise.
5400 (improve_allocation): Likewise.
5401 (calculate_spill_cost): Likewise.
5402 * ira-emit.c (modify_move_list): Likewise.
5403 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
5404 (ira_hard_reg_in_set_p): Likewise.
5405 * ira.c (setup_reg_mode_hard_regset): Likewise.
5406 (clarify_prohibited_class_mode_regs): Likewise.
5407 (check_allocation): Likewise.
5408 * lra-assigns.c (find_hard_regno_for_1): Likewise.
5409 (lra_setup_reg_renumber): Likewise.
5410 (setup_try_hard_regno_pseudos): Likewise.
5411 (spill_for): Likewise.
5412 (assign_hard_regno): Likewise.
5413 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
5414 * lra-constraints.c (in_class_p): Likewise.
5415 (lra_constraint_offset): Likewise.
5416 (simplify_operand_subreg): Likewise.
5417 (lra_constraints): Likewise.
5418 (split_reg): Likewise.
5419 (split_if_necessary): Likewise.
5420 (invariant_p): Likewise.
5421 (inherit_in_ebb): Likewise.
5422 * lra-lives.c (process_bb_lives): Likewise.
5423 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
5424 (get_hard_regs): Likewise.
5425 (do_remat): Likewise.
5426 * lra-spills.c (assign_spill_hard_regs): Likewise.
5427 * mode-switching.c (create_pre_exit): Likewise.
5428 * postreload.c (reload_combine_recognize_pattern): Likewise.
5429 * recog.c (peep2_find_free_register): Likewise.
5430 * regcprop.c (kill_value_regno): Likewise.
5431 (set_value_regno): Likewise.
5432 (copy_value): Likewise.
5433 (maybe_mode_change): Likewise.
5434 (find_oldest_value_reg): Likewise.
5435 (copyprop_hardreg_forward_1): Likewise.
5436 * regrename.c (check_new_reg_p): Likewise.
5437 (regrename_do_replace): Likewise.
5438 * reload.c (push_reload): Likewise.
5439 (combine_reloads): Likewise.
5440 (find_dummy_reload): Likewise.
5441 (operands_match_p): Likewise.
5442 (find_reloads): Likewise.
5443 (find_equiv_reg): Likewise.
5444 (reload_adjust_reg_for_mode): Likewise.
5445 * reload1.c (count_pseudo): Likewise.
5446 (count_spilled_pseudo): Likewise.
5447 (find_reg): Likewise.
5448 (clear_reload_reg_in_use): Likewise.
5449 (free_for_value_p): Likewise.
5450 (allocate_reload_reg): Likewise.
5451 (choose_reload_regs): Likewise.
5452 (reload_adjust_reg_for_temp): Likewise.
5453 (emit_reload_insns): Likewise.
5454 (delete_output_reload): Likewise.
5455 * rtlanal.c (subreg_get_info): Likewise.
5456 * sched-deps.c (sched_analyze_reg): Likewise.
5457 * sel-sched.c (init_regs_for_mode): Likewise.
5458 (mark_unavailable_hard_regs): Likewise.
5459 (choose_best_reg_1): Likewise.
5460 (verify_target_availability): Likewise.
5461 * valtrack.c (dead_debug_insert_temp): Likewise.
5462 * var-tracking.c (track_loc_p): Likewise.
5463 (emit_note_insn_var_location): Likewise.
5464 * varasm.c (make_decl_rtl): Likewise.
5465 * reginfo.c (choose_hard_reg_mode): Likewise.
5466 (init_reg_modes_target): Refer directly to
5467 this_target_regs->x_hard_regno_nregs.
5469 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5471 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
5472 instead of hard_regno_nregs.
5474 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5476 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
5477 end_hard_regno instead of hard_regno_nregs.
5478 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
5479 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
5480 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
5481 * ira-color.c (improve_allocation): Likewise.
5482 * lra-assigns.c (find_hard_regno_for_1): Likewise.
5483 * lra-lives.c (mark_regno_live): Likewise.
5484 (mark_regno_dead): Likewise.
5485 * lra-remat.c (operand_to_remat): Likewise.
5486 * lra.c (collect_non_operand_hard_regs): Likewise.
5487 * postreload.c (reload_combine_note_store): Likewise.
5488 (move2add_valid_value_p): Likewise.
5489 * reload.c (regno_clobbered_p): Likewise.
5491 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5493 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
5495 * config/v850/v850.c (v850_reorg): Likewise.
5496 * reload.c (refers_to_regno_for_reload_p): Likewise.
5497 (find_equiv_reg): Likewise.
5498 * reload1.c (reload_reg_reaches_end_p): Likewise.
5500 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5502 * caller-save.c (add_used_regs): Use REG_NREGS instead of
5504 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
5505 * config/arm/arm.c (output_move_neon): Likewise.
5506 (arm_attr_length_move_neon): Likewise.
5507 (neon_split_vcombine): Likewise.
5508 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
5509 (c6x_mark_reg_written): Likewise.
5510 (c6x_dwarf_register_span): Likewise.
5511 * config/i386/i386.c (ix86_save_reg): Likewise.
5512 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
5513 (rws_access_reg): Likewise.
5514 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
5515 * mode-switching.c (create_pre_exit): Likewise.
5516 * ree.c (combine_reaching_defs): Likewise.
5517 (add_removable_extension): Likewise.
5518 * regcprop.c (find_oldest_value_reg): Likewise.
5519 (copyprop_hardreg_forward_1): Likewise.
5520 * reload.c (reload_inner_reg_of_subreg): Likewise.
5521 (push_reload): Likewise.
5522 (combine_reloads): Likewise.
5523 (find_dummy_reload): Likewise.
5524 (reload_adjust_reg_for_mode): Likewise.
5525 * reload1.c (find_reload_regs): Likewise.
5526 (forget_old_reloads_1): Likewise.
5527 (reload_reg_free_for_value_p): Likewise.
5528 (reload_adjust_reg_for_temp): Likewise.
5529 (emit_reload_insns): Likewise.
5530 (delete_output_reload): Likewise.
5531 * sel-sched.c (choose_best_reg_1): Likewise.
5532 (choose_best_pseudo_reg): Likewise.
5534 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5535 Alan Hayward <alan.hayward@arm.com>
5536 David Sherwood <david.sherwood@arm.com>
5538 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
5539 * target.def (slow_unaligned_access): New hook.
5540 * targhooks.h (default_slow_unaligned_access): Declare.
5541 * targhooks.c (default_slow_unaligned_access): New function.
5542 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
5543 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
5544 * doc/tm.texi: Regenerate.
5545 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
5546 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
5547 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
5549 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
5550 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
5552 (rs6000_slow_unaligned_access): New function.
5553 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
5554 (expand_block_compare): Likewise.
5555 (expand_strn_compare): Likewise.
5556 (rs6000_rtx_costs): Likewise.
5557 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
5558 (riscv_slow_unaligned_access): Likewise.
5559 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
5560 (riscv_slow_unaligned_access_p): ...this and make static.
5561 (riscv_option_override): Update accordingly.
5562 (riscv_slow_unaligned_access): New function.
5563 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
5564 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
5565 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
5566 (rs6000_slow_unaligned_access): New function.
5567 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
5568 (rs6000_rtx_costs): Likewise.
5569 * config/rs6000/rs6000-string.c (expand_block_compare)
5570 (expand_strn_compare): Use targetm.slow_unaligned_access instead
5571 of SLOW_UNALIGNED_ACCESS.
5572 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
5573 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
5574 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
5575 of SLOW_UNALIGNED_ACCESS.
5576 * expmed.c (simple_mem_bitfield_p): Likewise.
5577 * expr.c (alignment_for_piecewise_move): Likewise.
5578 (emit_group_load_1): Likewise.
5579 (emit_group_store): Likewise.
5580 (copy_blkmode_from_reg): Likewise.
5581 (emit_push_insn): Likewise.
5582 (expand_assignment): Likewise.
5583 (store_field): Likewise.
5584 (expand_expr_real_1): Likewise.
5585 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
5586 * lra-constraints.c (simplify_operand_subreg): Likewise.
5587 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
5588 * gimple-ssa-store-merging.c: Likewise in block comment at start
5590 * tree-ssa-strlen.c: Include target.h.
5591 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
5592 of SLOW_UNALIGNED_ACCESS.
5593 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
5595 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5597 PR rtl-optimization/82185
5598 * expmed.c (emit_store_flag_int): Only test tem if it has been
5601 2017-09-12 Richard Biener <rguenther@suse.de>
5604 * match.pd ((FTYPE) N CMP CST): Fix typo.
5606 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
5608 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
5610 (mips_near_type_p): Add 'short_call' attribute as a synonym
5612 * doc/extend.texi (short_call): Document new function attribute.
5614 2017-09-12 Jakub Jelinek <jakub@redhat.com>
5617 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
5618 assertion check that in the condition.
5619 (get_atomic_generic_size): Likewise. Before testing if parameter
5620 has pointer type, if it has array type, call for C++
5621 default_conversion to perform array-to-pointer conversion.
5623 2017-09-12 Richard Biener <rguenther@suse.de>
5625 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
5626 for operations we cannot scalarize.
5628 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
5630 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
5631 vectors heap vectors. Clean up comments.
5632 Make visited_bbs a reference.
5633 (profitable_jump_thread_path): Make GC
5634 vectors heap vectors. Clean up comments.
5636 (convert_and_register_jump_thread_path): Make GC vectors heap
5638 (check_subpath_and_update_thread_path): Same. Clean up comments.
5639 Make visited_bbs a reference.
5640 (handle_phi): Abstract common code to to
5641 register_jump_thread_path_if_profitable.
5642 Rename VAR_BB to DEF_BB.
5644 Make GC vectors heap vectors.
5645 Make visited_bbs a reference.
5646 (handle_assignment): Same.
5647 (register_jump_thread_path_if_profitable): New.
5648 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
5650 Make GC vectors heap vectors. Clean up comments.
5651 Make visited_bbs a reference.
5652 (find_jump_threads_backwards): Make visited_bbs live in the stack.
5653 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
5656 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
5659 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
5660 words of E_DImode object are reachable by xtensa_uimm8x4 access.
5662 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
5664 Revert r251800 and r251799.
5666 2017-09-11 Martin Jambor <mjambor@suse.cz>
5669 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
5670 arguments in advance.
5671 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
5672 use it to find predecessor edges.
5673 (naive_outof_ssa): Collect vector of predecessors.
5675 2017-09-08 Jason Merrill <jason@redhat.com>
5677 PR c++/70029 - ICE with ref-qualifier and -flto
5678 * langhooks.h (struct lang_hooks_for_types): Add
5679 copy_lang_qualifiers.
5680 * attribs.c (build_type_attribute_qual_variant): Use it.
5681 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
5683 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
5684 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
5686 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
5689 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
5690 (*mulsi3_sp64): New instruction.
5691 (mulsi3): New expander.
5693 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
5695 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
5697 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5699 * sancov.c: Include memmodel.h.
5701 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
5704 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
5707 2017-09-07 Carl Love <cel@us.ibm.com>
5709 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
5710 the sldi instruction.
5712 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
5714 * sancov.c: Include tm_p.h.
5716 2017-09-07 Jakub Jelinek <jakub@redhat.com>
5719 * output.h (switch_to_other_text_partition): New declaration.
5720 * varasm.c (switch_to_other_text_partition): New function.
5721 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
5722 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
5723 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
5724 to the other text partition before emitting LCL label and switch back
5725 after emitting the word after it.
5727 2017-09-07 Richard Biener <rguenther@suse.de>
5729 * passes.def (pass_split_crit_edges): Remove instance before PRE.
5730 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
5731 critical edges here, after loop init.
5732 (pass_data_pre): Remove PROP_no_crit_edges flags.
5733 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
5734 for valueization of call args to avoid leaking VN_TOP.
5735 (visit_use): Assert we do not visit default defs.
5736 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
5737 Use error_mark_node to more easily detect leaking VN_TOP.
5738 All default-defs are varying, not VN_TOP. Mark them visited.
5739 (run_scc_vn): Make code match comment.
5741 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
5743 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
5744 OPTION_MASK_FLOAT128_KEYWORD.
5745 (POWERPC_MASKS): Likewise.
5746 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
5747 support for the -mfloat128-type option, and make -mfloat128
5748 default on PowerPC Linux systems. Define or undefine
5749 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
5750 Define __float128 to be __ieee128 if IEEE 128-bit support is
5751 enabled, or undefine it.
5752 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
5753 Delete defining __FLOAT128_TYPE__.
5754 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
5755 -mfloat128-type option and make -mfloat128 default on PowerPC
5757 (TARGET_FLOAT128_TYPE): Likewise.
5758 (-mfloat128-type): Likewise.
5759 * config/rs6000/rs6000.c (rs6000_option_override_internal):
5760 Delete the -mfloat128-type option and make -mfloat128 default on
5761 PowerPC Linux systems. Always use __ieee128 to be the keyword for
5762 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
5763 128-bit floating point is enabled. Change tests from using
5764 -mfloat128-type to -mfloat128.
5765 (rs6000_mangle_type): Use the correct mangling for the __float128
5766 type even if normal long double is restricted to 64-bits.
5767 (floatn_mode): Enable the _Float128 type by default on VSX Linux
5769 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
5770 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
5771 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
5773 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
5774 documentation for -mfloat128.
5776 2017-09-06 Olivier Hainque <hainque@adacore.com>
5778 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
5780 2017-09-06 Wish Wu <wishwu007@gmail.com>
5781 Jakub Jelinek <jakub@redhat.com>
5783 * asan.c (initialize_sanitizer_builtins): Add
5784 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
5785 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
5786 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
5787 BT_FN_VOID_UINT64_PTR variables.
5788 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
5789 (BT_FN_VOID_UINT16_UINT16): Likewise.
5790 (BT_FN_VOID_UINT32_UINT32): Likewise.
5791 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
5792 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
5793 (BT_FN_VOID_UINT64_PTR): Likewise.
5794 * common.opt (flag_sanitize_coverage): New variable.
5795 (fsanitize-coverage=trace-pc): Remove.
5796 (fsanitize-coverage=): Add.
5797 * flag-types.h (enum sanitize_coverage_code): New enum.
5798 * fold-const.c (fold_range_test): Disable non-short-circuit
5799 optimization if flag_sanitize_coverage.
5800 (fold_truth_andor): Likewise.
5801 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
5802 * opts.c (COVERAGE_SANITIZER_OPT): Define.
5803 (coverage_sanitizer_opts): New array.
5804 (get_closest_sanitizer_option): Add OPTS argument, handle also
5805 OPT_fsanitize_coverage_.
5806 (parse_sanitizer_options): Adjusted to also handle
5807 OPT_fsanitize_coverage_.
5808 (common_handle_option): Add OPT_fsanitize_coverage_.
5809 * sancov.c (instrument_comparison, instrument_switch): New function.
5810 (sancov_pass): Add trace-cmp support.
5811 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
5812 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
5813 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
5814 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
5815 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
5816 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
5817 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
5818 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
5819 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
5820 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
5822 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
5824 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
5825 error. Only quit immediately if parsing is complete.
5826 (BEGIN): Initialize fatal_err and parse_done.
5827 (begin fpu, end fpu): Check number of arguments.
5828 (begin arch, end arch): Likewise.
5829 (begin cpu, end cpu): Likewise.
5830 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
5831 (optalias): Likewise.
5833 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
5835 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
5836 * config/arm/arm-isa.h: Delete. Move definitions to ...
5837 * arm-cpus.in: ... here. Use new feature and fgroup values.
5838 * config/arm/arm.c (arm_option_override): Use lower case for feature
5840 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
5841 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
5842 * config/arm/parsecpu.awk (END): Add new command 'isa'.
5844 (print_isa_bits_for): New function.
5845 (gen_isa): New function.
5846 (gen_comm_data): Use print_isa_bits_for.
5847 (define feature): New keyword.
5848 (define fgroup): New keyword.
5849 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
5850 (arm-isa.h): Add rule to generate file.
5851 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
5852 case for feature bit names.
5854 2017-09-06 Richard Biener <rguenther@suse.de>
5856 * tree-ssa-pre.c (NECESSARY): Remove.
5857 (create_expression_by_pieces): Do not touch pass-local flags.
5858 (insert_into_preds_of_block): Likewise.
5859 (do_pre_regular_insertion): Likewise.
5860 (eliminate_insert): Likewise.
5861 (eliminate_dom_walker::before_dom_children): Likewise.
5862 (fini_eliminate): Do not look at inserted_exprs.
5863 (mark_operand_necessary): Remove.
5864 (remove_dead_inserted_code): Replace with simple work-list
5865 algorithm based on inserted_exprs and SSA uses.
5866 (pass_pre::execute): Re-order fini_eliminate and
5867 remove_dead_inserted_code.
5869 2017-09-06 Olivier Hainque <hainque@adacore.com>
5871 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
5872 for VxWorks 7. Adjust surrounding comments.
5874 2017-09-06 Richard Biener <rguenther@suse.de>
5876 * gimple-ssa-strength-reduction.c
5877 (find_candidates_dom_walker::before_dom_children): Also allow
5880 2017-09-06 Richard Biener <rguenther@suse.de>
5882 PR tree-optimization/82108
5883 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
5884 for gap in the non-permutation SLP case.
5886 2017-09-06 Martin Jambor <mjambor@suse.cz>
5888 PR tree-optimization/82078
5889 * tree-sra.c (sort_and_splice_var_accesses): Move call to
5890 add_access_to_work_queue...
5891 (build_accesses_from_assign): ...here.
5892 (propagate_all_subaccesses): Make sure racc is the group
5893 representative, if there is one.
5895 2017-09-06 Jakub Jelinek <jakub@redhat.com>
5898 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
5901 2017-09-06 Richard Biener <rguenther@suse.de>
5903 * gimple-ssa-strength-reduction.c
5904 (find_candidates_dom_walker::before_doom_children): Use a
5905 type and not a mode check.
5907 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
5910 * config/arm/predicates.md (arm_general_adddi_operand): Create new
5912 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
5914 2017-09-05 Jeff Law <law@redhat.com>
5916 PR tree-optimization/64910
5917 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
5918 cases where we have 3 or more operands.
5920 2017-09-05 Jakub Jelinek <jakub@redhat.com>
5923 * omp-low.c (lower_omp_for): Recompute tree invariant if
5924 gimple_omp_for_initial/final is ADDR_EXPR.
5927 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
5928 into gimple val before gimplification fo the COND_EXPR.
5930 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
5932 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
5933 REGION_COPY argument.
5934 (thread_through_all_blocks): Remove unused argument to
5935 duplicate_thread_path.
5937 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5938 Alan Hayward <alan.hayward@arm.com>
5939 David Sherwood <david.sherwood@arm.com>
5941 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
5942 Take a scalar_mode rather than a machine_mode.
5943 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
5944 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
5945 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
5946 (aarch64_gen_adjusted_ldpstp): Likewise.
5947 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
5949 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5950 Alan Hayward <alan.hayward@arm.com>
5951 David Sherwood <david.sherwood@arm.com>
5953 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
5954 Take a scalar_int_mode instead of a machine_mode.
5955 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
5956 (aarch64_output_scalar_simd_mov_immediate): Likewise.
5957 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
5958 (aarch64_simd_attr_length_rglist): Delete.
5959 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
5960 a scalar_int_mode instead of a machine_mode.
5961 (aarch64_add_offset): Likewise.
5962 (aarch64_internal_mov_immediate): Likewise
5963 (aarch64_add_constant_internal): Likewise.
5964 (aarch64_add_constant): Likewise.
5965 (aarch64_movw_imm): Likewise.
5966 (aarch64_rtx_arith_op_extract_p): Likewise.
5967 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
5968 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
5969 Remove assert that the mode isn't a vector.
5970 (aarch64_output_scalar_simd_mov_immediate): Likewise.
5971 (aarch64_expand_mov_immediate): Update calls after above changes.
5972 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
5973 (aarch64_and_bitmask_imm): Check for scalar integer modes.
5974 (aarch64_move_imm): Likewise.
5975 (aarch64_can_const_movi_rtx_p): Likewise.
5976 (aarch64_strip_extend): Likewise.
5977 (aarch64_extr_rtx_p): Likewise.
5978 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
5979 a CONST_INT when the mode parameter is VOIDmode.
5980 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
5982 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5984 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
5985 * stor-layout.c (bitwise_mode_for_mode): Likewise.
5986 (bitwise_type_for_mode): Update accordingly.
5988 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5990 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
5991 * stor-layout.c (mode_for_size_tree): Likewise.
5992 (mode_for_array): Update accordingly.
5993 (layout_decl): Likewise.
5994 (compute_record_mode): Likewise. Only set the mode once.
5996 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
5998 * target.def (get_mask_mode): Change return type to opt_mode.
6000 * doc/tm.texi: Regenerate.
6001 * targhooks.h (default_get_mask_mode): Return an opt_mode.
6002 * targhooks.c (default_get_mask_mode): Likewise.
6003 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
6004 * optabs-query.c (can_vec_mask_load_store_p): Update use of
6005 targetm.get_mask_mode.
6006 * tree.c (build_truth_vector_type): Likewise.
6008 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6010 * machmode.h (mode_for_vector): Return an opt_mode.
6011 * stor-layout.c (mode_for_vector): Likewise.
6012 (mode_for_int_vector): Update accordingly.
6013 (layout_type): Likewise.
6014 * config/i386/i386.c (emit_memmov): Likewise.
6015 (ix86_expand_set_or_movmem): Likewise.
6016 (ix86_expand_vector_init): Likewise.
6017 (ix86_get_mask_mode): Likewise.
6018 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
6020 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
6021 * expmed.c (extract_bit_field_1): Likewise.
6022 * expr.c (expand_expr_real_2): Likewise.
6023 * optabs-query.c (can_vec_perm_p): Likewise.
6024 (can_vec_mask_load_store_p): Likewise.
6025 * optabs.c (expand_vec_perm): Likewise.
6026 * targhooks.c (default_get_mask_mode): Likewise.
6027 * tree-vect-stmts.c (vectorizable_store): Likewise.
6028 (vectorizable_load): Likewise.
6029 (get_vectype_for_scalar_type_and_size): Likewise.
6031 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6033 * machmode.h (mode_for_int_vector): New function.
6034 * stor-layout.c (mode_for_int_vector): Likewise.
6035 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
6036 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
6037 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
6038 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
6039 (s390_expand_vcond): Likewise.
6041 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6043 * machmode.h (opt_machine_mode): New type.
6044 (opt_mode<T>): Allow construction from anything that can be
6046 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
6047 (mode_for_size): Return an opt_machine_mode.
6048 * stor-layout.c (mode_for_size): Likewise.
6049 (mode_for_size_tree): Update call accordingly.
6050 (bitwise_mode_for_mode): Likewise.
6051 (make_fract_type): Likewise.
6052 (make_accum_type): Likewise.
6053 * caller-save.c (replace_reg_with_saved_mem): Update call
6055 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
6056 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
6057 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
6058 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
6059 * expmed.c (extract_bit_field_1): Likewise.
6060 * reload.c (get_secondary_mem): Likewise.
6061 * varasm.c (assemble_integer): Likewise.
6062 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
6065 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6067 * machmode.h (decimal_float_mode_for_size): New function.
6068 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
6069 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
6070 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
6071 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
6072 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
6073 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
6075 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6077 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
6078 (get_builtin_sync_mode): Likewise.
6079 (expand_ifn_atomic_compare_exchange): Likewise.
6080 (expand_builtin_atomic_clear): Likewise.
6081 (expand_builtin_atomic_test_and_set): Likewise.
6082 (fold_builtin_atomic_always_lock_free): Likewise.
6083 * calls.c (compute_argument_addresses): Likewise.
6084 (emit_library_call_value_1): Likewise.
6085 (store_one_arg): Likewise.
6086 * combine.c (combine_instructions): Likewise.
6087 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
6088 * config/arm/arm.c (arm_function_value): Likewise.
6089 (aapcs_allocate_return_reg): Likewise.
6090 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
6091 * config/i386/i386.c (construct_container): Likewise.
6092 (ix86_gimplify_va_arg): Likewise.
6093 (ix86_expand_sse_cmp): Likewise.
6094 (emit_memmov): Likewise.
6095 (emit_memset): Likewise.
6096 (expand_small_movmem_or_setmem): Likewise.
6097 (ix86_expand_pextr): Likewise.
6098 (ix86_expand_pinsr): Likewise.
6099 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
6100 * config/microblaze/microblaze.c (microblaze_block_move_straight):
6102 * config/mips/mips.c (mips_function_value_1) Likewise.
6103 (mips_block_move_straight): Likewise.
6104 (mips_expand_ins_as_unaligned_store): Likewise.
6105 * config/powerpcspe/powerpcspe.c
6106 (rs6000_darwin64_record_arg_advance_flush): Likewise.
6107 (rs6000_darwin64_record_arg_flush): Likewise.
6108 * config/rs6000/rs6000.c
6109 (rs6000_darwin64_record_arg_advance_flush): Likewise.
6110 (rs6000_darwin64_record_arg_flush): Likewise.
6111 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
6112 (sparc_function_value_1): Likewise.
6113 * config/spu/spu.c (adjust_operand): Likewise.
6114 (spu_emit_branch_or_set): Likewise.
6115 (arith_immediate_p): Likewise.
6116 * emit-rtl.c (gen_lowpart_common): Likewise.
6117 * expr.c (expand_expr_real_1): Likewise.
6118 * function.c (assign_parm_setup_block): Likewise.
6119 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
6120 * reload1.c (alter_reg): Likewise.
6121 * stor-layout.c (mode_for_vector): Likewise.
6122 (layout_type): Likewise.
6124 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6126 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
6127 (spu_convert_move): Likewise.
6128 * lower-subreg.c (resolve_simple_move): Likewise.
6130 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6133 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
6134 define_insn to a define_expand.
6135 (altivec_vsum2sws_direct): New define_insn.
6136 (altivec_vsumsws): Convert from a define_insn to a define_expand.
6138 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
6140 * config/arm/arm.c (arm_option_params_internal): Improve setting of
6143 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
6148 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
6149 realign stack if stack alignment needed is less than incoming
6152 2017-09-05 Marek Polacek <polacek@redhat.com>
6155 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
6158 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
6160 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
6162 2017-09-05 Richard Biener <rguenther@suse.de>
6164 PR tree-optimization/82084
6165 * fold-const.c (can_native_encode_string_p): Handle wide characters.
6167 2017-09-05 Richard Biener <rguenther@suse.de>
6169 PR tree-optimization/82102
6170 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
6172 2017-09-05 Martin Liska <mliska@suse.cz>
6174 PR tree-optimization/82032
6175 * tree-cfg.c (generate_range_test): New function.
6176 * tree-cfg.h (generate_range_test): Declared here.
6177 * tree-cfgcleanup.c (convert_single_case_switch): New function.
6178 (cleanup_control_expr_graph): Use it.
6179 * tree-switch-conversion.c (try_switch_expansion): Remove
6181 (emit_case_nodes): Use generate_range_test.
6183 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
6186 * config/i386/i386.md (*<btsc><mode>_mask): Add
6187 TARGET_USE_BT to insn constraint.
6188 (*btr<mode>_mask): Ditto.
6190 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
6192 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
6193 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
6195 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
6198 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
6199 TARGET_NEON and TARGET_IWMMXT.
6200 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
6201 TARGET_NEON and TARGET_IWMMXT.
6202 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
6204 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
6206 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
6207 (ix86_rewrite_tls_address): Ditto.
6208 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
6209 (ix86_rewrite_tls_address_1): Ditto.
6210 (ix86_rewrite_tls_address): Ditto.
6211 * config/i386/predicates.md (tls_address_pattern): New predicate.
6212 * config/i386/i386.md (TLS address splitter): New splitter.
6214 2017-09-04 Richard Biener <rguenther@suse.de>
6216 PR tree-optimization/82084
6217 * fold-const.h (can_native_encode_string_p): Declare.
6218 * fold-const.c (can_native_encode_string_p): Factor out from ...
6219 (native_encode_string): ... here.
6220 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
6221 vectorizing stores from constants we later cannot handle.
6223 2017-09-04 Marek Polacek <polacek@redhat.com>
6226 * doc/invoke.texi: Update -Wtautological-compare documentation.
6228 2017-09-04 Jeff Law <law@redhat.com>
6230 PR tree-optimization/64910
6231 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
6232 swap the first and last operand if the last is a constant.
6234 2017-09-04 Marek Polacek <polacek@redhat.com>
6237 * convert.c (do_narrow): When sanitizing signed integer overflows,
6238 bail out for signed types.
6239 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
6241 2017-09-04 Richard Biener <rguenther@suse.de>
6243 PR tree-optimization/82060
6244 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
6245 Move devirtualization after stmt folding and before EH/AB/noreturn
6246 cleanup to get the stmt refs canonicalized. Use a bool instead
6247 of gimple_modified_p since that doesn't work for NOPs. Schedule
6248 NOPs generated by folding for removal.
6250 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6251 Alan Hayward <alan.hayward@arm.com>
6252 David Sherwood <david.sherwood@arm.com>
6254 * coretypes.h (pad_direction): New enum.
6255 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
6256 (FUNCTION_ARG_PADDING): Likewise.
6257 * target.def (function_arg_padding): New hook.
6258 * targhooks.h (default_function_arg_padding): Declare.
6259 * targhooks.c (default_function_arg_padding): New function.
6260 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
6261 (TARGET_FUNCTION_ARG_PADDING): ...this.
6262 * doc/tm.texi: Regenerate.
6263 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
6264 instead of direction.
6265 (compute_argument_addresses): Likewise.
6266 (load_register_parameters): Likewise.
6267 (emit_library_call_value_1): Likewise.
6268 (store_one_arg): Use targetm.calls.function_arg_padding instead
6269 of FUNCTION_ARG_PADDING.
6270 (must_pass_in_stack_var_size_or_pad): Likewise.
6271 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
6272 (emit_group_store): Likewise.
6273 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
6274 instead of FUNCTION_ARG_PADDING.
6275 (emit_push_insn): Likewise, and propagate enum change throughout
6277 * function.h (direction): Delete.
6278 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
6280 * function.c (assign_parm_find_stack_rtl): Likewise.
6281 (assign_parm_setup_block_p): Likewise.
6282 (assign_parm_setup_block): Likewise.
6283 (gimplify_parameters): Likewise.
6284 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
6285 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
6287 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
6288 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
6289 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
6290 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
6291 (aarch64_function_arg_padding): ...this new function.
6292 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
6293 (TARGET_FUNCTION_ARG_PADDING): Redefine.
6294 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
6295 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
6296 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
6297 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6298 (arm_pad_arg_upward): Replace with...
6299 (arm_function_arg_padding): ...this new function.
6300 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
6302 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
6303 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
6304 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6305 (ia64_hpux_function_arg_padding): Replace with...
6306 (ia64_function_arg_padding): ...this new function. Use pad_direction
6307 instead of direction. Check for TARGET_HPUX.
6308 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
6309 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6310 (iq2000_function_arg_padding): New function.
6311 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
6312 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
6313 (mips_function_arg_padding): ...this new function.
6314 (mips_pad_reg_upward): Update accordingly.
6315 (TARGET_FUNCTION_ARG_PADDING): Redefine.
6316 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
6317 targetm.calls.function_arg_padding.
6318 (FUNCTION_ARG_PADDING): Delete.
6319 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
6320 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
6321 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
6322 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
6323 (nios2_block_reg_padding): Return pad_direction instead of direction.
6324 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
6325 instead of direction.
6326 (nios2_function_arg_padding): Likewise. Make static.
6327 (TARGET_FUNCTION_ARG_PADDING): Redefine.
6328 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
6329 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
6330 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
6331 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6332 (pa_function_arg_padding): Make static. Return pad_direction instead
6334 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
6335 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
6336 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
6337 instead of direction. Use targetm.calls.function_arg_padding.
6338 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
6339 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
6340 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
6341 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
6342 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
6344 (function_arg_padding): Rename to...
6345 (rs6000_function_arg_padding): ...this. Make static. Return
6346 pad_direction instead of direction.
6347 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
6348 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
6349 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
6350 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
6351 instead of direction. Use targetm.calls.function_arg_padding.
6352 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
6353 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
6354 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
6355 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
6356 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6357 (function_arg_padding): Rename to...
6358 (rs6000_function_arg_padding): ...this. Make static. Return
6359 pad_direction instead of direction.
6360 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
6361 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
6362 * config/s390/s390.c (s390_function_arg_padding): New function.
6363 (TARGET_FUNCTION_ARG_PADDING): Redefine.
6364 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
6365 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
6366 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6367 (function_arg_padding): Rename to...
6368 (sparc_function_arg_padding): ...this. Make static. Return
6369 pad_direction instead of direction.
6370 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
6371 * config/spu/spu.c (spu_function_arg_padding): New function.
6372 (TARGET_FUNCTION_ARG_PADDING): Redefine.
6373 * system.h (FUNCTION_ARG_PADDING): Poison.
6375 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6376 Alan Hayward <alan.hayward@arm.com>
6377 David Sherwood <david.sherwood@arm.com>
6379 * target.def (modes_tieable_p): New hook.
6380 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
6381 (TARGET_MODES_TIEABLE_P): ...this.
6382 * doc/tm.texi.in: Regenerate.
6383 * hooks.h (hook_bool_mode_mode_true): Declare.
6384 * hooks.c (hook_bool_mode_mode_true): New function.
6385 * combine.c (subst): Use targetm.modes_tieable_p instead of
6387 * dse.c (find_shift_sequence): Likewise.
6388 * expmed.c (extract_low_bits): Likewise.
6389 * lower-subreg.c: Include target.h.
6390 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
6392 * rtlanal.c (rtx_cost): Likewise.
6393 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
6394 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
6395 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
6396 (TARGET_MODES_TIEABLE_P): Redefine.
6397 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
6398 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
6399 (TARGET_MODES_TIEABLE_P): Redefine.
6400 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
6401 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
6402 (arc_modes_tieable_p): New function.
6403 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
6404 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
6405 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
6406 (arm_modes_tieable_p): Make static.
6407 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
6408 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
6409 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
6410 (TARGET_MODES_TIEABLE_P): Redefine.
6411 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
6412 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
6413 (TARGET_MODES_TIEABLE_P): Redefine.
6414 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
6415 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
6416 (cr16_modes_tieable_p): New function.
6417 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
6418 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
6419 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
6420 (TRULY_NOOP_TRUNCATION): Update comment.
6421 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
6422 (TRULY_NOOP_TRUNCATION): Update comment.
6423 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
6424 (frv_modes_tieable_p): New function.
6425 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
6426 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
6427 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
6428 (TARGET_MODES_TIEABLE_P): Redefine.
6429 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
6430 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
6431 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
6432 (TARGET_MODES_TIEABLE_P): Redefine.
6433 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
6434 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
6435 (ia64_modes_tieable_p): New function.
6436 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
6437 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
6438 (iq2000_modes_tieable_p): New function.
6439 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
6440 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
6441 (lm32_modes_tieable_p): New function.
6442 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
6443 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
6444 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
6445 (TARGET_MODES_TIEABLE_P): Redefine.
6446 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
6447 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
6448 (m32r_modes_tieable_p): New function.
6449 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
6450 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
6451 (m68k_modes_tieable_p): New function.
6452 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
6453 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
6454 (mcore_modes_tieable_p): New function.
6455 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
6456 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
6458 (TARGET_MODES_TIEABLE_P): Redefine.
6459 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
6460 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
6461 * config/mips/mips.c (mips_modes_tieable_p): Make static.
6462 (TARGET_MODES_TIEABLE_P): Redefine.
6463 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
6464 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
6465 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
6466 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
6467 (mn10300_modes_tieable_p): ...this and make static.
6468 (TARGET_MODES_TIEABLE_P): Redefine.
6469 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
6470 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
6471 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
6472 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
6473 (msp430_modes_tieable_p): Make static.
6474 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
6475 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
6476 (TARGET_MODES_TIEABLE_P): Redefine.
6477 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
6478 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
6479 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
6480 (TARGET_MODES_TIEABLE_P): Redefine.
6481 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
6482 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
6483 * config/pa/pa.c (pa_modes_tieable_p): Make static.
6484 (TARGET_MODES_TIEABLE_P): Redefine.
6485 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
6486 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
6487 (pdp11_modes_tieable_p): New function.
6488 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
6489 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
6490 (rs6000_modes_tieable_p): New function.
6491 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
6492 * config/powerpcspe/powerpcspe.md: Update comment.
6493 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
6494 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
6495 (TARGET_MODES_TIEABLE_P): Redefine.
6496 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
6497 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
6498 (rl78_modes_tieable_p): New function.
6499 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
6500 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
6501 (rs6000_modes_tieable_p): New function.
6502 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
6503 * config/rs6000/rs6000.md: Update comment.
6504 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
6505 * config/rx/rx.c (rx_modes_tieable_p): New function.
6506 (TARGET_MODES_TIEABLE_P): Redefine.
6507 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
6508 * config/s390/s390.c (s390_modes_tieable_p): New function.
6509 (TARGET_MODES_TIEABLE_P): Redefine.
6510 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
6511 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
6512 (sh_modes_tieable_p): New function.
6513 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
6514 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
6515 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
6516 (sparc_modes_tieable_p): Make static.
6517 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
6518 * config/spu/spu.c (spu_modes_tieable_p): New function.
6519 (TARGET_MODES_TIEABLE_P): Redefine.
6520 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
6521 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
6522 (TARGET_MODES_TIEABLE_P): Redefine.
6523 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
6524 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
6525 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
6526 * config/v850/v850.c (v850_modes_tieable_p): New function.
6527 (TARGET_MODES_TIEABLE_P): Redefine.
6528 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
6529 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
6530 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
6531 (visium_modes_tieable_p): New function.
6532 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
6533 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
6534 (xtensa_modes_tieable_p): New function.
6535 * system.h (MODES_TIEABLE_P): Poison.
6537 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6538 Alan Hayward <alan.hayward@arm.com>
6539 David Sherwood <david.sherwood@arm.com>
6541 * target.def (hard_regno_mode_ok): New hook.
6542 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
6543 (TARGET_HARD_REGNO_MODE_OK): ...this.
6544 * doc/tm.texi.in: Regenerate.
6545 * hooks.h (hook_bool_uint_mode_true): Declare.
6546 * hooks.c (hook_bool_uint_mode_true): New function.
6547 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
6549 * genpreds.c (write_insn_preds_c): Add an include of target.h.
6550 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
6551 instead of HARD_REGNO_MODE_OK.
6552 * caller-save.c: Include target.h.
6553 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
6555 * combine.c (can_combine_p): Likewise.
6556 (combinable_i3pat): Likewise.
6557 (can_change_dest_mode): Likewise.
6558 * expr.c (init_expr_target): Likewise.
6559 (convert_move): Likewise.
6560 (convert_modes): Likewise.
6561 * ira.c (setup_prohibited_class_mode_regs): Likewise.
6562 (setup_prohibited_mode_move_regs): Likewise.
6563 * ira.h (target_ira): Likewise.
6564 * lra-assigns.c (find_hard_regno_for_1): Likewise.
6565 * lra-constraints.c (process_alt_operands): Likewise.
6566 (split_reg): Likewise.
6567 * recog.c (peep2_find_free_register): Likewise.
6568 * ree.c (combine_reaching_defs): Likewise.
6569 * regcprop.c (maybe_mode_change): Likewise.
6570 * reginfo.c (init_reg_sets_1): Likewise.
6571 (choose_hard_reg_mode): Likewise.
6572 (simplifiable_subregs): Likewise.
6573 * regrename.c (check_new_reg_p): Likewise.
6574 * reload.c (find_valid_class): Likewise.
6575 (find_valid_class_1): Likewise.
6576 (reload_inner_reg_of_subreg): Likewise.
6577 (push_reload): Likewise.
6578 (combine_reloads): Likewise.
6579 (find_dummy_reload): Likewise.
6580 (find_reloads): Likewise.
6581 * reload1.c (find_reg): Likewise.
6582 (set_reload_reg): Likewise.
6583 (allocate_reload_reg): Likewise.
6584 (choose_reload_regs): Likewise.
6585 (reload_adjust_reg_for_temp): Likewise.
6586 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
6587 (simplify_subreg_regno): Likewise.
6588 * sel-sched.c (init_regs_for_mode): Likewise.
6589 * varasm.c (make_decl_rtl): Likewise.
6590 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
6591 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
6593 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
6594 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
6595 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6596 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
6597 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
6598 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6599 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
6600 (arc_mode_class): Delete.
6601 (HARD_REGNO_MODE_OK): Delete.
6602 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6603 (arc_hard_regno_mode_ok): Rename old array to...
6604 (arc_hard_regno_mode_ok_modes): ...this.
6605 (arc_conditional_register_usage): Update accordingly.
6606 (arc_mode_class): Make static.
6607 (arc_hard_regno_mode_ok): New function.
6608 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
6609 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
6610 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6611 (arm_hard_regno_mode_ok): Make static.
6612 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
6614 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
6615 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
6616 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
6618 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6619 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
6620 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
6621 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
6622 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
6623 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6624 * config/bfin/predicates.md (valid_reg_operand): Use
6625 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
6626 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
6627 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
6628 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6629 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
6630 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
6631 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6632 (cr16_hard_regno_mode_ok): Make static and return a bool.
6633 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
6634 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6635 (cris_hard_regno_mode_ok): New function.
6636 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
6637 (epiphany_mode_class): Delete.
6638 (HARD_REGNO_MODE_OK): Delete.
6639 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
6640 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6641 (hard_regno_mode_ok): Rename to...
6642 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
6643 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
6644 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
6646 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
6647 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
6648 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6649 (frv_hard_regno_mode_ok): Make static and return a bool.
6650 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
6652 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
6653 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
6654 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
6655 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
6657 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6658 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
6659 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
6660 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
6662 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6663 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
6664 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6665 (ia64_hard_regno_mode_ok): New function.
6666 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
6667 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6668 (iq2000_hard_regno_mode_ok): New function.
6669 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
6670 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6671 (lm32_hard_regno_mode_ok): New function.
6672 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
6673 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
6674 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
6675 instead of HARD_REGNO_MODE_OK.
6676 (m32c_hard_regno_ok): Rename to...
6677 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
6678 (m32c_cannot_change_mode_class): Update accordingly.
6679 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6680 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
6681 (m32r_mode_class): Delete.
6682 (HARD_REGNO_MODE_OK): Delete.
6683 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6684 (m32r_hard_regno_mode_ok): Rename to...
6685 (m32r_hard_regno_modes): ...this.
6686 (m32r_mode_class): Make static.
6687 (m32r_hard_regno_mode_ok): New function.
6688 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
6689 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
6690 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6691 (m68k_hard_regno_mode_ok): Make static.
6692 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
6693 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6694 (mcore_hard_regno_mode_ok): New function.
6695 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
6696 (HARD_REGNO_MODE_OK): Delete.
6697 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
6699 (microblaze_hard_regno_mode_ok_p): ...this and make static.
6700 (microblaze_hard_regno_mode_ok): New function.
6701 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6702 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
6703 (mips_hard_regno_mode_ok): Delete.
6704 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
6705 (mips_hard_regno_mode_ok_p): ...this and make static.
6706 (mips_hard_regno_mode_ok_p): Rename to...
6707 (mips_hard_regno_mode_ok_uncached): ...this.
6708 (mips_hard_regno_mode_ok): New function.
6709 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
6710 of HARD_REGNO_MODE_OK.
6711 (mips_option_override): Update after above name changes.
6712 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6713 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
6714 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
6715 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
6716 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
6717 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6718 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
6719 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
6720 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
6721 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6722 (msp430_hard_regno_mode_ok): Make static and return a bool.
6723 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
6724 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
6725 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
6727 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6728 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
6729 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
6730 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
6731 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
6732 (PA_HARD_REGNO_MODE_OK): ...this
6733 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
6734 (PA_HARD_REGNO_MODE_OK): ...this.
6735 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6736 (pa_hard_regno_mode_ok): New function.
6737 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
6738 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6739 (pdp11_hard_regno_mode_ok): New function.
6740 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
6741 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
6743 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
6745 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6746 (rs6000_hard_regno_mode_ok): Rename to...
6747 (rs6000_hard_regno_mode_ok_uncached): ...this.
6748 (rs6000_init_hard_regno_mode_ok): Update accordingly.
6749 (rs6000_hard_regno_mode_ok): New function.
6750 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
6751 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
6752 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
6753 (riscv_hard_regno_mode_ok): ...this and make static.
6754 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6755 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
6756 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
6757 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6758 (rl78_hard_regno_mode_ok): Make static and return bool.
6759 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
6760 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
6762 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
6763 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6764 (rs6000_hard_regno_mode_ok): Rename to...
6765 (rs6000_hard_regno_mode_ok_uncached): ...this.
6766 (rs6000_init_hard_regno_mode_ok): Update accordingly.
6767 (rs6000_hard_regno_mode_ok): New function.
6768 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
6769 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
6770 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6771 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
6772 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
6773 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
6774 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6775 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
6776 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
6777 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6778 (sh_hard_regno_mode_ok): Make static.
6779 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
6780 instead of HARD_REGNO_MODE_OK.
6781 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
6782 (sparc_mode_class): Delete.
6783 (HARD_REGNO_MODE_OK): Delete.
6784 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6785 (hard_regno_mode_classes): Make static.
6786 (sparc_mode_class): Likewise.
6787 (sparc_hard_regno_mode_ok): New function.
6788 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
6789 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
6790 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
6792 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6793 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
6794 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
6795 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
6796 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
6797 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6798 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
6799 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
6800 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
6801 (visium_hard_regno_mode_ok): New function.
6802 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
6803 instead of HARD_REGNO_MODE_OK.
6804 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
6805 (HARD_REGNO_MODE_OK): Delete.
6806 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
6807 (xtensa_hard_regno_mode_ok_p): ...this and make static.
6808 (xtensa_option_override): Update accordingly.
6809 (TARGET_HARD_REGNO_MODE_OK): Redefine.
6810 (xtensa_hard_regno_mode_ok): New function.
6811 * system.h (HARD_REGNO_MODE_OK): Poison.
6813 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6814 Alan Hayward <alan.hayward@arm.com>
6815 David Sherwood <david.sherwood@arm.com>
6817 * target.def (hard_regno_call_part_clobbered): New hook.
6818 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
6819 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
6820 * doc/tm.texi: Regenerate.
6821 * hooks.h (hook_bool_uint_mode_false): Declare.
6822 * hooks.c (hook_bool_uint_mode_false): New function.
6823 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6824 * cselib.c (cselib_process_insn): Use
6825 targetm.hard_regno_call_part_clobbered instead of
6826 HARD_REGNO_CALL_PART_CLOBBERED.
6827 * ira-conflicts.c (ira_build_conflicts): Likewise.
6828 * ira-costs.c (ira_tune_allocno_costs): Likewise.
6829 * lra-constraints.c (need_for_call_save_p): Likewise.
6830 * lra-lives.c: Include target.h.
6831 (check_pseudos_live_through_calls): Use
6832 targetm.hard_regno_call_part_clobbered instead of
6833 HARD_REGNO_CALL_PART_CLOBBERED.
6834 * regcprop.c: Include target.h.
6835 (copyprop_hardreg_forward_1): Use
6836 targetm.hard_regno_call_part_clobbered instead of
6837 HARD_REGNO_CALL_PART_CLOBBERED.
6838 * reginfo.c (choose_hard_reg_mode): Likewise.
6839 * regrename.c (check_new_reg_p): Likewise.
6840 * reload.c (find_equiv_reg): Likewise.
6841 * reload1.c (emit_reload_insns): Likewise.
6842 * sched-deps.c (deps_analyze_insn): Likewise.
6843 * sel-sched.c (init_regs_for_mode): Likewise.
6844 (mark_unavailable_hard_regs): Likewise.
6845 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
6846 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6847 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
6849 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6850 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6851 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
6853 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
6855 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6856 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6857 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
6859 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6860 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6861 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
6863 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6864 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
6866 * config/powerpcspe/powerpcspe.c
6867 (rs6000_hard_regno_call_part_clobbered): New function.
6868 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6869 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6870 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
6872 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6873 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6874 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
6876 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
6877 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
6878 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
6880 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6881 Alan Hayward <alan.hayward@arm.com>
6882 David Sherwood <david.sherwood@arm.com>
6884 * rtl.h (subreg_memory_offset): Declare.
6885 * emit-rtl.c (subreg_memory_offset): New function.
6886 * expmed.c (store_bit_field_1): Use it.
6887 * expr.c (undefined_operand_subword_p): Likewise.
6888 * simplify-rtx.c (simplify_subreg): Likewise.
6890 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
6892 PR rtl-optimization/57448
6895 * optabs.c (expand_atomic_load): Place compiler memory barriers if
6896 using atomic_load pattern.
6897 (expand_atomic_store): Likewise.
6899 2017-09-04 Jakub Jelinek <jakub@redhat.com>
6902 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
6903 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
6904 handling. Use replace_call_with_value with NULL instead of
6905 gsi_replace, unlink_stmt_vdef and release_defs.
6907 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
6910 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
6912 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6915 * rtl.h (emit_library_call_value_1): Declare.
6916 (emit_library_call): Replace declaration with a series of overloads.
6917 Remove the parameter count argument.
6918 (emit_library_call_value): Likewise.
6919 * calls.c (emit_library_call_value_1): Make global. Replace varargs
6920 with an "rtx_mode_t *".
6921 (emit_library_call_value): Delete.
6922 (emit_library_call): Likewise.
6923 * asan.c (asan_emit_stack_protection): Update calls accordingly.
6924 (asan_emit_allocas_unpoison): Likewise.
6925 * builtins.c (expand_builtin_powi): Likewise.
6926 (expand_asan_emit_allocas_unpoison): Likewise.
6927 * cfgexpand.c (expand_main_function): Likewise.
6928 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
6929 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
6930 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
6931 * config/arm/arm.c (arm_trampoline_init): Likewise.
6932 (arm_call_tls_get_addr): Likewise.
6933 (arm_expand_divmod_libfunc): Likewise.
6934 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
6935 (smulsi3_highpart): Likewise.
6936 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
6937 (c6x_expand_compare): Likewise.
6938 (c6x_expand_movmem): Likewise.
6939 * config/frv/frv.c (frv_trampoline_init): Likewise.
6940 * config/i386/i386.c (ix86_trampoline_init): Likewise.
6941 (ix86_expand_divmod_libfunc): Likewise.
6942 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
6943 (ia64_expand_compare): Likewise.
6944 (ia64_profile_hook): Likewise.
6945 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
6946 (nonlocal_goto): Likewise.
6947 (restore_stack_nonlocal): Likewise.
6948 * config/m32r/m32r.c (block_move_call): Likewise.
6949 (m32r_trampoline_init): Likewise.
6950 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
6951 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
6952 (m68k_call_m68k_read_tp): Likewise.
6953 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
6954 (microblaze_expand_divide): Likewise.
6955 * config/mips/mips.h (mips_args): Likewise.
6956 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
6957 (MIPS_ICACHE_SYNC): Likewise.
6958 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
6959 (nios2_trampoline_init): Likewise.
6960 * config/pa/pa.c (hppa_tls_call): Likewise.
6961 (pa_trampoline_init): Likewise.
6962 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
6963 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
6964 (expand_strn_compare): Likewise.
6965 (rs6000_generate_compare): Likewise.
6966 (rs6000_expand_float128_convert): Likewise.
6967 (output_profile_hook): Likewise.
6968 (rs6000_trampoline_init): Likewise.
6969 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
6970 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
6971 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
6972 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
6973 (rs6000_generate_compare): Likewise.
6974 (rs6000_expand_float128_convert): Likewise.
6975 (output_profile_hook): Likewise.
6976 (rs6000_trampoline_init): Likewise.
6977 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
6978 * config/sh/sh.c (sh_trampoline_init): Likewise.
6979 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
6980 (sparc_emit_float_lib_cmp): Likewise.
6981 (sparc32_initialize_trampoline): Likewise.
6982 (sparc64_initialize_trampoline): Likewise.
6983 (sparc_profile_hook): Likewise.
6984 * config/spu/spu.c (ea_load_store): Likewise.
6985 * config/spu/spu.md (floatunssidf2): Likewise.
6986 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
6987 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
6988 * config/visium/visium.c (expand_block_move_4): Likewise.
6989 (expand_block_move_2): Likewise.
6990 (expand_block_move_1): Likewise.
6991 (expand_block_set_4): Likewise.
6992 (expand_block_set_2): Likewise.
6993 (expand_block_set_1): Likewise.
6994 (visium_trampoline_init): Likewise.
6995 (visium_profile_hook): Likewise.
6996 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
6997 (xtensa_setup_frame_addresses): Likewise.
6998 (xtensa_trampoline_init): Likewise.
6999 * except.c (sjlj_emit_function_enter): Likewise.
7000 (sjlj_emit_function_exit): Likewise.
7001 * explow.c (allocate_dynamic_stack_space): Likewise.
7002 (probe_stack_range): Likewise.
7003 * expr.c (convert_mode_scalar): Likewise.
7004 * optabs.c (expand_binop): Likewise.
7005 (expand_twoval_binop_libfunc): Likewise.
7006 (expand_unop): Likewise.
7007 (prepare_cmp_insn): Likewise.
7008 (prepare_float_lib_cmp): Likewise.
7009 (expand_float): Likewise.
7010 (expand_fix): Likewise.
7011 (expand_fixed_convert): Likewise.
7012 (maybe_emit_sync_lock_test_and_set): Likewise.
7013 (expand_atomic_compare_and_swap): Likewise.
7014 (expand_mem_thread_fence): Likewise.
7015 (expand_atomic_fetch_op): Likewise.
7017 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
7019 * doc/generic.texi (OpenACC): Adjust URL.
7020 * doc/invoke.texi (C Dialect Options): Ditto.
7022 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
7024 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
7025 predicate for operand 1. Add (m,<S>) constraint.
7026 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
7027 Prevent memory operand 1 with register operand 2.
7029 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
7031 PR rtl-optimization/82024
7032 * combine.c (try_combine): If the combination result is a PARALLEL,
7033 and we only need to retain the SET in there that would be placed
7034 at I2, check that we can place that at I3 instead, before doing so.
7036 2017-09-01 Jakub Jelinek <jakub@redhat.com>
7039 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
7041 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
7042 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
7045 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
7046 Jeff Law <law@redhat.com>
7048 * varasm.c (bss_initializer_p): Do not put constants into .bss
7049 (categorize_decl_for_section): Handle bss_initializer_p returning
7050 false when DECL_INITIAL is NULL.
7052 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7055 * config/s390/s390.c (s390_can_inline_p): New function.
7057 2017-09-01 Jeff Law <law@redhat.com>
7059 PR tree-optimization/82052
7060 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
7061 Always initialize the returned slot after a hash table miss
7062 when INSERT is true.
7064 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
7066 * config/s390/s390.md (mem_signal_fence): Remove.
7067 * doc/md.texi (mem_signal_fence): Remove.
7068 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
7070 * target-insns.def (mem_signal_fence): Remove.
7072 2017-09-01 Jakub Jelinek <jakub@redhat.com>
7075 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
7078 * asan.c (create_odr_indicator): Strip name encoding from assembler
7079 name before appending it after __odr_asan_.
7081 2017-09-01 Martin Liska <mliska@suse.cz>
7083 PR tree-optimization/82059
7084 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
7085 frequency only when an edge is redirected.
7087 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
7089 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
7090 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
7091 (arc_conditional_register_usage): Remove ARC600 lp_count
7093 (arc_file_start): Emit Tag_ARC_CPU_variation.
7094 (arc_can_use_doloop_p): New conditions to use ZOLs.
7095 (hwloop_fail): New function.
7096 (hwloop_optimize): Likewise.
7097 (hwloop_pattern_reg): Likewise.
7098 (arc_doloop_hooks): New struct, to be used with reorg_loops.
7099 (arc_reorg_loops): New function, calls reorg_loops.
7100 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
7101 (arc600_corereg_hazard): Remove ZOL checking, case handled by
7103 (arc_loop_hazard): Remove function, functionality moved into
7105 (arc_hazard): Remove arc_loop_hazard call.
7106 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
7107 into hwloop_optimize.
7108 (arc_label_align): Remove ZOL handling.
7109 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
7110 * config/arc/arc.md (doloop_begin): Remove pattern.
7111 (doloop_begin_i): Likewise.
7112 (doloop_end_i): Likewise.
7113 (doloop_fallback): Likewise.
7114 (doloop_fallback_m): Likewise.
7115 (doloop_end): Reimplement expand.
7116 (arc_lp): New pattern for LP instruction.
7117 (loop_end): New pattern.
7118 (loop_fail): Likewise.
7119 (decrement_and_branch_until_zero): Likewise.
7120 * config/arc/arc.opt (mlpc-width): New option.
7121 * doc/invoke.texi (mlpc-width): Document option.
7123 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
7125 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
7126 (arc_ccfsm_advance): Fix checking for delay slots.
7127 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
7129 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
7131 * config/arc/arc.md (movqi_insn): Add stores to save constant long
7133 (movhi_insn): Update store instruction constraint which are saving
7134 6-bit short immediates.
7135 (movsi_insn): Consider also short scaled load operations.
7136 (zero_extendhisi2_i): Use Usd constraint instead of T.
7137 (extendhisi2_i): Add q constraint.
7138 (arc_clzsi2): Add type and length attributes.
7139 (arc_ctzsi2): Likewise.
7140 * config/arc/constraints.md (Usc): Update constraint, the
7141 assembler can parse two relocations for a single instruction.
7143 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
7145 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
7146 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
7148 2017-08-31 Olivier Hainque <hainque@adacore.com>
7150 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
7151 match as powerpc-wrs-vxworks*.
7153 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
7155 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
7156 register constraint for by-element operand.
7157 (aarch64_mls_elt_merge<mode>): Likewise.
7159 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
7161 * config/arc/arc.c (arc_can_follow_jump): Check for short
7164 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
7166 * config.gcc: Use g.opt for arc.
7167 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
7168 functionality moved to ...
7169 (legitimate_scaled_address_p): New function, ...here.
7170 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
7171 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
7172 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
7174 (arc_override_options): Handle G option.
7175 (arc_output_pic_addr_const): Correct function definition.
7176 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
7177 (arc_decl_anon_ns_mem_p): Delete.
7178 (arc_in_small_data_p): Overhaul this function to take into
7179 consideration the value given via G option.
7180 (arc_rewrite_small_data_1): Renamed and corrected old
7181 arc_rewrite_small_data function.
7182 (arc_rewrite_small_data): New function.
7183 (small_data_pattern): Don't use pic_offset_table_rtx.
7184 * config/arc/arc.h (CC1_SPEC): Recognize G option.
7185 * config/arc/simdext.md (movmisalignv2hi): Use
7186 prepare_move_operands function.
7188 (movmisalign*): Likewise.
7189 * doc/invoke.texi (ARC options): Document -G option.
7191 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
7193 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
7195 * config/arc/arc.c (arc_print_operand): Output scalled address for
7196 sdata whenever is possible.
7197 (arc_in_small_data_p): Allow sdata for 64bit datum when double
7198 load/stores are available.
7199 (compact_sda_memory_operand): Check for the alignment required by
7200 code density instructions.
7201 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
7203 * config/arc/constraints.md (Usd): Update constraint.
7204 (Us0): New constraint.
7205 (Usc): Update constraint.
7207 2017-08-31 Richard Biener <rguenther@suse.de>
7210 * dwarf2out.c (dwarf2out_early_global_decl): Process each
7213 2017-08-31 Tamar Christina <tamar.christina@arm.com>
7215 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
7216 Resize type_signature.
7218 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
7219 Alan Hayward <alan.hayward@arm.com>
7220 David Sherwood <david.sherwood@arm.com>
7222 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
7223 subregs whose inner modes can be stored in GPRs.
7224 (aarch64_classify_index): Likewise.
7226 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
7227 Alan Hayward <alan.hayward@arm.com>
7228 David Sherwood <david.sherwood@arm.com>
7230 * config/aarch64/iterators.md (V_cmp_result): Rename to...
7231 (V_INT_EQUIV): ...this.
7232 (v_cmp_result): Rename to...
7233 (v_int_equiv): ...this.
7234 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
7235 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
7236 (copysign<mode>3): Likewise.
7237 (aarch64_simd_bsl<mode>_internal): Likewise.
7238 (aarch64_simd_bsl<mode>): Likewise.
7239 (vec_cmp<mode><mode>): Likewise.
7240 (vcond<mode><mode>): Likewise.
7241 (vcond<v_cmp_mixed><mode>): Likewise.
7242 (vcondu<mode><v_cmp_mixed>): Likewise.
7243 (aarch64_cm<optab><mode>): Likewise.
7244 (aarch64_cmtst<mode>): Likewise.
7245 (aarch64_fac<optab><mode>): Likewise.
7246 (vec_perm_const<mode>): Likewise.
7247 (vcond_mask_<mode><v_cmp_result>): Rename to...
7248 (vcond_mask_<mode><v_int_equiv>): ...this.
7249 (vec_cmp<mode><v_cmp_result>): Rename to...
7250 (vec_cmp<mode><v_int_equiv>): ...this.
7252 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
7253 Alan Hayward <alan.hayward@arm.com>
7254 David Sherwood <david.sherwood@arm.com>
7256 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
7258 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
7259 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
7260 (UNSPEC_LD4_DREG): New unspecs.
7261 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
7262 (aarch64_ld2<mode>_dreg_be): Replace with...
7263 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
7265 (aarch64_ld3<mode>_dreg_le)
7266 (aarch64_ld3<mode>_dreg_be): Replace with...
7267 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
7269 (aarch64_ld4<mode>_dreg_le)
7270 (aarch64_ld4<mode>_dreg_be): Replace with...
7271 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
7274 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7276 PR tree-optimization/81987
7277 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
7278 insert an initializer in a location not dominated by the stride
7281 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
7283 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
7284 on the entire header of the finally block in the fallthru case.
7286 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
7288 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
7290 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
7292 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
7293 rs6000_emit_move_from_cr and call renamed function.
7294 (rs6000_emit_prologue): Call renamed functions.
7295 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
7296 movesi_from_cr, remove volatile CRs.
7298 2017-08-30 Jon Beniston <jon@beniston.com>
7299 Richard Biener <rguenther@suse.de>
7301 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
7302 of VECTOR_MODE_P check.
7303 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
7304 element vector types.
7306 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7308 * df.h (df_read_modify_subreg_p): Remove in favor of...
7309 * rtl.h (read_modify_subreg_p): ...this new function. Take a
7310 const_rtx instead of an rtx.
7311 * cprop.c (local_cprop_find_used_regs): Update accordingly.
7312 * df-problems.c (df_word_lr_mark_ref): Likewise.
7313 * ira-lives.c (mark_pseudo_reg_live): Likewise.
7314 (mark_pseudo_reg_dead): Likewise.
7315 (mark_ref_dead): Likewise.
7316 * reginfo.c (init_subregs_of_mode): Likewise.
7317 * sched-deps.c (sched_analyze_1): Likewise.
7318 * df-scan.c (df_def_record_1): Likewise.
7319 (df_uses_record): Likewise.
7320 (df_read_modify_subreg_p): Remove in favor of...
7321 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
7322 const_rtx instead of an rtx.
7324 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7325 Alan Hayward <alan.hayward@arm.com>
7326 David Sherwood <david.sherwood@arm.com>
7328 * rtl.h (partial_subreg_p): New function.
7329 * caller-save.c (save_call_clobbered_regs): Use it.
7330 * calls.c (expand_call): Likewise.
7331 * combine.c (combinable_i3pat): Likewise.
7332 (simplify_set): Likewise.
7333 (make_extraction): Likewise.
7334 (make_compound_operation_int): Likewise.
7335 (gen_lowpart_or_truncate): Likewise.
7336 (force_to_mode): Likewise.
7337 (make_field_assignment): Likewise.
7338 (reg_truncated_to_mode): Likewise.
7339 (record_truncated_value): Likewise.
7340 (move_deaths): Likewise.
7341 * cse.c (record_jump_cond): Likewise.
7342 (cse_insn): Likewise.
7343 * cselib.c (cselib_lookup_1): Likewise.
7344 * expmed.c (extract_bit_field_using_extv): Likewise.
7345 * function.c (assign_parm_setup_reg): Likewise.
7346 * ifcvt.c (noce_convert_multiple_sets): Likewise.
7347 * ira-build.c (create_insn_allocnos): Likewise.
7348 * lra-coalesce.c (merge_pseudos): Likewise.
7349 * lra-constraints.c (match_reload): Likewise.
7350 (simplify_operand_subreg): Likewise.
7351 (curr_insn_transform): Likewise.
7352 * lra-lives.c (process_bb_lives): Likewise.
7353 * lra.c (new_insn_reg): Likewise.
7354 (lra_substitute_pseudo): Likewise.
7355 * regcprop.c (mode_change_ok): Likewise.
7356 (maybe_mode_change): Likewise.
7357 (copyprop_hardreg_forward_1): Likewise.
7358 * reload.c (push_reload): Likewise.
7359 (find_reloads): Likewise.
7360 (find_reloads_subreg_address): Likewise.
7361 * reload1.c (alter_reg): Likewise.
7362 (eliminate_regs_1): Likewise.
7363 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7365 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
7367 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
7368 back to if statements, including unpack.
7370 2017-08-30 Martin Liska <mliska@suse.cz>
7373 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
7375 (func_checker::compare_asm_inputs_outputs): ... this function.
7376 (func_checker::compare_gimple_asm): Use the function to compare
7377 also ASM constrains.
7378 * ipa-icf-gimple.h: Rename the function.
7380 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7381 Alan Hayward <alan.hayward@arm.com>
7382 David Sherwood <david.sherwood@arm.com>
7384 * coretypes.h (complex_mode): New type.
7385 * gdbhooks.py (build_pretty_printer): Handle it.
7386 * machmode.h (complex_mode): New class.
7387 (complex_mode::includes_p): New function.
7388 (is_complex_int_mode): Likewise.
7389 (is_complex_float_mode): Likewise.
7390 * genmodes.c (get_mode_class): Handle complex mode classes.
7391 * function.c (expand_function_end): Use is_complex_int_mode.
7393 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7394 Alan Hayward <alan.hayward@arm.com>
7395 David Sherwood <david.sherwood@arm.com>
7397 * coretypes.h (scalar_mode_pod): New typedef.
7398 * gdbhooks.py (build_pretty_printer): Handle it.
7399 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
7400 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
7401 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
7402 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
7405 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7406 Alan Hayward <alan.hayward@arm.com>
7407 David Sherwood <david.sherwood@arm.com>
7409 * machmode.h (mode_for_vector): Take a scalar_mode instead
7411 * stor-layout.c (mode_for_vector): Likewise.
7412 * explow.c (promote_mode): Use as_a <scalar_mode>.
7413 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
7415 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7416 Alan Hayward <alan.hayward@arm.com>
7417 David Sherwood <david.sherwood@arm.com>
7419 * target.def (preferred_simd_mode): Take a scalar_mode
7420 instead of a machine_mode.
7421 * targhooks.h (default_preferred_simd_mode): Likewise.
7422 * targhooks.c (default_preferred_simd_mode): Likewise.
7423 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
7424 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
7425 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
7426 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
7427 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
7428 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
7429 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
7430 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
7432 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
7433 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
7434 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
7435 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
7436 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
7437 * doc/tm.texi: Regenerate.
7438 * optabs-query.c (can_vec_mask_load_store_p): Return false for
7441 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7442 Alan Hayward <alan.hayward@arm.com>
7443 David Sherwood <david.sherwood@arm.com>
7445 * target.def (scalar_mode_supported_p): Take a scalar_mode
7446 instead of a machine_mode.
7447 * targhooks.h (default_scalar_mode_supported_p): Likewise.
7448 * targhooks.c (default_scalar_mode_supported_p): Likewise.
7449 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
7450 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
7451 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
7452 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
7453 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
7454 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
7455 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
7456 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
7457 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
7458 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
7459 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
7460 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
7462 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
7463 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
7464 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
7465 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
7466 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
7468 * doc/tm.texi: Regenerate.
7470 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7471 Alan Hayward <alan.hayward@arm.com>
7472 David Sherwood <david.sherwood@arm.com>
7474 * coretypes.h (opt_scalar_mode): New typedef.
7475 * gdbhooks.py (build_pretty_printers): Handle it.
7476 * machmode.h (mode_iterator::get_2xwider): Add overload for
7478 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
7480 * expr.c (convert_mode_scalar): Likewise.
7481 * omp-low.c (omp_clause_aligned_alignment): Likewise.
7482 * optabs.c (expand_float): Likewise.
7483 (expand_fix): Likewise.
7484 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
7486 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7487 Alan Hayward <alan.hayward@arm.com>
7488 David Sherwood <david.sherwood@arm.com>
7490 * optabs.c (expand_float): Explicitly check for scalars before
7491 using a branching expansion.
7492 (expand_fix): Likewise.
7494 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7495 Alan Hayward <alan.hayward@arm.com>
7496 David Sherwood <david.sherwood@arm.com>
7498 * expr.c (convert_mode): Split scalar handling out into...
7499 (convert_mode_scalar): ...this new function. Treat the modes
7502 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7503 Alan Hayward <alan.hayward@arm.com>
7504 David Sherwood <david.sherwood@arm.com>
7506 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
7508 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
7510 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7511 Alan Hayward <alan.hayward@arm.com>
7512 David Sherwood <david.sherwood@arm.com>
7514 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
7515 rather than a machine_mode.
7516 (fixed_from_string): Likewise.
7517 (fixed_convert): Likewise.
7518 (fixed_convert_from_int): Likewise.
7519 (fixed_convert_from_real): Likewise.
7520 (real_convert_from_fixed): Likewise.
7521 * fixed-value.c (fixed_from_double_int): Likewise.
7522 (fixed_from_string): Likewise.
7523 (fixed_convert): Likewise.
7524 (fixed_convert_from_int): Likewise.
7525 (fixed_convert_from_real): Likewise.
7526 (real_convert_from_fixed): Likewise.
7527 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
7529 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7530 Alan Hayward <alan.hayward@arm.com>
7531 David Sherwood <david.sherwood@arm.com>
7533 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
7534 of separate mode class checks. Do not allow vector modes here.
7535 (immed_wide_int_const): Use as_a <scalar_mode>.
7536 * explow.c (trunc_int_for_mode): Likewise.
7537 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
7538 (wi::shwi): Likewise.
7539 (wi::min_value): Likewise.
7540 (wi::max_value): Likewise.
7541 * dwarf2out.c (loc_descriptor): Likewise.
7542 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
7545 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7546 Alan Hayward <alan.hayward@arm.com>
7547 David Sherwood <david.sherwood@arm.com>
7549 * tree.h (SCALAR_TYPE_MODE): New macro.
7550 * expr.c (expand_expr_addr_expr_1): Use it.
7551 (expand_expr_real_2): Likewise.
7552 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
7553 (fold_convert_const_fixed_from_int): Likewise.
7554 (fold_convert_const_fixed_from_real): Likewise.
7555 (native_encode_fixed): Likewise
7556 (native_encode_complex): Likewise
7557 (native_encode_vector): Likewise.
7558 (native_interpret_fixed): Likewise.
7559 (native_interpret_real): Likewise.
7560 (native_interpret_complex): Likewise.
7561 (native_interpret_vector): Likewise.
7562 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
7563 (simd_clone_adjust_argument_types): Likewise.
7564 (simd_clone_init_simd_arrays): Likewise.
7565 (simd_clone_adjust): Likewise.
7566 * stor-layout.c (layout_type): Likewise.
7567 * tree.c (build_minus_one_cst): Likewise.
7568 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
7569 * tree-inline.c (estimate_move_cost): Likewise.
7570 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
7571 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
7572 (vectorizable_reduction): Likewise.
7573 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
7574 (vect_recog_mixed_size_cond_pattern): Likewise.
7575 (check_bool_pattern): Likewise.
7576 (adjust_bool_pattern): Likewise.
7577 (search_type_for_mask_1): Likewise.
7578 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
7579 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
7580 (vectorizable_load): Likewise.
7581 (vectorizable_store): Likewise.
7582 * ubsan.c (ubsan_encode_value): Likewise.
7583 * varasm.c (output_constant): Likewise.
7585 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7586 Alan Hayward <alan.hayward@arm.com>
7587 David Sherwood <david.sherwood@arm.com>
7589 * coretypes.h (scalar_mode): New class.
7590 * machmode.h (scalar_mode): Likewise.
7591 (scalar_mode::includes_p): New function.
7592 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
7593 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
7594 * genmodes.c (get_mode_class): Handle remaining scalar modes.
7595 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
7596 * expmed.c (store_bit_field_1): Likewise.
7597 (extract_bit_field_1): Likewise.
7598 * expr.c (write_complex_part): Likewise.
7599 (read_complex_part): Likewise.
7600 (emit_move_complex_push): Likewise.
7601 (expand_expr_real_2): Likewise.
7602 * function.c (assign_parm_setup_reg): Likewise.
7603 (assign_parms_unsplit_complex): Likewise.
7604 * optabs.c (expand_binop): Likewise.
7605 * rtlanal.c (subreg_get_info): Likewise.
7606 * simplify-rtx.c (simplify_immed_subreg): Likewise.
7607 * varasm.c (output_constant_pool_2): Likewise.
7609 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7610 Alan Hayward <alan.hayward@arm.com>
7611 David Sherwood <david.sherwood@arm.com>
7613 * expmed.c (extract_high_half): Use scalar_int_mode and remove
7615 (expmed_mult_highpart_optab): Likewise.
7616 (expmed_mult_highpart): Likewise.
7618 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7619 Alan Hayward <alan.hayward@arm.com>
7620 David Sherwood <david.sherwood@arm.com>
7622 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
7623 instead of a machine_mode.
7624 (builtin_memset_read_str): Likewise.
7625 * builtins.c (c_readstr): Likewise.
7626 (builtin_memcpy_read_str): Likewise.
7627 (builtin_strncpy_read_str): Likewise.
7628 (builtin_memset_read_str): Likewise.
7629 (builtin_memset_gen_str): Likewise.
7630 (expand_builtin_signbit): Use scalar_int_mode for local variables.
7631 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
7632 instead of a machine_mode.
7633 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
7635 (make_extraction): Likewise.
7636 (try_widen_shift_mode): Take and return scalar_int_modes instead
7638 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
7639 a scalar_int_mode instead of a machine_mode.
7640 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
7641 (avr_addr_space_pointer_mode): Likewise.
7642 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
7643 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
7644 (msp430_unwind_word_mode): Likewise.
7645 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
7646 (spu_addr_space_pointer_mode): Likewise.
7647 (spu_addr_space_address_mode): Likewise.
7648 (spu_libgcc_cmp_return_mode): Likewise.
7649 (spu_libgcc_shift_count_mode): Likewise.
7650 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
7651 (rl78_addr_space_pointer_mode): Likewise.
7652 (fl78_unwind_word_mode): Likewise.
7653 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
7655 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
7656 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
7657 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
7658 (mips_valid_pointer_mode): Likewise.
7659 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
7660 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
7661 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
7663 (ft32_addr_space_address_mode): Likewise.
7664 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
7665 scalar_int_mode instead of a machine_mode.
7666 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
7668 (m32c_addr_space_address_mode): Likewise.
7669 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
7670 (rs6000_eh_return_filter_mode): Likewise.
7671 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
7672 (rs6000_eh_return_filter_mode): Likewise.
7673 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
7674 (s390_libgcc_shift_count_mode): Likewise.
7675 (s390_unwind_word_mode): Likewise.
7676 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
7678 * target.def (mode_rep_extended): Likewise.
7679 (valid_pointer_mode): Likewise.
7680 (addr_space.valid_pointer_mode): Likewise.
7681 (eh_return_filter_mode): Return a scalar_int_mode rather than
7683 (libgcc_cmp_return_mode): Likewise.
7684 (libgcc_shift_count_mode): Likewise.
7685 (unwind_word_mode): Likewise.
7686 (addr_space.pointer_mode): Likewise.
7687 (addr_space.address_mode): Likewise.
7688 * doc/tm.texi: Regenerate.
7689 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
7691 (do_jump): Use scalar_int_mode for local variables.
7692 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
7693 rather than a machine_mode.
7694 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
7695 (scompare_loc_descriptor_wide): Likewise.
7696 (scompare_loc_descriptor_narrow): Likewise.
7697 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
7699 * except.c (sjlj_emit_dispatch_table): Likewise.
7700 (expand_builtin_eh_copy_values): Likewise.
7701 * explow.c (convert_memory_address_addr_space_1): Likewise.
7702 Take a scalar_int_mode rather than a machine_mode.
7703 (convert_memory_address_addr_space): Take a scalar_int_mode rather
7704 than a machine_mode.
7705 (memory_address_addr_space): Use scalar_int_mode for local variables.
7706 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
7707 rather than a machine_mode.
7708 * expmed.c (mask_rtx): Likewise.
7709 (init_expmed_one_conv): Likewise.
7710 (expand_mult_highpart_adjust): Likewise.
7711 (extract_high_half): Likewise.
7712 (expmed_mult_highpart_optab): Likewise.
7713 (expmed_mult_highpart): Likewise.
7714 (expand_smod_pow2): Likewise.
7715 (expand_sdiv_pow2): Likewise.
7716 (emit_store_flag_int): Likewise.
7717 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
7719 (extract_low_bits): Likewise.
7720 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
7722 * expr.c (pieces_addr::adjust): Likewise.
7723 (can_store_by_pieces): Likewise.
7724 (store_by_pieces): Likewise.
7725 (clear_by_pieces_1): Likewise.
7726 (expand_expr_addr_expr_1): Likewise.
7727 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
7728 (expand_expr_real_1): Likewise.
7729 (try_casesi): Likewise.
7730 * final.c (shorten_branches): Likewise.
7731 * fold-const.c (fold_convert_const_int_from_fixed): Change the
7732 type of "mode" to machine_mode.
7733 * internal-fn.c (expand_arith_overflow_result_store): Take a
7734 scalar_int_mode rather than a machine_mode.
7735 (expand_mul_overflow): Use scalar_int_mode for local variables.
7736 * loop-doloop.c (doloop_modify): Likewise.
7737 (doloop_optimize): Likewise.
7738 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
7739 than a machine_mode.
7740 (expand_doubleword_shift_condmove): Likewise.
7741 (expand_doubleword_shift): Likewise.
7742 (expand_doubleword_clz): Likewise.
7743 (expand_doubleword_popcount): Likewise.
7744 (expand_doubleword_parity): Likewise.
7745 (expand_absneg_bit): Use scalar_int_mode for local variables.
7746 (prepare_float_lib_cmp): Likewise.
7747 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
7748 rather than a machine_mode.
7749 (convert_memory_address_addr_space): Likewise.
7750 (get_mode_bounds): Likewise.
7751 (get_address_mode): Return a scalar_int_mode rather than a
7753 * rtlanal.c (get_address_mode): Likewise.
7754 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
7755 than a machine_mode.
7756 * targhooks.c (default_mode_rep_extended): Likewise.
7757 (default_valid_pointer_mode): Likewise.
7758 (default_addr_space_valid_pointer_mode): Likewise.
7759 (default_eh_return_filter_mode): Return a scalar_int_mode rather
7760 than a machine_mode.
7761 (default_libgcc_cmp_return_mode): Likewise.
7762 (default_libgcc_shift_count_mode): Likewise.
7763 (default_unwind_word_mode): Likewise.
7764 (default_addr_space_pointer_mode): Likewise.
7765 (default_addr_space_address_mode): Likewise.
7766 * targhooks.h (default_eh_return_filter_mode): Likewise.
7767 (default_libgcc_cmp_return_mode): Likewise.
7768 (default_libgcc_shift_count_mode): Likewise.
7769 (default_unwind_word_mode): Likewise.
7770 (default_addr_space_pointer_mode): Likewise.
7771 (default_addr_space_address_mode): Likewise.
7772 (default_mode_rep_extended): Take a scalar_int_mode rather than
7774 (default_valid_pointer_mode): Likewise.
7775 (default_addr_space_valid_pointer_mode): Likewise.
7776 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
7778 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
7779 rather than a machine_mode.
7780 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
7781 for local variables.
7782 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
7783 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
7784 than a machine_mode.
7786 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7787 Alan Hayward <alan.hayward@arm.com>
7788 David Sherwood <david.sherwood@arm.com>
7790 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
7791 the mode argument to scalar_int_mode.
7792 (do_jump_by_parts_zero_rtx): Likewise.
7793 (do_jump_by_parts_equality_rtx): Likewise.
7794 (do_jump_by_parts_greater): Take a mode argument.
7795 (do_jump_by_parts_equality): Likewise.
7796 (do_jump_1): Update calls accordingly.
7798 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7799 Alan Hayward <alan.hayward@arm.com>
7800 David Sherwood <david.sherwood@arm.com>
7802 * is-a.h (safe_dyn_cast): New function.
7803 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
7804 (jump_table_for_label): Likewise.
7805 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
7806 instead of an rtx_insn *.
7807 (shorten_branches): Use dyn_cast instead of LABEL_P and
7808 JUMP_TABLE_DATA_P. Use jump_table_for_label and
7809 rtx_jump_table_data::get_data_mode.
7810 (final_scan_insn): Likewise.
7812 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7813 Alan Hayward <alan.hayward@arm.com>
7814 David Sherwood <david.sherwood@arm.com>
7816 * combine.c (try_combine): Use is_a <scalar_int_mode> when
7817 trying to combine a full-register integer set with a subreg
7820 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7821 Alan Hayward <alan.hayward@arm.com>
7822 David Sherwood <david.sherwood@arm.com>
7824 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
7825 that is always either address_mode or pointer_mode.
7827 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7828 Alan Hayward <alan.hayward@arm.com>
7829 David Sherwood <david.sherwood@arm.com>
7831 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
7832 when the two are known to be equal.
7834 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7835 Alan Hayward <alan.hayward@arm.com>
7836 David Sherwood <david.sherwood@arm.com>
7838 * simplify-rtx.c (simplify_const_unary_operation): Use
7839 is_a <scalar_int_mode> instead of checking for a nonzero
7840 precision. Forcibly convert op_mode to a scalar_int_mode
7841 in that case. More clearly differentiate the operand and
7842 result modes and use the former when deciding what the value
7843 of a count-bits operation should be. Use is_int_mode instead
7844 of checking for a MODE_INT. Remove redundant check for whether
7845 this mode has a zero precision.
7847 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7848 Alan Hayward <alan.hayward@arm.com>
7849 David Sherwood <david.sherwood@arm.com>
7851 * optabs.c (widen_leading): Change the type of the mode argument
7852 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
7853 (widen_bswap): Likewise.
7854 (expand_parity): Likewise.
7855 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
7856 (expand_ffs): Likewise.
7857 (epand_unop): Check for scalar integer modes before calling the
7860 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7861 Alan Hayward <alan.hayward@arm.com>
7862 David Sherwood <david.sherwood@arm.com>
7864 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
7866 (expand_expr_real_1): Update call accordingly.
7868 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7869 Alan Hayward <alan.hayward@arm.com>
7870 David Sherwood <david.sherwood@arm.com>
7872 * expmed.c (store_bit_field_using_insv): Add op0_mode and
7873 value_mode arguments. Use scalar_int_mode internally.
7874 (store_bit_field_1): Rename the new integer mode from imode
7875 to op0_mode and use it instead of GET_MODE (op0). Update calls
7876 to store_split_bit_field, store_bit_field_using_insv and
7877 store_fixed_bit_field.
7878 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
7879 Use scalar_int_mode internally. Use a bit count rather than a mode
7880 when calculating the largest bit size for get_best_mode.
7881 Update calls to store_split_bit_field and store_fixed_bit_field_1.
7882 (store_fixed_bit_field_1): Add mode and value_mode arguments.
7883 Remove assertion that OP0 has a scalar integer mode.
7884 (store_split_bit_field): Add op0_mode and value_mode arguments.
7885 Update calls to extract_fixed_bit_field.
7886 (extract_bit_field_using_extv): Add an op0_mode argument.
7887 Use scalar_int_mode internally.
7888 (extract_bit_field_1): Rename the new integer mode from imode to
7889 op0_mode and use it instead of GET_MODE (op0). Update calls to
7890 extract_split_bit_field, extract_bit_field_using_extv and
7891 extract_fixed_bit_field.
7892 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
7893 to extract_split_bit_field and extract_fixed_bit_field_1.
7894 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
7895 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
7897 (extract_split_bit_field): Add an op0_mode argument. Update call
7898 to extract_fixed_bit_field.
7900 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7901 Alan Hayward <alan.hayward@arm.com>
7902 David Sherwood <david.sherwood@arm.com>
7904 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
7905 * explow.c (hard_function_value): Likewise.
7906 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
7907 convert_to_mode call outside the loop.
7908 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
7909 for the mode iterator. Require the mode specified by max_pieces
7911 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
7913 (copy_blkmode_to_reg): Likewise.
7914 (set_storage_via_setmem): Likewise.
7915 * optabs.c (prepare_cmp_insn): Likewise.
7916 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
7917 * stor-layout.c (finish_bitfield_representative): Likewise.
7919 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7920 Alan Hayward <alan.hayward@arm.com>
7921 David Sherwood <david.sherwood@arm.com>
7923 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
7924 * expr.c (convert_move): Use them.
7925 (convert_modes): Likewise.
7926 (store_expr_with_bounds): Likewise.
7928 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7929 Alan Hayward <alan.hayward@arm.com>
7930 David Sherwood <david.sherwood@arm.com>
7932 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
7933 parameter for the mode of "x". Remove the "known_x", "known_mode"
7934 and "known_ret" arguments. Change the type of the mode argument
7936 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
7937 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
7938 (reg_num_sign_bit_copies_for_combine): Likewise.
7939 * rtlanal.c (nonzero_bits1): Likewise.
7940 (num_sign_bit_copies1): Likewise.
7941 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
7942 (reg_num_sign_bit_copies_general): Likewise.
7943 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
7944 (reg_nonzero_bits_general): Likewise.
7946 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7947 Alan Hayward <alan.hayward@arm.com>
7948 David Sherwood <david.sherwood@arm.com>
7950 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
7951 than in subroutines. Return 1 for non-integer modes.
7952 (cached_num_sign_bit_copies): Change the type of the mode parameter
7954 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
7955 classes. Handle CONST_INT_P first and then check whether X also
7956 has a scalar integer mode. Check the same thing for inner registers
7957 of a SUBREG and for values that are being extended or truncated.
7959 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7960 Alan Hayward <alan.hayward@arm.com>
7961 David Sherwood <david.sherwood@arm.com>
7963 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
7964 in subroutines. Return the mode mask for non-integer modes.
7965 (cached_nonzero_bits): Change the type of the mode parameter
7967 (nonzero_bits1): Likewise. Remove early exit for other mode
7968 classes. Handle CONST_INT_P first and then check whether X
7969 also has a scalar integer mode.
7971 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7972 Alan Hayward <alan.hayward@arm.com>
7973 David Sherwood <david.sherwood@arm.com>
7975 * expr.c (widest_int_mode_for_size): Make the comment match the code.
7976 Return a scalar_int_mode and assert that the size is greater than
7978 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
7979 (op_by_pieces_d::op_by_pieces_d): Likewise.
7980 (op_by_pieces_d::run): Likewise.
7981 (can_store_by_pieces): Likewise.
7983 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7984 Alan Hayward <alan.hayward@arm.com>
7985 David Sherwood <david.sherwood@arm.com>
7987 * combine.c (extract_left_shift): Add a mode argument and update
7989 (make_compound_operation_int): Change the type of the mode parameter
7990 to scalar_int_mode and update the call to extract_left_shift.
7992 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7993 Alan Hayward <alan.hayward@arm.com>
7994 David Sherwood <david.sherwood@arm.com>
7996 * combine.c (simplify_and_const_int): Change the type of the mode
7997 parameter to scalar_int_mode.
7998 (simplify_and_const_int_1): Likewise. Update recursive call.
8000 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8001 Alan Hayward <alan.hayward@arm.com>
8002 David Sherwood <david.sherwood@arm.com>
8004 * combine.c (simplify_compare_const): Check that the mode is a
8005 scalar_int_mode (rather than VOIDmode) before testing its
8007 (simplify_comparison): Move COMPARISON_P handling out of the
8008 loop and restrict the latter part of the loop to scalar_int_modes.
8009 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
8010 and when considering SUBREG_REGs. Use is_int_mode instead of
8011 checking GET_MODE_CLASS against MODE_INT.
8013 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8014 Alan Hayward <alan.hayward@arm.com>
8015 David Sherwood <david.sherwood@arm.com>
8017 * combine.c (try_widen_shift_mode): Move check for equal modes to...
8018 (simplify_shift_const_1): ...here. Use scalar_int_mode for
8019 shift_unit_mode and for modes involved in scalar shifts.
8021 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8022 Alan Hayward <alan.hayward@arm.com>
8023 David Sherwood <david.sherwood@arm.com>
8025 * combine.c (force_int_to_mode): New function, split out from...
8026 (force_to_mode): ...here. Keep xmode up-to-date and use it
8027 instead of GET_MODE (x).
8029 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8030 Alan Hayward <alan.hayward@arm.com>
8031 David Sherwood <david.sherwood@arm.com>
8033 * optabs-query.h (extraction_insn::struct_mode): Change type to
8034 opt_scalar_int_mode and update comment.
8035 (extraction_insn::field_mode): Change type to scalar_int_mode.
8036 (extraction_insn::pos_mode): Likewise.
8037 * combine.c (make_extraction): Update accordingly.
8038 * optabs-query.c (get_traditional_extraction_insn): Likewise.
8039 (get_optab_extraction_insn): Likewise.
8040 * recog.c (simplify_while_replacing): Likewise.
8041 * expmed.c (narrow_bit_field_mem): Change the type of the mode
8042 parameter to opt_scalar_int_mode.
8044 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8045 Alan Hayward <alan.hayward@arm.com>
8046 David Sherwood <david.sherwood@arm.com>
8048 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
8049 to a scalar_int_mode instead of a machine_mode.
8050 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
8051 (get_best_mode): Return a boolean and use a pointer argument to store
8052 the selected mode. Replace the limit mode parameter with a bit limit.
8053 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
8054 for the values returned by bit_field_mode_iterator::next_mode.
8055 (store_bit_field): Update call to get_best_mode.
8056 (store_fixed_bit_field): Likewise.
8057 (extract_fixed_bit_field): Likewise.
8058 * expr.c (optimize_bitfield_assignment_op): Likewise.
8059 * fold-const.c (optimize_bit_field_compare): Likewise.
8060 (fold_truth_andor_1): Likewise.
8061 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
8062 Update for new type of m_mode.
8063 (get_best_mode): As above.
8065 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8066 Alan Hayward <alan.hayward@arm.com>
8067 David Sherwood <david.sherwood@arm.com>
8069 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
8070 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
8071 (store_bit_field): Check is_a <scalar_int_mode> before calling
8072 strict_volatile_bitfield_p.
8073 (extract_bit_field): Likewise.
8075 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8076 Alan Hayward <alan.hayward@arm.com>
8077 David Sherwood <david.sherwood@arm.com>
8079 * target.def (cstore_mode): Return a scalar_int_mode.
8080 * doc/tm.texi: Regenerate.
8081 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
8082 * targhooks.h (default_cstore_mode): Likewise.
8083 * targhooks.c (default_cstore_mode): Likewise, using a forced
8085 * expmed.c (emit_cstore): Expect the target of the cstore to be
8088 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8089 Alan Hayward <alan.hayward@arm.com>
8090 David Sherwood <david.sherwood@arm.com>
8092 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
8094 (niter_desc): Likewise mode.
8095 (iv_analyze): Add a mode parameter.
8097 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
8098 and change its type to scalar_int_mode.
8099 * loop-iv.c: Update commentary at head of file.
8100 (iv_constant): Pass the mode paraeter before the rtx it describes
8101 and change its type to scalar_int_mode. Remove VOIDmode handling.
8102 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
8103 (iv_extend): Likewise.
8104 (shorten_into_mode): Likewise.
8105 (iv_add): Use scalar_int_mode.
8106 (iv_mult): Likewise.
8107 (iv_shift): Likewise.
8108 (canonicalize_iv_subregs): Likewise.
8109 (get_biv_step_1): Pass the outer_mode parameter before the rtx
8110 it describes and change its mode to scalar_int_mode. Also change
8111 the type of the returned inner_mode to scalar_int_mode.
8112 (get_biv_step): Likewise, turning outer_mode from a pointer
8113 into a direct parameter. Update call to get_biv_step_1.
8114 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
8115 iv_constant and get_biv_step.
8116 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
8117 and change its type to scalar_int_mode. Don't initialise iv->mode
8118 to VOIDmode and remove later checks for its still being VOIDmode.
8119 Update calls to iv_analyze_op and iv_analyze_expr. Check
8120 is_a <scalar_int_mode> when changing the mode under consideration.
8121 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
8122 Update call to iv_analyze_expr.
8123 (iv_analyze_op): Add a mode parameter. Reject subregs whose
8124 inner register is not also a scalar_int_mode. Update call to
8126 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
8127 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
8128 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
8129 separate mode class checks. Update calls to iv_analyze. Remove
8130 fix-up of VOIDmodes after iv_analyze_biv.
8131 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
8132 don't have a scalar_int_mode. Update call to biv_p.
8134 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8135 Alan Hayward <alan.hayward@arm.com>
8136 David Sherwood <david.sherwood@arm.com>
8138 * cfgexpand.c (convert_debug_memory_address): Use
8139 as_a <scalar_int_mode>.
8140 * combine.c (expand_compound_operation): Likewise.
8141 (make_extraction): Likewise.
8142 (change_zero_ext): Likewise.
8143 (simplify_comparison): Likewise.
8144 * cse.c (cse_insn): Likewise.
8145 * dwarf2out.c (minmax_loc_descriptor): Likewise.
8146 (mem_loc_descriptor): Likewise.
8147 (loc_descriptor): Likewise.
8148 * expmed.c (init_expmed_one_mode): Likewise.
8149 (synth_mult): Likewise.
8150 (emit_store_flag_1): Likewise.
8151 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
8152 of a comparison with size.
8153 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
8154 (reduce_to_bit_field_precision): Likewise.
8155 * function.c (expand_function_end): Likewise.
8156 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
8157 * loop-doloop.c (doloop_modify): Likewise.
8158 * optabs.c (expand_binop): Likewise.
8159 (expand_unop): Likewise.
8160 (expand_copysign_absneg): Likewise.
8161 (prepare_cmp_insn): Likewise.
8162 (maybe_legitimize_operand): Likewise.
8163 * recog.c (const_scalar_int_operand): Likewise.
8164 * rtlanal.c (get_address_mode): Likewise.
8165 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8166 (simplify_cond_clz_ctz): Likewise.
8167 * tree-nested.c (get_nl_goto_field): Likewise.
8168 * tree.c (build_vector_type_for_mode): Likewise.
8169 * var-tracking.c (use_narrower_mode): Likewise.
8171 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8172 Alan Hayward <alan.hayward@arm.com>
8173 David Sherwood <david.sherwood@arm.com>
8175 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
8176 * builtins.c (expand_builtin_signbit): Use it.
8177 * cfgexpand.c (expand_debug_expr): Likewise.
8178 * dojump.c (do_jump): Likewise.
8179 (do_compare_and_jump): Likewise.
8180 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
8181 * expmed.c (make_tree): Likewise.
8182 * expr.c (expand_expr_real_2): Likewise.
8183 (expand_expr_real_1): Likewise.
8184 (try_casesi): Likewise.
8185 * fold-const-call.c (fold_const_call_ss): Likewise.
8186 * fold-const.c (unextend): Likewise.
8187 (extract_muldiv_1): Likewise.
8188 (fold_single_bit_test): Likewise.
8189 (native_encode_int): Likewise.
8190 (native_encode_string): Likewise.
8191 (native_interpret_int): Likewise.
8192 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
8193 * internal-fn.c (expand_addsub_overflow): Likewise.
8194 (expand_neg_overflow): Likewise.
8195 (expand_mul_overflow): Likewise.
8196 (expand_arith_overflow): Likewise.
8197 * match.pd: Likewise.
8198 * stor-layout.c (layout_type): Likewise.
8199 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
8200 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
8201 * tree-ssanames.c (get_range_info): Likewise.
8202 * tree-switch-conversion.c (array_value_type) Likewise.
8203 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
8204 (vect_recog_divmod_pattern): Likewise.
8205 (vect_recog_mixed_size_cond_pattern): Likewise.
8206 * tree-vrp.c (extract_range_basic): Likewise.
8207 (simplify_float_conversion_using_ranges): Likewise.
8208 * tree.c (int_fits_type_p): Likewise.
8209 * ubsan.c (instrument_bool_enum_load): Likewise.
8210 * varasm.c (mergeable_string_section): Likewise.
8211 (narrowing_initializer_constant_valid_p): Likewise.
8212 (output_constant): Likewise.
8214 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8215 Alan Hayward <alan.hayward@arm.com>
8216 David Sherwood <david.sherwood@arm.com>
8218 * machmode.h (NARROWEST_INT_MODE): New macro.
8219 * expr.c (alignment_for_piecewise_move): Use it instead of
8220 GET_CLASS_NARROWEST_MODE (MODE_INT).
8221 (push_block): Likewise.
8222 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
8224 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
8226 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8227 Alan Hayward <alan.hayward@arm.com>
8228 David Sherwood <david.sherwood@arm.com>
8230 * postreload.c (move2add_valid_value_p): Change the type of the
8231 mode parameter to scalar_int_mode.
8232 (move2add_use_add2_insn): Add a mode parameter and use it instead
8234 (move2add_use_add3_insn): Likewise.
8235 (reload_cse_move2add): Update accordingly.
8237 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8238 Alan Hayward <alan.hayward@arm.com>
8239 David Sherwood <david.sherwood@arm.com>
8241 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
8243 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
8244 * optabs.c (expand_unop): Likewise.
8246 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8247 Alan Hayward <alan.hayward@arm.com>
8248 David Sherwood <david.sherwood@arm.com>
8250 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
8251 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
8252 (popcount_loc_descriptor): Likewise.
8253 (bswap_loc_descriptor): Likewise.
8254 (rotate_loc_descriptor): Likewise.
8255 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
8256 calling the functions above.
8258 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8259 Alan Hayward <alan.hayward@arm.com>
8260 David Sherwood <david.sherwood@arm.com>
8262 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
8264 (try_combine): Likewise.
8265 (simplify_if_then_else): Likewise.
8266 * cse.c (cse_insn): Likewise.
8267 * dwarf2out.c (mem_loc_descriptor): Likewise.
8268 * emit-rtl.c (gen_lowpart_common): Likewise.
8269 * simplify-rtx.c (simplify_truncation): Likewise.
8270 (simplify_binary_operation_1): Likewise.
8271 (simplify_const_relational_operation): Likewise.
8272 (simplify_ternary_operation): Likewise.
8273 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
8275 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8276 Alan Hayward <alan.hayward@arm.com>
8277 David Sherwood <david.sherwood@arm.com>
8279 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
8280 * reload.c (push_reload): Likewise.
8281 (find_reloads): Likewise.
8283 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8284 Alan Hayward <alan.hayward@arm.com>
8285 David Sherwood <david.sherwood@arm.com>
8287 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
8288 (make_compound_operation_int): Likewise.
8289 (change_zero_ext): Likewise.
8290 * expr.c (convert_move): Likewise.
8291 (convert_modes): Likewise.
8292 * fwprop.c (forward_propagate_subreg): Likewise.
8293 * loop-iv.c (get_biv_step_1): Likewise.
8294 * optabs.c (widen_operand): Likewise.
8295 * postreload.c (move2add_valid_value_p): Likewise.
8296 * recog.c (simplify_while_replacing): Likewise.
8297 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8298 (simplify_binary_operation_1): Likewise. Remove redundant
8299 mode equality check.
8301 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8302 Alan Hayward <alan.hayward@arm.com>
8303 David Sherwood <david.sherwood@arm.com>
8305 * combine.c (combine_simplify_rtx): Add checks for
8306 is_a <scalar_int_mode>.
8307 (simplify_if_then_else): Likewise.
8308 (make_field_assignment): Likewise.
8309 (simplify_comparison): Likewise.
8310 * ifcvt.c (noce_try_bitop): Likewise.
8311 * loop-invariant.c (canonicalize_address_mult): Likewise.
8312 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8314 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8315 Alan Hayward <alan.hayward@arm.com>
8316 David Sherwood <david.sherwood@arm.com>
8318 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
8319 is_a <scalar_int_mode> instead of != BLKmode.
8321 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8322 Alan Hayward <alan.hayward@arm.com>
8323 David Sherwood <david.sherwood@arm.com>
8325 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
8326 instead of != VOIDmode.
8327 * combine.c (if_then_else_cond): Likewise.
8328 (change_zero_ext): Likewise.
8329 * dwarf2out.c (mem_loc_descriptor): Likewise.
8330 (loc_descriptor): Likewise.
8331 * rtlanal.c (canonicalize_condition): Likewise.
8332 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
8334 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8335 Alan Hayward <alan.hayward@arm.com>
8336 David Sherwood <david.sherwood@arm.com>
8338 * simplify-rtx.c (simplify_binary_operation_1): Use
8339 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
8341 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8342 Alan Hayward <alan.hayward@arm.com>
8343 David Sherwood <david.sherwood@arm.com>
8345 * wide-int.h (int_traits<unsigned char>) New class.
8346 (int_traits<unsigned short>) Likewise.
8347 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
8348 Use GET_MODE_UNIT_PRECISION and remove redundant test for
8350 * combine.c (set_nonzero_bits_and_sign_copies): Use
8351 is_a <scalar_int_mode>.
8352 (find_split_point): Likewise.
8353 (combine_simplify_rtx): Likewise.
8354 (simplify_logical): Likewise.
8355 (expand_compound_operation): Likewise.
8356 (expand_field_assignment): Likewise.
8357 (make_compound_operation): Likewise.
8358 (extended_count): Likewise.
8359 (change_zero_ext): Likewise.
8360 (simplify_comparison): Likewise.
8361 * dwarf2out.c (scompare_loc_descriptor): Likewise.
8362 (ucompare_loc_descriptor): Likewise.
8363 (minmax_loc_descriptor): Likewise.
8364 (mem_loc_descriptor): Likewise.
8365 (loc_descriptor): Likewise.
8366 * expmed.c (init_expmed_one_mode): Likewise.
8367 * lra-constraints.c (lra_constraint_offset): Likewise.
8368 * optabs.c (prepare_libcall_arg): Likewise.
8369 * postreload.c (move2add_note_store): Likewise.
8370 * reload.c (operands_match_p): Likewise.
8371 * rtl.h (load_extend_op): Likewise.
8372 * rtlhooks.c (gen_lowpart_general): Likewise.
8373 * simplify-rtx.c (simplify_truncation): Likewise.
8374 (simplify_unary_operation_1): Likewise.
8375 (simplify_binary_operation_1): Likewise.
8376 (simplify_const_binary_operation): Likewise.
8377 (simplify_const_relational_operation): Likewise.
8378 (simplify_subreg): Likewise.
8379 * stor-layout.c (bitwise_mode_for_mode): Likewise.
8380 * var-tracking.c (adjust_mems): Likewise.
8381 (prepare_call_arguments): Likewise.
8383 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8384 Alan Hayward <alan.hayward@arm.com>
8385 David Sherwood <david.sherwood@arm.com>
8387 * machmode.h (is_int_mode): New fuction.
8388 * combine.c (find_split_point): Use it.
8389 (combine_simplify_rtx): Likewise.
8390 (simplify_if_then_else): Likewise.
8391 (simplify_set): Likewise.
8392 (simplify_shift_const_1): Likewise.
8393 (simplify_comparison): Likewise.
8394 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
8395 * cse.c (notreg_cost): Likewise.
8396 (cse_insn): Likewise.
8397 * cselib.c (cselib_lookup_1): Likewise.
8398 * dojump.c (do_jump_1): Likewise.
8399 (do_compare_rtx_and_jump): Likewise.
8400 * dse.c (get_call_args): Likewise.
8401 * dwarf2out.c (rtl_for_decl_init): Likewise.
8402 (native_encode_initializer): Likewise.
8403 * expmed.c (emit_store_flag_1): Likewise.
8404 (emit_store_flag): Likewise.
8405 * expr.c (convert_modes): Likewise.
8406 (store_field): Likewise.
8407 (expand_expr_real_1): Likewise.
8408 * fold-const.c (fold_read_from_constant_string): Likewise.
8409 * gimple-ssa-sprintf.c (get_format_string): Likewise.
8410 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
8411 * optabs.c (expand_binop): Likewise.
8412 (expand_unop): Likewise.
8413 (expand_abs_nojump): Likewise.
8414 (expand_one_cmpl_abs_nojump): Likewise.
8415 * simplify-rtx.c (mode_signbit_p): Likewise.
8416 (val_signbit_p): Likewise.
8417 (val_signbit_known_set_p): Likewise.
8418 (val_signbit_known_clear_p): Likewise.
8419 (simplify_relational_operation_1): Likewise.
8420 * tree.c (vector_type_mode): Likewise.
8422 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8423 Alan Hayward <alan.hayward@arm.com>
8424 David Sherwood <david.sherwood@arm.com>
8426 * machmode.h (smallest_mode_for_size): Fix formatting.
8427 (smallest_int_mode_for_size): New function.
8428 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
8429 instead of smallest_mode_for_size.
8430 * combine.c (make_extraction): Likewise.
8431 * config/arc/arc.c (arc_expand_movmem): Likewise.
8432 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
8433 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
8434 * config/s390/s390.c (s390_expand_insv): Likewise.
8435 * config/sparc/sparc.c (assign_int_registers): Likewise.
8436 * config/spu/spu.c (spu_function_value): Likewise.
8437 (spu_function_arg): Likewise.
8438 * coverage.c (get_gcov_type): Likewise.
8439 (get_gcov_unsigned_t): Likewise.
8440 * dse.c (find_shift_sequence): Likewise.
8441 * expmed.c (store_bit_field_1): Likewise.
8442 * expr.c (convert_move): Likewise.
8443 (store_field): Likewise.
8444 * internal-fn.c (expand_arith_overflow): Likewise.
8445 * optabs-query.c (get_best_extraction_insn): Likewise.
8446 * optabs.c (expand_twoval_binop_libfunc): Likewise.
8447 * stor-layout.c (layout_type): Likewise.
8448 (initialize_sizetypes): Likewise.
8449 * targhooks.c (default_get_mask_mode): Likewise.
8450 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
8452 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8453 Alan Hayward <alan.hayward@arm.com>
8454 David Sherwood <david.sherwood@arm.com>
8456 * machmode.h (opt_mode::else_blk): New function.
8457 (int_mode_for_mode): Declare.
8458 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
8459 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
8461 * cfgexpand.c (expand_debug_expr): Likewise.
8462 * combine.c (gen_lowpart_or_truncate): Likewise.
8463 (gen_lowpart_for_combine): Likewise.
8464 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
8465 * config/avr/avr.c (avr_to_int_mode): Likewise.
8466 (avr_out_plus_1): Likewise.
8467 (avr_out_plus): Likewise.
8468 (avr_out_round): Likewise.
8469 * config/i386/i386.c (ix86_split_to_parts): Likewise.
8470 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
8471 (s390_expand_vcond): Likewise.
8472 * config/spu/spu.c (spu_split_immediate): Likewise.
8473 (spu_expand_mov): Likewise.
8474 * dse.c (get_stored_val): Likewise.
8475 * expmed.c (store_bit_field_1): Likewise.
8476 (convert_extracted_bit_field): Use int_mode_for_mode instead of
8478 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
8479 (extract_low_bits): Likewise.
8480 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
8481 handling rather than repeating the check.
8482 (emit_group_store): Likewise.
8483 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
8484 * optabs.c (expand_absneg_bit): Likewise.
8485 (expand_copysign_absneg): Likewise.
8486 (expand_copysign_bit): Likewise.
8487 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
8488 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
8489 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
8490 * var-tracking.c (prepare_call_arguments): Likewise.
8491 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
8492 int_mode_for_mode instead of mode_for_size.
8493 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
8495 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8496 Alan Hayward <alan.hayward@arm.com>
8497 David Sherwood <david.sherwood@arm.com>
8499 * machmode.h (int_mode_for_size): New function.
8500 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
8501 instead of mode_for_size.
8502 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
8504 * combine.c (expand_field_assignment): Use int_mode_for_size
8505 instead of mode_for_size.
8506 (make_extraction): Likewise.
8507 (simplify_shift_const_1): Likewise.
8508 (simplify_comparison): Likewise.
8509 * dojump.c (do_jump): Likewise.
8510 * dwarf2out.c (mem_loc_descriptor): Likewise.
8511 * emit-rtl.c (init_derived_machine_modes): Likewise.
8512 * expmed.c (flip_storage_order): Likewise.
8513 (convert_extracted_bit_field): Likewise.
8514 * expr.c (copy_blkmode_from_reg): Likewise.
8515 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
8516 * internal-fn.c (expand_mul_overflow): Likewise.
8517 * lower-subreg.c (simple_move): Likewise.
8518 * optabs-libfuncs.c (init_optabs): Likewise.
8519 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8520 * tree.c (vector_type_mode): Likewise.
8521 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
8522 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
8523 * tree-vect-generic.c (expand_vector_parallel): Likewise.
8524 * tree-vect-stmts.c (vectorizable_load): Likewise.
8525 (vectorizable_store): Likewise.
8527 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8528 Alan Hayward <alan.hayward@arm.com>
8529 David Sherwood <david.sherwood@arm.com>
8531 * coretypes.h (pod_mode): New type.
8532 (scalar_int_mode_pod): New typedef.
8533 * machmode.h (pod_mode): New class.
8534 (int_n_data_t::m): Change type to scalar_int_mode_pod.
8535 * genmodes.c (emit_mode_int_n): Update accordingly.
8536 * lower-subreg.h (target_lower_subreg): Change type to
8537 scalar_int_mode_pod.
8538 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
8539 scalar_int_mode_pod.
8541 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8542 Alan Hayward <alan.hayward@arm.com>
8543 David Sherwood <david.sherwood@arm.com>
8545 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
8546 machine_mode to scalar_int_mode.
8547 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
8548 (rs6000_option_override_internal): Remove cast to int.
8549 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
8550 machine_mode to scalar_int_mode.
8551 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
8552 (rs6000_option_override_internal): Remove cast to int.
8553 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
8554 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
8556 * config/s390/s390.c (s390_expand_builtin): Likewise.
8557 * coretypes.h (scalar_int_mode): New type.
8558 (opt_scalar_int_mode): New typedef.
8559 * machmode.h (scalar_int_mode): New class.
8560 (scalar_int_mode::includes_p): New function.
8561 (byte_mode): Change type to scalar_int_mode.
8562 (word_mode): Likewise.
8563 (ptr_mode): Likewise.
8564 * emit-rtl.c (byte_mode): Likewise.
8565 (word_mode): Likewise.
8566 (ptr_mode): Likewise.
8567 (init_derived_machine_modes): Update accordingly.
8568 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
8569 and MODE_PARTIAL_INT.
8570 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
8571 opt_scalar_int_mode.
8573 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8574 Alan Hayward <alan.hayward@arm.com>
8575 David Sherwood <david.sherwood@arm.com>
8577 * target.def (libgcc_floating_mode_supported_p): Take a
8579 * doc/tm.texi: Regenerate.
8580 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
8582 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
8583 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
8586 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8587 Alan Hayward <alan.hayward@arm.com>
8588 David Sherwood <david.sherwood@arm.com>
8590 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
8591 * doc/tm.texi: Regenerate.
8592 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
8593 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
8594 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
8595 * targhooks.h (default_floatn_mode): Likewise.
8596 * targhooks.c (default_floatn_mode): Likewise.
8597 * tree.c (build_common_tree_nodes): Update accordingly.
8599 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8600 Alan Hayward <alan.hayward@arm.com>
8601 David Sherwood <david.sherwood@arm.com>
8603 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
8604 (mode_iterator::iterate_p): Likewise.
8605 (mode_iterator::get_wider): Likewise.
8606 * expr.c (init_expr_target): Use opt_scalar_float_mode.
8608 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8609 Alan Hayward <alan.hayward@arm.com>
8610 David Sherwood <david.sherwood@arm.com>
8612 * coretypes.h (opt_scalar_float_mode): New typedef.
8613 * machmode.h (float_mode_for_size): New function.
8614 * emit-rtl.c (double_mode): Delete.
8615 (init_emit_once): Use float_mode_for_size.
8616 * stor-layout.c (layout_type): Likewise.
8617 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
8619 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8620 Alan Hayward <alan.hayward@arm.com>
8621 David Sherwood <david.sherwood@arm.com>
8623 * output.h (assemble_real): Take a scalar_float_mode.
8624 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
8625 * config/arm/arm.md (consttable_4): Likewise.
8626 (consttable_8): Likewise.
8627 (consttable_16): Likewise.
8628 * config/mips/mips.md (consttable_float): Likewise.
8629 * config/s390/s390.c (s390_output_pool_entry): Likewise.
8630 * varasm.c (assemble_real): Take a scalar_float_mode.
8631 (output_constant_pool_2): Update accordingly.
8632 (output_constant): Likewise.
8634 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8635 Alan Hayward <alan.hayward@arm.com>
8636 David Sherwood <david.sherwood@arm.com>
8638 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
8639 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
8640 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
8641 (native_encode_real): Likewise.
8642 (native_interpret_real): Likewise.
8643 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
8644 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
8646 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8647 Alan Hayward <alan.hayward@arm.com>
8648 David Sherwood <david.sherwood@arm.com>
8650 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
8651 <scalar_float_mode>. Simplify.
8652 (gen_extend_conv_libfunc): Likewise.
8654 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8655 Alan Hayward <alan.hayward@arm.com>
8656 David Sherwood <david.sherwood@arm.com>
8658 * coretypes.h (scalar_float_mode): New type.
8659 * machmode.h (mode_traits::from_int): Use machine_mode if
8660 USE_ENUM_MODES is defined.
8661 (is_a): New function.
8663 (dyn_cast): Likewise.
8664 (scalar_float_mode): New class.
8665 (scalar_float_mode::includes_p): New function.
8666 (is_float_mode): Likewise.
8667 * gdbhooks.py (MachineModePrinter): New class.
8668 (build_pretty_printer): Use it for scalar_float_mode.
8669 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
8670 (format_helper::format_helper): Turn into a template.
8671 * genmodes.c (get_mode_class): New function.
8672 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
8673 or machine_mode if none.
8674 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
8675 as_a <scalar_float_mode>.
8676 * dwarf2out.c (mem_loc_descriptor): Likewise.
8677 (insert_float): Likewise.
8678 (add_const_value_attribute): Likewise.
8679 * simplify-rtx.c (simplify_immed_subreg): Likewise.
8680 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
8681 (expand_unop): Update accordingly.
8682 (expand_abs_nojump): Likewise.
8683 (expand_copysign_absneg): Take a scalar_float_mode.
8684 (expand_copysign_bit): Likewise.
8685 (expand_copysign): Update accordingly.
8687 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8688 Alan Hayward <alan.hayward@arm.com>
8689 David Sherwood <david.sherwood@arm.com>
8691 * coretypes.h (opt_mode): New class.
8692 * machmode.h (opt_mode): Likewise.
8693 (opt_mode::else_void): New function.
8694 (opt_mode::require): Likewise.
8695 (opt_mode::exists): Likewise.
8696 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
8697 (GET_MODE_2XWIDER_MODE): Likewise.
8698 (mode_iterator::get_wider): Update accordingly.
8699 (mode_iterator::get_2xwider): Likewise.
8700 (mode_iterator::get_known_wider): Likewise, turning into a template.
8701 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
8702 forcing a wider mode to exist.
8703 * config/cr16/cr16.h (LONG_REG_P): Likewise.
8704 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
8705 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
8706 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
8707 * lower-subreg.c (init_lower_subreg): Likewise.
8708 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
8709 on the final iteration.
8710 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
8711 a wider mode exists before asking for a move pattern.
8712 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
8713 forcing a wider mode to exist.
8714 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
8715 returning false if no such mode exists.
8716 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
8717 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
8718 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
8719 Avoid checking for a MODE_INT if we already know the mode is not a
8721 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
8722 forcing a wider mode to exist.
8723 (expmed_mult_highpart_optab): Likewise.
8724 (expmed_mult_highpart): Likewise.
8725 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
8727 * lto-streamer-in.c (lto_input_mode_table): Likewise.
8728 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
8729 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
8730 * internal-fn.c (expand_mul_overflow): Update use of
8731 GET_MODE_2XWIDER_MODE.
8732 * omp-low.c (omp_clause_aligned_alignment): Likewise.
8733 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
8734 GET_MODE_WIDER_MODE.
8735 (convert_plusminus_to_widen): Likewise.
8736 * tree-switch-conversion.c (array_value_type): Likewise.
8737 * var-tracking.c (emit_note_insn_var_location): Likewise.
8738 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
8739 Return false inside rather than outside the loop if no wider mode
8741 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
8742 and GET_MODE_2XWIDER_MODE
8743 (can_compare_p): Use else_void.
8744 * gdbhooks.py (OptMachineModePrinter): New class.
8745 (build_pretty_printer): Use it for opt_mode.
8747 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8748 Alan Hayward <alan.hayward@arm.com>
8749 David Sherwood <david.sherwood@arm.com>
8751 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
8752 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
8754 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8755 Alan Hayward <alan.hayward@arm.com>
8756 David Sherwood <david.sherwood@arm.com>
8758 * machmode.h (mode_traits): New structure.
8759 (get_narrowest_mode): New function.
8760 (mode_iterator::start): Likewise.
8761 (mode_iterator::iterate_p): Likewise.
8762 (mode_iterator::get_wider): Likewise.
8763 (mode_iterator::get_known_wider): Likewise.
8764 (mode_iterator::get_2xwider): Likewise.
8765 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
8766 (FOR_EACH_MODE): Likewise.
8767 (FOR_EACH_MODE_FROM): Likewise.
8768 (FOR_EACH_MODE_UNTIL): Likewise.
8769 (FOR_EACH_WIDER_MODE): Likewise.
8770 (FOR_EACH_2XWIDER_MODE): Likewise.
8771 * builtins.c (expand_builtin_strlen): Use new mode iterators.
8772 * combine.c (simplify_comparison): Likewise
8773 * config/i386/i386.c (type_natural_mode): Likewise.
8774 * cse.c (cse_insn): Likewise.
8775 * dse.c (find_shift_sequence): Likewise.
8776 * emit-rtl.c (init_derived_machine_modes): Likewise.
8777 (init_emit_once): Likewise.
8778 * explow.c (hard_function_value): Likewise.
8779 * expmed.c (extract_fixed_bit_field_1): Likewise.
8780 (extract_bit_field_1): Likewise.
8781 (expand_divmod): Likewise.
8782 (emit_store_flag_1): Likewise.
8783 * expr.c (init_expr_target): Likewise.
8784 (convert_move): Likewise.
8785 (alignment_for_piecewise_move): Likewise.
8786 (widest_int_mode_for_size): Likewise.
8787 (emit_block_move_via_movmem): Likewise.
8788 (copy_blkmode_to_reg): Likewise.
8789 (set_storage_via_setmem): Likewise.
8790 (compress_float_constant): Likewise.
8791 * omp-low.c (omp_clause_aligned_alignment): Likewise.
8792 * optabs-query.c (get_best_extraction_insn): Likewise.
8793 * optabs.c (expand_binop): Likewise.
8794 (expand_twoval_unop): Likewise.
8795 (expand_twoval_binop): Likewise.
8796 (widen_leading): Likewise.
8797 (widen_bswap): Likewise.
8798 (expand_parity): Likewise.
8799 (expand_unop): Likewise.
8800 (prepare_cmp_insn): Likewise.
8801 (prepare_float_lib_cmp): Likewise.
8802 (expand_float): Likewise.
8803 (expand_fix): Likewise.
8804 (expand_sfix_optab): Likewise.
8805 * postreload.c (move2add_use_add2_insn): Likewise.
8806 * reg-stack.c (reg_to_stack): Likewise.
8807 * reginfo.c (choose_hard_reg_mode): Likewise.
8808 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
8809 * stor-layout.c (mode_for_size): Likewise.
8810 (smallest_mode_for_size): Likewise.
8811 (mode_for_vector): Likewise.
8812 (finish_bitfield_representative): Likewise.
8813 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
8814 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
8815 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
8816 * var-tracking.c (prepare_call_arguments): Likewise.
8818 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8819 Alan Hayward <alan.hayward@arm.com>
8820 David Sherwood <david.sherwood@arm.com>
8822 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
8823 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
8824 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
8825 * machmode.h (mode_size): Move earlier in file.
8826 (mode_precision): Likewise.
8827 (mode_inner): Likewise.
8828 (mode_nunits): Likewise.
8829 (mode_unit_size): Likewise.
8830 (unit_unit_precision): Likewise.
8831 (mode_wider): Likewise.
8832 (mode_2xwider): Likewise.
8833 (machine_mode): New class.
8834 (mode_to_bytes): New function.
8835 (mode_to_bits): Likewise.
8836 (mode_to_precision): Likewise.
8837 (mode_to_inner): Likewise.
8838 (mode_to_unit_size): Likewise.
8839 (mode_to_unit_precision): Likewise.
8840 (mode_to_nunits): Likewise.
8841 (GET_MODE_SIZE): Use mode_to_bytes.
8842 (GET_MODE_BITSIZE): Use mode_to_bits.
8843 (GET_MODE_PRECISION): Use mode_to_precision.
8844 (GET_MODE_INNER): Use mode_to_inner.
8845 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
8846 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
8847 (GET_MODE_NUNITS): Use mode_to_nunits.
8848 * system.h (ALWAYS_INLINE): New macro.
8849 * config/powerpcspe/powerpcspe-c.c
8850 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
8851 int for arg1_mode and arg2_mode.
8853 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8854 Alan Hayward <alan.hayward@arm.com>
8855 David Sherwood <david.sherwood@arm.com>
8857 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
8858 Prefix mode names with E_ in case statements.
8859 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8860 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
8861 (aarch64_split_simd_move): Likewise.
8862 (aarch64_gen_storewb_pair): Likewise.
8863 (aarch64_gen_loadwb_pair): Likewise.
8864 (aarch64_gen_store_pair): Likewise.
8865 (aarch64_gen_load_pair): Likewise.
8866 (aarch64_get_condition_code_1): Likewise.
8867 (aarch64_constant_pool_reload_icode): Likewise.
8868 (get_rsqrte_type): Likewise.
8869 (get_rsqrts_type): Likewise.
8870 (get_recpe_type): Likewise.
8871 (get_recps_type): Likewise.
8872 (aarch64_gimplify_va_arg_expr): Likewise.
8873 (aarch64_simd_container_mode): Likewise.
8874 (aarch64_emit_load_exclusive): Likewise.
8875 (aarch64_emit_store_exclusive): Likewise.
8876 (aarch64_expand_compare_and_swap): Likewise.
8877 (aarch64_gen_atomic_cas): Likewise.
8878 (aarch64_emit_bic): Likewise.
8879 (aarch64_emit_atomic_swap): Likewise.
8880 (aarch64_emit_atomic_load_op): Likewise.
8881 (aarch64_evpc_trn): Likewise.
8882 (aarch64_evpc_uzp): Likewise.
8883 (aarch64_evpc_zip): Likewise.
8884 (aarch64_evpc_ext): Likewise.
8885 (aarch64_evpc_rev): Likewise.
8886 (aarch64_evpc_dup): Likewise.
8887 (aarch64_gen_ccmp_first): Likewise.
8888 (aarch64_gen_ccmp_next): Likewise.
8889 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
8890 (alpha_emit_xfloating_libcall): Likewise.
8891 (emit_insxl): Likewise.
8892 (alpha_arg_type): Likewise.
8893 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
8894 (arc_preferred_simd_mode): Likewise.
8895 (arc_secondary_reload): Likewise.
8896 (get_arc_condition_code): Likewise.
8897 (arc_print_operand): Likewise.
8898 (arc_legitimate_constant_p): Likewise.
8899 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8900 * config/arc/arc.md (casesi_load): Likewise.
8901 (casesi_compact_jump): Likewise.
8902 * config/arc/predicates.md (proper_comparison_operator): Likewise.
8903 (cc_use_register): Likewise.
8904 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8905 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
8906 (arm_init_iwmmxt_builtins): Likewise.
8907 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
8908 (neon_expand_vector_init): Likewise.
8909 (arm_attr_length_move_neon): Likewise.
8910 (maybe_get_arm_condition_code): Likewise.
8911 (arm_emit_vector_const): Likewise.
8912 (arm_preferred_simd_mode): Likewise.
8913 (arm_output_iwmmxt_tinsr): Likewise.
8914 (thumb1_output_casesi): Likewise.
8915 (thumb2_output_casesi): Likewise.
8916 (arm_emit_load_exclusive): Likewise.
8917 (arm_emit_store_exclusive): Likewise.
8918 (arm_expand_compare_and_swap): Likewise.
8919 (arm_evpc_neon_vuzp): Likewise.
8920 (arm_evpc_neon_vzip): Likewise.
8921 (arm_evpc_neon_vrev): Likewise.
8922 (arm_evpc_neon_vtrn): Likewise.
8923 (arm_evpc_neon_vext): Likewise.
8924 (arm_validize_comparison): Likewise.
8925 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
8926 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
8927 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
8928 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
8929 (c6x_preferred_simd_mode): Likewise.
8930 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
8931 (epiphany_rtx_costs): Likewise.
8932 * config/epiphany/predicates.md (proper_comparison_operator):
8934 * config/frv/frv.c (condexec_memory_operand): Likewise.
8935 (frv_emit_move): Likewise.
8936 (output_move_single): Likewise.
8937 (output_condmove_single): Likewise.
8938 (frv_hard_regno_mode_ok): Likewise.
8939 (frv_matching_accg_mode): Likewise.
8940 * config/h8300/h8300.c (split_adds_subs): Likewise.
8941 (h8300_rtx_costs): Likewise.
8942 (h8300_print_operand): Likewise.
8943 (compute_mov_length): Likewise.
8944 (output_logical_op): Likewise.
8945 (compute_logical_op_length): Likewise.
8946 (compute_logical_op_cc): Likewise.
8947 (h8300_shift_needs_scratch_p): Likewise.
8948 (output_a_shift): Likewise.
8949 (compute_a_shift_length): Likewise.
8950 (compute_a_shift_cc): Likewise.
8951 (expand_a_rotate): Likewise.
8952 (output_a_rotate): Likewise.
8953 * config/i386/i386.c (classify_argument): Likewise.
8954 (function_arg_advance_32): Likewise.
8955 (function_arg_32): Likewise.
8956 (function_arg_64): Likewise.
8957 (function_value_64): Likewise.
8958 (ix86_gimplify_va_arg): Likewise.
8959 (ix86_legitimate_constant_p): Likewise.
8960 (put_condition_code): Likewise.
8961 (split_double_mode): Likewise.
8962 (ix86_avx256_split_vector_move_misalign): Likewise.
8963 (ix86_expand_vector_logical_operator): Likewise.
8964 (ix86_split_idivmod): Likewise.
8965 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
8966 (ix86_build_const_vector): Likewise.
8967 (ix86_build_signbit_mask): Likewise.
8968 (ix86_match_ccmode): Likewise.
8969 (ix86_cc_modes_compatible): Likewise.
8970 (ix86_expand_branch): Likewise.
8971 (ix86_expand_sse_cmp): Likewise.
8972 (ix86_expand_sse_movcc): Likewise.
8973 (ix86_expand_int_sse_cmp): Likewise.
8974 (ix86_expand_vec_perm_vpermi2): Likewise.
8975 (ix86_expand_vec_perm): Likewise.
8976 (ix86_expand_sse_unpack): Likewise.
8977 (ix86_expand_int_addcc): Likewise.
8978 (ix86_split_to_parts): Likewise.
8979 (ix86_vectorize_builtin_gather): Likewise.
8980 (ix86_vectorize_builtin_scatter): Likewise.
8981 (avx_vpermilp_parallel): Likewise.
8982 (inline_memory_move_cost): Likewise.
8983 (ix86_tieable_integer_mode_p): Likewise.
8984 (x86_maybe_negate_const_int): Likewise.
8985 (ix86_expand_vector_init_duplicate): Likewise.
8986 (ix86_expand_vector_init_one_nonzero): Likewise.
8987 (ix86_expand_vector_init_one_var): Likewise.
8988 (ix86_expand_vector_init_concat): Likewise.
8989 (ix86_expand_vector_init_interleave): Likewise.
8990 (ix86_expand_vector_init_general): Likewise.
8991 (ix86_expand_vector_set): Likewise.
8992 (ix86_expand_vector_extract): Likewise.
8993 (emit_reduc_half): Likewise.
8994 (ix86_emit_i387_round): Likewise.
8995 (ix86_mangle_type): Likewise.
8996 (ix86_expand_round_sse4): Likewise.
8997 (expand_vec_perm_blend): Likewise.
8998 (canonicalize_vector_int_perm): Likewise.
8999 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
9000 (expand_vec_perm_1): Likewise.
9001 (expand_vec_perm_interleave3): Likewise.
9002 (expand_vec_perm_even_odd_pack): Likewise.
9003 (expand_vec_perm_even_odd_1): Likewise.
9004 (expand_vec_perm_broadcast_1): Likewise.
9005 (ix86_vectorize_vec_perm_const_ok): Likewise.
9006 (ix86_expand_vecop_qihi): Likewise.
9007 (ix86_expand_mul_widen_hilo): Likewise.
9008 (ix86_expand_sse2_abs): Likewise.
9009 (ix86_expand_pextr): Likewise.
9010 (ix86_expand_pinsr): Likewise.
9011 (ix86_preferred_simd_mode): Likewise.
9012 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
9013 * config/i386/sse.md (*andnot<mode>3): Likewise.
9014 (<mask_codefor><code><mode>3<mask_name>): Likewise.
9015 (*<code><mode>3): Likewise.
9016 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
9017 (ia64_expand_atomic_op): Likewise.
9018 (ia64_arg_type): Likewise.
9019 (ia64_mode_to_int): Likewise.
9020 (ia64_scalar_mode_supported_p): Likewise.
9021 (ia64_vector_mode_supported_p): Likewise.
9022 (expand_vec_perm_broadcast): Likewise.
9023 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
9024 (iq2000_function_arg_advance): Likewise.
9025 (iq2000_function_arg): Likewise.
9026 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
9027 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
9028 (m68k_libcall_value): Likewise.
9029 (m68k_function_value): Likewise.
9030 (sched_attr_op_type): Likewise.
9031 * config/mcore/mcore.c (mcore_output_move): Likewise.
9032 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
9034 (microblaze_function_arg): Likewise.
9035 * config/mips/mips.c (mips16_build_call_stub): Likewise.
9036 (mips_print_operand): Likewise.
9037 (mips_mode_ok_for_mov_fmt_p): Likewise.
9038 (mips_vector_mode_supported_p): Likewise.
9039 (mips_preferred_simd_mode): Likewise.
9040 (mips_expand_vpc_loongson_even_odd): Likewise.
9041 (mips_expand_vec_unpack): Likewise.
9042 (mips_expand_vi_broadcast): Likewise.
9043 (mips_expand_vector_init): Likewise.
9044 (mips_expand_vec_reduc): Likewise.
9045 (mips_expand_msa_cmp): Likewise.
9046 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
9047 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
9048 (cc_flags_for_mode): Likewise.
9049 * config/msp430/msp430.c (msp430_print_operand): Likewise.
9050 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
9051 (nds32_output_casesi_pc_relative): Likewise.
9052 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9053 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
9054 (nvptx_gen_unpack): Likewise.
9055 (nvptx_gen_pack): Likewise.
9056 (nvptx_gen_shuffle): Likewise.
9057 (nvptx_gen_wcast): Likewise.
9058 (nvptx_preferred_simd_mode): Likewise.
9059 * config/pa/pa.c (pa_secondary_reload): Likewise.
9060 * config/pa/predicates.md (base14_operand): Likewise.
9061 * config/powerpcspe/powerpcspe-c.c
9062 (altivec_resolve_overloaded_builtin): Likewise.
9063 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
9065 (rs6000_preferred_simd_mode): Likewise.
9066 (output_vec_const_move): Likewise.
9067 (rs6000_expand_vector_extract): Likewise.
9068 (rs6000_split_vec_extract_var): Likewise.
9069 (reg_offset_addressing_ok_p): Likewise.
9070 (rs6000_legitimate_offset_address_p): Likewise.
9071 (rs6000_legitimize_address): Likewise.
9072 (rs6000_emit_set_const): Likewise.
9073 (rs6000_const_vec): Likewise.
9074 (rs6000_emit_move): Likewise.
9075 (spe_build_register_parallel): Likewise.
9076 (rs6000_darwin64_record_arg_recurse): Likewise.
9077 (swap_selector_for_mode): Likewise.
9078 (spe_init_builtins): Likewise.
9079 (paired_init_builtins): Likewise.
9080 (altivec_init_builtins): Likewise.
9081 (do_load_for_compare): Likewise.
9082 (rs6000_generate_compare): Likewise.
9083 (rs6000_expand_float128_convert): Likewise.
9084 (emit_load_locked): Likewise.
9085 (emit_store_conditional): Likewise.
9086 (rs6000_output_function_epilogue): Likewise.
9087 (rs6000_handle_altivec_attribute): Likewise.
9088 (rs6000_function_value): Likewise.
9089 (emit_fusion_gpr_load): Likewise.
9090 (emit_fusion_p9_load): Likewise.
9091 (emit_fusion_p9_store): Likewise.
9092 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
9093 (fusion_gpr_mem_load): Likewise.
9094 (fusion_addis_mem_combo_load): Likewise.
9095 (fusion_addis_mem_combo_store): Likewise.
9096 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
9097 (fusion_gpr_mem_load): Likewise.
9098 (fusion_addis_mem_combo_load): Likewise.
9099 (fusion_addis_mem_combo_store): Likewise.
9100 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9102 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
9103 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
9104 (rs6000_preferred_simd_mode): Likewise.
9105 (output_vec_const_move): Likewise.
9106 (rs6000_expand_vector_extract): Likewise.
9107 (rs6000_split_vec_extract_var): Likewise.
9108 (reg_offset_addressing_ok_p): Likewise.
9109 (rs6000_legitimate_offset_address_p): Likewise.
9110 (rs6000_legitimize_address): Likewise.
9111 (rs6000_emit_set_const): Likewise.
9112 (rs6000_const_vec): Likewise.
9113 (rs6000_emit_move): Likewise.
9114 (rs6000_darwin64_record_arg_recurse): Likewise.
9115 (swap_selector_for_mode): Likewise.
9116 (paired_init_builtins): Likewise.
9117 (altivec_init_builtins): Likewise.
9118 (rs6000_expand_float128_convert): Likewise.
9119 (emit_load_locked): Likewise.
9120 (emit_store_conditional): Likewise.
9121 (rs6000_output_function_epilogue): Likewise.
9122 (rs6000_handle_altivec_attribute): Likewise.
9123 (rs6000_function_value): Likewise.
9124 (emit_fusion_gpr_load): Likewise.
9125 (emit_fusion_p9_load): Likewise.
9126 (emit_fusion_p9_store): Likewise.
9127 * config/rx/rx.c (rx_gen_move_template): Likewise.
9128 (flags_from_mode): Likewise.
9129 * config/s390/predicates.md (s390_alc_comparison): Likewise.
9130 (s390_slb_comparison): Likewise.
9131 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
9132 (s390_vector_mode_supported_p): Likewise.
9133 (s390_cc_modes_compatible): Likewise.
9134 (s390_match_ccmode_set): Likewise.
9135 (s390_canonicalize_comparison): Likewise.
9136 (s390_emit_compare_and_swap): Likewise.
9137 (s390_branch_condition_mask): Likewise.
9138 (s390_rtx_costs): Likewise.
9139 (s390_secondary_reload): Likewise.
9140 (__SECONDARY_RELOAD_CASE): Likewise.
9141 (s390_expand_cs): Likewise.
9142 (s390_preferred_simd_mode): Likewise.
9143 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
9144 * config/sh/sh.c (sh_print_operand): Likewise.
9145 (dump_table): Likewise.
9146 (sh_secondary_reload): Likewise.
9147 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9148 * config/sh/sh.md (casesi_worker_1): Likewise.
9149 (casesi_worker_2): Likewise.
9150 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
9151 (fcc_comparison_operator): Likewise.
9152 * config/sparc/sparc.c (sparc_expand_move): Likewise.
9153 (emit_soft_tfmode_cvt): Likewise.
9154 (sparc_preferred_simd_mode): Likewise.
9155 (output_cbranch): Likewise.
9156 (sparc_print_operand): Likewise.
9157 (sparc_expand_vec_perm_bmask): Likewise.
9158 (vector_init_bshuffle): Likewise.
9159 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
9160 (spu_vector_mode_supported_p): Likewise.
9161 (spu_expand_insv): Likewise.
9162 (spu_emit_branch_or_set): Likewise.
9163 (spu_handle_vector_attribute): Likewise.
9164 (spu_builtin_splats): Likewise.
9165 (spu_builtin_extract): Likewise.
9166 (spu_builtin_promote): Likewise.
9167 (spu_expand_sign_extend): Likewise.
9168 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
9169 (tilegx_simd_int): Likewise.
9170 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
9171 (tilepro_simd_int): Likewise.
9172 * config/v850/v850.c (const_double_split): Likewise.
9173 (v850_print_operand): Likewise.
9174 (ep_memory_offset): Likewise.
9175 * config/vax/vax.c (vax_rtx_costs): Likewise.
9176 (vax_output_int_move): Likewise.
9177 (vax_output_int_add): Likewise.
9178 (vax_output_int_subtract): Likewise.
9179 * config/visium/predicates.md (visium_branch_operator): Likewise.
9180 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
9181 (visium_print_operand_address): Likewise.
9182 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9183 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
9184 (xtensa_expand_conditional_branch): Likewise.
9185 (xtensa_copy_incoming_a7): Likewise.
9186 (xtensa_output_literal): Likewise.
9187 * dfp.c (decimal_real_maxval): Likewise.
9188 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
9190 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9191 Alan Hayward <alan.hayward@arm.com>
9192 David Sherwood <david.sherwood@arm.com>
9194 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
9195 (mode_nunits_inline): Likewise.
9196 (mode_inner_inline): Likewise.
9197 (mode_unit_size_inline): Likewise.
9198 (mode_unit_precision_inline): Likewise.
9199 (emit_insn_modes_h): Likewise. Also emit a #define of the
9201 (emit_mode_wider): Add an E_ prefix to mode names.
9202 (emit_mode_complex): Likewise.
9203 (emit_mode_inner): Likewise.
9204 (emit_mode_adjustments): Likewise.
9205 (emit_mode_int_n): Likewise.
9206 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
9207 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
9208 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
9209 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
9210 (CRC32_BUILTIN, ENTRY): Likewise.
9211 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
9212 (aarch64_pop_regs): Likewise.
9213 (aarch64_process_components): Likewise.
9214 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
9215 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
9216 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
9217 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
9218 * config/arm/arm.c (arm_init_libfuncs): Likewise.
9219 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
9221 * config/i386/i386-builtin.def (pcmpestr): Likewise.
9222 (pcmpistr): Likewise.
9223 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
9224 * config/mmix/mmix.c (mmix_output_condition): Likewise.
9225 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
9227 * config/rl78/rl78.c (mduc_regs): Likewise.
9228 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
9229 (htm_expand_builtin): Likewise.
9230 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
9231 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
9232 * config/xtensa/xtensa.c (print_operand): Likewise.
9233 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
9234 (NUM_MODE_VECTOR_INT): Likewise.
9235 * genoutput.c (null_operand): Likewise.
9236 (output_operand_data): Likewise.
9237 * genrecog.c (print_parameter_value): Likewise.
9238 * lra.c (debug_operand_data): Likewise.
9240 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9242 * dwarf2out.c (scompare_loc_descriptor_wide)
9243 (scompare_loc_descriptor_narrow): New functions, split out from...
9244 (scompare_loc_descriptor): ...here.
9245 * expmed.c (emit_store_flag_int): New function, split out from...
9246 (emit_store_flag): ...here.
9248 2017-08-30 Richard Biener <rguenther@suse.de>
9250 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
9251 (dwarf2out_early_finish): Move setting of AT_pubnames from
9252 early debug output to early finish.
9254 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
9256 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
9257 and -mdata-region to the assembler.
9259 2017-08-30 Richard Biener <rguenther@suse.de>
9261 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
9263 (gen_subprogram_die): Add DW_AT_object_pointer only early.
9264 (dwarf2out_early_global_decl): Only generate a DIE for the
9265 abstract origin if it doesn't already exist or is a declaration DIE.
9266 (resolve_addr): Do not add the linkage name twice when
9267 generating a stub DIE for the DW_TAG_GNU_call_site target.
9269 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9271 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9272 Use machine_mode rather than int for arg1_mode.
9274 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
9277 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
9278 that the second argument of the built-in functions to unpack
9279 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
9280 switch statement instead a lot of if statements.
9281 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
9282 Allow 64-bit values to be in Altivec registers as well as
9283 traditional floating point registers.
9284 (pack<mode>, FMOVE128_VSX iterator): Likewise.
9286 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
9288 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
9289 MAX_REGS_PER_ADDRESS == 1.
9291 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
9293 * config/i386/i386.opt (flag_fentry): Do not init to -1.
9294 * config/i386/i386.c (ix86_option_override_internal): Simplify
9295 setting of opts->x_flag_entry.
9297 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9298 Jakub Jelinek <jakub@redhat.com>
9299 Richard Biener <rguenther@suse.de>
9301 PR tree-optimization/81503
9302 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
9303 folded constant fits in the target type; reorder tests for clarity.
9305 2017-08-29 Martin Liska <mliska@suse.cz>
9307 * passes.def: Include pass_lower_switch.
9308 * stmt.c (dump_case_nodes): Remove and move to
9309 tree-switch-conversion.
9310 (case_values_threshold): Likewise.
9311 (expand_switch_as_decision_tree_p): Likewise.
9312 (emit_case_decision_tree): Likewise.
9313 (expand_case): Likewise.
9314 (balance_case_nodes): Likewise.
9315 (node_has_low_bound): Likewise.
9316 (node_has_high_bound): Likewise.
9317 (node_is_bounded): Likewise.
9318 (emit_case_nodes): Likewise.
9319 (struct simple_case_node): New struct.
9320 (add_case_node): Remove.
9321 (emit_case_dispatch_table): Use vector instead of case_list.
9322 (reset_out_edges_aux): Remove.
9323 (compute_cases_per_edge): Likewise.
9324 (expand_case): Build list of simple_case_node.
9325 (expand_sjlj_dispatch_table): Use it.
9326 * tree-switch-conversion.c (struct case_node): Moved from
9327 stmt.c and adjusted.
9328 (emit_case_nodes): Likewise.
9329 (node_has_low_bound): Likewise.
9330 (node_has_high_bound): Likewise.
9331 (node_is_bounded): Likewise.
9332 (case_values_threshold): Likewise.
9333 (reset_out_edges_aux): Likewise.
9334 (compute_cases_per_edge): Likewise.
9335 (add_case_node): Likewise.
9336 (dump_case_nodes): Likewise.
9337 (balance_case_nodes): Likewise.
9338 (expand_switch_as_decision_tree_p): Likewise.
9339 (emit_jump): Likewise.
9340 (emit_case_decision_tree): Likewise.
9341 (try_switch_expansion): Likewise.
9342 (do_jump_if_equal): Likewise.
9343 (emit_cmp_and_jump_insns): Likewise.
9344 (fix_phi_operands_for_edge): New function.
9345 (record_phi_operand_mapping): Likewise.
9346 (class pass_lower_switch): New pass.
9347 (pass_lower_switch::execute): New function.
9348 (make_pass_lower_switch): Likewise.
9349 (conditional_probability):
9350 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
9351 * tree-pass.h: Add make_pass_lower_switch.
9353 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
9356 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
9359 2017-08-29 Richard Biener <rguenther@suse.de>
9361 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
9362 we do not add a DW_AT_inline attribute twice.
9363 (gen_subprogram_die): Remove code setting DW_AT_inline on
9364 DECL_ABSTRACT_P nodes.
9366 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
9368 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
9369 calls to internal functions.
9370 (gimplify_modify_expr): Likewise.
9371 * tree-call-cdce.c (use_internal_fn): Likewise.
9372 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
9373 (convert_to_divmod): Set the nothrow flag.
9374 * tree-if-conv.c (predicate_mem_writes): Likewise.
9375 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
9376 (vectorizable_call): Likewise.
9377 (vectorizable_store): Likewise.
9378 (vectorizable_load): Likewise.
9379 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
9380 (vect_recog_mask_conversion_pattern): Likewise.
9382 2017-08-29 Martin Liska <mliska@suse.cz>
9385 * gcc.c (driver_handle_option): Add new argument.
9386 * opts-common.c (handle_option): Pass
9387 target_option_override_hook.
9388 * opts-global.c (lang_handle_option): Add new option.
9389 (set_default_handlers): Add new argument.
9390 (decode_options): Likewise.
9391 * opts.c (target_handle_option): Likewise.
9392 (common_handle_option): Call target_option_override_hook.
9393 * opts.h (struct cl_option_handler_func): Add hook for
9394 target option override.
9395 (struct cl_option_handlers): Likewise.
9396 (set_default_handlers): Add new argument.
9397 (decode_options): Likewise.
9398 (common_handle_option): Likewise.
9399 (target_handle_option): Likewise.
9400 * toplev.c (toplev::main): Pass targetm.target_option.override
9403 2017-08-29 Richard Biener <rguenther@suse.de>
9404 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
9406 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
9407 life to the active subtree.
9409 2017-08-28 Jeff Law <law@redhat.com>
9411 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
9412 derive_equivalences.
9413 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
9415 (edge_info::derive_equivalences): New private member function
9417 * tree-ssa-dom.c (class edge_info): Changed from a struct
9418 to a class. Add ctor/dtor, methods and data members.
9419 (edge_info::edge_info): Renamed from allocate_edge_info.
9420 Initialize additional members.
9421 (edge_info::~edge_info): New.
9422 (free_dom_edge_info): Delete the edge info.
9423 (record_edge_info): Use new class & associated member functions.
9424 Tighten forms for testing for edge equivalences.
9425 (record_temporary_equivalences): Iterate over the simple
9426 equivalences rather than assuming there's only one per edge.
9427 (cprop_into_successor_phis): Iterate over the simple
9428 equivalences rather than assuming there's only one per edge.
9429 (optimize_stmt): Use operand_equal_p rather than pointer
9430 equality for mini-DSE code.
9432 2017-08-28 Nathan Sidwell <nathan@acm.org>
9434 * gcc.c (execute): Fold SIGPIPE handling into switch
9435 statement. Adjust internal error message.
9437 2017-08-28 Richard Biener <rguenther@suse.de>
9440 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
9441 Do nothing for removed DIEs.
9443 2017-08-28 Richard Biener <rguenther@suse.de>
9445 PR tree-optimization/81977
9446 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
9449 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
9452 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
9453 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
9454 using targetm.gen_mem_thread_fence.
9456 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
9459 * config/i386/i386.md (*<btsc><mode>): Change operand 2
9460 predicate to register_operand. Reorder operands.
9461 (*btr<mode>): Ditto.
9462 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
9463 (*btr<mode>_mask): Ditto.
9465 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
9467 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
9468 * config/rs6000/xmmintrin.h: New file.
9469 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
9471 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9474 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
9475 parameter and_insn and return it.
9476 (recombine_lvx_pattern): Insert a copy to ensure availability of
9477 the base register of the copied masking operation at the point of
9478 the instruction replacement.
9479 (recombine_stvx_pattern): Likewise.
9481 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
9483 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
9484 undocumented switches.
9485 (-mpower9-dform-vector): Likewise.
9486 (-mpower9-dform): Likewise.
9487 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
9488 comments to delete references to -mpower9-dform* switches.
9489 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
9490 Delete reference to -mpower9-dform* switches, test for
9491 -mpower9-vector instead.
9492 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
9493 (OTHER_P9_VECTOR_MASKS): Likewise.
9494 (POWERPC_MASKS): Likewise.
9495 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
9496 tests against -mpower9-dform* to -mpower9-vector. Delete code
9497 that checked for -mpower9-dform* consistancy with other options.
9498 Add test for -mpower9-misc to enable other power9 switches.
9499 (rs6000_init_hard_regno_mode_ok): Likewise.
9500 (rs6000_option_override_internal): Likewise.
9501 (rs6000_emit_prologue): Likewise.
9502 (rs6000_emit_epilogue): Likewise.
9503 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
9504 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
9505 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
9507 (emit_fusion_p9_store): Likewise.
9508 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
9509 resetting these macros if the assembler does not support ISA 3.0
9511 (TARGET_P9_DFORM_VECTOR): Likewise.
9512 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
9513 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
9515 2017-08-25 Alan Modra <amodra@gmail.com>
9517 PR rtl-optimization/81747
9518 * cse.c (cse_extended_basic_block): Don't attempt to record
9519 equivalences for degenerate conditional branches that branch
9520 to their fall-through.
9522 2017-08-24 Martin Sebor <msebor@redhat.com>
9525 * gimple-fold.c (size_must_be_zero_p): New function.
9526 (gimple_fold_builtin_memory_op): Call it.
9528 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
9530 * config/rs6000/mm_malloc.h: New file.
9532 2017-08-24 Bin Cheng <bin.cheng@arm.com>
9534 PR tree-optimization/81913
9535 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
9536 analysis when either IVs in condition can wrap.
9538 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
9540 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
9541 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
9543 2017-08-24 Richard Biener <rguenther@suse.de>
9546 * targhooks.c (default_target_can_inline_p): Properly
9547 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
9548 is present and always compare.
9549 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
9550 infer -mfpmath=sse from TARGET_SSE_P.
9551 (ix86_can_inline_p): Properly use target_option_default_node when
9552 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
9554 2017-08-24 Richard Biener <rguenther@suse.de>
9557 * dwarf2out.c (output_die): Handle flag_generate_offload like
9559 (output_comp_unit): Likewise.
9560 (gen_array_type_die): Likewise.
9561 (dwarf2out_early_finish): Likewise.
9562 (note_variable_value_in_expr): Likewise.
9563 (dwarf2out_finish): Likewise. Adjust assert.
9564 * cgraphunit.c (symbol_table::compile): Move setting of
9565 flag_generate_offload earlier ...
9566 (symbol_table::finalize_compilation_unit): ... here, before
9567 early debug finalization.
9569 2017-08-24 Richard Biener <rguenther@suse.de>
9571 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
9572 and ipa-fnsummary.h.
9573 (ix86_can_inline_p): When ix86_fpmath flags do not match
9574 check whether the callee uses FP math at all.
9576 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
9579 * tree-ssanames.c (get_nonzero_bits): Use element_precision
9580 instead of TYPE_PRECISION.
9582 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
9583 Alan Hayward <alan.hayward@arm.com>
9584 David Sherwood <david.sherwood@arm.com>
9586 * combine.c (make_extraction): Use subreg_offset_from_lsb.
9588 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
9590 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
9592 (ix86_frame::stack_realign_allocate): New field.
9593 (struct machine_frame_state): Modify comments.
9594 (machine_frame_state::sp_realigned_fp_end): New field.
9595 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
9597 (sp_valid_at): Add assertion to assure no attempt to access invalid
9598 offset of a realigned stack.
9599 (fp_valid_at): Likewise.
9600 (choose_baseaddr): Modify comments.
9601 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
9602 ix86_expand_prologue.
9603 (ix86_expand_prologue): Modify stack realignment and allocation.
9604 (ix86_expand_epilogue): Modify comments.
9605 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
9606 avx2_runtime, avx512f, and avx512f_runtime.
9608 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
9610 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
9611 (mstackrealign): Do not init to -1.
9612 * config/i386/i386.c (ix86_option_override_internal):
9613 Check opts_set, not opts when setting default value of
9614 opts->x_ix86_force_align_arg_pointer.
9616 2017-08-23 Richard Biener <rguenther@suse.de>
9618 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
9619 lang_hooks.decl_printable_name.
9620 * print-rtl-function.c (print_rtx_function): Likewise.
9621 * tree-pretty-print.c (dump_function_header): Likewise.
9623 2017-08-23 Richard Biener <rguenther@suse.de>
9626 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
9627 -g0 at compile-time.
9629 2017-08-23 Tamar Christina <tamar.christina@arm.com>
9632 * doc/sourcebuild.texi (Other hardware attributes):
9635 2017-08-23 Tamar Christina <tamar.christina@arm.com>
9638 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
9639 Fix single-use check.
9641 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
9643 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
9645 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
9647 * doc/install.texi: Modify to add more details on running selected
9650 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
9652 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
9653 is combined with -mabi=ms.
9654 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
9657 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9659 PR tree-optimization/81488
9660 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
9661 and cached_basis fields.
9662 (MAX_SPREAD): New constant.
9663 (alloc_cand_and_find_basis): Initialize new fields.
9664 (clear_visited): New function.
9665 (create_phi_basis_1): Rename from create_phi_basis, set visited
9666 and cached_basis fields.
9667 (create_phi_basis): New wrapper function.
9668 (phi_add_costs_1): Rename from phi_add_costs, add spread
9669 parameter, set visited field, short-circuit when limits reached.
9670 (phi_add_costs): New wrapper function.
9671 (record_phi_increments_1): Rename from record_phi_increments, set
9673 (record_phi_increments): New wrapper function.
9674 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
9675 (phi_incr_cost): New wrapper function.
9676 (all_phi_incrs_profitable_1): Rename from
9677 all_phi_incrs_profitable, set visited field.
9678 (all_phi_incrs_profitable): New wrapper function.
9680 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
9681 Alan Hayward <alan.hayward@arm.com>
9682 David Sherwood <david.sherwood@arm.com>
9684 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
9685 that takes the outer and inner modes.
9686 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
9687 comparison as the canonical test for a paradoxical subreg.
9688 * combine.c (simplify_set): Use paradoxical_subreg_p.
9689 (make_extraction): Likewise.
9690 (force_to_mode): Likewise.
9691 (rtx_equal_for_field_assignment_p): Likewise.
9692 (gen_lowpart_for_combine): Likewise.
9693 (simplify_comparison): Likewise.
9694 * cse.c (equiv_constant): Likewise.
9695 * expmed.c (store_bit_field_1): Likewise.
9696 * final.c (alter_subreg): Likewise.
9697 * fwprop.c (propagate_rtx): Likewise.
9698 (forward_propagate_subreg): Likewise.
9699 * ira-conflicts.c (ira_build_conflicts): Likewise.
9700 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
9701 * lra-constraints.c (curr_insn_transform): Likewise.
9702 (split_reg): Likewise.
9703 * lra-eliminations.c (move_plus_up): Likewise.
9704 (lra_eliminate_regs_1): Likewise.
9705 * recog.c (general_operand): Likewise.
9706 * ree.c (combine_reaching_defs): Likewise.
9707 * reload.c (push_reload): Likewise.
9708 (find_reloads): Likewise.
9709 * reload1.c (elimination_effects): Likewise.
9710 (compute_reload_subreg_offset): Likewise.
9711 (choose_reload_regs): Likewise.
9712 * rtlanal.c (subreg_lsb_1): Likewise.
9713 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9714 (simplify_subreg): Likewise.
9715 * var-tracking.c (track_loc_p): Likewise.
9716 * emit-rtl.c (byte_lowpart_offset): Likewise.
9717 (paradoxical_subreg_p): Delete out-of-line definition.
9719 2017-08-22 Jeff Law <law@redhat.com>
9721 PR tree-optimization/81741
9722 PR tree-optimization/71947
9723 * tree-ssa-dom.c: Include tree-inline.h.
9724 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
9725 equivalences if one is more expensive to compute than the other.
9726 * tree-ssa-scopedtables.h (class const_or_copies): Make
9727 record_const_or_copy_raw method private.
9728 (class avail_exprs_stack): New method simplify_binary_operation.
9729 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
9730 avail_exprs_stack::simplify_binary_operation as needed.
9731 (avail_exprs_stack::simplify_binary_operation): New function.
9733 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
9735 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
9736 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
9737 (DOT_SYMBOLS): Likewise.
9738 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
9739 (RELOCATABLE_NEEDS_FIXUP): Likewise.
9740 (RS6000_ABI_NAME): Likewise.
9741 (TARGET_CMODEL): Likewise.
9742 (TOC_SECTION_ASM_OP): Likewise.
9743 (SET_CMODEL): New macro.
9744 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
9746 2017-08-22 Richard Biener <rguenther@suse.de>
9748 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
9749 to allow for free-lang-data replacements similar to verify_type_variant.
9751 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
9752 Alan Hayward <alan.hayward@arm.com>
9753 David Sherwood <david.sherwood@arm.com>
9755 * config/aarch64/aarch64.md (casesi): Use DImode rather than
9756 VOIDmode for the LABEL_REF.
9758 2017-08-22 Richard Biener <rguenther@suse.de>
9760 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
9762 2017-08-22 Richard Biener <rguenther@suse.de>
9764 * common.opt (feliminate-dwarf2-dups): Ignore.
9765 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
9766 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
9767 same_die_p_wrap, compute_section_prefix,
9768 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
9769 (comdat_symbol_id, comdat_symbol_number): Likewise.
9770 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
9772 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
9773 (output_die): Mark unreachable path unreachable.
9774 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
9775 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
9776 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
9777 (dwarf2out_early_finish): Likewise.
9779 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
9781 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
9783 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
9786 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
9787 not VAR_P. Filter attribute warnings with OPT_Wattributes.
9788 (avr_attribute_table) <io, io_low, address>: Initialize
9789 .decl_required with true.
9791 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
9793 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
9794 undocumented debugging options.
9795 (-mvsx-scalar-double): Likewise.
9796 (-mallow-df-permute): Likewise.
9797 (-mvectorize-builtins): Likewise.
9798 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
9799 (rs6000_builtin_vectorized_function): Likewise.
9800 (rs6000_builtin_md_vectorized_function): Likewise.
9801 (rs6000_opt_vars): Likewise.
9803 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
9806 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
9807 (*btrq_imm): Rename from *btrq.
9808 (*btcq_imm): Rename from *btcq.
9809 (btsc): New code attribute.
9810 (*<btsc><mode>): New insn pattern.
9811 (*btr<mode>): Ditto.
9812 (*<btsc><mode>_mask): New insn_and_split pattern.
9813 (*btr<mode>_mask): Ditto.
9815 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9816 Alan Hayward <alan.hayward@arm.com>
9817 David Sherwood <david.sherwood@arm.com>
9819 * function.c (pad_below): Simplify padding calculation.
9821 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9822 Alan Hayward <alan.hayward@arm.com>
9823 David Sherwood <david.sherwood@arm.com>
9825 * target.def (function_prologue): Remove frame size argument.
9826 (function_epilogue): Likewise.
9827 * doc/tm.texi: Regenerate.
9828 * final.c (final_start_function): Update call to function_prologue.
9829 (final_end_function): Update call to function_epilogue.
9830 (default_function_pro_epilogue): Remove frame size argument.
9831 * output.h (default_function_pro_epilogue): Likewise.
9832 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
9833 (arm_output_function_prologue): Likewise.
9834 * config/frv/frv.c (frv_function_prologue): Likewise.
9835 (frv_function_epilogue): Likewise.
9836 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
9837 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
9838 (ia64_output_function_epilogue): Likewise.
9839 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
9840 (m32r_output_function_epilogue): Likewise.
9841 * config/microblaze/microblaze.c (microblaze_function_prologue)
9842 (microblaze_function_epilogue): Likewise.
9843 * config/mips/mips.c (mips_output_function_prologue): Likewise.
9844 (mips_output_function_epilogue): Likewise.
9845 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
9846 (mmix_target_asm_function_epilogue): Likewise.
9847 * config/msp430/msp430.c (msp430_start_function): Likewise.
9848 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
9849 (nds32_asm_function_epilogue): Likewise.
9850 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
9851 * config/pa/pa.c (pa_output_function_prologue): Likewise.
9852 (pa_output_function_epilogue): Likewise.
9853 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
9854 (rs6000_output_function_epilogue): Likewise.
9855 * config/rl78/rl78.c (rl78_start_function): Likewise.
9856 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
9857 (rs6000_output_function_epilogue): Likewise.
9858 * config/rx/rx.c (rx_output_function_prologue): Likewise.
9859 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
9860 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
9861 (sparc_asm_function_epilogue): Likewise.
9863 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9865 * tree.h (type_has_mode_precision_p): New function.
9866 * convert.c (convert_to_integer_1): Use it.
9867 * expr.c (expand_expr_real_2): Likewise.
9868 (expand_expr_real_1): Likewise.
9869 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
9870 * match.pd: Likewise.
9871 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
9872 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
9873 * tree-tailcall.c (process_assignment): Likewise.
9874 * tree-vect-loop.c (vectorizable_reduction): Likewise.
9875 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
9876 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
9877 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
9878 (vectorizable_assignment): Likewise.
9879 (vectorizable_shift): Likewise.
9880 (vectorizable_operation): Likewise.
9881 * tree-vrp.c (register_edge_assert_for_2): Likewise.
9883 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
9885 * match.pd: Add pow (C, x) simplification.
9887 2017-08-21 Richard Biener <rguenther@suse.de>
9889 PR tree-optimization/81900
9890 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
9891 for blocks with abnormal predecessors.
9892 (compute_antic): Do not set visited flag prematurely.
9894 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
9897 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
9899 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
9901 * stor-layout.h (vector_type_mode): Move to...
9902 * tree.h (vector_type_mode): ...here.
9903 * stor-layout.c (vector_type_mode): Move to...
9904 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
9906 2017-08-21 Richard Biener <rguenther@suse.de>
9908 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
9909 register_external_die hooks.
9910 (debug_false_tree_charstarstar_uhwistar): Declare.
9911 (debug_nothing_tree_charstar_uhwi): Likewise.
9912 * debug.c (do_nothing_debug_hooks): Adjust.
9913 (debug_false_tree_charstarstar_uhwistar): New do nothing.
9914 (debug_nothing_tree_charstar_uhwi): Likewise.
9915 * dbxout.c (dbx_debug_hooks): Adjust.
9916 (xcoff_debug_hooks): Likewise.
9917 * sdbout.c (sdb_debug_hooks): Likewise.
9918 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
9919 * dwarf2out.c (macinfo_label_base): New global.
9920 (dwarf2out_register_external_die): New function for the
9921 register_external_die hook.
9922 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
9923 (dwarf2_debug_hooks): Use them.
9924 (dwarf2_lineno_debug_hooks): Adjust.
9925 (struct die_struct): Add with_offset flag.
9926 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
9927 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
9928 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
9929 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
9930 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
9931 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
9932 defining section names for the early LTO debug variants.
9933 (reset_indirect_string): New helper.
9934 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
9935 (print_dw_val): Add support for offsetted symbol references.
9936 (get_ultimate_context): Split out from is_cxx.
9937 (is_cxx): Use get_ultimate_context.
9938 (is_fortran): Add decl overload.
9939 (compute_comp_unit_symbol): Split out worker from
9940 compute_section_prefix.
9941 (compute_section_prefix): Call compute_comp_unit_symbol and
9942 set comdat_type_p here.
9943 (output_die): Skip DIE symbol output for the LTO added one.
9944 Handle DIE symbol references with offset.
9945 (output_comp_unit): Guard section name mangling properly.
9946 For LTO debug sections emit a symbol at the section beginning
9947 which we use to refer to its DIEs.
9948 (add_abstract_origin_attribute): For DIEs registered via
9949 dwarf2out_register_external_die directly refer to the early
9950 DIE rather than indirectly through the shadow one we created.
9951 Remove obsolete call to dwarf2out_abstract_function for
9952 non-function/block origins.
9953 (gen_array_type_die): When generating early LTO debug do
9954 not emit DW_AT_string_length.
9955 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
9956 late when in LTO. As suggested place a gcc_unreachable for
9957 the DECL_ABSTRACT_P case.
9958 (gen_subprogram_die): Avoid another specification DIE
9959 for early built declarations/definitions for the late LTO case.
9960 (gen_variable_die): Add type references for late duplicated VLA dies
9962 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
9963 we have the abstract instance already.
9964 (process_scope_var): Adjust decl DIE contexts in LTO which
9965 first puts them in limbo.
9966 (gen_decl_die): Do not generate type DIEs late apart from
9967 types for VLAs or for decls we do not yet have a DIE. Do not
9968 call dwarf2out_abstract_function late.
9969 (dwarf2out_early_global_decl): Make sure to create DIEs
9970 for abstract instances of a decl first.
9971 (dwarf2out_late_global_decl): Adjust comment.
9972 (output_macinfo_op): With multiple macro sections use
9973 macinfo_label_base to distinguish labels.
9974 (output_macinfo): Likewise. Update macinfo_label_base.
9975 Pass in the line info label.
9976 (note_variable_value_in_expr): When generating LTO resolve
9977 all variable values here by generating DIEs as needed.
9978 (init_sections_and_labels): Add early LTO debug flag parameter
9979 and generate different sections and names if set. Add generation
9980 counter for the labels so we can have multiple of them.
9981 (reset_dies): Helper to allow DIEs to be output multiple times.
9982 (dwarf2out_finish): When outputting DIEs to the fat part of an
9983 LTO object first reset DIEs.
9984 (dwarf2out_early_finish): Output early DIEs when generating LTO.
9985 (modified_type_die): Check for decl_ultimate_origin being self
9987 (gen_type_die_with_usage): Likewise.
9988 (gen_typedef_die): Allow decl_ultimate_origin being self.
9989 (set_decl_abstract_flags): Remove.
9990 (set_block_abstract_flags): Likewise.
9991 (dwarf2out_abstract_function): Treat the early generated DIEs
9992 as the abstract copy and only add DW_AT_inline and
9993 DW_AT_artificial here and call set_decl_origin_self.
9994 If the DIE has an abstract origin don't do anything.
9995 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
9996 if we have none yet (Go fails to build one, PR78628).
9997 (variably_modified_type_p): Prevent endless recursion for Ada
9998 cyclic pointer types.
9999 * lto-streamer-in.c: Include debug.h.
10000 (dref_queue): New global.
10001 (lto_read_tree_1): Stream in DIE references.
10002 (lto_input_tree): Register DIE references.
10003 (input_function): Stream DECL_DEBUG_ARGS.
10004 * lto-streamer-out.c: Include debug.h.
10005 (lto_write_tree_1): Output DIE references.
10006 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
10007 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
10008 (output_function): Stream DECL_DEBUG_ARGS.
10009 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
10010 Stream DECL_ABSTRACT_ORIGIN.
10011 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
10012 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
10013 DECL_CONTEXT for file-scope decls.
10014 * lto-streamer.h (struct dref_entry): Declare.
10015 (dref_queue): Likewise.
10016 * cfgexpand.c (pass_expand::execute): Do not call the
10017 outlining_inline_function hook here.
10018 * lto-wrapper.c (debug_obj): New global.
10019 (tool_cleanup): Unlink it if required.
10020 (debug_objcopy): New function.
10021 (run_gcc): Handle early debug sections in the IL files by
10022 extracting them to separate files, partially linkin them and
10023 feeding the result back as result to the linker.
10024 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
10025 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
10026 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
10027 sections into a separate segment.
10028 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
10030 (darwin_asm_dwarf_section): Likewise.
10031 (darwin_asm_output_dwarf_offset): Likewise.
10032 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
10034 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10035 Alan Hayward <alan.hayward@arm.com>
10036 David Sherwood <david.sherwood@arm.com>
10038 * read-md.h (md_reader::record_potential_iterator_use): Replace
10039 pointer argument with an rtx and an index.
10040 * read-rtl.c (iterator_group::apply_iterator): Likewise.
10041 (apply_mode_iterator): Likewise.
10042 (apply_code_iterator): Likewise.
10043 (apply_int_iterator): Likewise.
10044 (apply_subst_iterator): Likewise.
10045 (record_iterator_use): Likewise.
10046 (record_attribute_use): Likewise.
10047 (md_reader::record_potential_iterator_use): Likewise. Update calls
10048 to record_iterator_use and apply_iterator.
10049 (iterator_use): Replace ptr with x and index.
10050 (attribute_use): Likewise.
10051 (apply_attribute_uses): Update calls to apply_iterator.
10052 (apply_iterators): Likewise. Update initialization of iterator_use.
10053 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
10054 and record_potential_iterator_use.
10055 (rtx_reader::read_rtx_operand): Likewise.
10057 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10058 Alan Hayward <alan.hayward@arm.com>
10059 David Sherwood <david.sherwood@arm.com>
10061 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
10064 2017-08-21 Richard Biener <rguenther@suse.de>
10066 PR middle-end/81884
10067 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
10068 at struct end conservatively when comparing common bases.
10070 2017-08-21 Richard Biener <rguenther@suse.de>
10072 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
10073 (mem_ref_in_stmt): Remove.
10074 (determine_max_movement): Use ref index to get at the reference.
10075 (invariantness_dom_walker::before_dom_children): Deal with
10076 lim data already initialized.
10077 (gather_mem_refs_stmt): Initialize lim data and record ref index.
10079 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
10081 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
10082 (TARGET_ISA_ROUND): Ditto.
10083 (TARGET_ROUND): Ditto.
10084 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
10085 * config/i386/i386.md: Ditto.
10086 * config/i386/sse.md: Ditto.
10087 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
10088 with OPTION_MASK_ISA_SSE4_1.
10090 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
10093 * doc/extend.texi (x86 Built-in Functions): Correct the name of
10094 __builtin_ia32_lzcnt_u16.
10096 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
10099 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
10100 (rs6000_set_current_function): Rewrite function to use it.
10102 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
10105 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
10106 and TYPE_WARN_IF_NOT_ALIGN.
10107 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
10108 (handle_warn_if_not_align): New.
10109 (place_union_field): Call handle_warn_if_not_align.
10110 (place_field): Call handle_warn_if_not_align.
10111 Copy TYPE_WARN_IF_NOT_ALIGN.
10112 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
10113 (layout_type): Likewise.
10114 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
10116 (tree_decl_common): Add warn_if_not_align.
10117 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
10118 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
10119 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
10120 (DECL_WARN_IF_NOT_ALIGN): Likewise.
10121 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
10122 * doc/extend.texi: Document warn_if_not_aligned attribute.
10123 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
10125 2017-08-17 Martin Liska <mliska@suse.cz>
10128 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
10129 (get_data_dependence): Use it as pointer type.
10130 (distribute_loop): Likewise.
10132 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10134 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
10135 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
10136 (p8_vmrgow_<mode>_direct): New define_insn.
10137 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
10138 handle endianness for vmrgew and vmrgow permute patterns.
10140 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
10142 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
10143 * config/rs6000/rs6000-cpus.def: Remove comment.
10144 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
10145 (POWERPC_MASKS): Likewise.
10146 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
10147 use of TARGET_VSX_TIMODE.
10148 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
10149 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
10150 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
10151 (rs6000_option_override_internal): Remove dead code.
10152 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
10153 (rs6000_legitimize_reload_address): Likewise.
10154 (rs6000_legitimate_address_p): Likewise.
10155 (rs6000_opt_masks): Delete "vsx-timode".
10156 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
10157 from function comment.
10158 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
10159 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
10160 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
10162 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
10163 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
10164 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
10165 (VSX_TI): Likewise.
10167 (define_peephole2): Likewise.
10169 2017-08-17 Martin Sebor <msebor@redhat.com>
10172 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
10173 past the end of an array.
10174 (test_pp_format): Add test cases.
10176 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
10178 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
10179 missing ECF_NOTHROW flags.
10181 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
10184 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
10185 operands residing in integer registers.
10186 (*vsx_le_perm_load_<mode>): Likewise.
10187 (*vsx_le_perm_store_<mode>): Likewise.
10188 (define_peephole2): Add peepholes to optimize the above.
10190 2017-08-17 Marek Polacek <polacek@redhat.com>
10192 PR middle-end/81814
10193 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
10194 to mimic what shorten_compare did. Change the return type to bool.
10195 (fold_cond_expr_with_comparison): Update call to
10196 operand_equal_for_comparison_p.
10197 (fold_ternary_loc): Likewise.
10199 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
10201 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
10203 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
10205 2017-08-17 Richard Biener <rguenther@suse.de>
10207 * tree-ssa-structalias.c (solve_graph): When propagating
10208 to successors update the graphs succ edges and avoid duplicate work.
10210 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
10213 * config/i386/i386.c (ix86_option_override_internal): Save target
10214 specific options after ix86_stack_protector_guard_reg was changed.
10216 2017-08-17 Richard Biener <rguenther@suse.de>
10218 PR tree-optimization/81827
10219 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
10220 (new_var_info): Initialize it conservatively.
10221 (get_call_vi): Mark register vars.
10222 (new_scalar_tmp_constraint_exp): Likewise.
10223 (handle_rhs_call): Likewise.
10224 (handle_const_call): Likewise.
10225 (create_function_info_for): Likewise.
10226 (solve_constraints): Sort varinfos to separate register from
10227 non-register vars to pack points-to solution bitmaps during
10230 2017-08-17 Marek Polacek <polacek@redhat.com>
10232 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
10234 2017-08-17 Richard Biener <rguenther@suse.de>
10236 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
10237 to true when overflow is undefined and we saturated the result.
10239 2017-08-17 Alan Modra <amodra@gmail.com>
10242 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
10243 Don't use store multiple if only one reg needs saving.
10244 (interesting_frame_related_regno): New function.
10245 (rs6000_frame_related): Don't emit frame info for regs that
10247 (rs6000_emit_epilogue): Likewise.
10249 2017-08-16 Nathan Sidwell <nathan@acm.org>
10251 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
10252 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
10253 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
10254 * tree.c (free_lang_data_in_type): Use else-if chain. Always
10255 clear TYPE_LANG_1. Remove obsolete member-function stripping.
10256 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
10257 (verify_type): Adjust for TYPE_BINFO move.
10258 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
10259 process TYPE_BINFO directly.
10260 (hash_tree): Likewise.
10261 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
10263 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
10266 2017-08-16 David Malcolm <dmalcolm@redhat.com>
10268 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
10270 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
10273 * config/i386/i386.md (*anddi_1_btr): Change predicates of
10274 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
10275 Add ix86_binary_operator_ok to insn constraint.
10276 (*iordi_1_bts): Ditto.
10277 (*xordi_1_btc): Ditto.
10278 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
10279 Update corresponding peephole2 pattern.
10283 2017-08-16 Bin Cheng <bin.cheng@arm.com>
10285 PR tree-optimization/81832
10286 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
10287 copy loop header which has IFN_LOOP_DIST_ALIAS call.
10289 2017-08-16 Marek Polacek <polacek@redhat.com>
10292 * fold-const.c (fold_indirect_ref_1): Restore original behavior
10293 regarding size_zero_node.
10295 2017-08-16 Martin Liska <mliska@suse.cz>
10298 * config.gcc: Respect previously set extra_objs in case
10301 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
10303 PR tree-optimization/81835
10304 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
10305 the phi SSA_NAME. Check that the condition in a COND_EXPR does
10306 not depend on the phi.
10308 2017-08-16 Alan Modra <amodra@gmail.com>
10310 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
10313 2017-08-16 Alan Modra <amodra@gmail.com>
10315 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
10316 (save_reg_p): ..into this. Update all callers.
10317 (first_reg_to_save): Simplify.
10319 2017-08-16 Alan Modra <amodra@gmail.com>
10321 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
10324 2017-08-15 Joseph Myers <joseph@codesourcery.com>
10328 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
10329 constant count if that count is less than 32.
10331 2017-08-15 Nathan Sidwell <nathan@acm.org>
10333 * gcc.c (execute): Emit friendlier message if inferior is killed
10334 by an external cause.
10336 2017-08-15 Richard Biener <rguenther@suse.de>
10338 PR tree-optimization/81790
10339 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
10340 CONSTRUCTORs from simplifying and VN.
10342 2017-08-14 Martin Sebor <msebor@redhat.com>
10344 * builtin-attrs.def: Add comments.
10346 2017-08-14 Martin Sebor <msebor@redhat.com>
10349 * doc/extend.texi (attribute nonstring): Document new attribute.
10351 2017-08-14 Martin Sebor <msebor@redhat.com>
10354 * tree-diagnostic.c (default_tree_printer): Handle %G.
10355 * gimple-pretty-print.h (percent_G_format): Declare new function.
10356 * gimple-pretty-print.c (percent_G_format): Define.
10357 * tree-pretty-print.c (percent_K_format): Add argument.
10359 2017-08-14 Martin Sebor <msebor@redhat.com>
10361 PR translation/79998
10362 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
10363 Remove a stray space.
10365 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
10368 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
10369 (*iordi_1_bts): Ditto.
10370 (*xordi_1_btc): Ditto.
10372 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10375 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
10376 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10378 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
10379 quoted strings, and make more translator-friendly.
10380 (darwin_rs6000_override_options): Likewise.
10381 (rs6000_option_override_internal): Likewise.
10382 (rs6000_return_in_memory): Fix overlong line.
10383 (init_cmulative_args): Use quoted strings, and make more
10384 translator-friendly.
10385 (rs6000_pass_by_reference): Fix overlong line.
10386 (def_builtin): Use quoted strings.
10387 (altivec_expand_predicate_builtin): Use quoted strings, and make
10388 more translator-friendly.
10389 (htm_expand_builtin): Use quoted strings.
10390 (cpu_expand_builtin): Use quoted strings, and make more
10391 translator-friendly.
10392 (altivec_expand_builtin): Likewise.
10393 (paired_expand_predicate_builtin): Likewise.
10394 (rs6000_invalid_builtin): Likewise.
10395 (builtin_function_type): Use quoted strings.
10396 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
10397 more translator-friendly.
10398 (rs6000_trampoline_init): Likewise.
10399 (rs6000_handle_altivec_attribute): Likewise.
10400 (rs6000_inner_target_options): Use quoted strings.
10401 (rs6000_disable_incompatible_switches): Likewise.
10402 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
10403 strings, and make more translator-friendly.
10404 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
10406 2017-08-14 Bin Cheng <bin.cheng@arm.com>
10408 PR tree-optimization/81799
10409 * tree-loop-distribution.c (version_loop_by_alias_check): Force
10410 cond_expr to simple gimple operand.
10412 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
10414 PR middle-end/46932
10415 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
10417 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
10421 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
10422 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
10423 TARGET_GASISR_PROLOGUES.
10424 * config/avr/avr.c (avr_option_override): Same.
10425 (avr_pass_pre_proep::execute): Same.
10427 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
10430 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
10431 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
10433 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
10435 * config/i386/i386.md (*load_tp_<mode>): Redefine as
10436 define_insn_and_split. Split to a memory load from 0 in
10437 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
10438 using PTR mode iterator.
10439 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
10440 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
10441 (*add_tp_<mode>): Redefine as define_insn_and_split.
10442 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
10443 address space. Merge with *add_tp_x32 using PTR mode iterator.
10444 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
10445 Split to an add with a memory load from 0 in
10446 DEFAULT_TLS_SEG_REG address space.
10448 2017-08-12 Andrew Pinski <apinski@cavium.com>
10450 * config/aarch64/aarch64-option-extensions.def (rdma):
10451 Fix feature string to what Linux prints out in /proc/cpuinfo.
10453 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
10456 * dwarf2out.c (modified_type_die): For C typedef types that have
10457 an ultimate origin, process the ultimate origin instead of the
10459 (gen_typedef_die): Assert that input DECLs have no ultimate
10461 (gen_type_die_with_usage): For typedef variants that have an
10462 ultimate origin, just call gen_decl_die on the original DECL.
10463 (process_scope_var): Avoid creating DIEs for local typedefs and
10464 concrete static variables.
10466 2017-08-12 Alan Modra <amodra@gmail.com>
10470 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
10471 match gnu-user.h startfile.
10472 (ENDFILE_LINUX_SPEC): Similarly.
10474 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
10477 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
10479 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
10481 2017-08-11 Tamar Christina <tamar.christina@arm.com>
10482 * config/aarch64/aarch64.md (mov<mode>): Change.
10483 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
10484 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
10485 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
10487 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
10489 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
10490 for storage order barriers.
10492 2017-08-11 Martin Liska <mliska@suse.cz>
10495 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
10496 variables of void type.
10498 2017-08-11 Martin Liska <mliska@suse.cz>
10500 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
10501 TARGET_SUPPORTS_ALIASES.
10502 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
10503 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
10504 (optimize_weakref): Likewise.
10505 * symtab.c (symtab_node::noninterposable_alias): Likewise.
10506 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
10507 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
10509 2017-08-11 Martin Liska <mliska@suse.cz>
10512 * config/i386/i386.c (make_resolver_func): Do complete
10513 refactoring of the function.
10515 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
10518 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
10519 * config/i386/i386.c (ix86_stack_protect_guard): Use
10520 ix86_stack_protect_guard_symbol_str to generate varible declaration.
10521 * doc/invoke.texi (x86 Options): Document
10522 -mstack-protector-guard-symbol= option.
10524 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
10526 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
10527 * config/i386/i386.c (ix86_split_stack_guard): New function.
10528 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
10529 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
10530 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
10531 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
10532 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
10533 (split_stack_space_check): Call ix86_split_stack_guard.
10535 2017-08-10 Martin Sebor <msebor@redhat.com>
10537 * print-tree.c (print_node): Print location using the established
10539 Replace spaces with colons.
10540 (debug_raw, debug): Ditto.
10542 2017-08-10 Martin Sebor <msebor@redhat.com>
10545 * pretty-print.c (pp_format): Correct the handling of %s precision.
10547 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
10550 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
10552 (ix86_finalize_stack_frame_flags): This. Also clear
10553 frame_pointer_needed if -fno-omit-frame-pointer is used without
10555 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
10556 with ix86_finalize_stack_frame_flags.
10557 (ix86_expand_epilogue): Likewise.
10558 (ix86_expand_split_stack_prologue): Likewise.
10559 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
10561 2017-08-10 Martin Liska <mliska@suse.cz>
10564 * c-attribs.c (handle_target_attribute):
10565 Report warning for an empty string argument of target attribute.
10567 2017-08-09 Jakub Jelinek <jakub@redhat.com>
10570 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
10572 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
10573 or DECL_NONLOCAL labels.
10574 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
10575 or DECL_NONLOCAL labels here.
10577 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
10579 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
10580 to indicate when early gimple folding has been disabled.
10581 (rs6000_gimple_fold_builtin): Add debug content.
10582 (rs6000_invalid_builtin): Fix whitespace.
10583 (rs6000_expand_builtin): Fix whitespace.
10584 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
10586 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
10589 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
10590 SAVE_MULTIPLE if not all the registers that saves, should be saved.
10592 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
10594 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
10595 (qdf24xx): Likewise.
10596 * config/aarch64/aarch64.md: Include falkor.md.
10597 * config/aarch64/falkor.md: New.
10599 2017-08-09 Marek Polacek <polacek@redhat.com>
10602 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
10603 * diagnostic.c (emit_diagnostic): Add a comment.
10604 (emit_diagnostic_valist): New function.
10606 2017-08-09 Marek Polacek <polacek@redhat.com>
10609 * input.c (make_location): New overload.
10610 * input.h (make_location): Declare.
10612 2017-08-08 Alan Modra <amodra@gmail.com>
10613 H.J. Lu <hongjiu.lu@intel.com>
10616 * gcc.c (NO_PIE_SPEC): Delete.
10617 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
10619 (LINK_PIE_SPEC): ..to here.
10620 (LINK_COMMAND_SPEC): Support -no-pie.
10621 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
10622 chain of crtbegin*.o selection, update for PIE_SPEC changes and
10624 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
10625 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
10626 (ENDFILE_CRTEND_SPEC): Similarly.
10628 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
10631 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
10632 (mstack-protector-guard-offset=): Ditto.
10633 * config/i386/i386.c (ix86_option_override): Handle
10634 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
10636 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
10637 ix86_stack_protect_guard_offset variables.
10638 (TARGET_STACK_PROTECT_GUARD): Always define.
10639 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
10640 and -mstack-protector-guard-offset= options.
10642 2017-08-08 Bin Cheng <bin.cheng@arm.com>
10644 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
10645 boundary case for the last candidate.
10647 2017-08-08 Bin Cheng <bin.cheng@arm.com>
10649 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
10650 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
10652 2017-08-08 Tamar Christina <tamar.christina@arm.com>
10654 PR middle-end/19706
10655 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
10656 * config/aarch64/aarch64-builtins.c
10657 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
10658 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
10659 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
10661 2017-08-08 Tamar Christina <tamar.christina@arm.com>
10662 Andrew Pinski <pinskia@gmail.com>
10664 PR middle-end/19706
10665 * internal-fn.def (XORSIGN): New.
10666 * optabs.def (xorsign_optab): New.
10667 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
10668 (convert_expand_mult_copysign): New.
10669 (pass_optimize_widening_mul::execute): Call
10670 convert_expand_mult_copysign.
10672 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10674 PR tree-optimization/81354
10675 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
10676 Insert on edges rather than explicitly creating landing pads.
10677 (analyze_candidates_and_replace): Commit edge inserts.
10679 2017-08-08 Richard Biener <rguenther@suse.de>
10681 PR middle-end/81719
10682 * tree-ssa-loop-niter.c: Include tree-dfa.h.
10683 (expand_simple_operations): Also look through ADDR_EXPRs with
10684 MEM_REF bases treating them as POINTER_PLUS_EXPR.
10686 2017-08-08 Richard Biener <rguenther@suse.de>
10688 PR tree-optimization/81723
10689 * tree-vect-slp.c (struct bst_traits): New hash traits.
10690 (bst_fail): New global.
10691 (vect_build_slp_tree_2): New worker, split out from ...
10692 (vect_build_slp_tree): ... this now wrapping it with using
10693 bst_fail set to cache SLP tree build fails. Properly handle
10695 (vect_analyze_slp_instance): Allocate and free bst_fail.
10697 2017-08-08 Martin Liska <mliska@suse.cz>
10700 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
10701 LABEL_DECLs that can be from a different function.
10703 2017-08-08 Bin Cheng <bin.cheng@arm.com>
10705 PR tree-optimization/81744
10706 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
10707 loop's number of iterations.
10709 2017-08-08 Martin Liska <mliska@suse.cz>
10711 * asan.c: Include header files.
10712 * attribs.c (build_decl_attribute_variant): New function moved
10714 (build_type_attribute_qual_variant): Likewise.
10715 (cmp_attrib_identifiers): Likewise.
10716 (simple_cst_list_equal): Likewise.
10717 (omp_declare_simd_clauses_equal): Likewise.
10718 (attribute_value_equal): Likewise.
10719 (comp_type_attributes): Likewise.
10720 (build_type_attribute_variant): Likewise.
10721 (lookup_ident_attribute): Likewise.
10722 (remove_attribute): Likewise.
10723 (merge_attributes): Likewise.
10724 (merge_type_attributes): Likewise.
10725 (merge_decl_attributes): Likewise.
10726 (merge_dllimport_decl_attributes): Likewise.
10727 (handle_dll_attribute): Likewise.
10728 (attribute_list_equal): Likewise.
10729 (attribute_list_contained): Likewise.
10730 * attribs.h (lookup_attribute): New function moved from tree.[ch].
10731 (lookup_attribute_by_prefix): Likewise.
10732 * bb-reorder.c: Include header files.
10733 * builtins.c: Likewise.
10734 * calls.c: Likewise.
10735 * cfgexpand.c: Likewise.
10736 * cgraph.c: Likewise.
10737 * cgraphunit.c: Likewise.
10738 * convert.c: Likewise.
10739 * dwarf2out.c: Likewise.
10740 * final.c: Likewise.
10741 * fold-const.c: Likewise.
10742 * function.c: Likewise.
10743 * gimple-expr.c: Likewise.
10744 * gimple-fold.c: Likewise.
10745 * gimple-pretty-print.c: Likewise.
10746 * gimple.c: Likewise.
10747 * gimplify.c: Likewise.
10748 * hsa-common.c: Likewise.
10749 * hsa-gen.c: Likewise.
10750 * internal-fn.c: Likewise.
10751 * ipa-chkp.c: Likewise.
10752 * ipa-cp.c: Likewise.
10753 * ipa-devirt.c: Likewise.
10754 * ipa-fnsummary.c: Likewise.
10755 * ipa-inline.c: Likewise.
10756 * ipa-visibility.c: Likewise.
10758 * lto-cgraph.c: Likewise.
10759 * omp-expand.c: Likewise.
10760 * omp-general.c: Likewise.
10761 * omp-low.c: Likewise.
10762 * omp-offload.c: Likewise.
10763 * omp-simd-clone.c: Likewise.
10764 * opts-global.c: Likewise.
10765 * passes.c: Likewise.
10766 * predict.c: Likewise.
10767 * sancov.c: Likewise.
10768 * sanopt.c: Likewise.
10769 * symtab.c: Likewise.
10770 * toplev.c: Likewise.
10771 * trans-mem.c: Likewise.
10772 * tree-chkp.c: Likewise.
10773 * tree-eh.c: Likewise.
10774 * tree-into-ssa.c: Likewise.
10775 * tree-object-size.c: Likewise.
10776 * tree-parloops.c: Likewise.
10777 * tree-profile.c: Likewise.
10778 * tree-ssa-ccp.c: Likewise.
10779 * tree-ssa-live.c: Likewise.
10780 * tree-ssa-loop.c: Likewise.
10781 * tree-ssa-sccvn.c: Likewise.
10782 * tree-ssa-structalias.c: Likewise.
10783 * tree-ssa.c: Likewise.
10784 * tree-streamer-in.c: Likewise.
10785 * tree-vectorizer.c: Likewise.
10786 * tree-vrp.c: Likewise.
10787 * tsan.c: Likewise.
10788 * ubsan.c: Likewise.
10789 * varasm.c: Likewise.
10790 * varpool.c: Likewise.
10791 * tree.c: Remove functions moved to attribs.[ch].
10792 * tree.h: Likewise.
10793 * config/aarch64/aarch64.c: Add attrs.h header file.
10794 * config/alpha/alpha.c: Likewise.
10795 * config/arc/arc.c: Likewise.
10796 * config/arm/arm.c: Likewise.
10797 * config/avr/avr.c: Likewise.
10798 * config/bfin/bfin.c: Likewise.
10799 * config/c6x/c6x.c: Likewise.
10800 * config/cr16/cr16.c: Likewise.
10801 * config/cris/cris.c: Likewise.
10802 * config/darwin.c: Likewise.
10803 * config/epiphany/epiphany.c: Likewise.
10804 * config/fr30/fr30.c: Likewise.
10805 * config/frv/frv.c: Likewise.
10806 * config/ft32/ft32.c: Likewise.
10807 * config/h8300/h8300.c: Likewise.
10808 * config/i386/winnt.c: Likewise.
10809 * config/ia64/ia64.c: Likewise.
10810 * config/iq2000/iq2000.c: Likewise.
10811 * config/lm32/lm32.c: Likewise.
10812 * config/m32c/m32c.c: Likewise.
10813 * config/m32r/m32r.c: Likewise.
10814 * config/m68k/m68k.c: Likewise.
10815 * config/mcore/mcore.c: Likewise.
10816 * config/microblaze/microblaze.c: Likewise.
10817 * config/mips/mips.c: Likewise.
10818 * config/mmix/mmix.c: Likewise.
10819 * config/mn10300/mn10300.c: Likewise.
10820 * config/moxie/moxie.c: Likewise.
10821 * config/msp430/msp430.c: Likewise.
10822 * config/nds32/nds32-isr.c: Likewise.
10823 * config/nds32/nds32.c: Likewise.
10824 * config/nios2/nios2.c: Likewise.
10825 * config/nvptx/nvptx.c: Likewise.
10826 * config/pa/pa.c: Likewise.
10827 * config/pdp11/pdp11.c: Likewise.
10828 * config/powerpcspe/powerpcspe.c: Likewise.
10829 * config/riscv/riscv.c: Likewise.
10830 * config/rl78/rl78.c: Likewise.
10831 * config/rx/rx.c: Likewise.
10832 * config/s390/s390.c: Likewise.
10833 * config/sh/sh.c: Likewise.
10834 * config/sol2.c: Likewise.
10835 * config/sparc/sparc.c: Likewise.
10836 * config/spu/spu.c: Likewise.
10837 * config/stormy16/stormy16.c: Likewise.
10838 * config/tilegx/tilegx.c: Likewise.
10839 * config/tilepro/tilepro.c: Likewise.
10840 * config/v850/v850.c: Likewise.
10841 * config/vax/vax.c: Likewise.
10842 * config/visium/visium.c: Likewise.
10843 * config/xtensa/xtensa.c: Likewise.
10845 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
10848 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
10849 constraints since the -mupper-regs-* switches have been
10851 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
10852 into a vector from a double word element that was extracted from
10853 another vector, and eliminate extra XXPERMDI instructions.
10854 (vsx_concat_<mode>_2): Likewise.
10855 (vsx_concat_<mode>_3): Likewise.
10856 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
10857 concat to allow optimizing inserts from previous extracts.
10859 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
10861 * config/i386/i386.c (ix86_stack_protect_guard): Generate
10862 memory reference to a SSP offset in TLS address space.
10863 (ix86_print_operand) <case '@'>: Remove.
10864 (ix86_print_operand_punct_valid_p): Remove '@' code.
10865 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
10866 UNSPEC_SP_TLS_TEST.
10867 (stack_tls_protect_set_<mode>): Remove.
10868 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
10869 (stack_tls_protect_test_<mode>): Remove.
10870 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
10872 2017-08-07 Olivier Hainque <hainque@adacore.com>
10875 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
10877 2017-08-07 Douglas Rupp <rupp@adacore.com>
10879 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
10880 variable was referenced as multidir in command.
10882 2017-08-07 Jakub Jelinek <jakub@redhat.com>
10885 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
10888 2017-08-07 Martin Liska <mliska@suse.cz>
10890 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
10891 * config/rl78/rl78.c: Add include of attribs.h.
10892 * config/sh/sh.c: Likewise.
10893 * config/v850/v850.c: Likewise.
10895 2017-08-07 Tom de Vries <tom@codesourcery.com>
10897 PR middle-end/78266
10898 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
10900 2017-08-07 Martin Liska <mliska@suse.cz>
10902 * config/mips/mips.c: Include attribs.h.
10904 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
10907 * doc/invoke.texi: Document change in behvaior for -Ofast for
10910 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
10912 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
10914 (aarch64_pop_regs): Likewise.
10915 (aarch64_gen_load_pair): Likewise.
10916 (aarch64_save_callee_saves): Likewise.
10917 (aarch64_restore_callee_saves): Likewise.
10919 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
10921 * config/i386/i386.c: Revert the last change.
10923 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
10926 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
10928 (ix86_finalize_stack_frame_flags): This. Also clear
10929 frame_pointer_needed if -fno-omit-frame-pointer is used without
10931 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
10932 with ix86_finalize_stack_frame_flags.
10933 (ix86_expand_epilogue): Likewise.
10934 (ix86_expand_split_stack_prologue): Likewise.
10936 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
10939 * config/i386/i386.c (get_builtin_code_for_version): Set priority
10940 to P_AES for Westmere.
10942 2017-08-07 Jonathan Yong <10walls@gmail.com>
10944 * config/i386/mingw.opt (fset-stack-executable): Removed.
10945 * config/i386/cygming.opt (fset-stack-executable): Moved
10947 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
10949 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
10951 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
10953 2017-08-07 Marek Polacek <polacek@redhat.com>
10955 PR middle-end/81737
10956 * fold-const.c (fold_indirect_ref_1): Check type_domain.
10958 2017-08-07 Martin Liska <mliska@suse.cz>
10960 * attribs.h (canonicalize_attr_name): New function.
10961 (cmp_attribs): Move from c-format.c and adjusted.
10962 (is_attribute_p): Moved from tree.h.
10963 * tree-inline.c: Add new includes.
10964 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
10965 (private_is_attribute_p): Remove.
10966 (private_lookup_attribute): Likewise.
10967 (private_lookup_attribute_by_prefix): Simplify.
10968 (remove_attribute): Use is_attribute_p.
10969 * tree.h: Remove removed declarations.
10971 2017-08-07 Jakub Jelinek <jakub@redhat.com>
10973 PR middle-end/81698
10974 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
10975 instead of computing it in the function. Formatting fix.
10976 (expand_case): Don't rely on default_edge being the first edge,
10977 clear it if removing it, pass default_edge to
10978 emit_case_dispatch_table.
10979 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
10982 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
10984 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
10985 insn in the function, emit NOP after the insn.
10987 2017-08-06 Tom de Vries <tom@codesourcery.com>
10989 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
10992 2017-08-06 Tom de Vries <tom@codesourcery.com>
10994 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
10997 2017-08-04 Yury Gribov <tetra2005@gmail.com>
10999 PR tree-optimization/57371
11000 * match.pd: New pattern.
11002 2017-08-04 Marek Polacek <polacek@redhat.com>
11004 PR middle-end/81695
11005 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
11006 perform the computation in offset_int.
11008 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
11010 PR tree-optimization/81136
11011 * tree-vectorizer.h: Include tree-hash-traits.h.
11012 (vec_base_alignments): New typedef.
11013 (vec_info): Add a base_alignments field.
11014 (vect_record_base_alignments): Declare.
11015 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
11017 (DR_IS_CONDITIONAL_IN_STMT): New macro.
11018 (create_data_ref): Add an is_conditional_in_stmt argument.
11019 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
11020 the is_conditional_in_stmt field.
11021 (data_ref_loc): Add an is_conditional_in_stmt field.
11022 (get_references_in_stmt): Set the is_conditional_in_stmt field.
11023 (find_data_references_in_stmt): Update call to create_data_ref.
11024 (graphite_find_data_references_in_stmt): Likewise.
11025 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
11026 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
11027 (vect_record_base_alignment): New function.
11028 (vect_record_base_alignments): Likewise.
11029 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
11030 for nested statements even if we fail to compute a misalignment.
11031 Use pooled base alignments for unconditional references.
11032 (vect_find_same_alignment_drs): Compare base addresses instead
11034 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
11035 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
11037 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
11039 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
11040 Add an explicit name for the enum. Use auto_vec for slp_instances
11041 and grouped_stores.
11042 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
11044 (_bb_vec_info): Add a constructor and destructor.
11045 (vinfo_for_stmt): Return NULL for uids of -1 as well.
11046 (destroy_loop_vec_info): Delete.
11047 (vect_destroy_datarefs): Likewise.
11048 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
11049 (vec_info::vec_info): New function.
11050 (vec_info::~vec_info): Likewise.
11051 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
11052 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
11053 destroy_loop_vec_info.
11054 * tree-vect-loop.c (new_loop_vec_info): Replace with...
11055 (_loop_vec_info::_loop_vec_info): ...this.
11056 (destroy_loop_vec_info): Replace with...
11057 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
11058 the stmt_vec_infos. Leave handling of vec_info information to its
11059 destructor. Remove explicit vector releases.
11060 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
11061 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
11062 * tree-vect-slp.c (new_bb_vec_info): Replace with...
11063 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
11064 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
11065 (destroy_bb_vec_info): Replace with...
11066 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
11067 information to its destructor.
11068 (vect_slp_analyze_bb_1): Use new and delete instead of
11069 new_bb_vec_info and destroy_bb_vec_info.
11070 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
11073 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
11075 * tree-data-ref.h (subscript): Add access_fn field.
11076 (data_dependence_relation): Add could_be_independent_p.
11077 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
11078 (same_access_functions): Move to tree-data-ref.c.
11079 * tree-data-ref.c (ref_contains_union_access_p): New function.
11080 (access_fn_component_p): Likewise.
11081 (access_fn_components_comparable_p): Likewise.
11082 (dr_analyze_indices): Add a reference to access_fn_component_p.
11083 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
11085 (constant_access_functions): Likewise.
11086 (add_other_self_distances): Likewise.
11087 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
11088 (initialize_data_dependence_relation): Use XCNEW and remove
11089 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
11090 of access functions that have the same type. Allow the
11091 subsequence to end with different bases in some circumstances.
11092 Record the chosen access functions in SUB_ACCESS_FN.
11093 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
11094 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
11095 (subscript_dependence_tester_1): Likewise dra and drb.
11096 (build_classic_dist_vector): Update calls accordingly.
11097 (subscript_dependence_tester): Likewise.
11098 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
11099 DDR_COULD_BE_INDEPENDENT_P.
11100 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
11101 comp_alias_ddrs instead of may_alias_ddrs.
11102 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
11104 (vect_analyze_data_ref_dependence): Use it if
11105 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
11106 distance vectors if that fails.
11107 (dependence_distance_ge_vf): New function.
11108 (vect_prune_runtime_alias_test_list): Use it. Don't clear
11109 LOOP_VINFO_MAY_ALIAS_DDRS.
11111 2017-08-04 Richard Biener <rguenther@suse.de>
11113 PR middle-end/81705
11114 * fold-const.c (fold_binary_loc): Properly restrict
11115 minus_var0 && minus_var1 case when associating undefined overflow
11118 2017-08-04 Tom de Vries <tom@codesourcery.com>
11120 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
11122 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11124 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11125 Don't start diagnostic messages with a capital letter.
11126 * config/rs6000/rs6000.c (rs6000_option_override_internal):
11128 (rs6000_invalid_builtin): Likewise.
11129 (rs6000_trampoline_init): Likewise.
11131 2017-08-03 Jakub Jelinek <jakub@redhat.com>
11134 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
11135 after setting changeable df flags.
11137 2017-08-03 Richard Biener <rguenther@suse.de>
11139 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
11140 up if the use is in USE - X.
11142 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
11144 * toplev.c (dumpfile.h): New include.
11145 (internal_error_reentered): New static function. Use it...
11146 (internal_error_function): ...here to handle reentered internal_error.
11148 2017-08-03 Richard Biener <rguenther@suse.de>
11150 PR middle-end/81148
11151 * fold-const.c (split_tree): Add minus_var and minus_con
11152 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
11153 here but always use minus_*.
11154 (associate_trees): Assert we never associate with MINUS_EXPR
11155 and NULL first operand. Do not recurse for PLUS_EXPR operands
11156 when associating as MINUS_EXPR either.
11157 (fold_binary_loc): Track minus_var and minus_con.
11159 2017-08-03 Tom de Vries <tom@codesourcery.com>
11162 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
11163 ACCEL_COMPILER, apply finish_options on
11164 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
11166 2017-08-03 Tom de Vries <tom@codesourcery.com>
11169 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
11170 function_entry_patch_area_size > 0.
11172 2017-08-03 Jakub Jelinek <jakub@redhat.com>
11175 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
11176 instead of 10??LU, perform unit multiplication in wide_int,
11177 don't change alloc_object_size_limit if the limit is larger
11180 PR tree-optimization/81655
11181 PR tree-optimization/81588
11182 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
11183 the case when ranges[i].low and high are 1 for unsigned type with
11186 PR middle-end/81052
11187 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
11188 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
11190 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
11192 * tree-vrp.h: Add include guard.
11194 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
11197 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
11198 (ud2): New insn pattern.
11199 * config/i386/i386.c (ix86_expand_epilogue):
11200 For naked functions, generate ud2 instead of trap insn.
11202 2017-08-02 Marek Polacek <polacek@redhat.com>
11205 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
11207 2017-08-02 Tom de Vries <tom@codesourcery.com>
11208 Cesar Philippidis <cesar@codesourcery.com>
11210 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
11211 Add missing edge probabilities.
11213 2017-08-02 Tamar Christina <tamar.christina@arm.com>
11215 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
11216 Correct endianness.
11218 2017-08-02 Jakub Jelinek <jakub@redhat.com>
11220 PR middle-end/79499
11221 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
11222 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
11223 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
11224 prologue_seq sequences - if any.
11226 2017-08-02 Richard Biener <rguenther@suse.de>
11228 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
11229 via vectors if supported, integer extracts via punning if supported
11230 or otherwise vector extracts.
11232 2017-08-02 Richard Biener <rguenther@suse.de>
11234 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
11236 (bitmap_insert_into_set): ... this.
11238 2017-08-02 Richard Biener <rguenther@suse.de>
11240 PR tree-optimization/81633
11242 2015-08-17 Alan Hayward <alan.hayward@arm.com>
11244 PR tree-optimization/71752
11245 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
11247 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
11249 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
11250 (machine_function::call_ms2sysv_pad_out): Remove field.
11251 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
11252 (ix86_compute_frame_layout): Likewise.
11254 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
11257 * config/i386/i386.c (ix86_set_func_type): Disallow naked
11258 attribute with interrupt attribute.
11260 2017-08-01 Andrew Pinski <apinski@cavium.com>
11262 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
11263 BIT_INSERT_EXPR's operand 1
11264 to see if the types precision matches.
11266 2017-08-01 Martin Liska <mliska@suse.cz>
11268 PR middle-end/70140
11269 * builtins.c (expand_builtin_memcpy_args): Remove.
11270 (expand_builtin_memcpy): Call newly added function
11271 expand_builtin_memory_copy_args.
11272 (expand_builtin_memcpy_with_bounds): Likewise.
11273 (expand_builtin_mempcpy): Remove last argument.
11274 (expand_builtin_mempcpy_with_bounds): Likewise.
11275 (expand_builtin_memory_copy_args): New function created from
11276 expand_builtin_mempcpy_args with small modifications.
11277 (expand_builtin_mempcpy_args): Remove.
11278 (expand_builtin_stpcpy): Remove unused argument.
11279 (expand_builtin): Likewise.
11280 (expand_builtin_with_bounds): Likewise.
11282 2017-08-01 Martin Liska <mliska@suse.cz>
11285 Make mempcpy more optimal (PR middle-end/70140).
11287 2017-08-01 Jakub Jelinek <jakub@redhat.com>
11290 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
11291 __builtin_vec_cmpne verify both arguments are compatible vectors
11292 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
11293 verify arg1_type is a pointer or array type. For __builtin_vec_st,
11294 move computation of aligned to after checking the argument types.
11298 * config/rs6000/vsx.md (vextract_fp_from_shorth,
11299 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
11302 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
11303 Doug Rupp <rupp@adacore.com>
11304 Olivier Hainque <hainque@adacore.com>
11306 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
11307 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
11309 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
11310 for TARGET_OS_CPP_BUILTIN.
11311 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
11312 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
11314 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
11315 passing required abi options to the assembler for EABI configurations.
11316 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
11317 of .text.hot and .text.unlikely sections for kernel modules when
11318 using ARM style exceptions.
11319 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
11320 options. Add EXTRA_CC1_SPEC.
11321 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
11323 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
11325 (ARM_TARGET2_DWARF_FORMAT): Define.
11326 * config/arm/t-vxworks: Adjust multilib control to removal of the
11327 Diab command line options.
11329 2017-08-01 Martin Liska <mliska@suse.cz>
11331 PR gcov-profile/81561
11332 * gcov.c (unblock): Make unblocking safe as we need to preserve
11333 index correspondence of blocks and block_lists.
11335 2017-08-01 Richard Biener <rguenther@suse.de>
11337 PR tree-optimization/81181
11338 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
11339 (compute_antic): ... end of iteration here.
11341 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
11343 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
11344 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
11345 (ftree-slp-vectorize): Likewise.
11346 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
11347 can no longer be set independent of flag_tree_loop_vectorize.
11348 * omp-general.c (emp_max_vf): Likewise.
11349 * opts.c (enable_fdo_optimizations): Remove references to
11350 flag_tree_vectorize, these are now implicit.
11351 (common_handle_option): Remove handling for OPT_ftree_vectorize,
11352 and leave it for the options machinery.
11354 2017-08-01 Martin Liska <mliska@suse.cz>
11356 PR middle-end/70140
11357 * builtins.c (expand_builtin_memcpy_args): Remove.
11358 (expand_builtin_memcpy): Call newly added function
11359 expand_builtin_memory_copy_args.
11360 (expand_builtin_memcpy_with_bounds): Likewise.
11361 (expand_builtin_mempcpy): Remove last argument.
11362 (expand_builtin_mempcpy_with_bounds): Likewise.
11363 (expand_builtin_memory_copy_args): New function created from
11364 expand_builtin_mempcpy_args with small modifications.
11365 (expand_builtin_mempcpy_args): Remove.
11366 (expand_builtin_stpcpy): Remove unused argument.
11367 (expand_builtin): Likewise.
11368 (expand_builtin_with_bounds): Likewise.
11370 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
11373 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
11374 print "ds:" only for immediates in generic address space.
11376 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
11379 * config/i386/i386.c (ix86_funciton_naked): New prototype.
11380 (ix86_function_ok_for_sibcall): Return false for naked functions.
11382 2017-08-01 Richard Biener <rguenther@suse.de>
11384 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
11385 (compute_antic): Seed worklist with exit block predecessors.
11386 * cfganal.c (dfs_find_deadend): For a cycle return the source
11387 of the edge closing it.
11389 2017-08-01 Tamar Christina <tamar.christina@arm.com>
11391 * config/aarch64/aarch64.c
11392 (aarch64_can_const_movi_rtx_p): Move 0 check.
11394 2017-08-01 Bin Cheng <bin.cheng@arm.com>
11396 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
11397 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
11399 * match.pd: Ditto in address comparison pattern.
11401 2017-08-01 Bin Cheng <bin.cheng@arm.com>
11403 PR tree-optimization/81627
11404 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
11405 closed ssa form for store-store chain.
11407 2017-08-01 Bin Cheng <bin.cheng@arm.com>
11409 PR tree-optimization/81620
11410 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
11411 for store-store chain.
11413 2017-08-01 Jakub Jelinek <jakub@redhat.com>
11415 PR tree-optimization/81588
11416 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
11417 ranges[i].in_p, invert comparison code ccode. For >/>=,
11418 swap rhs1 and rhs2 and comparison code unconditionally,
11419 for </<= don't do that. Don't swap rhs1/rhs2 again if
11420 ranges[i].in_p, instead invert comparison code ccode if
11421 opcode or oe->rank is BIT_IOR_EXPR.
11424 * optabs.def (vec_extract_optab, vec_init_optab): Change from
11425 a direct optab to conversion optab.
11426 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
11427 with GET_MODE_INNER as last argument instead of optab_handler.
11428 * expmed.c (extract_bit_field_1): Likewise. Use vector from
11429 vector extraction if possible and optab is available.
11430 * expr.c (store_constructor): Use convert_optab_handler instead
11431 of optab_handler. Use vector initialization from smaller
11432 vectors if possible and optab is available.
11433 * tree-vect-stmts.c (vectorizable_load): Likewise.
11434 * doc/md.texi (vec_extract, vec_init): Document that the optabs
11435 now have two modes.
11436 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
11437 of vec_init from half-sized vectors with the same element mode.
11438 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
11439 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
11440 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
11441 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
11442 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
11443 after mode in gen_vec_extract* calls.
11444 (vec_extract<mode>): Renamed to ...
11445 (vec_extract<mode><ssescalarmodelower>): ... this.
11446 (vec_extract<mode><ssehalfvecmodelower>): New expander.
11447 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
11448 element mode after mode in gen_vec_init* calls.
11449 (VEC_INIT_HALF_MODE): New mode iterator.
11450 (vec_init<mode>): Renamed to ...
11451 (vec_init<mode><ssescalarmodelower>): ... this.
11452 (vec_init<mode><ssehalfvecmodelower>): New expander.
11453 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
11454 (vec_extractv2sfsf): ... this.
11455 (vec_initv2sf): Renamed to ...
11456 (vec_initv2sfsf): ... this.
11457 (vec_extractv2si): Renamed to ...
11458 (vec_extractv2sisi): ... this.
11459 (vec_initv2si): Renamed to ...
11460 (vec_initv2sisi): ... this.
11461 (vec_extractv4hi): Renamed to ...
11462 (vec_extractv4hihi): ... this.
11463 (vec_initv4hi): Renamed to ...
11464 (vec_initv4hihi): ... this.
11465 (vec_extractv8qi): Renamed to ...
11466 (vec_extractv8qiqi): ... this.
11467 (vec_initv8qi): Renamed to ...
11468 (vec_initv8qiqi): ... this.
11469 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
11470 (vec_init<mode>): Renamed to ...
11471 (vec_init<mode><VEC_base_l>): ... this.
11472 (vec_extract<mode>): Renamed to ...
11473 (vec_extract<mode><VEC_base_l>): ... this.
11474 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
11475 (vec_initv2sfsf): ... this.
11476 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
11477 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
11478 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
11479 element mode after mode in gen_vec_init* calls.
11480 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
11481 (vec_init<mode><Vel>): ... this.
11482 (vec_extract<mode>): Renamed to ...
11483 (vec_extract<mode><Vel>): ... this.
11484 * config/aarch64/iterators.md (Vel): New mode attribute.
11485 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
11486 Add element mode after mode in gen_vec_extract* calls.
11487 * config/s390/vector.md (non_vec_l): New mode attribute.
11488 (vec_extract<mode>): Renamed to ...
11489 (vec_extract<mode><non_vec_l>): ... this.
11490 (vec_init<mode>): Renamed to ...
11491 (vec_init<mode><non_vec_l>): ... this.
11492 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
11493 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
11495 * config/arm/iterators.md (V_elem_l): New mode attribute.
11496 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
11497 (vec_extract<mode><V_elem_l>): ... this.
11498 (vec_extractv2di): Renamed to ...
11499 (vec_extractv2didi): ... this.
11500 (vec_init<mode>): Renamed to ...
11501 (vec_init<mode><V_elem_l>): ... this.
11502 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
11503 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
11504 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
11505 Add element mode after gen_vec_extract* calls.
11506 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
11507 (vec_init<mode><unitmode>): ... this.
11508 (vec_extract<mode>): Renamed to ...
11509 (vec_extract<mode><unitmode>): ... this.
11510 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
11511 (vec_init<mode><unitmode>): ... this.
11512 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
11513 (vec_initv2sfsf): ... this.
11514 (vec_extractv2sf): Renamed to ...
11515 (vec_extractv2sfsf): ... this.
11516 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
11517 Add element mode after gen_vec_extract* calls.
11518 * config/mips/mips.md (unitmode): New mode iterator.
11519 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
11520 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
11521 * config/spu/spu.md (inner_l): New mode attribute.
11522 (vec_init<mode>): Renamed to ...
11523 (vec_init<mode><inner_l>): ... this.
11524 (vec_extract<mode>): Renamed to ...
11525 (vec_extract<mode><inner_l>): ... this.
11526 * config/sparc/sparc.md (veltmode): New mode iterator.
11527 (vec_init<VMALL:mode>): Renamed to ...
11528 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
11529 * config/ia64/vect.md (vec_initv2si): Renamed to ...
11530 (vec_initv2sisi): ... this.
11531 (vec_initv2sf): Renamed to ...
11532 (vec_initv2sfsf): ... this.
11533 (vec_extractv2sf): Renamed to ...
11534 (vec_extractv2sfsf): ... this.
11535 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
11536 (vec_init<mode>): Renamed to ...
11537 (vec_init<mode><VEC_base_l>): ... this.
11538 (vec_extract<mode>): Renamed to ...
11539 (vec_extract<mode><VEC_base_l>): ... this.
11540 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
11541 (vec_initv2sfsf): ... this.
11542 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
11543 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
11544 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
11545 gen_vec_init* calls.
11547 2017-08-01 Richard Biener <rguenther@suse.de>
11549 PR tree-optimization/81297
11550 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
11551 TREE_OVERFLOW from INTEGER_CSTs.
11553 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
11555 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
11557 2017-07-31 Carl Love <cel@us.ibm.com>
11559 * config/rs6000/rs6000-c: Add support for built-in functions
11560 vector signed char vec_xl_be (signed long long, signed char *);
11561 vector unsigned char vec_xl_be (signed long long, unsigned char *);
11562 vector signed int vec_xl_be (signed long long, signed int *);
11563 vector unsigned int vec_xl_be (signed long long, unsigned int *);
11564 vector signed long long vec_xl_be (signed long long, signed long long *);
11565 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
11566 vector signed short vec_xl_be (signed long long, signed short *);
11567 vector unsigned short vec_xl_be (signed long long, unsigned short *);
11568 vector double vec_xl_be (signed long long, double *);
11569 vector float vec_xl_be (signed long long, float *);
11570 * config/rs6000/altivec.h (vec_xl_be): Add #define.
11571 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
11572 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
11574 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
11575 (altivec_expand_builtin): Add switch statement to call
11576 altivec_expand_xl_be for each builtin.
11577 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
11578 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
11579 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
11580 __builtin_vsx_le_be_v16qi.
11581 * doc/extend.texi: Update the built-in documentation file for the
11582 new built-in functions.
11584 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
11587 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
11589 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
11591 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11593 * config.gcc: Add z14.
11594 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
11595 CPU model numbers for z13s and z14.
11596 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
11598 * config/s390/s390-opts.h (enum processor_type): Rename
11599 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
11600 * config/s390/s390.c (processor_table): Add field for CPU name to
11601 be passed to Binutils.
11602 (s390_asm_output_machine_for_arch): Use the new field in
11603 processor_table for Binutils.
11604 (s390_expand_builtin): Replace arch12 with z14.
11605 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
11606 (s390_get_sched_attrmask): Likewise.
11607 (s390_get_unit_mask): Likewise.
11608 * config/s390/s390.opt: Add z14 to processor_type enum.
11610 2017-07-31 Martin Jambor <mjambor@suse.cz>
11613 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
11614 regardless of optimization level.
11616 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
11617 Martin Liska <mliska@suse.cz>
11619 * predict.def: Remove old comment and adjust probability.
11620 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
11621 PREDICT statements.
11623 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
11626 * config/i386/i386.c (ix86_function_naked): New function.
11627 (ix86_can_use_return_insn_p): Return false for naked functions.
11628 (ix86_expand_prologue): Skip prologue for naked functions.
11629 (ix86_expand_epilogue): Skip epilogue for naked functions
11630 and emit trap instruction.
11631 (ix86_warn_func_return): New function.
11632 (ix86_attribute_table): Add "naked" attribute specification.
11633 (TARGET_WARN_FUNC_RETURN): Define.
11634 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
11636 2017-07-31 Martin Liska <mliska@suse.cz>
11638 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
11639 (dump_gimple_bb_header): Always dump BB info.
11640 (pp_cfg_jump): Do not append info about BB when dumping a jump.
11642 2017-07-31 Martin Liska <mliska@suse.cz>
11645 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
11646 also with current_function_decl non-null equality.
11648 2017-07-31 Jakub Jelinek <jakub@redhat.com>
11651 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
11652 change type to the element type, instead add eltype variable and
11653 use it where we are interested in the element type.
11655 PR tree-optimization/81603
11656 * ipa-polymorphic-call.c
11657 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
11658 offset arithmetic in offset_int, bail out if the resulting bit offset
11659 doesn't fit into shwi.
11661 2017-07-31 Martin Liska <mliska@suse.cz>
11663 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
11664 (gimplify_save_expr): Fix comment.
11666 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
11669 * config/i386/i386.c (ix86_function_arg): Update arguments for
11671 (ix86_compute_frame_layout): Set the initial stack offset to
11672 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
11673 INCOMING_FRAME_SP_OFFSET.
11674 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
11675 stack before exception handler returns.
11676 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
11677 the 'ERROR_CODE' for exception handler.
11679 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
11681 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
11682 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
11683 (ASM_OUTPUT_REG_POP): Ditto.
11684 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
11685 instead of asm_fprintf to output pure string.
11687 2017-07-29 Jakub Jelinek <jakub@redhat.com>
11689 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
11690 to imported_module_or_decl hook.
11691 (debug_nothing_tree_tree_tree_bool): Remove.
11692 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
11693 * debug.c (do_nothing_debug_hooks): Use
11694 debug_nothing_tree_tree_tree_bool_bool instead of
11695 debug_nothing_tree_tree_tree_bool.
11696 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
11697 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
11698 * sdbout.c (sdb_debug_hooks): Likewise.
11699 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
11700 (gen_namespace_die): Add DW_AT_export_symbols attribute if
11702 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
11703 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
11704 attribute, don't add anything.
11706 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11708 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
11709 (fold_build2_stat_loc): Likewise.
11710 (fold_build3_stat_loc): Likewise.
11711 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
11712 (fold_build1_loc): Remove macro.
11713 (fold_build2_loc): Likewise.
11714 (fold_build3_loc): Likewise.
11716 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11718 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
11719 (gimple_build_debug_bind_source_stat): Likewise.
11720 * gimple.h (gimple_build_debug_bind): Remove macro.
11721 (gimple_build_debug_bind_source): Likewise.
11723 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11725 * bitmap.c (bitmap_alloc): Adjust.
11726 (bitmap_gc_alloc): Likewise.
11727 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
11729 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11731 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
11732 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
11733 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
11734 (bitmap_gc_alloc_stat): Likewise.
11735 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
11737 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11739 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
11740 * rtl.h (shallow_copy_rtx): Remove macro.
11742 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11744 * emit-rtl.c (gen_raw_REG): Adjust.
11745 * gengenrtl.c (gendef): Likewise.
11746 * rtl.c (rtx_alloc_stat): Remove _stat from name.
11747 * rtl.h (rtx_alloc): Remove macro.
11749 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11751 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
11752 (build_tree_list_stat): Likewise.
11753 * tree.h (build_tree_list): Remove macro.
11754 (build_tree_list_vec): Likewise.
11756 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11758 * tree.c (make_vector_stat): Remove _stat from name.
11759 (build_vector_stat): Likewise.
11760 * tree.h (make_vector_stat): Remove macro.
11761 (build_vector_stat): Likewise.
11763 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11765 * tree.h (build_var_debug_value): Remove prototype.
11767 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11769 * tree.c (tree_cons_stat): Remove _stat from name.
11770 * tree.h (tree_cons): Remove macro.
11772 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11774 * tree.c (build_vl_exp_stat): Remove _stat from name.
11775 * tree.h (build_vl_exp): Remove macro.
11777 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11779 * tree.c (build_decl_stat): Remove _stat from name.
11780 * tree.h (build_decl): Remove macro.
11782 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11784 * gimple.c (gimple_build_with_ops_stat): Adjust.
11785 (gimple_alloc_stat): Remove _stat from name.
11786 * gimple.h (gimple_alloc): Remove macro.
11788 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11790 * tree.c (make_tree_vec_stat): Remove _stat from name.
11791 (grow_tree_vec_stat): Likewise.
11792 * tree.h (make_tree_vec_stat): Adjust prototype.
11793 (grow_tree_vec_stat): Likewise.
11794 (make_tree_vec): Remove macro.
11795 (grow_tree_vec): Likewise.
11797 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11799 * fold-const.c (fold_build1_stat_loc): Adjust.
11800 (fold_build2_stat_loc): Likewise.
11801 (fold_build3_stat_loc): Likewise.
11802 * tree.c (build0_stat): Remove _stat from name.
11803 (build1_stat): Likewise.
11804 (build2_stat): Likewise.
11805 (build3_stat): Likewise.
11806 (build4_stat): Likewise.
11807 (build5_stat): Likewise.
11808 * tree.h (build1_loc): Remove macro, and rename _stat function
11810 (build2_loc): Likewise.
11811 (build3_loc): Likewise.
11812 (build4_loc): Likewise.
11813 (build5_loc): Likewise.
11815 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11817 * tree.c (make_int_cst_stat): Remove _stat from name.
11818 * tree.h (make_int_cst_stat): Adjust prototype.
11819 (make_int_cst): Remove macro.
11821 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11823 * tree.c (make_tre_binfo_stat): Remove _stat from name.
11824 * tree.h (make_tree_binfo_stat): Adjust prototype.
11825 (make_tree_binfo): Remove.
11827 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11829 * tree.c (copy_node_stat): Rename to copy_node.
11830 (build_distinct_type_copy): Adjust.
11831 * tree.h (copy_node_stat): Adjust prototype.
11832 (copy_node): Remove macro.
11834 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
11836 * tree.c (make_node_stat): rename to make_node.
11837 (build_tree_list_stat): Adjust.
11838 (build0_stat): Likewise.
11839 (build2_stat): Likewise.
11840 (build3_stat): Likewise.
11841 (build4_stat): Likewise.
11842 (build5_stat): Likewise.
11843 (build_decl_stat): Likewise.
11844 * tree.h (make_node_stat): Adjust prototype.
11845 (make_node): remove macro.
11847 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
11849 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
11850 (PPC_FEATURE2_SCV): Likewise.
11851 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
11853 2017-07-28 Tamar Christina <tamar.christina@arm.com>
11855 * config/aarch64/aarch64.c
11856 (aarch64_internal_mov_immediate): Add new special pattern.
11857 * config/aarch64/aarch64.md (*movdi_aarch64):
11858 Add reg/32bit const mov case.
11860 2017-07-28 Tamar Christina <tamar.christina@arm.com>
11861 Richard Sandiford <richard.sandiford@linaro.org>
11863 * config/aarch64/aarch64.md (mov<mode>): Generalize.
11864 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
11865 Add integer and movi cases.
11866 (movi-split-hf-df-sf split, fp16): New.
11867 (enabled): Added TARGET_FP_F16INST.
11868 * config/aarch64/iterators.md (GPF_HF): New.
11869 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
11871 2017-07-28 Tamar Christina <tamar.christina@arm.com>
11873 * config/aarch64/aarch64.c
11874 (aarch64_simd_container_mode): Add prototype.
11875 (aarch64_expand_mov_immediate): Add HI support.
11876 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
11877 (aarch64_can_const_movi_rtx_p): New.
11878 (aarch64_preferred_reload_class):
11879 Remove restrictions of using FP registers for certain SIMD operations.
11880 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
11881 (aarch64_valid_floating_const): Add integer move validation.
11882 (aarch64_simd_imm_scalar_p): Remove.
11883 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
11884 (aarch64_legitimate_constant_p): Expand list of supported cases.
11885 * config/aarch64/aarch64-protos.h
11886 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
11887 (aarch64_reinterpret_float_as_int): New.
11888 (aarch64_simd_imm_scalar_p): Remove.
11889 * config/aarch64/constraints.md (Uvi): New.
11890 (Dd): Split into Ds and new Dd.
11891 * config/aarch64/aarch64.md (*movsi_aarch64):
11893 (*movdi_aarch64): Add SIMD mov case.
11895 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11897 * tree-predcom.c: (struct chain): Handle store-store chain in which
11898 stores for elimination only store loop invariant values.
11899 (execute_pred_commoning_chain): Ditto.
11900 (prepare_initializers_chain_store_elim): Ditto.
11901 (prepare_finalizers): Ditto.
11902 (is_inv_store_elimination_chain): New function.
11903 (initialize_root_vars_store_elim_1): New function.
11905 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11907 * tree-predcom.c: Revise general description of the pass.
11908 (enum chain_type): New enum type for store elimination.
11909 (struct chain): New field supporting store elimination.
11910 (struct component): Ditto.
11911 (dump_chain): Dump store-stores chain.
11912 (release_chain): Release resources.
11913 (split_data_refs_to_components): Compute and create component
11914 contains only stores for elimination.
11915 (get_chain_last_ref_at): New function.
11916 (make_invariant_chain): Initialization.
11917 (make_rooted_chain): Specify chain type in parameter and record it.
11918 (add_looparound_copies): Skip for store-stores chain.
11919 (determine_roots_comp): Compute type of chain and pass it to
11921 (initialize_root_vars_store_elim_2): New function.
11922 (finalize_eliminated_stores): New function.
11923 (remove_stmt): Handle store for elimination.
11924 (execute_pred_commoning_chain): Execute predictive commoning on
11925 store-store chains.
11926 (determine_unroll_factor): Skip unroll for store-stores chain.
11927 (prepare_initializers_chain_store_elim): New function.
11928 (prepare_initializers_chain): Hanlde store-store chain.
11929 (prepare_finalizers_chain, prepare_finalizers): New function.
11930 (tree_predictive_commoning_loop): Return integer value indicating
11931 if loop is unrolled or lcssa form is corrupted.
11932 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
11934 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11936 * tree-predcom.c (initialize_root): Delete.
11937 (execute_pred_commoning_chain): Initialize root vars and replace
11938 reference of non-combined chain directly, rather than call above
11941 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11943 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
11944 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
11946 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11948 * tree-predcom.c (struct chain): New field init_seq.
11949 (release_chain): Release init_seq.
11950 (prepare_initializers_chain): Record intialization stmts in above
11952 (insert_init_seqs): New function.
11953 (tree_predictive_commoning_loop): Call insert_init_seqs.
11955 2017-07-28 Bin Cheng <bin.cheng@arm.com>
11957 * tree-predcom.c (determine_roots_comp): Skip trivial components.
11959 2017-07-28 Richard Biener <rguenther@suse.de>
11961 * match.pd: Remove superfluous :c.
11962 * genmatch.c (simplify::id): Add member.
11963 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
11965 (current_id): New global.
11966 (dt_node::parent): Move from ...
11967 (dt_operand::parent): ... here. Add for_id member.
11968 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
11969 (decision_tree::find_node): Relax order requirement when
11970 merging DT_TRUE nodes to ones inbetween the current simplify
11971 and the one we try to merge with. Add diagnostic whenever
11972 we need to enforce pattern order by not merging.
11973 (decision_tree::insert): Set current_id.
11974 (decision_tree::print_node): Dump parent node and for_id.
11975 (parser::last_id): Add member.
11976 (parser::push_simplify): Assign unique id.
11977 (parser::parser): Initialize last_id.
11979 2017-07-28 Martin Liska <mliska@suse.cz>
11982 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
11983 gimple_build_debug_bind.
11985 2017-07-28 Richard Biener <rguenther@suse.de>
11987 PR tree-optimization/81502
11988 * match.pd: Add pattern combining BIT_INSERT_EXPR with
11990 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
11992 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
11993 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
11994 for BIT_FIELD_REF args.
11995 * fold-const.c (make_bit_field_ref): Likewise.
11996 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
11998 2017-07-28 Jakub Jelinek <jakub@redhat.com>
12001 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
12002 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
12003 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
12004 Or it into SANITIZER_UNDEFINED.
12005 * ubsan.c: Include gimple-fold.h and varasm.h.
12006 (ubsan_expand_ptr_ifn): New function.
12007 (instrument_pointer_overflow): New function.
12008 (maybe_instrument_pointer_overflow): New function.
12009 (instrument_object_size): Formatting fix.
12010 (pass_ubsan::execute): Call instrument_pointer_overflow
12011 and maybe_instrument_pointer_overflow.
12012 * internal-fn.c (expand_UBSAN_PTR): New function.
12013 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
12014 * sanitizer.def (__ubsan_handle_pointer_overflow,
12015 __ubsan_handle_pointer_overflow_abort): New builtins.
12016 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
12017 * internal-fn.def (UBSAN_PTR): New internal function.
12018 * opts.c (sanitizer_opts): Add pointer-overflow.
12019 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
12020 * fold-const.c (build_range_check): Compute pointer range check in
12021 integral type if pointer arithmetics would be needed. Formatting
12024 2017-07-28 Martin Liska <mliska@suse.cz>
12027 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
12028 parameters that are of a variable-length.
12030 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
12032 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
12034 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
12035 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
12036 (CRT_CALL_STATIC_FUNCTION): Likewise.
12037 (ASM_DEFAULT_SPEC): New define.
12038 (ASM_SPEC32): Likewise.
12039 (ASM_SPEC64): Likewise.
12040 (ASM_SPEC_COMMON): Likewise.
12041 (ASM_SPEC): Likewise.
12042 (INVALID_64BIT): Likewise.
12043 (LINK_OS_DEFAULT_SPEC): Likewise.
12044 (LINK_OS_SPEC32): Likewise.
12045 (LINK_OS_SPEC64): Likewise.
12046 (POWERPC_LINUX): Likewise.
12047 (PTRDIFF_TYPE): Likewise.
12048 (RESTORE_FP_PREFIX): Likewise.
12049 (RESTORE_FP_SUFFIX): Likewise.
12050 (SAVE_FP_PREFIX): Likewise.
12051 (SAVE_FP_SUFFIX): Likewise.
12052 (SIZE_TYPE): Likewise.
12053 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
12054 (TARGET_64BIT): Likewise.
12055 (TARGET_64BIT): Likewise.
12056 (TARGET_AIX): Likewise.
12057 (WCHAR_TYPE_SIZE): Likewise.
12058 (WCHAR_TYPE): Undefine.
12059 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
12060 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
12061 (CPP_OS_RTEMS_SPEC): Delete.
12062 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
12063 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
12065 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
12067 2017-07-28 Jakub Jelinek <jakub@redhat.com>
12069 PR tree-optimization/81578
12070 * tree-parloops.c (build_new_reduction): Bail out if
12071 reduction_code isn't one of the standard OpenMP reductions.
12072 Move the details printing after that decision.
12074 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
12076 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
12077 related to reload_in_progress.
12078 (splat_input_operand): Likewise.
12079 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
12081 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
12083 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
12084 (TARGET_INSTANTIATE_DECLS): Likewise.
12085 (legitimate_indexed_address_p): Delete reload_in_progress code.
12086 (rs6000_debug_legitimate_address_p): Likewise.
12087 (rs6000_eliminate_indexed_memrefs): Likewise.
12088 (rs6000_emit_le_vsx_store): Likewise.
12089 (rs6000_emit_move_si_sf_subreg): Likewise.
12090 (rs6000_emit_move): Likewise.
12091 (register_to_reg_type): Likewise.
12092 (rs6000_pre_atomic_barrier): Likewise.
12093 (rs6000_machopic_legitimize_pic_address): Likewise.
12094 (rs6000_allocate_stack_temp): Likewise.
12095 (rs6000_address_for_fpconvert): Likewise.
12096 (rs6000_address_for_altivec): Likewise.
12097 (rs6000_secondary_memory_needed_rtx): Delete function.
12098 (rs6000_check_sdmode): Likewise.
12099 (rs6000_alloc_sdmode_stack_slot): Likewise.
12100 (rs6000_instantiate_decls): Likewise.
12101 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
12102 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
12103 Delete reload_in_progress.
12104 (*vec_reload_and_plus_<mptrsize>): Likewise.
12105 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
12106 (vsx_div_v2di): Likewise.
12107 (vsx_udiv_v2di): Likewise.
12109 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
12111 * config/rs6000/rs6000.opt (mlra): Replace with stub.
12112 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
12113 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
12114 (rs6000_debug_reg_global): Delete print of LRA status.
12115 (rs6000_option_override_internal): Delete dead LRA related code.
12116 (rs6000_lra_p): Delete function.
12117 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
12119 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
12121 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
12122 * config/riscv/rtems.h: New file.
12124 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12125 Sudakshina Das <sudi.das@arm.com>
12127 * config/aarch64/aarch64.md
12128 (define_split for and<mode>3nr_compare): Move
12129 non aarch64_logical_operand to a register.
12130 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
12131 register immediate operand to a register.
12132 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
12134 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
12136 PR middle-end/81564
12137 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
12139 2017-07-27 Richard Biener <rguenther@suse.de>
12141 PR tree-optimization/81573
12142 PR tree-optimization/81494
12143 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
12144 multi defuse cycle case.
12146 2017-07-27 Richard Biener <rguenther@suse.de>
12148 PR tree-optimization/81571
12149 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
12152 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
12154 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
12155 earlier and only if MASK_FPU is set. Adjust formatting.
12157 2017-07-27 Martin Liska <mliska@suse.cz>
12159 * opt-functions.awk: Add validation of value of Init.
12160 * optc-gen.awk: Pass new argument.
12162 2017-07-27 Martin Liska <mliska@suse.cz>
12164 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
12165 Fix wrong condition.
12167 2017-07-27 Martin Liska <mliska@suse.cz>
12169 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
12170 BBs and edges seen by autoFDO.
12172 2017-07-27 Richard Biener <rguenther@suse.de>
12174 PR tree-optimization/81502
12175 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
12176 with incompatible but same sized type.
12177 (execute_update_addresses_taken): Likewise.
12179 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
12181 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
12182 flag_tree_loop_vectorize rather than flag_tree_vectorize.
12184 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12187 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
12188 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
12189 Change s_operand to memory_operand.
12191 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
12193 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
12194 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
12195 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
12196 Emit instructions rather than returning an expression. Handle TFmode
12197 and KFmode by casting to TImode.
12198 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
12199 (rs6000_emit_le_vsx_store): Likewise.
12200 * config/rs6000/vsx.md (VSX_TI): New iterator.
12201 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
12202 (*vsx_le_undo_permute_<mode>): Likewise.
12203 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
12204 emit the split sequence.
12205 (*vsx_le_perm_store_<mode>): Likewise.
12207 2017-07-27 Jakub Jelinek <jakub@redhat.com>
12209 PR tree-optimization/81555
12210 PR tree-optimization/81556
12211 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
12212 if true, force CHANGED for the recursive invocation.
12213 (reassociate_bb): Remember original length of ops array, pass
12214 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
12216 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
12217 attributes for noipa attribute. For naked attribute use
12218 lookup_attribute first before lookup_attribute_spec.
12219 * final.c (rest_of_handle_final): Disable IPA RA for functions with
12221 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
12222 for functions with noipa attribute.
12223 (cgraph_externally_visible_p): Return true for functions with noipa
12225 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
12226 for functions with noipa attribute.
12227 * doc/extend.texi: Document noipa function attribute.
12228 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
12229 also for functions with noipa attribute.
12230 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
12232 2017-07-26 Andrew Pinski <apinski@cavium.com>
12234 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
12235 vec_unalign_load_cost and vec_unalign_store_cost.
12237 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
12239 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
12240 -mvsx-small-integer option.
12241 (ISA_3_0_MASKS_IEEE): Likewise.
12242 (OTHER_VSX_VECTOR_MASKS): Likewise.
12243 (POWERPC_MASKS): Likewise.
12244 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
12245 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
12246 code, only testing for DImode being allowed in non-VSX floating
12248 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
12249 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
12251 (rs6000_option_override_internal): Delete -mvsx-small-integer.
12252 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
12253 TARGET_P8_VECTOR test.
12254 (rs6000_secondary_reload_simple_move): Likewise.
12255 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
12256 since TARGET_P9_VECTOR was already tested.
12257 (rs6000_opt_masks): Remove -mvsx-small-integer.
12258 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
12259 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
12261 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
12262 test for TARGET_VEXTRACTUB was used, and that uses
12264 (p9 extract splitter): Likewise.
12265 (vsx_extract_<mode>_di_p9): Likewise.
12266 (vsx_extract_<mode>_store_p9): Likewise.
12267 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
12268 for TARGET_P9_VECTOR was used. Delete code that is now dead with
12269 the elimination of TARGET_VSX_SMALL_INTEGER.
12270 (vsx_extract_<mode>_p8): Likewise.
12271 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
12272 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
12273 (vsx_set_<mode>_p9): Likewise.
12274 (vsx_set_v4sf_p9): Likewise.
12275 (vsx_set_v4sf_p9_zero): Likewise.
12276 (vsx_insert_extract_v4sf_p9): Likewise.
12277 (vsx_insert_extract_v4sf_p9_2): Likewise.
12278 * config/rs6000/rs6000.md (sign extend splitter): Change
12279 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
12280 (floatsi<mode>2_lfiwax_mem): Likewise.
12281 (floatunssi<mode>2_lfiwzx_mem): Likewise.
12282 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
12283 since a test for TARGET_P9_VECTOR was used.
12284 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
12285 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
12286 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
12287 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
12288 TARGET_P8_VECTOR test.
12289 (fix_trunc<mode>si2_stfiwx): Likewise.
12290 (fix_trunc<mode>si2_internal): Likewise.
12291 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
12292 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
12294 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
12295 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
12296 TARGET_P8_VECTOR test.
12297 (fixuns_trunc<mode>si2_stfiwx): Likewise.
12298 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
12299 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
12301 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
12302 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
12303 since a test for TARGET_P9_VECTOR was used.
12304 (splitter for loading small constants): Likewise.
12306 2017-07-26 Andrew Pinski <apinski@cavium.com>
12308 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
12311 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
12314 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
12315 (fp_valid_at): Likewise.
12317 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
12319 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
12320 (qdf24xx_addrcost_table): Likewise.
12321 (cortexa57_tunings): Update to use generic_branch_cost.
12322 (cortexa72_tunings): Likewise.
12323 (cortexa73_tunings): Likewise.
12324 (qdf24xx_tunings): Likewise.
12326 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
12328 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
12329 (thunderx2t99_branch_cost): Likewise.
12330 (cortexa35_tunings): Update to use generic_branch_cost.
12331 (cortexa53_tunings): Likewise.
12332 (cortexa57_tunings): Likewise.
12333 (cortexa72_tunings): Likewise.
12334 (cortexa73_tunings): Likewise.
12335 (thunderx2t99_tunings): Likewise.
12337 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
12339 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
12340 (sparc_option_override): Honour MASK_FSMULD.
12341 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
12342 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
12343 * config/sparc/sparc.opt (mfsmuld): New option.
12344 * doc/invoke.texi (mfsmuld): Document option.
12346 2017-07-26 Marek Polacek <polacek@redhat.com>
12348 PR middle-end/70992
12349 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
12351 2017-07-26 Richard Biener <rguenther@suse.de>
12353 * gimple-match-head.c (do_valueize): Return OP if valueize
12355 (get_def): New helper to get at the def stmt of a SSA name
12356 if valueize allows.
12357 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
12358 do_valueize to get at the def stmt.
12359 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
12361 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
12363 PR middle-end/46932
12364 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
12366 2017-07-26 Martin Liska <mliska@suse.cz>
12369 * function.c (expand_function_start): Make expansion of
12370 nonlocal_goto_save_area after parm_birth_insn.
12372 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
12374 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
12375 from all CPU target flags enable members.
12377 2017-07-26 Richard Biener <rguenther@suse.de>
12379 * genmatch.c (dt_simplify::gen): Make iterator vars const.
12380 (decision_tree::gen): Make 'type' const.
12381 (write_predicate): Likewise.
12383 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
12385 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
12386 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
12387 (rs6000_option_override_internal): Likewise.
12388 (rs6000_expand_vector_set): Likewise.
12389 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
12390 (TARGET_UPPER_REGS_SF): Likewise.
12391 (TARGET_UPPER_REGS_DI): Likewise.
12392 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
12393 (TARGET_DIRECT_MOVE_64BIT): Likewise.
12394 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
12395 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
12396 (Splitters for DI constants in Altivec registers): Likewise.
12397 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
12398 (vsx_set_v4sf_p9): Likewise.
12399 (vsx_set_v4sf_p9_zero): Likewise.
12400 (vsx_insert_extract_v4sf_p9): Likewise.
12401 (vsx_insert_extract_v4sf_p9_2): Likewise.
12403 2017-07-25 Carl Love <cel@us.ibm.com>
12405 * doc/extend.texi: Update the built-in documentation file for the
12406 existing built-in functions
12407 vector signed char vec_cnttz (vector signed char);
12408 vector unsigned char vec_cnttz (vector unsigned char);
12409 vector signed short vec_cnttz (vector signed short);
12410 vector unsigned short vec_cnttz (vector unsigned short);
12411 vector signed int vec_cnttz (vector signed int);
12412 vector unsigned int vec_cnttz (vector unsigned int);
12413 vector signed long long vec_cnttz (vector signed long long);
12414 vector unsigned long long vec_cnttz (vector unsigned long long);
12416 2017-07-25 Andrew Pinski <apinski@cavium.com>
12418 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
12419 accesses where the use is for the first operand of a BIT_INSERT.
12421 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
12424 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
12425 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
12427 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
12429 * config/i386/gstabs.h: Delete.
12430 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
12432 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
12434 * config/i386/i386.c (ix86_decompose_address): Do not check for
12435 register RTX when looking at index_reg or base_reg.
12436 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
12438 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
12440 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
12441 to update EH info here.
12443 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
12445 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
12447 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
12449 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
12451 2017-07-25 Torsten Duwe <duwe@suse.de>
12453 * common.opt: Introduce -fpatchable-function-entry
12454 command line option, and its variables function_entry_patch_area_size
12455 and function_entry_patch_area_start.
12456 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
12457 including a two-value parser.
12458 * target.def (print_patchable_function_entry): New target hook.
12459 * targhooks.h (default_print_patchable_function_entry): New function.
12460 * targhooks.c (default_print_patchable_function_entry): Likewise.
12461 * toplev.c (process_options): Switch off IPA-RA if
12462 patchable function entries are being generated.
12463 * varasm.c (assemble_start_function): Look at the
12464 patchable-function-entry command line switch and current
12465 function attributes and maybe generate NOP instructions by
12466 calling the print_patchable_function_entry hook.
12467 * doc/extend.texi: Document patchable_function_entry attribute.
12468 * doc/invoke.texi: Document -fpatchable_function_entry
12469 command line option.
12470 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
12472 * doc/tm.texi: Re-generate.
12474 2017-07-25 Jakub Jelinek <jakub@redhat.com>
12477 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
12478 TARGET_AVX512DQ rather than TARGET_AVX512BW.
12480 2017-07-25 Tamar Christina <tamar.christina@arm.com>
12482 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
12484 2017-07-25 Richard Biener <rguenther@suse.de>
12486 PR tree-optimization/81455
12487 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
12488 not walk in cycles when looking for guards.
12490 2017-07-25 Richard Biener <rguenther@suse.de>
12492 PR tree-optimization/81529
12493 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
12494 when optimizing backedge uses.
12496 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
12498 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
12500 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
12501 to dl_section_ref. On AIX, append an expression to subtract
12502 the size of the section length to dl_section_ref.
12504 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
12506 * configure.ac: If any of the config.* scripts fail, exit 1.
12507 * configure: Regenerate.
12509 2017-07-25 Richard Biener <rguenther@suse.de>
12511 PR middle-end/81546
12512 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
12513 of immediate uses, be more verbose on errors.
12515 2017-07-25 Richard Biener <rguenther@suse.de>
12517 PR tree-optimization/81510
12518 * tree-vect-loop.c (vect_is_simple_reduction): When the
12519 reduction stmt is not inside the loop bail out.
12521 2017-07-25 Richard Biener <rguenther@suse.de>
12523 PR tree-optimization/81303
12524 * tree-vect-loop-manip.c (vect_loop_versioning): Build
12525 profitability check against LOOP_VINFO_NITERSM1.
12527 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
12529 * domwalk.c (cmp_bb_postorder): Simplify.
12530 (sort_bbs_postorder): New function. Use it...
12531 (dom_walker::walk): ...here to optimize common cases.
12533 2017-07-25 Martin Liska <mliska@suse.cz>
12536 * ipa-visibility.c (function_and_variable_visibility): Make the
12537 redirection just on target that supports aliasing.
12538 Fix GNU coding style.
12540 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
12543 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
12545 * config/arm/rtems.h: Likewise.
12546 * config/bfin/rtems.h: Likewise.
12547 * config/i386/rtemself.h: Likewise.
12548 * config/lm32/rtems.h: Likewise.
12549 * config/m32c/rtems.h: Likewise.
12550 * config/m68k/rtemself.h: Likewise.
12551 * config/microblaze/rtems.h: Likewise.
12552 * config/mips/rtems.h: Likewise.
12553 * config/moxie/rtems.h: Likewise.
12554 * config/nios2/rtems.h: Likewise.
12555 * config/powerpcspe/rtems.h: Likewise.
12556 * config/rs6000/rtems.h: Likewise.
12557 * config/rtems.h: Likewise.
12558 * config/sh/rtems.h: Likewise.
12559 * config/sh/rtemself.h: Likewise.
12560 * config/sparc/rtemself.h: Likewise.
12562 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
12565 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
12566 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
12567 * tree-ssa-structalias.c (alias_get_name): Same.
12569 2017-07-25 Bin Cheng <bin.cheng@arm.com>
12572 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
12573 instructions if no du chain is found.
12575 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
12577 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
12579 2017-07-25 Richard Biener <rguenther@suse.de>
12581 PR middle-end/81505
12582 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
12585 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
12587 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
12588 upper-regs options.
12589 (ISA_2_7_MASKS_SERVER): Likewise.
12590 (ISA_3_0_MASKS_IEEE): Likewise.
12591 (OTHER_P8_VECTOR_MASKS): Likewise.
12592 (OTHER_VSX_VECTOR_MASKS): Likewise.
12593 (POWERPC_MASKS): Likewise.
12594 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
12595 duplicate list of options.
12596 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
12597 explicit -mupper-regs options.
12598 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
12599 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
12600 alias for -mupper-regs-df.
12601 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
12602 (rs6000_init_hard_regno_mode_ok): Likewise.
12603 (rs6000_option_override_internal): Likewise.
12604 (rs6000_opt_masks): Likewise.
12605 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
12606 options in terms of whether -mvsx or -mpower8-vector was used.
12607 (TARGET_UPPER_REGS_DI): Likewise.
12608 (TARGET_UPPER_REGS_SF): Likewise.
12609 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
12610 -mupper-regs-* options.
12612 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
12614 * passes.c (emergency_dump_function): Print some empty lines and a
12615 header before the RTL dump.
12617 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
12619 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
12621 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
12624 * config/aarch64/aarch64.c (aarch64_classify_symbol):
12625 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
12627 2017-07-24 Carl Love <cel@us.ibm.com>
12629 * config/rs6000/rs6000-c.c: Add support for built-in functions
12630 vector float vec_extract_fp32_from_shorth (vector unsigned short);
12631 vector float vec_extract_fp32_from_shortl (vector unsigned short);
12632 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
12633 vec_extract_fp_from_shortl): Add defines for the two builtins.
12634 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
12635 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
12637 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
12638 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
12639 * doc/extend.texi: Update the built-in documentation file for the
12640 new built-in function.
12642 2017-07-24 Jakub Jelinek <jakub@redhat.com>
12645 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
12647 * doc/generic.texi: Likewise.
12648 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
12649 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
12651 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
12653 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
12654 (aarch64_mls_elt_merge<mode>): Likewise.
12656 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
12658 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
12659 having __cxa_atexit.
12661 2017-07-23 Michael Collison <michael.collison@arm.com>
12663 * config/arm/arm.c (arm_option_override): Deprecate
12664 use of -mstructure-size-boundary.
12665 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
12666 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
12668 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12671 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
12672 Reduce cost estimate for direct moves.
12674 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
12677 * config/i386/i386.c (ix86_option_override_internal): Disable
12678 BMI, BMI2 and TBM instructions for -m16.
12680 2017-07-21 Carl Love <cel@us.ibm.com>
12682 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12683 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
12684 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
12685 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
12686 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
12687 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
12688 VMULOSW): New enum "unspec" values.
12689 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
12690 altivec_vmulosw): New patterns.
12691 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
12692 VMULOSW): Add definitions.
12694 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
12696 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
12697 (qdf24xx): Likewise.
12698 * config/aarch64/aarch64-options-extensions.def (rdma); New.
12699 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
12700 (AARCH64_FL_V8_1): Renumber.
12701 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
12702 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
12703 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
12704 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
12705 rdma to feature modifiers list.
12707 2017-07-21 Yury Gribov <tetra2005@gmail.com>
12709 PR middle-end/56727
12710 * ipa-visibility (function_and_variable_visibility): Convert
12711 recursive PLT call to direct call if appropriate.
12713 2017-07-21 Andrew Pinski <apinski@cavium.com>
12715 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
12716 operand 1 to see if the types precision matches.
12717 * fold-const.c (operand_equal_p): Likewise.
12719 2017-07-21 Richard Biener <rguenther@suse.de>
12721 PR tree-optimization/81303
12722 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
12723 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
12724 (vect_peeling_hash_get_lowest_cost): Adjust.
12725 (vect_enhance_data_refs_alignment): Likewise. Use
12726 vect_get_peeling_costs_all_drs to compute the penalty for no
12727 peeling to match up costs.
12729 2017-07-21 Richard Biener <rguenther@suse.de>
12731 PR tree-optimization/81500
12732 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
12733 we didn't identify a reduction path.
12735 2017-07-21 Tom de Vries <tom@codesourcery.com>
12736 Cesar Philippidis <cesar@codesourcery.com>
12738 PR gcov-profile/81442
12739 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
12742 2017-07-21 Tom de Vries <tom@codesourcery.com>
12745 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
12747 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
12748 nvptx_override_options_after_change.
12750 2017-07-21 Ulrich Drepper <drepper@redhat.com>
12752 * dwarf2out.c (output_file_names): Avoid double testing for
12753 dwarf_version >= 5.
12755 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
12757 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
12759 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
12761 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
12763 (try_crossjump_to_edge): Do not punt on partitioned functions.
12765 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
12767 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
12768 Put all BBs reachable only via paths crossing cold region to cold
12770 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
12772 2016-07-21 Richard Biener <rguenther@suse.de>
12774 PR tree-optimization/81303
12775 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
12776 into account prologue and epilogue iterations when raising
12777 min_profitable_iters to sth at least covering one vector iteration.
12779 2017-07-21 Tamar Christina <tamar.christina@arm.com>
12781 * config/arm/arm.c (arm_test_cpu_arch_dat):
12784 2017-07-20 Nathan Sidwell <nathan@acm.org>
12786 Remove TYPE_METHODS.
12787 * tree.h (TYPE_METHODS): Delete.
12788 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
12789 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
12790 (dbxout_type_methods): Scan TYPE_FIELDS.
12791 (dbxout_type): Don't check TYPE_METHODS here.
12792 * function.c (use_register_for_decl): Always ignore register for
12793 class types when not optimizing.
12794 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
12795 * tree.c (free_lang_data_in_type): Stitch out member functions and
12796 templates from TYPE_FIELDS.
12797 (build_distinct_type_copy, verify_type_variant,
12798 verify_type): Member fns are on TYPE_FIELDS.
12799 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
12800 * tree-pretty-print.c (dump_generic_node): Likewise.
12802 2017-07-20 Jakub Jelinek <jakub@redhat.com>
12805 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
12806 V2TImode and V4TImode.
12807 (ix86_expand_vector_extract): Likewise.
12808 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
12809 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
12810 (ssescalarmode): Handle V4TImode and V2TImode.
12811 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
12812 (*vec_extractv2ti, *vec_extractv4ti): New insns.
12813 (VEXTRACTI128_MODE): New mode iterator.
12814 (splitter for *vec_extractv?ti first element): New.
12815 (VEC_INIT_MODE): New mode iterator.
12816 (vec_init<mode>): Consolidate 3 expanders into one using
12817 VEC_INIT_MODE mode iterator.
12819 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
12821 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
12822 non_spilled_static_chain_regno_p.
12824 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
12826 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
12828 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
12830 * bb-reorder.c (connect_traces): Allow copying of blocks within
12833 2017-07-20 Richard Biener <rguenther@suse.de>
12835 * gimple.h (gimple_phi_result): Add gphi * overload.
12836 (gimple_phi_result_ptr): Likewise.
12837 (gimple_phi_arg): Likewise. Adjust index assert to only
12838 allow actual argument accesses rather than all slots available
12840 (gimple_phi_arg_def): Add gphi * overload.
12841 * tree-phinodes.c (make_phi_node): Initialize only actual
12843 (resize_phi_node): Clear memory not covered by old node,
12844 do not initialize excess argument slots.
12845 (reserve_phi_args_for_new_edge): Initialize new argument slot
12848 2017-07-20 Bin Cheng <bin.cheng@arm.com>
12850 PR tree-optimization/81388
12852 2016-07-21 Bin Cheng <bin.cheng@arm.com>
12854 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
12855 by removing computation of may_be_zero.
12857 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
12858 Tom de Vries <tom@codesourcery.com>
12860 PR middle-end/81030
12861 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
12862 when gimple level profile disagrees with what RTL expander did.
12864 2017-07-20 Richard Biener <rguenther@suse.de>
12866 PR tree-optimization/61171
12867 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
12868 (vect_analyze_stmt): Add slp instance parameter.
12869 (vectorizable_reduction): Likewise.
12870 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
12871 (vect_is_simple_reduction): Deal with chains not detected
12872 as SLP reduction chain, specifically not properly associated
12873 chains containing a mix of plus/minus.
12874 (get_reduction_op): Remove.
12875 (get_initial_defs_for_reduction): Simplify, pass in whether
12876 this is a reduction chain, pass in the SLP node for the PHIs.
12877 (vect_create_epilog_for_reduction): Get the SLP instance as
12879 (vectorizable_reduction): Get the SLP instance as arg.
12880 During analysis remember the SLP node with the PHIs in the
12881 instance. Simplify getting at the vectorized reduction PHIs.
12882 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
12883 through SLP instance.
12884 (vect_slp_analyze_operations): Likewise.
12885 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
12886 (vect_transform_stmt): Likewise.
12888 2017-07-20 Tom de Vries <tom@codesourcery.com>
12890 PR tree-optimization/81489
12891 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
12892 read of phi arg location to before loop that modifies phi.
12894 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
12896 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
12899 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
12901 PR middle-end/81331
12902 * except.c (execute): Fix ordering issue.
12904 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
12906 PR rtl-optimization/81423
12907 * combine.c (make_compound_operation_int): Don't try to optimize
12908 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
12910 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
12912 PR rtl-optimization/81423
12913 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
12914 with a constant that is -1 in the truncated to mode.
12916 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
12918 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
12919 (determine_unlikely_bbs): ... here.
12920 * predict.h (propagate_unlikely_bbs_forward): Declare.
12921 * cfgexpand.c (pass_expand::execute): Use it.
12922 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
12924 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
12925 propagate_unlikely_bbs_forward.
12927 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
12929 PR middle-end/81331
12930 * except.c (maybe_add_nop_after_section_switch): New function.
12933 2017-07-19 Tom de Vries <tom@codesourcery.com>
12935 * gimple.h (gimple_phi_set_arg): Make assert more strict.
12937 2017-07-19 Tom de Vries <tom@codesourcery.com>
12939 * gimple.h (gimple_phi_arg): Make assert more strict.
12941 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
12943 * config.gcc (powerpc*-*-*): Add mmintrin.h.
12944 * config/rs6000/mmintrin.h: New file.
12945 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
12947 2017-07-19 Jakub Jelinek <jakub@redhat.com>
12949 PR tree-optimization/81346
12950 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
12952 2017-07-19 Tom de Vries <tom@codesourcery.com>
12954 * config/nvptx/nvptx.md (VECIM): Add V2DI.
12956 2017-07-19 Tom de Vries <tom@codesourcery.com>
12958 * config/nvptx/nvptx-modes.def: Add V2DImode.
12959 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
12960 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
12961 (nvptx_output_mov_insn): Handle lack of mov.b128.
12962 (nvptx_print_operand): Handle 'H' and 'L' codes.
12963 (nvptx_vector_mode_supported): Allow V2DImode.
12964 (nvptx_preferred_simd_mode): New function.
12965 (nvptx_data_alignment): New function.
12966 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
12967 nvptx_preferred_simd_mode.
12968 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
12970 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
12972 2017-07-19 Tom de Vries <tom@codesourcery.com>
12974 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
12975 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
12976 (nvptx_vector_mode_supported): New function. Allow V2SImode.
12977 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
12978 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
12979 (mov<VECIM>_insn): New define_insn.
12980 (define_expand "mov<VECIM>): New define_expand.
12982 2017-07-19 Tom de Vries <tom@codesourcery.com>
12984 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
12986 2017-07-19 Jakub Jelinek <jakub@redhat.com>
12988 PR tree-optimization/81346
12989 * fold-const.h (fold_div_compare, range_check_type): Declare.
12990 * fold-const.c (range_check_type): New function.
12991 (build_range_check): Use range_check_type.
12992 (fold_div_compare): No longer static, rewritten into
12993 a match.pd helper function.
12994 (fold_comparison): Don't call fold_div_compare here.
12995 * match.pd (X / C1 op C2): New optimization using fold_div_compare
12996 as helper function.
12998 2017-07-19 Nathan Sidwell <nathan@acm.org>
13000 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
13001 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
13002 * tree.c (find_decls_types_r, verify_type): Use
13003 TYPE_{MIN,MAX}_VALUE_RAW.
13004 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
13005 (hash_tree): Likewise.
13006 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
13008 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
13011 2017-07-18 Tom de Vries <tom@codesourcery.com>
13013 PR middle-end/81464
13014 * omp-expand.c (expand_omp_for_static_chunk): Handle
13015 equal-argument loop exit phi.
13017 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
13020 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
13021 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
13022 operand 2 predicate.
13023 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
13024 operand 2 predicate.
13025 (ror,rol -> rorx splitters): Use const_int_operand as
13026 operand 2 predicate.
13028 2017-06-18 Richard Biener <rguenther@suse.de>
13030 PR tree-optimization/81410
13031 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
13032 the gap in the ! slp_perm SLP case after each group.
13034 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
13036 PR middle-end/81463
13037 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
13040 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
13042 PR middle-end/81462
13043 * predict.c (set_even_probabilities): Cleanup; do not affect
13044 probabilities that are already known.
13045 (combine_predictions_for_bb): Call even when count is set.
13047 2017-07-18 Nathan Sidwell <nathan@acm.org>
13049 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
13052 2017-07-18 Bin Cheng <bin.cheng@arm.com>
13055 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
13056 optimization for loop niter analysis.
13058 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
13061 * config/avr/avr.c (avr_optimize_casesi): Don't use
13062 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
13064 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
13066 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
13067 body_cost_vec from _vect_peel_extended_info.
13068 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
13069 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
13072 2017-07-18 Bin Cheng <bin.cheng@arm.com>
13074 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
13076 2017-07-18 Richard Biener <rguenther@suse.de>
13078 PR tree-optimization/80620
13079 PR tree-optimization/81403
13080 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
13081 info when re-using a VN table entry.
13083 2017-07-18 Richard Biener <rguenther@suse.de>
13085 PR tree-optimization/81418
13086 * tree-vect-loop.c (vectorizable_reduction): Properly compute
13087 vectype_in. Verify that with lane-reducing reduction operations
13088 we have a single def-use cycle.
13090 2017-07-17 Carl Love <cel@us.ibm.com>
13092 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
13094 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13095 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
13096 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
13097 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
13098 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
13099 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
13100 VMULOSW): New enum "unspec" values.
13101 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
13102 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
13103 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
13104 altivec_vmulosw): New patterns.
13105 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
13106 VMULOSW): Add definitions.
13107 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
13109 * config/alpha/alpha.c: Include predict.h.
13111 2017-07-17 Yury Gribov <tetra2005@gmail.com>
13113 * tree-vrp.c (compare_assert_loc): Fix comparison function
13114 to return predictable results.
13116 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
13118 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
13120 (subdi3): Likewise.
13121 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
13122 * doc/invoke.texi (mexpand-adddi): Update text.
13124 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
13126 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
13127 that also clobbers the CC register. The old expand code is moved
13129 (*arc_clzsi2): ... here.
13130 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
13131 the CC register. The old expand code is moved to ...
13132 (arc_ctzsi2): ... here.
13134 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
13136 * config/arc/arc.opt (mindexed-loads): Use initial value
13137 TARGET_INDEXED_LOADS_DEFAULT.
13138 (mauto-modify-reg): Use initial value
13139 TARGET_AUTO_MODIFY_REG_DEFAULT.
13140 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
13141 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
13142 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
13143 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
13145 2017-07-17 Martin Liska <mliska@suse.cz>
13148 * opts.c (finish_options): Do not allow -fgnu-tm
13149 w/ -fsanitize={kernel-,}address. Say sorry.
13151 2017-07-17 Bin Cheng <bin.cheng@arm.com>
13154 * tree-loop-distribution.c (classify_partition): Only assert on
13155 numer of iterations.
13156 (merge_dep_scc_partitions): Delete prameter. Update function call.
13157 (distribute_loop): Remove code handling loop with unknown niters.
13158 (pass_loop_distribution::execute): Skip loop with unknown niters.
13160 2017-07-17 Bin Cheng <bin.cheng@arm.com>
13163 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
13164 function sort_partitions_by_post_order.
13166 2017-07-17 Bin Cheng <bin.cheng@arm.com>
13168 PR tree-optimization/81374
13169 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
13170 the max index of basic blocks, rather than number of basic blocks.
13172 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
13174 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
13176 (arc_legitimate_pic_operand_p): Likewise.
13177 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
13179 (arc_needs_pcl_p): Likewise.
13180 (arc_legitimate_pc_offset_p): Likewise.
13181 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
13182 function is also used in constrains.md.
13183 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
13184 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
13185 PLUS. Only return true/false in known cases, otherwise assert.
13186 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
13187 is already called in arc_legitimate_constant_p.
13188 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
13190 (LEGITIMATE_PIC_OPERAND_P): Use
13191 arc_raw_symbolic_reference_mentioned_p function.
13192 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
13197 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
13198 Andrew Burgess <andrew.burgess@embecosm.com>
13200 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
13201 (arc_return_address_register): New function.
13202 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
13203 (arc_handle_fndecl_attribute): Add naked attribute.
13204 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
13205 (TARGET_WARN_FUNC_RETURN): Likewise.
13206 (arc_allocate_stack_slots_for_args): New function.
13207 (arc_warn_func_return): Likewise.
13208 (machine_function): Change type fn_type.
13209 (arc_compute_function_type): Consider new naked function type,
13210 change function return type.
13211 (arc_must_save_register): Adapt to handle new
13212 arc_compute_function_type's return type.
13213 (arc_expand_prologue): Likewise.
13214 (arc_expand_epilogue): Likewise.
13215 (arc_return_address_regs): Delete.
13216 (arc_return_address_register): New function.
13217 (arc_epilogue_uses): Use above function.
13218 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
13219 (arc_function_type): Change encoding, add naked type.
13220 (ARC_INTERRUPT_P): Change to handle the new encoding.
13221 (ARC_FAST_INTERRUPT_P): Likewise.
13222 (ARC_NORMAL_P): Define.
13223 (ARC_NAKED_P): Likewise.
13224 (arc_compute_function_type): Delete prototype.
13225 * config/arc/arc.md (in_ret_delay_slot): Use
13226 arc_return_address_register function.
13227 (simple_return): Likewise.
13228 (p_return_i): Likewise.
13230 2017-07-17 Jakub Jelinek <jakub@redhat.com>
13232 PR tree-optimization/81428
13233 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
13234 can't be built for those types.
13236 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
13238 Remove stuff dead since r239246.
13240 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
13241 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
13242 (avr_inform_devices): Remove dead stuff.
13244 2017-07-17 Tamar Christina <tamar.christina@arm.com>
13246 * config/arm/arm_neon.h: Fix softp typo.
13248 2017-07-17 Jakub Jelinek <jakub@redhat.com>
13250 PR tree-optimization/81365
13251 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
13252 aggregate moves onto bb predecessor edges, make sure there are no
13253 loads that could alias the lhs in between the start of bb and the
13256 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
13259 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
13260 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
13261 [LSHIFTRT, outer_code = TRUNCATE]: Same.
13263 2017-07-17 Jakub Jelinek <jakub@redhat.com>
13265 PR tree-optimization/81396
13266 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
13267 (init_symbolic_number): Initialize it to 1.
13268 (perform_symbolic_merge): Add n_ops from both operands into the new
13270 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
13271 without base_addr as useless if they need more than one operation.
13272 (bswap_replace): Handle !bswap case for NULL base_addr.
13274 2017-07-17 Tom de Vries <tom@codesourcery.com>
13277 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
13280 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
13282 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
13283 conditional builtin define __FIX_LEON3FT_B2BST.
13285 2017-07-17 Daniel Cederman <cederman@gaisler.com>
13287 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
13288 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
13291 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
13293 PR rtl-optimization/81424
13294 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
13295 to remove potential trapping from operands if -fnon-call-exceptions.
13297 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13299 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
13300 profile_proability for scalling.
13301 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
13303 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13305 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
13307 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13309 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
13310 fixpoint arithmetics.
13312 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13314 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
13315 fixpoint arithmetics.
13317 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13319 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
13320 fixpoint arithmetics.
13322 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13324 * profile-count.h (profile_probability::from_reg_br_prob_note,
13325 profile_probability::to_reg_br_prob_note): New functions.
13326 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
13327 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
13328 * predict.c (probability_reliable_p): Update.
13329 (edge_probability_reliable_p): Update.
13330 (br_prob_note_reliable_p): Update.
13331 (invert_br_probabilities): Update.
13332 (add_reg_br_prob_note): New function.
13333 (combine_predictions_for_insn): Update.
13334 * asan.c (asan_clear_shadow): Update.
13335 * cfgbuild.c (compute_outgoing_frequencies): Update.
13336 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
13337 (update_br_prob_note): Update.
13338 (rtl_verify_edges): Update.
13339 (purge_dead_edges): Update.
13340 (fixup_reorder_chain): Update.
13341 * emit-rtl.c (try_split): Update.
13342 * ifcvt.c (cond_exec_process_insns): Update.
13343 (cond_exec_process_if_block): Update.
13344 (dead_or_predicable): Update.
13345 * internal-fn.c (expand_addsub_overflow): Update.
13346 (expand_neg_overflow): Update.
13347 (expand_mul_overflow): Update.
13348 * loop-doloop.c (doloop_modify): Update.
13349 * loop-unroll.c (compare_and_jump_seq): Update.
13350 * optabs.c (emit_cmp_and_jump_insn_1): Update.
13351 * predict.h: Update.
13352 * reorg.c (mostly_true_jump): Update.
13354 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
13355 * config/alpha/alpha.c (emit_unlikely_jump): Update.
13356 * config/arc/arc.c: (emit_unlikely_jump): Update.
13357 * config/arm/arm.c: (emit_unlikely_jump): Update.
13358 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
13359 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
13360 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
13361 (ix86_print_operand): Update.
13362 (ix86_split_fp_branch): Update.
13363 (predict_jump): Update.
13364 * config/ia64/ia64.c (ia64_print_operand): Update.
13365 * config/mmix/mmix.c (mmix_print_operand): Update.
13366 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
13367 (rs6000_expand_split_stack_prologue): Update.
13368 * config/rs6000/rs6000.c: Update.
13369 * config/s390/s390.c (s390_expand_vec_strlen): Update.
13370 (s390_expand_vec_movstr): Update.
13371 (s390_expand_cs_tdsi): Update.
13372 (s390_expand_split_stack_prologue): Update.
13373 * config/sh/sh.c (sh_print_operand): Update.
13374 (expand_cbranchsi4): Update.
13375 (expand_cbranchdi4): Update.
13376 * config/sparc/sparc.c (output_v9branch): Update.
13377 * config/spu/spu.c (get_branch_target): Update.
13378 (ea_load_store_inline): Update.
13379 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
13380 * config/tilepro/tilepro.c: Update.
13382 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
13384 * gimplify.c (mostly_copy_tree_r): Revert latest change.
13385 (gimplify_save_expr): Likewise.
13387 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
13389 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
13391 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
13393 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
13395 * timevar.def (TV_IPA_FNSUMMARY): Define.
13397 2017-07-16 Daniel Cederman <cederman@gaisler.com>
13399 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
13400 to back store errata sensitive sequence from being generated.
13401 (sqrtdf2_fix): Likewise.
13403 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
13405 * tree-ssa-threadupdate.c (compute_path_counts,
13406 update_joiner_offpath_counts): Use profile_probability.
13408 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
13411 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
13413 * config/arm/arm-c.c (arm_cpu_builtins): Define
13414 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
13416 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
13418 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13419 array entries to represent __ieee128 versions of the
13420 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
13421 scalar_extract_sig, and scalar_insert_exp built-in functions.
13422 (altivec_resolve_overloaded_builtin): Add special case handling
13423 for the __builtin_scalar_insert_exp function, as represented by
13424 the P9V_BUILTIN_VEC_VSIEDP constant.
13425 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
13426 exponent support for __ieee128 argument.
13427 (VSESQP): Add scalar extract signature support for __ieee128
13429 (VSTDCNQP): Add scalar test negative support for __ieee128
13431 (VSIEQP): Add scalar insert exponent support for __int128 argument
13432 with __ieee128 result.
13433 (VSIEQPF): Add scalar insert exponent support for __ieee128
13434 argument with __ieee128 result.
13435 (VSTDCQP): Add scalar test data class support for __ieee128
13437 (VSTDCNQP): Add overload support for scalar test negative with
13438 __ieee128 argument.
13439 (VSTDCQP): Add overload support for scalar test data class
13440 __ieee128 argument.
13441 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
13442 UNSPEC_VSX_SXSIGDP.
13443 (UNSPEC_VSX_SIEXPQP): New constant.
13444 (xsxexpqp): New insn for VSX scalar extract exponent quad
13446 (xsxsigqp): New insn for VSX scalar extract significand quad
13448 (xsiexpqpf): New insn for VSX scalar insert exponent quad
13449 precision with floating point argument.
13450 (xststdcqp): New expand for VSX scalar test data class quad
13452 (xststdcnegqp): New expand for VSX scalar test negative quad
13454 (xststdcqp): New insn to match expansions for VSX scalar test data
13455 class quad precision and VSX scalar test negative quad precision.
13456 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
13457 special case operand checking to enforce that second operand of
13458 VSX scalar test data class with quad precision argument is a 7-bit
13460 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
13461 prototypes and descriptions of __ieee128 versions of
13462 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
13463 scalar_test_data_class, and scalar_test_neg built-in functions.
13465 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13467 PR tree-optimization/81162
13468 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
13469 replace a negate with an add.
13471 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
13473 * doc/invoke.texi (arm/-mcpu): Document +crypto.
13475 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
13477 * config/arm/arm-c.c (arm_cpu_builtins): Define
13478 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
13480 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
13482 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
13483 (armv8-r): Set ARM Cortex-R52 as default CPU.
13484 * config/arm/arm-tables.opt: Regenerate.
13485 * config/arm/arm-tune.md: Regenerate.
13486 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
13488 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
13489 extension for -mcpu=cortex-r52.
13491 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
13493 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
13494 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
13495 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
13496 (fp-armv8): Define it as FP_ARMv8 only.
13497 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
13498 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
13500 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
13501 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
13502 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
13503 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
13504 than TARGET_FPU_ARMV8.
13505 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
13506 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
13507 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
13509 * config/arm/neon.md (neon_vrint): Likewise.
13510 (neon_vcvt): Likewise.
13511 (neon_<fmaxmin_op><mode>): Likewise.
13512 (<fmaxmin><mode>3): Likewise.
13513 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
13514 * config/arm/predicates.md (arm_cond_move_operator): Check against
13515 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
13517 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
13519 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
13520 to top of function.
13522 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13524 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
13525 loop in comment with memset.
13527 2017-07-14 Martin Liska <mliska@suse.cz>
13529 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
13530 * dwarf2out.c (is_java): Remove the function.
13531 (output_pubname): Remove usage of the function.
13532 (lower_bound_default): Remove usage of DW_LANG_Java.
13533 (gen_compile_unit_die): Likewise.
13534 * gcc.c: Remove compiler defaults for .java and .zip files.
13535 * gimple-expr.c (remove_suffix): Change as there's no longer
13536 extension than 4-letter one.
13537 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
13538 (gimplify_save_expr): Likewise.
13539 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
13540 as it's possible even for other languages than Java.
13541 * langhooks.h (struct lang_hooks): Remove Java from a comment.
13542 * lto-opts.c (lto_write_options): Remove reference to Java.
13543 * opts.c (strip_off_ending): Update file extension handling.
13544 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
13545 * tree-eh.c (lower_resx): Likewise.
13546 * tree.c (free_lang_data_in_type): Remove dead code.
13547 (find_decls_types_r): Likewise.
13548 (build_common_builtin_nodes): Remove Java from a comment.
13549 (verify_type): Remove dead code.
13550 * varasm.c (assemble_external): Remove Java from a comment.
13552 2017-07-14 Martin Liska <mliska@suse.cz>
13554 * opts.c (finish_options): Add quotes.
13555 (common_handle_option): Likewise.
13557 2017-07-14 Martin Liska <mliska@suse.cz>
13559 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
13560 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
13561 Remove N_SO_PASCAL.
13562 * dwarf2out.c (lower_bound_default): Do not handle
13564 (gen_compile_unit_die): Likewise.
13565 * gcc.c: Remove default extension binding for GNU Pascal.
13566 * stmt.c: Remove Pascal language from a comment.
13567 * xcoffout.c: Likewise.
13569 2017-07-13 David Malcolm <dmalcolm@redhat.com>
13572 * diagnostic-show-locus.c (fixit_cmp): New function.
13573 (layout::layout): Sort m_fixit_hints.
13574 (column_range::column_range): Assert that the values are valid.
13575 (struct char_span): New struct.
13576 (correction::overwrite): New method.
13577 (struct source_line): New struct.
13578 (line_corrections::add_hint): Add assertions. Reimplement memcpy
13579 calls in terms of classes source_line and char_span, and
13580 correction::overwrite.
13581 (selftest::test_overlapped_fixit_printing_2): New function.
13582 (selftest::diagnostic_show_locus_c_tests): Call it.
13584 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
13586 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
13587 early if there is no lhs.
13589 2017-07-13 Martin Liska <mliska@suse.cz>
13591 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
13592 (gen_reference_type_die): Likewise.
13593 * stor-layout.c: Remove Pascal-related comment.
13595 2017-07-13 Martin Liska <mliska@suse.cz>
13597 * opts.c (finish_options): Add quotes to error messages.
13598 (parse_sanitizer_options): Likewise.
13600 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
13602 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
13604 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
13606 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
13608 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
13610 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
13612 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
13614 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
13617 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
13618 provides the hardware capability bits, define the macro
13619 __BUILTIN_CPU_SUPPORTS__.
13620 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
13621 if GLIBC does not provide the hardware capability bits. Add a
13622 gcc_unreachable call if the built-in cpu function is neither
13623 __builtin_cpu_is nor __builtin_cpu_supports.
13624 (rs6000_get_function_versions_dispatcher): Change the warning
13625 that an old GLIBC is used which does not export the capability
13626 bits to be an error.
13627 * doc/extend.texi (target_clones attribute): Document the
13628 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
13629 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
13630 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
13631 the macros defined by GCC if the newer GLIBC is available.
13633 2017-07-12 Jeff Law <law@redhat.com>
13635 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
13636 remaining includes slightly.
13637 * config/riscv/riscv-builtins.c: Include profile-count.h.
13639 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
13642 * config/avr/avr.c (avr_set_current_function): In diagnostic
13643 messages: Quote keywords and (parts of) identifiers.
13644 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
13647 2017-07-12 Carl Love <cel@us.ibm.com>
13649 * config/rs6000/rs6000-c.c: Add support for built-in functions
13650 vector bool char vec_revb (vector bool char);
13651 vector bool short vec_revb (vector short char);
13652 vector bool int vec_revb (vector bool int);
13653 vector bool long long vec_revb (vector bool long long);
13654 * doc/extend.texi: Update the built-in documentation file for the
13655 new built-in functions.
13657 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13659 * config/s390/s390.md: Remove movcc splitter.
13661 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13663 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
13664 load/store on condition.
13666 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
13669 * config/avr/avr.c (avr_encode_section_info)
13670 [progmem && !TREE_READONLY]: Error if progmem object needs
13673 2017-07-11 Michael Collison <michael.collison@arm.com>
13675 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
13678 2017-07-11 Carl Love <cel@us.ibm.com>
13680 * config/rs6000/rs6000-c.c: Add support for builtins
13681 vector unsigned int vec_parity_lsbb (vector signed int);
13682 vector unsigned int vec_parity_lsbb (vector unsigned int);
13683 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
13684 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
13685 vector unsigned long long vec_parity_lsbb (vector signed long long);
13686 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
13687 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
13688 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
13689 * doc/extend.texi: Update the built-in documentation file for the
13690 new built-in functions.
13692 2017-07-11 David Malcolm <dmalcolm@redhat.com>
13694 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
13695 (layout::m_primary_loc): New field.
13696 (layout::layout): Initialize new field. Move location filtering
13697 logic from here to...
13698 (layout::maybe_add_location_range): ...this new method. Add
13699 support for filtering to just the lines already specified by other
13701 (layout::will_show_line_p): New method.
13702 (gcc_rich_location::add_location_if_nearby): New method.
13703 (selftest::test_add_location_if_nearby): New test function.
13704 (selftest::diagnostic_show_locus_c_tests): Call it.
13705 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
13708 2017-07-11 Tom de Vries <tom@codesourcery.com>
13710 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
13711 (bb_first_real_insn): New function.
13712 (nvptx_single): Add extra initialization of broadcasted condition
13715 2017-07-11 Nathan Sidwell <nathan@acm.org>
13717 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
13719 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
13721 * doc/extend.texi (AVR Function Attributes): Remove weblink to
13722 Binutils doc as TEXI will mess them up.
13723 * doc/invoke.texi (AVR Options): Same here.
13725 2017-07-11 Daniel Cederman <cederman@gaisler.com>
13727 * config/sparc/sparc.opt (mfix-ut700): New option.
13728 (mfix-gr712rc): Likewise.
13729 (sparc_fix_b2bst): New variable.
13730 * doc/invoke.texi (SPARC options): Document them.
13731 (ARM options): Fix warnings.
13732 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
13733 instructions to prevent sequences that can trigger the store-store
13734 errata for certain LEON3FT processors.
13735 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
13736 (sparc_option_override): Set sparc_fix_b2bst appropriately.
13737 * config/sparc/sparc.md (fix_b2bst): New attribute.
13738 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
13740 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
13743 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
13745 (*rsqrtsf2_sse): Ditto.
13747 (div<mode>3): Macroize insn from divdf3 and divsf3
13748 using MODEF mode iterator.
13750 2017-07-10 Martin Sebor <msebor@redhat.com>
13752 PR tree-optimization/80397
13753 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
13754 instead of testing for equality to INTEGER_TYPE.
13756 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
13758 * config.gcc: Remove uclibc from arc target spec.
13760 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
13762 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
13764 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
13767 * lto-wrapper.c (remove_option): New function.
13768 (merge_and_complain): Merge PIC/PIE options more realistically.
13770 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
13772 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
13776 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
13777 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
13778 * config.in: Regenerate.
13779 * configure: Regenerate.
13780 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
13781 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
13782 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
13783 (TARGET_GASISR_PROLOGUES): ...target mask.
13784 * common/config/avr/avr-common.c
13785 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
13786 Set -mgas-isr-prologues.
13787 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
13788 INSERT_PASS_BEFORE for it.
13789 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
13790 * config/avr/avr.c (avr_option_override)
13791 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
13792 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
13793 (avr_attribute_table) <no_gccisr>: Add new function attribute.
13794 (avr_set_current_function) <is_no_gccisr>: Init machine field.
13795 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
13797 (make_avr_pass_pre_proep): New function.
13798 (emit_push_sfr) <treg>: Add argument to function and use it
13799 instead of TMP_REG.
13800 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
13801 and set machine->gasisr.yes.
13802 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
13803 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
13804 __gcc_isr.n_pushed to .L__stack_usage.
13805 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
13806 (avr_asm_final_postscan_insn): ...this new static function.
13807 * config/avr/avr.h (machine_function)
13808 <is_no_gccisr, use_L__stack_usage>: New fields.
13809 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
13810 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
13811 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
13812 (gasisr, *gasisr): New expander and insn.
13813 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
13814 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
13815 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
13817 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
13819 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
13822 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
13824 Move jump-tables out of .text again.
13827 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
13828 (ASM_OUTPUT_ADDR_VEC): New function.
13829 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
13830 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
13831 INSN_ADDRESSes as asm comment.
13832 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
13833 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
13834 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
13835 * config/avr/avr.md (*tablejump): Adjust comment.
13836 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
13837 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
13839 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
13840 (avr_output_addr_vec): New proto.
13841 (avr_log_t) <insn_addresses>: New field.
13843 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
13846 * config/i386/i386.c (ix86_function_arg_advance): Set
13847 outgoing_args_on_stack to true if there are outgoing arguments
13849 (ix86_function_arg): Likewise.
13850 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
13851 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
13852 * config/i386/i386.h (machine_function): Add
13853 outgoing_args_on_stack.
13855 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
13857 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
13858 supporting pthreds.
13859 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
13861 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
13863 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
13864 (REAL_H): Remove $(MACHMODE_H).
13865 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
13867 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
13868 $(MACHMODE_H) and double-int.h.
13869 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
13871 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
13874 2017-07-07 Andrew Pinski <apinski@cavium.com>
13876 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
13877 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
13879 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13881 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
13882 Add warning if GCC was not configured to link against a GLIBC that
13883 exports the hardware capability bits.
13884 (make_resolver_func): Make resolver function private and not a
13885 COMDAT function. Create the name with clone_function_name instead
13886 of make_unique_name.
13889 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
13890 correct operand in doing the split.
13892 2017-07-07 Carl Love <cel@us.ibm.com>
13894 * config/rs6000/rs6000-c: Add support for built-in function
13895 vector unsigned short vec_pack_to_short_fp32 (vector float,
13897 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
13898 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
13899 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
13900 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
13901 (convert_4f32_8i16): Add define_expand.
13902 * doc/extend.texi: Update the built-in documentation file for the
13903 new built-in function.
13905 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13907 * config/sparc/m8.md: New file.
13908 * config/sparc/sparc.md: Include m8.md.
13910 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13912 * config/sparc/sparc.opt: New option -mvis4b.
13913 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
13914 (sparc_option_override): Handle VIS4B.
13915 (enum sparc_builtins): Define
13916 SPARC_BUILTIN_DICTUNPACK{8,16,32},
13917 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
13918 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
13919 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
13920 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
13921 (check_constant_argument): New function.
13922 (sparc_vis_init_builtins): Define builtins
13923 __builtin_vis_dictunpack{8,16,32},
13924 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
13925 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
13926 __builtin_vis_fpcmpde{8,16,32}shl and
13927 __builtin_vis_fpcmpur{8,16,32}shl.
13928 (sparc_expand_builtin): Check that the constant operands to
13929 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
13930 constant and in range.
13931 * config/sparc/sparc-c.c (sparc_target_macros): Handle
13933 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
13934 (SPARC_IMM5_P): Likewise.
13935 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
13936 (enabled): Handle vis4b.
13937 (UNSPEC_DICTUNPACK): New unspec.
13938 (UNSPEC_FPCMPSHL): Likewise.
13939 (UNSPEC_FPUCMPSHL): Likewise.
13940 (UNSPEC_FPCMPDESHL): Likewise.
13941 (UNSPEC_FPCMPURSHL): Likewise.
13942 (cpu_feature): New CPU feature `vis4b'.
13943 (dictunpack{8,16,32}): New insns.
13944 (FPCSMODE): New mode iterator.
13945 (fpcscond): New code iterator.
13946 (fpcsucond): Likewise.
13947 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
13948 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
13949 (fpcmpde{8,16,32}{si,di}shl): Likewise.
13950 (fpcmpur{8,16,32}{si,di}shl): Likewise.
13951 * config/sparc/constraints.md: Define constraints `q' for unsigned
13952 2-bit integer constants and `t' for unsigned 5-bit integer
13954 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
13956 (imm5_operand_dictunpack16): Likewise.
13957 (imm5_operand_dictunpack32): Likewise.
13958 (imm2_operand): Likewise.
13959 * doc/invoke.texi (SPARC Options): Document -mvis4b.
13960 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
13961 ditunpack* and fpcmp*shl builtins.
13963 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13965 * config.gcc: Handle m8 in --with-{cpu,tune} options.
13966 * config.in: Add HAVE_AS_SPARC6 define.
13967 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
13969 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
13971 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
13972 (CPP_CPU_SPEC): Handle m8.
13973 (ASM_CPU_SPEC): Likewise.
13974 * config/sparc/sparc-opts.h (enum processor_type): Add
13976 * config/sparc/sparc.c (m8_costs): New struct.
13977 (sparc_option_override): Handle TARGET_CPU_m8.
13978 (sparc32_initialize_trampoline): Likewise.
13979 (sparc64_initialize_trampoline): Likewise.
13980 (sparc_issue_rate): Likewise.
13981 (sparc_register_move_cost): Likewise.
13982 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
13983 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
13984 (ASM_CPU64_DEFAULT_SPEC): Likewise.
13985 (CPP_CPU_SPEC): Handle M8.
13986 (ASM_CPU_SPEC): Likewise.
13987 (AS_M8_FLAG): Define.
13988 * config/sparc/sparc.md: Add m8 to the cpu attribute.
13989 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
13990 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
13992 * configure: Regenerate.
13993 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
13996 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
13998 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
14000 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
14001 ("*movdi_insn_sp32"): Do not set v3pipe.
14002 ("*movsi_insn"): Likewise.
14003 ("*movdi_insn_sp64"): Likewise.
14004 ("*movsf_insn"): Likewise.
14005 ("*movdf_insn_sp32"): Likewise.
14006 ("*movdf_insn_sp64"): Likewise.
14007 ("*zero_extendsidi2_insn_sp64"): Likewise.
14008 ("*sign_extendsidi2_insn"): Likewise.
14009 ("*mov<VM32:mode>_insn"): Likewise.
14010 ("*mov<VM64:mode>_insn_sp64"): Likewise.
14011 ("*mov<VM64:mode>_insn_sp32"): Likewise.
14012 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
14013 ("<vlop:code><VL:mode>3"): Likewise.
14014 ("*not_<vlop:code><VL:mode>3"): Likewise.
14015 ("*nand<VL:mode>_vis"): Likewise.
14016 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
14017 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
14018 ("one_cmpl<VL:mode>2"): Likewise.
14019 ("faligndata<VM64:mode>_vis"): Likewise.
14020 ("alignaddrsi_vis"): Likewise.
14021 ("alignaddrdi_vis"): Likweise.
14022 ("alignaddrlsi_vis"): Likewise.
14023 ("alignaddrldi_vis"): Likewise.
14024 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
14025 ("bmaskdi_vis"): Likewise.
14026 ("bmasksi_vis"): Likewise.
14027 ("bshuffle<VM64:mode>_vis"): Likewise.
14028 ("cmask8<P:mode>_vis"): Likewise.
14029 ("cmask16<P:mode>_vis"): Likewise.
14030 ("cmask32<P:mode>_vis"): Likewise.
14031 ("pdistn<P:mode>_vis"): Likewise.
14032 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
14034 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14036 * config/sparc/sparc.md ("subtype"): New insn attribute.
14037 ("*wrgsr_sp64"): Set insn subtype.
14038 ("*rdgsr_sp64"): Likewise.
14039 ("alignaddrsi_vis"): Likewise.
14040 ("alignaddrdi_vis"): Likewise.
14041 ("alignaddrlsi_vis"): Likewise.
14042 ("alignaddrldi_vis"): Likewise.
14043 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
14044 ("fexpand_vis"): Likewise.
14045 ("fpmerge_vis"): Likewise.
14046 ("faligndata<VM64:mode>_vis"): Likewise.
14047 ("bshuffle<VM64:mode>_vis"): Likewise.
14048 ("cmask8<P:mode>_vis"): Likewise.
14049 ("cmask16<P:mode>_vis"): Likewise.
14050 ("cmask32<P:mode>_vis"): Likewise.
14051 ("fchksm16_vis"): Likewise.
14052 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
14053 ("fmean16_vis"): Likewise.
14054 ("fp<plusminus_insn>64_vis"): Likewise.
14055 ("<plusminus_insn>v8qi3"): Likewise.
14056 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
14057 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
14058 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
14059 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
14060 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
14061 ("*movqi_insn"): Likewise.
14062 ("*movhi_insn"): Likewise.
14063 ("*movsi_insn"): Likewise.
14064 ("movsi_pic_gotdata_op"): Likewise.
14065 ("*movdi_insn_sp32"): Likewise.
14066 ("*movdi_insn_sp64"): Likewise.
14067 ("movdi_pic_gotdata_op"): Likewise.
14068 ("*movsf_insn"): Likewise.
14069 ("*movdf_insn_sp32"): Likewise.
14070 ("*movdf_insn_sp64"): Likewise.
14071 ("*zero_extendhisi2_insn"): Likewise.
14072 ("*zero_extendqihi2_insn"): Likewise.
14073 ("*zero_extendqisi2_insn"): Likewise.
14074 ("*zero_extendqidi2_insn"): Likewise.
14075 ("*zero_extendhidi2_insn"): Likewise.
14076 ("*zero_extendsidi2_insn_sp64"): Likewise.
14077 ("ldfsr"): Likewise.
14078 ("prefetch_64"): Likewise.
14079 ("prefetch_32"): Likewise.
14080 ("tie_ld32"): Likewise.
14081 ("tie_ld64"): Likewise.
14082 ("*tldo_ldub_sp32"): Likewise.
14083 ("*tldo_ldub1_sp32"): Likewise.
14084 ("*tldo_ldub2_sp32"): Likewise.
14085 ("*tldo_ldub_sp64"): Likewise.
14086 ("*tldo_ldub1_sp64"): Likewise.
14087 ("*tldo_ldub2_sp64"): Likewise.
14088 ("*tldo_ldub3_sp64"): Likewise.
14089 ("*tldo_lduh_sp32"): Likewise.
14090 ("*tldo_lduh1_sp32"): Likewise.
14091 ("*tldo_lduh_sp64"): Likewise.
14092 ("*tldo_lduh1_sp64"): Likewise.
14093 ("*tldo_lduh2_sp64"): Likewise.
14094 ("*tldo_lduw_sp32"): Likewise.
14095 ("*tldo_lduw_sp64"): Likewise.
14096 ("*tldo_lduw1_sp64"): Likewise.
14097 ("*tldo_ldx_sp64"): Likewise.
14098 ("*mov<VM32:mode>_insn"): Likewise.
14099 ("*mov<VM64:mode>_insn_sp64"): Likewise.
14100 ("*mov<VM64:mode>_insn_sp32"): Likewise.
14102 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14104 * config/sparc/sparc.md ("type"): New insn type viscmp.
14105 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
14107 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
14108 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
14109 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
14110 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
14112 ("n7_vis_logical_11cycle"): Likewise.
14113 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
14114 * config/sparc/niagara2.md ("niag3_vis": Likewise.
14115 * config/sparc/niagara.md ("niag_vis"): Likewise.
14116 * config/sparc/ultra3.md ("us3_fga"): Likewise.
14117 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
14119 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14121 * config/sparc/sparc.md: New instruction type `bmask'.
14122 (bmaskdi_vis): Use the `bmask' type.
14123 (bmasksi_vis): Likewise.
14124 * config/sparc/ultra3.md (us3_array): Likewise.
14125 * config/sparc/niagara7.md (n7_array): Likewise.
14126 * config/sparc/niagara4.md (n4_array): Likewise.
14127 * config/sparc/niagara2.md (niag2_vis): Likewise.
14128 (niag3_vis): Likewise.
14129 * config/sparc/niagara.md (niag_vis): Likewise.
14131 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
14133 * ipa-comdats.c: Remove optimize check from gate.
14134 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
14135 for functions not optimized.
14136 (ipa_fn_summary_read): Skip optimize check.
14137 (ipa_fn_summary_write): Likewise.
14138 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
14140 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
14142 (can_inline_edge_p): Check flag_pcc_struct_return for match.
14143 (check_callers): Give up on caller which is not optimized.
14144 (inline_small_functions): Likewise.
14145 (ipa_inline): Do not give up when not optimizing.
14146 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
14147 away unoptimizes cdtors.
14148 (whole_program_function_and_variable_visibility): Do
14149 ipa_discover_readonly_nonaddressable_vars in LTO mode.
14150 * ipa.c (process_references): Do not check optimize.
14151 (symbol_table::remove_unreachable_nodes): Update optimize check.
14152 (set_writeonly_bit): Update optimize check.
14153 (pass_ipa_cdtor_merge::gate): Do not check optimize.
14154 (pass_ipa_single_use::gate): Remove.
14156 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
14158 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
14159 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
14160 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
14161 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
14162 permute_load, permute_store, adjust_extract, adjust_splat,
14163 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
14164 replace_swap_with_copy, dump_swap_insn_table,
14165 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
14166 recombine_lvx_pattern, recombine_stvx_pattern,
14167 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
14168 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
14169 to file rs6000-p8swap.c.
14170 * config/rs6000/rs6000-p8swap.c: New file.
14171 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
14172 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
14173 and rs6000*-*-* targets.
14175 2017-07-06 David Malcolm <dmalcolm@redhat.com>
14177 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
14179 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
14181 * lto-wrapper.c (merge_and_complain): Do not merge
14182 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
14183 fsigned_zeros, ftrapping_math, fwrapv.
14184 (append_compiler_options): Do not track these options.
14185 (append_linker_options): Likewie
14187 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
14189 * cgraphunit.c (cgraph_node::finalize_function): When
14190 !flag_toplevel_reorde set no_reorder flag.
14191 (varpool_node::finalize_decl): Likewise.
14192 (symbol_table::compile): Drop no toplevel reorder path.
14194 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
14196 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
14197 edges; zero probability is not better than uninitialized.
14199 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
14201 * asan.h (asan_sanitize_allocas_p): Declare.
14202 * asan.c (asan_sanitize_allocas_p): New function.
14203 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
14204 (handle_builtin_alloca): Likewise.
14205 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
14206 if !asan_sanitize_allocas_p.
14207 * params.def (asan-instrument-allocas): Add new option.
14208 * params.h (ASAN_PROTECT_ALLOCAS): Define.
14209 * opts.c (common_handle_option): Disable allocas sanitization for
14212 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
14214 * asan.c: Include gimple-fold.h.
14215 (get_last_alloca_addr): New function.
14216 (handle_builtin_stackrestore): Likewise.
14217 (handle_builtin_alloca): Likewise.
14218 (asan_emit_allocas_unpoison): Likewise.
14219 (get_mem_refs_of_builtin_call): Add new parameter, remove const
14220 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
14221 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
14222 (instrument_builtin_call): Pass gimple iterator to
14223 get_mem_refs_of_builtin_call.
14224 (last_alloca_addr): New global.
14225 * asan.h (asan_emit_allocas_unpoison): Declare.
14226 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
14227 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
14228 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
14229 if function calls alloca.
14230 * gimple-fold.c (replace_call_with_value): Remove static keyword.
14231 * gimple-fold.h (replace_call_with_value): Declare.
14232 * internal-fn.c: Include asan.h.
14233 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
14234 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
14236 2017-07-06 David Malcolm <dmalcolm@redhat.com>
14238 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
14239 (C_SELFTEST_FLAGS): New.
14240 (CPP_SELFTEST_FLAGS): New.
14241 (SELFTEST_DEPS): New, from deps of s-selftest.
14242 (C_SELFTEST_DEPS): New, from deps of s-selftest.
14243 (CPP_SELFTEST_DEPS): New.
14244 (selftest): Add dependency on s-selftest-c++.
14245 (s-selftest): Rename to...
14246 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
14247 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
14248 than SELFTEST_FLAGS.
14249 (selftest-gdb): Rename to...
14250 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
14252 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
14253 (selftest-valgrind): Rename to...
14254 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
14256 (selftest-valgrind): Reintroduce as an alias for
14257 selftest-c-valgrind.
14258 (s-selftest-c++): New.
14259 (selftest-c++-gdb): New.
14260 (selftest-c++-valgrind): New.
14262 2017-07-06 Olivier Hainque <hainque@adacore.com>
14264 * gcc.c (process_command): When deciding if undefined variables
14265 should be ignored when processing specs, accept "gcc -v" as well.
14267 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
14269 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
14270 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
14272 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
14274 * config/arm/arm-cpus.in (armv8-r): Add new entry.
14275 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
14276 * config/arm/arm-tables.opt: Regenerate.
14277 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
14279 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
14281 2017-07-06 Carl Love <cel@us.ibm.com>
14283 * ChangeLog: Clean up from mid air collision
14285 2017-07-06 Carl Love <cel@us.ibm.com>
14287 * config/rs6000/rs6000-c.c: Add support for built-in functions
14288 vector signed int vec_subc (vector signed int, vector signed int);
14289 vector signed __int128 vec_subc (vector signed __int128,
14290 vector signed __int128);
14291 vector unsigned __int128 vec_subc (vector unsigned __int128,
14292 vector unsigned __int128);
14293 vector signed int vec_sube (vector signed int, vector signed int,
14294 vector signed int);
14295 vector unsigned int vec_sube (vector unsigned int,
14296 vector unsigned int,
14297 vector unsigned int);
14298 vector signed __int128 vec_sube (vector signed __int128,
14299 vector signed __int128,
14300 vector signed__int128);
14301 vector unsigned __int128 vec_sube (vector unsigned __int128,
14302 vector unsigned __int128,
14303 vector unsigned __int128);
14304 vector signed int vec_subec (vector signed int, vector signed int,
14305 vector signed int);
14306 vector unsigned int vec_subec (vector unsigned int,
14307 vector unsigned int,
14308 vector unsigned int);
14309 vector signed __int128 vec_subec (vector signed __int128,
14310 vector signed __int128,
14311 vector signed__int128);
14312 vector unsigned __int128 vec_subec (vector unsigned __int128,
14313 vector unsigned __int128,
14314 vector unsigned __int128);
14315 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
14316 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
14317 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
14318 BU_ALTIVEC_OVERLOAD_X definitions.
14319 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
14320 * doc/extend.texi: Update the built-in documentation file for the new
14321 built-in functions.
14323 2017-07-06 David Malcolm <dmalcolm@redhat.com>
14326 * diagnostic-show-locus.c (layout::layout): Use start and finish
14327 spelling location for the start and finish of each range.
14328 * genmatch.c (linemap_client_expand_location_to_spelling_point):
14329 Add unused aspect param.
14330 * input.c (expand_location_1): Add "aspect" param, and use it
14331 to access the correct part of the location.
14332 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
14334 (expand_location_to_spelling_point): Likewise.
14335 (linemap_client_expand_location_to_spelling_point): Add "aspect"
14336 param, and pass it to expand_location_1.
14338 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
14340 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
14341 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
14342 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
14343 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
14344 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
14345 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
14346 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
14347 _mm_maskz_getmant_ss): New intrinsics.
14348 (__builtin_ia32_getexpss128_mask): Changed to ...
14349 __builtin_ia32_getexpss128_round ... this.
14350 (__builtin_ia32_getexpsd128_mask): Changed to ...
14351 __builtin_ia32_getexpsd128_round ... this.
14352 * config/i386/i386-builtin-types.def
14353 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
14354 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
14355 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
14356 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
14357 __builtin_ia32_getmantss_mask_round): New builtins.
14358 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
14359 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
14360 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
14361 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
14362 * config/i386/sse.md
14363 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
14364 avx512f_sgetexp<mode><mask_scalar_name>
14365 <round_saeonly_scalar_name> ... this.
14366 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
14367 %0, %1, %2<round_saeonly_op3>}): Changed to ...
14368 vgetexp<ssescalarmodesuffix>
14369 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
14370 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
14371 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
14372 avx512f_vgetmant<mode><mask_scalar_name>
14373 <round_saeonly_scalar_name> ... this.
14374 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
14375 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
14376 vgetmant<ssescalarmodesuffix>
14377 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
14378 %0<mask_scalar_operand4>, %1, %2
14379 <round_saeonly_scalar_mask_op4>, %3} ... this.
14380 * config/i386/subst.md (mask_scalar_operand4,
14381 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
14382 round_saeonly_scalar_nimm_predicate): New subst attributes.
14384 2017-07-06 Julia Koval <julia.koval@intel.com>
14386 * config/i386/i386.c (ix86_erase_embedded_rounding):
14387 Remove code for old rounding pattern.
14389 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
14391 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
14393 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
14395 * doc/sourcebuild.texi (Test Directives, Variants of
14396 dg-require-support): Add documentation for dg-require-stack-check.
14398 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
14400 * config/i386/subst.md (mask_scalar, round_scalar,
14401 round_saeonly_scalar): New meta-templates.
14402 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
14403 round_scalar_mask_operand3, round_scalar_mask_op3,
14404 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
14405 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
14406 round_saeonly_scalar_constraint,
14407 round_saeonly_scalar_prefix): New subst attribute.
14408 * config/i386/sse.md
14409 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
14410 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
14411 <round_scalar_name> ... this.
14412 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
14413 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
14414 <round_scalar_name> ... this.
14415 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
14416 <sse>_vm<code><mode>3<mask_scalar_name>
14417 <round_saeonly_scalar_name> ... this.
14418 (v<plusminus_mnemonic><ssescalarmodesuffix>
14419 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
14420 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
14421 v<plusminus_mnemonic><ssescalarmodesuffix>
14422 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
14423 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
14424 (v<multdiv_mnemonic><ssescalarmodesuffix>
14425 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
14426 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
14427 v<multdiv_mnemonic><ssescalarmodesuffix>
14428 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
14429 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
14430 (v<maxmin_float><ssescalarmodesuffix>
14431 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
14432 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
14433 v<maxmin_float><ssescalarmodesuffix>
14434 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
14435 %0<mask_scalar_operand3>, %1, %<iptr>2
14436 <round_saeonly_scalar_mask_op3>} ... this.
14438 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
14440 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
14441 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
14443 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
14444 Alan Hayward <alan.hayward@arm.com>
14445 David Sherwood <david.sherwood@arm.com>
14447 * combine.c (simplify_if_then_else): Remove "enum" before
14449 * compare-elim.c (can_eliminate_compare): Likewise.
14450 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
14452 (aarch64_lookup_simd_builtin_type): Likewise.
14453 (aarch64_simd_builtin_type): Likewise.
14454 (aarch64_init_simd_builtin_types): Likewise.
14455 (aarch64_simd_expand_args): Likewise.
14456 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
14458 (aarch64_reverse_mask): Likewise.
14459 (aarch64_simd_emit_reg_reg_move): Likewise.
14460 (aarch64_gen_adjusted_ldpstp): Likewise.
14461 (aarch64_ccmp_mode_to_code): Likewise.
14462 (aarch64_operands_ok_for_ldpstp): Likewise.
14463 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
14464 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
14466 (aarch64_min_divisions_for_recip_mul): Likewise.
14467 (aarch64_reassociation_width): Likewise.
14468 (aarch64_get_condition_code_1): Likewise.
14469 (aarch64_simd_emit_reg_reg_move): Likewise.
14470 (aarch64_simd_attr_length_rglist): Likewise.
14471 (aarch64_reverse_mask): Likewise.
14472 (aarch64_operands_ok_for_ldpstp): Likewise.
14473 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
14474 (aarch64_gen_adjusted_ldpstp): Likewise.
14475 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
14477 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
14478 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
14479 (arm_lookup_simd_builtin_type): Likewise.
14480 (arm_simd_builtin_type): Likewise.
14481 (arm_init_simd_builtin_types): Likewise.
14482 (arm_expand_builtin_args): Likewise.
14483 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
14484 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
14485 (ft32_setup_incoming_varargs): Likewise.
14486 (ft32_function_arg): Likewise.
14487 (ft32_function_arg_advance): Likewise.
14488 (ft32_pass_by_reference): Likewise.
14489 (ft32_arg_partial_bytes): Likewise.
14490 (ft32_valid_pointer_mode): Likewise.
14491 (ft32_addr_space_pointer_mode): Likewise.
14492 (ft32_addr_space_legitimate_address_p): Likewise.
14493 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
14495 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
14496 (ix86_emit_outlined_ms2sysv_restore): Likewise.
14497 (iamcu_alignment): Likewise.
14498 (canonicalize_vector_int_perm): Likewise.
14499 (ix86_noce_conversion_profitable_p): Likewise.
14500 (ix86_mpx_bound_mode): Likewise.
14501 (ix86_operands_ok_for_move_multiple): Likewise.
14502 * config/microblaze/microblaze-protos.h
14503 (microblaze_expand_conditional_branch_reg): Likewise.
14504 * config/microblaze/microblaze.c
14505 (microblaze_expand_conditional_branch_reg): Likewise.
14506 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
14508 (rs6000_reassociation_width): Likewise.
14509 (rs6000_invalid_binary_op): Likewise.
14510 (fusion_p9_p): Likewise.
14511 (emit_fusion_p9_load): Likewise.
14512 (emit_fusion_p9_store): Likewise.
14513 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
14515 (riscv_hard_regno_mode_ok_p): Likewise.
14516 (riscv_address_insns): Likewise.
14517 (riscv_split_symbol): Likewise.
14518 (riscv_legitimize_move): Likewise.
14519 (riscv_function_value): Likewise.
14520 (riscv_hard_regno_nregs): Likewise.
14521 (riscv_expand_builtin): Likewise.
14522 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
14523 (riscv_build_integer): Likewise.
14524 (riscv_split_integer): Likewise.
14525 (riscv_legitimate_constant_p): Likewise.
14526 (riscv_cannot_force_const_mem): Likewise.
14527 (riscv_regno_mode_ok_for_base_p): Likewise.
14528 (riscv_valid_base_register_p): Likewise.
14529 (riscv_valid_offset_p): Likewise.
14530 (riscv_valid_lo_sum_p): Likewise.
14531 (riscv_classify_address): Likewise.
14532 (riscv_legitimate_address_p): Likewise.
14533 (riscv_address_insns): Likewise.
14534 (riscv_load_store_insns): Likewise.
14535 (riscv_force_binary): Likewise.
14536 (riscv_split_symbol): Likewise.
14537 (riscv_force_address): Likewise.
14538 (riscv_legitimize_address): Likewise.
14539 (riscv_move_integer): Likewise.
14540 (riscv_legitimize_const_move): Likewise.
14541 (riscv_legitimize_move): Likewise.
14542 (riscv_address_cost): Likewise.
14543 (riscv_subword): Likewise.
14544 (riscv_output_move): Likewise.
14545 (riscv_canonicalize_int_order_test): Likewise.
14546 (riscv_emit_int_order_test): Likewise.
14547 (riscv_function_arg_boundary): Likewise.
14548 (riscv_pass_mode_in_fpr_p): Likewise.
14549 (riscv_pass_fpr_single): Likewise.
14550 (riscv_pass_fpr_pair): Likewise.
14551 (riscv_get_arg_info): Likewise.
14552 (riscv_function_arg): Likewise.
14553 (riscv_function_arg_advance): Likewise.
14554 (riscv_arg_partial_bytes): Likewise.
14555 (riscv_function_value): Likewise.
14556 (riscv_pass_by_reference): Likewise.
14557 (riscv_setup_incoming_varargs): Likewise.
14558 (riscv_print_operand): Likewise.
14559 (riscv_elf_select_rtx_section): Likewise.
14560 (riscv_save_restore_reg): Likewise.
14561 (riscv_for_each_saved_reg): Likewise.
14562 (riscv_register_move_cost): Likewise.
14563 (riscv_hard_regno_mode_ok_p): Likewise.
14564 (riscv_hard_regno_nregs): Likewise.
14565 (riscv_class_max_nregs): Likewise.
14566 (riscv_memory_move_cost): Likewise.
14567 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
14568 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
14569 (rl78_addr_space_address_mode): Likewise.
14570 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14572 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
14573 (rs6000_reassociation_width): Likewise.
14574 (rs6000_invalid_binary_op): Likewise.
14575 (fusion_p9_p): Likewise.
14576 (emit_fusion_p9_load): Likewise.
14577 (emit_fusion_p9_store): Likewise.
14578 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
14579 (ok_for_simple_move_operands): Likewise.
14580 (ok_for_simple_move_strict_operands): Likewise.
14581 (ok_for_simple_arith_logic_operands): Likewise.
14582 (visium_legitimize_reload_address): Likewise.
14583 (visium_select_cc_mode): Likewise.
14584 (output_cbranch): Likewise.
14585 (visium_split_double_move): Likewise.
14586 (visium_expand_copysign): Likewise.
14587 (visium_expand_int_cstore): Likewise.
14588 (visium_expand_fp_cstore): Likewise.
14589 * config/visium/visium.c (visium_pass_by_reference): Likewise.
14590 (visium_function_arg): Likewise.
14591 (visium_function_arg_advance): Likewise.
14592 (visium_libcall_value): Likewise.
14593 (visium_setup_incoming_varargs): Likewise.
14594 (visium_legitimate_constant_p): Likewise.
14595 (visium_legitimate_address_p): Likewise.
14596 (visium_legitimize_address): Likewise.
14597 (visium_secondary_reload): Likewise.
14598 (visium_register_move_cost): Likewise.
14599 (visium_memory_move_cost): Likewise.
14600 (prepare_move_operands): Likewise.
14601 (ok_for_simple_move_operands): Likewise.
14602 (ok_for_simple_move_strict_operands): Likewise.
14603 (ok_for_simple_arith_logic_operands): Likewise.
14604 (visium_function_value_1): Likewise.
14605 (rtx_ok_for_offset_p): Likewise.
14606 (visium_legitimize_reload_address): Likewise.
14607 (visium_split_double_move): Likewise.
14608 (visium_expand_copysign): Likewise.
14609 (visium_expand_int_cstore): Likewise.
14610 (visium_expand_fp_cstore): Likewise.
14611 (visium_split_cstore): Likewise.
14612 (visium_select_cc_mode): Likewise.
14613 (visium_split_cbranch): Likewise.
14614 (output_cbranch): Likewise.
14615 (visium_print_operand_address): Likewise.
14616 * expmed.c (flip_storage_order): Likewise.
14617 * expmed.h (emit_cstore): Likewise.
14618 (flip_storage_order): Likewise.
14619 * genrecog.c (validate_pattern): Likewise.
14620 * hsa-gen.c (gen_hsa_addr): Likewise.
14621 * internal-fn.c (expand_arith_overflow): Likewise.
14622 * ira-color.c (allocno_copy_cost_saving): Likewise.
14623 * lra-assigns.c (find_hard_regno_for_1): Likewise.
14624 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
14625 (process_invariant_for_inheritance): Likewise.
14626 * lra-eliminations.c (move_plus_up): Likewise.
14627 * omp-low.c (lower_oacc_reductions): Likewise.
14628 * simplify-rtx.c (simplify_subreg): Likewise.
14629 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
14630 (TARGET_CHKP_BOUND_MODE): Likewise..
14631 * targhooks.c (default_chkp_bound_mode): Likewise.
14632 (default_setup_incoming_vararg_bounds): Likewise.
14633 * targhooks.h (default_chkp_bound_mode): Likewise.
14634 (default_setup_incoming_vararg_bounds): Likewise.
14635 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
14636 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
14637 (have_whole_vector_shift): Likewise.
14638 * tree-vect-stmts.c (vectorizable_load): Likewise.
14639 * doc/tm.texi: Regenerate.
14641 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
14643 Graceful degrade if Binutils PR21472 is not available.
14646 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
14647 .rodata in flash test fails.
14648 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
14649 * confgure: Regenerate.
14650 * config.in: Regenerate.
14651 * config/avr/avr.c (avr_asm_named_section)
14652 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
14653 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
14654 (avr_asm_init_sections): Same.
14656 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14658 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
14659 (fma<VH:mode>4_intrinsic): Likewise.
14660 (*fmsub<VCVTF:mode>4): Likewise.
14661 (*fmsub<VH:mode>4_intrinsic): Likewise.
14663 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
14666 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
14667 Don't depend on "optimize > 0".
14668 (out_movhi_r_mr, out_movqi_mr_r): Same.
14669 (out_movhi_mr_r, out_movqi_r_mr): Same.
14670 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
14671 io_address_operand on "optimize > 0".
14673 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14675 * tree-loop-distribution.c: Add general explanantion on the pass.
14676 (generate_loops_for_partition): Mark distributed loop.
14677 (pg_add_dependence_edges): New parameter. Handle alias data
14678 dependence specially and record it in the parameter if asked.
14679 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
14680 (init_partition_graph_vertices, add_partition_graph_edge): New.
14681 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
14682 (free_partition_graph_vdata, build_partition_graph): New.
14683 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
14684 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
14685 (data_ref_segment_size, latch_dominated_by_data_ref): New.
14686 (compute_alias_check_pairs, version_loop_by_alias_check): New.
14687 (version_for_distribution_p, finalize_partitions): New.
14688 (distribute_loop): Handle alias data dependence specially. Factor
14689 out loop fusion code as functions and call these functions.
14691 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14693 * tree-loop-distribution.c (classify_partition): New parameter and
14694 better handle reduction statement.
14695 (rdg_build_partitions): Revise comment.
14696 (distribute_loop): Compute statements in all partitions and pass it
14697 to classify_partition.
14699 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14701 * tree-loop-distribution.c (enum partition_type): New.
14702 (struct partition): New field type.
14703 (partition_merge_into): Add parameter. Update partition type.
14704 (data_dep_in_cycle_p, update_type_for_merge): New functions.
14705 (build_rdg_partition_for_vertex): Compute partition type.
14706 (rdg_build_partitions): Dump partition type.
14707 (distribute_loop): Update calls to partition_merge_into.
14709 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14711 * tree-loop-distribution.c (struct ddr_hasher): New.
14712 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
14714 (classify_partition): Call get_data_dependence.
14715 (pg_add_dependence_edges): Ditto.
14716 (distribute_loop): Release data dependence hash table.
14718 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14720 * tree-loop-distribution.c (ref_base_address): Delete.
14721 (similar_memory_accesses): Rename ...
14722 (share_memory_accesses): ... to this. Check if partitions access
14723 the same memory reference.
14724 (distribute_loop): Call share_memory_accesses.
14726 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14728 * tree-loop-distribution.c (struct partition): New field recording
14729 its data reference.
14730 (partition_alloc, partition_free): Init and release data refs.
14731 (partition_merge_into): Merge data refs.
14732 (build_rdg_partition_for_vertex): Collect data refs for partition.
14733 (pg_add_dependence_edges): Change parameters from vector to bitmap.
14735 (distribute_loop): Remve data refs from vertice data of partition
14738 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14740 * tree-loop-distribution.c (params.h): Include header file.
14741 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
14742 (datarefs_vec): New global var.
14743 (create_rdg_vertices): Use datarefs_vec directly.
14744 (free_rdg): Don't free data references.
14745 (build_rdg): Update use. Don't free data references.
14746 (distribute_loop): Compute global variable for data references.
14747 Bail out if there are too many data references.
14749 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14751 * tree-loop-distribution.c (loop_nest): New global var.
14752 (build_rdg): Use loop directly, rather than loop nest.
14753 (pg_add_dependence_edges): Remove loop nest parameter. Use global
14755 (distribute_loop): Compute global variable loop nest. Update use.
14757 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14759 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
14760 (partition_merge_into): New parameter. Dump reason for fusion.
14761 (distribute_loop): Update use of partition_merge_into.
14763 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14765 * tree-loop-distribution.c (bb_top_order_index): New.
14766 (bb_top_order_index_size, bb_top_order_cmp): New.
14767 (stmts_from_loop): Use topological order.
14768 (pass_loop_distribution::execute): Compute and release topological
14769 order for basic blocks.
14771 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14773 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
14776 2017-07-05 Bin Cheng <bin.cheng@arm.com>
14778 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
14779 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
14780 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
14781 * internal-fn.def (LOOP_DIST_ALIAS): New.
14782 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
14783 (fold_loop_internal_call): ... this.
14784 (vect_loop_dist_alias_call): New function.
14785 (set_uid_loop_bbs): Call fold_loop_internal_call.
14786 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
14789 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
14792 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
14793 Require dead FLAGS_REG at the beginning of a peephole.
14795 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
14798 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
14799 arguments in the call to __builtin_ia32_sbb_u32.
14800 (_subborrow_u64): Swap _X and _Y arguments in the call to
14801 __builtin_ia32_sbb_u64.
14803 2017-07-04 Jakub Jelinek <jakub@redhat.com>
14806 * tree-vrp.c (compare_assert_loc): Turn into a function template
14807 with stable template parameter. Only test if a->e is NULL,
14808 !a->e == !b->e has been verified already. Use e == NULL or
14809 e != NULL instead of e or ! e tests. If stable is true, don't use
14810 iterative_hash_expr, on the other side allow a or b or both NULL
14811 and sort the NULLs last.
14812 (process_assert_insertions): Sort using compare_assert_loc<false>
14813 instead of compare_assert_loc, later sort using
14814 compare_assert_loc<true> before calling process_assert_insertions_for
14815 in a loop. Use break instead of continue once seen NULL pointer.
14817 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
14819 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
14820 Cortex-R7 and Cortex-R8 processors.
14822 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
14824 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
14825 uninitialized while src is not.
14827 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
14829 * common/config/arm/arm-common.c: Adjust include path for
14831 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
14832 (arm-cpu.h): Create in build directory. Adjust dependency rules.
14833 (arm-cpu-data.h): Likewise.
14834 (arm-cpu-cdata.h): Likewise.
14835 * config/arm/arm-cpu.h: Delete.
14836 * config/arm/arm-cpu-cdata.h: Delete.
14837 * config/arm/arm-cpu-data.h: Delete.
14839 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
14841 * config/arm/arm-cpus.in (cortex-a55): New.
14842 (cortex-a75): Likewise.
14843 (cortex-a75.cortex-a55): Likewise.
14844 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
14846 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
14847 * config/arm/arm-cpu-cdata.h: Regenerate.
14848 * config/arm/arm-cpu-data.h: Regenerate.
14849 * config/arm/arm-cpu.h: Regenerate.
14850 * config/arm/arm-tables.opt: Regenerate.
14851 * config/arm/arm-tune.md: Regenerate.
14853 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
14855 * haifa-sched.c (sched_create_recovery_edges): Update profile.
14857 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
14859 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
14862 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
14864 PR tree-optimization/81292
14865 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
14866 full_string_p, also call adjust_related_strinfos if the adjustment
14867 is simple, otherwise invalidate related strinfos.
14869 2017-07-04 Martin Liska <mliska@suse.cz>
14872 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
14873 newly created variable as DECL_IGNORED_P.
14875 2017-07-04 Martin Liska <mliska@suse.cz>
14878 * ipa-inline.c (inline_small_functions):
14879 Use xstrdup_for_dump.
14881 2017-07-04 Tom de Vries <tom@codesourcery.com>
14883 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
14885 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
14888 * config/darwin.c (darwin_function_switched_text_sections):
14891 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
14893 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
14895 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
14897 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
14899 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14901 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
14902 min_profitable_iters, and th as inclusive lower bounds.
14903 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
14904 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
14905 for min_profitable_iters and min_profitable_estimate.
14906 (vect_transform_loop): Treat th as an inclusive lower bound.
14907 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
14909 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
14912 * config/darwin.c (darwin_function_switched_text_sections):
14913 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
14914 in two pieces, and suppress the use of buf.
14916 2017-07-03 Nathan Sidwell <nathan@acm.org>
14918 * hash-table.h (hash_table_mod1): Fix indentation.
14920 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14922 PR middle-end/81290
14923 * predict.c (force_edge_cold): Be more careful about propagation
14925 * profile-count.h (profile_probability::guessed,
14926 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
14928 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
14930 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
14932 * doc/invoke.texi (rcpc architecture extension): Document it.
14934 2017-07-03 Richard Biener <rguenther@suse.de>
14936 PR tree-optimization/60510
14937 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
14938 the scalar reduction PHI and use it.
14939 (vectorizable_reduction): Properly guard the single_defuse_cycle
14940 path for non-SLP reduction chains where we cannot use it.
14941 Rework reduc_def/index and vector type deduction. Rework
14942 vector operand gathering during reduction op code-gen.
14943 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
14944 chains dissolve the chain and leave it to non-SLP reduction
14947 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14949 * tree-data-ref.h (dr_alignment): Declare.
14950 * tree-data-ref.c (dr_alignment): New function.
14951 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
14952 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
14954 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
14956 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14958 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
14959 and base_misalignment fields.
14960 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
14961 * tree-data-ref.c: Include builtins.h.
14962 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
14963 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
14964 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
14965 * tree-vect-data-refs.c: Include tree-cfg.h.
14966 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
14967 fields instead of calculating an alignment here.
14968 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
14969 innermost_loop_behavior fields.
14971 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14973 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
14975 (DR_STEP_ALIGNMENT): New macro.
14976 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
14977 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
14978 (create_data_ref): Print it.
14979 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
14980 to tell whether the step preserves vector (mis)alignment.
14981 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
14982 Move the check for an integer step and generalise to all INTEGER_CST.
14983 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
14984 Print the outer step alignment.
14986 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
14988 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
14989 with offset_alignment.
14990 (DR_ALIGNED_TO): Delete.
14991 (DR_OFFSET_ALIGNMENT): New macro.
14992 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
14993 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
14994 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
14995 (create_data_ref): Likewise.
14996 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
14997 (vect_analyze_data_refs): Likewise.
14998 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
14999 creating dummy innermost behavior.
15001 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15003 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
15004 with a "innermost_loop_behavior *" and refeence tree.
15005 * tree-data-ref.c (dr_analyze_innermost): Likewise.
15006 (create_data_ref): Update call accordingly.
15007 * tree-predcom.c (find_looparound_phi): Likewise.
15009 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15011 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
15012 fields with dr_wrt_vec_loop.
15013 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
15014 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
15015 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
15016 (vect_dr_behavior): New function.
15017 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
15018 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
15019 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
15020 track whether the step preserves the misalignment.
15021 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
15022 Use vect_dr_behavior.
15023 (vect_setup_realignment): Update call accordingly.
15024 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
15025 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
15026 call to vect_create_addr_base_for_vector_ref.
15027 (vect_create_cond_for_align_checks): Likewise.
15028 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
15029 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
15030 (vect_recog_mask_conversion_pattern): Likewise.
15031 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
15032 (new_stmt_vec_info): Remove redundant zeroing.
15034 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
15036 * common/config/arm/arm-common.c (arm_be8_option): New function.
15037 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
15038 (ISA_ARMv6): Add isa_bit_be8.
15039 * config/arm/arm.h (arm_be8_option): Add prototype.
15040 (BE8_SPEC_FUNCTION): New define.
15041 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
15042 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
15043 (mlittle-endian): Similarly.
15044 (mbe8, mbe32): New options.
15045 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
15046 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
15048 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15050 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
15052 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15054 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
15055 (cleanup_tree_cfg_bb): Use it.
15056 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
15058 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
15060 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15063 * loop-doloop.c (add_test): Update profile.
15065 2017-07-03 Martin Liska <mliska@suse.cz>
15068 * sanopt.c (rewrite_usage_of_param): New function.
15069 (sanitize_rewrite_addressable_params): Likewise.
15070 (pass_sanopt::execute): Call rewrite_usage_of_param.
15072 2017-07-03 Richard Biener <rguenther@suse.de>
15074 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
15075 back to using VIEW_CONVERT_EXPR.
15077 2017-07-03 Martin Liska <mliska@suse.cz>
15080 * doc/extend.texi: Document when a resolver function is
15081 generated for target_clones.
15083 2017-07-03 Martin Liska <mliska@suse.cz>
15085 * asan.c (asan_emit_stack_protection): Unpoison just red zones
15086 and shadow memory of auto variables which are subject of
15087 use-after-scope sanitization.
15088 (asan_expand_mark_ifn): Add do set only when is_poison.
15090 2016-07-03 Richard Biener <rguenther@suse.de>
15092 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
15094 (vect_force_simple_reduction): Record reduction def -> phi mapping.
15095 (vectorizable_reduction): Perform reduction PHI creation when
15096 visiting a reduction PHI and adjust and simplify code generation
15097 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
15098 (vect_transform_loop): Visit reduction PHIs.
15099 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
15100 defs into the SLP tree.
15101 (vect_build_slp_tree): Reduction defs terminate the recursion.
15102 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
15104 (vect_get_vec_defs_for_stmt_copy): Export.
15105 (vect_get_vec_defs): Likewise.
15106 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
15108 (vect_get_vec_defs_for_stmt_copy): Declare.
15109 (vect_get_vec_defs): Likewise.
15111 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15113 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
15114 parameter with a "loop" parameter and use it instead of the
15115 loop containing DR_STMT. Don't check simple_iv when doing
15116 BB analysis. Describe the two analysis modes in the comment.
15118 2017-07-03 Tom de Vries <tom@codesourcery.com>
15120 PR tree-optimization/69468
15121 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
15122 (find_same_succ_bb): Handle ignore_edge_flags.
15124 2017-07-03 Tom de Vries <tom@codesourcery.com>
15126 PR tree-optimization/81192
15127 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
15129 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
15131 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
15133 2017-07-03 Tom de Vries <tom@codesourcery.com>
15135 PR tree-optimization/81192
15136 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
15137 BB_SAME_SUCC (bb) == NULL.
15139 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15141 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
15144 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15146 * dumpfile.c: Include profile-count.h
15147 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
15149 (insert_cond_bb): Update profile.
15150 * tree-cfg.h (insert_cond_bb): Update prototype.
15151 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
15152 * tree-dump.c: Do not include tree-cfg.
15154 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15156 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
15158 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15160 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
15163 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15165 * tree-complex.c (expand_complex_div_wide): update profile.
15167 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
15168 Alan Hayward <alan.hayward@arm.com>
15169 David Sherwood <david.sherwood@arm.com>
15171 * Makefile.in (MACHMODE_H): Remove insn-modes.h
15172 (CORETYPES_H): New define.
15173 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
15174 (insn-modes-inline.h, s-modes-inline-h): New rules.
15175 (generated_files): Add insn-modes-inline.h.
15176 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
15177 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
15178 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
15179 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
15180 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
15181 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
15182 (build/gencodes.o, build/genconditions.o): Likewise.
15183 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
15184 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
15185 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
15186 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
15187 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
15188 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
15189 * coretypes.h: Include everything up to real.h for generators.
15190 Include insn-modes.h first. Include wide-int-print.h after
15191 wide-int.h. Include insn-modes-inline.h and then machmode.h.
15192 * machmode.h: Don't include insn-modes.h here.
15193 * function-tests.c: Remove includes of signop.h, machmode.h,
15194 double-int.h and wide-int.h.
15196 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
15198 * optc-save-gen.awk: Likewise.
15199 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
15200 * godump.c: Remove include of wide-int-print.h.
15201 * pretty-print.h: Likewise.
15202 * wide-int-print.cc: Likewise.
15203 * wide-int.cc: Likewise.
15204 * hash-map-tests.c: Remove include of signop.h.
15205 * hash-set-tests.c: Likewise.
15206 * rtl-tests.c: Likewise.
15207 * mkconfig.sh: Remove include of machmode.h.
15208 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
15210 (emit_insn_modes_inline_h): ...this new function. Emit the code
15211 into an insn-modes-inline.h header file, adding appropriate
15212 include guards and end comments.
15213 (emit_insn_modes_c_header): Remove include of machmode.h.
15214 (emit_min_insn_modes_c_header): Include coretypes.h rather than
15216 (main): Handle -i flag and call emit_insn_modes_inline_h when
15219 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
15221 * tree-ssa-strlen.c (strinfo): Rename the length field to
15222 nonzero_chars. Add a full_string_p field.
15223 (compare_nonzero_chars, zero_length_string_p): New functions.
15224 (get_addr_stridx): Add an offset_out parameter.
15225 Use compare_nonzero_chars.
15226 (get_stridx): Update accordingly. Use compare_nonzero_chars.
15227 (new_strinfo): Update after above changes to strinfo.
15228 (set_endptr_and_length): Set full_string_p.
15229 (get_string_length): Update after above changes to strinfo.
15230 (unshare_strinfo): Update call to new_strinfo.
15231 (maybe_invalidate): Likewise.
15232 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
15233 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
15234 as a uhwi instead of an shwi. Update after above changes to
15235 strinfo and new_strinfo.
15236 (zero_length_string): Assert that chainsi contains full strings.
15237 Use zero_length_string_p. Update call to new_strinfo.
15238 (adjust_related_strinfos): Update after above changes to strinfo.
15239 Copy full_string_p from origsi.
15240 (adjust_last_stmt): Use zero_length_string_p.
15241 (handle_builtin_strlen): Update after above changes to strinfo and
15242 new_strinfo. Install the lhs as the string length if the previous
15243 entry didn't describe a full string.
15244 (handle_builtin_strchr): Update after above changes to strinfo
15246 (handle_builtin_strcpy): Likewise.
15247 (handle_builtin_strcat): Likewise.
15248 (handle_builtin_malloc): Likewise.
15249 (handle_pointer_plus): Likewise.
15250 (handle_builtin_memcpy): Likewise. Track nonzero characters
15251 that aren't necessarily followed by a nul terminator.
15252 (handle_char_store): Likewise.
15254 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
15256 PR tree-optimization/80769
15257 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
15258 for malloc and calloc. Document the new invariant that all related
15259 strinfos have delayed lengths or none do.
15260 (verify_related_strinfos): Move earlier in file.
15261 (set_endptr_and_length): New function, split out from...
15262 (get_string_length): ...here. Also set the lengths of related
15264 (zero_length_string): Assert that chainsi has known (rather than
15266 (adjust_related_strinfos): Likewise.
15268 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
15270 PR tree-optimization/81136
15271 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
15272 assert that two references with the same misalignment have the same
15273 compile-time misalignment if those compile-time misalignments
15276 2017-07-01 Andi Kleen <ak@linux.intel.com>
15278 * print-tree.c (print_node): Print all attributes.
15280 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15282 * cfg.c (scale_bbs_frequencies): New function.
15283 * cfg.h (scale_bbs_frequencies): Declare it.
15284 * cfgloopanal.c (single_likely_exit): Cleanup.
15285 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
15287 (scale_loop_profile): Likewise.
15288 (loop_version): Likewise.
15289 (create_empty_loop_on_edge): Update.
15290 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
15291 scale_loop_frequencies, scale_loop_profile, loopify,
15292 loop_version): Update prototypes.
15293 * modulo-sched.c (sms_schedule): Update.
15294 * predict.c (unlikely_executed_edge_p): Also check probability.
15295 (probably_never_executed_edge_p): Fix typo.
15296 * tree-if-conv.c (version_loop_for_if_conversion): Update.
15297 * tree-parloops.c (gen_parallel_loop): Update.
15298 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
15299 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
15300 * tree-ssa-loop-split.c (split_loop): Update.
15301 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
15302 * tree-vect-loop-manip.c (vect_do_peeling): Update.
15303 (vect_loop_versioning): Update.
15304 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
15306 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15308 * trans-mem.c (split_bb_make_tm_edge): Update profile.
15310 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15312 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
15313 to keep profile consistent.
15315 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15317 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
15318 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
15319 * profile-count.h (max_safe_multiplier): Make unsigned.
15320 (profile_count::guessed_zero): New.
15322 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15324 * bb-reorder.c (fix_up_crossing_landing_pad,
15325 fix_crossing_conditional_branches): Use make_single_succ_edge
15326 to keep profile consistent.
15328 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15330 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
15333 2017-07-01 Jakub Jelinek <jakub@redhat.com>
15336 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
15337 the right scopes, make sure cond_jump isn't preserved between multiple
15338 iterations. Search for fallthru edge whenever there are 3+ edges and
15339 use find_fallthru_edge for it.
15341 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15343 Patch by Alexander Monakov <amonakov@ispras.ru>
15344 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
15345 probabilities consistently.
15347 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15349 * pa.c (pa_expand_compare_and_swap_loop): Update call of
15350 emit_cmp_and_jump_insns.
15352 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15355 * tree-inline.c (expand_call_inline): Combine profile statuses.
15357 2017-06-30 Andrew Pinski <apinski@cavium.com>
15359 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
15360 fold_stmt returned true.
15362 2017-06-30 Nathan Sidwell <nathan@acm.org>
15364 * ggc.h (empty_string): Delete.
15365 * cfgexpand.c (expand_asm_stmt): Use plain "".
15366 * optabs.c (expand_asm_memory_barrier): Likewise.
15367 * stringpool.c (empty_string): Delete.
15368 (digit_vector, digit_string): Delete.
15369 (ggc_alloc_string): Use plain "", don't optimize single digit
15370 strings. Use ggc_alloc_atomic.
15372 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
15374 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
15375 comparison set and one other set, use the cost of the non-comparison
15378 2017-06-30 Nathan Sidwell <nathan@acm.org>
15380 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
15383 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
15385 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
15386 loops. Remove now unneeded calls to gimple_switch_set_label() that
15387 just set removed labels to NULL_TREE.
15389 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
15391 * tree-ssanames.c (set_range_info_raw): Abstract from ...
15392 (set_range_info): ...here. Only call set_range_info_raw if domain
15394 (set_nonzero_bits): Call set_range_info_raw.
15395 * tree-ssanames.h (set_range_info_raw): New.
15397 2017-06-30 Jakub Jelinek <jakub@redhat.com>
15400 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
15401 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
15402 of nonimmediate_operand and <store_mask_constraint> instead of m
15403 for the input operand. For V8FI iterator, always split if input
15404 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
15405 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
15406 <store_mask_predicate> instead of register_operand and
15407 <store_mask_constraint> instead of v for the input operand. Make
15408 sure both operands aren't MEMs for if not <mask_applied>.
15410 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
15412 * lto-wrapper.c (copy_file) Close both file descriptors before
15415 2017-06-30 Martin Liska <mliska@suse.cz>
15418 * multiple_target.c (create_dispatcher_calls): Make ifunc
15419 also for function that don't have calls or are not referenced.
15421 2017-06-30 Richard Biener <rguenther@suse.de>
15423 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
15424 analyze the first scalar stmt. Move vector type computation
15425 for the BB case here from ...
15426 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
15427 live operation processing in the SLP case properly.
15429 2017-06-30 Richard Biener <rguenther@suse.de>
15431 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
15433 2017-06-30 Martin Liska <mliska@suse.cz>
15436 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
15437 before BUILT_IN_UNWIND_RESUME when ASAN is used.
15439 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
15441 * doc/invoke.texi (AArch64): Add missing options and remove redundant
15444 2017-06-30 Richard Biener <rguenther@suse.de>
15446 PR tree-optimization/81249
15447 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
15448 condition reduction result to original scalar type.
15450 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15452 * profile-count.h (enum profile_quality): Fix typos and whitespace
15455 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15457 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
15458 type for branch probabilities.
15460 2017-06-29 Julian Brown <julian@codesourcery.com>
15461 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15463 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
15464 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
15465 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
15466 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
15468 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15470 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
15471 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
15472 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
15473 CC usage from generic code to here.
15474 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
15475 CC usage into the target macros.
15477 2017-06-29 Maya Rashish <coypu@sdf.org>
15479 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
15482 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15484 * arm/arm-builtins.c: Include profile-count.h
15485 * except.c (sjlj_emit_function_enter): Use
15486 profile_probability::unlikely.
15488 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15490 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
15491 and tocrel_offset be pointer args rather than implicitly using
15493 (legitimate_constant_pool_address_p, rs6000_emit_move,
15494 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
15495 tocrel_offset and use in toc_relative_expr_p call.
15496 (print_operand, print_operand_address): Use static tocrel_base_oac
15497 and tocrel_offset_oac.
15498 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
15501 2017-06-29 Maya Rashish <coypu@sdf.org>
15503 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
15505 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
15507 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
15508 objects, take into account only the alignment of 'op0' and 'mode1' if
15511 2017-06-29 Steve Ellcey <sellcey@cavium.com>
15513 * ccmp.c (ccmp_tree_comparison_p): New function.
15514 (ccmp_candidate_p): Update to use above function.
15515 (get_compare_parts): New function.
15516 (expand_ccmp_next): Update to use new functions.
15517 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
15519 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
15520 take mode as argument.
15521 * ccmp.h (expand_ccmp_expr): Add mode as argument.
15522 * expr.c (expand_expr_real_1): Pass mode as argument.
15524 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
15526 * combine.c (combine_instructions): Print insns to dump_file, together
15529 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15531 * asan.c (asan_emit_stack_protection): Update.
15532 (create_cond_insert_point): Update.
15533 * auto-profile.c (afdo_propagate_circuit): Update.
15534 * basic-block.h (struct edge_def): Turn probability to
15535 profile_probability.
15536 (EDGE_FREQUENCY): Update.
15537 * bb-reorder.c (find_traces_1_round): Update.
15538 (better_edge_p): Update.
15539 (sanitize_hot_paths): Update.
15540 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
15541 (make_single_succ_edge): Update.
15542 (check_bb_profile): Update.
15543 (dump_edge_info): Update.
15544 (update_bb_profile_for_threading): Update.
15545 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
15546 probabilitycount to 0.
15547 * cfgbuild.c (compute_outgoing_frequencies): Update.
15548 * cfgcleanup.c (try_forward_edges): Update.
15549 (outgoing_edges_match): Update.
15550 (try_crossjump_to_edge): Update.
15551 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
15552 (expand_gimple_tailcall): Update.
15553 (construct_init_block): Use make_single_succ_edge.
15554 (construct_exit_block): Use make_single_succ_edge.
15555 * cfghooks.c (verify_flow_info): Update.
15556 (redirect_edge_succ_nodup): Update.
15557 (split_edge): Update.
15558 (account_profile_record): Update.
15559 * cfgloopanal.c (single_likely_exit): Update.
15560 * cfgloopmanip.c (scale_loop_profile): Update.
15561 (set_zero_probability): Remove.
15562 (duplicate_loop_to_header_edge): Update.
15563 * cfgloopmanip.h (loop_version): Update prototype.
15564 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
15565 (force_nonfallthru_and_redirect): Update.
15566 (update_br_prob_note): Update.
15567 (rtl_verify_edges): Update.
15568 (purge_dead_edges): Update.
15569 (rtl_lv_add_condition_to_bb): Update.
15570 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
15571 * cgraphunit.c (init_lowered_empty_function): Update.
15572 (cgraph_node::expand_thunk): Update.
15573 * cilk-common.c: Include profile-count.h
15574 * dojump.c (inv): Remove.
15575 (jumpifnot): Update.
15576 (jumpifnot_1): Update.
15577 (do_jump_1): Update.
15579 (do_jump_by_parts_greater_rtx): Update.
15580 (do_compare_rtx_and_jump): Update.
15581 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
15582 do_jump_1. do_compare_rtx_and_jump): Update prototype.
15583 * dwarf2cfi.c: Include profile-count.h
15584 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
15585 (sjlj_emit_dispatch_table): Likewise.
15586 * explow.c: Include profile-count.h
15587 * expmed.c (emit_store_flag_force): Update.
15588 (do_cmp_and_jump): Update.
15589 * expr.c (compare_by_pieces_d::generate): Update.
15590 (compare_by_pieces_d::finish_mode): Update.
15591 (emit_block_move_via_loop): Update.
15592 (store_expr_with_bounds): Update.
15593 (store_constructor): Update.
15594 (expand_expr_real_2): Update.
15595 (expand_expr_real_1): Update.
15596 * expr.h (try_casesi, try_tablejump): Update prototypes.
15597 * gimple-pretty-print.c (dump_probability): Update.
15598 (dump_profile): New.
15599 (dump_gimple_label): Update.
15600 (dump_gimple_bb_header): Update.
15601 * graph.c (draw_cfg_node_succ_edges): Update.
15602 * hsa-gen.c (convert_switch_statements): Update.
15603 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
15604 (find_if_case_1): Update.
15605 (find_if_case_2): Update.
15606 * internal-fn.c (expand_arith_overflow_result_store): Update.
15607 (expand_addsub_overflow): Update.
15608 (expand_neg_overflow): Update.
15609 (expand_mul_overflow): Update.
15610 (expand_vector_ubsan_overflow): Update.
15611 * ipa-cp.c (good_cloning_opportunity_p): Update.
15612 * ipa-split.c (split_function): Use make_single_succ_edge.
15613 * ipa-utils.c (ipa_merge_profiles): Update.
15614 * loop-doloop.c (add_test): Update.
15615 (doloop_modify): Update.
15616 * loop-unroll.c (compare_and_jump_seq): Update.
15617 (unroll_loop_runtime_iterations): Update.
15618 * lra-constraints.c (lra_inheritance): Update.
15619 * lto-streamer-in.c (input_cfg): Update.
15620 * lto-streamer-out.c (output_cfg): Update.
15621 * mcf.c (adjust_cfg_counts): Update.
15622 * modulo-sched.c (sms_schedule): Update.
15623 * omp-expand.c (expand_omp_for_init_counts): Update.
15624 (extract_omp_for_update_vars): Update.
15625 (expand_omp_ordered_sink): Update.
15626 (expand_omp_for_ordered_loops): Update.
15627 (expand_omp_for_generic): Update.
15628 (expand_omp_for_static_nochunk): Update.
15629 (expand_omp_for_static_chunk): Update.
15630 (expand_cilk_for): Update.
15631 (expand_omp_simd): Update.
15632 (expand_omp_taskloop_for_outer): Update.
15633 (expand_omp_taskloop_for_inner): Update.
15634 * omp-simd-clone.c (simd_clone_adjust): Update.
15635 * optabs.c (expand_doubleword_shift): Update.
15636 (expand_abs): Update.
15637 (emit_cmp_and_jump_insn_1): Update.
15638 (expand_compare_and_swap_loop): Update.
15639 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
15640 * predict.c (predictable_edge_p): Update.
15641 (edge_probability_reliable_p): Update.
15642 (set_even_probabilities): Update.
15643 (combine_predictions_for_insn): Update.
15644 (combine_predictions_for_bb): Update.
15645 (propagate_freq): Update.
15646 (estimate_bb_frequencies): Update.
15647 (force_edge_cold): Update.
15648 * profile-count.c (profile_count::dump): Add missing space into dump.
15649 (profile_count::debug): Add newline.
15650 (profile_count::differs_from_p): Explicitly convert to unsigned.
15651 (profile_count::stream_in): Update.
15652 (profile_probability::dump): New member function.
15653 (profile_probability::debug): New member function.
15654 (profile_probability::differs_from_p): New member function.
15655 (profile_probability::differs_lot_from_p): New member function.
15656 (profile_probability::stream_in): New member function.
15657 (profile_probability::stream_out): New member function.
15658 * profile-count.h (profile_count_quality): Rename to ...
15659 (profile_quality): ... this one.
15660 (profile_probability): New.
15661 (profile_count): Update.
15662 * profile.c (compute_branch_probabilities): Update.
15663 * recog.c (peep2_attempt): Update.
15664 * sched-ebb.c (schedule_ebbs): Update.
15665 * sched-rgn.c (find_single_block_region): Update.
15666 (compute_dom_prob_ps): Update.
15667 (schedule_region): Update.
15668 * sel-sched-ir.c (compute_succs_info): Update.
15669 * stmt.c (struct case_node): Update.
15670 (do_jump_if_equal): Update.
15671 (get_outgoing_edge_probs): Update.
15672 (conditional_probability): Update.
15673 (emit_case_dispatch_table): Update.
15674 (expand_case): Update.
15675 (expand_sjlj_dispatch_table): Update.
15676 (emit_case_nodes): Update.
15677 * targhooks.c: Update.
15678 * tracer.c (better_p): Update.
15679 (find_best_successor): Update.
15680 * trans-mem.c (expand_transaction): Update.
15681 * tree-call-cdce.c: Update.
15682 * tree-cfg.c (gimple_split_edge): Upate.
15683 (move_sese_region_to_fn): Upate.
15684 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
15685 * tree-eh.c (lower_resx): Upate.
15686 (cleanup_empty_eh_move_lp): Upate.
15687 * tree-if-conv.c (version_loop_for_if_conversion): Update.
15688 * tree-inline.c (copy_edges_for_bb): Update.
15689 (copy_cfg_body): Update.
15690 * tree-parloops.c (gen_parallel_loop): Update.
15691 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
15692 (gimple_gen_time_profiler): Update.
15693 * tree-ssa-dce.c (remove_dead_stmt): Update.
15694 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
15695 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
15696 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
15697 (unloop_loops): Update.
15698 (try_peel_loop): Update.
15699 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
15700 * tree-ssa-loop-split.c (connect_loops): Update.
15701 (split_loop): Update.
15702 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
15703 (hoist_guard): Update.
15704 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
15705 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
15706 (value_replacement): Update.
15707 * tree-ssa-reassoc.c (branch_fixup): Update.
15708 * tree-ssa-tail-merge.c (replace_block_by): Update.
15709 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
15710 (create_edge_and_update_destination_phis): Update.
15711 (compute_path_counts): Update.
15712 (recompute_probabilities): Update.
15713 (update_joiner_offpath_counts): Update.
15714 (freqs_to_counts_path): Update.
15715 (duplicate_thread_path): Update.
15716 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
15717 (struct switch_conv_info): Update.
15718 (gen_inbound_check): Update.
15719 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
15720 (vect_do_peeling): Update.
15721 (vect_loop_versioning): Update.
15722 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
15723 (optimize_mask_stores): Update.
15724 * ubsan.c (ubsan_expand_null_ifn): Update.
15725 * value-prof.c (gimple_divmod_fixed_value): Update.
15726 (gimple_divmod_fixed_value_transform): Update.
15727 (gimple_mod_pow2): Update.
15728 (gimple_mod_pow2_value_transform): Update.
15729 (gimple_mod_subtract): Update.
15730 (gimple_mod_subtract_transform): Update.
15731 (gimple_ic): Update.
15732 (gimple_stringop_fixed_value): Update.
15733 (gimple_stringops_transform): Update.
15734 * value-prof.h: Update.
15736 2017-06-29 Carl Love <cel@us.ibm.com>
15738 * config/rs6000/rs6000-c.c: Add support for built-in functions
15739 vector signed int vec_signed (vector float);
15740 vector signed long long vec_signed (vector double);
15741 vector signed int vec_signed2 (vector double, vector double);
15742 vector signed int vec_signede (vector double);
15743 vector signed int vec_signedo (vector double);
15744 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
15745 instruction generator.
15746 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
15747 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
15748 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
15750 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
15751 vunsignede_v2df): Add define_expands.
15752 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
15753 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
15754 VEC_UNSIGNEDO): Add definitions.
15755 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
15756 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
15757 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
15758 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
15759 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
15760 * config/rs6000/altivec.h (vec_signed, vec_signed2,
15761 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
15762 vec_unsignede, vec_unsignedo): Add builtin defines.
15763 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
15765 * doc/extend.texi: Update the built-in documentation file for the
15766 new built-in functions.
15768 2017-06-29 Richard Biener <rguenther@suse.de>
15770 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
15771 reduction chains to LOOP_VINFO_REDUCTIONS.
15772 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
15773 SLP reductions after processing reduction chains.
15775 2017-06-29 Nathan Sidwell <nathan@acm.org>
15777 * builtins.c (fold_builtin_FUNCTION): Use
15778 lang_hooks.decl_printable_name.
15780 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
15782 PR middle-end/81194
15783 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
15784 with only one label.
15785 * stmt.c (expand_case): Assert NCASES is greater than one.
15787 2017-06-29 Richard Biener <rguenther@suse.de>
15789 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
15791 (group_case_labels): Likewise.
15792 (find_taken_edge): Push sanity checking on val to workers...
15793 (find_taken_edge_cond_expr): ... here
15794 (find_taken_edge_switch_expr): ... and here, handle cases
15795 with just a default label.
15796 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
15797 (group_case_labels): Likewise.
15798 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
15799 group_case_labels does anything cleanup the CFG again.
15801 2017-06-29 Bin Cheng <bin.cheng@arm.com>
15803 PR tree-optimization/81196
15804 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
15805 exit condition comparing two IVs.
15807 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
15809 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
15810 profile to the dummy entry at the end of the list of architectures.
15811 * config/arm/arm-cpu-cdata.h: Regenerated.
15813 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15814 Michael Collison <michael.collison@arm.com>
15817 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
15819 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
15820 (*aarch64_reg_<mode>3_minus_mask): New pattern.
15821 (*aarch64_<optab>_reg_di3_mask2): New pattern.
15822 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
15823 of shift when the shift amount is masked with constant equal to
15824 the size of the mode.
15825 * config/aarch64/predicates.md (subreg_lowpart_operator): New
15828 2017-06-29 Martin Liska <mliska@suse.cz>
15830 * config/i386/i386.opt: Change range from [1,5] to [0,5].
15832 2017-06-29 Yury Gribov <tetra2005@gmail.com>
15835 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
15838 (edge_growth_cache_entry::edge_growth_cache_entry): New
15840 (reset_edge_growth_cache): Update to use constructor.
15842 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15844 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
15845 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
15846 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
15848 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
15850 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
15851 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
15853 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
15855 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
15856 (*-linux-uclibc*): Add t-uclibc tmake_file.
15857 * config/t-musl: New.
15858 * config/t-uclibc: New.
15860 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
15862 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
15864 (gen_comm_data): Emit architectural setting of arch_prof.
15865 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
15867 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
15868 (armv8-m.base, armv8-m.main): Likewise.
15869 * arm-protos.h (arm_build_target): Add profile field.
15870 (arch_option): Likewise.
15871 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
15873 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
15874 arm_active_target.profile.
15876 2017-06-28 Richard Biener <rguenther@suse.de>
15878 PR middle-end/81227
15879 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
15880 TYPE_OVERFLOW_WRAPS.
15881 * match.pd (negate_expr_p): Likewise.
15882 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
15883 fold_build2, not fold_binary.
15885 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
15887 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
15888 Convert memory address to Pmode.
15889 (aarch64_print_operand): Assert MEM operands are always Pmode.
15891 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
15894 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
15895 Remove redundant if.
15896 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
15897 * config/arm/aarch-common-protos.h
15898 (aarch_forward_to_shift_is_not_shifted_re): Remove.
15899 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
15901 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
15904 * multiple_target.c (create_dispatcher_calls): Set the default
15905 clone to be static, not public.
15907 2017-06-28 Richard Biener <rguenther@suse.de>
15909 * tree-vect-loop.c (vectorizable_reduction): Move special
15910 cond reduction IV var creation ...
15911 (vect_create_epilog_for_reduction): ... here. Remove induction_index
15912 parameter. Use STMT_VINFO_VECTYPE.
15913 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
15916 2017-06-28 Martin Liska <mliska@suse.cz>
15919 * ipa-visibility.c (non_local_p): Handle visibility.
15921 2017-06-28 Martin Liska <mliska@suse.cz>
15924 * common.opt: Add IntegerRange to various options.
15925 * opt-functions.awk (integer_range_info): New function.
15926 * optc-gen.awk: Add integer_range_info to cl_options struct.
15927 * opts-common.c (decode_cmdline_option): Handle
15928 CL_ERR_INT_RANGE_ARG.
15929 (cmdline_handle_error): Likewise.
15930 * opts.c (print_filtered_help): Show valid interval in
15931 when --help is provided.
15932 * opts.h (struct cl_option): Add range_min and range_max fields.
15933 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
15935 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
15937 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
15938 (x * C EQ/NE y * C): New transformation.
15940 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
15942 * genmultilib (combination_space): Accept '+' in option names.
15944 2017-06-28 Martin Liska <mliska@suse.cz>
15947 * asan.c (instrument_derefs): Bail out inner references
15948 that are hard register variables.
15950 2017-06-28 Jakub Jelinek <jakub@redhat.com>
15953 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
15954 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
15956 2017-06-28 Richard Biener <rguenther@suse.de>
15958 * tree-vectorizer.h (vect_get_vec_defs): Remove.
15959 (vect_get_slp_defs): Adjust.
15960 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
15962 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
15964 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
15965 get_initial_defs_for_reduction instead of vect_get_vec_defs.
15966 (vectorizable_reduction): Adjust.
15967 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
15969 (vect_get_slp_defs): Likewise.
15970 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
15971 (vectorizable_bswap): Adjust.
15972 (vectorizable_call): Likewise.
15973 (vectorizable_conversion): Likewise.
15974 (vectorizable_assignment): Likewise.
15975 (vectorizable_shift): Likewise.
15976 (vectorizable_operation): Likewise.
15977 (vectorizable_store): Likewise.
15978 (vectorizable_condition): Likewise.
15979 (vectorizable_comparison): Likewise.
15981 2017-06-28 Michael Collison <michael.collison@arm.com>
15984 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
15986 (arm_gen_movmemqi): Removed unused variable 'i'.
15987 Convert 'for' loop into 'while' loop.
15988 (arm_expand_prologue): Remove last unnecessary set of insn.
15989 (thumb_pop): Remove unused variable 'pushed_words'.
15990 (thumb_exit): Remove last unnecessary set of regs_to_pop.
15992 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15994 * config/s390/predicates.md: Use s390_rel_address_ok_p.
15995 * config/s390/s390-protos.h: Add prototype of
15996 s390_rel_address_ok_p.
15997 * config/s390/s390.c (s390_got_symbol): New function.
15998 (s390_rel_address_ok_p): New function.
15999 (legitimize_pic_address): Use s390_rel_address_ok_p.
16000 (s390_load_got): Use s390_got_symbol.
16001 (s390_option_override): Issue error if
16002 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
16003 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
16005 * config/s390/s390.opt: New option mpic-data-is-text-relative.
16007 2017-06-27 Andrew Pinski <apinski@cavium.com>
16009 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
16010 (X * copysign (1.0, X)): New pattern.
16011 (X * copysign (1.0, -X)): New pattern.
16012 (copysign (-1.0, CST)): New pattern.
16014 2017-06-27 Joseph Myers <joseph@codesourcery.com>
16016 * genmultilib (combination_space): Remove variable.
16017 Validate reuse rules against regular expression for any sequence
16018 of multilib options in any order.
16020 2017-06-27 Michael Collison <michael.collison@arm.com>
16022 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
16023 call aarch64_split_simd_combine.
16024 * (aarch64_combine_internal<mode>): Delete pattern.
16025 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
16026 Allow register and subreg operands.
16028 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16030 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
16031 specific need, just fallback on defaults.
16032 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
16034 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16035 Olivier Hainque <hainque@adacore.com>
16037 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
16039 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
16040 targets. Pick a default if no particular attempt applied.
16041 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
16044 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16046 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
16047 (x86_64-wrs-vxworks7): Likewise.
16049 2017-06-27 Marek Polacek <polacek@redhat.com>
16052 * ubsan.c (instrument_null): Check get_base_address's result for null.
16054 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
16056 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
16058 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
16060 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
16061 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
16062 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
16063 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
16064 New function types.
16065 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
16066 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
16067 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
16068 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
16069 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
16070 BUILT_IN_FEUPDATEENV): New builtins.
16071 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
16072 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
16073 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
16074 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
16076 (builtin_structptr_types): Adjust size.
16077 * tree.c (builtin_structptr_types): Add four entries.
16079 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16080 Olivier Hainque <hainque@adacore.com>
16082 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
16083 (TLS_SYM): New local macro, forcing reference to __tls__ on
16084 link command lines for VxWorks 7 RTPs, triggering initialization
16086 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
16087 OS features TLS support, true for RTPs on VxWorks 7.
16088 * config/vxworks.c (vxworks_override_options): Setup emutls
16091 2017-06-27 Jakub Jelinek <jakub@redhat.com>
16093 * predict.c (test_prediction_value_range): Use -1U instead of -1
16094 to avoid narrowing conversion warning.
16095 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
16096 to avoid narrowing conversion warning.
16097 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
16099 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
16101 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16103 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
16104 64bit configurations.
16105 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
16106 (SIZE_TYPE): Likewise.
16107 * config/vxworks.c (vxworks_emutls_var_fields): Use
16108 long_unsigned_type_node instead of unsigned_type_node as the offset
16109 field type, which is "pointer" mode in emutls.c.
16111 2017-06-27 Jakub Jelinek <jakub@redhat.com>
16114 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
16116 PR middle-end/81207
16117 * gimple-fold.c (replace_call_with_call_and_fold): Handle
16118 gimple_vuse copying separately from gimple_vdef copying.
16120 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16122 * value-prof.c (free_hist): Remove call to memset and the enclosing if
16125 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
16126 Olivier Hainque <hainque@adacore.com>
16128 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
16129 for all vxworks7 targets.
16130 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
16131 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
16132 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
16133 variations for VX6/VX7 and 32/64bits later on in ...
16134 (VXWORKS_LIB_SPEC): Leverage new macros.
16135 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
16136 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
16138 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
16140 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
16141 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
16143 2017-06-26 Carl Love <cel@us.ibm.com>
16145 * config/rs6000/rs6000-c.c: Add support for built-in functions
16146 vector bool char vec_reve (vector bool char);
16147 vector signed char vec_reve (vector signed char);
16148 vector unsigned char vec_reve (vector unsigned char);
16149 vector bool int vec_reve (vector bool int);
16150 vector signed int vec_reve (vector signed int);
16151 vector unsigned int vec_reve (vector unsigned int);
16152 vector bool long long vec_reve (vector bool long long);
16153 vector signed long long vec_reve (vector signed long long);
16154 vector unsigned long long vec_reve (vector unsigned long long);
16155 vector bool short vec_reve (vector bool short);
16156 vector signed short vec_reve (vector signed short);
16157 vector double vec_reve (vector double);
16158 vector float vec_reve (vector float);
16159 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
16160 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
16161 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
16162 (altivec_vreve): New pattern.
16163 * config/rs6000/altivec.h (vec_reve): New define.
16164 * doc/extend.texi (vec_rev): Update the built-in documentation file
16165 for the new built-in functions.
16167 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16169 PR tree-optimization/71815
16170 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
16172 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
16174 (slsr_process_phi): Likewise.
16175 (replace_uncond_cands_and_profitable_phis): Don't replace a
16176 multiply candidate with a stride of 1 (copy or cast).
16177 (phi_incr_cost): Call uses_consumed_by_stmt rather than
16179 (lowest_cost_path): Likewise.
16180 (total_savings): Likewise.
16182 2017-06-26 Richard Biener <rguenther@suse.de>
16185 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
16186 Use def_builtin_pure for all gather builtins.
16188 2017-06-26 Richard Biener <rguenther@suse.de>
16190 PR tree-optimization/81203
16191 * tree-tailcall.c (find_tail_calls): Do not move stmts into
16192 non-dominating BBs.
16194 2017-06-26 Marek Polacek <polacek@redhat.com>
16197 * doc/invoke.texi: Document -Wmultistatement-macros.
16199 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
16201 * doc/sourcebuild.texi (ARM-specific attributes): Document new
16202 arm_neon_ok_no_float_abi effective target.
16204 2017-06-26 Richard Biener <rguenther@suse.de>
16206 PR tree-optimization/80928
16207 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
16208 (copy_bbs): Set BB_DUPLICATED flag early.
16209 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
16211 (execute_on_shrinking_pred): Likewise.
16212 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
16213 BB_DUPLICATED blocks.
16214 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
16215 iterate over all PHIs considering removal of *gsi.
16217 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
16219 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
16222 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16224 * config/rs6000/rs6000-string.c: (expand_block_clear,
16225 do_load_for_compare, select_block_compare_mode,
16226 compute_current_alignment, expand_block_compare,
16227 expand_strncmp_align_check, expand_strn_compare,
16228 expand_block_move, rs6000_output_load_multiple)
16229 Move functions related to string/block move/compare
16230 to a separate file.
16231 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
16232 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
16233 for this function which is now used in two files.
16234 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
16235 * config.gcc: Add rs6000-string.o to extra_objs for
16236 targets powerpc*-*-* and rs6000*-*-*.
16238 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
16241 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
16242 32-bit, since indexed is not valid for DImode.
16243 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
16244 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
16245 (define_peephole2 for Altivec d-form load): Add 32-bit support.
16246 (define_peephole2 for Altivec d-form store): Likewise.
16249 * multiple_target.c (create_dispatcher_calls): Only create the
16250 dispatcher call if the function is the default clone of a
16251 versioned function.
16253 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
16255 PR middle-end/80902
16256 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
16257 a call, force the call to not be a tail call.
16259 2017-06-23 Jeff Law <law@redhat.com>
16261 * doc/contrib.texi: Add entry for Steven Pemberton's work on
16264 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
16266 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
16267 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
16268 handling for early expansion of vector shifts (sl,sr,sra,rl).
16269 (builtin_function_type): Add vector shift right instructions
16270 to the unsigned argument list.
16272 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
16274 rtl-optimizatoin/79286
16275 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
16276 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
16277 trap. PIC register plus a const unspec without offset can never trap.
16279 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
16281 * tree.h (builtin_structptr_type): New type.
16282 (builtin_structptr_types): Declare new array.
16283 * tree.c (builtin_structptr_types): New array.
16284 (free_lang_data, build_common_tree_nodes): Use it.
16286 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
16289 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
16292 2017-06-22 Matt Turner <mattst88@gmail.com>
16294 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
16295 Lake models to skylake case. Assume skylake for unknown
16296 models with clflushopt.
16298 2017-06-22 Jeff Law <law@redhat.com>
16300 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
16301 frame sizes that do not satisfy aarch64_uimm12_shift.
16303 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
16305 * profile-count.h (apply_probability,
16306 apply_scale, probability_in): Fix checks for zero.
16308 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16310 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
16311 * doc/cppdiropts.texi (-I @var{dir}): Document it.
16313 2016-06-22 Richard Biener <rguenther@suse.de>
16315 * tree-vect-loop.c (vect_model_reduction_cost): Handle
16316 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
16317 REDUC_MAX_EXPR support.
16318 (vectorizable_reduction): Likewise.
16319 (vect_create_epilog_for_reduction): Likewise.
16321 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
16323 * match.pd (A / (1 << B) -> A >> B): New.
16324 * generic-match-head.c: Include optabs-tree.h.
16325 * gimple-match-head.c: Likewise.
16326 * optabs-tree.h (target_supports_op_p): New.
16327 * optabs-tree.c (target_supports_op_p): New.
16329 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16331 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
16332 $gcc_cv_ld --help output.
16333 (gcc_cv_ld_demangle): Likewise.
16334 (gcc_cv_ld_eh_frame_hdr): Likewise.
16335 (gcc_cv_ld_pie): Likewise.
16336 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
16337 (gcc_cv_ld_buildid): Likewise.
16338 (gcc_cv_ld_sysroot): Likewise.
16339 (ld_bndplt_support): Likewise.
16340 (ld_pushpopstate_support): Likewise.
16341 * configure: Regenerate.
16342 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
16344 2017-06-21 Jakub Jelinek <jakub@redhat.com>
16347 * config/i386/sse.md (round<mode>2): Renumber match_dup and
16348 operands indexes to avoid gap between operands and match_dups.
16350 2017-06-21 Andrew Pinski <apinski@cavium.com>
16352 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
16353 Increment Arith_shift and Arith_shift_reg by 1.
16354 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
16356 * config/aarch64/aarch64.c (thunderx_tunings): Enable
16357 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
16358 (aarch64_strip_extend): Add new argument and test for it.
16359 (aarch64_cheap_mult_shift_p): New function.
16360 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
16361 add a cost if it is true.
16362 Update calls to aarch64_strip_extend.
16363 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
16365 2017-06-21 Andrew Pinski <apinski@cavium.com>
16367 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
16369 (thunderxt88): Likewise.
16370 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
16371 (thunderx_prefetch_tune): New variable.
16372 (thunderx2t99_prefetch_tune): Update for the correct values.
16373 (thunderxt88_tunings): New variable.
16374 (thunderx_tunings): Use thunderx_prefetch_tune instead of
16375 generic_prefetch_tune.
16376 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
16378 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16380 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
16381 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
16382 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
16383 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
16384 (aarch64_atomic_cas<mode>, GPI): Likewise.
16386 2017-06-21 Martin Liska <mliska@suse.cz>
16388 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
16389 statements on cold and hot labels.
16390 * predict.c (tree_estimate_probability_bb): Remove the
16391 prediction from this place.
16393 2017-06-21 Martin Liska <mliska@suse.cz>
16395 PR tree-optimization/79489
16396 * gimplify.c (maybe_add_early_return_predict_stmt): New
16398 (gimplify_return_expr): Call the function.
16399 * predict.c (tree_estimate_probability_bb): Remove handling
16401 * predict.def: Update comment about early return predictor.
16402 * gimple-predict.h (is_gimple_predict): New function.
16403 * predict.def: Change default value of early return to 66.
16404 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
16406 * passes.def: Put pass_strip_predict_hints to the beginning of
16409 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
16411 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
16412 FUNCTION_DECL declarations.
16413 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
16415 (dwaf2out_decl): Likewise.
16416 * godump.c (go_early_global_decl): Skip call to the real debug hook
16417 for FUNCTION_DECL declarations.
16418 * passes.c (rest_of_decl_compilation): Skip call to the
16419 early_global_decl debug hook for FUNCTION_DECL declarations, unless
16420 -fdump-go-spec is passed.
16422 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
16424 * config/i386/i386.c (struct builtin_isa): New field pure_p.
16425 Reorder for compactness.
16426 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
16427 (def_builtin_pure, def_builtin_pure2): New functions.
16428 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
16430 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
16432 * match.pd (nop_convert): New predicate.
16433 ((A +- CST1) +- CST2): Allow some NOP conversions.
16435 2017-06-21 Jakub Jelinek <jakub@redhat.com>
16438 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
16439 with ctors/dtors if GOVD_SHARED is set.
16441 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
16443 * config/aarch64/aarch64.md (movti_aarch64):
16444 Emit mov rather than orr.
16445 (movtf_aarch64): Likewise.
16446 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
16447 Emit mov rather than orr.
16449 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
16451 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
16452 Swap alternatives, make integer dup more expensive.
16454 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
16456 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
16457 Return true for non-tls symbols.
16459 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
16461 * config/aarch64/aarch64-cores.def (cortex-a55): New.
16462 (cortex-a75): Likewise.
16463 (cortex-a75.cortex-a55): Likewise.
16464 * config/aarch64/aarch64-tune.md: Regenerate.
16465 * doc/invoke.texi (-mtune): Document new values for -mtune.
16467 2017-06-21 Tom de Vries <tom@codesourcery.com>
16469 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
16470 stack_size feature.
16471 (Effective-Target Keywords, Other attributes): Suggest using
16472 dg-add-options stack_size feature to get stack limit in stack_size
16473 effective target documentation.
16475 2017-06-21 Julian Brown <julian@codesourcery.com>
16476 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16478 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
16479 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
16480 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
16482 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
16483 attribute type list for neon_multiply.
16484 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
16485 attribute type list for neon_multiply.
16486 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
16487 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
16488 attribute type list for neon_multiply.
16489 * config/arm/types.md (crypto_pmull): Add.
16490 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
16491 attribute type list.
16493 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
16495 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
16498 2017-06-20 Jakub Jelinek <jakub@redhat.com>
16500 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
16501 to make sure not to dereference a NULL cost_classes_ptr pointer.
16503 2017-06-20 Carl Love <cel@us.ibm.com>
16505 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16506 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16507 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16508 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16509 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16510 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16511 VMULOSW): New enum "unspec" values.
16512 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
16513 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
16514 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16515 altivec_vmulosw): New patterns.
16516 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16517 VMULOSW): Add definitions.
16519 2017-06-20 Julia Koval <julia.koval@intel.com>
16521 * config/i386/i386.c: Fix rounding expand for new pattern.
16522 * config/i386/subst.md: Fix pattern (parallel -> unspec).
16524 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
16526 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
16527 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
16529 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
16531 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
16534 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
16536 * config/aarch64/aarch64-cores.def: Rearrange to sort by
16537 architecture, then by implementer ID.
16538 * config/aarch64/aarch64-tune.md: Regenerate.
16540 2017-06-20 Richard Biener <rguenther@suse.de>
16542 PR middle-end/81097
16543 * fold-const.c (split_tree): Fold to type before negating.
16545 2017-06-20 David Malcolm <dmalcolm@redhat.com>
16547 * diagnostic-show-locus.c
16548 (selftest::test_fixit_deletion_affecting_newline): New function.
16549 (selftest::diagnostic_show_locus_c_tests): Call it.
16551 2017-06-20 Andreas Schwab <schwab@suse.de>
16554 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
16557 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
16559 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
16560 __ARM_FEATURE_COPROC according to support.
16562 2017-06-20 Jakub Jelinek <jakub@redhat.com>
16564 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
16565 Rewritten to avoid overflow for > 32-bit pointers.
16568 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
16569 by removing enum keyword.
16570 (ubsan_type_descriptor): Likewise. Formatting fix.
16573 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
16574 splitter): Require TARGET_SSE2 in the condition.
16576 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
16579 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
16580 for doing vector set of SFmode on ISA 3.0.
16581 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
16582 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
16584 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
16585 SFmode value into a V4SF variable that was extracted from another
16586 V4SF variable without converting the element to double precision
16587 and back to single precision vector format.
16588 (vsx_insert_extract_v4sf_p9_2): Likewise.
16590 2017-06-19 Jakub Jelinek <jakub@redhat.com>
16592 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
16593 in UWHI to avoid undefined overflow.
16596 * ubsan.h (enum ubsan_encode_value_phase): New.
16597 (ubsan_encode_value): Change second argument to
16598 enum ubsan_encode_value_phase with default value of
16599 UBSAN_ENCODE_VALUE_GENERIC.
16600 * ubsan.c (ubsan_encode_value): Change second argument to
16601 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
16602 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
16603 create_tmp_var_raw instead of create_tmp_var and use a
16605 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
16606 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
16607 ubsan_encode_value callers.
16610 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
16611 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
16612 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
16614 2017-06-19 Richard Biener <rguenther@suse.de>
16616 PR middle-end/81118
16617 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
16618 estimates if we changed anything.
16620 2017-06-19 Richard Biener <rguenther@suse.de>
16622 PR tree-optimization/80887
16623 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
16624 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
16625 simplified lookups, then reset mprts_hook.
16626 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
16628 (try_to_simplify): Likewise.
16630 2017-06-19 Martin Liska <mliska@suse.cz>
16633 * gimplify.c (gimplify_switch_expr):
16634 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
16636 2017-06-19 Martin Liska <mliska@suse.cz>
16638 * doc/install.texi: Document that PGO runs in 4 stages.
16640 2017-06-19 Martin Liska <mliska@suse.cz>
16643 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
16644 to dispatcher function name.
16645 * multiple_target.c (replace_function_decl): New function.
16646 (create_dispatcher_calls): Redirect both edges and references.
16648 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
16650 * profile-count.c (profile_count::dump): Dump quality.
16651 (profile_count::differs_from_p): Update for unsigned val.
16652 * profile-count.h (profile_count_quality): New enum.
16653 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
16655 2017-06-19 Richard Biener <rguenther@suse.de>
16657 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
16658 struct function as arg.
16659 (estimate_numbers_of_iterations): Export overload with loop arg.
16660 (free_numbers_of_iterations_estimates_loop): Use an overload of
16661 free_numbers_of_iterations_estimates instead.
16662 * tree-cfg.c (remove_bb): Adjust.
16663 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
16664 * tree-parloops.c (gen_parallel_loop): Likewise.
16665 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
16667 (tree_unroll_loops_completely): Likewise.
16668 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
16669 Use an overload instead and export.
16670 (estimated_loop_iterations): Adjust.
16671 (max_loop_iterations): Likewise.
16672 (likely_max_loop_iterations): Likewise.
16673 (estimate_numbers_of_iterations): Take struct function as arg
16675 (loop_exits_before_overflow): Adjust.
16676 (free_numbers_of_iterations_estimates_loop): Use an overload.
16677 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
16678 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
16680 2017-06-19 Richard Biener <rguenther@suse.de>
16683 * ipa-prop.c (find_constructor_constant_at_offset): Handle
16684 RANGE_EXPR conservatively.
16686 2017-06-16 Carl Love <cel@us.ibm.com>
16688 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16689 definitions for vec_float, vec_float2, vec_floato,
16690 vec_floate built-ins.
16691 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
16692 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
16694 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
16695 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
16696 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
16697 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
16698 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
16699 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
16700 vec_floato): Add builtin defines.
16701 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
16702 Update the built-in documentation file for the new built-in
16705 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16707 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
16708 (mthumb): Mark as the negative of -marm.
16710 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16712 * doc/invoke.texi (ARM Options, -mcpu): Document supported
16714 (ARM Options, -mtune): Document that this accepts the same
16715 extension options as -mcpu.
16716 (ARM Options, -mfpu): Document addition of -mfpu=auto.
16718 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16720 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
16721 permitted extensions.
16723 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16725 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
16726 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
16727 (armv8-m.main): Add option +nodsp.
16728 * config/arm/arm-cpu-cdata.h: Regenerated.
16730 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16732 * config/arm/t-fuchsia: New file.
16733 * config.gcc (arm*-*-fuchsia*): Use it.
16735 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16737 * config/arm/t-symbian: Rewrite for new option infrastructure.
16739 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16741 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
16742 (MULTILIB_REQUIRED): Likewise.
16744 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16746 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
16747 (MULTILIB_RESUE): Likewise.
16748 (MULTILIB_MATCHES): Likewise.
16749 (MULTLIB_REQUIRED): Likewise.
16751 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16753 * config/arm/t-rtems: Rewrite for new option framework.
16755 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16757 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
16758 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
16759 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
16760 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
16761 * config/arm/t-multilib: ... here.
16762 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
16763 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
16764 armv7-a and armv8*-a when A-profile libraries have not been built.
16765 * config/arm/t-rmprofile: Rewrite.
16767 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16769 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
16770 with a backslash. Remove the backslash after substituting unescaped
16772 * doc/fragments.texi (MULTILIB_REUSE): Document it.
16774 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16776 * config.gcc: (arm*-*-*): When building a-profile libraries, force
16777 the driver to pass through the default setting of -mfloat-abi.
16778 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
16780 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
16781 (all_feat_combs): New rule.
16782 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
16784 * config/arm/t-aprofile: Rewrite.
16786 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16788 * config/arm/arm.h (FPUTYPE_AUTO): Define.
16789 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
16790 fpu is not specified by the user/command-line.
16791 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
16792 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
16793 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
16794 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
16795 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
16796 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
16798 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16800 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
16801 * config/arm/t-arm-elf: Rewritten.
16803 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16805 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
16806 have some floating-point instructions.
16807 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
16808 (TARGET_MAYBE_HARD_FLOAT): New macro.
16809 * config/arm/arm-builtins.c (arm_init_builtins): Use
16810 TARGET_MAYBE_HARD_FLOAT.
16811 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
16813 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16815 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
16816 (configargs.h): Include it.
16817 (arm_print_hint_for_fpu_option): New function.
16818 (arm_parse_fpu_option): New function.
16819 (candidate_extension): New class.
16820 (arm_canon_for_multilib): New function.
16821 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
16822 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
16823 (ARCH_CANONICAL_SPECS): New macro.
16824 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
16826 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16828 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
16829 are set after handling multilib fragments. Set target_cpu_default2
16832 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16834 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
16836 (arm*-*-*): Set target_cpu_default2 to a quoted string.
16837 * config/arm/parsecpu.awk (check_cpu): Validate any extension
16839 (check_arch): Likewise.
16840 * config/arm/arm.c (arm_configure_build_target): Handle
16841 TARGET_CPU_DEFAULT being a string constant. Scan any feature
16842 options in the default.
16844 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16846 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
16847 when an option is an alias of another.
16848 * config/arm/parsecpu.awk (optalias): New parser token.
16849 (gen_comm_data): Mark non-alias options as such. Emit entries
16850 for extension aliases.
16851 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
16852 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
16853 (armv6kz, armv6zk, armv6t2): Likewise.
16854 (armv7): Make vfpv3-d16 an alias.
16855 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
16857 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
16858 Sort in canonical order.
16859 (armv8-a): Sort in canonical order.
16860 (armv8.1-a, armv8.2-a): Likewise.
16861 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
16863 (cortex-a9): Sort in canonical order.
16864 * config/arm/arm.c (selftests.h): Include it.
16865 (arm_test_cpu_arch_data): New function.
16866 (arm_run_self_tests): New function.
16867 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
16868 (targetm): Move declaration to the end of the file.
16869 * arm-cpu-cdata.h: Regenerated.
16871 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16873 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
16874 call to target_mode_check describing the type of option passed.
16875 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
16876 (arm_target_thumb_only): Use arm_parse_arch_option_name or
16877 arm_parse_cpu_option_name to match parameters against list of
16879 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
16880 arm_arch_core_flags data structure.
16881 * config/arm/arm-cpu_cdata.h: Regenerated.
16883 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16885 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
16887 (arm_print_hint_for_cpu_option): Likewise.
16888 (arm_print_hint_for_arch_option): Likewise.
16889 (arm_parse_cpu_option_name): Likewise.
16890 (arm_parse_arch_option_name): Likewise.
16891 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
16892 of entries in the all_fpus list.
16893 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
16894 (arm_parse_cpu_option_name): Declare.
16895 (arm_parse_arch_option_name): Declare.
16896 (arm_parse_option_features): Declare.
16897 (arm_intialize_isa): Declare.
16898 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
16900 (gen_comm_data): ... here. Make definitions non-static.
16901 * config/arm/arm-cpu-data.h: Regenerated.
16902 * config/arm/arm-cpu-cdata.h: Regenerated.
16904 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16906 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
16907 (cpu_arch_extension): New structure.
16908 (cpu_arch_option, arch_option, cpu_option): New structures.
16909 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
16910 architecture types.
16911 (gen_data): Generate new format data tables.
16912 * config/arm/arm.c (cpu_tune): New structure.
16913 (cpu_option, processors): Delete.
16914 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
16915 (arm_print_hint_for_cpu_option): ... this and ...
16916 (arm_print_hint_for_arch_option): ... this.
16917 (arm_parse_arch_cpu_name): Delete. Replace with ...
16918 (arm_parse_cpu_option_name): ... this and ...
16919 (arm_parse_arch_option_name): ... this.
16920 (arm_unrecognized_feature): Change type of target parameter to
16922 (arm_parse_arch_cpu_features): Delete. Replace with ...
16923 (arm_parse_option_features): ... this.
16924 (arm_configure_build_target): Rework to use new configuration data
16926 (arm_print_tune_info): Rework for new configuration data tables.
16927 * config/arm/arm-cpu-data.h: Regenerated.
16928 * config/arm/arm-cpu.h: Regenerated.
16930 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16932 * Makefile.in (OBJS): Move sbitmap.o from here ...
16933 (OBJS-libcommon): ... to here.
16935 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16937 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
16938 (ISA_ALL_CRYPTO): New macro.
16939 (ISA_ALL_SIMD): New macro
16940 (ISA_ALL_FP): New macro.
16941 * config/arm/arm.c (fpu_bitlist): Update initializer.
16942 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
16944 (arm9e): Add fpu. Add option for nofp
16945 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
16946 (arm926ej-s, arm1026ej-s): Likewise.
16947 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
16948 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
16949 neon-fp16, neon-vfpv4, nofp and nosimd.
16950 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
16951 (cortex-a8): Add fpu. Add option for nofp.
16952 (cortex-a9): Add fpu. Add options for nosimd and nofp.
16953 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
16954 (cortex-r4f): Add fpu.
16955 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
16956 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
16958 (cortex-r8): Likewise.
16959 (cortex-m4): Add fpu. Add option for nofp.
16960 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
16961 (cortex-a17.cortex-a7): Likewise.
16962 (cortex-a32): Add fpu. Add options for crypto and nofp.
16963 (cortex-a35, cortex-a53): Likewise.
16964 (cortex-a57): Add fpu. Add option for crypto.
16965 (cortex-a72, cortex-a73): Likewise.
16966 (exynos-m1): Likewise.
16967 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
16968 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
16969 (cortex-m33): Add fpu. Add option for nofp.
16970 * config/arm/arm-cpu-cdata.h: Regenerated
16971 * config/arm/arm-cpu-data.h: Regenerated.
16973 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16975 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
16976 (armv5te, armv5tej): Likewise.
16977 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
16978 (armv7): Add options fp and vfpv3-d16.
16979 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
16980 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
16982 (armv7ve): Likewise.
16983 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
16984 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
16985 (armv8-a): Add nocrypto option.
16986 (armv8.1-a, armv8.2-a): Likewise.
16987 (armv8-m.main): add options fp, fp.dp and nofp.
16989 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
16991 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
16993 (armv8-a+crc): Delete.
16994 (armv8.1-a): Add options simd, crypto and nofp.
16995 (armv8.2-a): Add options fp16, simd, crypto and nofp.
16996 (armv8.2-a+fp16): Delete.
16997 (armv8-m.main): Add option dsp.
16998 (armv8-m.main+dsp): Delete.
16999 (cortex-a8): Add fpu. Add nofp option.
17000 (cortex-a9): Add fpu. Add nofp and nosimd options.
17001 * config/arm/parsecpu.awk (gen_data): Generate option tables and
17002 link to main cpu and architecture data structures.
17003 (gen_comm_data): Only put isa attributes from the main architecture
17005 (option): New statement for architecture and CPU entries.
17006 * arm.c (struct cpu_option): New structure.
17007 (struct processors): Add entry for options.
17008 (arm_unrecognized_feature): New function.
17009 (arm_parse_arch_cpu_name): Ignore any characters after the first
17011 (arm_parse_arch_cpu_feature): New function.
17012 (arm_configure_build_target): Separate out any CPU and architecture
17013 features and parse separately. Don't error out if -mfpu=auto is
17014 used with only an architecture string.
17015 (arm_print_asm_arch_directives): New function.
17016 (arm_file_start): Call it.
17017 * config/arm/arm-cpu-cdata.h: Regenerated.
17018 * config/arm/arm-cpu-data.h: Likewise.
17019 * config/arm/arm-tables.opt: Likewise.
17021 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17023 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
17024 assembler when it is not -mfpu=auto.
17026 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17028 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
17029 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
17030 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
17031 (ASM_CPU_SPEC): Rewrite.
17032 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
17033 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
17034 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
17035 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
17036 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
17037 copied string is NUL-terminated. Also strip any characters prefixed
17039 (arm_rewrite_selected_arch): New function.
17040 (arm_rewrite_march): New function.
17042 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17044 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
17045 (x_arm_cpu_string, x_arm_tune_string): Likewise.
17046 (march, mcpu, mtune): Convert to string-based options.
17047 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
17048 (arm_parse_arch_cpu_name): New function.
17049 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
17050 identify selected architecture or CPU.
17051 (arm_option_save): New function.
17052 (TARGET_OPTION_SAVE): Redefine.
17053 (arm_option_restore): Restore string options.
17054 (arm_option_print): Print string options.
17056 2017-06-16 Martin Sebor <msebor@redhat.com>
17058 PR tree-optimization/80933
17059 PR tree-optimization/80934
17060 * builtins.c (fold_builtin_3): Do not handle bcmp here.
17061 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
17062 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
17063 (gimple_fold_builtin): Call them.
17065 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17067 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
17068 as unlikely; update profile.
17070 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17072 * predict.c (force_edge_cold): Handle declaring edges impossible
17075 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17077 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
17079 (try_unroll_loop_completely): Fix reporting.
17081 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17083 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
17085 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
17088 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
17089 if given a non-constant argument for an intrinsic which requires a
17092 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17094 * profile.c (compare_freqs): New function.
17095 (branch_prob): Sort edge list.
17096 (find_spanning_tree): Assume that the list is priority sorted.
17098 2017-06-16 Richard Biener <rguenther@suse.de>
17100 PR tree-optimization/81090
17101 * passes.def (pass_record_bounds): Remove.
17102 * tree-pass.h (make_pass_record_bounds): Likewise.
17103 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
17104 make_pass_record_bounds): Likewise.
17105 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
17106 not free niter estimates at the beginning but at the end.
17107 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
17109 2017-06-16 Richard Biener <rguenther@suse.de>
17111 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
17112 initializer to workaround ICE in host GCC 4.8.
17114 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17116 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
17118 (clone_inlined_nodes): Update.
17120 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17122 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
17123 prefetch settings, and enable prefetching by default at -O3.
17125 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17127 * config/aarch64/aarch64.c (aarch64_override_options_internal):
17128 Set flag_prefetch_loop_arrays according to tuning data.
17130 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17132 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
17133 New tune structure.
17134 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
17135 [Unrelated to main purpose of the patch] Place the pointer field last
17136 to enable type checking errors when tune structure are wrongly merged.
17137 * config/aarch64/aarch64.c (generic_prefetch_tune,)
17138 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
17139 (thunderx2t99_prefetch_tune): New tune constants.
17140 (tune_params *_tunings): Update all tunings (no functional change).
17141 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
17142 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
17143 from tunings structures.
17145 2017-06-16 Jakub Jelinek <jakub@redhat.com>
17148 * ubsan.c (instrument_null): Add T argument, use it instead
17149 of computing it based on IS_LHS.
17150 (instrument_object_size): Likewise.
17151 (pass_ubsan::execute): Adjust instrument_null and
17152 instrument_object_size callers to pass gimple_get_lhs or
17153 gimple_assign_rhs1 result to it. Use instrument_null instead of
17154 calling get_base_address and instrument_mem_ref. Handle
17155 aggregate call arguments for object-size sanitization.
17157 2017-06-16 Yury Gribov <tetra2005@gmail.com>
17159 PR tree-optimization/81089
17160 * tree-vrp.c (is_masked_range_test): Validate operands of
17163 2017-06-15 Martin Sebor <msebor@redhat.com>
17166 * dumpfile.c (dump_register): Avoid calling memset to initialize
17167 a class with a default ctor.
17168 * gcc.c (struct compiler): Remove const qualification.
17169 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
17170 * hash-table.h: Ditto.
17171 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
17173 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
17174 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
17176 * params.h (struct param_info): Make struct members non-const.
17177 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
17178 with default initialization.
17179 * vec.h (vec_copy_construct, vec_default_construct): New helper
17181 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
17182 with vec_copy_construct.
17183 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
17184 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
17185 * doc/invoke.texi (-Wclass-memaccess): Document.
17187 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17189 * emit-rtl.h (is_leaf): Update comment about local
17190 register allocator.
17192 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
17195 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
17196 for a variable to have a section before checking if the section has a
17198 Set section to.persistent if persistent attribute is set.
17199 Warn if .persistent attribute is used on an automatic variable.
17201 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
17203 PR rtl-optimization/80474
17204 * reorg.c (update_block): Do not ignore instructions in a delay slot.
17206 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
17208 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
17211 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
17213 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
17214 (casesi): Emit bounds checking as RTL.
17215 (casesi_internal_mips16_<mode>): Remove bounds checking.
17217 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
17219 * config/xtensa/xtensa.c (xtensa_option_override): Append
17220 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
17221 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
17222 xtensa_doloop_hooks): Define unconditionally.
17223 (xtensa_reorg_loops): Only call reorg_loops in the presence of
17225 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
17226 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
17227 for it in xtensa_option_override.
17228 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
17229 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
17231 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
17233 * doc/cppopts.texi: Document '-' special value to -MF.
17235 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
17237 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
17238 (cortex_a53_fconst): Likewise.
17239 (cortex_a53_fpmul): Likewise.
17240 (cortex_a53_f_load_64): Likewise.
17241 (cortex_a53_f_load_many): Likewise.
17242 (cortex_a53_advsimd_alu): Likewise.
17243 (cortex_a53_advsimd_alu_q): Likewise.
17244 (cortex_a53_advsimd_mul): Likewise.
17245 (cortex_a53_advsimd_mul_q): Likewise.
17246 (fpmac bypass): Add new bypass for fpmac-fpmac case.
17247 Add missing fmul, r2f_cvt and fconst cases.
17249 2017-06-14 Richard Biener <rguenther@suse.de>
17251 PR middle-end/81088
17252 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
17254 (fold_binary_loc): When associating do not treat pre-existing
17255 TREE_OVERFLOW on literal constants as a reason to allow
17256 TREE_OVERFLOW on associated literal constants.
17258 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
17260 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
17261 (MASK_FEATURES): New macro.
17262 * config/sparc/sparc.c (sparc_option_override): Remove the special
17263 handling of -mfpu and generalize it to all MASK_FEATURES switches.
17265 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
17267 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
17268 a division of 0 if non-call exceptions are enabled.
17270 2017-06-14 Andrew Pinski <apinski@cavium.com>
17271 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17274 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
17275 Improve vector initialization code gen for only variable case.
17277 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
17279 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
17281 2017-06-14 Richard Biener <rguenther@suse.de>
17283 PR tree-optimization/81083
17284 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
17287 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17289 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
17290 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
17291 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
17292 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
17293 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
17294 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
17296 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17298 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
17299 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
17301 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17303 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
17305 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17307 * config/rs6000/t-rtems: Don't handle SPE.
17309 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17311 * config/rs6000/t-linux: Don't handle SPE.
17313 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17315 * config/rs6000/eabispe.h: Delete file.
17317 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17319 * config/rs6000/t-spe: Delete file.
17321 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17323 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
17324 (rs6000_legitimate_offset_address_p): Return false for anything in
17325 V2SImode or V2SFmode.
17327 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17329 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
17330 except V2SF and V2SI. Rearrange the vector modes, and add comments.
17331 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
17333 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
17334 (rs6000_legitimate_offset_address_p): Ditto.
17335 (rs6000_emit_move): Ditto.
17336 (rs6000_init_builtins): Remove V4HI_type_node.
17338 2017-06-13 Martin Liska <mliska@suse.cz>
17341 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
17342 (gate_asan): Likewise.
17343 * asan.h (asan_no_sanitize_address_p): Remove the function.
17344 (sanitize_flags_p): New function.
17345 * builtins.def: Fix coding style.
17346 * common.opt: Use renamed enum value.
17347 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
17348 * doc/extend.texi: Document no_sanitize attribute.
17349 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
17350 to SANITIZE_UNDEFINED_NONDEFAULT.
17351 * gcc.c (sanitize_spec_function): Use the renamed enum value.
17352 * gimple-fold.c (optimize_atomic_compare_exchange_p):
17353 Use sanitize_flags_p.
17354 * gimplify.c (gimplify_function_tree): Likewise.
17355 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
17356 * opts.c (parse_no_sanitize_attribute): New function.
17357 (common_handle_option): Use renamed enum value.
17358 * opts.h (parse_no_sanitize_attribute): Declare.
17359 * tree.c (sanitize_flags_p): New function.
17360 * tree.h: Declared here.
17361 * tsan.c: Use sanitize_flags_p.
17362 * ubsan.c (ubsan_expand_null_ifn): Likewise.
17363 (instrument_mem_ref): Likewise.
17364 (instrument_bool_enum_load): Likewise.
17365 (do_ubsan_in_current_function): Remove the function.
17366 (pass_ubsan::execute): Use sanitize_flags_p.
17367 * ubsan.h: Remove do_ubsan_in_current_function
17368 * tree-cfg.c (print_no_sanitize_attr_value): New function.
17369 (dump_function_to_file): Use it here.
17371 2017-06-13 Martin Jambor <mjambor@suse.cz>
17373 PR tree-optimization/80803
17374 PR tree-optimization/81063
17375 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
17376 (propagate_subaccesses_across_link): Enqueue subtree whenever
17377 necessary instead of relying on the caller.
17379 2017-06-13 Martin Jambor <mjambor@suse.cz>
17381 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
17382 that have a first_link.
17383 (sort_and_splice_var_accesses): Do not check first_link before
17385 (subtree_mark_written_and_enqueue): Likewise.
17386 (propagate_all_subaccesses): Likewise and do not stop at first
17387 parent with a first_link.
17389 2017-06-13 Martin Jambor <mjambor@suse.cz>
17391 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
17394 2017-06-13 Yury Gribov <tetra2005@gmail.com>
17396 * match.pd: New pattern.
17398 2017-06-13 Yury Gribov <tetra2005@gmail.com>
17400 * tree-vrp.c (is_masked_range_test): New function.
17401 (register_edge_assert_for): Determine ranges for
17404 2017-06-13 Yury Gribov <tetra2005@gmail.com>
17406 PR tree-optimization/67328
17407 * fold-const.c (maskable_range_p): New function.
17408 (build_range_check): Generate bittests if possible.
17410 2017-06-13 Martin Liska <mliska@suse.cz>
17412 * gimple-pretty-print.c (dump_probability): Add new argument.
17413 (dump_edge_probability): Dump both probability and count.
17414 (dump_gimple_label): Likewise.
17415 (dump_gimple_bb_header): Likewise.
17417 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
17420 * config/avr/avr-devices.c: Fix indentation.
17421 * config/avr/gen-avr-mmcu-specs.c: Dito.
17423 2017-06-13 Richard Biener <rguenther@suse.de>
17425 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
17426 instead get vector type from stmt_info.
17427 (vectorizable_reduction): Adjust. Remove dead code.
17429 2017-06-13 Richard Biener <rguenther@suse.de>
17431 PR middle-end/81065
17432 * fold-const.c (extract_muldiv_1): Remove bogus distribution
17433 case of C * (x * C2 + C3).
17434 (fold_addr_of_array_ref_difference): Properly fold index difference.
17436 2017-06-12 David S. Miller <davem@davemloft.net>
17439 * config/sparc/sparc.md (return expander): Emit frame blockage if
17440 function uses alloca.
17442 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
17444 * combine.c (make_field_assignment): Check len rather than the mode
17445 precision when calling force_to_mode.
17447 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
17449 Support multilibs and devices that see flash in RAM address range.
17452 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
17453 (avr_mcu_t) <flash_pm_offset>: New field.
17454 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
17455 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
17456 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
17457 (AVR_TINY_PM_OFFSET): Remove macro.
17458 * config/avr/avr.opt (-mshort-calls): New option.
17459 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
17460 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
17461 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
17462 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
17463 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
17464 instead of avr_arch->have_jmp_call.
17465 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
17466 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
17467 avr_arch->flash_pm_offset to define.
17468 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
17469 new field flash_pm_offset. Add entry for avrxmega3.
17470 (avr_texinfo): Add entry for avrxmega3.
17471 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
17472 attiny212, attiny214,
17473 attiny412, attiny414, attiny416, attiny417,
17474 attiny814, attiny816, attiny817,
17475 attiny1614, attiny1616, attiny1617,
17476 attiny3214, attiny3216, attiny3217.
17477 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
17478 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
17479 (avr_print_operand_address) [AVR_TINY]: Same.
17480 (avr_asm_init_sections) <readonly_data_section>: Only patch
17481 callback if avr_arch->flash_pm_offset = 0.
17482 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
17483 for rodata if avr_arch->flash_pm_offset != 0.
17484 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
17485 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
17486 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
17487 (m_options): Append opt_rcall.
17488 (m_dirnames): Append dir_rcall.
17489 * config/avr/t-multilib: Regenerate.
17491 * configure.ac [target=avr]: Check whether avrxmega3 default
17492 linker description file works as needed.
17493 * configure: Regenerate.
17494 * doc/avr-mmcu.texi: Regenerate.
17495 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
17496 <__AVR_ARCH__>: Document avrxmega3 and 103.
17497 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
17498 <__AVR_SHORT_CALLS__>: Document it.
17499 <__AVR_PM_BASE_ADDRESS__>: Document it.
17500 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
17501 (AVR Variable Attributes) <progmem>: Document this is
17502 not needed for avrxmega3.
17503 (AVR Named Address Spaces) <__flash>: Dito.
17505 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
17507 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
17509 2017-06-12 Doug Rupp <rupp@adacore.com>
17511 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
17512 Append vxworks-stdint.h to the tm_file list.
17513 * config/vxworks-stdint.h: New file.
17515 2017-06-12 Martin Liska <mliska@suse.cz>
17517 PR tree-optimization/81041
17518 * tree-profile.c (gimple_gen_ic_func_profiler):
17519 Create an extra BB in profile-generate
17520 (gimple_gen_time_profiler): Likewise.
17522 2017-06-12 Jakub Jelinek <jakub@redhat.com>
17524 PR tree-optimization/81003
17525 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
17526 (update_range_test): Use it instead of force_gimple_operand_gsi.
17528 2017-06-12 Richard Biener <rguenther@suse.de>
17530 PR tree-optimization/81053
17531 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
17532 with backedge value not defined in loop. Simplify def stmt
17535 2017-06-11 Tom de Vries <tom@codesourcery.com>
17538 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
17540 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
17541 nvptx_cannot_force_const_mem.
17543 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17545 * opts.c (finish_options): Move test for flag_split_stack after
17546 it has been initialized.
17548 2017-06-11 Jason Merrill <jason@redhat.com>
17550 * tree.h (id_equal): New.
17551 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
17552 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
17553 instead of strcmp of IDENTIFIER_POINTER.
17555 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17557 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
17558 (mark_all_inlined_calls_cdtor): Fix formating.
17559 (inline_transform): Rescale profile before inlining.
17561 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17563 * cgraph.h (cgraph_edge::clone): Update prototype.
17564 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
17565 (cgraph_node::create_clone): Update.
17566 (cgraph_node::create_version_clone): Update.
17567 * tree-inline.c (copy_bb): Update.
17568 (expand_call_inline): Update.
17570 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
17572 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
17573 factored out from ...
17574 (rs6000_emit_prologue): ... here.
17576 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
17578 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
17579 factored out from ...
17580 (rs6000_emit_prologue): ... here.
17582 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17584 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
17586 (handle_missing_profiles): Fix computation of tp_first_run.
17587 (counts_to_freqs): Do not touch freqs when count is 0.
17589 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
17591 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
17594 2017-06-10 Tom de Vries <tom@codesourcery.com>
17596 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
17597 attributes): Document signal effective target.
17599 2017-06-10 Tom de Vries <tom@codesourcery.com>
17601 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
17602 Document effective target stack_size.
17604 2017-06-09 David Malcolm <dmalcolm@redhat.com>
17606 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
17607 to the edit_context if they can be auto-applied.
17609 2017-06-9 Ian Lance Taylor <iant@golang.org>
17611 * opts.c (finish_options): If -fsplit-stack, disable implicit
17612 -forder-blocks-and-partition.
17613 * doc/invoke.texi (Optimize Options): Document that when using
17614 -fsplit-stack -forder-blocks-and-partition is not implicitly
17617 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
17619 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
17620 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
17621 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
17622 * builtins.def (abort, trap, unreachable): Declare cold.
17623 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
17624 * tree-core.h (ECF_COLD): New.
17625 * tree.c (set_call_expr_flags): Handle ECF_COLD.
17626 (build_common_builtin_nodes): Mark unreachable and abort as cold.
17628 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
17630 * predict.c (unlikely_executed_stmt_p): Cleanup.
17632 2017-06-09 Richard Biener <rguenther@suse.de>
17634 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
17635 model if the ref is always written to.
17637 2017-06-09 Tamar Christina <tamar.christina@arm.com>
17639 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
17641 2017-06-09 Tamar Christina <tamar.christina@arm.com>
17643 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
17646 2017-06-09 Tom de Vries <tom@codesourcery.com>
17649 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
17650 "target cannot support label values" when encountering LABEL_REF.
17652 2017-06-09 Martin Liska <mliska@suse.cz>
17654 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
17655 (gimple_gen_ic_func_profiler): Emit direct comparison
17656 of __gcov_indirect_call_callee with NULL.
17657 (gimple_gen_time_profiler): Change probability from
17658 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
17660 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
17662 * profile.c (edge_gcov_counts): Turn to pointer.
17663 (compute_branch_probabilities, compute_branch_probabilities): Update.
17664 (branch_prob): Do not clear edge_gcov_count.
17665 * profile.h (edge_gcov_counts): Turn to pointer.
17666 (edge_gcov_count): Update.
17668 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
17670 * gimple.h (gimple_check_failed): Mark cold.
17672 2017-06-09 Richard Biener <rguenther@suse.de>
17674 PR tree-optimization/66623
17675 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
17676 refactor check_reduction into two parts, properly computing
17677 whether we have to check reduction validity for outer loop
17680 2017-06-09 Richard Biener <rguenther@suse.de>
17682 PR tree-optimization/79483
17683 * graphite-scop-detection.c (order): New global.
17684 (get_order): Compute bb to order mapping that satisfies code
17685 generation constraints.
17686 (cmp_pbbs): New helper.
17687 (build_scops): Start domwalk at entry block, sort generated
17690 2017-06-09 Richard Biener <rguenther@suse.de>
17692 PR middle-end/81007
17693 * ipa-polymorphic-call.c
17694 (ipa_polymorphic_call_context::restrict_to_inner_class):
17695 Skip FIELD_DECLs with error_mark_node type.
17696 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
17699 2017-06-09 Martin Liska <mliska@suse.cz>
17701 * predict.c (struct branch_predictor): New struct.
17702 (test_prediction_value_range): New test.
17703 (predict_c_tests): New function.
17704 * selftest-run-tests.c (selftest::run_tests): Run the function.
17705 * selftest.h: Declare new tests.
17707 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
17710 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
17711 gen_add3_insn did not fail.
17712 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
17713 r0, construct that number in a temporary reg and add that reg to r0.
17714 If asked to put the result in r0 as well, fail.
17716 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
17718 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
17719 for early expansion of vec_eqv.
17721 2017-06-08 Jakub Jelinek <jakub@redhat.com>
17723 PR middle-end/81005
17724 * ubsan.c (instrument_null): Avoid pointless code temporary.
17725 (pass_ubsan::execute): Instrument aggregate arguments of calls.
17727 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
17731 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
17734 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
17737 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17739 * predict.c (unlikely_executed_edge_p): Move ahead.
17740 (probably_never_executed_edge_p): Use it.
17742 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
17744 PR middle-end/79988
17745 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
17746 gimple_call_builtin_p call.
17748 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17750 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
17751 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
17752 rtl_check_failed_type2, rtl_check_failed_code1,
17753 rtl_check_failed_code2, rtl_check_failed_code_mode,
17754 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
17755 rtvec_check_failed_bounds, rtl_check_failed_flag,
17756 _fatal_insn_not_found, _fatal_insn): Likewise.
17757 * tree.h (tree_contains_struct_check_failed,
17758 tree_check_failed, tree_not_check_failed,
17759 tree_class_check_failed, tree_range_check_failed,
17760 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
17761 tree_vec_elt_check_failed, phi_node_elt_check_failed,
17762 tree_operand_check_failed, omp_clause_check_failed,
17763 omp_clause_operand_check_failed, omp_clause_range_check_failed):
17766 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17768 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
17769 flag_branch_probabilities.
17770 * ipa-inline.c (edge_badness): Likewise.
17771 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
17772 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
17773 * predict.c (maybe_hot_frequency_p): Likewise.
17774 (probably_never_executed): Likewise.
17775 * sched-ebb.c (schedule_ebbs): Likewise.
17776 * sched-rgn.c (find_single_block_region): Likewise.
17777 * tracer.c (tail_duplicate): Likewise.
17779 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17781 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
17782 longer requires x_flag_profile_use.
17784 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17786 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
17787 instead of flag_reorder_blocks_and_partition.
17788 * dbxout.c (dbxout_function_end): Likewise.
17789 * dwarf2out.c (gen_subprogram_die): Likewise.
17790 * haifa-sched.c (sched_create_recovery_edges): Likewise.
17791 * hw-doloop.c (reorg_loops): Likewise.
17792 * varasm.c (assemble_start_function,
17793 assemble_end_function): Likewise.
17794 (decide_function_section): Do not check for
17795 flag_reorder_blocks_and_partition.
17797 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
17799 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
17801 (chkp_get_hard_register_fake_addr_expr): Ditto.
17802 (chkp_build_addr_expr): Add check for hard reg case.
17803 (chkp_parse_array_and_component_ref): Ditto.
17804 (chkp_find_bounds_1): Ditto.
17805 (chkp_process_stmt): Don't generate bounds store for
17808 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17810 * predict.c (maybe_hot_bb_p): Do not check profile status.
17811 (maybe_hot_edge_p): Likewise.
17812 (probably_never_executed): Check for zero counts even if profile
17814 (unlikely_executed_edge_p): New function.
17815 (unlikely_executed_stmt_p): New function.
17816 (unlikely_executed_bb_p): New function.
17817 (set_even_probabilities): Use unlikely predicates.
17818 (combine_predictions_for_bb): Likewise.
17819 (predict_paths_for_bb): Likewise.
17820 (predict_paths_leading_to_edge): Likewise.
17821 (determine_unlikely_bbs): New function.
17822 (estimate_bb_frequencies): Use it.
17823 (compute_function_frequency): Use zero counts even if profile is
17825 * profile-count.h: Fix typo.
17827 2017-08-08 Julia Koval <julia.koval@intel.com>
17829 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
17830 _mm512_mask_cvtsepi16_storeu_epi8,
17831 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
17832 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
17833 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
17834 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
17835 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
17836 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
17837 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
17838 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
17839 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
17840 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
17841 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
17842 __builtin_ia32_pmovuswb256mem_mask,
17843 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
17844 __builtin_ia32_pmovwb512mem_mask): New builtins.
17846 2017-08-08 Julia Koval <julia.koval@intel.com>
17848 PR target/73350,80862
17849 * config/i386/subst.md (round): Fix round pattern.
17850 * config/i386/i386.c (ix86_erase_embedded_rounding):
17851 Fix erasing rounding for the fixed pattern.
17853 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
17855 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
17857 2017-06-08 Martin Liska <mliska@suse.cz>
17859 PR gcov-profile/80911
17860 * gcov.c (block_info::block_info): New constructor.
17862 2017-06-07 Carl Love <cel@us.ibm.com>
17864 * config/rs6000/rs6000-c: The return type of the following
17865 built-in functions was implemented as int not long long. Fix sign
17866 of return value for the unsigned version of vec_mulo and vec_mule.
17867 vector unsigned long long vec_bperm (vector unsigned long long,
17868 vector unsigned char)
17869 vector signed long long vec_mule (vector signed int,
17871 vector unsigned long long vec_mule (vector unsigned int,
17872 vector unsigned int)
17873 vector signed long long vec_mulo (vector signed int,
17875 vector unsigned long long vec_mulo (vector unsigned int,
17876 vector unsigned int)
17877 * doc/extend.texi: Fix the documentation for the built-in
17880 2017-06-07 Carl Love <cel@us.ibm.com>
17883 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
17886 2017-06-07 Carl Love <cel@us.ibm.com>
17888 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
17889 support, Generate doublehv for signed int/float for BE case only.
17891 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
17893 * doc/invoke.texi (mcx16): Rewrite.
17895 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17897 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
17898 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
17899 *mov<mode>_softfloat, and an anonymous splitter): Use
17900 nonimmediate_operand instead of rs6000_nonimmediate_operand.
17902 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17904 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
17906 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
17907 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
17908 (rs6000_debug_reg_global): Adjust.
17909 (rs6000_init_hard_regno_mode_ok): Adjust.
17910 (rs6000_dbx_register_number): Adjust.
17911 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
17912 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
17913 Remove SPE_ACC and SPEFSCR.
17914 (REG_ALLOC_ORDER): Ditto.
17915 (FRAME_POINTER_REGNUM): Change to 111.
17916 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
17917 (REG_CLASS_NAMES): Ditto.
17918 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
17919 (REGISTER_NAMES): Ditto.
17920 (ADDITIONAL_REG_NAMES): Ditto.
17921 (rs6000_reg_names): Ditto.
17922 * config/rs6000/rs6000.md: Renumber some register number
17925 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17927 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
17929 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
17930 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
17932 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
17933 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
17934 Delete the SPE high registers.
17935 (REG_ALLOC_ORDER): Ditto.
17936 (enum reg_class): Remove SPE_HIGH_REGS.
17937 (REG_CLASS_NAMES): Ditto.
17938 (REG_CLASS_CONTENTS): Delete the SPE high registers.
17939 (REGISTER_NAMES): Ditto.
17940 (rs6000_reg_names): Ditto.
17941 * doc/tm.texi.in: Remove SPE as example.
17942 * doc/tm.texi: Regenerate.
17944 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17946 * config/rs6000/8540.md (ppc8540_brinc): Delete.
17947 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
17948 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
17949 * config/rs6000/rs6000.md (type): Remove "brinc".
17951 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17953 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
17954 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
17955 * config/rs6000/linuxspe.h: Delete file.
17956 * config/rs6000/rs6000.md: Don't include spe.md.
17957 * config/rs6000/spe.h: Delete file.
17958 * config/rs6000/spe.md: Delete file.
17959 * config/rs6000/t-rs6000: Remove spe.md.
17961 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17963 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
17964 (reg_or_none500mem_operand): Delete.
17965 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
17966 instead of reg_or_none500mem_operand.
17968 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17970 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
17971 handling of SPE flags.
17972 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
17974 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
17976 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
17978 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
17979 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
17980 paired_divv2sf3): Similar.
17981 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
17982 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
17983 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
17985 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
17986 Rename the paired_* instruction patterns.
17987 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
17989 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
17990 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
17991 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
17992 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
17993 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
17994 PAIRED_VECTOR_MODE.
17995 (struct machine_function): Delete field spe_insn_chain_scanned_p.
17996 (spe_func_has_64bit_regs_p): Delete.
17997 (spe_expand_predicate_builtin): Delete.
17998 (spe_expand_evsel_builtin): Delete.
17999 (TARGET_DWARF_REGISTER_SPAN): Do not define.
18000 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
18001 (invalid_e500_subreg): Delete.
18002 (rs6000_legitimize_address): Always force_reg op2 as well, for
18003 paired single memory accesses.
18004 (rs6000_member_type_forces_blk): Delete.
18005 (rs6000_spe_function_arg): Delete.
18006 (rs6000_expand_unop_builtin): Delete SPE handling.
18007 (rs6000_expand_binop_builtin): Ditto.
18008 (spe_expand_stv_builtin): Delete.
18009 (bdesc_2arg_spe): Delete.
18010 (spe_expand_builtin): Delete.
18011 (spe_expand_predicate_builtin): Delete.
18012 (spe_expand_evsel_builtin): Delete.
18013 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
18014 (spe_init_builtins): Delete.
18015 (spe_func_has_64bit_regs_p): Delete.
18016 (savres_routine_name): Delete "info" parameter. Adjust callers.
18017 (rs6000_emit_stack_reset): Ditto.
18018 (rs6000_dwarf_register_span): Delete.
18019 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
18020 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
18021 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
18023 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
18025 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
18026 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
18027 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
18028 mulv2sf3, divv2sf3): Delete expanders.
18030 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18032 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
18034 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18036 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
18037 * config/rs6000/rs6000.c: Ditto.
18039 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18041 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
18042 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
18043 comparison_operator.
18045 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18047 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
18048 * config/rs6000/rs6000.opt: Ditto.
18049 * config/rs6000/t-rtems: Ditto.
18051 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18053 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
18054 TARGET_E500_SINGLE by 0, simplify.
18055 * config/rs6000/rs6000.c: Ditto.
18056 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
18057 (spe_build_register_parallel): Delete.
18058 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
18059 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
18060 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
18061 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
18062 (E500_CONVERT): Delete.
18063 * config/rs6000/spe.md: Remove many patterns and all define_constants.
18065 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18067 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
18068 * config/rs6000/dfp.md: Ditto.
18069 (negdd2, *negdd2_fpr): Merge.
18070 (absdd2, *absdd2_fpr): Merge.
18071 (negtd2, *negtd2_fpr): Merge.
18072 (abstd2, *abstd2_fpr): Merge.
18073 * config/rs6000/e500.h: Delete file.
18074 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
18075 TARGET_FPRS by 1 and simplify.
18076 * config/rs6000/rs6000-c.c: Ditto.
18077 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
18078 TARGET_DF_SPE by 0.
18079 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
18081 * config/rs6000/rs6000.md: Ditto.
18082 (floatdidf2, *floatdidf2_fpr): Merge.
18083 (move_from_CR_gt_bit): Delete.
18084 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
18085 (E500_CR_IOR_COMPARE): Delete.
18086 (All patterns that require !TARGET_FPRS): Delete.
18087 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
18089 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18091 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
18093 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18095 * graphds.c (add_edge): Intitialize edge's attached data.
18096 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
18097 pointer parameter. Call pointed function on each edge during
18098 graph traversing. Skip traversing the edge when the function
18100 (graphds_dfs, graphds_scc): Ditto.
18101 (for_each_edge): New parameter. Pass the new parameter to callback
18103 * graphds.h (skip_edge_callback): New function pointer type.
18104 (graphds_dfs, graphds_scc): New function pointer parameter.
18105 (graphds_edge_callback, for_each_edge): New parameter.
18107 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18109 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
18110 out code checking if runtime alias check is possible to below ...
18111 Call the new function.
18112 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
18113 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
18115 2017-06-07 Marek Polacek <polacek@redhat.com>
18118 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
18119 TYPE_OVERFLOW_WRAPS check.
18121 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18123 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
18124 if versioning is required.
18125 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
18126 peeling with the check for versioning.
18128 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18130 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
18131 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
18132 Set true to new parameter if new ssa variable is defined.
18133 (vect_gen_vector_loop_niters): Refactor. Set range information
18134 for the new vector loop bound variable.
18135 (vect_do_peeling): Ditto.
18137 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18139 * tree-affine.c (ssa.h): Include header file.
18140 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
18141 has wrapping overflow behavior.
18143 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18145 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
18147 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18149 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
18150 (tree_to_aff_combination): ... here.
18152 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18154 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
18155 reg_pressure model function.
18156 (ivopts_global_cost_for_size): Delete.
18157 (determine_set_costs, iv_ca_recount_cost): Call new model function
18158 ivopts_estimate_reg_pressure.
18160 2017-06-07 Tamar Christina <tamar.christina@arm.com>
18162 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
18163 expensive than udiv. Remove floating point cases from mod.
18165 2017-06-07 Tamar Christina <tamar.christina@arm.com>
18167 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
18168 Increase idiv cost.
18170 2017-06-07 Tamar Christina <tamar.christina@arm.com>
18172 * config/aarch64/aarch64.md
18173 (copysignsf3): Fix mask generation.
18175 2017-06-07 Jakub Jelinek <jakub@redhat.com>
18177 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
18179 (class dump_manager): Add register_dumps method.
18180 * dumpfile.c: Include langhooks.h.
18181 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
18182 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
18183 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
18184 (dump_manager::dump_register): Start with 512 entries instead of 32.
18185 (dump_manager::register_dumps): New method.
18186 * toplev.c (general_init): Instead of invoking register_dumps
18187 langhook, invoke register_dumps method on the dump manager.
18188 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
18191 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
18193 * doc/md.texi: Clarify the restrictions on a define_insn condition.
18194 Say that # requires an associated define_split to exist, and that
18195 the define_split must be suitable for use after register allocation.
18197 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
18199 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
18200 (compute_outgoing_frequencies): Also initialize zero counts.
18201 (find_many_sub_basic_blocks): Do not produce uninitialized profile
18202 around loops; preserve more of profile when nothing changes.
18204 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
18206 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
18208 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
18209 * config/arm/arm-cpu-cdata.h: Regenerate.
18210 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
18211 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
18212 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
18213 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
18214 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
18216 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
18218 * config/arm/t-rmprofile: Likewise.
18219 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
18221 2017-06-06 David S. Miller <davem@davemloft.net>
18224 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
18225 blockage if function uses alloca.
18227 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
18229 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
18230 New "uid" fields to hold pretty-print IDs of group and ref.
18231 Memory references are now identified as <group_id>:<ref_id>
18232 instead of using [random] addresses.
18233 (dump_mem_details): Simplify, no functional change.
18234 (dump_mem_ref): Simplify and make output more concise.
18235 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
18236 (find_or_create_group): Initialize group uid.
18237 (record_ref): Initialize ref uid. Improve debug output.
18238 (prune_group_by_reuse, should_issue_prefetch_p,)
18239 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
18240 (mark_nontemporal_store, determine_loop_nest_reuse):
18241 Improve debug output.
18243 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
18245 * dbgcnt.def (prefetch): New debug counter.
18246 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
18247 (schedule_prefetches): Stop issueing prefetches if debug counter
18250 2017-06-06 Tom de Vries <tom@codesourcery.com>
18252 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
18253 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
18255 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18257 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
18258 Use aarch64_reg_or_zero predicate for operand 4.
18259 (aarch64_compare_and_swap<mode> define_insn_and_split):
18260 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
18261 (aarch64_store_exclusive<mode>): Likewise for operand 2.
18263 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
18265 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
18266 (arm_compute_save_core_reg_mask): This.
18267 (thumb1_compute_save_reg_mask): Rename into ...
18268 (thumb1_compute_save_core_reg_mask): This.
18269 (arm_compute_save_reg0_reg12_mask): Adapt comment.
18270 (arm_compute_frame_layout): Likewise.
18272 2017-06-06 Richard Biener <rguenther@suse.de>
18274 PR tree-optimization/80974
18275 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
18276 keep or clear leaders SSA info.
18278 2017-06-06 Tom de Vries <tom@codesourcery.com>
18280 * config/nvptx/nvptx.c (split_mode_p): New function.
18281 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
18283 2017-06-06 Tom de Vries <tom@codesourcery.com>
18285 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
18287 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
18290 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
18293 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
18295 * shrink-wrap.c (handle_simple_exit): Update profile.
18296 (try_shrink_wrapping): Upate profile.
18298 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
18300 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
18301 (tree_guess_outgoing_edge_probabilities): New.
18302 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
18303 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
18305 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
18307 * ipa-split.c (split_function): Initialize return bb profile.
18309 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
18311 * profile.c (compute_branch_probabilities): Also initialize
18312 EXIT_BLOCK profile.
18314 2017-06-06 Richard Biener <rguenther@suse.de>
18316 PR tree-optimization/80928
18317 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
18318 (vect_analyze_loop_operations): Properly guard analysis for
18320 (vect_transform_loop): Likewise.
18321 (vect_analyze_loop_2): Also reset SLP type on PHIs.
18322 (vect_model_induction_cost): Do not cost for pure SLP.
18323 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
18324 of induction in inner loop vectorization.
18325 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
18326 (vect_get_and_check_slp_defs): Handle vect_induction_def.
18327 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
18329 (vect_analyze_slp_cost_1): Cost induction.
18330 (vect_detect_hybrid_slp_stmts): Handle PHIs.
18331 (vect_get_slp_vect_defs): Likewise.
18332 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
18333 (vect_transform_stmt): Handle SLP reductions.
18334 * tree-vectorizer.h (vectorizable_induction): Adjust.
18336 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
18338 * config/rs6000/rs6000.c (make_resolver_func): Update
18339 init_lowered_empty_function call.
18341 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
18343 * doc/invoke.texi: Document the -fprofile-abs-path option.
18344 * common.opt (fprofile-abs-path): New option.
18345 * gcov-io.h (gcov_write_filename): Declare.
18346 * gcov-io.c (gcov_write_filename): New function.
18347 * coverage.c (coverage_begin_function): Use gcov_write_filename.
18348 * profile.c (output_location): Likewise.
18350 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
18352 * shring-wrap.c: Revert accidental commit.
18354 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
18356 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
18358 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
18360 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
18362 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
18363 profile in callgraph edge.
18364 * profile-count.h (apply_probability): If THIS is 0, then result is 0
18365 (apply_scale): Likewise.
18366 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
18367 Also scale profile when inlining function with zero profile.
18368 (initialize_cfun): Update exit block profile even when it is zero.
18369 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
18370 when profile is read.
18372 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
18374 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
18375 (CLONE_*): New constants to define the processors we can generate
18376 code for with the target_clone attribute.
18377 (rs6000_clone_map): New array to identify which clone processors
18378 the current program is running on.
18379 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
18380 target_clone attribute.
18381 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
18382 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
18383 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
18384 (cpu_expand_builtin): Add support for target_clone attribute.
18385 (rs6000_valid_attribute_p): Allow "default" attribute.
18386 (get_decl_name): New debug function to simplify printing the
18387 current function name in debugging statements.
18388 (rs6000_clone_priority): New functions to support the target_clone
18389 attribute, and be able to generate code to switch between ISA 2.05
18390 through ISA 3.0 (power6 through power9).
18391 (rs6000_compare_version_priority): Likewise.
18392 (rs6000_get_function_versions_dispatcher): Likewise.
18393 (make_resolver_func): Likewise.
18394 (add_condition_to_bb): Likewise.
18395 (dispatch_function_versions): Likewise.
18396 (rs6000_generate_version_dispatcher_body): Likewise.
18397 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
18398 (fusion_gpr_load_p): Fix a spacing issue.
18399 * doc/extend.texi (Common Function Attributes): Document that the
18400 PowerPC supports the target_clone attribute.
18402 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
18404 * config/arm/arm.h: explain F symbol found in description of ARM
18405 register allocation in its legend.
18407 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
18409 * config/mips/frame-header-opt.c: Include profile-count.h.
18410 * config/riscv/riscv.c: Include profile-count.h
18412 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
18414 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
18416 (sm_set_flag_if_changed): Add bbs field.
18417 (execute_sm_if_changed_flag_set): Pass BBS.
18418 (execute_sm): Update.
18420 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18422 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
18425 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18427 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
18428 (peephole2): New peephole2 to emit the above.
18429 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
18431 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18433 * config/aarch64/aarch64.c (define_peephole2 above
18434 *sub_<shift>_<mode>): New peephole.
18436 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
18438 * config/i386/i386.c (make_resolver_func): Update.
18439 * Makefile.in: Add profile-count.h and profile-count.o
18440 * auto-profile.c (afdo_indirect_call): Update to new API.
18441 (afdo_set_bb_count): Update.
18442 (afdo_propagate_edge): Update.
18443 (afdo_propagate_circuit): Update.
18444 (afdo_calculate_branch_prob): Update.
18445 (afdo_annotate_cfg): Update.
18446 * basic-block.h: Include profile-count.h
18447 (struct edge_def): Turn count to profile_count.
18448 (struct basic_block_def): Likewie.
18449 (REG_BR_PROB_BASE): Move to profile-count.h
18450 (RDIV): Move to profile-count.h
18451 * bb-reorder.c (max_entry_count): Turn to profile_count.
18452 (find_traces): Update.
18453 (rotate_loop):Update.
18454 (connect_traces):Update.
18455 (sanitize_hot_paths):Update.
18456 * bt-load.c (migrate_btr_defs): Update.
18457 * cfg.c (RDIV): Remove.
18458 (init_flow): Use alloc_block.
18459 (alloc_block): Uninitialize count.
18460 (unchecked_make_edge): Uninitialize count.
18461 (check_bb_profile): Update.
18462 (dump_edge_info): Update.
18463 (dump_bb_info): Update.
18464 (update_bb_profile_for_threading): Update.
18465 (scale_bbs_frequencies_int): Update.
18466 (scale_bbs_frequencies_gcov_type): Update.
18467 (scale_bbs_frequencies_profile_count): New.
18468 * cfg.h (update_bb_profile_for_threading): Update.
18469 (scale_bbs_frequencies_profile_count): Declare.
18470 * cfgbuild.c (compute_outgoing_frequencies): Update.
18471 (find_many_sub_basic_blocks): Update.
18472 * cfgcleanup.c (try_forward_edges): Update.
18473 (try_crossjump_to_edge): Update.
18474 * cfgexpand.c (expand_gimple_tailcall): Update.
18475 (construct_exit_block): Update.
18476 * cfghooks.c (verify_flow_info): Update.
18477 (dump_bb_for_graph): Update.
18478 (split_edge): Update.
18479 (make_forwarder_block): Update.
18480 (duplicate_block): Update.
18481 (account_profile_record): Update.
18482 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
18483 (get_estimated_loop_iterations): Update.
18484 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
18485 (single_likely_exit): Update.
18486 * cfgloopmanip.c (scale_loop_profile): Update.
18488 (set_zero_probability): Update.
18489 (lv_adjust_loop_entry_edge): Update.
18490 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
18491 (purge_dead_edges): Update.
18492 (rtl_account_profile_record): Update.
18493 * cgraph.c (cgraph_node::create): Uninitialize count.
18494 (symbol_table::create_edge): Uninitialize count.
18495 (cgraph_update_edges_for_call_stmt_node): Update.
18496 (cgraph_edge::dump_edge_flags): Update.
18497 (cgraph_node::dump): Update.
18498 (cgraph_edge::maybe_hot_p): Update.
18499 * cgraph.h: Include profile-count.h
18500 (create_clone), create_edge, create_indirect_edge): Update.
18501 (cgraph_node): Turn count to profile_count.
18502 (cgraph_edge0: Likewise.
18503 (make_speculative, clone): Update.
18504 (create_edge): Update.
18505 (init_lowered_empty_function): Update.
18506 * cgraphclones.c (cgraph_edge::clone): Update.
18507 (duplicate_thunk_for_node): Update.
18508 (cgraph_node::create_clone): Update.
18509 * cgraphunit.c (cgraph_node::analyze): Update.
18510 (cgraph_node::expand_thunk): Update.
18511 * final.c (dump_basic_block_info): Update.
18512 * gimple-streamer-in.c (input_bb): Update.
18513 * gimple-streamer-out.c (output_bb): Update.
18514 * graphite.c (print_global_statistics): Update.
18515 (print_graphite_scop_statistics): Update.
18516 * hsa-brig.c: Include basic-block.h.
18517 * hsa-dump.c: Include basic-block.h.
18518 * hsa-gen.c (T sum_slice): Update.
18519 (convert_switch_statements):Update.
18520 * hsa-regalloc.c: Include basic-block.h.
18521 * ipa-chkp.c (chkp_produce_thunks): Update.
18522 * ipa-cp.c (struct caller_statistics): Update.
18523 (init_caller_stats): Update.
18524 (gather_caller_stats): Update.
18525 (ipcp_cloning_candidate_p): Update.
18526 (good_cloning_opportunity_p): Update.
18527 (get_info_about_necessary_edges): Update.
18528 (dump_profile_updates): Update.
18529 (update_profiling_info): Update.
18530 (update_specialized_profile): Update.
18531 (perhaps_add_new_callers): Update.
18532 (decide_about_value): Update.
18533 (ipa_cp_c_finalize): Update.
18534 * ipa-devirt.c (struct odr_type_warn_count): Update.
18535 (struct decl_warn_count): Update.
18536 (struct final_warning_record): Update.
18537 (possible_polymorphic_call_targets): Update.
18538 (ipa_devirt): Update.
18539 * ipa-fnsummary.c (redirect_to_unreachable): Update.
18540 * ipa-icf.c (sem_function::merge): Update.
18541 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
18542 * ipa-inline.c (compute_uninlined_call_time): Update.
18543 (compute_inlined_call_time): Update.
18544 (want_inline_small_function_p): Update.
18545 (want_inline_self_recursive_call_p): Update.
18546 (edge_badness): Update.
18547 (lookup_recursive_calls): Update.
18548 (recursive_inlining): Update.
18549 (inline_small_functions): Update.
18550 (dump_overall_stats): Update.
18551 (dump_inline_stats): Update.
18552 * ipa-profile.c (ipa_profile_generate_summary): Update.
18553 (ipa_propagate_frequency): Update.
18554 (ipa_profile): Update.
18555 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
18556 * ipa-utils.c (ipa_merge_profiles): Update.
18557 * loop-doloop.c (doloop_modify): Update.
18558 * loop-unroll.c (report_unroll): Update.
18559 (unroll_loop_runtime_iterations): Update.
18560 * lto-cgraph.c (lto_output_edge): Update.
18561 (lto_output_node): Update.
18562 (input_node): Update.
18563 (input_edge): Update.
18564 (merge_profile_summaries): Update.
18565 * lto-streamer-in.c (input_cfg): Update.
18566 * lto-streamer-out.c (output_cfg): Update.
18567 * mcf.c (create_fixup_graph): Update.
18568 (adjust_cfg_counts): Update.
18569 (sum_edge_counts): Update.
18570 * modulo-sched.c (sms_schedule): Update.
18571 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
18572 * predict.c (maybe_hot_count_p): Update.
18573 (probably_never_executed): Update.
18574 (dump_prediction): Update.
18575 (combine_predictions_for_bb): Update.
18576 (propagate_freq): Update.
18577 (handle_missing_profiles): Update.
18578 (counts_to_freqs): Update.
18579 (rebuild_frequencies): Update.
18580 (force_edge_cold): Update.
18581 * predict.h: Include profile-count.h
18582 (maybe_hot_count_p, counts_to_freqs): UPdate.
18583 * print-rtl-function.c: Do not include cfg.h
18584 * print-rtl.c: Include basic-block.h
18585 * profile-count.c: New file.
18586 * profile-count.h: New file.
18587 * profile.c (is_edge_inconsistent): Update.
18588 (correct_negative_edge_counts): Update.
18589 (is_inconsistent): Update.
18590 (set_bb_counts): Update.
18591 (read_profile_edge_counts): Update.
18592 (compute_frequency_overlap): Update.
18593 (compute_branch_probabilities): Update; Initialize and deinitialize
18595 (branch_prob): Update.
18596 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
18597 (edge_gcov_count): New.
18598 (bb_gcov_count): New.
18599 * shrink-wrap.c (try_shrink_wrapping): Update.
18600 * tracer.c (better_p): Update.
18601 * trans-mem.c (expand_transaction): Update.
18602 (ipa_tm_insert_irr_call): Update.
18603 (ipa_tm_insert_gettmclone_call): Update.
18604 * tree-call-cdce.c: Update.
18605 * tree-cfg.c (gimple_duplicate_sese_region): Update.
18606 (gimple_duplicate_sese_tail): Update.
18607 (gimple_account_profile_record): Update.
18608 (execute_fixup_cfg): Update.
18609 * tree-inline.c (copy_bb): Update.
18610 (copy_edges_for_bb): Update.
18611 (initialize_cfun): Update.
18612 (freqs_to_counts): Update.
18613 (copy_cfg_body): Update.
18614 (expand_call_inline): Update.
18615 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
18616 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
18617 (try_unroll_loop_completely): Update.
18618 (try_peel_loop): Update.
18619 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18620 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
18621 * tree-ssa-loop-split.c (connect_loops): Update.
18622 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
18623 * tree-ssa-reassoc.c (branch_fixup): Update.
18624 * tree-ssa-tail-merge.c (replace_block_by): Update.
18625 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
18626 (compute_path_counts): Update.
18627 (update_profile): Update.
18628 (recompute_probabilities): Update.
18629 (update_joiner_offpath_counts): Update.
18630 (estimated_freqs_path): Update.
18631 (freqs_to_counts_path): Update.
18632 (clear_counts_path): Update.
18633 (ssa_fix_duplicate_block_edges): Update.
18634 (duplicate_thread_path): Update.
18635 * tree-switch-conversion.c (case_bit_test_cmp): Update.
18636 (struct switch_conv_info): Update.
18637 * tree-tailcall.c (decrease_profile): Update.
18638 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
18639 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18640 * value-prof.c (check_counter): Update.
18641 (gimple_divmod_fixed_value): Update.
18642 (gimple_mod_pow2): Update.
18643 (gimple_mod_subtract): Update.
18644 (gimple_ic_transform): Update.
18645 (gimple_stringop_fixed_value): Update.
18646 * value-prof.h (gimple_ic): Update.
18648 2017-06-02 Carl Love <cel@us.ibm.com>
18650 * config/rs6000/rs6000-c: Add support for built-in functions
18651 vector double vec_doublee (vector signed int);
18652 vector double vec_doublee (vector unsigned int);
18653 vector double vec_doublee (vector float);
18654 vector double vec_doubleh (vector signed int);
18655 vector double vec_doubleh (vector unsigned int);
18656 vector double vec_doubleh (vector float);
18657 vector double vec_doublel (vector signed int);
18658 vector double vec_doublel (vector unsigned int);
18659 vector double vec_doublel (vector float);
18660 vector double vec_doubleo (vector signed int);
18661 vector double vec_doubleo (vector unsigned int);
18662 vector double vec_doubleo (vector float);.
18663 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
18664 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
18666 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
18667 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
18668 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
18670 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
18671 vec_doublel, vec_doubleh.
18672 * doc/extend.texi: Update the built-in documentation file for the
18673 new built-in functions.
18675 2017-06-02 David Malcolm <dmalcolm@redhat.com>
18678 * ipa-inline-analysis.c (free_growth_caches): Set
18679 edge_removal_hook_holder to NULL after removing it.
18681 2017-06-02 Sudakshina Das <sudi.das@arm.com>
18683 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
18684 comparision with zero.
18686 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
18687 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
18688 for early expansion of vec_min and vec_max builtins.
18689 (builtin_function_type): Add min/max unsigned variants to those
18690 identified as having unsigned arguments.
18692 2017-06-02 Olivier Hainque <hainque@adacore.com>
18694 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
18696 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18698 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
18699 Use VALL_F16 iterator rather than VALL.
18701 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18703 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
18704 Emit CBNZ inside loop when doing a strong exchange and comparing
18705 against zero. Generate the CC flags after the loop.
18707 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
18709 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
18710 (dl_section_ref): New.
18711 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
18712 On AIX, append an expression to subtract the size of the
18713 section length to dl_section_ref.
18715 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
18717 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
18718 for early expansion of vector absolute builtins.
18720 2017-06-02 Richard Biener <rguenther@suse.de>
18722 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
18723 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
18725 2017-06-02 Richard Biener <rguenther@suse.de>
18727 PR tree-optimization/80948
18728 * tree-tailcall.c (find_tail_calls): Track stmts to move in
18729 stmt order as well.
18731 2017-06-02 Richard Biener <rguenther@suse.de>
18733 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
18735 * tree-vect-stmts.c (process_use): Do not mark backedge defs
18736 for inductions as relevant.
18738 2017-06-02 Richard Biener <rguenther@suse.de>
18740 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
18741 (vectorizable_induction): ... this. Remove dead code.
18743 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
18745 * builtins. (expand_builtin_alloca): Remove second parameter and
18746 infer its value from the first parameter instead.
18747 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
18749 2017-06-02 Jakub Jelinek <jakub@redhat.com>
18751 PR rtl-optimization/80903
18752 * loop-doloop.c (add_test): Unshare sequence.
18754 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
18756 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
18758 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
18760 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
18762 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
18763 xlogue_layout::get_instance, logue_layout::xlogue_layout,
18764 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
18765 (xlogue_layout::get_stub_rtx): Make static.
18766 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
18767 (xlogue_layout::compute_stub_managed_regs): Rename to...
18768 (xlogue_layout::count_stub_managed_regs): ...this.
18769 (xlogue_layout::is_stub_managed_reg): New function.
18770 (xlogue_layout::m_stub_names): Rename to...
18771 (xlogue_layout::s_stub_names): ...this, make static.
18772 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
18773 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
18774 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
18775 xlogue_layout::s_stub_names): Instantiate statics.
18776 (stub_managed_regs): Remove.
18777 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
18778 (disable_call_ms2sysv_xlogues): Rename to...
18779 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
18780 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
18782 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
18783 change after reload_completed.
18784 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
18786 (ix86_expand_prologue): Likewise.
18787 (ix86_expand_epilogue): Likewise.
18788 (ix86_expand_split_stack_prologue): Likewise.
18789 (ix86_compute_frame_layout): Remove frame parameter ...
18790 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
18791 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
18793 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
18794 (ix86_frame): Move from here ...
18795 * config/i386/i386.h (ix86_frame): ... to here.
18796 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
18797 complete ix86_frame data structure instead. Remove some_ld_name.
18799 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
18801 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
18802 symbols that hold a DECL_VALUE_EXPR.
18804 2017-06-01 Martin Jambor <mjambor@suse.cz>
18806 PR tree-optimization/80898
18807 * tree-sra.c (process_subtree_disqualification): Removed.
18808 (disqualify_candidate): Do not acll
18809 process_subtree_disqualification.
18810 (subtree_mark_written_and_enqueue): New function.
18811 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
18812 RHS has been disqualified and re-queue LHS if necessary. Apart
18813 from that, ignore disqualified RHS.
18815 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18817 * config/s390/s390.c (s390_emit_epilogue): Disable early return
18818 address fetch for z10 or later.
18820 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18822 * config/arc/arc.md (tst_movb): Add guard when splitting.
18824 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18826 * config/arc/arc.c (arc_can_eliminate): Test against
18827 arc_frame_pointer_needed.
18829 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18831 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
18832 to prevent store reordering.
18833 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
18834 (type): Add block type.
18835 (stack_tie): Define special instruction to be used in
18838 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18840 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
18841 constraint. It is not valid for the pattern.
18842 (noncommutative_binary_comparison): Likewise.
18844 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18846 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
18849 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18851 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
18852 be used by the reg-alloc.
18854 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18856 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
18857 reg-alloc when having mul64 or mul32x16 instructions.
18858 (mulsidi3): Likewise.
18859 (umulsidi3): Likewise.
18860 (mulsi32x16): New pattern.
18861 (mulsi64): Likewise.
18862 (mulsidi64): Likewise.
18863 (umulsidi64): Likewise.
18864 (MUL32x16_REG): Define.
18865 (mul64_600): Use MUL32x16_REG.
18866 (mac64_600): Likewise.
18867 (umul64_600): Likewise.
18868 (umac64_600): Likewise.
18870 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
18872 * config/arc/arc.md (mulsi3_700): Make it commutative.
18874 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
18876 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
18878 (*sign_extendsidi2_insn): Likewise for movstosw.
18880 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
18882 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
18883 the type of the input discriminant value. Convert the
18884 discriminant value of signedness vary.
18886 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
18888 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
18889 Add to -Wall section.
18891 2017-06-01 Richard Biener <rguenther@suse.de>
18893 PR middle-end/66313
18894 * fold-const.c (fold_plusminus_mult_expr): If the factored
18895 factor may be zero use a wrapping type for the inner operation.
18896 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
18897 and handle moved defs.
18898 (process_assignment): Properly guard the unary op case. Return a
18899 tri-state indicating that moving the stmt before the call may allow
18900 to continue. Pass through to_move.
18901 (find_tail_calls): Handle moving unrelated defs before
18904 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
18907 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
18908 splitter result in the canonical way.
18910 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
18912 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
18913 also for 32bit target. Update insn attributes.
18914 (zero-extendsidi2 splitter): Allow all registers for operand 1.
18916 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
18918 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
18919 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
18920 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
18921 (_mm_maskz_min_ss): New intrinsics.
18923 2017-05-31 Martin Liska <mliska@suse.cz>
18925 * tree-vect-loop.c (vect_create_epilog_for_reduction):
18926 Change comment style to one we normally use.
18927 (vectorizable_reduction): Likewise.
18928 (vectorizable_induction): Likewise.
18929 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
18930 (vectorizable_call): Likewise.
18931 (vectorizable_simd_clone_call): Likewise.
18932 (vectorizable_conversion): Likewise.
18933 (vectorizable_assignment): Likewise.
18934 (vectorizable_shift): Likewise.
18935 (vectorizable_operation): Likewise.
18936 (vectorizable_store): Likewise.
18937 (vectorizable_load): Likewise.
18938 * tree-vectorizer.h: Likewise.
18940 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
18942 * passes.c (emergency_dump_function): New.
18943 * tree-pass.h (emergency_dump_function): Declare.
18944 * plugin.c (plugins_internal_error_function): Remove.
18945 * plugin.h (plugins_internal_error_function): Remove declaration.
18946 * toplev.c (internal_error_function): New static function. Use it...
18947 (general_init): ...here.
18949 2017-05-31 Graham Markall <graham.markall@embecosm.com>
18951 * config/arc/arc.c (arc_print_operand): Handle constant operands.
18952 (arc_rtx_costs): Add costs for new patterns.
18953 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
18954 * config/arc/predicates.md: Add _1_2_3_operand predicate.
18956 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
18958 * tree-ssa-strlen.c (get_next_strinfo): New function.
18959 (get_stridx_plus_constant): Use it.
18960 (zero_length_string): Likewise.
18961 (adjust_related_strinfos): Likewise.
18962 (adjust_last_stmt): Likewise.
18964 2017-05-31 Richard Biener <rguenther@suse.de>
18967 * config/i386/i386.c (ix86_expand_builtin): Remove assert
18968 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
18970 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
18972 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
18973 loop_vinfo argument and use of dependence distance vectors.
18974 Check instead whether the two references differ only in their
18975 initial value and assume that they have the same alignment if the
18976 difference is a multiple of the vector alignment.
18977 (vect_analyze_data_refs_alignment): Update call accordingly.
18979 2017-05-31 Martin Liska <mliska@suse.cz>
18982 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
18984 2017-05-31 Bin Cheng <bin.cheng@arm.com>
18986 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
18987 (create_intersect_range_checks): Move from ...
18988 * tree-data-ref.c (create_intersect_range_checks_index)
18989 (create_intersect_range_checks): ... to here.
18990 (create_runtime_alias_checks): New function factored from ...
18991 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
18992 here. Call above function.
18993 * tree-data-ref.h (create_runtime_alias_checks): New function.
18995 2017-05-31 Bin Cheng <bin.cheng@arm.com>
18997 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
18998 segment length for dr_b and compute it in wide_int.
19000 2017-05-31 Richard Biener <rguenther@suse.de>
19002 PR tree-optimization/80906
19003 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
19004 and pass through iv_map.
19005 (copy_bb_and_scalar_dependences): Adjust.
19006 (translate_pending_phi_nodes): Likewise.
19007 (copy_loop_close_phi_args): Handle code-generating IVs instead
19010 2017-05-30 David Malcolm <dmalcolm@redhat.com>
19012 * diagnostic-color.c (color_dict): Add "type-diff".
19013 (parse_gcc_colors): Update comment.
19014 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
19015 -fdiagnostics-show-template-tree and -fno-elide-type.
19016 (GCC_COLORS): Add type-diff to example.
19018 (-fdiagnostics-show-template-tree): New.
19019 (-fno-elide-type): New.
19020 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
19021 the pp_format_decoder callback. Call any m_format_postprocessor's
19023 (pretty_printer::pretty_printer): Initialize
19024 m_format_postprocessor.
19025 (pretty_printer::~pretty_printer): Delete any
19026 m_format_postprocessor.
19027 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
19028 (class format_postprocessor): New class.
19029 (struct pretty_printer::format_decoder): Document the new parameters.
19030 (struct pretty_printer::m_format_postprocessor): New field.
19031 * tree-diagnostic.c (default_tree_printer): Update for new
19032 bool and const char ** params.
19033 * tree-diagnostic.h (default_tree_printer): Likewise.
19035 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
19037 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
19038 (lwa_operand): Delete rs6000_gen_cell_microcode test.
19039 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
19040 rs6000_gen_cell_microcode code.
19041 (rs6000_final_prescan_insn): Delete.
19042 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
19043 "warn-cell-microcode" entries.
19044 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
19045 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
19046 throughout. Change cc_reg_not_micro_cr0_operand to
19047 cc_reg_not_cr0_operand throughout.
19048 (*extendhi<mode>2_noload): Delete.
19049 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
19050 (mwarn-cell-microcode): Delete.
19051 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
19052 -mgen-cell-microcode and -mwarn-cell-microcode.
19054 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
19057 * config/i386/constraints.md (Yd): New constraint.
19059 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
19060 and (?Yd, r) alternatives. Update insn attributes.
19061 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
19062 and (?*Yd, r) alternatives. Update insn attributes.
19063 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
19065 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
19067 * gimplify.c (gimplify_modify_expr): Don't create a
19068 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
19071 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
19073 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
19075 2017-05-30 Richard Biener <rguenther@suse.de>
19077 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
19078 and reduc_def fields.
19079 (STMT_VINFO_REDUC_TYPE): New define.
19080 (STMT_VINFO_REDUC_DEF): Likewise.
19081 (vect_force_simple_reduction): Adjust prototype.
19082 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
19083 (vect_is_simple_reduction): Remove check_reduction argument.
19084 (vect_force_simple_reduction): Adjust and set
19085 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
19086 (vectorizable_reduction): Do not re-do reduction analysis
19087 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
19088 * tree-parloops.c (gather_scalar_reductions): Adjust.
19090 2017-05-30 Richard Biener <rguenther@suse.de>
19092 PR middle-end/80901
19093 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
19096 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
19098 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
19099 Introduce unknown_misalignment parameter and remove vf.
19100 (vect_peeling_hash_get_lowest_cost):
19101 Pass unknown_misalignment parameter.
19102 (vect_enhance_data_refs_alignment):
19103 Fix unsupportable data ref treatment.
19105 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
19107 * tree-vect-data-refs.c (vect_get_data_access_cost):
19108 Workaround for SLP handling.
19109 (vect_enhance_data_refs_alignment):
19110 Compute costs for doing no peeling at all, compare to the best
19111 peeling costs so far and avoid peeling if cheaper.
19113 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
19115 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
19116 Return peeling info and set costs to zero for unlimited cost
19118 (vect_enhance_data_refs_alignment): Also inspect all datarefs
19119 with unknown misalignment. Compute and costs for unknown
19120 misalignment, compare them to the costs for known misalignment
19121 and choose the cheapest for peeling.
19123 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
19125 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
19126 (vect_get_peeling_costs_all_drs): Create function.
19127 (vect_peeling_hash_get_lowest_cost):
19128 Use vect_get_peeling_costs_all_drs.
19129 (vect_peeling_supportable): Create function.
19130 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
19132 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
19134 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
19135 DR_HAS_NEGATIVE_STEP.
19136 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
19137 (vect_enhance_data_refs_alignment): Use.
19138 (vect_duplicate_ssa_name_ptr_info): Use.
19139 * tree-vectorizer.h (dr_misalignment): Use.
19140 (known_alignment_for_access_p): Use.
19142 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19145 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
19147 (has_section_name): New function.
19149 2017-05-30 Martin Liska <mliska@suse.cz>
19152 * auto-profile.c (get_function_decl_from_block): Fix
19155 2017-05-30 Richard Biener <rguenther@suse.de>
19157 PR middle-end/80876
19158 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
19160 2017-05-30 Martin Liska <mliska@suse.cz>
19162 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
19163 * dumpfile.h (struct dump_file_info): Remove ctors.
19165 2017-05-30 Martin Liska <mliska@suse.cz>
19167 * predict.def: Fix GNU coding style.
19169 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
19171 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
19172 Mark 'to' argument with ATTRIBUTE_UNUSED.
19174 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
19176 * config/xtensa/xtensa.c (xtensa_emit_call): Use
19177 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
19178 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
19181 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
19183 * doc/install.texi (Options specification): Restore entry of
19184 --enable-sjlj-exceptions.
19186 2017-05-27 Michael Eager <eager@eagercon.com>
19189 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
19191 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
19193 * config/microblaze/microblaze.h
19194 (FIXED_REGISTERS): Update in macro.
19195 (CALL_USED_REGISTERS): Update in macro.
19197 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
19199 * doc/install.texi: Add links to macOS binary distributions.
19201 2017-05-27 Jakub Jelinek <jakub@redhat.com>
19205 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
19207 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
19209 2017-05-26 Martin Liska <mliska@suse.cz>
19211 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
19213 2017-05-26 Martin Liska <mliska@suse.cz>
19215 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
19216 always leading ';; '.
19217 (dump_bb_info): Likewise.
19218 (brief_dump_cfg): Likewise.
19219 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
19220 * dumpfile.c: Remove usage of TDF_VERBOSE.
19221 * dumpfile.h (enum dump_kind): Likewise.
19222 (dump_gimple_bb_header): Do not use TDF_COMMENT.
19223 * print-tree.c (debug_verbose): Remove.
19224 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
19225 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
19226 * tree-diagnostic.c (default_tree_printer): Replace
19227 TDF_DIAGNOSTIC with TDF_SLIM.
19229 2017-05-26 Bin Cheng <bin.cheng@arm.com>
19231 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
19232 in parameter loop, rather than loop_vinfo.
19233 (create_intersect_range_checks): Ditto.
19234 (vect_create_cond_for_alias_checks): Update call to above functions.
19236 2017-05-26 Bin Cheng <bin.cheng@arm.com>
19238 PR tree-optimization/80815
19239 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
19240 for merging runtime alias checks. Handle negative DR_STEPs.
19242 2017-05-26 Bin Cheng <bin.cheng@arm.com>
19244 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
19246 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
19247 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
19248 out code pruning runtime alias checks.
19249 * tree-data-ref.c (prune_runtime_alias_test_list): New function
19250 factored out from above.
19251 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
19253 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
19255 (prune_runtime_alias_test_list): New decalaration.
19257 2017-05-26 Bin Cheng <bin.cheng@arm.com>
19259 * tree-vect-data-refs.c (compare_tree): Rename and move ...
19260 * tree-data-ref.c (data_ref_compare_tree): ... to here.
19261 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
19262 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
19263 (operator==, comp_dr_with_seg_len_pair): Ditto.
19264 (vect_prune_runtime_alias_test_list): Ditto.
19266 2017-05-26 Martin Liska <mliska@suse.cz>
19269 * params.def: Bound partial-inlining-entry-probability param.
19271 2017-05-26 Marek Polacek <polacek@redhat.com>
19274 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
19277 2017-05-26 Richard Biener <rguenther@suse.de>
19279 PR tree-optimization/80842
19280 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
19283 2017-05-26 Richard Biener <rguenther@suse.de>
19285 PR tree-optimization/80844
19286 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
19288 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
19290 * doc/md.texi (Machine Constraints): Update x86 family
19291 machine constraints section to match 'config/i386/constraints.md'.
19293 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
19295 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
19297 2017-05-25 Nathan Sidwell <nathan@acm.org>
19299 * doc/invoke.texi (--enable-languages): Update documentation.
19301 2017-05-25 Martin Liska <mliska@suse.cz>
19303 * dumpfile.c: Add TDF_FOLDING.
19304 * dumpfile.h (enum dump_kind): Likewise.
19305 * genmatch.c (dt_simplify::gen_1): Use it.
19307 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
19309 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
19311 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
19313 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
19314 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
19316 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
19318 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
19319 * match.pd (X == C): Rewrite it here.
19320 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
19321 with_certain_nonzero_bits2): New predicates.
19322 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
19324 2017-05-24 Nathan Sidwell <nathan@acm.org>
19326 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
19329 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
19332 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
19334 * config/powerpcspe: New port. Files are copied from the rs6000
19335 port, with "rs6000" in filenames replaced by "powerpcspe".
19337 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
19339 PR rtl-optimization/80754
19340 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
19342 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
19344 * config/sparc/sparc.md (length): Return the correct value for -mflat
19345 sibcalls to match output_sibcall.
19347 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
19351 * config/rs6000/rs6000.c (struct machine_function): Add new field
19353 (rs6000_get_separate_components): Init that field, use it.
19354 (rs6000_components_for_bb): Use the field.
19356 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
19358 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
19360 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
19362 PR middle-end/80823
19363 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
19365 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19368 * config/s390/s390.c (s390_check_qrst_address): Check incoming
19369 address against address_operand predicate.
19370 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
19372 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
19374 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
19375 parameters passed indirectly.
19377 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
19379 * config/i386/i386.md (*movdi_internal): Remove SSE4
19380 alternative 18 (?r, *v). Update insn attributes.
19381 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
19382 Update insn attributes.
19383 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
19384 Update insn attributes.
19385 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
19386 alternative 1 (r, v). Remove isa attribute.
19387 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
19388 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
19389 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
19391 2017-05-23 Tom de Vries <tom@codesourcery.com>
19393 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
19396 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
19398 * cgraphunit.c (symbol_table::process_new_functions): Update.
19399 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
19400 (inline_generate_summary): Rename to ...
19401 (ipa_fn_summary_generate): ... this one.
19402 (inline_read_summary): Rename to ...
19403 (ipa_fn_summary_read): ... this one.
19404 (inline_write_summary): Rename to ...
19405 (ipa_fn_summary_write): ... this one.
19406 (inline_free_summary): Rename to ...
19407 (ipa_free_fn_summary): ... this one.
19408 (pass_data_local_fn_summary, pass_local_fn_summary,
19409 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
19410 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
19411 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
19412 make_pass_ipa_fn_summary): New.
19413 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
19414 inline_write_summary, inline_free_summary): Remove.
19415 (ipa_free_fn_summary) : New.
19416 * ipa-inline.c (ipa_inline): Update.
19417 (pass_ipa_inline): Do not generate summaries.
19418 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
19420 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
19421 and add pass_ipa_fn_summary.
19422 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
19424 (make_pass_inline_parameters): Remove.
19426 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
19428 * omp-low.c (struct omp_context): Remove "default_kind" member.
19431 * omp-offload.c (execute_oacc_device_lower): Remove the
19432 parallelism dimensions function attributes for unparallelized
19433 OpenACC kernels constructs.
19435 2017-05-23 Martin Liska <mliska@suse.cz>
19437 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
19439 (cgraph_edge::make_speculative): Likewise.
19440 (cgraph_edge::resolve_speculation): Likewise.
19441 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
19442 (cgraph_node::dump): Likewise.
19443 * cgraph.h: Likewise.
19444 * cgraphunit.c (analyze_functions): Likewise.
19445 (symbol_table::compile): Likewise.
19446 * ipa-cp.c (print_all_lattices): Likewise.
19447 (determine_versionability): Likewise.
19448 (initialize_node_lattices): Likewise.
19449 (ipcp_verify_propagated_values): Likewise.
19450 (estimate_local_effects): Likewise.
19451 (update_profiling_info): Likewise.
19452 (create_specialized_node): Likewise.
19453 (perhaps_add_new_callers): Likewise.
19454 (decide_about_value): Likewise.
19455 (decide_whether_version_node): Likewise.
19456 (identify_dead_nodes): Likewise.
19457 (ipcp_store_bits_results): Likewise.
19458 * ipa-devirt.c (dump_targets): Likewise.
19459 (ipa_devirt): Likewise.
19460 * ipa-icf.c (sem_item::dump): Likewise.
19461 (sem_function::equals): Likewise.
19462 (sem_variable::equals): Likewise.
19463 (sem_item_optimizer::read_section): Likewise.
19464 (sem_item_optimizer::execute): Likewise.
19465 (congruence_class::dump): Likewise.
19466 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
19467 (dump_inline_summary): Likewise.
19468 (estimate_node_size_and_time): Likewise.
19469 (inline_analyze_function): Likewise.
19470 * ipa-inline-transform.c (inline_call): Likewise.
19471 * ipa-inline.c (report_inline_failed_reason): Likewise.
19472 (want_early_inline_function_p): Likewise.
19473 (edge_badness): Likewise.
19474 (update_edge_key): Likewise.
19475 (inline_small_functions): Likewise.
19476 * ipa-profile.c (ipa_profile): Likewise.
19477 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
19478 (ipa_make_edge_direct_to_target): Likewise.
19479 (remove_described_reference): Likewise.
19480 (ipa_impossible_devirt_target): Likewise.
19481 (propagate_controlled_uses): Likewise.
19482 (ipa_print_node_params): Likewise.
19483 (ipcp_transform_function): Likewise.
19484 * ipa-pure-const.c (pure_const_read_summary): Likewise.
19485 (propagate_pure_const): Likewise.
19486 * ipa-reference.c (generate_summary): Likewise.
19487 (read_write_all_from_decl): Likewise.
19488 (propagate): Likewise.
19489 (ipa_reference_read_optimization_summary): Likewise.
19490 * ipa-utils.c (ipa_merge_profiles): Likewise.
19491 * ipa.c (walk_polymorphic_call_targets): Likewise.
19492 (symbol_table::remove_unreachable_nodes): Likewise.
19493 (ipa_single_use): Likewise.
19494 * passes.c (execute_todo): Likewise.
19495 * predict.c (drop_profile): Likewise.
19496 * symtab.c (symtab_node::get_dump_name): New function.
19497 (symtab_node::dump_name): Likewise.
19498 (symtab_node::dump_asm_name): Likewise.
19499 (symtab_node::dump_references): Likewise.
19500 (symtab_node::dump_referring): Likewise.
19501 (symtab_node::dump_base): Likewise.
19502 (symtab_node::debug_symtab): Likewise.
19503 * tree-sra.c (convert_callers_for_node): Likewise.
19504 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
19505 * value-prof.c (init_node_map): Likewise.
19507 2017-05-23 Martin Liska <mliska@suse.cz>
19509 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
19510 and symtab_node::debug_symtab to symbol_table::debug.
19511 * cgraphunit.c (analyze_functions): Use the renamed function.
19512 (symbol_table::compile): Likewise.
19513 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
19514 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
19515 * passes.c (execute_todo): Likewise.
19516 * symtab.c (symbol_table::dump): New function.
19517 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
19519 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
19521 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
19522 that nonconst implies exec.
19524 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
19526 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
19527 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
19528 (inline_edge_summary_vec): Turn into ...
19529 (ipa_call_summaries): ... this one.
19530 (redirect_to_unreachable, edge_set_predicate,
19531 evaluate_properties_for_edge, inline_summary_alloc,
19532 reset_ipa_call_summary, reset_inline_summary,
19533 inline_summary_t::duplicate): Update.
19534 (inline_edge_duplication_hook): Turn to ...
19535 (ipa_call_summary_t::duplicate): ... this one.
19536 (inline_edge_removal_hook): Turn to ...
19537 (ipa_call_summary_t::remove): ... this one.
19538 (dump_inline_edge_summary): Turn to ...
19539 (dump_ipa_call_summary): ... this one.
19540 (estimate_function_body_sizes): Update.
19541 (inline_update_callee_summaries): Update.
19542 (remap_edge_change_prob): Update.
19543 (remap_edge_summaries): Update.
19544 (inline_merge_summary): Update.
19545 (do_estimate_edge_time): Update.
19546 (inline_generate_summary): Update.
19547 (inline_read_section): Update.
19548 (inline_read_summary): Update.
19549 (inline_free_summary): Update.
19550 * ipa-inline.c (can_inline_edge_p): Update.
19551 (compute_inlined_call_time): Update.
19552 (want_inline_small_function_p): Update.
19553 (edge_badness): Update.
19554 (early_inliner): Update.
19555 * ipa-inline.h (inline_edge_summary): Turn to ...
19556 (ipa_call_summary): ... this one.
19557 (ipa_call_summary_t): New class.
19558 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
19559 (ipa_call_summaries): New.
19560 (inline_edge_summary): Remove.
19561 (estimate_edge_growth): Update.
19562 * ipa-profile.c (ipa_propagate_frequency_1): Update.
19563 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
19564 * ipa-split.c (execute_split_functions): Update.
19565 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
19567 2017-05-23 Tom de Vries <tom@codesourcery.com>
19569 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
19570 attributes): Document rdrand effective target.
19572 2017-05-23 Tom de Vries <tom@codesourcery.com>
19574 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
19575 attributes): Sort alphabetically.
19577 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
19579 * config/avr/genmultilib.awk: Use gsub instead of gensub.
19581 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
19584 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
19585 V2DF/V2DI splat into two separate patterns, one that handles
19586 registers, and the other that only handles memory. Drop support
19587 for splatting from a GPR on ISA 2.07 and then splitting the
19588 splat into direct move and splat.
19589 (vsx_splat_<mode>_reg): Likewise.
19590 (vsx_splat_<mode>_mem): Likewise.
19592 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
19594 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
19596 2017-05-22 Jakub Jelinek <jakub@redhat.com>
19598 PR middle-end/80809
19599 * omp-low.c (finish_taskreg_remap): New function.
19600 (finish_taskreg_scan): If unit size of ctx->record_type
19601 is non-constant, unshare the size expression and replace
19602 decls in it with possible outer var refs.
19604 PR middle-end/80809
19605 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
19606 GOVD_SHARED rather than GOVD_PRIVATE with it.
19607 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
19608 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
19610 PR middle-end/80853
19611 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
19612 as last argument to build_outer_var_ref for pointer bases of array
19613 section reductions.
19615 2017-05-19 Martin Sebor <msebor@redhat.com>
19617 * print-tree.c (print_node): Print DECL_READ_P flag.
19619 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19621 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
19622 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
19623 * cgraph.c: Likewise.
19624 * cgraphunit.c: Likewise.
19625 * gengtype.c: Likewise.
19626 * ipa-cp.c: Likewise.
19627 * ipa-devirt.c: Likewise.
19628 * ipa-icf.c: Likewise.
19629 * ipa-predicate.c: Likewise.
19630 * ipa-profile.c: Likewise.
19631 * ipa-prop.c: Likewise.
19632 * ipa-split.c: Likewise.
19634 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
19635 edge_predicate_pool, dump_inline_hints,
19636 inline_summary::account_size_time, redirect_to_unreachable,
19637 edge_set_predicate, set_hint_predicate,
19638 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
19639 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
19640 inline_summary_t::remove, remap_hint_predicate_after_duplication,
19641 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
19642 ipa_call_summary_t::remove, initialize_growth_caches,
19643 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
19644 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
19645 mark_modified, unmodified_parm_1, unmodified_parm,
19646 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
19647 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
19648 compute_bb_predicates, will_be_nonconstant_expr_predicate,
19649 will_be_nonconstant_predicate, record_modified_bb_info,
19650 get_minimal_bb, record_modified, param_change_prob,
19651 phi_result_unknown_predicate, predicate_for_phi_result,
19652 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
19653 estimate_function_body_sizes, compute_inline_parameters,
19654 compute_inline_parameters_for_curren, pass_data_inline_parameters,
19655 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
19656 inline_update_callee_summaries, remap_edge_change_prob,
19657 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
19658 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
19659 inline_analyze_function, inline_summary_t::insert,
19660 inline_generate_summary, read_ipa_call_summary, inline_read_section,
19661 inline_read_summary, write_ipa_call_summary, inline_write_summary,
19662 inline_free_summary): Move to ipa-fnsummary.h
19663 (predicate_t): Remove.
19664 * ipa-fnsummary.c: New file.
19665 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
19666 (enum inline_hints_vals, inline_hints, agg_position_info,
19667 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
19668 inline_summaries, ipa_call_summary, ipa_call_summary_t,
19669 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
19670 dump_inline_summary, dump_inline_hints, inline_generate_summary,
19671 inline_read_summary, inline_write_summary, inline_free_summary,
19672 inline_analyze_function, initialize_inline_failed,
19673 inline_merge_summary, inline_update_overall_summary,
19674 compute_inline_parameters): Move to ipa-fnsummary.h
19675 * ipa-fnsummary.h: New file.
19676 * ipa-inline-transform.h: Include ipa-inline.h.
19677 * ipa-inline.c: LIkewise.
19679 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19681 * ipa-inline.c (edge_badness): Use inlined_time instead of
19682 inline_summaries->get.
19684 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19686 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
19688 2017-05-22 Nathan Sidwell <nathan@acm.org>
19690 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
19691 (fdump-lang): Document 'raw' option.
19692 * dumpfile.h (TDI_tu): Delete.
19693 * dumpfile.c (dump_files): Remove translation-unit.
19694 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
19696 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
19698 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
19699 command option from $(AWK) call.
19700 * config/avr/genmultilib.awk: Simplify and rewrite so that it
19701 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
19702 [FORMAT]: Remove handling of variable.
19703 * config/avr/t-multilib: Regenerate.
19705 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19707 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
19709 (dump_inline_summary): Do not print self_time.
19710 (estimate_function_body_sizes): Do not set self_time.
19711 (compute_inline_parameters): Likewise.
19712 (inline_read_section, inline_write_summary): Do not stream self_time.
19713 * ipa-inline.h (inline_summary): Drop self_time.
19715 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
19717 * ipa-inline-analysis.c (account_size_time): Rename to ...
19718 (inline_summary::account_size_time): ... this one.
19719 (reset_ipa_call_summary): Turn to ...
19720 (ipa_call_summary::reset): ... this one.
19721 (reset_inline_summary): Turn to ...
19722 (inline_summary::reset): ... this one.
19723 (inline_summary_t::remove): Update.
19724 (inline_summary_t::duplicate): Update.
19725 (ipa_call_summary_t::remove): Update.
19726 (dump_inline_summary): Update.
19727 (estimate_function_body_sizes): Update.
19728 (compute_inline_parameters): Update.
19729 (estimate_node_size_and_time): Update.
19730 (inline_merge_summary): Update.
19731 (inline_update_overall_summary): Update.
19732 (inline_read_section): Update.
19733 (inline_write_summary): Update.
19734 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
19735 add account_size_time and reset member functions.
19736 (ipa_call_summary): Add reset function.
19737 * ipa-predicate.h (predicate::operator &): Constify.
19739 2017-05-22 Richard Biener <rguenther@suse.de>
19741 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
19743 2017-05-19 Jason Merrill <jason@redhat.com>
19745 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
19747 2017-05-19 Marek Polacek <polacek@redhat.com>
19750 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
19751 TYPE_OVERFLOW_WRAPS checks.
19753 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
19755 * tree-core.h (enum omp_clause_default_kind): Add
19756 "OMP_CLAUSE_DEFAULT_PRESENT".
19757 * tree-pretty-print.c (dump_omp_clause): Handle it.
19758 * gimplify.c (enum gimplify_omp_var_data): Add
19759 "GOVD_MAP_FORCE_PRESENT".
19760 (gimplify_adjust_omp_clauses_1): Map it to
19761 "GOMP_MAP_FORCE_PRESENT".
19762 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
19764 * gimplify.c (oacc_default_clause): Clarify.
19766 2017-05-19 Nathan Sidwell <nathan@acm.org>
19768 LANG_HOOK_REGISTER_DUMPS
19769 * toplev.c (general_init): Call register dump lang hook.
19770 * doc/invoke.texi: Document -fdump-lang option family.
19771 * dumpfile.c (dump_files): Remove class dump here.
19772 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
19773 * dumpfile.h (tree_dump_index): Remove TDI_class.
19774 * langhooks-def.h (lhd_register_dumps): Declare.
19775 (LANG_HOOKS_REGISTER_DUMPS): Define.
19776 (LANG_HOOKS_INITIALIZER): Add it.
19777 * langhooks.c (lhd_register_dumps): Define.
19778 * langhooks.h (struct lang_hooks): Add register_dumps.
19780 2017-05-19 Nathan Sidwell <nathan@acm.org>
19782 * context.h (context::set_passes): New.
19783 * context.c (context::context): Do not create pass manager.
19784 * toplev.c (general_init): Create pass manager here.
19786 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
19788 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
19789 use this splitter if two add or or instructions would also work for
19790 the constant we want to generate.
19792 2017-05-19 Richard Biener <rguenther@suse.de>
19795 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
19796 predicate evaluation.
19798 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
19800 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
19802 * ipa-inline.c (want_inline_small_function_p): Do not cast to
19805 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
19807 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
19808 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
19809 (inline_edge_summary_vec): Turn into ...
19810 (ipa_call_summaries): ... this one.
19811 (redirect_to_unreachable, edge_set_predicate,
19812 evaluate_properties_for_edge, inline_summary_alloc,
19813 reset_ipa_call_summary, reset_inline_summary,
19814 inline_summary_t::duplicate): Update.
19815 (inline_edge_duplication_hook): Turn to ...
19816 (ipa_call_summary_t::duplicate): ... this one.
19817 (inline_edge_removal_hook): Turn to ...
19818 (ipa_call_summary_t::remove): ... this one.
19819 (dump_inline_edge_summary): Turn to ...
19820 (dump_ipa_call_summary): ... this one.
19821 (estimate_function_body_sizes): Update.
19822 (inline_update_callee_summaries): Update.
19823 (remap_edge_change_prob): Update.
19824 (remap_edge_summaries): Update.
19825 (inline_merge_summary): Update.
19826 (do_estimate_edge_time): Update.
19827 (inline_generate_summary): Update.
19828 (inline_read_section): Update.
19829 (inline_read_summary): Update.
19830 (inline_free_summary): Update.
19831 * ipa-inline.c (can_inline_edge_p): Update.
19832 (compute_inlined_call_time): Update.
19833 (want_inline_small_function_p): Update.
19834 (edge_badness): Update.
19835 (early_inliner): Update.
19836 * ipa-inline.h (inline_edge_summary): Turn to ...
19837 (ipa_call_summary): ... this one.
19838 (ipa_call_summary_t): New class.
19839 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
19840 (ipa_call_summaries): New.
19841 (inline_edge_summary): Remove.
19842 (estimate_edge_growth): Update.
19843 * ipa-profile.c (ipa_propagate_frequency_1): Update.
19844 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
19845 * ipa-split.c (execute_split_functions): Update.
19846 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
19848 2017-05-19 Richard Biener <rguenther@suse.de>
19850 PR middle-end/80764
19851 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
19853 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
19855 * config/rs6000/rs6000.c (struct machine_function): Add field
19856 fpr_is_wrapped_separately.
19857 (rs6000_get_separate_components): Use 64 components. Handle the
19858 new FPR components.
19859 (rs6000_components_for_bb): Handle the FPR components.
19860 (rs6000_emit_prologue_components): Handle the FPR components.
19861 (rs6000_emit_epilogue_components): Handle the FPR components.
19862 (rs6000_set_handled_components): Handle the FPR components.
19863 (rs6000_emit_prologue): Don't output prologue code for those FPRs
19864 that are already separately shrink-wrapped.
19865 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
19866 that are already separately shrink-wrapped.
19868 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
19871 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
19874 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
19875 (define_peephole2 for Altivec d-form load): Add peepholes to catch
19876 cases where the register allocator uses a move and an offsettable
19877 memory operation to/from a FPR register on ISA 2.06/2.07.
19878 (define_peephole2 for Altivec d-form store): Likewise.
19880 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
19883 * config/i386/mmx.md (*mov<mode>_internal): Enable
19884 alternatives 11, 12, 13 and 14 also for 32bit targets.
19885 Remove alternatives 15, 16, 17 and 18.
19886 * config/i386/sse.md (vec_concatv2di): Change
19887 alternative (!x, *y) to (x, ?!*Yn).
19889 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
19891 * dumpfile.h (enum dump_kind): Remove stray comma.
19893 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
19895 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
19896 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
19897 predicate::num_conditions
19898 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
19899 (CHANGED): turn into predicate::changed.
19900 (agg_position_info): Move to ipa-predicate.h
19901 (add_condition, predicate::add_clause, predicate::operator &=,
19902 predicate::or_with, predicate::evaluate, predicate::probability,
19903 dump_condition, dump_clause, predicate::dump,
19904 predicate::remap_after_duplication, predicate::remap_after_inlining,
19905 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
19906 (evaluate_conditions_for_known_args): Update.
19907 (set_cond_stmt_execution_predicate): Update.
19908 * ipa-inline.h: Include ipa-predicate.h
19909 (condition, inline_param_summary, conditions, agg_position_info,
19910 predicate): Move to ipa-predicate.h
19911 * ipa-predicate.c: New file.
19912 * ipa-predicate.h: New file.
19914 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
19916 * final.c (leaf_function_p): Check we are not in a sequence.
19918 2017-05-18 Martin Liska <mliska@suse.cz>
19920 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
19921 * dumpfile.c (dump_register): Use new enum dump_kind.
19922 (get_dump_file_name): Likewise.
19923 (dump_enable_all): Likewise.
19924 (dump_switch_p_1): Likewise.
19925 (enable_rtl_dump_file): Remove usage of TDF_RTL.
19926 * dumpfile.h (enum dump_kind): New enum type.
19927 (struct dump_file_info): Create constructor and
19928 format fields and comments.
19929 * passes.c (pass_manager::register_one_dump_file):
19931 * statistics.c (statistics_early_init): Likewise.
19932 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
19933 TDF_TREE with TDF_SLIM.
19934 (gather_memory_references_ref): Likewise.
19936 2017-05-18 Martin Liska <mliska@suse.cz>
19938 * vec.h (struct vnull): Use it.
19940 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
19942 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
19943 (true_predicate, false_predicate, true_predicate_p,
19944 false_predicate_p): Remove.
19945 (single_cond_predicate, not_inlined_predicate): Turn to member function
19947 (add_condition): Update.
19948 (add_clause): Turn to...
19949 (predicate::add_clause): ... this one; update; allow passing NULL
19951 (and_predicates): Turn to ...
19952 (predicate::operator &=): ... this one.
19953 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
19954 (or_predicates): Turn to ...
19955 (predicate::or_with): ... this one.
19956 (evaluate_predicate): Turn to ...
19957 (predicate::evaluate): ... this one.
19958 (predicate_probability): Turn to ...
19959 (predicate::probability): ... this one.
19960 (dump_condition): Update.
19961 (dump_predicate): Turn to ...
19962 (predicate::dump): ... this one.
19963 (account_size_time): Update.
19964 (edge_set_predicate): Update.
19965 (set_hint_predicate): UPdate.
19966 (evaluate_conditions_for_known_args): Update.
19967 (evaluate_properties_for_edge): Update.
19968 (remap_predicate_after_duplication): Turn to...
19969 (predicate::remap_after_duplication): ... this one.
19970 (remap_hint_predicate_after_duplication): Update.
19971 (inline_summary_t::duplicate): UPdate.
19972 (dump_inline_edge_summary): Update.
19973 (dump_inline_summary): Update.
19974 (set_cond_stmt_execution_predicate): Update.
19975 (set_switch_stmt_execution_predicate): Update.
19976 (compute_bb_predicates): Update.
19977 (will_be_nonconstant_expr_predicate): Update.
19978 (will_be_nonconstant_predicate): Update.
19979 (phi_result_unknown_predicate): Update.
19980 (predicate_for_phi_result): Update.
19981 (array_index_predicate): Update.
19982 (estimate_function_body_sizes): Update.
19983 (estimate_node_size_and_time): Update.
19984 (estimate_ipcp_clone_size_and_time): Update.
19985 (remap_predicate): Rename to ...
19986 (predicate::remap_after_inlining): ... this one.
19987 (remap_hint_predicate): Update.
19988 (inline_merge_summary): Update.
19989 (inline_update_overall_summary): Update.
19990 (estimate_size_after_inlining): Update.
19991 (read_predicate): Rename to ...
19992 (predicate::stream_in): ... this one.
19993 (read_inline_edge_summary): Update.
19994 (write_predicate): Rename to ...
19995 (predicate::stream_out): ... this one.
19996 (write_inline_edge_summary): Update.
19997 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
19998 (clause_t): Turn to uint32_t
19999 (predicate): Turn to class; implement constructor and operators
20001 (size_time_entry): Update.
20002 (inline_summary): Update.
20003 (inline_edge_summary): Update.
20005 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
20007 * fold-const.c (fold_binary_loc): Move transformation...
20008 * match.pd (C - X CMP X): ... here.
20010 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
20012 * config/sparc/sparc.c (sparc_option_override): Set function
20013 alignment for -mcpu=niagara7 to 64 to match the I$ line.
20014 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
20016 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
20018 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
20020 2017-05-18 Marek Polacek <polacek@redhat.com>
20023 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
20024 (pass_ubsan::execute): Call gimple_assign_single_p instead of
20025 gimple_assign_load_p.
20027 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
20029 PR middle-end/80692
20030 * real.c (do_compare): Give decimal_do_compare preference over
20031 comparing just the signs.
20033 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
20035 * doc/md.texi (Canonicalization of Instructions): Describe the
20036 canonical form of instructions that inherently set a condition
20039 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
20041 PR middle-end/80775
20042 * tree-cfg.c: Move deletion of unreachable case statements to after
20043 the merging of consecutive case labels.
20045 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
20047 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
20048 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
20049 restoring of callee-saved registers.
20051 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
20053 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
20054 * config/visium/visium.c (single_set_and_flags): Likewise.
20055 * config/visium/visium.md (Substitutions): Likewise.
20057 2017-05-17 Martin Liska <mliska@suse.cz>
20059 * cfg.c: Introduce dump_flags_t type and
20060 use it instead of int type.
20062 * cfghooks.c: Likewise.
20063 * cfghooks.h (struct cfg_hooks): Likewise.
20064 * cfgrtl.c: Likewise.
20065 * cfgrtl.h: Likewise.
20066 * cgraph.c (cgraph_node::get_body): Likewise.
20067 * coretypes.h: Likewise.
20068 * domwalk.c: Likewise.
20069 * domwalk.h: Likewise.
20070 * dumpfile.c (struct dump_option_value_info): Likewise.
20071 (dump_enable_all): Likewise.
20072 (dump_switch_p_1): Likewise.
20073 (opt_info_switch_p): Likewise.
20074 * dumpfile.h (enum tree_dump_index): Likewise.
20075 (struct dump_file_info): Likewise.
20076 * genemit.c: Likewise.
20077 * generic-match-head.c: Likewise.
20078 * gengtype.c (open_base_files): Likewise.
20079 * gimple-pretty-print.c: Likewise.
20080 * gimple-pretty-print.h: Likewise.
20081 * graph.c (print_graph_cfg): Likewise.
20082 * graphite-scop-detection.c (dot_all_sese): Likewise.
20083 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
20084 * loop-unroll.c (report_unroll): Likewise.
20085 * passes.c (pass_manager::register_one_dump_file): Likewise.
20086 * print-tree.c: Likewise.
20087 * statistics.c: Likewise.
20088 * tree-cfg.c: Likewise.
20089 * tree-cfg.h: Likewise.
20090 * tree-dfa.c: Likewise.
20091 * tree-dfa.h: Likewise.
20092 * tree-dump.c (dump_function): Likewise.
20093 * tree-dump.h (struct dump_info): Likewise.
20094 * tree-pretty-print.c: Likewise.
20095 * tree-pretty-print.h: Likewise.
20096 * tree-ssa-live.c: Likewise.
20097 * tree-ssa-live.h: Likewise.
20098 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
20099 * tree-vect-loop.c: Likewise.
20100 * tree-vect-slp.c: Likewise.
20102 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
20103 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20105 PR tree-optimization/80457
20106 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
20107 of all arguments to a statement as scalar_to_vec operations.
20108 (vectorizable_call): Adjust call to vect_model_simple_cost for
20110 (vectorizable_conversion): Likewise.
20111 (vectorizable_assignment): Likewise.
20112 (vectorizable_shift): Likewise.
20113 (vectorizable_operation): Likewise.
20114 (vectorizable_comparison): Likewise.
20115 (vect_is_simple_cond): Record the def types for operands.
20116 (vectorizable_condition): Likewise, call vect_model_simple_cost.
20117 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
20118 for statement argument count.
20120 2017-05-16 Carl Love <cel@us.ibm.com>
20122 * config/rs6000/rs6000-c: Add support for built-in functions
20123 vector unsigned long long vec_bperm (vector unsigned long long,
20124 vector unsigned char)
20125 vector signed long long vec_mule (vector signed int,
20127 vector unsigned long long vec_mule (vector unsigned int,
20128 vector unsigned int)
20129 vector signed long long vec_mulo (vector signed int,
20131 vector unsigned long long vec_mulo (vector unsigned int,
20132 vector unsigned int)
20133 vector signed char vec_sldw (vector signed char,
20134 vector signed char,
20136 vector unsigned char vec_sldw (vector unsigned char,
20137 vector unsigned char,
20139 vector signed short vec_sldw (vector signed short,
20140 vector signed short,
20142 vector unsigned short vec_sldw (vector unsigned short,
20143 vector unsigned short,
20145 vector signed int vec_sldw (vector signed int,
20148 vector unsigned int vec_sldw (vector unsigned int,
20149 vector unsigned int,
20151 vector signed long long vec_sldw (vector signed long long,
20152 vector signed long long,
20154 vector unsigned long long vec_sldw (vector unsigned long long,
20155 vector unsigned long long,
20157 * config/rs6000/rs6000-c: Add support for built-in functions
20158 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
20159 * config/rs6000/altivec.h: Add defintion for vec_sldw.
20160 * doc/extend.texi: Update the built-in documentation for the
20161 new built-in functions.
20163 2017-05-16 Marek Polacek <polacek@redhat.com>
20167 * tree.c (save_expr): Don't fold the expression.
20169 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
20171 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
20172 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
20173 and (?*y,m). Update insn attributes.
20175 2017-05-16 Martin Liska <mliska@suse.cz>
20177 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
20178 flags argument of print_gimple_stmt, print_gimple_expr,
20179 print_generic_stmt and print_generic_expr.
20180 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
20181 * coretypes.h: Likewise.
20182 * except.c (dump_eh_tree): Likewise.
20183 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
20184 * gimple-pretty-print.h: Likewise.
20185 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
20186 (backprop::push_to_worklist): Likewise.
20187 (backprop::pop_from_worklist): Likewise.
20188 (backprop::process_use): Likewise.
20189 (backprop::intersect_uses): Likewise.
20190 (note_replacement): Likewise.
20191 * gimple-ssa-store-merging.c
20192 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
20193 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
20194 (pass_store_merging::execute): Likewise.
20195 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
20196 (ssa_base_cand_dump_callback): Likewise.
20197 (dump_incr_vec): Likewise.
20198 (replace_refs): Likewise.
20199 (replace_mult_candidate): Likewise.
20200 (create_add_on_incoming_edge): Likewise.
20201 (create_phi_basis): Likewise.
20202 (insert_initializers): Likewise.
20203 (all_phi_incrs_profitable): Likewise.
20204 (introduce_cast_before_cand): Likewise.
20205 (replace_one_candidate): Likewise.
20206 * gimplify.c (gimplify_expr): Likewise.
20207 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
20208 (set_rename): Likewise.
20209 (rename_uses): Likewise.
20210 (copy_loop_phi_nodes): Likewise.
20211 (add_close_phis_to_merge_points): Likewise.
20212 (copy_loop_close_phi_args): Likewise.
20213 (copy_cond_phi_args): Likewise.
20214 (graphite_copy_stmts_from_block): Likewise.
20215 (translate_pending_phi_nodes): Likewise.
20216 * graphite-poly.c (print_pdr): Likewise.
20217 (dump_gbb_cases): Likewise.
20218 (dump_gbb_conditions): Likewise.
20219 (print_scop_params): Likewise.
20220 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
20221 (build_cross_bb_scalars_use): Likewise.
20222 (gather_bbs::before_dom_children): Likewise.
20223 * hsa-dump.c (dump_hsa_immed): Likewise.
20224 * ipa-cp.c (print_ipcp_constant_value): Likewise.
20225 (get_replacement_map): Likewise.
20226 * ipa-inline-analysis.c (dump_condition): Likewise.
20227 (estimate_function_body_sizes): Likewise.
20228 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
20229 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
20230 * ipa-prop.c (ipa_dump_param): Likewise.
20231 (ipa_print_node_jump_functions_for_edge): Likewise.
20232 (ipa_modify_call_arguments): Likewise.
20233 (ipa_modify_expr): Likewise.
20234 (ipa_dump_param_adjustments): Likewise.
20235 (ipa_dump_agg_replacement_values): Likewise.
20236 (ipcp_modif_dom_walker::before_dom_children): Likewise.
20237 * ipa-pure-const.c (check_stmt): Likewise.
20238 (pass_nothrow::execute): Likewise.
20239 * ipa-split.c (execute_split_functions): Likewise.
20240 * omp-offload.c (dump_oacc_loop_part): Likewise.
20241 (dump_oacc_loop): Likewise.
20242 * trans-mem.c (tm_log_emit): Likewise.
20243 (tm_memopt_accumulate_memops): Likewise.
20244 (dump_tm_memopt_set): Likewise.
20245 (dump_tm_memopt_transform): Likewise.
20246 * tree-cfg.c (gimple_verify_flow_info): Likewise.
20247 (print_loop): Likewise.
20248 * tree-chkp-opt.c (chkp_print_addr): Likewise.
20249 (chkp_gather_checks_info): Likewise.
20250 (chkp_get_check_result): Likewise.
20251 (chkp_remove_check_if_pass): Likewise.
20252 (chkp_use_outer_bounds_if_possible): Likewise.
20253 (chkp_reduce_bounds_lifetime): Likewise.
20254 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
20255 (chkp_mark_completed_bounds): Likewise.
20256 (chkp_register_incomplete_bounds): Likewise.
20257 (chkp_mark_invalid_bounds): Likewise.
20258 (chkp_maybe_copy_and_register_bounds): Likewise.
20259 (chkp_build_returned_bound): Likewise.
20260 (chkp_get_bound_for_parm): Likewise.
20261 (chkp_build_bndldx): Likewise.
20262 (chkp_get_bounds_by_definition): Likewise.
20263 (chkp_generate_extern_var_bounds): Likewise.
20264 (chkp_get_bounds_for_decl_addr): Likewise.
20265 * tree-chrec.c (chrec_apply): Likewise.
20266 * tree-data-ref.c (dump_data_reference): Likewise.
20267 (dump_subscript): Likewise.
20268 (dump_data_dependence_relation): Likewise.
20269 (analyze_overlapping_iterations): Likewise.
20270 * tree-inline.c (expand_call_inline): Likewise.
20271 (tree_function_versioning): Likewise.
20272 * tree-into-ssa.c (dump_defs_stack): Likewise.
20273 (dump_currdefs): Likewise.
20274 (dump_names_replaced_by): Likewise.
20275 (dump_update_ssa): Likewise.
20276 (update_ssa): Likewise.
20277 * tree-object-size.c (pass_object_sizes::execute): Likewise.
20278 * tree-parloops.c (build_new_reduction): Likewise.
20279 (try_create_reduction_list): Likewise.
20280 (ref_conflicts_with_region): Likewise.
20281 (oacc_entry_exit_ok_1): Likewise.
20282 (oacc_entry_exit_single_gang): Likewise.
20283 * tree-pretty-print.h: Likewise.
20284 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
20285 (get_scalar_evolution): Likewise.
20286 (add_to_evolution): Likewise.
20287 (get_loop_exit_condition): Likewise.
20288 (analyze_evolution_in_loop): Likewise.
20289 (analyze_initial_condition): Likewise.
20290 (analyze_scalar_evolution): Likewise.
20291 (instantiate_scev): Likewise.
20292 (number_of_latch_executions): Likewise.
20293 (gather_chrec_stats): Likewise.
20294 (final_value_replacement_loop): Likewise.
20295 (scev_const_prop): Likewise.
20296 * tree-sra.c (dump_access): Likewise.
20297 (disqualify_candidate): Likewise.
20298 (create_access): Likewise.
20299 (reject): Likewise.
20300 (maybe_add_sra_candidate): Likewise.
20301 (create_access_replacement): Likewise.
20302 (analyze_access_subtree): Likewise.
20303 (analyze_all_variable_accesses): Likewise.
20304 (sra_modify_assign): Likewise.
20305 (initialize_constant_pool_replacements): Likewise.
20306 (find_param_candidates): Likewise.
20307 (decide_one_param_reduction): Likewise.
20308 (replace_removed_params_ssa_names): Likewise.
20309 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
20310 * tree-ssa-copy.c (dump_copy_of): Likewise.
20311 (copy_prop_visit_cond_stmt): Likewise.
20312 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
20313 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
20314 (record_equivalences_from_stmt): Likewise.
20315 * tree-ssa-dse.c (compute_trims): Likewise.
20316 (delete_dead_call): Likewise.
20317 (delete_dead_assignment): Likewise.
20318 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
20319 (forward_propagate_into_cond): Likewise.
20320 (pass_forwprop::execute): Likewise.
20321 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
20322 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
20324 (move_computations_worker): Likewise.
20325 (execute_sm): Likewise.
20326 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
20327 (remove_exits_and_undefined_stmts): Likewise.
20328 (remove_redundant_iv_tests): Likewise.
20329 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
20330 (adjust_iv_update_pos): Likewise.
20331 * tree-ssa-math-opts.c (bswap_replace): Likewise.
20332 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
20333 (value_replacement): Likewise.
20334 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
20335 * tree-ssa-pre.c (print_pre_expr): Likewise.
20336 (get_representative_for): Likewise.
20337 (create_expression_by_pieces): Likewise.
20338 (insert_into_preds_of_block): Likewise.
20339 (eliminate_insert): Likewise.
20340 (eliminate_dom_walker::before_dom_children): Likewise.
20341 (eliminate): Likewise.
20342 (remove_dead_inserted_code): Likewise.
20343 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
20344 * tree-ssa-reassoc.c (get_rank): Likewise.
20345 (eliminate_duplicate_pair): Likewise.
20346 (eliminate_plus_minus_pair): Likewise.
20347 (eliminate_not_pairs): Likewise.
20348 (undistribute_ops_list): Likewise.
20349 (eliminate_redundant_comparison): Likewise.
20350 (update_range_test): Likewise.
20351 (optimize_range_tests_var_bound): Likewise.
20352 (optimize_vec_cond_expr): Likewise.
20353 (rewrite_expr_tree): Likewise.
20354 (rewrite_expr_tree_parallel): Likewise.
20355 (linearize_expr): Likewise.
20356 (break_up_subtract): Likewise.
20357 (linearize_expr_tree): Likewise.
20358 (attempt_builtin_powi): Likewise.
20359 (attempt_builtin_copysign): Likewise.
20360 (transform_stmt_to_copy): Likewise.
20361 (transform_stmt_to_multiply): Likewise.
20362 (dump_ops_vector): Likewise.
20363 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
20364 (print_scc): Likewise.
20365 (set_ssa_val_to): Likewise.
20366 (visit_reference_op_store): Likewise.
20367 (visit_use): Likewise.
20368 (sccvn_dom_walker::before_dom_children): Likewise.
20369 (run_scc_vn): Likewise.
20370 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
20372 (expr_hash_elt::print): Likewise.
20373 (const_and_copies::pop_to_marker): Likewise.
20374 (const_and_copies::record_const_or_copy_raw): Likewise.
20375 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
20376 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
20377 (dump_predicates): Likewise.
20378 (find_uninit_use): Likewise.
20379 (warn_uninitialized_phi): Likewise.
20380 (pass_late_warn_uninitialized::execute): Likewise.
20381 * tree-ssa.c (verify_vssa): Likewise.
20382 (verify_ssa): Likewise.
20383 (maybe_optimize_var): Likewise.
20384 * tree-vrp.c (dump_value_range): Likewise.
20385 (dump_all_value_ranges): Likewise.
20386 (dump_asserts_for): Likewise.
20387 (register_edge_assert_for_2): Likewise.
20388 (vrp_visit_cond_stmt): Likewise.
20389 (vrp_visit_switch_stmt): Likewise.
20390 (vrp_visit_stmt): Likewise.
20391 (vrp_visit_phi_node): Likewise.
20392 (simplify_cond_using_ranges_1): Likewise.
20393 (fold_predicate_in): Likewise.
20394 (evrp_dom_walker::before_dom_children): Likewise.
20395 (evrp_dom_walker::push_value_range): Likewise.
20396 (evrp_dom_walker::pop_value_range): Likewise.
20397 (execute_early_vrp): Likewise.
20399 2017-05-16 Richard Biener <rguenther@suse.de>
20401 * dwarf2out.c (loc_list_from_tree_1): Do not create
20402 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
20404 2017-05-16 Richard Biener <rguenther@suse.de>
20406 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
20408 (note_variable_value_in_expr): If we resolved the decl ref
20409 do not push to the stack.
20411 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
20413 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
20414 operations in fast-math mode.
20415 (vaddq_f16): Likewise.
20416 (vmul_f16): Likewise.
20417 (vmulq_f16): Likewise.
20418 (vsub_f16): Likewise.
20419 (vsubq_f16): Likewise.
20420 * config/arm/neon.md (add<mode>3): New.
20422 (fma:<VH:mode>3): New. Also remove outdated comment.
20425 2017-05-16 Martin Liska <mliska@suse.cz>
20429 * ipa-devirt.c (warn_types_mismatch): Fix typo.
20430 (odr_types_equivalent_p): Likewise.
20432 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
20434 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
20436 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
20439 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
20440 non-interunit SSE move alternatives with '?'.
20441 (zero-extendsidi peephole2): New peephole to skip intermediate
20442 general register in SSE zero-extend sequence.
20444 2017-05-15 Jeff Law <law@redhat.com>
20446 * reorg.c (relax_delay_slots): Create a new variable to hold
20447 the temporary target rather than clobbering TARGET_LABEL.
20449 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
20450 missing argument to extract_bit_field call.
20451 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
20453 2017-05-15 Martin Liska <mliska@suse.cz>
20456 * gcc.c (process_command): Do not allow empty argument of -o option.
20458 2017-05-15 Renlin Li <renlin.li@arm.com>
20460 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
20461 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
20462 * config/aarch64/constraints.md (Usf): Add long call check.
20463 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
20464 (call_value): Likewise.
20465 (sibcall): Likewise.
20466 (sibcall_value): Likewise.
20468 (call_value_insn): New.
20469 (sibcall_insn): Update rtx pattern.
20470 (sibcall_value_insn): Likewise.
20471 (call_internal): Remove.
20472 (call_value_internal): Likewise.
20473 (sibcall_internal): Likewise.
20474 (sibcall_value_internal): Likewise.
20475 (call_reg): Likewise.
20476 (call_symbol): Likewise.
20477 (call_value_reg): Likewise.
20478 (call_value_symbol): Likewise.
20480 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
20483 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
20485 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
20487 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
20488 compatible with CCGOCmode and with CCZmode.
20490 2017-05-14 Martin Sebor <msebor@redhat.com>
20492 PR middle-end/77671
20493 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
20494 (gimple_fold_builtin_snprintf): Same.
20495 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
20496 (gimple_fold_builtin_snprintf): Same.
20497 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
20498 of character types.
20499 (is_call_safe): New function.
20500 (try_substitute_return_value): Call it.
20501 (try_simplify_call): New function.
20502 (pass_sprintf_length::handle_gimple_call): Call it.
20504 2017-05-14 Martin Sebor <msebor@redhat.com>
20506 PR middle-end/80669
20507 * builtins.c (expand_builtin_stpncpy): Simplify.
20509 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
20511 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
20512 * config/i386/i386.h
20513 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
20514 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
20515 (struct machine_function): Add new members call_ms2sysv,
20516 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
20517 (struct machine_frame_state): New fields sp_realigned and
20518 sp_realigned_offset.
20519 * config/i386/i386.c
20520 (enum xlogue_stub): New enum.
20521 (enum xlogue_stub_sets): New enum.
20522 (class xlogue_layout): New class.
20523 (struct ix86_frame): New fields stack_realign_allocate_offset,
20524 stack_realign_offset and outlined_save_offset. Modify comments to
20525 detail stack layout when using out-of-line stubs.
20526 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
20527 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
20528 -mcall-ms2sysv-xlogues.
20529 (stub_managed_regs): New static variable.
20530 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
20531 registers managed by out-of-line stub.
20532 (disable_call_ms2sysv_xlogues): New function.
20533 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
20534 m->call_ms2sysv when appropriate and compute frame layout for
20536 (sp_valid_at, fp_valid_at): New inline functions.
20537 (choose_basereg): New function.
20538 (choose_baseaddr): Add align parameter, use choose_basereg and modify
20540 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
20541 Use align parameter of choose_baseaddr to generated aligned SSE movs
20543 (pro_epilogue_adjust_stack): Modify to track
20544 machine_frame_state::sp_realigned.
20545 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
20546 (ix86_nsaved_sseregs): Likewise.
20547 (ix86_emit_save_regs): Likewise.
20548 (ix86_emit_save_regs_using_mov): Likewise.
20549 (ix86_emit_save_sse_regs_using_mov): Likewise.
20550 (get_scratch_register_on_entry): Likewise.
20551 (gen_frame_set): New function.
20552 (gen_frame_load): Likewise.
20553 (gen_frame_store): Likewise.
20554 (emit_outlined_ms2sysv_save): Likewise.
20555 (emit_outlined_ms2sysv_restore): Likewise.
20556 (ix86_expand_prologue): Modify stack re-alignment code and call
20557 emit_outlined_ms2sysv_save when appropriate.
20558 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
20559 parameter rtx_insn *insn, which allows the function to be used to only
20560 generate the notes.
20561 (ix86_expand_epilogue): Modify validity checks of frame and stack
20562 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
20563 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
20564 * config/i386/predicates.md
20565 (save_multiple): New predicate.
20566 (restore_multiple): Likewise.
20567 * config/i386/sse.md
20568 (save_multiple<mode>): New pattern.
20569 (save_multiple_realign<mode>): Likewise.
20570 (restore_multiple<mode>): Likewise.
20571 (restore_multiple_and_return<mode>): Likewise.
20572 (restore_multiple_leave_return<mode>): Likewise.
20573 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
20575 2017-05-14 Julia Koval <julia.koval@intel.com>
20577 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
20578 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
20579 (__builtin_ia32_xsetbv): New builtins.
20580 * config/i386/i386.c (ix86_expand_special_args_builtin):
20582 (ix86_expand_builtin): Special expand for new intrinsics.
20583 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
20584 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
20585 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
20587 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20589 * cfganal.c (inverted_post_order_compute): Change argument type
20591 * cfganal.h (inverted_post_order_compute): Adjust prototype.
20592 * df-core.c (rest_of_handle_df_initialize): Adjust.
20593 (rest_of_handle_df_finish): Likewise.
20594 (df_analyze_1): Likewise.
20595 (df_analyze): Likewise.
20596 (loop_inverted_post_order_compute): Change argument to be a vec *.
20597 (df_analyze_loop): Adjust.
20598 (df_get_n_blocks): Likewise.
20599 (df_get_postorder): Likewise.
20600 * df.h (struct df_d): Change field to be a vec.
20601 * lcm.c (compute_laterin): Adjust.
20602 (compute_available): Likewise.
20603 * lra-lives.c (lra_create_live_ranges_1): Likewise.
20604 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
20605 * tree-ssa-pre.c (compute_antic): Likewise.
20607 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20609 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
20610 (depth_first_search::depth_first_search): Change structure init
20611 function to this constructor.
20612 (depth_first_search::add_bb): Rename function to this member.
20613 (depth_first_search::execute): Likewise.
20614 (flow_dfs_compute_reverse_finish): Adjust.
20616 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20618 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
20619 (longest_simple_path): Likewise.
20620 * shrink-wrap.c (spread_components): Likewise.
20621 (disqualify_problematic_components): Likewise.
20622 (emit_common_heads_for_components): Likewise.
20623 (emit_common_tails_for_components): Likewise.
20624 (insert_prologue_epilogue_for_components): Likewise.
20626 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20628 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
20631 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20633 * df-core.c (df_set_blocks): Start using auto_bitmap.
20634 (df_compact_blocks): Likewise.
20635 * df-problems.c (df_rd_confluence_n): Likewise.
20636 * df-scan.c (df_insn_rescan_all): Likewise.
20637 (df_process_deferred_rescans): Likewise.
20638 (df_update_entry_block_defs): Likewise.
20639 (df_update_exit_block_uses): Likewise.
20640 (df_entry_block_bitmap_verify): Likewise.
20641 (df_exit_block_bitmap_verify): Likewise.
20642 (df_scan_verify): Likewise.
20643 * lra-constraints.c (lra_constraints): Likewise.
20644 (undo_optional_reloads): Likewise.
20645 (lra_undo_inheritance): Likewise.
20646 * lra-remat.c (calculate_gen_cands): Likewise.
20647 (do_remat): Likewise.
20648 * lra-spills.c (assign_spill_hard_regs): Likewise.
20649 (spill_pseudos): Likewise.
20650 * tree-ssa-pre.c (bitmap_set_and): Likewise.
20651 (bitmap_set_subtract_values): Likewise.
20653 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20655 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
20656 management with auto_bitmap.
20657 (fix_inter_tick): Likewise.
20658 (fix_recovery_deps): Likewise.
20659 * ira.c (add_store_equivs): Likewise.
20660 (find_moveable_pseudos): Likewise.
20661 (split_live_ranges_for_shrink_wrap): Likewise.
20662 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
20663 (rtx_reuse_manager::seen_def_p): Likewise.
20664 (rtx_reuse_manager::set_seen_def): Likewise.
20665 * print-rtl.h (class rtx_reuse_manager): Likewise.
20667 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20669 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
20671 (migrate_btr_def): Likewise.
20672 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
20673 * df-core.c (loop_post_order_compute): Likewise.
20674 (loop_inverted_post_order_compute): Likewise.
20675 * hsa-common.h: Likewise.
20676 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
20677 * init-regs.c (initialize_uninitialized_regs): Likewise.
20678 * ipa-inline.c (resolve_noninline_speculation): Likewise.
20679 (inline_small_functions): Likewise.
20680 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
20681 * ira.c (combine_and_move_insns): Likewise.
20682 (build_insn_chain): Likewise.
20683 * loop-invariant.c (find_invariants): Likewise.
20684 * lower-subreg.c (propagate_pseudo_copies): Likewise.
20685 * predict.c (tree_predict_by_opcode): Likewise.
20686 (predict_paths_leading_to): Likewise.
20687 (predict_paths_leading_to_edge): Likewise.
20688 (estimate_loops_at_level): Likewise.
20689 (estimate_loops): Likewise.
20690 * shrink-wrap.c (try_shrink_wrapping): Likewise.
20691 (spread_components): Likewise.
20692 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
20693 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
20694 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
20695 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
20696 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
20697 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
20698 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
20699 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
20700 (mark_threaded_blocks): Likewise.
20701 (thread_through_all_blocks): Likewise.
20702 * tree-ssa.c (verify_ssa): Likewise.
20703 (execute_update_addresses_taken): Likewise.
20704 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
20706 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20708 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
20710 (post_order_compute): Likewise.
20711 (inverted_post_order_compute): Likewise.
20712 (pre_and_rev_post_order_compute_fn): Likewise.
20714 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20716 * genrecog.c (int_set::int_set): Explicitly construct our
20717 auto_vec base class.
20718 * vec.h (auto_vec::auto_vec): New constructor.
20720 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20722 * bitmap.h (class auto_bitmap): New constructor taking
20723 bitmap_obstack * argument.
20725 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20727 * bitmap.h (class auto_bitmap): Change type of m_bits to
20728 bitmap_head, and adjust ctor / dtor and member operators.
20730 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
20732 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
20733 when returned register mode doesn't match original mode.
20735 2017-05-12 Jeff Law <law@redhat.com>
20736 Jakub Jelinek <jakub@redhat.com>
20738 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
20739 we look for cc setter after the compare-elim changes.
20740 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
20741 within the vector to match what compare-elim now expects.
20742 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
20743 (xorsi3_flags, one_cmplsi2_flags): Likewise.
20745 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
20746 after the compare-elim changes.
20747 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
20748 the vector to match what compare-elim now expects.
20749 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
20750 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
20751 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
20752 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
20753 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
20755 * config/visium/visium.c (single_set_and_flags): Fix where
20756 we look for cc setter after the compare-elim changes.
20757 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
20758 with the vector to match what compare-elim now expects.
20759 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
20760 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
20761 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
20762 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
20763 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
20764 (neg<mode>2_insn_set_overflow): Likewise.
20766 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
20768 PR middle-end/79794
20769 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
20770 maybe_expand_insn call, set ops[0].target. If still set after call,
20771 set alt_rtl. Add extra arg to recursive calls.
20772 (extract_bit_field): Add alt_rtl argument. Pass to
20774 * expmed.h (extract_bit_field): Fix prototype.
20775 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
20776 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
20777 to extract_bit_field_calls.
20778 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
20779 Pass alt_rtl to extract_bit_field calls.
20780 * calls.c (store_unaligned_arguments_into_psuedos)
20781 load_register_parameters): Pass extra NULL to extract_bit_field calls.
20782 * optabs.c (maybe_legitimize_operand): Clear op->target when call
20784 * optabs.h (struct expand_operand): Add target bitfield.
20786 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
20788 * compare-elim.c (try_eliminate_compare): Canonicalize
20789 operation with embedded compare to
20790 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
20791 (set (reg) (operation)].
20793 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
20795 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
20798 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
20799 cost of adding a carry flag for ADC instruction.
20800 [case MINUS]: Ignore the cost of subtracting a carry flag
20801 for SBB instruction.
20803 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
20805 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
20807 * config/rs6000/bmiintrin.h: New file.
20808 * config/rs6000/bmi2intrin.h: New file.
20809 * config/rs6000/x86intrin.h: New file.
20811 2017-05-12 Jeff Law <law@redhat.com>
20813 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
20816 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
20818 PR middle-end/80707
20819 * tree-cfg.c: Remove cfg edges of unreachable case statements.
20821 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
20823 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
20824 early expansion of vector divide builtins.
20825 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
20826 builtins identified as having unsigned arguments.
20828 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
20830 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
20831 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
20832 expansion of vector logical operations (and, andc, or, xor,
20835 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
20837 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
20838 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
20840 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
20842 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
20843 early GIMPLE expansion of vector multiplies.
20845 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20847 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
20848 TARGET_HAVE_MOVT conditional.
20849 (movt splitter): Likewise.
20851 2017-05-12 Richard Biener <rguenther@suse.de>
20853 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
20854 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
20855 Fold all stmts not inplace.
20857 2017-05-12 Richard Biener <rguenther@suse.de>
20859 PR tree-optimization/80713
20860 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
20861 inserted_exprs bit for not removed stmts.
20863 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
20865 PR middle-end/69921
20866 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
20867 parallelized" attribute for parallelized OpenACC kernels.
20868 * omp-offload.c (execute_oacc_device_lower): Use it.
20870 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
20871 Set "oacc kernels" attribute.
20872 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
20873 parameter. Adjust all users.
20874 (oacc_fn_attrib_kernels_p): Remove function.
20875 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
20876 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
20877 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
20878 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
20879 assert "oacc kernels" attribute is set.
20881 2017-05-11 Carl Love <cel@us.ibm.com>
20883 * config/rs6000/rs6000-c: Add support for built-in functions
20884 vector unsigned char vec_popcnt (vector signed char)
20885 vector unsigned char vec_popcnt (vector unsigned char)
20886 vector unsigned short vec_popcnt (vector signed short)
20887 vector unsigned short vec_popcnt (vector unsigned short)
20888 vector unsigned int vec_popcnt (vector signed int)
20889 vector unsigned int vec_popcnt (vector unsigned int)
20890 vector unsigned long long vec_popcnt (vector signed long long)
20891 vector unsigned long long vec_popcnt (vector unsigned long long)
20892 vector signed long long vec_slo (vector signed long long,
20893 vector signed char)
20894 vector signed long long vec_slo (vector signed long long,
20895 vector unsigned char)
20896 vector unsigned long long vec_slo (vector unsigned long long,
20897 vector signed char)
20898 vector unsigned long long vec_slo (vector unsigned long long,
20899 vector unsigned char)
20900 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
20901 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
20902 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
20903 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
20904 * doc/extend.texi: Update the built-in documentation file for the
20905 new built-in functions.
20907 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
20909 * attribs.h (sorted_attr_string): Move machine independent
20910 functions for target clone support from the i386 port to common
20911 code. Rename ix86_function_versions to common_function_versions.
20912 Rename make_name to make_unique_name.
20913 (common_function_versions): Likewise.
20914 (make_unique_name): Likewise.
20915 (make_dispatcher_decl): Likewise.
20916 (is_function_default_version): Likewise.
20917 * attribs.c (attr_strcmp): Likewise.
20918 (sorted_attr_string): Likewise.
20919 (common_function_versions): Likewise.
20920 (make_unique_name): Likewise.
20921 (make_dispatcher_decl): Likewise.
20922 (is_function_default_version): Likewise.
20923 * config/i386/i386.c (attr_strcmp): Likewise.
20924 (sorted_attr_string): Likewise.
20925 (ix86_function_versions): Likewise.
20926 (make_name): Likewise.
20927 (make_dispatcher_decl): Likewise.
20928 (is_function_default_version): Likewise.
20929 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
20931 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20934 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
20935 Account for direct move costs for vec_construct of integer
20938 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
20941 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
20942 (UNSPEC_STX_ATOMIC): Ditto.
20943 (loaddi_via_sse): New insn.
20944 (storedi_via_sse): Ditto.
20945 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
20946 Update corresponding peephole2 patterns.
20947 (atomic_storedi_fpu): Ditto.
20949 2017-05-11 Julia Koval <julia.koval@intel.com>
20951 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
20952 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
20954 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
20955 (__builtin_ia32_rsqrt14ss_mask): New builtins.
20956 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
20958 2017-05-11 Nathan Sidwell <nathan@acm.org>
20960 * graphite-poly.c: Include dumpfile.h.
20962 * dumpfle.h (dump_function): Declare here ...
20963 * tree-dump.h (dump_function): ... not here.
20964 * dumpfile.c: #include tree-cfg.h.
20965 (dump_function): Move here from ...
20966 * tree-dump.c (dump_function): ... here.
20967 * gimplify.c: #include splay-tree.h, not tree-dump.h.
20968 * graphite-poly.c: Don't include tree-dump.h.
20969 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
20970 * print-tree.c: Likewise.
20971 * stor-layout.c: Likewise.
20972 * tree-nested.c: Likewise.
20974 * dumpfile.c (dump_start): Use TDF_FLAGS.
20975 (dump_enable_all): Fix TDF_KIND check thinko.
20977 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
20979 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20980 array entries to represent two legal parameterizations of the
20981 overloaded __builtin_cmpb function, as represented by the
20982 P6_OV_BUILTIN_CMPB constant.
20983 (altivec_resolve_overloaded_builtin): Add special case handling
20984 for the __builtin_cmpb function, as represented by the
20985 P6_OV_BUILTIN_CMPB constant.
20986 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
20987 (BU_P6_64BIT_2): New macro.
20988 (BU_P6_OVERLOAD_2): New macro
20989 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
20990 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
20991 (CMPB): Add overload support to represent both 32-bit and 64-bit
20992 compare-bytes function.
20993 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
20994 support for TARGET_CMPB.
20995 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
20996 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
20997 documentation of the __builtin_cmpb overloaded built-in function.
20999 2017-05-11 Richard Biener <rguenther@suse.de>
21001 PR tree-optimization/80705
21002 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
21003 bases are not vectorizable.
21005 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21007 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
21008 when counting register pressure.
21010 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21012 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
21013 (iv_ca_more_deps): Renamed to ...
21014 (iv_ca_compare_deps): ... this.
21015 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
21017 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21019 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
21021 (determine_group_iv_costs): ... here.
21022 (find_inv_vars_cb): Record inv var if it's not recorded before.
21024 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21026 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
21027 (get_shiftadd_cost): Ditto.
21029 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21031 * tree-ssa-address.c: Include header file.
21032 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
21034 (add_to_parts): Refactor.
21035 (addr_to_parts): New parameter. Update use of move_hint_to_base.
21036 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
21039 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21041 PR tree-optimization/53090
21042 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
21044 (extract_cond_operands): Detect condition with IV on both sides
21045 and return COMP_IV_EXPR_2.
21046 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
21047 (rewrite_use_compare): Simplify by removing call to function
21048 extract_cond_operands.
21050 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21052 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
21053 (extract_cond_operands): Detect condition comparing against non-
21054 invariant bound and return appropriate enum value.
21055 (find_interesting_uses_cond): Update use of extract_cond_operands.
21056 Handle its return value accordingly.
21057 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
21059 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21061 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
21062 nonlinear iv_use computation in loop invariant sensitive way.
21064 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21066 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
21067 (find_iv_candidates): Call relate_compare_use_with_all_cands.
21069 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21071 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
21072 (dump_cand): Support iv_cand.inv_exprs.
21073 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
21075 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
21078 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21080 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
21082 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
21083 as local function. Include necessary header files.
21084 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
21086 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21088 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
21090 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21092 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
21093 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
21094 RSHIFT_EXPR and BIT_NOT_EXPR.
21096 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21098 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
21099 (adjust_setup_cost): New parameter supporting round up adjustment.
21100 (struct address_cost_data): Delete.
21101 (force_expr_to_var_cost): Don't bound cost with spill_cost.
21102 (split_address_cost, ptr_difference_cost): Delete.
21103 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
21104 (struct ainc_cost_data): New struct.
21105 (get_address_cost_ainc): New function.
21106 (get_address_cost, get_computation_cost): Reimplement.
21107 (determine_group_iv_cost_address): Record inv_expr for all uses of
21109 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
21110 (iv_ca_has_deps): Reimplemented to ...
21111 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
21113 (iv_ca_extend): Call iv_ca_more_deps.
21115 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21117 * tree-ssa-address.c (struct mem_address): Move to header file.
21118 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
21119 * tree-ssa-address.h (struct mem_address): Move from C file.
21120 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
21122 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21124 * tree-affine.h (aff_combination_type): New interface.
21125 (aff_combination_zero_p): Remove static.
21126 (aff_combination_const_p): New interface.
21127 (aff_combination_singleton_var_p): New interfaces.
21129 2017-05-11 Richard Biener <rguenther@suse.de>
21131 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
21132 Skip unreachable blocks and destinations.
21133 (eliminate): Move stmt removal and fixup ...
21134 (fini_eliminate): ... here. Skip inserted exprs.
21135 (pass_pre::execute): Move fini_pre after fini_eliminate.
21136 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
21137 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
21138 PRE to get rid of dead code that has invalid SSA form and
21139 split critical edges again.
21141 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21143 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
21145 2017-05-11 Richard Biener <rguenther@suse.de>
21147 * passes.c (execute_function_todo): Verify loops if they are
21148 said to be up-to-date.
21149 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
21150 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
21152 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
21155 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
21156 handle calling assemble_external ourself.
21159 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
21160 modes with zero size. Enhance comment.
21162 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21164 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
21165 built-ins for vec_xl and vec_xst with short and char pointer
21168 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
21170 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
21171 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
21172 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
21173 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
21174 (_mm_maskz_min_round_ss): New intrinsics.
21175 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
21176 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
21177 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
21178 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
21179 (__builtin_ia32_minss_mask_round): New builtins.
21180 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
21181 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
21182 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
21184 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
21185 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
21187 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
21190 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
21192 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
21193 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
21194 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
21195 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
21196 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
21197 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
21198 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
21199 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
21200 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
21201 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
21202 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
21203 (__builtin_ia32_mulss_mask_round): New builtins.
21204 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
21205 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
21206 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
21208 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
21209 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
21211 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
21214 2017-05-10 Julia Koval <julia.koval@intel.com>
21216 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
21217 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
21218 (_mm256_setr_m128i): New intrinsics.
21220 2017-05-10 Julia Koval <julia.koval@intel.com>
21222 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
21223 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
21224 (_mm_maskz_rcp14_ss): New intrinsics.
21225 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
21226 (__builtin_ia32_rcp14ss_mask): New builtins.
21227 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
21229 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
21231 PR tree-optimization/51513
21232 * tree-cfg.c (gimple_seq_unreachable_p): New function.
21233 (assert_unreachable_fallthru_edge_p): Use it.
21234 (group_case_labels_stmt): Likewise.
21235 * tree-cfg.h: Prototype it.
21236 * stmt.c: Include cfghooks.h and tree-cfg.h.
21237 (emit_case_dispatch_table) <gap_label>: New local variable.
21238 Use it to fill dispatch table gaps.
21239 Test for default_label before updating probabilities.
21240 (expand_case) <default_label>: Remove unneeded initialization.
21241 Test for unreachable default case statement and remove its edge.
21242 Set default_label accordingly.
21243 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
21245 2017-05-10 Carl Love <cel@us.ibm.com>
21247 * config/rs6000/rs6000-c: Add support for built-in functions
21248 vector signed char vec_neg (vector signed char)
21249 vector signed short int vec_neg (vector short int)
21250 vector signed int vec_neg (vector signed int)
21251 vector signed long long vec_neg (vector signed long long)
21252 vector float vec_neg (vector float)
21253 vector double vec_neg (vector double)
21254 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
21256 * config/rs6000/altivec.h: Add define for vec_neg
21257 * doc/extend.texi: Update the built-in documentation for the
21258 new built-in functions.
21260 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21262 PR tree-optimization/77644
21263 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
21265 2017-05-10 Nathan Sidwell <nathan@acm.org>
21267 * dumpfile.h (TDI_lang_all): New.
21268 (TDF_KIND): New. Renumber others
21269 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
21271 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
21273 (get_dump_file_name): Adjust suffix generation.
21274 (dump_enable_all): Use TDF_KIND.
21275 * doc/invoke.texi (-fdump-lang-all): Document.
21277 * dumpfile.h: Tabify.
21279 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
21282 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
21283 Move member access before delete.
21285 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
21287 * tree-inline.c (expand_call_inline): Split block at stmt
21290 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
21293 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
21294 are now unused after splitting mov{sf,sd}_hardfloat.
21295 (f32_lr2): Likewise.
21296 (f32_lm): Likewise.
21297 (f32_lm2): Likewise.
21298 (f32_li): Likewise.
21299 (f32_li2): Likewise.
21300 (f32_lv): Likewise.
21301 (f32_sr): Likewise.
21302 (f32_sr2): Likewise.
21303 (f32_sm): Likewise.
21304 (f32_sm2): Likewise.
21305 (f32_si): Likewise.
21306 (f32_si2): Likewise.
21307 (f32_sv): Likewise.
21308 (f32_dm): Likewise.
21309 (f32_vsx): Likewise.
21310 (f32_av): Likewise.
21311 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
21312 For movsf, order stores so the VSX stores occur before the GPR
21313 store which encourages the register allocator to use a traditional
21314 FPR instead of a GPR. For movsd, order the stores so that the GPR
21315 store comes before the VSX stores to allow the power6 to work.
21316 This is due to the power6 not having a 32-bit integer store
21317 instruction from a FPR.
21318 (movsf_hardfloat): Likewise.
21319 (movsd_hardfloat): Likewise.
21321 2017-05-09 Martin Sebor <msebor@redhat.com>
21323 PR translation/80280
21324 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
21327 PR translation/80280
21328 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
21329 data member added in r247778.
21330 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
21332 2017-05-09 Nathan Sidwell <nathan@acm.org>
21334 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
21336 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
21339 2017-05-09 Marek Polacek <polacek@redhat.com>
21341 * doc/invoke.texi: Fix typo.
21343 2017-05-09 Richard Biener <rguenther@suse.de>
21345 * tree-vrp.c (vrp_val_is_max): Adjust comment.
21346 (vrp_val_is_min): Likewise.
21347 (set_value_range_to_value): Likewise.
21348 (set_value_range_to_nonnegative): Likewise.
21349 (gimple_assign_nonzero_p): Likewise.
21350 (gimple_stmt_nonzero_p): Likewise.
21351 (vrp_int_const_binop): Likewise. Remove unreachable case.
21352 (adjust_range_with_scev): Adjust comments.
21353 (compare_range_with_value): Likewise.
21354 (extract_range_from_phi_node): Likewise.
21355 (test_for_singularity): Likewise.
21357 2017-05-09 Richard Biener <rguenther@suse.de>
21359 * tree-vrp.c (get_single_symbol): Add assert that we don't
21360 get overflowed constants as invariant part.
21361 (compare_values_warnv): Add comment before the TREE_NO_WARNING
21362 checks. Use wi::cmp instead of recursing for integer constants.
21363 (compare_values): Just ignore whether we assumed undefined
21364 overflow instead of failing the compare.
21365 (extract_range_for_var_from_comparison_expr): Add comment before the
21366 TREE_NO_WARNING sets.
21367 (test_for_singularity): Likewise.
21368 (extract_range_from_comparison): Do not disable optimization
21369 when we assumed undefined overflow.
21370 (extract_range_basic): Remove init of unused var.
21372 2017-05-09 Richard Biener <rguenther@suse.de>
21374 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
21375 (extract_range_from_multiplicative_op_1): Adjust.
21376 (extract_range_from_binary_expr_1): Use int_const_binop.
21378 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
21381 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
21382 rs6000_store_data_bypass_p in seven define_bypass directives and
21383 in several comments.
21384 * config/rs6000/rs6000-protos.h: Add prototype for
21385 rs6000_store_data_bypass_p function.
21386 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
21387 function implements slightly different (rs6000-specific) semantics
21388 than store_data_bypass_p, returning false rather than aborting
21389 with assertion error when arguments do not satisfy the
21390 requirements of store data bypass.
21391 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
21392 rs6000_store_data_bypass_p.
21394 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
21396 * config/xtensa/xtensa-protos.h
21397 (xtensa_initial_elimination_offset): New declaration.
21398 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
21399 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
21400 macro definition, add case for FRAME_POINTER_REGNUM when
21401 FRAME_GROWS_DOWNWARD.
21402 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
21403 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
21404 xtensa_initial_elimination_offset.
21406 2017-05-08 Nathan Sidwell <nathan@acm.org>
21408 * doc/invoke.texi: Alphabetize -fdump options.
21410 2017-05-08 Martin Sebor <msebor@redhat.com>
21412 PR translation/80280
21413 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
21415 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
21417 * target.def (compute_frame_layout): New optional target hook.
21418 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
21419 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
21420 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
21422 * reload1.c (verify_initial_elim_offsets): Likewise.
21423 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
21424 (use_simple_return_p): Call arm_compute_frame_layout if needed.
21425 (arm_get_frame_offsets): Split up into this ...
21426 (arm_compute_frame_layout): ... and this function.
21428 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
21430 * config/aarch64/constraints.md (Usa): New constraint.
21431 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
21433 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
21435 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
21436 with_multilib_list after it has been checked.
21438 2017-05-08 Richard Biener <rguenther@suse.de>
21440 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
21441 (bitmap_set_subtract_values): Likewise.
21443 2017-05-08 Richard Biener <rguenther@suse.de>
21445 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
21446 (gimple_assign_nonzero): ... this and remove strict_overflow_p
21448 (gimple_stmt_nonzero_warnv_p): Rename to ...
21449 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
21451 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
21452 (extract_range_basic): Adjust, do not disable propagation on
21453 strict overflow sensitive simplification.
21454 (vrp_visit_cond_stmt): Likewise.
21456 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
21458 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
21459 body size unconditionally.
21461 2017-05-07 Jeff Law <law@redhat.com>
21464 2017-05-06 Jeff Law <law@redhat.com>
21465 PR tree-optimization/78496
21466 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
21469 PR tree-optimization/78496
21470 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
21471 (simplify_stmt_using_ranges): Call it.
21472 (vrp_dom_walker::before_dom_children): Extract equivalences
21473 from an ASSERT_EXPR with an equality comparison against a
21476 2017-05-06 Jeff Law <law@redhat.com>
21478 PR tree-optimization/78496
21479 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
21482 PR tree-optimization/78496
21483 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
21484 (simplify_stmt_using_ranges): Call it.
21485 (vrp_dom_walker::before_dom_children): Extract equivalences
21486 from an ASSERT_EXPR with an equality comparison against a
21489 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
21491 * lra-constraints.c (lra_copy_reg_equiv): New function.
21492 (split_reg): Use it to copy equivalence information from the
21493 original register to the spill register.
21495 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
21497 PR rtl-optimization/75964
21498 * simplify-rtx.c (simplify_const_relational_operation): Remove
21499 invalid handling of comparisons of integer ABS.
21501 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
21503 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
21504 initialize to zero.
21505 (init_regs): Remove declaration.
21506 (function_arg_advance_32): Initialize error_p as boolean variable.
21508 2017-05-05 Nathan Sidwell <nathan@acm.org>
21510 * store-motion.c (remove_reachable_equiv_notes): Reformat long
21511 lines. Use for (;;).
21513 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21515 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
21516 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
21517 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
21518 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
21519 VF=2 that require versioning.
21521 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21523 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
21526 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21528 * diagnostic.h (diagnostic_override_option_index): Convert from
21529 macro to inline function.
21531 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21533 * diagnostic.c (last_module_changed_p): New function.
21534 (set_last_module): New function.
21535 (diagnostic_report_current_module): Convert macro usage to
21536 the above functions.
21537 * diagnostic.h (diagnostic_context::last_module): Strengthen
21538 from const line_map * to const line_map_ordinary *.
21539 (diagnostic_last_module_changed): Delete macro.
21540 (diagnostic_set_last_module): Delete macro.
21542 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21544 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
21545 with diagnostic_report_diagnostic.
21546 (diagnostic_n_impl_richloc): Likewise.
21547 * diagnostic.h (report_diagnostic): Delete macro.
21548 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
21549 with diagnostic_report_diagnostic.
21550 * substring-locations.c (format_warning_va): Likewise.
21552 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21554 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
21555 save/restor of format_spec. Move option-printing code to...
21556 (print_option_information): ...this new function, and
21557 reimplement by simply printing to the pretty_printer,
21558 rather than appending to the format string.
21560 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21562 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
21563 handling logic into...
21564 (update_effective_level_from_pragmas): ...this new function.
21566 2017-05-04 Andrew Waterman <andrew@sifive.com>
21568 * config/riscv/riscv.opt (mstrict-align): New option.
21569 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
21570 (SLOW_UNALIGNED_ACCESS): Define.
21571 (riscv_slow_unaligned_access): Declare.
21572 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
21574 (riscv_slow_unaligned_access): New variable.
21575 (rocket_tune_info): Set slow_unaligned_access to true.
21576 (optimize_size_tune_info): Set slow_unaligned_access to false.
21577 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
21578 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
21579 (riscv_option_override): Set riscv_slow_unaligned_access.
21580 * doc/invoke.texi: Add -mstrict-align to RISC-V.
21582 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
21584 * config/riscv/riscv.md: Unify indentation.
21586 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21591 * config/rs6000/rs6000.md (u code attribute): Add FIX and
21593 (extendsi<mode>2): Add support for doing sign extension via
21594 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
21595 don't have ISA 3.0 instructions.
21596 (extendsi<mode>2 splitter): Likewise.
21597 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
21598 generate the normal insns since SImode can now go in vector
21599 registers. Disallow the special UNSPECs needed for previous
21600 machines to hide SImode being used. Add new insns
21601 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
21602 (fix_trunc<mode>si2_stfiwx): Likewise.
21603 (fix_trunc<mode>si2_internal): Likewise.
21604 (fixuns_trunc<mode>si2): Likewise.
21605 (fixuns_trunc<mode>si2_stfiwx): Likewise.
21606 (fctiw<u>z_<mode>_smallint): Likewise.
21607 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
21608 of floating point to 32-bit integer from doing a direct move to
21609 the GPR registers to do a store.
21610 (fctiwz_<mode>): Break long line.
21612 2017-05-05 Bin Cheng <bin.cheng@arm.com>
21614 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
21615 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
21616 (addr_list, addr_offset_valid_p): New.
21617 (split_address_groups): Check offset validity with above function.
21618 (gt-tree-ssa-loop-ivopts.h): Include header file.
21620 2017-05-05 Nathan Sidwell <nathan@acm.org>
21622 * config.gcc (arm*-*-*): Add missing 'fi'.
21624 2017-05-05 Steve Ellcey <sellcey@cavium.com>
21626 * doc/invoke.texi (-fopt-info): Explicitly say order of options
21627 included in -fopt-info does not matter.
21628 * doc/optinfo.texi (-fopt-info): Fix description of default
21629 behavour. Explicitly say order of options included in -fopt-info
21632 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
21634 * config.gcc: Allow combinations of aprofile and rmprofile values for
21635 --with-multilib-list.
21636 * config/arm/t-multilib: New file.
21637 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
21638 variables. Remove setting of ISA and floating-point ABI in
21639 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
21640 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
21641 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
21642 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
21644 * config/arm/t-rmprofile: Likewise except for the matches changes.
21645 * doc/install.texi (--with-multilib-list): Document the combination of
21646 aprofile and rmprofile values and warn about pitfalls in doing that.
21648 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
21650 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
21651 (movdi_aarch64): Likewise.
21653 2017-05-05 Jakub Jelinek <jakub@redhat.com>
21655 PR tree-optimization/80632
21656 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
21658 (build_arrays): Initialize it for virtual phis.
21659 (fix_phi_nodes): Use it for virtual phis.
21661 PR tree-optimization/80558
21662 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
21663 [x, y] op z into [x op, y op z] for op & or | if conditions
21666 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
21667 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21670 * config/arm/arm.md (use_literal_pool): Remove.
21671 (64-bit immediate split): No longer takes cost into consideration
21672 if arm_disable_literal_pool is enabled.
21673 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
21674 used when arm_disable_literal_pool is enabled.
21675 (arm_max_const_double_inline_cost): Remove use of
21676 arm_disable_literal_pool.
21677 (push_minipool_fix): Add assert.
21678 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
21679 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
21680 (no_literal_pool_sf_immediate): New.
21682 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21684 PR tree-optimization/80613
21685 * tree-ssa-dce.c (propagate_necessity): Remove cases for
21686 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
21688 2017-05-05 Richard Biener <rguenther@suse.de>
21690 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
21692 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
21694 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
21695 of this flag from insn conditions due to removal from r247495.
21697 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
21699 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
21701 (arm_early_store_addr_dep_ptr): Likewise.
21702 * config/arm/aarch-common-protos.h
21703 (arm_early_load_addr_dep_ptr): Add prototype.
21704 (arm_early_store_addr_dep_ptr): Likewise.
21705 * config/arm/cortex-a53.md: Add new bypasses.
21707 2017-05-05 Jakub Jelinek <jakub@redhat.com>
21709 * tree.c (next_type_uid): Change type to unsigned.
21710 (type_hash_canon): Decrement back next_type_uid if
21711 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
21712 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
21715 2017-05-04 Martin Sebor <msebor@redhat.com>
21717 * builtins.c: Fix a trivial typo in a comment.
21719 PR middle-end/79234
21720 * builtins.c (check_sizes): Adjust to handle reading past the end.
21721 Avoid printing excessive upper bound of ranges. Use %E to print
21722 tree nodes instead of converting them to %wu.
21723 (expand_builtin_memchr): New function.
21724 (compute_dest_size): Rename...
21725 (compute_objsize): ...to this.
21726 (expand_builtin_memcpy): Adjust.
21727 (expand_builtin_mempcpy): Adjust.
21728 (expand_builtin_strcat): Adjust.
21729 (expand_builtin_strcpy): Adjust.
21730 (check_strncat_sizes): Adjust.
21731 (expand_builtin_strncat): Adjust.
21732 (expand_builtin_strncpy): Adjust and simplify.
21733 (expand_builtin_memset): Adjust.
21734 (expand_builtin_bzero): Adjust.
21735 (expand_builtin_memcmp): Adjust.
21736 (expand_builtin): Handle memcmp.
21737 (maybe_emit_chk_warning): Check strncat just once.
21739 2017-05-04 Martin Sebor <msebor@redhat.com>
21741 PR preprocessor/79214
21742 PR middle-end/79222
21743 PR middle-end/79223
21744 * builtins.c (check_sizes): Add inlining context and issue
21745 warnings even when -Wno-system-headers is set.
21746 (check_strncat_sizes): Same.
21747 (expand_builtin_strncat): Same.
21748 (expand_builtin_memmove): New function.
21749 (expand_builtin_stpncpy): Same.
21750 (expand_builtin): Handle memmove and stpncpy.
21752 2017-05-04 Bin Cheng <bin.cheng@arm.com>
21754 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
21755 which is not used any more.
21757 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
21759 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
21761 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
21763 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
21764 (cortexa53_tunings): Likewise.
21765 (cortexa57_tunings): Likewise.
21766 (cortexa72_tunings): Likewise.
21767 (cortexa73_tunings): Likewise.
21769 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
21771 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
21772 Set loop alignment to 8.
21774 2017-05-04 Martin Sebor <msebor@redhat.com>
21776 PR translation/80280
21777 * builtins.c (expand_builtin_object_size): Add missing quoting to
21778 %D and like directives.
21779 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
21780 (hsa_type_for_tree_type): Same.
21781 (verify_function_arguments): Same.
21782 * symtab.c (symbol_table::change_decl_assembler_name): Same.
21783 * varasm.c (get_section): Same.
21786 2017-05-04 Martin Sebor <msebor@redhat.com>
21788 PR translation/80280
21789 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
21791 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
21793 * config/aarch64/aarch64.c (generic_addrcost_table):
21794 Change HI/TI mode setting.
21796 2017-05-04 Martin Jambor <mjambor@suse.cz>
21798 PR tree-optimization/80622
21799 * tree-sra.c (comes_initialized_p): New function.
21800 (build_accesses_from_assign): Only set write lazily when
21801 comes_initialized_p is false.
21802 (analyze_access_subtree): Use comes_initialized_p.
21803 (propagate_subaccesses_across_link): Assert !comes_initialized_p
21804 instead of testing for PARM_DECL.
21806 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21808 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
21809 constraint on operand 0 to allow more general addressing modes.
21810 Adjust output template.
21811 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
21813 * config/aarch64/aarch64-protos.h
21814 (aarch64_address_valid_for_prefetch_p): Declare prototype.
21815 * config/aarch64/constraints.md (Dp): New address constraint.
21816 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
21819 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
21821 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
21822 update use of estimate_ipcp_clone_size_and_time.
21823 (estimate_local_effects): Update use of
21824 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
21825 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
21826 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
21827 Return nonspecialized time.
21829 2017-05-04 Richard Biener <rguenther@suse.de>
21831 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
21832 for the last VUSE which def dominates the PHI. Directly call
21834 (get_continuation_for_phi_1): Remove.
21836 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
21838 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
21839 to explain the use of truncating division. Cap the number of
21840 iterations to the maximum given by nb_iterations_upper_bound,
21843 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
21845 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
21846 * configure: Regenerate.
21847 * config.in: Regenerate.
21848 * config/i386/driver-mingw32.c: new file.
21849 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
21850 * config.host: Link driver-mingw32.o on MinGW host.
21851 * doc/install.texi: Document new --enable-mingw-wildcard configure
21854 2017-05-04 Marek Polacek <polacek@redhat.com>
21856 PR tree-optimization/80612
21857 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
21859 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21860 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
21862 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
21863 (movt splitter): Likewise.
21864 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
21865 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
21866 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
21867 block for Thumb-1 with MOVT.
21868 (thumb2_legitimate_address_p): Move code block ...
21869 (can_avoid_literal_pool_for_label_p): ... into this new function.
21870 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
21872 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
21873 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
21874 "M-profile targets with the MOVT instruction".
21876 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21878 * config/arm/arm-builtins.c (arm_init_builtins): Rename
21879 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
21880 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
21882 2017-05-04 Martin Liska <mliska@suse.cz>
21884 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
21885 variable cond_code.
21887 2017-05-04 Richard Biener <rguenther@suse.de>
21889 * tree.c (array_at_struct_end_p): Handle arrays at struct
21890 end with flexarrays more conservatively. Refactor and treat
21891 arrays of arrays or aggregates more strict. Fix
21892 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
21893 * tree.c (array_at_struct_end_p): Adjust prototype.
21894 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
21895 * gimple-fold.c (get_range_strlen): Likewise.
21896 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
21898 2017-05-04 Richard Biener <rguenther@suse.de>
21900 PR tree-optimization/31130
21901 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
21903 (supports_overflow_infinity): Likewise.
21904 (is_negative_overflow_infinity): Likewise.
21905 (is_positive_overflow_infinity): Likewise.
21906 (is_overflow_infinity): Likewise.
21907 (stmt_overflow_infinity): Likewise.
21908 (overflow_infinity_range_p): Likewise.
21909 (usable_range_p): Remove as always returning true.
21910 (make_overflow_infinity): Remove.
21911 (negative_overflow_infinity): Likewise.
21912 (positive_overflow_infinity): Likewise.
21913 (avoid_overflow_infinity): Likewise.
21914 (set_value_range): Adjust accordingly.
21915 (set_value_range_to_nonnegative): Likewise, remove now unused
21916 overflow_infinity arg.
21917 (vrp_operand_equal_p): Adjust.
21918 (update_value_range): Likewise.
21919 (range_int_cst_singleton_p): Likewise.
21920 (operand_less_p): Likewise.
21921 (compare_values_warnv): Likewise.
21922 (extract_range_for_var_from_comparison_expr): Likewise.
21923 (vrp_int_const_binop): Likewise.
21924 (zero_nonzero_bits_from_vr): Likewise.
21925 (extract_range_from_multiplicative_op_1): Likewise.
21926 (extract_range_from_binary_expr_1): Likewise.
21927 (extract_range_from_unary_expr): Likewise.
21928 (extract_range_from_comparison): Likewise.
21929 (extract_range_basic): Likewise.
21930 (adjust_range_with_scev): Likewise.
21931 (compare_ranges): Likewise.
21932 (compare_range_with_value): Likewise.
21933 (dump_value_range): Likewise.
21934 (test_for_singularity): Likewise, remove strict_overflow_p parameter
21936 (simplify_cond_using_ranges): Adjust.
21938 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
21940 * brig-builtins.def: Added a builtin for class_f64.
21941 * builtin-types.def: Added a builtin type needed by class_f64.
21943 2017-05-03 Jason Merrill <jason@redhat.com>
21945 * timevar.def: Add TV_CONSTEXPR.
21947 2017-05-03 David Malcolm <dmalcolm@redhat.com>
21949 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
21951 2017-05-03 Martin Jambor <mjambor@suse.cz>
21953 * ipa-prop.c (ipa_update_after_lto_read): Removed.
21954 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
21955 * ipa-cp.c (ipcp_propagate_stage): Do not call
21956 ipa_update_after_lto_read.
21957 * ipa-inline.c (ipa_inline): Likewise.
21959 2017-05-03 Martin Jambor <mjambor@suse.cz>
21961 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
21962 tag. Added a default constructor and a destructor.
21963 (ipa_edge_args_sum_t): New class;
21964 (ipa_edge_args_sum): Declare.
21965 (ipa_edge_args_vector): Remove declaration.
21966 (IPA_EDGE_REF): Use ipa_edge_args_sum.
21967 (ipa_free_edge_args_substructures): Remove declaration.
21968 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
21969 (ipa_edge_args_info_available_for_edge_p): Likewise.
21970 * ipa-prop.c (ipa_edge_args_vector): Removed.
21971 (edge_removal_hook_holder): Likewise.
21972 (edge_duplication_hook_holder): Likewise.
21973 (ipa_edge_args_sum): New variable.
21974 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
21975 ipa_edge_args_vector.
21976 (ipa_free_edge_args_substructures): Likewise.
21977 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
21978 ipa_edge_args_vector.
21979 (ipa_edge_removal_hook): Turned into method
21980 ipa_edge_args_sum_t::remove.
21981 (ipa_edge_duplication_hook): Turned into method
21982 ipa_edge_args_sum_t::duplicate.
21983 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
21984 registering edge hooks.
21985 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
21986 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
21987 ipa_edge_args_sum instead of ipa_edge_args_vector.
21988 * ipa-profile.c (ipa_profile): Likewise.
21990 2017-05-03 Martin Jambor <mjambor@suse.cz>
21992 * symbol-summary.h (function_summary): New method exists.
21993 (function_summary::symtab_removal): Deallocate through release.
21994 (call_summary): New class.
21995 (gt_ggc_mx): New overload.
21996 (gt_pch_nx): Likewise.
21997 (gt_pch_nx): Likewise.
21999 2017-05-03 Jeff Law <law@redhat.com>
22001 PR tree-optimization/78496
22002 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
22003 from simplify_cond_using_ranges. Split off code to walk
22004 backwards through casts into ...
22005 (simplify_cond_using_ranges_2): New function.
22006 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
22007 (execute_vrp): After identifying jump threads, call
22008 simplify_cond_using_ranges_2.
22010 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
22013 * ipa-inline.h (inline_summary): Add ctor.
22014 (create_ggc): Do not use ggc_cleared_alloc.
22016 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
22017 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22019 * gcc.c (handle_braces): Support escaping in switch matching
22021 * doc/invoke.texi (Spec Files): Document it.
22022 Remove superfluous @code markup in items.
22024 2017-05-03 David Malcolm <dmalcolm@redhat.com>
22026 * diagnostic-show-locus.c (struct column_range): New struct.
22027 (get_affected_columns): New function.
22028 (get_printed_columns): New function.
22029 (struct correction): New struct.
22030 (correction::ensure_capacity): New function.
22031 (correction::ensure_terminated): New function.
22032 (struct line_corrections): New struct.
22033 (line_corrections::~line_corrections): New dtor.
22034 (line_corrections::add_hint): New function.
22035 (layout::print_trailing_fixits): Reimplement in terms of the new
22037 (selftest::test_overlapped_fixit_printing): New function.
22038 (selftest::diagnostic_show_locus_c_tests): Call it.
22040 2017-05-03 Nathan Sidwell <nathan@acm.org>
22042 Canonicalize canonical type hashing
22043 * tree.h (type_hash_canon_hash): Declare.
22044 * tree.c (type_hash_list, attribute_hash_list): Move into
22045 type_hash_canon_hash.
22046 (build_type_attribute_qual_variant): Break out hash code calc into
22047 type_hash_canon_hash.
22048 (type_hash_canon_hash): New. Generic type hash computation.
22049 (build_range_type_1, build_array_type_1, build_function_type,
22050 build_method_type_directly, build_offset_type, build_complex_type,
22051 make_vector_type): Call it.
22053 2017-05-03 Richard Biener <rguenther@suse.de>
22055 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
22056 When all DRs have unknown misaligned do not always peel
22057 when there is a store but apply the same costing model as if
22058 there were only loads.
22060 2017-05-03 Richard Biener <rguenther@suse.de>
22063 PR tree-optimization/80492
22064 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
22065 compare_base_decls returning dont-know properly.
22067 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
22069 * config/arm/iterators.md (CCSI): New mode iterator.
22070 (arch): New mode attribute.
22071 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
22072 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
22073 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
22074 code iterator for success result mode.
22075 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
22076 the corresponding new insn generators.
22078 2017-05-03 Bin Cheng <bin.cheng@arm.com>
22081 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22082 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
22084 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
22086 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
22087 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
22088 (DDR_A): Wrap DDR argument in brackets.
22089 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
22090 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
22091 (DDR_REVERSED_P): Likewise.
22093 2017-05-03 Jakub Jelinek <jakub@redhat.com>
22095 PR tree-optimization/79472
22096 * tree-switch-conversion.c (struct switch_conv_info): Add
22097 contiguous_range and default_case_nonstandard fields.
22098 (collect_switch_conv_info): Compute contiguous_range and
22099 default_case_nonstandard fields, don't clear final_bb if
22100 contiguous_range and only the default case doesn't have the required
22102 (check_all_empty_except_final): Set default_case_nonstandard instead
22103 of failing if contiguous_range and the default case doesn't have empty
22105 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
22106 and only the default case doesn't have the required constants. Skip
22108 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
22109 if default_case_nonstandard.
22110 (build_constructors): Build constant 1 just once. Assert that default
22111 values aren't inserted in between cases if contiguous_range. Skip
22113 (build_arrays): Skip virtual phis.
22114 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
22115 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
22116 Handle virtual phis.
22117 (gen_inbound_check): Handle default_case_nonstandard case.
22118 (process_switch): Adjust check_final_bb caller. Call
22119 gather_default_values with the first non-default case instead of
22120 default case if default_case_nonstandard.
22122 2017-05-02 Nathan Sidwell <nathan@acm.org>
22124 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
22125 check. Fix formatting.
22127 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
22129 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
22130 errors when comparing specialized and unspecialized times.
22132 2017-05-02 David Malcolm <dmalcolm@redhat.com>
22134 * diagnostic-show-locus.c
22135 (layout::should_print_annotation_line_p): Make private.
22136 (layout::print_annotation_line): Make private.
22137 (layout::annotation_line_showed_range_p): Make private.
22138 (layout::show_ruler): Make private.
22139 (layout::print_source_line): Make private. Pass in line and
22140 line_width, rather than calling location_get_source_line. Drop
22142 (layout::print_leading_fixits): New method.
22143 (layout::print_any_fixits): Rename to...
22144 (layout::print_trailing_fixits): ...this, and make private.
22145 Don't print newline fixits.
22146 (diagnostic_show_locus): Move logic for printing one row into...
22147 (layout::print_line): ...this new function. Move the
22148 location_get_source_line call and error-handling from
22149 print_source_line to here. Call print_leading_fixits, and rename
22150 print_any_fixits to print_trailing_fixits.
22151 (selftest::test_fixit_insert_containing_newline): Update now that
22152 newlines are partially supported.
22153 (selftest::test_fixit_insert_containing_newline_2): New test.
22154 (selftest::test_fixit_replace_containing_newline): Update comments.
22155 (selftest::diagnostic_show_locus_c_tests): Call the new test.
22156 * edit-context.c (class added_line): New class.
22157 (class edited_line): Describe newline handling in comment.
22158 (edited_line::actually_edited_p): New method.
22159 (edited_line::print_content): Delete redundant decl.
22160 (edited_line::m_predecessors): New field.
22161 (edited_file::print_content): Call edited_line::print_content.
22162 (edited_file::print_diff): Update to support newlines.
22163 (edited_file::print_diff_hunk): Likewise.
22164 (edited_file::print_run_of_changed_lines): New function.
22165 (edited_file::print_diff_line): Convert to...
22166 (print_diff_line): ...this.
22167 (edited_file::get_effective_line_count): New function.
22168 (edited_line::edited_line): Initialize new field m_predecessors.
22169 (edited_line::~edited_line): Clean up m_predecessors.
22170 (edited_line::apply_fixit): Handle newlines.
22171 (edited_line::get_effective_line_count): New function.
22172 (edited_line::print_content): New function.
22173 (edited_line::print_diff_lines): New function.
22174 (selftest::test_applying_fixits_insert_containing_newline): New
22176 (selftest::test_applying_fixits_replace_containing_newline): New
22178 (selftest::insert_line): New function.
22179 (selftest::test_applying_fixits_multiple_lines): Add example of
22181 (selftest::edit_context_c_tests): Call the new tests.
22183 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22185 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
22186 parameter cand. Update dump information.
22187 (get_computation_cost): Update uses.
22189 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22191 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
22192 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
22193 (get_computation_at, rewrite_use_address): Update use of
22194 get_computation_aff.
22196 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22198 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
22199 (get_computation): Delete.
22200 (get_computation_cost): Implement like get_computation_cost_at.
22201 Use get_computation_at.
22202 (get_computation_cost_at): Delete.
22203 (rewrite_use_nonlinear_expr): Use get_computation_at.
22204 (rewrite_use_compare, remove_unused_ivs): Ditto.
22206 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22208 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
22210 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22212 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
22213 (ivopts_global_cost_for_size): Rename parameter and update uses.
22214 (iv_ca_recount_cost): Update uses.
22215 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
22216 candidates seperately in n_invs and n_cands.
22217 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
22219 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22221 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
22222 (find_inv_vars_cb): New.
22223 (find_depends): Renamed to ...
22224 (find_inv_vars): ... this.
22225 (add_candidate_1, force_var_cost): Call find_inv_vars.
22226 (split_address_cost, determine_group_iv_cost_cond): Ditto.
22228 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22230 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
22231 inv_vars. Add inv_exprs.
22232 (struct iv_cand): Rename depends_on to inv_vars.
22233 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
22234 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
22235 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
22236 (dump_cand): Dump inv_vars.
22237 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
22238 (record_invariant, find_depends, add_candidate_1): Ditto.
22239 (set_group_iv_cost, force_var_cost): Ditto.
22240 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
22241 (get_computation_cost_at, get_computation_cost): Ditto.
22242 (determine_group_iv_cost_generic): Ditto.
22243 (determine_group_iv_cost_address): Ditto.
22244 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
22245 (determine_group_iv_costs): Ditto.
22246 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
22247 (iv_ca_set_remove_invariants): Renamed to ...
22248 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
22249 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
22250 (iv_ca_set_add_invariants): Renamed to ...
22251 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
22252 (iv_ca_set_cp): Use iv_ca_set_add_invs.
22253 (iv_ca_has_deps): Support inv_vars and inv_exprs.
22254 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
22255 (create_new_ivs): Remove useless dump.
22257 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22259 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
22261 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
22262 (iv_ca_set_no_cp, create_new_iv): Ditto.
22264 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22266 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
22268 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
22270 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
22271 function tree_check2.
22273 2017-05-02 Martin Liska <mliska@suse.cz>
22275 * doc/gcov.texi: Add missing preposition.
22276 * gcov.c (function_info::function_info): Properly fill up
22277 all member variables.
22279 2017-05-02 Tamar Christina <tamar.christina@arm.com>
22281 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
22283 2017-05-02 Tamar Christina <tamar.christina@arm.com>
22285 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
22287 2017-05-02 Martin Liska <mliska@suse.cz>
22290 * lto-streamer-in.c (lto_read_tree_1): Remove
22291 LTO_STREAMER_DEBUG.
22292 * lto-streamer.c (struct tree_hash_entry): Likewise.
22293 (struct tree_entry_hasher): Likewise.
22294 (tree_entry_hasher::hash): Likewise.
22295 (tree_entry_hasher::equal): Likewise.
22296 (lto_streamer_init): Likewise.
22297 (lto_orig_address_map): Likewise.
22298 (lto_orig_address_get): Likewise.
22299 (lto_orig_address_remove): Likewise.
22300 * lto-streamer.h: Likewise.
22301 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
22302 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
22304 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
22306 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
22307 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
22308 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
22309 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
22310 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
22311 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
22312 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
22313 (mm_maskz_sub_ss): New intrinsics.
22314 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
22315 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
22316 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
22317 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
22318 (__builtin_ia32_subss_mask_round): New builtins.
22319 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
22320 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
22321 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
22323 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
22324 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
22326 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
22329 2017-05-02 Martin Jambor <mjambor@suse.cz>
22331 PR tree-optimization/78687
22332 * tree-sra.c (access): New field parent.
22333 (process_subtree_disqualification): New function.
22334 (disqualify_candidate): Call it.
22335 (build_accesses_from_assign): Reset write flag if creating an
22337 (build_access_subtree): Fill in parent field and also prpagate
22338 down grp_write flag.
22339 (create_artificial_child_access): New parameter set_grp_write, set
22340 grp_write to its value.
22341 (propagate_subaccesses_across_link): Also propagate grp_write flag
22343 (propagate_all_subaccesses): Push the closest parent back to work
22344 queue if add_access_to_work_queue returned true.
22346 2017-05-02 Richard Biener <rguenther@suse.de>
22348 * common.opt (fstrict-overflow): Alias negative to fwrapv.
22349 * doc/invoke.texi (fstrict-overflow): Remove all traces of
22350 -fstrict-overflow documentation.
22351 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
22352 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
22353 flag_strict_overflow.
22354 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
22355 * lto-opts.c (lto_write_options): Do not stream it.
22356 * lto-wrapper.c (merge_and_complain): Do not handle it.
22357 * opts.c (default_options_table): Do not set -fstrict-overflow.
22358 (finish_options): Likewise do not clear it when sanitizing.
22359 * simplify-rtx.c (simplify_const_relational_operation): Do not
22360 test flag_strict_overflow.
22362 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
22364 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
22365 using enabled attribute.
22366 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
22367 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
22368 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
22369 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
22370 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
22371 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
22372 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
22373 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
22374 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
22375 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
22377 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
22379 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
22381 2017-05-02 Richard Biener <rguenther@suse.de>
22383 PR tree-optimization/80591
22385 2017-04-10 Richard Biener <rguenther@suse.de>
22387 * tree-ssa-structalias.c (find_func_aliases): Properly handle
22390 2017-05-02 Richard Biener <rguenther@suse.de>
22392 PR tree-optimization/80549
22393 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
22394 (cleanup_tree_cfg_noloop): Create forwarders to known loop
22395 headers if they do not have a preheader.
22397 2017-05-02 Martin Liska <mliska@suse.cz>
22400 * common.opt: Fix typo.
22401 * doc/invoke.texi: Likewise.
22403 2017-05-01 Jan Beulich <jbeulich@suse.com>
22405 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
22406 swapping, add (x,x,m,x,n) alternative.
22408 2017-05-01 Nathan Sidwell <nathan@acm.org>
22410 * calls.c (combine_pending_stack_adjustment_and_call): Remove
22411 unnecessary unadjusted_alignment check.
22413 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
22416 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
22418 * gimplify.c (gimplify_cilk_detach): New function.
22419 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
22420 * tree-core.h: Document EXPR_CILK_SPAWN.
22421 * tree.h (EXPR_CILK_SPAWN): Define.
22423 2017-05-01 David Malcolm <dmalcolm@redhat.com>
22425 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
22426 to use new fixit_hint representation, using the "replace" logic.
22427 (get_line_span_for_fixit_hint): Likewise.
22428 (layout::print_any_fixits): Likewise.
22429 (selftest::test_one_liner_many_fixits): Rename to...
22430 (selftest::test_one_liner_many_fixits_1): ...this, and update
22431 comment and expected output to reflect that the multiple fix-it
22432 hints are now consolidated into one insertion.
22433 (selftest::test_one_liner_many_fixits_2): New test.
22434 (selftest::test_diagnostic_show_locus_one_liner): Update for
22436 (selftest::test_fixit_consolidation): Update for fix-it API
22438 * diagnostic.c (print_parseable_fixits): Likewise.
22439 * edit-context.c (edited_line::m_line_events): Convert from
22440 auto_vec <line_event *> to auto_vec <line_event>.
22441 (class line_event): Convert from abstract base class to a concrete
22442 class, taking over the role of replace_event.
22443 (class insert_event): Delete.
22444 (class replace_event): Rename to class line_event. Convert to
22446 (edit_context::add_fixits): Reimplement.
22447 (edit_context::apply_insert): Delete.
22448 (edit_context::apply_replace): Rename to...
22449 (edit_context::apply_fixit): ...this. Convert to half-open range.
22450 (edited_file::apply_insert): Delete.
22451 (edited_file::apply_replace): Rename to...
22452 (edited_file::apply_fixit): ...this.
22453 (edited_line::~edited_line): Drop deletion of events.
22454 (edited_line::apply_insert): Delete.
22455 (edited_line::apply_replace): Rename to...
22456 (edited_line::apply_fixit): ...this. Convert to half-open range.
22457 Update for change to type of m_line_events.
22458 * edit-context.h (edit_context::apply_insert): Delete.
22459 (edit_context::apply_replace): Rename to...
22460 (edit_context::apply_fixit): ...this.
22462 2017-05-01 Martin Sebor <msebor@redhat.com>
22464 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
22467 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
22470 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
22471 __get_cpuid_max returns 0.
22472 (__get_cpuid_count): Ditto.
22474 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
22476 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
22477 replacement expression is another instance of one of its arguments.
22479 2017-05-01 Jakub Jelinek <jakub@redhat.com>
22482 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
22483 check for stack push/pop autoinc.
22484 * config/i386/i386.c (ix86_agi_dependent): Return false
22485 if the only reason why modified_in_p returned true is that
22486 addr is SP based and set_insn is a push or pop.
22488 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
22490 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
22493 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
22496 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
22497 (account_size_time): Use two predicates - exec_pred and
22499 (evaluate_conditions_for_known_args): Compute both clause and
22501 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
22502 (inline_summary_t::duplicate): Update.
22503 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
22505 (compute_inline_parameters): Likewise.
22506 (estimate_edge_size_and_time): Update caluclation of time.
22507 (estimate_node_size_and_time): Compute both time and nonspecialized
22509 (estimate_ipcp_clone_size_and_time): Update.
22510 (inline_merge_summary): Update.
22511 (do_estimate_edge_time): Update.
22512 (do_estimate_edge_size): Update.
22513 (do_estimate_edge_hints): Update.
22514 (inline_read_section, inline_write_summary): Stream both new predicates.
22515 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
22517 (compute_inlined_call_time): Cleanup.
22518 (big_speedup_p): Update.
22519 (edge_badness): Update.
22520 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
22521 (size_time_entry): Replace predicate by exec_predicate and
22522 nonconst_predicate.
22523 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
22524 (estimate_edge_time): Return also nonspec_time.
22525 (reset_edge_growth_cache): Update.
22527 2017-04-29 Jakub Jelinek <jakub@redhat.com>
22529 PR rtl-optimization/80491
22530 * ifcvt.c (noce_process_if_block): When looking for x setter
22531 with missing else_bb, don't check only the insn right before
22532 cond_earliest, but look for the last insn that x is modified in
22533 within the same bb.
22535 PR rtl-optimization/80491
22536 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
22538 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
22540 PR tree-optimization/80487
22541 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
22543 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22545 PR tree-optimization/79697
22546 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
22547 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
22548 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
22550 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
22551 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
22553 2017-04-28 Martin Sebor <msebor@redhat.com>
22555 PR tree-optimization/80523
22556 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
22557 (init_target_to_host_charmap, target_to_host, target_strtol10): New
22559 (maybe_warn, format_directive, parse_directive): Use new functions.
22560 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
22562 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
22564 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
22566 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
22568 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
22569 target_header_dir): Set correctly.
22570 * configure: Regenerated.
22571 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
22572 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
22573 instead of SYSTEM_HEADER_DIR.
22575 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
22577 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
22578 (estimate_local_effects): Likewise.
22579 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
22580 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
22581 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
22582 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
22583 do_estimate_edge_time, estimate_edge_time): Likewise.
22584 * ipa-inline-analysis.c (estimate_node_size_and_time,
22585 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
22586 (estimate_time_after_inlining): Remove.
22588 2017-04-28 Martin Liska <mliska@suse.cz>
22590 * doc/gcov.texi: Enhance documentation of gcov.
22592 2017-04-28 Martin Liska <mliska@suse.cz>
22594 * doc/gcov.texi: Sort options in alphabetic order.
22595 * doc/gcov-dump.texi: Likewise.
22596 * doc/gcov-tool.texi: Likewise.
22597 * gcov.c (print_usage): Likewise.
22598 * gcov-dump.c (print_usage): Likewise.
22599 * gcov-tool.c (print_merge_usage_message): Likewise.
22600 (print_rewrite_usage_message): Likewise.
22601 (print_overlap_usage_message): Likewise.
22603 2017-04-28 Martin Liska <mliska@suse.cz>
22605 PR gcov-profile/53915
22606 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
22608 2017-04-28 Martin Liska <mliska@suse.cz>
22610 PR gcov-profile/79891
22611 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
22612 is marked by compiler as living on a line.
22613 (get_cycles_count): Remove usage of the union.
22614 (output_intermediate_file): Likewise.
22615 (find_source): Fix GNU coding style.
22616 (accumulate_line_counts): Remove old non-all block mode.
22617 (output_lines): Remove usage of the union.
22618 * profile.c (output_location): Include all BBs, even if
22619 belonging to a same line (and file) as a previous BB.
22621 2017-04-28 Martin Liska <mliska@suse.cz>
22623 * gcov.c (process_args): Handle new argument 'w'.
22624 (read_graph_file): Assign ID to BBs.
22625 (output_branch_count): Display BB # if verbose flag is set.
22626 (output_lines): Likewise for arcs.
22627 (print_usage): Add '--verbose' option help.
22628 * doc/gcov.texi: Document --verbose (-w) option.
22630 2017-04-28 Martin Liska <mliska@suse.cz>
22632 * gcov.c (struct block_location_info): New struct.
22633 (process_file): Fill up the new structure.
22634 (read_graph_file): Replace usage of encoding by the newly added
22636 (add_line_counts): Likewise.
22637 (accumulate_line_counts): Remove usage of the union.
22638 (function_info::function_info): New function.
22639 (function_info::~function_info): Likewise.
22640 (process_file): Call delete instead of release_function.
22641 (release_function): Release the function.
22642 (release_structures): Call delete instead of release_function.
22643 (solve_flow_graph): Replace usage of num_blocks.
22644 (find_exception_blocks): Likewise.
22645 (output_lines): Fix GNU coding style.
22647 2017-04-28 Martin Liska <mliska@suse.cz>
22650 * coverage.c (coverage_remove_note_file): New function.
22651 * coverage.h: Declare the function.
22652 * toplev.c (finalize): Clean if an error has been seen.
22654 2017-04-28 Martin Liska <mliska@suse.cz>
22656 PR gcov-profile/80031
22657 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
22658 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
22659 * gcov.c (read_graph_file): Read just number of blocks.
22660 * profile.c (branch_prob): Do not stream 0 flags per a basic
22663 2017-04-28 Martin Liska <mliska@suse.cz>
22665 * gcov-dump.c (tag_*): Add new argument to declarations.
22666 (dump_gcov_file): Likewise.
22667 (tag_blocks): Add and use new argument depth.
22668 (tag_arcs): Likewise.
22669 (tag_lines): Likewise.
22670 (tag_counters): Likewise.
22671 (tag_summary): Likewise.
22672 (dump_working_sets): Use depth to do a proper indentation.
22674 2017-04-28 Jakub Jelinek <jakub@redhat.com>
22677 * cgraph.h (symtab_node::debug_symtab): No longer inline.
22678 * symtab.c (symtab_node::debug_symtab): Move definition here.
22680 2017-04-28 Richard Biener <rguenther@suse.de>
22682 * lto-streamer.h (LTO_major_version): Bump to 7.
22684 2017-04-28 Richard Biener <rguenther@suse.de>
22686 * tree-vrp.c (assert_info): New struct.
22687 (add_assert_info): New helper.
22688 (register_edge_assert_for_2): Refactor to add asserts to a vector
22690 (register_edge_assert_for_1): Likewise.
22691 (register_edge_assert_for): Likewise.
22692 (finish_register_edge_assert_for): New helper actually registering
22693 asserts where live on edge.
22694 (find_conditional_asserts): Adjust.
22695 (find_switch_asserts): Likewise.
22696 (evrp_dom_walker::try_find_new_range): Generalize.
22697 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
22699 2017-04-27 Marek Polacek <polacek@redhat.com>
22702 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
22703 arg10 and arg11 to itype.
22705 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
22707 * doc/extend.texi (Object Size Checking): Improve grammar.
22709 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
22712 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
22713 that the logic for permitting reciprocal estimates matches that
22716 2017-04-27 Jakub Jelinek <jakub@redhat.com>
22719 * tree.c (type_cache_hasher::equal): Only compare
22720 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
22721 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
22722 non-aggregate element types.
22723 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
22724 about the flag on ARRAY_TYPEs in the comment, formatting fix.
22726 2017-04-27 Richard Biener <rguenther@suse.de>
22728 PR middle-end/80533
22729 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
22730 stripping ARRAY_REFs from MEM_EXPR make sure we're not
22731 keeping a reference to a trailing array.
22733 2017-04-27 Richard Biener <rguenther@suse.de>
22735 PR middle-end/80539
22736 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
22737 being in loop-closed SSA form conservatively.
22738 (chrec_fold_multiply_poly_poly): Likewise.
22740 2017-04-27 Tamar Christina <tamar.christina@arm.com>
22742 PR middle-end/79665
22743 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
22744 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
22746 2017-04-27 Jakub Jelinek <jakub@redhat.com>
22749 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
22750 (aarch64_function_arg_alignment): Return unsigned int again, but still
22751 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
22752 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
22753 Don't emit -Wpsabi note.
22754 (aarch64_function_arg_boundary): Likewise.
22755 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
22758 2017-04-26 Nathan Sidwell <nathan@acm.org>
22760 * tree.h (crc32_unsigned_n): Declare.
22761 (crc32_unsigned, crc32_unsigned): Make inline.
22762 * tree.c (crc32_unsigned_bits): Replace with ...
22763 (crc32_unsigned_n): ... this.
22764 (crc32_unsigned, crc32_byte): Remove.
22765 (crc32_string): Remove unnecessary braces.
22767 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
22769 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
22770 * ipa-inline-analysis.c (MAX_TIME): Remove.
22771 (account_size_time): Use sreal for time.
22772 (dump_inline_summary): Update.
22773 (estimate_function_body_sizes): Update.
22774 (estimate_edge_size_and_time): Update.
22775 (estimate_calls_size_and_time): Update.
22776 (estimate_node_size_and_time): Update.
22777 (inline_merge_summary): Update.
22778 (inline_update_overall_summary): Update.
22779 (estimate_time_after_inlining): Update.
22780 (inline_read_section): Update.
22781 (inline_write_summary): Update.
22782 * ipa-inline.c (compute_uninlined_call_time): Update.
22783 (compute_inlined_call_time): Update.
22784 (recursive_inlining): Update.
22785 (inline_small_functions): Update.
22786 (dump_overall_stats): Update.
22787 * ipa-inline.h: Include sreal.h.
22788 (size_time_entry): Turn time to sreal.
22789 (inline_summary): Turn self_time nad time to sreal.
22791 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
22793 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
22795 (sreal::stream_out, sreal::stream_in): New.
22796 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
22798 2017-04-25 Jakub Jelinek <jakub@redhat.com>
22800 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
22803 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
22806 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
22808 (dimode_scalar_chain::compute_convert_gain): Ditto.
22809 (dimode_scalar_chain::make_vector_copies): Ditto.
22810 (dimode_scalar_chain::convert_reg): Ditto.
22811 (dimode_scalar_chain::convert_insn): Ditto.
22812 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
22813 (VI248_AVX512BW_1): New mode iterator.
22814 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
22815 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
22818 2017-04-25 Martin Sebor <msebor@redhat.com>
22820 PR tree-optimization/80497
22821 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
22822 constants are representable in HOST_WIDE_INT.
22823 (parse_directive): Ditto.
22825 2017-04-25 Martin Sebor <msebor@redhat.com>
22828 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
22829 (new_zero_array): Adjust signature.
22830 (dom_info::dom_init): Used unsigned rather that size_t.
22831 (dom_info::dom_info): Same.
22833 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22834 Jakub Jelinek <jakub@redhat.com>
22837 * config/arm/arm.c: Include gimple.h.
22838 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
22839 returns negative, increment ncrn only if it returned positive.
22840 (arm_needs_doubleword_align): Return int instead of bool,
22841 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
22842 members, but if there is any such non-FIELD_DECL
22843 > PARM_BOUNDARY aligned decl, return -1 instead of false.
22844 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
22845 returns negative, increment nregs only if it returned positive.
22846 (arm_setup_incoming_varargs): Likewise.
22847 (arm_function_arg_boundary): Emit -Wpsabi note if
22848 arm_needs_doubleword_align returns negative, return
22849 DOUBLEWORD_ALIGNMENT only if it returned positive.
22851 2017-04-25 Marek Polacek <polacek@redhat.com>
22854 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
22855 first argument to type.
22857 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
22860 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
22861 type checks to test for compatibility instead of equality.
22863 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22864 Jakub Jelinek <jakub@redhat.com>
22867 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
22869 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
22870 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
22871 the alignment computation, but return their maximum in warn_alignment.
22872 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
22873 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
22875 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
22876 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
22879 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22881 * config/arc/simdext.md (dmpyh): Fix typo.
22883 2017-04-25 Richard Biener <rguenther@suse.de>
22885 PR tree-optimization/80492
22886 * alias.c (compare_base_decls): Handle registers with asm
22887 specification conservatively.
22888 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
22889 compare_base_decls returning dont-know properly.
22891 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22893 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
22894 (legitimate_offset_address_p): New function.
22895 (arc_legitimate_address_p): Use above function.
22897 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22899 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
22901 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22903 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
22904 ACCH registers whenever they are available.
22906 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22908 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
22909 double regs fix when not used.
22911 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22913 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
22915 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
22916 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
22918 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22920 * config/arc/arc.c (arc_output_addsi): Check for h-register class
22921 when emitting short ADD instructions.
22923 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
22925 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
22927 (cmpsi_cc_c_insn): Likewise.
22928 (cbranchsi4_scratch): Compute proper instruction length using
22929 compact_hreg_operand.
22930 * config/arc/predicates.md (compact_hreg_operand): New predicate.
22932 2017-04-25 Richard Biener <rguenther@suse.de>
22934 PR middle-end/80509
22935 * passes.c (pass_manager::pass_manager): Initialize
22936 m_name_to_pass_map.
22938 2017-04-25 Richard Biener <rguenther@suse.de>
22940 PR tree-optimization/79201
22941 * tree-ssa-sink.c (statement_sink_location): Handle calls.
22943 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22946 * config/s390/vector.md: Split MEM->GPR vector moves for
22947 non-s_operand addresses.
22949 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22952 * config/s390/predicates.md (reload_const_wide_int_operand): New
22954 * config/s390/s390.md ("movti"): Remove d/P alternative.
22955 ("movti_bigconst"): New pattern definition.
22957 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
22960 * s390-protos.h (s390_expand_cs_hqi): Removed.
22961 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
22962 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
22963 modes as well as CCZ1mode and CCZmode.
22964 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
22965 signature of s390_emit_compare_and_swap.
22966 (s390_expand_cs_hqi): Likewise, make static.
22967 (s390_expand_cs_tdsi): Generate an explicit compare before trying
22968 compare-and-swap, in some cases.
22969 (s390_expand_cs): Wrapper function.
22970 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
22972 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
22973 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
22974 patterns for small and large integers. Forbid symref memory operands.
22975 Move expander to s390.c. Require cc register.
22976 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
22977 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
22978 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
22979 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
22980 symref memory operands. Remove CC mode and call s390_match_ccmode
22982 ("atomic_exchange<mode>"): Allow and implement all integer modes.
22984 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
22986 * config/s390/s390.md (define_peephole2): New peephole to help
22987 combining the load-and-test pattern with volatile memory.
22989 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
22991 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
22992 with CCZmode for TARGET_Z196.
22994 2017-04-25 Jakub Jelinek <jakub@redhat.com>
22996 PR rtl-optimization/80501
22997 * combine.c (make_compound_operation_int): Set subreg_code to SET
22998 even for AND with mask of the sign bit of mode.
23000 PR rtl-optimization/80500
23001 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
23002 sum's initial value.
23004 2017-04-25 Julian Brown <julian@codesourcery.com>
23005 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23007 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
23009 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
23011 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
23013 2017-04-25 Julian Brown <julian@codesourcery.com>
23014 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23016 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
23017 (thunderx2t99_sha): New Reservation.
23019 2017-04-25 Julian Brown <julian@codesourcery.com>
23020 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23022 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
23023 type for 1-element load.
23025 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
23027 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
23029 2017-04-24 Martin Jambor <mjambor@suse.cz>
23031 PR tree-optimization/80293
23032 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
23033 char arrays not totally scalarizable if it is false.
23034 (analyze_all_variable_accesses): Pass correct value in the new
23035 parameter. Add a statistics counter.
23037 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
23039 PR middle-end/79931
23040 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
23042 2017-04-24 Richard Biener <rguenther@suse.de>
23044 PR tree-optimization/80494
23045 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
23046 out for complex types.
23048 2017-04-24 Richard Biener <rguenther@suse.de>
23050 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
23051 * tree-ssa-sccvn.c (print_scc): Print SCC size.
23052 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
23053 (DFS): Adjust and never fail.
23054 (sccvn_dom_walker::fail): Remove.
23055 (sccvn_dom_walker::before_dom_children): Adjust.
23056 (run_scc_vn): Likewise and never fail.
23057 * tree-ssa-pre.c (pass_pre::execute): Adjust.
23058 (pass_fre::execute): Likewise.
23060 2017-04-24 Richard Biener <rguenther@suse.de>
23062 PR tree-optimization/79725
23063 * tree-ssa-sink.c (statement_sink_location): Return whether
23064 failure reason was zero uses. Move that check later.
23065 (sink_code_in_bb): Deal with zero uses by removing the stmt
23068 2017-04-24 Richard Biener <rguenther@suse.de>
23071 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
23072 pointer-based references.
23074 2017-04-24 Richard Biener <rguenther@suse.de>
23077 * pass_manager.h (pass_manager::operator new): Remove.
23078 (pass_manager::operator delete): Likewise.
23079 * passes.c (pass_manager::operator new): Remove.
23080 (pass_manager::operator delete): Likewise.
23081 (pass_manager::pass_manager): Zero individual pass members.
23083 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
23086 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
23087 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
23088 Check "XEXP (src, 1)" operand here.
23089 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
23090 Check "XEXP (src, 1)" operand here.
23091 (dimode_scalar_chain::make_vector_copies): Detect count register
23092 of a shift instruction. Zero extend count register from QImode
23093 to DImode to satisfy vector shift pattern count operand predicate.
23094 Substitute vector shift count operand with a DImode copy.
23095 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
23098 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
23100 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
23101 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
23102 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
23103 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
23104 (UNSPEC_NOREX_MEM): Remove definition.
23106 2017-04-21 Richard Biener <rguenther@suse.de>
23108 PR tree-optimization/79547
23109 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
23110 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
23111 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
23112 without any constraints.
23114 2017-04-21 Richard Biener <rguenther@suse.de>
23116 PR tree-optimization/78847
23117 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
23119 2017-04-21 Richard Biener <rguenther@suse.de>
23121 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
23122 (build_distinct_type_copy): Likewise.
23123 (build_variant_type_copy): Likewise.
23124 * tree.c (build_qualified_type): Pass down mem-stat info.
23125 (build_distinct_type_copy): Likewise.
23126 (build_variant_type_copy): Likewise.
23128 2017-04-21 Richard Biener <rguenther@suse.de>
23130 PR tree-optimization/80237
23131 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
23133 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
23134 for a simplified result.
23136 2016-04-21 Richard Biener <rguenther@suse.de>
23138 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
23139 sth as strict as a simple_iv but a chrec without symbols and an
23140 operand defined in the loop we are peeling (and not some subloop).
23141 (propagate_constants_for_unrolling): Propagate all constants.
23143 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
23146 * config/i386/i386.c (print_reg): Remove assert for disalowed
23147 regno values, call output_operand_lossage instead.
23149 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
23152 * config/i386/constraints.md (Yc): New register constraint.
23153 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
23154 Use Yc constraint for alternative 2 of operand 0. Remove
23155 preferred_for_speed attribute.
23157 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
23159 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
23160 lastprivate clauses in SIMT case.
23162 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
23164 * doc/invoke.texi (-Wextra-semi): Document new warning option.
23166 2017-04-20 Richard Biener <rguenther@suse.de>
23168 PR tree-optimization/57796
23169 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
23170 as N scalar stores.
23171 (vect_model_load_cost): Cost gathers as N scalar loads.
23173 2017-04-20 Richard Biener <rguenther@suse.de>
23175 * ggc-page.c (ggc_allocated_p): Rename to ...
23176 (safe_lookup_page_table_entry): ... this and return the lookup
23178 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
23180 2017-04-20 Richard Biener <rguenther@suse.de>
23182 PR tree-optimization/80453
23183 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
23184 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
23185 from the conditions.
23186 (vn_phi_eq): Pass them down.
23187 (vn_phi_lookup): Record them.
23188 (vn_phi_insert): Likewise.
23190 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
23192 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
23193 uninitialized variable warning to avoid buffer overrun.
23195 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
23198 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
23199 is suppressed for '{ 0 }' in C.
23201 2017-04-20 Jakub Jelinek <jakub@redhat.com>
23203 * BASE-VER: Set to 8.0.0.
23205 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
23207 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
23208 priority .init_array and .fini_array section with SECTION_NOTYPE
23211 2017-04-20 Jakub Jelinek <jakub@redhat.com>
23213 PR middle-end/80423
23214 * tree.h (build_array_type): Add typeless_storage default argument.
23215 * tree.c (type_cache_hasher::equal): Also compare
23216 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
23217 (build_array_type): Add typeless_storage argument, set
23218 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
23220 (build_nonshared_array_type): Adjust build_array_type_1 caller.
23221 (build_array_type): Likewise. Add typeless_storage argument.
23223 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
23224 Jakub Jelinek <jakub@redhat.com>
23226 PR tree-optimization/80426
23227 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
23228 operation on symbolic operands, also compute the overflow for the
23229 invariant part when the operation degenerates into a negation.
23231 2017-04-19 Jakub Jelinek <jakub@redhat.com>
23234 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
23235 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
23238 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
23240 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
23243 * config/avr/avr.c (tree.h): Include it.
23244 (cgraph.h): Include it.
23245 (avr_encode_section_info): Don't warn for uninitialized progmem
23246 variable if it's just an alias.
23248 2017-04-19 Richard Biener <rguenther@suse.de>
23251 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
23252 when needed by AutoPGO.
23254 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
23257 * doc/lto.texi: Remove an extra 'that'.
23259 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
23261 PR rtl-optimization/80429
23262 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
23263 are only used in debug insns.
23265 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
23266 Vladimir Makarov <vmakarov@redhat.com>
23268 * config/sparc/predicates.md (input_operand): Add comment. Return
23269 true for any memory operand when LRA is in progress.
23270 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
23272 2017-04-18 Jeff Law <law@redhat.com>
23275 * mips.md ({return,simple_return}_internal): Do not overwrite
23278 2017-04-18 Jakub Jelinek <jakub@redhat.com>
23280 PR tree-optimization/80443
23281 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
23282 instead of adding 1, subtract -1 and similarly instead of subtracting
23285 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
23287 PR rtl-optimization/80357
23288 * haifa-sched.c (tmp_bitmap): New variable.
23289 (model_recompute): Handle duplicate use records.
23290 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
23291 (free_global_sched_pressure_data): Free it.
23293 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
23296 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
23297 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
23298 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
23299 instead of SYSTEM_HEADER_DIR.
23301 2017-04-18 Jeff Law <law@redhat.com>
23303 PR middle-end/80422
23304 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
23305 predecessors after walking up the insn chain.
23307 2017-04-18 Jakub Jelinek <jakub@redhat.com>
23310 * dwarf2out.c (modified_type_die): Try harder not to emit internal
23311 sizetype type into debug info.
23313 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
23316 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
23317 unneeded test for TARGET_UPPER_REGS_SF.
23318 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
23320 2017-04-18 Jakub Jelinek <jakub@redhat.com>
23323 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
23324 instead of gsi_after_labels.
23326 2017-04-18 Jeff Law <law@redhat.com>
23328 * regcprop.c (maybe_mode_change): Avoid creating copies of the
23332 2017-04-13 Jeff Law <law@redhat.com>
23333 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
23334 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
23336 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
23339 * config/avr/avr.c (intl.h): Include it.
23340 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
23342 2017-04-18 Martin Liska <mliska@suse.cz>
23344 PR gcov-profile/78783
23345 * gcov-tool.c (gcov_output_files): Validate that destination
23346 file is either removed by the tool or by a user.
23348 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
23349 Guy Benyei <guybe@mellanox.com>
23351 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
23352 block, and do not negate it, the stored id is already negative.
23354 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
23356 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
23358 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
23361 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
23362 masks of options that should be turned off if the VSX vector
23363 options are turned off.
23364 (OTHER_P8_VECTOR_MASKS): Likewise.
23365 (OTHER_VSX_VECTOR_MASKS): Likewise.
23366 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
23367 rs6000_disable_incompatible_switches to validate no type switches
23369 (rs6000_incompatible_switch): New function to disallow turning on
23370 other vector options if -mno-vsx, -mno-power8-vector, or
23371 -mno-power9-vector are specified.
23373 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
23375 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
23377 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
23379 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
23380 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
23381 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
23382 (ARG_POINTER_CFA_OFFSET): Likewise.
23384 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
23386 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
23387 conditions to take advantage of various optimizations.
23389 2017-04-13 Jeff Law <law@redhat.com>
23391 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
23392 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
23393 (zero_extendsidi2_dext): Likewise.
23395 2017-04-13 Jakub Jelinek <jakub@redhat.com>
23398 * fold-const.c (fold_ternary_loc): Revert
23399 use op0 instead of fold_convert_loc (loc, type, arg0) part of
23402 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
23404 PR rtl-optimization/80343
23405 * lra-remat.c (update_scratch_ops): Assign original hard reg to
23406 new scratch pseudo.
23408 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
23411 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
23412 to ubsan_encode_value.
23414 2017-04-13 Jeff Law <law@redhat.com>
23416 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
23417 appearing in DEBUG_INSNs.
23419 2017-04-13 Martin Liska <mliska@suse.cz>
23421 PR gcov-profile/80413
23422 * gcov-io.c (gcov_write_string): Copy to buffer just when
23423 allocated size is greater than zero.
23425 2017-04-13 Jakub Jelinek <jakub@redhat.com>
23428 * dwarf2out.c (decls_for_scope): Ignore declarations of
23429 current_function_decl in BLOCK_NONLOCALIZED_VARS.
23431 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
23434 * ipa-visibility.c (non_local_p): Fix typos.
23435 (localize_node): When localizing symbol in same comdat group,
23436 dissolve the group only when we know external symbols are going
23438 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
23440 2017-04-12 Jakub Jelinek <jakub@redhat.com>
23442 PR tree-optimization/79390
23443 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
23444 order does not result in usable sequence, retry with reversed operand
23450 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
23451 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
23452 op0 instead of fold_convert_loc (loc, type, arg0).
23454 2017-04-12 Jeff Law <law@redhat.com>
23456 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
23457 has a delay slot in the generated code.
23459 * config/cris/cris.md (cris_preferred_reload_class): Return
23460 GENNONACR_REGS rather than GENERAL_REGS.
23462 2017-04-12 Jakub Jelinek <jakub@redhat.com>
23465 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
23466 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
23467 signedness of the result type.
23469 2017-04-12 Richard Biener <rguenther@suse.de>
23470 Jeff Law <law@redhat.com>
23472 PR tree-optimization/80359
23473 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
23474 trim stores to TARGET_MEM_REFs.
23476 2017-04-12 Richard Biener <rguenther@suse.de>
23478 PR tree-optimization/79390
23479 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
23480 threading case even more.
23482 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
23485 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
23486 for quad_address_p for TImode, instead of just not indexed_address.
23488 2017-04-12 Richard Biener <rguenther@suse.de>
23489 Bernd Edlinger <bernd.edlinger@hotmail.de>
23491 PR middle-end/79671
23492 * alias.c (component_uses_parent_alias_set_from): Handle
23493 TYPE_TYPELESS_STORAGE.
23494 (get_alias_set): Likewise.
23495 * tree-core.h (tree_type_common): Add typeless_storage flag.
23496 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
23497 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
23498 for types containing members with TYPE_TYPELESS_STORAGE.
23499 (place_field): Likewise.
23500 (layout_type): Likewise for ARRAY_TYPE.
23501 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
23502 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
23503 TYPE_TYPELESS_STORAGE.
23504 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
23506 2017-04-12 Jakub Jelinek <jakub@redhat.com>
23509 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
23510 first argument to type.
23512 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23516 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
23517 CONST0_RTX (mode) rather than const0_rtx where appropriate.
23518 (rs6000_expand_binop_builtin): Likewise.
23519 (rs6000_expand_ternop_builtin): Likewise; also add missing
23520 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
23521 vshasigma built-ins.
23522 * doc/extend.texi: Document that vec_xxpermdi's third argument
23523 must be a constant.
23525 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
23527 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
23528 Use shift_const cost parameter when calculating gain of STV shifts.
23530 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
23532 PR rtl-optimization/70478
23533 * lra-constraints.c (process_alt_operands): Check memory for
23534 disfavoring memory insn operand.
23536 2017-04-11 Jakub Jelinek <jakub@redhat.com>
23538 PR middle-end/80100
23539 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
23540 left shift in unsigned HOST_WIDE_INT type.
23542 PR rtl-optimization/80385
23543 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
23544 (not (neg X)) into (plus X -1) for complex or non-integral modes.
23547 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
23548 if they have any depend clauses.
23550 2017-04-11 Martin Liska <mliska@suse.cz>
23553 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
23554 * ipa-split.c (split_function): Create a local comdat symbol
23555 if caller is in a comdat group.
23557 2017-04-11 Martin Liska <mliska@suse.cz>
23560 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
23563 2017-04-11 Martin Sebor <msebor@redhat.com>
23565 PR middle-end/80364
23566 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
23567 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
23569 (directive::set_width, directive::set_precision, format_character):
23571 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
23574 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
23577 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
23578 conflict, set target->arch_name instead of target->cpu_name.
23580 2017-04-11 Richard Biener <rguenther@suse.de>
23582 PR tree-optimization/80374
23583 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
23584 build_zero_cst, remove fold_convertible_p check again.
23586 2017-04-11 Martin Liska <mliska@suse.cz>
23589 * ubsan.c (instrument_object_size): Do not instrument register
23592 2017-04-11 Jakub Jelinek <jakub@redhat.com>
23595 * config/i386/i386-builtin-types.def
23596 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
23597 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
23598 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
23599 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
23600 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
23601 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
23602 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
23603 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
23604 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
23605 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
23606 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
23607 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
23608 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
23609 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
23610 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
23611 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
23612 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
23613 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
23614 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
23615 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
23616 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
23617 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
23618 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
23619 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
23620 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
23621 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
23622 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
23623 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
23624 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
23625 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
23626 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
23627 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
23628 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
23629 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
23630 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
23631 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
23632 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
23633 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
23634 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
23635 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
23636 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
23637 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
23638 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
23639 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
23640 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
23641 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
23643 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
23644 flag to second_arg_count, handle 4 argument function type _COUNT
23645 aliases, handle second_arg_count on second argument rather than last.
23647 2017-04-10 Jeff Law <law@redhat.com>
23649 PR tree-optimization/80374
23650 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
23651 record anything if we can not convert integer_zero_node to the
23654 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
23657 * config/rs6000/rs6000.c (rs6000_option_override_internal):
23658 Enhance special handling given to the TARGET_P9_MINMAX option in
23659 relation to certain other options.
23661 2017-04-10 Bin Cheng <bin.cheng@arm.com>
23663 PR tree-optimization/80153
23664 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
23665 remove POINTER_PLUS_EXPR's base part directly, rather than through
23668 2017-04-10 Richard Biener <rguenther@suse.de>
23669 Bin Cheng <bin.cheng@arm.com>
23671 PR tree-optimization/80153
23672 * tree-affine.c (aff_combination_to_tree): Get base pointer from
23673 the first element of pointer type aff_tree. Build result expr in
23675 (add_elt_to_tree): Convert to type unconditionally. Remove other
23676 fold_convert calls.
23677 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
23678 (rewrite_use_nonlinear_expr): Check invariant using iv information.
23680 2017-04-10 Richard Biener <rguenther@suse.de>
23682 * tree-ssa-structalias.c (find_func_aliases): Properly handle
23685 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
23687 PR rtl-optimization/70478
23688 * lra-constraints.c (curr_small_class_check): New.
23689 (update_and_check_small_class_inputs): New.
23690 (process_alt_operands): Update curr_small_class_check. Disfavor
23691 alternative insn memory operands. Check available regs for small
23694 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
23697 * config/mips/mips.opt (-mvirt): Update description.
23698 * doc/invoke.texi (-mvirt): Likewise.
23700 2017-04-10 Richard Biener <rguenther@suse.de>
23702 PR middle-end/80362
23703 * fold-const.c (fold_binary_loc): Look at unstripped ops when
23704 looking for NEGATE_EXPR in -A / -B to A / B folding.
23706 2017-04-10 Martin Liska <mliska@suse.cz>
23708 PR gcov-profile/80224
23709 * gcov.c (print_usage): Fix usage string.
23710 (get_gcov_intermediate_filename): Remove.
23711 (output_gcov_file): Use both for normal and intermediate format.
23712 (generate_results): Do not initialize special file for
23713 intermediate format.
23715 2017-04-10 Richard Biener <rguenther@suse.de>
23717 PR tree-optimization/80304
23718 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
23721 2017-04-10 Nathan Sidwell <nathan@acm.org>
23724 * config/rs6000/rs6000.c (rs6000_vector_type): New.
23725 (rs6000_init_builtins): Use it.
23727 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23729 * config/arm/arm.md (<mrc>): Add mode to SET source.
23730 (<mrrc>): Likewise.
23732 2017-04-10 Richard Biener <rguenther@suse.de>
23734 PR middle-end/80344
23735 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
23737 2017-04-10 Jakub Jelinek <jakub@redhat.com>
23740 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
23741 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
23742 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
23743 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
23744 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
23745 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
23746 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
23747 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
23748 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
23749 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
23750 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
23751 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
23752 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
23753 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
23754 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
23755 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
23756 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
23757 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
23758 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
23759 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
23760 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
23761 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
23762 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
23764 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
23766 PR rtl-optimization/70478
23767 * lra-constraints.c: Reverse the last patch.
23769 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
23771 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
23772 Add comment for WCHAR_T.
23774 2017-04-08 Martin Liska <mliska@suse.cz>
23777 2017-04-07 Martin Liska <mliska@suse.cz>
23780 * ipa-split.c (split_function): Add function part to a same comdat
23783 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23786 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
23788 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
23790 * rs6000/rs6000.c (vec_load_pendulum): Rename...
23791 (vec_pairing): ...to this.
23792 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
23793 (rs6000_sched_init): Adjust for name change.
23794 (struct rs6000_sched_context): Likewise.
23795 (rs6000_init_sched_context): Likewise.
23796 (rs6000_set_sched_context): Likewise.
23798 2017-04-07 Jakub Jelinek <jakub@redhat.com>
23804 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
23806 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
23807 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
23808 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
23810 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
23812 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
23814 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
23816 PR rtl-optimization/70703
23817 * ira-color.c (update_conflict_hard_regno_costs): Use
23818 int64_t instead of HOST_WIDE_INT.
23820 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
23822 PR rtl-optimization/70478
23823 * lra-constraints.c (process_alt_operands): Disfavor alternative
23824 insn memory operands.
23826 2017-04-07 Jeff Law <law@redhat.com>
23828 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
23829 CALL and NOTE_INSN_CALL_ARG_LOCATION.
23831 2017-04-07 Martin Liska <mliska@suse.cz>
23834 * config/aarch64/aarch64.c (aarch64_process_target_attr):
23835 Show error message instead of an ICE.
23837 2017-04-07 Martin Liska <mliska@suse.cz>
23840 * ipa-split.c (split_function): Add function part to a same comdat
23843 2017-04-07 Richard Biener <rguenther@suse.de>
23845 PR middle-end/80341
23846 * tree.c (get_unwidened): Also handle ! for_type case for
23848 * convert.c (do_narrow): Split out from ...
23849 (convert_to_integer_1): ... here. Do not pass final truncation
23850 type to get_unwidened for TRUNC_DIV_EXPR.
23852 2017-04-07 Richard Biener <rguenther@suse.de>
23854 * tree-affine.c (wide_int_ext_for_comb): Take type rather
23856 (aff_combination_const): Adjust.
23857 (aff_combination_scale): Likewise.
23858 (aff_combination_add_elt): Likewise.
23859 (aff_combination_add_cst): Likewise.
23860 (aff_combination_convert): Likewise.
23861 (add_elt_to_tree): Likewise. Remove unused argument.
23862 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
23864 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
23866 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
23868 * config/arm/arm.c (arm_default_short_enums): Use
23869 ARM_DEFAULT_SHORT_ENUMS.
23870 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
23872 2017-04-06 Jakub Jelinek <jakub@redhat.com>
23875 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
23876 members with redundant out-of-class redeclaration.
23878 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
23881 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
23882 * config/i386/i386.md (*zero_extendsidi2):
23883 Add (?*x,*x) and (?*v,*v) alternatives.
23885 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
23888 * config/i386/i386.c (ix86_expand_builtin)
23889 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
23890 mode from insn data. Convert operands to insn operand mode.
23891 Copy operands that don't satisfy insn predicate to a register.
23893 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
23895 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
23898 2017-04-06 Richard Biener <rguenther@suse.de>
23900 PR tree-optimization/80334
23901 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
23902 preserve alignment of accesses.
23904 2017-04-06 Richard Biener <rguenther@suse.de>
23906 PR tree-optimization/80262
23907 * tree-sra.c (build_ref_for_offset): Preserve address-space
23909 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
23910 Drop useless address-space information on MEM_REF offsets.
23912 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
23914 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
23916 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
23918 PR rtl-optimization/70703
23919 * ira-color.c (update_conflict_hard_regno_costs): Use
23920 HOST_WIDE_INT instead of long.
23922 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
23925 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
23926 not defined for x86_64 target. Add -mmmx target option when __SSE2__
23928 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
23929 for x86_64 target. Handle -m3dnowa option.
23931 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
23933 PR rtl-optimization/70703
23934 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
23935 (update_conflict_hard_regno_costs): Use long instead of unsigned
23936 arithmetic for cost calculation.
23938 2017-04-05 Jakub Jelinek <jakub@redhat.com>
23939 Bernd Edlinger <bernd.edlinger@hotmail.de>
23942 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
23945 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
23948 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
23949 ptr_mode with Pmode throughout.
23950 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
23951 into probe_stack_range and use DImode.
23953 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
23956 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
23957 call_eh_return is true.
23959 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23961 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
23962 Initialize last_match_fntype_index.
23964 2017-04-05 Jakub Jelinek <jakub@redhat.com>
23967 * tree-nvr.c: Include internal-fn.h.
23968 (pass_return_slot::execute): Ignore internal calls without
23971 2017-04-04 Jakub Jelinek <jakub@redhat.com>
23972 Richard Biener <rguenther@suse.de>
23975 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
23976 captures used multiple times, except for the last use.
23977 * generic-match-head.c: Include gimplify.h.
23979 2017-04-04 Jakub Jelinek <jakub@redhat.com>
23981 PR tree-optimization/79390
23982 * target.h (struct noce_if_info): Declare.
23983 * targhooks.h (default_noce_conversion_profitable_p): Declare.
23984 * target.def (noce_conversion_profitable_p): New target hook.
23985 * ifcvt.h (struct noce_if_info): New type, moved from ...
23986 * ifcvt.c (struct noce_if_info): ... here.
23987 (noce_conversion_profitable_p): Renamed to ...
23988 (default_noce_conversion_profitable_p): ... this. No longer
23990 (noce_try_store_flag_constants, noce_try_addcc,
23991 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
23992 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
23993 instead of noce_conversion_profitable_p.
23994 * config/i386/i386.c: Include ifcvt.h.
23995 (ix86_option_override_internal): Don't override
23996 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
23997 (ix86_noce_conversion_profitable_p): New function.
23998 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
23999 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
24000 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
24001 * doc/tm.texi: Regenerated.
24003 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24005 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
24008 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24011 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
24012 instructions for small multiply cores.
24014 2017-04-04 Jeff Law <law@redhat.com>
24016 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
24018 (mips_expand_vec_perm_const): Initialize elements in orig_perm
24019 that are not set by the loop over the elements.
24021 2017-04-04 Jakub Jelinek <jakub@redhat.com>
24024 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
24025 int mode, convert_modes it to mode as unsigned, otherwise use
24026 lowpart_subreg to mode rather than SImode.
24027 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
24028 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
24029 Use DImode instead of SImode for the shift count operand.
24030 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
24033 2017-04-04 Richard Biener <rguenther@suse.de>
24035 PR middle-end/80281
24036 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
24037 arithmetic done for the negate or the plus. Simplify.
24038 (A - (-B) -> A + B): Likewise.
24039 * fold-const.c (split_tree): Make sure to not negate pointers.
24041 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
24043 PR rtl-optimization/60818
24044 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
24045 a compare of comparisons with the thing compared if this results
24046 in a different machine mode.
24048 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
24050 * alias.c (base_alias_check): Fix typo in comment.
24051 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
24052 * cgraphunit.c (symbol_table::compile): Likewise.
24053 * collect2.c (maybe_run_lto_and_relink): Likewise.
24054 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
24055 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
24056 * config/avr/avr.c (avr_map_op_t): Likewise.
24057 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
24058 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
24059 * config/epiphany/epiphany.md (movcc): Likewise.
24060 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
24061 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
24063 * config/mips/mips.c (mips_save_restore_reg): Likewise.
24064 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
24065 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
24066 * config/sh/sh.c (sh_rtx_costs): Likewise.
24067 * fold-const.c (fold_truth_andor): Likewise.
24068 * genautomata.c (collapse_flag): Likewise.
24069 * gengtype.h (struct type::u::s): Likewise.
24070 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
24071 * input.c (FORMAT_AMOUNT): Likewise.
24072 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
24073 (known_aggs_to_agg_replacement_list): Likewise.
24074 * ipa-inline-analysis.c: Likewise.
24075 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
24076 * ipa-polymorphic-call.c
24077 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
24078 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
24079 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
24081 * modulo-sched.c (apply_reg_moves): Likewise.
24082 * omp-expand.c (build_omp_regions_1): Likewise.
24083 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
24084 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
24085 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
24086 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
24087 * value-prof.c: Likewise.
24088 * var-tracking.c (val_reset): Likewise.
24090 2017-04-03 Richard Biener <rguenther@suse.de>
24092 PR tree-optimization/80275
24093 * fold-const.c (split_address_to_core_and_offset): Handle
24096 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
24098 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
24099 descriptors is at least equal to that of functions.
24101 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
24103 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
24105 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
24108 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
24109 (mov<IMOD4:mode>): New expander.
24110 (*mov<IMOD4:mode>_internal): New insn and split pattern.
24112 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
24114 PR rtl-optimization/79405
24115 * fwprop.c (propagations_left): New variable.
24116 (forward_propagate_into): Decrement it.
24117 (fwprop_init): Initialize it.
24118 (fw_prop): If the variable has reached zero, stop propagating.
24119 (fwprop_addr): Ditto.
24121 2017-03-31 Jakub Jelinek <jakub@redhat.com>
24124 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
24125 a FUNCTION_DECL, pass it as decl instead of origin to
24128 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
24130 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
24133 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
24136 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
24137 TARGET_VSX_SMALL_INTEGER.
24139 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24141 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
24142 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
24144 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
24146 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
24147 extraction from odd-numbered MSA register.
24149 2017-03-31 Jakub Jelinek <jakub@redhat.com>
24151 PR middle-end/80173
24152 * expmed.c (store_bit_field_1): Don't attempt to create
24153 a word subreg out of hard registers wider than word if they
24154 have HARD_REGNO_NREGS of 1 for their mode.
24156 PR middle-end/80163
24157 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
24158 conversions to integer types wider than word and pointer.
24161 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
24162 (rtx_equal_for_cselib_p): Pass 0 to it.
24163 * cselib.c (cselib_hasher::equal): Likewise.
24164 (rtx_equal_for_cselib_1): Add depth argument. If depth
24165 is 128, don't look up VALUE locs and punt. Increment
24166 depth in recursive calls when walking VALUE locs.
24168 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
24170 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
24171 (make_gcov_file_name): Use the canonical path name for generating
24173 (read_line): Fix handling of files with ascii null bytes.
24175 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
24177 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
24178 to initialise a vector register instead
24179 of using a const_int.
24181 2017-03-30 Jakub Jelinek <jakub@redhat.com>
24183 PR translation/80189
24184 * gimplify.c (omp_default_clause): Use %qs instead of %s in
24185 diagnostic messages.
24187 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
24190 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
24191 (dfp_diex_<mode>): Update mode of operand 1.
24192 * doc/extend.texi (dxex, dxexq): Document change to return type.
24193 (diex, diexq): Document change to argument type.
24195 2017-03-30 Martin Jambor <mjambor@suse.cz>
24198 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
24199 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
24200 it reflects the signature changes performed at the callee side.
24201 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
24202 to cgraph_build_function_type_skip_args.
24203 (build_function_decl_skip_args): Adjust call to the above function.
24205 2017-03-30 Jakub Jelinek <jakub@redhat.com>
24208 * config/i386/sse.md
24209 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
24210 register as dest whenever it is a MEM not rtx_equal_p to the
24211 corresponding dup operand, and when forcing into reg move the
24212 reg into the memory afterwards.
24213 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
24214 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
24215 for the force_reg mode.
24216 (avx512vl_vextractf128<mode>): Use register as dest either
24217 always when a MEM, or when it is a MEM not rtx_equal_p to the
24218 corresponding dup operand, or even not when it is a CONST_VECTOR
24219 depending on the mode and lo vs. hi.
24220 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
24222 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
24223 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
24224 Likewise. Require that operands[2] is even.
24225 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
24226 Remove extraneous parens. Require that operands[2] is a multiple
24228 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
24229 operands[0] is a MEM if <mask_applied>, the predicates/constraints
24230 disallow memory then.
24232 2017-03-30 Richard Biener <rguenther@suse.de>
24234 PR tree-optimization/77498
24235 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
24236 to non-constants over backedges.
24238 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
24240 PR rtl-optimization/80233
24241 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
24242 as last_combined_insn. Do not test for BARRIER_P separately.
24244 2017-03-29 Andreas Schwab <schwab@suse.de>
24247 * calls.c (prepare_call_address): Convert funexp to Pmode before
24248 copying to temp reg.
24250 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24252 PR tree-optimization/80158
24253 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
24254 Handle possible future case of more than one alternate
24256 (replace_rhs_if_not_dup): Likewise.
24257 (replace_one_candidate): Likewise.
24259 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
24261 PR rtl-optimization/80193
24262 * ira.c (ira): Do not check allocation for LRA.
24264 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
24266 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
24267 (nvptx_output_simt_exit): Declare.
24268 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
24269 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
24270 (init_softstack_frame): Move initialization of crtl->is_leaf to...
24271 (nvptx_declare_function_name): ...here. Emit declaration of local
24272 memory space buffer for omp_simt_enter insn.
24273 (nvptx_output_unisimt_switch): New.
24274 (nvptx_output_softstack_switch): New.
24275 (nvptx_output_simt_enter): New.
24276 (nvptx_output_simt_exit): New.
24277 * config/nvptx/nvptx.h (struct machine_function): New fields
24278 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
24279 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
24280 (UNSPECV_SIMT_EXIT): Ditto.
24281 (omp_simt_enter_insn): New insn.
24282 (omp_simt_enter): New expansion.
24283 (omp_simt_exit): New insn.
24284 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
24286 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
24287 (expand_GOMP_SIMT_ENTER_ALLOC): New.
24288 (expand_GOMP_SIMT_EXIT): New.
24289 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
24290 (GOMP_SIMT_ENTER_ALLOC): Ditto.
24291 (GOMP_SIMT_EXIT): Ditto.
24292 * target-insns.def (omp_simt_enter): New insn.
24293 (omp_simt_exit): Ditto.
24294 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
24296 (lower_rec_simd_input_clauses): Implement SIMT privatization.
24297 (lower_rec_input_clauses): Likewise.
24298 (lower_lastprivate_clauses): Handle SIMT privatization.
24300 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
24301 (ompdevlow_adjust_simt_enter): New.
24302 (find_simtpriv_var_op): New.
24303 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
24304 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
24306 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
24307 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
24308 (copy_decl_for_dup_finish): Ditto.
24310 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
24312 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
24315 * config/i386/i386.c (ix86_option_override_internal): Always
24316 allow -mpreferred-stack-boundary=3 for 64-bit targets.
24318 2017-03-28 Bin Cheng <bin.cheng@arm.com>
24320 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
24322 2017-03-28 Bin Cheng <bin.cheng@arm.com>
24324 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
24325 mark new edge's irreducible flag accordign to it.
24326 (vect_do_peeling): Check loop preheader edge's irreducible flag
24327 and pass it to function slpeel_add_loop_guard.
24329 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
24331 PR tree-optimization/80218
24332 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
24333 Update block frequencies and counts.
24335 2017-03-28 Richard Biener <rguenther@suse.de>
24337 PR tree-optimization/78644
24338 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
24339 of a simplification result we may not use it at all.
24341 2017-03-28 Richard Biener <rguenther@suse.de>
24344 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
24345 without arguments, generate default definition of a SSA name.
24347 2017-03-28 Richard Biener <rguenther@suse.de>
24349 PR middle-end/80222
24350 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
24351 TYPE_REF_CAN_ALIAS_ALL references.
24352 * fold-const.c (fold_indirect_ref_1): Likewise.
24354 2017-03-28 Martin Liska <mliska@suse.cz>
24357 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
24358 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
24360 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
24361 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
24363 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
24364 (EXTRA_SPECS): Define.
24365 (SUBTARGET_EXTRA_SPECS): Likewise.
24366 (SUBTARGET_CPP_SPEC): Likewise.
24367 * config/arc/elf.h (EXTRA_SPECS): Renamed to
24368 SUBTARGET_EXTRA_SPECS.
24369 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
24371 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
24373 * config/arc/simdext.md (vst64_insn): Update pattern.
24374 (vld32wh_insn): Likewise.
24375 (vld32wl_insn): Likewise.
24376 (vld64_insn): Likewise.
24377 (vld32_insn): Likewise.
24379 2017-03-28 Marek Polacek <polacek@redhat.com>
24382 * fold-const.c (fold_comparison): Use protected_set_expr_location
24383 instead of SET_EXPR_LOCATION.
24385 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
24387 * tree.c (add_expr): Avoid name lookup warning.
24389 2017-03-27 Jeff Law <law@redhat.com>
24391 PR tree-optimization/80216
24392 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
24393 function name. Limit recursion depth.
24394 (record_temporary_equivalences): Corresponding changes.
24396 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
24398 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
24401 2017-03-27 Jakub Jelinek <jakub@redhat.com>
24404 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
24406 * cfgcleanup.c (reg_note_cfa_p): New array.
24407 (insns_have_identical_cfa_notes): New function.
24408 (old_insns_match_p): Don't cross-jump in between /f
24409 and non-/f instructions. If both i1 and i2 are frame related,
24410 verify all CFA notes, their order and content.
24412 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
24415 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
24416 HImode and SImode with zero extend to DImode to one insn.
24417 (bswap<mode>2_extenddi): Likewise.
24418 (bswapsi2_extenddi): Likewise.
24419 (bswaphi2_extendsi): Likewise.
24420 (bswaphi2): Combine bswap HImode and SImode into one insn.
24421 Separate memory insns from swapping register.
24422 (bswapsi2): Likewise.
24423 (bswap<mode>2): Likewise.
24424 (bswaphi2_internal): Delete, no longer used.
24425 (bswapsi2_internal): Likewise.
24426 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
24427 store, and gpr<-gpr swap insns.
24428 (bswap<mode>2_store): Likewise.
24429 (bswaphi2_reg): Register only splitter, combine with the splitter.
24430 (bswaphi2 splitter): Likewise.
24431 (bswapsi2_reg): Likewise.
24432 (bswapsi2 splitter): Likewise.
24433 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
24434 the insns into load, store, and register/register insns.
24435 (bswapdi2_ldbrx): Likewise.
24436 (bswapdi2_load): Likewise.
24437 (bswapdi2_store): Likewise.
24438 (bswapdi2_reg): Likewise.
24440 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
24442 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
24443 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
24445 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
24448 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
24450 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
24451 special handling for target option conflicts between dform
24452 options (-mpower9-dform, -mpower9-dform-vector,
24453 -mpower9-dform-scalar) and -mno-direct-move.
24455 2017-03-27 Richard Biener <rguenther@suse.de>
24457 PR tree-optimization/80181
24458 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
24460 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24462 * config/arc/predicates.md (move_double_src_operand): Replace the
24463 call to move_double_src_operand with a call to address_operand.
24465 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24467 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
24468 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
24469 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
24471 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24473 * config/arc/predicates.md (long_immediate_loadstore_operand):
24474 Consider scaled addresses cases.
24476 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24478 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
24479 restored when in interrupt.
24480 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
24481 doesn't have delay slot.
24483 2017-03-27 Richard Biener <rguenther@suse.de>
24486 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
24487 inlined thunk clones.
24489 2017-03-27 Jakub Jelinek <jakub@redhat.com>
24492 * asan.c (instrument_derefs): Copy over last operand from
24493 original COMPONENT_REF to the new COMPONENT_REF with
24494 DECL_BIT_FIELD_REPRESENTATIVE.
24495 * ubsan.c (instrument_object_size): Likewise.
24497 2017-03-27 Richard Biener <rguenther@suse.de>
24499 PR tree-optimization/80170
24500 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
24501 sure DR/SCEV didnt fold in constants we do not see when looking
24502 at the reference base alignment.
24504 2017-03-27 Richard Biener <rguenther@suse.de>
24506 PR middle-end/80171
24507 * gimple-fold.c (fold_ctor_reference): Properly guard against
24508 NULL return value from canonicalize_constructor_val.
24510 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
24513 * config/i386/i386.c (ix86_expand_builtin)
24514 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
24515 flags reg setting and flags reg using instructions.
24516 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
24517 clobbering instructions to zero extend op2.
24519 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
24521 * doc/install.texi (Configuration) <--with-aix-soname>:
24522 Update link to AIX ld.
24524 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
24526 PR rtl-optimization/80160
24527 PR rtl-optimization/80159
24528 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
24529 reg_alternate_class into account.
24531 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
24534 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
24535 to consider in curr_insn_transform.
24537 2017-03-24 Jakub Jelinek <jakub@redhat.com>
24539 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
24540 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
24541 and emit_mode_inner.
24543 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24545 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
24546 argument to the overloaded builtin variants. Use the new flag to
24547 deprecate certain builtin variants.
24548 * config/s390/s390-builtin-types.def: Add new builtin types.
24549 * config/s390/s390-builtins.h: Support new flags field for
24550 overloaded builtins.
24551 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
24552 (s390_macro_to_expand): Enable vector float data type.
24553 (s390_cpu_cpp_builtins_internal): Indicate support of the new
24554 builtins by incrementing the __VEC__ version number.
24555 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
24557 (s390_resolve_overloaded_builtin): Emit error messages depending
24558 on the builtin flags.
24559 * config/s390/s390.c (s390_expand_builtin): Support additional
24560 flags argument. Change error message to match the messages
24561 emitted in s390-c.c.
24562 * config/s390/s390.md: New UNSPEC_* constants.
24563 (op_type): Add new instruction types.
24564 * config/s390/vecintrin.h: Add new builtins and test data class
24566 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
24567 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
24568 (VEC_INEXACT, VEC_NOINEXACT): New constants.
24569 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
24570 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
24571 ("vec_mergel<mode>"): V_HW -> VEC_HW.
24573 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
24574 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
24575 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
24576 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
24578 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
24579 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
24580 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
24581 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
24583 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
24584 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
24585 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
24586 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
24587 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
24588 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
24589 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
24591 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
24592 ("vec_scatter_element<V_HW_4:mode>_DI")
24593 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
24594 ("vec_fpint<mode>", "vflls")
24595 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
24596 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
24597 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
24598 ("*vec_cmphe<mode>_cc"): ... these.
24600 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
24601 mode constant instead of magic value.
24603 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24605 * config/s390/s390.c (s390_expand_vec_compare): Support other
24606 vector floating point modes than just V2DF.
24607 (s390_expand_vcond): Likewise.
24608 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
24609 (s390_cannot_change_mode_class): Prevent mode changes between TF
24610 and V1TF in vector registers.
24611 * config/s390/s390.md (DF, SF): New mode attributes.
24612 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
24613 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
24614 SFmode support for VRs.
24615 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
24617 (VFT, VF_HW): New mode iterators.
24618 (vw, sdx): New mode attributes.
24619 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
24620 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
24621 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
24622 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
24623 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
24624 also the new vector floating point modes. Renaming to ...
24626 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
24627 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
24628 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
24629 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
24630 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
24631 ("vec_unordered<mode>"): ... these.
24633 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
24634 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
24635 ("*vec_extendv2df"): New insn definitions.
24637 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24639 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
24640 ("mulditi3_2", "*muldi3_sign"): New patterns.
24641 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
24642 rename the pattern definition.
24644 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24646 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
24648 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
24650 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24652 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
24653 instruction if possible.
24654 * config/s390/vector.md (vec_halfnumelts): New mode
24656 ("*vec_vllezlf<mode>"): New pattern.
24658 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24660 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
24661 ("popcountv4si2", "popcountv2di2"): Rename to ...
24662 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
24663 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
24665 ("popcount<mode>2_vxe"): New pattern.
24667 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24669 * common/config/s390/s390-common.c (processor_flags_table): Add
24671 * config.gcc: Add arch12.
24672 * config/s390/driver-native.c (s390_host_detect_local_cpu):
24673 Default to arch12 for unknown CPU model numbers.
24674 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
24675 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
24676 PROCESSOR_max sanity check.
24677 * config/s390/s390-opts.h (enum processor_type): Add
24679 * config/s390/s390.c (processor_table): Add arch12.
24680 (s390_expand_builtin): Add check for B_VXE flag.
24681 (s390_issue_rate): Add PROCESSOR_ARCH12.
24682 (s390_get_sched_attrmask): Likewise.
24683 (s390_get_unit_mask): Likewise.
24684 (s390_sched_score): Enable z13 scheduling for arch12.
24685 (s390_sched_reorder): Likewise.
24686 (s390_sched_variable_issue): Likewise.
24687 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
24689 (s390_tune_attr): Use z13 scheduling also for arch12.
24690 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
24691 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
24692 (TARGET_VXE_P): New macros.
24693 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
24694 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
24695 * config/s390/s390.opt: Add arch12 as processor_type.
24697 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24699 * config/s390/s390.md
24700 ("fixuns_truncdddi2", "fixuns_trunctddi2")
24701 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
24702 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
24704 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
24705 Rename expanders to ...
24707 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
24708 ("fixuns_truncdddi2_emu"): ... these.
24710 ("fixuns_trunc<mode>si2_emu"): New expander.
24712 ("*fixuns_truncdfdi2_z13"): Rename to ...
24713 ("*fixuns_truncdfdi2_vx"): ... this.
24715 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24717 * config/s390/2964.md: Remove the single element vector compare
24718 instructions which are no longer used.
24719 * config/s390/s390.c (s390_select_ccmode): Remove handling of
24721 (s390_canonicalize_comparison): Remove handling of DFmode
24723 (s390_expand_vec_compare_scalar): Remove function.
24724 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
24725 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
24727 ("*cmp<mode>_ccs"): Add wfcdb instruction.
24729 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24731 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
24733 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
24734 will anyway by matched by mov<mode>_64dfp.
24736 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24738 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
24739 vlef/vstef. Add missing operand to vleif.
24741 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24743 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
24744 pair for all vector types with 64 bit elements.
24745 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
24746 * config/s390/vector.md (V_HW_64): ... here.
24747 (V_128_NOSINGLE): New mode iterator.
24748 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
24749 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
24750 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
24751 ("*vec_load_pairv2di"): Change to ...
24752 ("*vec_load_pair<mode>"): ... this one.
24754 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24756 * config/s390/constraints.md: Add comments.
24757 (jKK): Reject element sizes > 8 bytes.
24758 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
24760 * config/s390/s390.md: Add the s_operand checks formerly in
24761 s390_split_ok_p to various splitters where they are still
24763 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
24764 for 128 bit vectors. Plus two splitters.
24766 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24768 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
24771 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24774 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
24775 error if the boundary argument is not constant.
24777 2017-03-24 Jakub Jelinek <jakub@redhat.com>
24779 PR rtl-optimization/80112
24780 * loop-doloop.c (doloop_condition_get): Don't check condition
24781 if cmp isn't SET with IF_THEN_ELSE src.
24783 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24785 PR tree-optimization/80158
24786 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
24787 replacing a candidate statement, also replace it for the
24788 candidate's alternate interpretation.
24789 (replace_rhs_if_not_dup): Likewise.
24790 (replace_one_candidate): Likewise.
24792 2017-03-24 Richard Biener <rguenther@suse.de>
24794 PR tree-optimization/80167
24795 * graphite-isl-ast-to-gimple.c
24796 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
24798 (translate_isl_ast_to_gimple::get_rename): Likewise.
24800 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
24802 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
24803 handling of certain combinations of target options, including the
24804 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
24805 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
24807 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24810 * config/arm/arm.md (*load_multiple): Add reload_completed to
24811 matching condition.
24813 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24814 Richard Biener <rguenth@suse.de>
24816 PR tree-optimization/79908
24817 PR tree-optimization/80136
24818 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
24819 been cast away, gimplify_and_add suffices.
24821 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
24823 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
24825 2017-03-23 Richard Biener <rguenther@suse.de>
24827 PR tree-optimization/80032
24828 * gimplify.c (gimple_push_cleanup): Forced unconditional
24829 cleanups still have to go to the conditional_cleanups
24832 2017-03-22 Jakub Jelinek <jakub@redhat.com>
24834 PR tree-optimization/80072
24835 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
24837 (next_operand_entry_id): Change type to unsigned int.
24838 (sort_by_operand_rank): Make sure to return the right return value
24839 even if unsigned fields are bigger than INT_MAX.
24840 (struct oecount): Change cnt and id type to unsigned int.
24841 (oecount_hasher::equal): Formatting fix.
24842 (oecount_cmp): Make sure to return the right return value
24843 even if unsigned fields are bigger than INT_MAX.
24844 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
24847 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
24848 TREE_READONLY on result if writing it more than once.
24851 * doc/invoke.texi (-fsanitize=thread): Document that with
24852 -fnon-call-exceptions atomics are not able to throw
24856 * tsan.c: Include tree-eh.h.
24857 (instrument_builtin_call): Call maybe_clean_eh_stmt or
24858 maybe_clean_or_replace_eh_stmt where needed.
24859 (instrument_memory_accesses): Add cfg_changed argument.
24860 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
24861 if it returned true.
24862 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
24864 PR rtl-optimization/63191
24865 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
24866 wrapper function, moved the whole old content into ...
24867 (ix86_delegitimize_address_1): ... this. New inline function.
24868 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
24869 true as last argument instead of ix86_delegitimize_address.
24871 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
24873 * config/aarch64/aarch64.c (generic_branch_cost): Copy
24874 cortexa57_branch_cost.
24876 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
24878 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
24880 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24883 * doc/md.texi (Constraints): Document wA constraint.
24884 * config/rs6000/constraints.md (wA): New.
24885 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
24886 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
24887 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
24888 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
24890 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
24893 * gimplify.c (is_oacc_declared): New function.
24894 (oacc_default_clause): Use it to set default flags for acc declared
24895 variables inside parallel regions.
24896 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
24897 declared variables.
24898 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
24899 declare attribute to any decl as necessary.
24901 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
24904 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
24905 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
24906 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
24907 (arm_arch_lpae): This.
24908 * config/arm/arm.c (arm_arch7ve): Rename into ...
24909 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
24910 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
24913 2017-03-22 Martin Liska <mliska@suse.cz>
24916 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
24917 error message instead of an ICE.
24919 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24921 * doc/extend.texi (6.11 Additional Floating Types): Revise.
24923 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
24925 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
24927 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
24930 2017-03-21 Martin Sebor <msebor@redhat.com>
24932 * doc/extend.texi: Use "cannot" instead of "can't."
24933 * doc/hostconfig.texi: Same.
24934 * doc/install.texi: Same.
24935 * doc/invoke.texi: Same.
24936 * doc/loop.texi: Same.
24937 * doc/md.texi: Same.
24938 * doc/objc.texi: Same.
24939 * doc/rtl.texi: Same.
24940 * doc/tm.texi: Same.
24941 * doc/tm.texi.in: Same.
24942 * doc/trouble.texi: Same.
24944 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
24947 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
24948 (collect_checksum_attributes): Set it.
24949 (die_checksum_ordered): Use it.
24951 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24953 PR tree-optimization/79908
24954 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
24955 change: For a VA_ARG whose LHS has been cast away, use
24956 force_gimple_operand to construct the side effects.
24958 2017-03-21 David Malcolm <dmalcolm@redhat.com>
24960 PR translation/80001
24961 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
24962 more amenable to translation.
24963 (oacc_loop_auto_partitions): Likewise.
24965 2017-03-21 Marek Polacek <polacek@redhat.com>
24966 Martin Sebor <msebor@redhat.com>
24968 PR tree-optimization/80109
24969 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
24970 on INTEGRAL_TYPE_P.
24972 2017-03-21 Jakub Jelinek <jakub@redhat.com>
24973 Segher Boessenkool <segher@kernel.crashing.org>
24976 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
24977 check reg_used_between_p between insn and one of succ or succ2
24978 depending on if succ is artificial insn not inserted into insn
24981 2017-03-21 Martin Liska <mliska@suse.cz>
24983 PR gcov-profile/80081
24984 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
24985 * doc/gcc.texi: Include gcov-dump stuff.
24986 * doc/gcov-dump.texi: New file.
24988 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
24990 PR rtl-optimization/79150
24991 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
24992 conditional jump, if the jump is the last insn of the loop.
24994 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24995 Richard Biener <rguenth@suse.de>
24997 PR tree-optimization/79908
24998 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
24999 been cast away, use force_gimple_operand to construct the side
25002 2017-03-21 Martin Liska <mliska@suse.cz>
25004 PR libfortran/79956
25005 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
25008 2017-03-21 Brad Spengler <spender@grsecurity.net>
25011 * plugin.c (htab_hash_plugin): New function.
25012 (add_new_plugin): Use it and adjust.
25013 (parse_plugin_arg_opt): Adjust.
25014 (init_one_plugin): Likewise.
25016 2017-03-21 Richard Biener <rguenther@suse.de>
25018 PR tree-optimization/80032
25019 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
25020 if set force the cleanup to happen unconditionally.
25021 (gimplify_target_expr): Push inserted clobbers with force_uncond
25022 to avoid them being removed by control-dependent DCE.
25024 2017-03-21 Richard Biener <rguenther@suse.de>
25026 PR tree-optimization/80122
25027 * tree-inline.c (copy_bb): Do not expans va-arg packs or
25028 va_arg_pack_len when the inlined call stmt requires pack
25030 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
25032 2017-03-21 Jakub Jelinek <jakub@redhat.com>
25035 * tsan.c (instrument_builtin_call): If the memory model argument
25036 is not a constant, assume it is valid.
25039 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
25042 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
25044 PR rtl-optimization/79910
25045 * combine.c (can_combine_p): Do not allow combining an I0 or I1
25046 if its dest is used by an insn before I2 (other than the combined
25047 insns themselves, which are properly handled already).
25049 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
25052 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
25054 * combine.c (record_used_regs): New static function.
25055 (try_combine): Handle situations where there is an additional
25056 instruction between I2 and I3 which needs to have a LOG_LINK
25060 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
25062 * combine.c (try_combine): Delete redundant i1 test. Call
25063 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
25065 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25068 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
25069 alternatives 13/14.
25071 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25073 PR tree-optimization/80054
25074 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
25075 the optimization if a PHI or any of its arguments is not dominated
25076 by the candidate's basis. Use gphi* rather than gimple* as
25078 (replace_profitable_candidates): Clean up a gimple* variable that
25079 should be a gphi* variable.
25081 2017-03-20 Martin Sebor <msebor@redhat.com>
25084 * doc/extend.texi (attribute constructor): Document present limitation.
25086 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
25089 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
25090 __POWER9_VECTOR__ #ifdef control, change template definition to
25091 use Power9-specific built-in function.
25092 (vec_any_eq): Likewise.
25093 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
25094 to control outcomes from this test.
25095 (vector_ae_<mode>p): For VEC_F modes, likewise.
25097 2017-03-20 Ian Lance Taylor <iant@google.com>
25099 * config/i386/i386.c (ix86_function_regparm): Save an extra
25100 register for -fsplit-stack with DECL_STATIC_CHAIN.
25102 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
25105 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
25106 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
25108 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
25110 * config/riscv/riscv.c (riscv_print_operand): Use "fence
25112 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
25115 2017-03-20 Marek Polacek <polacek@redhat.com>
25118 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
25120 2017-03-20 Richard Biener <rguenther@suse.de>
25122 PR tree-optimization/80113
25123 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
25124 allocate extra SSA name for PHI def.
25125 (add_close_phis_to_outer_loops): Likewise.
25126 (add_close_phis_to_merge_points): Likewise.
25127 (copy_loop_close_phi_args): Likewise.
25128 (copy_cond_phi_nodes): Likewise.
25130 2017-03-20 Martin Liska <mliska@suse.cz>
25132 PR middle-end/79753
25133 * tree-chkp.c (chkp_build_returned_bound): Do not build
25134 returned bounds for a LHS that's not a BOUNDED_P type.
25136 2017-03-20 Martin Liska <mliska@suse.cz>
25140 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
25141 COMPLEX_CST and VECTOR_CST.
25143 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25146 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
25147 target operand. A new splitter adds the clobber statement in case
25148 the target operand is dead anyway.
25150 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
25152 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
25153 to age-old versions of binutils and glibc.
25155 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
25157 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
25159 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
25161 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
25163 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
25165 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
25166 requirement for binutils 2.13.
25168 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
25170 * combine.c (try_combine): Delete redundant i1 test. Call
25171 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
25173 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
25175 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
25176 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
25178 <riscv64-*-elf>: Re-arrange section
25179 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
25180 <riscv32-*-linux>: Likewise.
25181 <riscv64-*-elf>: Likewise
25182 <riscv64-*-linux>: Likewise.
25184 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
25187 * aarch64.opt(verbose-cost-dump): Fix typo.
25189 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
25192 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
25193 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
25195 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
25197 * reload.c (find_reloads): When reloading a nonoffsettable address,
25198 use RELOAD_OTHER for it and its address reloads.
25200 PR rtl-optimization/79910
25201 * combine.c (record_used_regs): New static function.
25202 (try_combine): Handle situations where there is an additional
25203 instruction between I2 and I3 which needs to have a LOG_LINK
25206 2017-03-17 Jeff Law <law@redhat.com>
25208 PR tree-optimization/71437
25209 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
25210 conditional in the hash table first.
25211 (vrp_dom_walker::before_dom_children): Extract condition from
25212 ASSERT_EXPR. Record condition, its inverion and any implied
25213 conditions as well.
25215 2017-03-17 Marek Polacek <polacek@redhat.com>
25216 Markus Trippelsdorf <markus@trippelsdorf.de>
25218 PR tree-optimization/80079
25219 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
25222 2017-03-17 Richard Biener <rguenther@suse.de>
25224 PR middle-end/80075
25225 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
25226 Properly verify the LHS before the RHS possibly claims to be
25228 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
25231 2017-03-17 Martin Jambor <mjambor@suse.cz>
25233 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
25234 (List of -O2 options): Likewise.
25235 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
25238 2017-03-17 Tom de Vries <tom@codesourcery.com>
25240 * gcov-dump.c (print_usage): Print bug_report_url.
25242 2017-03-17 Richard Biener <rguenther@suse.de>
25244 PR middle-end/80050
25245 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
25246 (parser::peek): Likewise.
25248 2017-03-17 Richard Biener <rguenther@suse.de>
25250 PR tree-optimization/80048
25251 * sese.c (free_sese_info): Properly release rename_map and
25252 copied_bb_map elements.
25254 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
25256 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
25257 Add linked-list forward and backlinks. Insert on
25258 construction, remove on destruction.
25259 (class pass_store_merging): Add m_stores_head field.
25260 (pass_store_merging::terminate_and_process_all_chains):
25261 Iterate over m_stores_head list.
25262 (pass_store_merging::terminate_all_aliasing_chains):
25264 (pass_store_merging::execute): Check for debug stmts first.
25265 Push new chains onto the m_stores_head stack.
25267 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
25270 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
25271 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
25272 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
25274 2017-03-16 Jeff Law <law@redhat.com>
25276 PR tree-optimization/71437
25277 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
25278 member function. Implementation moved into after_dom_children
25279 member function and into the threader's thread_outgoing_edges
25281 (dom_opt_dom_walker::after_dom_children): Simplify by moving
25282 some code into new thread_outgoing_edges.
25283 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
25284 definition. Simplify marker handling (do it here). Assume we always
25285 have the available expression and the const/copies tables.
25286 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
25288 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
25289 * tree-vrp.c (equiv_stack): No longer file scoped.
25290 (vrp_dom_walker): New class.
25291 (vrp_dom_walker::before_dom_children): New member function.
25292 (vrp_dom_walker::after_dom_children): Likewise.
25293 (identify_jump_threads): Setup domwalker. Use it rather than
25294 walking edges in a random order by hand. Simplify setup/finalization.
25295 (finalize_jump_threads): Remove.
25296 (vrp_finalize): Do not call identify_jump_threads here.
25297 (execute_vrp): Do it here instead and call thread_through_all_blocks
25300 PR tree-optimization/71437
25301 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
25303 (simplify_stmt_for_jump_threading): Add basic_block argument. All
25305 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
25306 (dom_opt_dom_walker::thread_across_edge): Remove
25307 handle_dominating_asserts argument. All callers changed.
25308 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
25309 changes. Remove calls to lhs_of_dominating_assert. Other
25310 uses of handle_dominating_asserts turn into unconditional code
25311 (simplify_control_stmt_condition_1): Likewise.
25312 (simplify_control_stmt_condition): Likewise.
25313 (thread_through_normal_block, thread_across_edge): Likewise.
25314 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
25315 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
25316 object if it is not an SSA_NAME.
25317 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
25318 before calling into the VRP specific simplifiers.
25319 (identify_jump_threads): Remove handle_dominating_asserts
25322 2017-03-16 Jakub Jelinek <jakub@redhat.com>
25325 * tree-diagnostic.c (default_tree_printer): No longer static.
25326 * tree-diagnostic.h (default_tree_printer): New prototype.
25328 2017-03-16 Tamar Christina <tamar.christina@arm.com>
25330 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
25331 Change ins into fmov.
25333 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25335 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
25336 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
25337 Use h_con constraint for operand 1.
25338 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
25339 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
25341 2017-03-15 Jeff Law <law@redhat.com>
25343 PR tree-optimization/71437
25344 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
25345 (record_temporary_equivalences): Use it.
25347 PR tree-optimization/71437
25348 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
25349 tree-ssa-scopedtables.
25350 (lookup_avail_expr, build_and_record_new_cond): Likewise.
25351 (record_conditions, record_cond, vuse_eq): Likewise.
25352 (record_edge_info): Adjust to API tweak of record_conditions.
25353 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
25354 (record_temporary_equivalences, optimize_stmt): Likewise.
25355 (eliminate_redundant_computations): Likewise.
25356 (record_equivalences_from_stmt): Likewise.
25357 * tree-ssa-scopedtables.c: Include options.h and params.h.
25358 (vuse_eq): New function, moved from tree-ssa-dom.c
25359 (build_and_record_new_cond): Likewise.
25360 (record_conditions): Likewise. Accept vector of conditions rather
25361 than edge_equivalence structure for first argument.
25362 for the first argument.
25363 (avail_exprs_stack::lookup_avail_expr): New member function, moved
25364 from tree-ssa-dom.c.
25365 (avail_exprs_stack::record_cond): Likewise.
25366 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
25367 from tree-ssa-dom.c.
25368 (avail_exprs_stack): Add new member functions lookup_avail_expr
25370 (record_conditions): Declare.
25372 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
25375 * lra-constraints.c (process_alt_operands): Increase reject for
25376 reloading an input/output operand.
25378 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
25381 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
25382 insns to convert from signed/unsigned char/short to IEEE 128-bit
25384 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
25386 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
25389 * config/i386/i386.c (ix86_vector_duplicate_value): Create
25390 subreg of inner mode for values already in registers.
25392 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
25394 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
25395 iteration reg is used after the loop.
25397 2017-03-14 Martin Sebor <msebor@redhat.com>
25399 PR tree-optimization/79800
25400 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
25401 precision in negative-positive range.
25402 (format_floating): Call non-const overload with adjusted precision.
25404 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
25407 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
25410 2017-03-14 Martin Sebor <msebor@redhat.com>
25412 PR middle-end/80020
25413 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
25414 * builtins.def (aligned_alloc): Use it.
25417 * Makefile.in (GTFILES): Add calls.c.
25418 * calls.c: Include "gt-calls.h".
25420 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
25422 PR rtl-optimization/79728
25423 * regs.h (struct target_regs): New field
25424 x_contains_allocatable_regs_of_mode.
25425 (contains_allocatable_regs_of_mode): New macro.
25426 * reginfo.c (init_reg_sets_1): Initialize it, and change
25427 contains_reg_of_mode so it includes global regs as well.
25428 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
25429 rather than contains_regs_of_mode.
25431 2017-03-14 Martin Liska <mliska@suse.cz>
25433 * doc/invoke.texi: Document options that can't be combined with
25434 -fcheck-pointer-bounds.
25436 2017-03-14 Martin Liska <mliska@suse.cz>
25438 PR middle-end/79831
25439 * doc/invoke.texi (-Wchkp): Document the option.
25441 2017-03-14 Martin Liska <mliska@suse.cz>
25443 * Makefile.in: Install gcov-dump.
25445 2017-03-14 Martin Liska <mliska@suse.cz>
25447 * multiple_target.c (expand_target_clones): Bail out for
25448 an invalid attribute.
25450 2017-03-14 Richard Biener <rguenther@suse.de>
25452 * alias.c (struct alias_set_entry): Pack properly.
25453 * cfgloop.h (struct loop): Likewise.
25454 * cse.c (struct set): Likewise.
25455 * ipa-utils.c (struct searchc_env): Likewise.
25456 * loop-invariant.c (struct invariant): Likewise.
25457 * lra-remat.c (struct cand): Likewise.
25458 * recog.c (struct change_t): Likewise.
25459 * rtl.h (struct address_info): Likewise.
25460 * symbol-summary.h (function_summary): Likewise.
25461 * tree-loop-distribution.c (struct partition): Likewise.
25462 * tree-object-size.c (struct object_size_info): Likewise.
25463 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
25464 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
25465 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
25466 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
25467 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
25468 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
25469 (struct _stmt_vec_info): Likewise.
25471 2017-03-14 Martin Liska <mliska@suse.cz>
25474 * multiple_target.c (create_dispatcher_calls): Check that
25475 a target can create a function dispatcher.
25477 2017-03-14 Martin Liska <mliska@suse.cz>
25480 * multiple_target.c (expand_target_clones): Drop local.local
25481 flag for default implementation.
25483 2017-03-14 Richard Biener <rguenther@suse.de>
25485 PR tree-optimization/80030
25486 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
25488 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
25490 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
25491 gcc_fallthrough() instead of __attribute__((fallthrough));
25493 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
25495 * doc/gcc.texi: Remove "up" link to (DIR).
25496 * doc/gccint.texi: Ditto.
25498 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
25500 * doc/install.texi (Specific) <avr>: Remove reference to
25503 2017-03-13 Jeff Law <law@redhat.com>
25505 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
25506 attribute rather than comments.
25508 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
25509 match_scratch operand is highest.
25511 2017-03-13 Martin Liska <mliska@suse.cz>
25513 PR middle-end/78339
25514 * ipa-pure-const.c (warn_function_noreturn): If the declarations
25515 is a CHKP clone, use original declaration.
25517 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25519 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
25520 (arc_conditional_register_usage): Use a different allocation order
25521 when optimizing for size.
25522 * common/config/arc/arc-common.c (arc_option_optimization_table):
25523 Section anchors default on when optimizing for size.
25525 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25527 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
25529 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25531 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
25532 * config/arc/arc.md (cpu_facility): Add cd variant.
25533 (*movqi_insn): Add code density variant.
25534 (*movhi_insn): Likewise.
25535 (*movqi_insn): Likewise.
25536 (*addsi3_mixed): Likewise.
25537 (subsi3_insn): Likewise.
25539 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25541 * config/arc/arc.md (movsi_cond_exec): Update constraint.
25543 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25545 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
25546 expressions with MINUS and UNARY ops.
25548 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25551 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
25553 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
25554 between vec_select and vector argument.
25555 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
25556 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
25557 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
25558 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
25559 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
25560 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
25562 2017-03-13 Richard Biener <rguenther@suse.de>
25565 * params.def (vect-max-peeling-for-alignment): Fix typo.
25567 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
25569 * doc/install.texi (Specific) <mips-*-*>: Remove description of
25570 issue that only occurred with binutils below 2.18.
25572 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
25574 * doc/install.texi (Specific) <cris-axis-elf>: No longer
25575 refer to binutils 2.11/2.12 minimum.
25577 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
25579 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
25580 ftp.kernel.org and simplify binutils requirement.
25582 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
25584 * doc/invoke.texi (Warning Options): Fix spelling of link-time
25586 (Optimize Options): Ditto. Also remove redundancy.
25588 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25590 PR translation/79848
25591 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
25593 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
25594 to G_ to avoid double translation.
25596 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25598 PR translation/79923
25599 * auto-profile.c (get_combined_location): Convert leading
25600 character of diagnostics to lower case and remove trailing period.
25601 (read_profile): Likewise for various diagnostics.
25602 * config/arm/arm.c (arm_option_override): Remove trailing period
25603 from various diagnostics.
25604 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
25605 (msp430_expand_delay_cycles): Likewise.
25607 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25610 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
25611 full command-line argument, rather than just "str".
25612 (aarch64_validate_march): Likewise.
25613 (aarch64_validate_mtune): Likewise.
25615 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
25617 PR rtl-optimization/78911
25618 * lra-assigns.c (must_not_spill_p): New function.
25619 (spill_for): Use it.
25621 2017-03-10 Jakub Jelinek <jakub@redhat.com>
25623 PR tree-optimization/79981
25624 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
25625 ATOMIC_COMPARE_EXCHANGE ifn result.
25626 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
25627 IFN_ATOMIC_COMPARE_EXCHANGE.
25629 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25632 * opts.c (parse_sanitizer_options): Add missing question mark to
25633 "did you mean" message.
25635 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25637 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
25639 (VMULEUH_UNS): Likewise.
25640 (VMULOUB_UNS): Likewise.
25641 (VMULOUH_UNS): Likewise.
25642 * config/rs6000/rs6000.c (builtin_function_type): Remove
25643 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
25645 2017-03-10 David Malcolm <dmalcolm@redhat.com>
25648 * read-rtl-function.c (function_reader::read_rtx_operand): Update
25649 x with result of extra_parsing_for_operand_code_0.
25650 (function_reader::extra_parsing_for_operand_code_0): Convert
25651 return type from void to rtx, returning x. When reading
25652 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
25653 larger size containing struct block_symbol.
25655 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
25657 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
25658 -mfloat128-hardware without -m64.
25660 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
25663 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
25664 entries to the case statement that marks unsigned arguments to
25665 overloaded functions.
25667 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
25669 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
25670 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
25672 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
25675 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
25676 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
25678 2017-03-10 Martin Liska <mliska@suse.cz>
25682 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
25683 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
25686 2017-03-10 Olivier Hainque <hainque@adacore.com>
25688 * tree-switch-conversion (array_value_type): Start by resetting
25689 candidate type to it's main variant.
25691 2017-03-10 Jakub Jelinek <jakub@redhat.com>
25693 PR rtl-optimization/79909
25694 * combine.c (try_combine): Use simplify_replace_rtx on individual
25695 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
25696 of the whole CALL_INSN_FUNCTION_USAGE.
25698 PR tree-optimization/79972
25699 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
25700 get_range_info on SSA_NAMEs. Formatting fixes.
25702 2017-03-10 Richard Biener <rguenther@suse.de>
25703 Jakub Jelinek <jakub@redhat.com>
25705 PR tree-optimization/77975
25706 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
25707 edge to be constant.
25708 (get_val_for): For constant x return it. Formatting fix.
25709 (loop_niter_by_eval): Avoid pointless looping if the next iteration
25710 would use the same bases as the current one.
25712 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25714 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
25715 instead of vec_select for V1TImode.
25716 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
25718 (VSX_LE_128): Add V1TI to this mode iterator.
25719 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
25720 (*vsx_le_perm_store_<mode>): Likewise.
25721 (pre-reload splitter for VSX stores): Likewise.
25722 (post-reload splitter for VSX stores): Likewise.
25723 (*vsx_xxpermdi2_le_<mode>): Likewise.
25724 (*vsx_lxvd2x2_le_<mode>): Likewise.
25725 (*vsx_stxvd2x2_le_<mode>): Likewise.
25727 2017-03-09 Michael Eager <eager@eagercon.com>
25729 Correct failures with --enable-checking=yes,rtl.
25731 * config/microblaze/microblaze.c (microblaze_expand_shift):
25732 Replace GET_CODE test with CONST_INT_P and INTVAL test with
25733 test for const0_rtx.
25734 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
25735 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
25737 2017-03-09 Richard Biener <rguenther@suse.de>
25739 PR tree-optimization/79977
25740 * graphite-scop-detection.c (scop_detection::merge_sese):
25741 Handle the case of extra exits to blocks dominating the entry.
25743 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
25745 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
25748 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
25750 PR rtl-optimization/79949
25751 * lra-constraints.c (process_alt_operands): Check memory when
25752 trying to predict a cycle. Print about the overall increase.
25754 2017-03-09 Richard Biener <rguenther@suse.de>
25756 PR middle-end/79971
25757 * gimple-expr.c (useless_type_conversion_p): Preserve
25758 TYPE_SATURATING for fixed-point types.
25760 2017-03-09 Richard Biener <rguenther@suse.de>
25763 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
25764 alignment of BLKmode params.
25766 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25769 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
25770 (VALL_NO_V2Q): Likewise.
25772 * config/aarch64/aarch64-simd.md
25773 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
25775 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
25776 VALL_NO_V2Q mode iterator.
25777 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
25779 2017-03-09 Martin Liska <mliska@suse.cz>
25781 PR tree-optimization/79631
25782 * tree-chkp-opt.c (chkp_is_constant_addr): Call
25783 tree_int_cst_sign_bit just for INTEGER constants.
25785 2017-03-09 Martin Liska <mliska@suse.cz>
25789 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
25792 2017-03-09 Marek Polacek <polacek@redhat.com>
25795 * tree.c (inchash::add_expr): Handle TREE_VEC.
25797 2017-03-09 Martin Liska <mliska@suse.cz>
25800 (chkp_narrow_size_and_offset): New function.
25801 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
25802 (void chkp_parse_bit_field_ref): New function.
25803 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
25804 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
25806 2017-03-09 Martin Liska <mliska@suse.cz>
25809 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
25810 (chkp_find_bounds_1): Remove gcc_unreachable.
25812 2017-03-09 Jakub Jelinek <jakub@redhat.com>
25815 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
25816 BUILT_IN_SYNC*, determine the access type from the size suffix and
25817 always build a MEM_REF with that type. Handle forgotten
25818 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
25821 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
25822 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
25823 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
25824 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
25825 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
25826 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
25827 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
25828 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
25829 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
25830 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
25831 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
25832 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
25833 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
25834 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
25835 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
25836 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
25837 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
25838 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
25839 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
25840 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
25841 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
25842 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
25843 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
25844 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
25845 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
25846 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
25847 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
25848 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
25849 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
25850 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
25851 definitions outside of __OPTIMIZE__ guarded section.
25854 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
25855 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
25856 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
25857 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
25860 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25862 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
25863 ("vfenez<mode>"): Add missing constraints.
25865 2017-03-08 Martin Sebor <msebor@redhat.com>
25868 * config/nds32/nds32.c (nds32_option_override):
25869 Fix misspelled diagnostic.
25871 2017-03-08 Jakub Jelinek <jakub@redhat.com>
25874 * gimplify.c (gimplify_omp_for): Replace index var in outer
25875 taskloop statement with an artificial variable and add
25876 OMP_CLAUSE_PRIVATE clause for it.
25878 2017-03-08 Richard Biener <rguenther@suse.de>
25880 PR tree-optimization/79955
25881 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
25882 for accesses that are completely outside of the variable.
25884 2017-03-08 Andrew Haley <aph@redhat.com>
25886 PR tree-optimization/79943
25887 * tree-ssa-loop-split.c (compute_new_first_bound): When
25888 calculating the new upper bound, (END-BEG) should be added, not
25891 2017-03-08 Jakub Jelinek <jakub@redhat.com>
25893 * config/avr/avr.md (setmemhi): Make sure match_dup
25894 operand number comes before match_scratch.
25896 2017-03-08 Richard Biener <rguenther@suse.de>
25898 PR tree-optimization/79920
25899 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
25900 with ncopies == 1 to ...
25901 (vect_transform_slp_perm_load): ... here. Properly compute
25902 all element loads by iterating VF times over the group. Do
25903 not handle ncopies (computed in a broken way) in
25904 vect_create_mask_and_perm.
25906 2017-03-08 Jakub Jelinek <jakub@redhat.com>
25909 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
25910 is a uniform vector, use uniform_vector_p return value instead of
25911 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
25913 2017-03-07 Marek Polacek <polacek@redhat.com>
25915 PR middle-end/79809
25916 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
25917 (alloca_call_type): Likewise.
25919 2017-03-07 Martin Liska <mliska@suse.cz>
25921 * gcov.c (process_args): Put comment to correct location.
25923 2017-03-07 Martin Liska <mliska@suse.cz>
25925 PR middle-end/68270
25926 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
25927 Use array_at_struct_end_p instead of DECL_CHAIN (field).
25928 (chkp_narrow_bounds_for_field): Likewise.
25929 (chkp_parse_array_and_component_ref): Pass one more argument to
25932 2017-03-07 Richard Biener <rguenther@suse.de>
25934 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
25937 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
25939 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
25940 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
25942 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25945 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
25946 to end of description.
25947 (PARAM_MAX_STORES_TO_MERGE): Likewise.
25949 2017-03-07 Jakub Jelinek <jakub@redhat.com>
25951 PR rtl-optimization/79901
25952 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
25954 (*avx512f_<code><mode>3<mask_name>): ... this.
25955 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
25956 iterator instead of VI8_AVX2_AVX512BW.
25958 PR rtl-optimization/79901
25959 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
25960 min/max expander, expand it using expand_vec_cond_expr.
25963 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
25966 2017-03-06 Jakub Jelinek <jakub@redhat.com>
25969 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
25970 to void * for PCH reasons.
25971 * dwarf2out.c (output_loc_operands, output_die): Cast
25972 v.val_vec.array to unsigned char *.
25974 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
25977 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
25980 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
25982 PR rtl-optimization/79571
25983 * lra-constraints.c (process_alt_operands): Calculate static
25984 reject and subtract it from overall when only addresses will be
25987 2017-03-06 Julia Koval <julia.koval@intel.com>
25990 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
25991 incoming stack boundary to 128 for 64-bit targets.
25993 2017-03-06 Richard Biener <rguenther@suse.de>
25995 PR tree-optimization/79894
25996 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
25997 to NULL after folding it.
25999 2017-03-06 Richard Biener <rguenther@suse.de>
26001 PR tree-optimization/79824
26002 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
26003 check disabling peeling for gaps.
26005 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
26007 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
26008 attributes): Document gettimeofday.
26010 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
26012 * config/s390/s390.c (s390_option_override_internal): Set
26013 PARAM_MIN_VECT_LOOP_BOUND
26015 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
26017 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
26018 * config/s390/s390.md: Likewise.
26020 2017-03-06 Jakub Jelinek <jakub@redhat.com>
26023 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
26024 (<avx2_avx512>_perm<mode>): Rename to ...
26025 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
26027 (<avx512>_perm<mode>_mask): Rename to ...
26028 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
26030 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
26031 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
26032 instead of VI8F_256_512.
26033 (avx512f_perm<mode>): New define_expand.
26034 (avx512f_perm<mode>_mask): Likewise.
26035 (avx512f_perm<mode>_1<mask_name>): New define_insn.
26036 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
26038 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
26040 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
26041 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
26043 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
26045 2017-03-06 Martin Liska <mliska@suse.cz>
26048 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
26049 when having a SSA NAME w/o VAR_DECL assigned to it.
26051 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
26053 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
26054 msa_dpsub_<su>_d): Fix MODE for vec_select.
26056 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
26058 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
26060 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
26062 2017-03-06 Richard Biener <rguenther@suse.de>
26064 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
26065 * plugin.c (register_plugin_info): Likewise.
26066 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
26068 2017-03-05 Jakub Jelinek <jakub@redhat.com>
26070 * config/i386/sse.md (sse_storehps, sse_storelps,
26071 avx_<castmode><avxsizesuffix>_<castmode>,
26072 avx512f_<castmode><avxsizesuffix>_<castmode>,
26073 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
26074 in condition that at least one operand is not a MEM.
26076 2017-03-03 Jakub Jelinek <jakub@redhat.com>
26078 PR middle-end/79805
26079 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
26080 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
26082 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
26083 gimple_call_nothrow_p flag based on whether original builtin can throw.
26084 If it can, emit following stmts on the fallthrough edge.
26085 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
26086 don't create new bb if inserting just debug stmts on the edge, try to
26087 insert them on the fallthru bb or just reset debug stmts.
26089 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
26092 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
26093 restore recog_data (including the operand rtxes inside it) around
26094 the call to get_insn_template.
26096 2017-03-03 Martin Sebor <msebor@redhat.com>
26098 PR tree-optimization/79699
26099 * context.c (context::~context): Free MPFR caches to avoid
26100 a memory leak on program exit.
26102 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26104 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
26105 Use wide_int::ulow () instead of .elt (0).
26107 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
26109 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
26110 (*pushxf): Limit oF constraint to 32bit targets and add oC
26111 constraint for 64bit targets.
26112 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
26113 (*pushdf): Change rmF constraint to rmC.
26115 2017-03-03 Martin Liska <mliska@suse.cz>
26117 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
26118 Remove unused variable.
26120 2017-03-03 Jakub Jelinek <jakub@redhat.com>
26123 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
26124 is a memory operand, increase num_memory.
26125 (ix86_expand_args_builtin): Likewise.
26127 2017-03-03 Jan Hubicka <jh@suse.cz>
26130 * ipa-devirt.c (maybe_record_node): Properly handle
26131 __cxa_pure_virtual visibility.
26133 2017-03-03 Martin Liska <mliska@suse.cz>
26135 PR tree-optimization/79803
26136 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
26138 (pass_loop_prefetch::execute): Disabled optimization if an
26139 assumption about L1 cache size is not met.
26141 2017-03-03 Martin Liska <mliska@suse.cz>
26143 PR rtl-optimization/79574
26144 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
26145 (hash_scan_set): Likewise.
26146 (dump_hash_table): Likewise.
26147 (hoist_code): Likewise.
26149 2017-03-03 Richard Biener <rguenther@suse.de>
26151 * fixed-value.c (fixed_from_string): Restore use of elt (1)
26152 in place of uhigh ().
26153 (fixed_convert_from_real): Likewise.
26155 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
26158 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
26160 2017-03-03 Richard Biener <rguenther@suse.de>
26162 PR middle-end/79818
26163 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
26164 TYPE_OVERFLOW_UNDEFINED check.
26166 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26168 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
26170 (vector_ae_<mode>_p): Likewise.
26171 (vector_nez_<mode>_p): Likewise.
26172 (vector_ne_v2di_p): Likewise.
26173 (vector_ae_v2di_p): Likewise.
26174 (vector_ne_<mode>_p): Likewise.
26175 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
26177 (vsx_tsqrt<mode>2_fe): Likewise.
26179 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
26182 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
26184 2017-03-02 Jakub Jelinek <jakub@redhat.com>
26186 PR rtl-optimization/79780
26187 * cprop.c (one_cprop_pass): When second and further conditional trap
26188 in a single basic block is turned into an unconditional trap, turn it
26189 into a deleted note to avoid RTL verification failures.
26191 2017-03-02 Richard Biener <rguenther@suse.de>
26193 * fold-const.c (const_binop): Use ulow () instead of elt (0).
26195 2017-03-02 Richard Biener <rguenther@suse.de>
26197 PR tree-optimization/79345
26199 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
26200 param and abort the walk, returning -1 if it is hit.
26201 (walk_aliased_vdefs): Take a limit param and pass it on.
26202 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
26203 defaulting to 0 and return a signed int.
26204 * tree-ssa-uninit.c (struct check_defs_data): New struct.
26205 (check_defs): New helper.
26206 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
26207 about uninitialized memory.
26208 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
26209 bogus uninitialized warning.
26210 (fixed_convert_from_real): Likewise.
26212 2017-03-02 Bin Cheng <bin.cheng@arm.com>
26214 PR tree-optimization/66768
26215 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
26216 iv_use if base object can't be determined.
26218 2017-03-02 Jakub Jelinek <jakub@redhat.com>
26220 PR tree-optimization/79345
26221 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
26222 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
26223 (get_pattern_stats): Initialize it.
26224 * genemit.c (gen_expand): Verify match_scratch numbers come after
26225 match_operand/match_dup numbers.
26226 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
26227 match_scratch numbers.
26228 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
26230 * config/s390/s390.md (trunctdsd2): Likewise.
26232 2017-03-02 Richard Biener <rguenther@suse.de>
26234 * wide-int.h (wide_int_storage::operator=): Implement in terms
26237 2017-03-02 Richard Biener <rguenther@suse.de>
26239 PR tree-optimization/79777
26240 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
26241 the to insert expression to sth existing.
26243 2017-03-01 Martin Sebor <msebor@redhat.com>
26245 PR middle-end/79692
26246 * gimple-ssa-sprintf.c
26247 (directive::known_width_and_precision): New function.
26248 (format_integer): Use it.
26249 (get_mpfr_format_length): Consider the full range of precision
26250 when computing %g output with the # flag. Set the likely byte
26251 count to 3 rather than 1 when precision is indeterminate.
26252 (format_floating): Correct the lower bound of precision.
26254 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26256 * doc/invoke.texi: Document default code model for 64-bit Linux.
26258 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26261 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
26262 udiv rather than div since input pattern is unsigned.
26264 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
26266 * config/i386/i386.c (print_reg): Warn for values of
26267 unsupported size in integer register.
26269 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
26272 * config/rs6000/predicates.md (current_file_function_operand): Do
26273 not allow self calls to be local if the function is replaceable.
26275 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
26278 * config/rs6000/altivec.h (vec_ctz and others): Change the
26279 preprocessor macro that controls conditional compilation from
26280 _ARCH_PWR9 to __POWER9_VECTOR__.
26281 (vec_all_ne): Change parameterization of __altivec_scalar_pred
26282 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
26283 control (instead of _ARCH_PWR9 control) so that template
26284 definition uses power9-specific function.
26285 (vec_any_eq): Likewise.
26286 (vec_all_ne): Change macro definition to use a power9-specific
26287 expansion under #ifdef __POWER9_VECTOR__ control (instead of
26288 _ARCH_PWR9 control).
26289 (vec_any_eq) Likewise.
26290 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
26291 expansion for CMPNEF to remove support for xvcmpnesp instruction.
26292 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
26293 support for xvcmpnedp instruction.
26294 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
26295 macro expansion so that Power9 implementation of vec_all_ne does
26296 not use the AltiVec predicate framework.
26297 (VCMPNEH_P): Likewise.
26298 (VCMPNEW_P): Likewise.
26299 (VCMPNED_P): Likewise.
26300 (VCMPNEFP_P): Likewise.
26301 (VCMPNEDP_P): Likewise.
26302 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
26303 implementation of vec_any_eq to not use AltiVec predicate
26305 (VCMPAEH_P): Likewise.
26306 (VCMPAEW_P): Likewise.
26307 (VCMPAED_P): Likewise.
26308 (VCMPAEFP_P): Likewise.
26309 (VCMPAEDP_P): Likewise.
26310 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
26311 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
26312 not use the AltiVec predicate framework.
26313 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
26314 of vec_any_eq to not use AltiVec predicate framework.
26315 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
26316 support for predefined __POWER9_VECTOR__ macro to indicate that
26317 Power9 instruction selection is enabled.
26318 (altivec_overloaded_builtins): Remove extraneous
26319 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
26320 function argument types RS6000_BTI_bool_V16QI and
26321 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
26322 entry for overloaded function argument types RS6000_BTI_bool_V4SI
26323 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
26324 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
26325 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
26326 Power9 for implementations of vec_cmpne. Change the signature for
26327 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
26328 (representing vec_all_ne) to remove the previously described first
26329 argument of type RS6000_BTI_INTSI, as this was an artifact of
26330 reliance on the AltiVec predicate framework, which is no longer
26331 used in the implementation of these functions. Add
26332 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
26333 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
26334 since, unlike the AltiVec predicate framework implementation, we
26335 do not share function descriptors between vec_alle and vec_anyeq.
26336 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
26337 set of modes that receive special treatment even when
26338 TARGET_P9_VECTOR is true. The special treatment emits code that
26339 does not depend on Power9 instructions.
26340 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
26341 define_expand to not rely on AltiVec predicate framework.
26342 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
26344 (vector_ne_v2di_p): Change this define_expand to not rely on
26345 AltiVec predicate framework.
26346 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
26348 (vector_ne_<mode>_p): Change this define_expand to not rely on
26349 AltiVec predicate framework.
26350 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
26352 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
26353 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
26354 define_insn pattern.
26355 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
26356 define_insn pattern because the xvcmpne<VSs>. instruction is not
26358 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
26359 instruction is not supported.
26361 2017-03-01 Jakub Jelinek <jakub@redhat.com>
26363 * config/nvptx/nvptx.c: Include intl.h.
26365 2017-03-01 Martin Jambor <mjambor@suse.cz>
26368 * ipa-prop.h (ipa_bits): Removed field known.
26369 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
26370 to pointers. Adjusted their comments to warn about their sharing.
26371 (ipcp_transformation_summary): Change bits to a vector of pointers.
26372 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
26373 (ipa_get_ipa_bits_for_value): Declare.
26374 * tree-vrp.h (value_range): Mark as GTY((for_user)).
26375 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
26376 (ipa_bits_hash_table): Likewise.
26377 (ipa_vr_ggc_hash_traits): Likewise.
26378 (ipa_vr_hash_table): Likewise.
26379 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
26380 being pointers and vr_known being removed.
26381 (ipa_set_jf_unknown): Likewise.
26382 (ipa_get_ipa_bits_for_value): New function.
26383 (ipa_set_jfunc_bits): Likewise.
26384 (ipa_get_value_range): New overloaded functions.
26385 (ipa_set_jfunc_vr): Likewise.
26386 (ipa_compute_jump_functions_for_edge): Use the above functions to
26387 construct bits and vr parts of jump functions.
26388 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
26389 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
26391 (ipcp_grow_transformations_if_necessary): Also allocate
26392 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
26394 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
26395 them. Fix too long lines.
26396 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
26397 vr_known being removed.
26398 (ipa_read_jump_function): Use new setter functions to construct bits
26399 and vr parts of jump functions or set them to NULL.
26400 (write_ipcp_transformation_info): Adjust for bits being pointers.
26401 (read_ipcp_transformation_info): Likewise.
26402 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
26404 Include gt-ipa-prop.h.
26405 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
26407 (ipcp_store_bits_results): Likewise.
26408 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
26409 Do not write to existing jump functions but use a temporary instead.
26411 2017-03-01 Jakub Jelinek <jakub@redhat.com>
26414 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
26415 attempt to use its first operand as BIT_FIELD_REF base.
26417 2017-03-01 Richard Biener <rguenther@suse.de>
26419 PR middle-end/79721
26420 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
26421 interpolating formula in wrapping arithmetic.
26422 (chrec_apply): Convert chrec_evaluate return value to wanted type.
26424 2017-03-01 Jakub Jelinek <jakub@redhat.com>
26426 PR tree-optimization/79734
26427 * tree-vect-generic.c (expand_vector_condition): Optimize
26428 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
26429 Handle VEC_COND_EXPR where comparison has different inner width from
26430 type's inner width.
26432 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
26434 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
26435 markup, and similar issues. Remove @opindex entries for things
26436 that aren't options. Add missing -mmpy-option entries.
26438 2017-02-28 Jakub Jelinek <jakub@redhat.com>
26440 PR tree-optimization/79737
26441 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
26442 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
26443 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
26444 instead of byte_size. Formatting fix.
26445 (shift_bytes_in_array_right): Formatting fix.
26447 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
26450 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
26451 condition on optimize for the leaf function test.
26453 2017-02-28 Martin Liska <mliska@suse.cz>
26456 * read-rtl-function.c (function_reader::handle_unknown_directive):
26457 Bail out when one uses -flto.
26459 2017-02-28 Martin Liska <mliska@suse.cz>
26461 * common.opt: Replace space with tabular for options of <number>
26463 * config/i386/i386.opt: Show <number> value for
26464 -mlarge-data-threshold.
26465 * opts.c (print_filtered_help): Do not display number in hexadecimal
26468 2017-02-28 Martin Liska <mliska@suse.cz>
26470 * common.opt: Fix --help=option -Q for options which are of
26473 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
26475 * config/i386/i386.c (print_reg): Error out for values
26476 of 8-bit size in invalid integer register.
26478 2017-02-28 Martin Sebor <msebor@redhat.com>
26480 PR tree-optimization/79691
26481 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
26483 2017-02-28 Jakub Jelinek <jakub@redhat.com>
26486 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
26487 gcc_unreachable with output_operand_lossage.
26489 2017-02-28 Richard Biener <rguenther@suse.de>
26491 PR tree-optimization/79740
26492 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
26494 (visit_nary_op): Insert the nary into the hashtable if we
26495 pattern-matched sth.
26496 * tree-ssa-pre.c (eliminate_insert): Robustify.
26498 2017-02-28 Richard Biener <rguenther@suse.de>
26500 PR middle-end/79731
26501 * fold-const.c (decode_field_reference): Reject out-of-bound
26504 2017-02-28 Jakub Jelinek <jakub@redhat.com>
26506 * config/i386/i386.c: Include intl.h.
26507 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
26508 instead of just cond ? "..." : "...".
26509 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
26510 * coverage.c (read_counts_file): Likewise.
26511 * omp-offload.c: Include intl.h.
26512 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
26513 of just cond ? "..." : "...".
26514 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
26515 of just cond ? "..." : "...".
26517 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
26520 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
26522 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
26523 'tune for' CPU name.
26524 * config/arm/arm-cpu-data.h: Regenerated.
26526 2017-02-28 Richard Biener <rguenther@suse.de>
26528 PR tree-optimization/79732
26529 * tree-inline.c (expand_call_inline): Do not shadow var.
26531 2017-02-28 Richard Biener <rguenther@suse.de>
26533 PR tree-optimization/79723
26534 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
26535 address-space properly.
26537 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
26539 * doc/optinfo.texi (Optimization groups): Fix option used for
26541 * doc/invoke.texi (-fopt-info): Document "omp".
26542 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
26543 (OPTGROUP_ALL): Add OPTGROUP_OMP.
26544 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
26545 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
26546 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
26548 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
26550 * dumpfile.c (optgroup_options): Instead of "openmp", associate
26551 OPTGROUP_OMP with "omp".
26553 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
26556 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
26557 for arithmetic shift of unsigned V2DI.
26559 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
26561 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
26562 arc/linux.h headers.
26563 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
26564 (LINK_SPEC): Likewise.
26565 (ARC_TLS_EXTRA_START_SPEC): Likewise.
26566 (EXTRA_SPECS): Likewise.
26567 (STARTFILE_SPEC): Likewise.
26568 (ENDFILE_SPEC): Likewise.
26569 (LIB_SPEC): Likewise.
26570 (TARGET_SDATA_DEFAULT): Likewise.
26571 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
26572 (MULTILIB_DEFAULTS): Likewise.
26573 (DWARF2_UNWIND_INFO): Likewise.
26574 * config/arc/big.h: New file.
26575 * config/arc/elf.h: Likewise.
26576 * config/arc/linux.h: Likewise.
26577 * config/arc/t-uClibc: Remove.
26579 2017-02-27 Bin Cheng <bin.cheng@arm.com>
26581 PR tree-optimization/77536
26582 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
26583 (tree_transform_and_unroll_loop): Use above function to compute the
26584 estimated niter of unrolled loop and use it when scaling profile.
26585 Also use count info rather than frequency if it's non-zero.
26586 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
26587 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
26588 (vect_transform_loop): Call above function.
26590 2017-02-27 Richard Biener <rguenther@suse.de>
26592 PR tree-optimization/45397
26593 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
26594 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
26595 (visit_nary_op): Add pattern matching for CSEing sign-changed
26596 or truncated operations with wider ones.
26598 2017-02-27 Richard Biener <rguenther@suse.de>
26600 PR tree-optimization/79690
26601 * tree-vect-stmts.c (vectorizable_store): Use vector type
26602 built from the DR with address-space.
26604 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
26606 * doc/invoke.texi (Optimize Options): Refine the description
26607 of asan-use-after-return.
26609 2017-02-25 Alan Modra <amodra@gmail.com>
26611 PR rtl-optimization/79584
26612 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
26613 base, not ad->base_term, the reg within base. Remove assertion
26614 that ad->base == ad->base_term. Replace gen_int_mode using
26615 bogus mode with const0_rtx.
26617 2017-02-25 Jakub Jelinek <jakub@redhat.com>
26619 PR middle-end/79396
26620 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
26621 FMA_EXPR like tcc_binary or tcc_unary.
26623 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
26626 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
26628 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
26629 (output_loc_operands): Handle DW_OP_call_ref and
26630 DW_OP_GNU_variable_value.
26631 (struct variable_value_struct): New type.
26632 (struct variable_value_hasher): Likewise.
26633 (variable_value_hash): New variable.
26634 (string_types): Remove.
26635 (copy_loc_descr): New function.
26636 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
26637 (prepend_loc_descr_to_each): New function.
26638 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
26639 instead of add_loc_descr_to_each if the first argument is single
26640 location list and the second has multiple.
26641 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
26642 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
26643 when looking for variable value which doesn't have other location info.
26644 (loc_list_from_tree): Formatting fix.
26645 (gen_array_type_die): Simplify DW_AT_string_length handling.
26646 (adjust_string_types): Remove.
26647 (gen_subprogram_die): Don't call adjust_string_types nor test/set
26648 string_types. Call resolve_variable_values.
26649 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
26650 (resolve_addr_in_expr): Likewise. Add A argument.
26651 (copy_deref_exprloc): Remove deref argument. Adjust for the
26652 original expression being DW_OP_GNU_variable_value with optionally
26653 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
26654 optionally after it.
26655 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
26656 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
26657 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
26658 (variable_value_hasher::hash, variable_value_hasher::equal): New
26660 (resolve_variable_value_in_expr, resolve_variable_value,
26661 resolve_variable_values, note_variable_value_in_expr,
26662 note_variable_value): New functions.
26663 (dwarf2out_early_finish): Call note_variable_value on all toplevel
26666 2017-02-24 Jakub Jelinek <jakub@redhat.com>
26669 * opts.h (handle_generated_option): Add GENERATED_P argument.
26670 * opts-common.c (handle_option): Adjust function comment.
26671 (handle_generated_option): Add GENERATED_P argument, pass it to
26673 (control_warning_option): Pass false to handle_generated_option
26675 * opts.c (maybe_default_option): Pass true to handle_generated_option
26677 * optc-gen.awk: Likewise.
26679 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26681 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
26682 a REG, look at the REG it is a SUBREG of.
26683 (splitter for cmpeqsi_t): Ditto.
26685 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26687 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
26688 the special USEs with the pattern of the insn, not the insn itself.
26690 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
26693 * doc/invoke.texi: Document -mload-store-pairs.
26695 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26696 Sandra Loosemore <sandra@codesourcery.com>
26698 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
26699 argument isn't a CONST_INT.
26700 (nios2_alternate_compare_const): Assert op is a CONST_INT.
26701 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
26702 (nios2_validate_compare): Bypass alternate compare logic if *op2
26703 is not a CONST_INT.
26704 (ldstwm_operation_p): Return false if first_base is not a REG or
26705 if first_offset is not a CONST_INT.
26707 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26709 * config/cris/cris.md: Use correct operand in a define_peephole2.
26711 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26713 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
26715 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
26717 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
26718 this_insn if it is an INSN or JUMP_INSN.
26719 (force_offsettable): Look at base, not at addr.
26720 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
26721 on things that aren't necessarily CONST_INTs.
26723 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
26725 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
26726 -mfpmath=sse is the default also for x86-32 targets with SSE2
26727 instruction set when @option{-ffast-math} is enabled
26729 2017-02-24 Jeff Law <law@redhat.com>
26731 PR rtl-optimizatoin/79286
26732 * ira.c (update_equiv_regs): Drop may_trap_p exception to
26735 2017-02-24 Richard Biener <rguenther@suse.de>
26737 PR tree-optimization/79389
26738 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
26741 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
26743 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
26744 function comment to reflect reality.
26745 (loop_exits_before_overflow): Fix typo in function description.
26747 2017-02-24 Richard Biener <rguenther@suse.de>
26749 PR tree-optimization/79389
26750 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
26751 properly that a threading opportunity exists. Detect conditional
26752 copy/constant propagation opportunities.
26754 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
26756 * config/visium/visium.md (type): Add trap.
26757 (b): New mode attribute.
26758 (*btst): Rename into...
26759 (*btst<mode>): ...this and adjust.
26760 (*cbranchsi4_btst_insn): Rename into...
26761 (*cbranch<mode>4_btst_insn): ...this and adjust.
26762 (trap): New define_insn.
26764 2017-02-23 Jakub Jelinek <jakub@redhat.com>
26766 PR tree-optimization/79389
26767 * ifcvt.c (struct noce_if_info): Add rev_cond field.
26768 (noce_reversed_cond_code): New function.
26769 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
26770 reversed_comparison_code. Formatting fix.
26771 (noce_try_store_flag): Test rev_cond != NULL in addition to
26772 reversed_comparison_code.
26773 (noce_try_store_flag_constants): Likewise.
26774 (noce_try_store_flag_mask): Likewise.
26775 (noce_try_addcc): Use rev_cond if non-NULL instead of
26776 reversed_comparison_code.
26777 (noce_try_cmove_arith): Likewise. Formatting fixes.
26778 (noce_try_minmax, noce_try_abs): Clear rev_cond.
26779 (noce_find_if_block): Initialize rev_cond.
26780 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
26781 instead of false as last argument never attempt to reverse it
26784 2017-02-23 Bin Cheng <bin.cheng@arm.com>
26786 PR tree-optimization/79663
26787 * tree-predcom.c (combine_chains): Process refs in reverse order
26788 only for ZERO length chains, and add explaining comment.
26790 2017-02-23 Jeff Law <law@redhat.com>
26792 PR tree-optimization/79578
26793 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
26794 in call to operand_equal_p.
26796 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
26799 * config/i386/cpuid.h: Fix another undefined behavior.
26801 2017-02-23 Richard Biener <rguenther@suse.de>
26803 PR tree-optimization/79683
26804 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
26805 vector types for data-refs.
26807 2017-02-23 Martin Liska <mliska@suse.cz>
26809 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
26811 2017-02-23 Jakub Jelinek <jakub@redhat.com>
26813 PR middle-end/79665
26814 * internal-fn.c (get_range_pos_neg): Moved to ...
26815 * tree.c (get_range_pos_neg): ... here. No longer static.
26816 * tree.h (get_range_pos_neg): New prototype.
26817 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
26818 are known to be in between 0 and signed maximum inclusive, try to
26819 expand both unsigned and signed divmod and use the cheaper one from
26822 2017-02-22 Jeff Law <law@redhat.com>
26824 PR tree-optimization/79578
26825 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
26826 to compare base operands.
26828 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
26831 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
26832 gpc_reg_operand instead of fpr_reg_operand.
26834 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
26836 * config/mips/mips.c (mips_return_in_memory): Force FP
26837 vector types to be returned in memory for o32 ABI.
26839 2017-02-22 Jakub Jelinek <jakub@redhat.com>
26841 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
26842 instead of DW_TAG_member for static data member declarations and don't
26843 set no_linkage_name for static inline data members.
26844 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
26847 2017-02-22 Martin Liska <mliska@suse.cz>
26849 * doc/invoke.texi: Replace inequality signs with square brackets
26852 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26854 PR tree-optimization/68644
26855 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
26857 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
26860 * lra-constraints.c (simplify_operand_subreg): Handle
26861 WORD_REGISTER_OPERATIONS targets.
26863 2017-02-22 Jakub Jelinek <jakub@redhat.com>
26866 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
26867 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
26868 elimination by swapping fld*.
26870 2017-02-22 Richard Biener <rguenther@suse.de>
26872 PR tree-optimization/79673
26873 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
26874 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
26875 irrelevant address-space qualifiers and avoiding a
26876 ADDR_SPACE_CONVERT_EXPR from fold_convert.
26878 2017-02-22 Richard Biener <rguenther@suse.de>
26880 PR tree-optimization/79666
26881 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
26882 to not symbolically negate if that may introduce undefined
26885 2017-02-22 Martin Liska <mliska@suse.cz>
26888 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
26889 * data-streamer-out.c (streamer_write_gcov_count_stream):
26891 * value-prof.c (stream_out_histogram_value): Make assert more
26892 precise based on type of counter.
26894 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
26897 * config/i386/i386.md (standard_x87sse_constant_load splitter):
26898 Use nonimmediate_operand instead of memory_operand for operand 1.
26899 (float-extend standard_x87sse_constant_load splitter): Ditto.
26901 2017-02-21 Jeff Law <law@redhat.com>
26903 PR tree-optimization/79621
26904 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
26905 blocks with edges to themselves.
26907 2017-02-21 Jakub Jelinek <jakub@redhat.com>
26910 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
26911 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
26912 Use gimple_call_builtin_p.
26915 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
26916 on temporarily removed DEBUG_INSNs.
26918 PR tree-optimization/79649
26919 * tree-loop-distribution.c (classify_partition): Give up on
26920 non-generic address space loads/stores.
26922 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
26924 * doc/loop.texi (Loop manipulation): Remove nonexistent
26925 tree_ssa_loop_version from the documentation.
26926 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
26928 2017-02-21 Jakub Jelinek <jakub@redhat.com>
26931 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
26932 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
26933 * config/rs6000/rs6000.c: Include except.h.
26934 (rs6000_expand_split_stack_prologue): Call
26935 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
26937 2017-02-21 Martin Jambor <mjambor@suse.cz>
26940 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
26941 have been analyzed.
26943 2017-02-21 Martin Jambor <mjambor@suse.cz>
26945 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
26946 for backward compatibility only.
26947 * doc/invoke.texi (Option Summary): Remove all references to
26948 -fipa-cp-alignment.
26950 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
26954 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
26956 * lra-constraints.c (curr_insn_transform): Handle
26957 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
26959 2017-02-21 Martin Liska <mliska@suse.cz>
26961 * config/i386/i386.opt: Replace -masm-dialect with -masm.
26963 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
26965 PR translation/79638
26966 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
26968 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
26971 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
26972 (arm_function_ok_for_sibcall): Return false for an indirect call by
26973 descriptor if all the argument registers are used.
26974 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
26975 alignment of the function.
26977 2017-02-21 Jakub Jelinek <jakub@redhat.com>
26979 PR tree-optimization/61441
26980 * simplify-rtx.c (simplify_const_unary_operation): For
26981 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
26982 the sNaN unmodified.
26984 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
26986 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26987 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26988 instead of SYSTEM_HEADER_DIR.
26990 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
26991 Martin Liška <mliska@suse.cz>
26993 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
26994 Fix typos and grammar, use active voice, and clarify.
26996 2017-02-20 Marek Polacek <polacek@redhat.com>
26998 PR middle-end/79537
26999 * gimplify.c (gimplify_expr): Handle unused *&&L;.
27002 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
27004 2017-02-20 Jakub Jelinek <jakub@redhat.com>
27007 * config/i386/i386.c (ix86_expand_builtin): Handle
27008 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
27009 ix86_builtins_isa[fcode].isa as a requirement of those
27010 flags and any other flag in the bitmask.
27011 (ix86_init_mmx_sse_builtins): Use 0 instead of
27012 ~OPTION_MASK_ISA_64BIT as mask.
27013 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
27014 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
27015 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
27016 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
27018 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27021 * lra-constraints.c (split_reg): Check requested split mode
27022 is supported by the register.
27024 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27026 * lra-constraints.c (simplify_operand_subreg): Remove early
27029 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27032 * lra-constraints.c (curr_insn_transform): Tighten condition
27033 for converting SUBREG reloads from OP_OUT to OP_INOUT.
27035 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27038 * lra-constraints.c (curr_insn_transform): Handle
27039 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
27041 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
27044 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
27046 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
27048 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
27051 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
27054 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27056 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
27059 2017-02-18 Jakub Jelinek <jakub@redhat.com>
27061 * final.c (last_columnnum, override_columnnum): New variables.
27062 (final_start_function): Set last_columnnum, pass it to begin_prologue
27063 hook and pass 0 to dwarf2out_begin_prologue.
27064 (final_scan_insn): Update override_columnnum. Pass last_columnnum
27065 to source_line debug hook.
27066 (notice_source_line): Compute last_columnnum and for debug_column_info
27067 return true on column changes.
27068 * debug.h (struct gcc_debug_hooks): Add column argument to
27069 source_line and begin_prologue hooks.
27070 (debug_nothing_int_charstar_int_bool): Remove prototype.
27071 (debug_nothing_int_int_charstar,
27072 debug_nothing_int_int_charstar_int_bool): New prototypes.
27073 (dwarf2out_begin_prologue): Add column argument.
27074 * debug.c (do_nothing_debug_hooks): Adjust source_line and
27075 begin_prologue hooks.
27076 (debug_nothing_int_charstar_int_bool): Remove.
27077 (debug_nothing_int_int_charstar,
27078 debug_nothing_int_int_charstar_int_bool): New functions.
27079 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
27080 through to dwarf2out_source_line.
27081 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
27082 (dwarf2out_source_line): Add column argument, emit it if requested.
27083 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
27085 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
27086 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
27087 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
27088 through to dwarf2out_begin_prologue.
27089 (vmsdbgout_source_line): Add column argument, pass it through to
27090 dwarf2out_source_line.
27091 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
27092 dbxout_source_line caller.
27093 (dbxout_source_line): Add column argument.
27095 * common.opt (gno-column-info, gcolumn-info): New options.
27096 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
27097 (check_die): Also test for multiple DW_AT_decl_column attributes.
27098 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
27099 DW_AT_decl_column if requested.
27100 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
27102 (gen_variable_die): Likewise.
27103 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
27104 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
27107 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
27108 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
27109 (ix86_handle_option): Handle OPT_m3dnowa.
27110 * doc/invoke.texi (-m3dnowa): Document.
27111 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
27112 -m3dnowa instead of -m3dnow -march=athlon.
27115 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
27116 instead of gcc_assert for K, r and R code checks. Formatting fixes.
27118 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27121 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
27122 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
27123 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
27124 generator for vsx_xxpermdi_<mode>_be.
27125 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
27126 force big-endian semantics.
27127 (vsx_xxpermdi_<mode>_be): New define_expand with same
27128 implementation as previous version of vsx_xxpermdi_<mode>.
27130 2017-02-17 Jakub Jelinek <jakub@redhat.com>
27132 PR tree-optimization/79327
27133 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
27134 variable, its initialization and use.
27136 2017-02-17 Julia Koval <julia.koval@intel.com>
27138 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
27139 (OPTION_MASK_ISA_PKU_UNSET): New.
27140 (ix86_handle_option): Handle -mrdpid.
27141 * config/i386/cpuid.h (bit_RDPID): New.
27142 * config/i386/driver-i386.c (host_detect_local_cpu):
27143 Detect RDPID feature.
27144 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
27145 * config/i386/i386-c.c (ix86_target_macros_internal):
27147 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
27148 (ix86_valid_target_attribute_inner_p): Add "rdpid".
27149 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
27150 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
27151 * config/i386/i386.md (define_insn "rdpid"): New.
27152 * config/i386/i386.opt Add -mrdpid.
27153 * config/i386/immintrin.h (_rdpid_u32): New.
27155 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
27157 PR rtl-optimization/79541
27158 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
27159 instead of transforming it into USE.
27161 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
27163 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
27164 If HONOR_SNANS (SFmode) force the input to a register.
27165 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
27166 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
27167 an frsp or similar insn.
27169 2017-02-17 Martin Liska <mliska@suse.cz>
27171 PR rtl-optimization/79577
27172 * params.def (selsched-max-sched-times): Increase minimum to 1.
27174 2017-02-17 Martin Liska <mliska@suse.cz>
27176 PR rtl-optimization/79574
27177 * gcse.c (want_to_gcse_p): Prevent integer overflow.
27179 2017-02-17 Martin Liska <mliska@suse.cz>
27181 PR tree-optimization/79529
27182 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
27183 ssa_defined_default_def_p to handle cases which are implicitly
27185 * tree-ssa.c (ssa_defined_default_def_p): New function.
27186 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
27187 which are implicitly defined.
27188 * tree-ssa.h (ssa_defined_default_def_p): Declare.
27190 2017-02-17 Richard Biener <rguenther@suse.de>
27192 PR middle-end/79576
27193 * params.def (max-ssa-name-query-depth): Limit to 10.
27195 2017-02-17 Richard Biener <rguenther@suse.de>
27197 PR tree-optimization/79552
27198 * tree-ssa-structalias.c (visit_loadstore): Properly verify
27201 2017-02-17 Richard Biener <rguenther@suse.de>
27204 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
27206 2017-02-17 Marek Polacek <polacek@redhat.com>
27208 PR middle-end/79536
27209 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
27210 (fold_negate_expr): New wrapper.
27212 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
27214 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
27215 Correct terminology and de-emphasize pre-standard behavior.
27217 2017-02-16 Alan Modra <amodra@gmail.com>
27219 PR rtl-optimization/79286
27220 * ira.c (def_dominates_uses): New function.
27221 (update_equiv_regs): Don't create an equivalence for insns that
27222 may trap where the register def does not dominate the use.
27224 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
27226 PR rtl-optimization/78127
27227 * lra.c (lra): Call lra_eliminate before finish the loop after
27230 2017-02-16 Richard Biener <rguenther@suse.de>
27232 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
27234 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
27235 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
27236 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
27237 (isl_val_int_from_wi): New function.
27238 (extract_affine_gmp): Rename to ...
27239 (extract_affine_wi): ... this, take a widest_int.
27240 (extract_affine_int): Just wrap extract_affine_wi.
27241 (add_param_constraints): Use isl_val_int_from_wi.
27242 (add_loop_constraints): Likewise, and extract_affine_wi.
27244 2017-02-15 Jeff Law <law@redhat.com>
27246 PR middle-end/79521
27247 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
27248 ira_init_register_move_cost_if_necessary.
27250 2017-02-15 Martin Sebor <msebor@redhat.com>
27252 PR middle-end/32003
27253 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
27254 removed in a prior commit.
27256 2017-02-15 Bin Cheng <bin.cheng@arm.com>
27258 PR tree-optimization/79347
27259 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
27260 counters during peeling.
27262 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
27264 * Makefile.in (site.exp): Remove "set ISLVER".
27266 2017-02-15 Jakub Jelinek <jakub@redhat.com>
27269 * real.c (real_from_integer): Call real_convert even for decimal.
27271 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
27274 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
27276 2017-02-14 Andrew Pinski <apinski@cavium.com>
27278 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
27279 cores and change the partno/implementer to be correct.
27280 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
27281 the 'B" as the implementer.
27282 * config/aarch64/aarch64-tune.md: Regenerate.
27284 2017-02-14 Carl Love <cel@us.ibm.com>
27286 * config/rs6000/rs6000.c: Add case statement entry to make the
27287 xvcvuxdsp built-in argument unsigned.
27288 * config/rs6000/vsx.md: Fix the source and return operand types so they
27289 match the instruction definitions from the ISA document. Fix typo
27290 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
27293 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
27296 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
27297 member early_clobber_alts.
27298 * lra-lives.c (reg_early_clobber_p): New.
27299 (process_bb_lives): Use it.
27300 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
27301 (debug_operand_data): Initialize early_clobber_alts.
27302 (setup_operand_alternative): Set up early_clobber_alts.
27303 (collect_non_operand_hard_regs): Ditto. Pass early clobber
27304 alternatives to new_insn_reg.
27305 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
27307 (lra_update_insn_regno_info): Pass the new arg.
27309 2017-02-14 Jakub Jelinek <jakub@redhat.com>
27311 PR middle-end/79505
27312 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
27313 (new_oacc_loop_raw): Don't clear already cleared fields.
27316 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
27317 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
27318 _mm512_prefetch_i64gather_ps): New inline functions and macros.
27320 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
27323 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
27325 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
27328 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
27329 the extra instruction to the right place to store 128-bit constant
27332 2017-02-14 Martin Sebor <msebor@redhat.com>
27334 PR middle-end/79448
27335 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
27336 warning for strings of unknown length.
27338 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
27340 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
27342 2017-02-14 Jeff Law <law@redhat.com>
27345 * ira-costs.c (scan_one_insn): Initialize register move costs
27346 for pseudos seen in USE/CLOBBER insns.
27348 PR tree-optimization/79095
27349 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
27350 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
27351 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
27352 if the operands are known to be not equal, then the resulting range
27354 (intersect_ranges): If the new range is ~[0,0] and the old range is
27355 wide, then prefer ~[0,0].
27356 * tree-vrp.c (overflow_comparison_p_1): New function.
27357 (overflow_comparison_p): New function.
27358 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
27359 if NAME is used in an overflow test.
27360 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
27361 overflow check that can be expressed as an equality test, then adjust
27362 ops to be that equality test.
27364 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27366 * config/s390/s390-builtin-types.def: Remove flags argument.
27367 * config/s390/s390.c (s390_init_builtins): Likewise.
27369 2017-02-14 Martin Liska <mliska@suse.cz>
27371 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
27372 vector. Fix trailing white spaces.
27374 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
27376 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
27379 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27381 PR rtl-optimization/68664
27382 * config/arm/arm.c (arm_sched_can_speculate_insn):
27383 New function. Declare prototype.
27384 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
27386 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27388 PR rtl-optimization/68664
27389 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
27391 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
27393 2017-02-14 Amit Pawar <amit.pawar@amd.com>
27395 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
27396 max skip bytes for function, loop and jump.
27398 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27400 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
27401 ABS_EXPR for gimple dump.
27403 2017-02-14 Jakub Jelinek <jakub@redhat.com>
27406 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
27408 PR tree-optimization/79408
27409 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
27410 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
27411 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
27412 also if rhs1 is INTEGER_CST.
27414 2017-02-14 Richard Biener <rguenther@suse.de>
27416 PR middle-end/79432
27417 * tree-into-ssa.c (insert_phi_nodes): When the function can
27418 have abnormal edges rewrite SSA names with broken use-def
27419 dominance out of SSA and register them for PHI insertion.
27421 2017-02-13 Martin Sebor <msebor@redhat.com>
27423 PR middle-end/79496
27424 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
27425 clearing info.nowrite flag when snprintf size argument is a range.
27427 2017-02-13 Jakub Jelinek <jakub@redhat.com>
27429 * cprop.c (cprop_jump): Add missing space in string literal.
27430 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
27431 (get_constraint_for_component_ref): Likewise.
27432 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
27433 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
27434 * lra-constraints.c (process_alt_operands): Likewise.
27435 * ipa-inline.c (inline_small_functions): Likewise.
27436 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
27437 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
27438 * trans-mem.c (diagnose_tm_1_op): Likewise.
27439 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
27440 (grid_parallel_clauses_gridifiable): Likewise.
27442 * config/nvptx/mkoffload.c (process): Add space in between
27445 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
27446 "MOD4_SSE_REGS" and "ALL_REGS".
27448 * spellcheck.c (test_data): Add , in between "foo" and "food".
27450 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27453 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
27454 boundary crossing check and subsequent code generation agree.
27456 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27458 * config/aarch64/aarch64.c (has_memory_op): Delete.
27459 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
27462 2017-02-13 Jakub Jelinek <jakub@redhat.com>
27464 PR rtl-optimization/79388
27465 PR rtl-optimization/79450
27466 * combine.c (distribute_notes): When removing TEM_INSN for which
27467 corresponding dest has last value recorded, invalidate that last
27470 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27472 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
27473 of explicit '@'. Add missing assembly comment marker on branch costs
27476 2017-02-13 Nathan Sidwell <nathan@acm.org>
27478 * gengtype-lex.l (<in_struct>): Add '/'.
27480 2017-02-13 Martin Liska <mliska@suse.cz>
27483 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
27485 2017-02-13 Richard Biener <rguenther@suse.de>
27487 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
27489 * configure: Re-generate.
27490 * config.in: Likewise.
27491 * graphite-dependences.c: Simplify as if
27492 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
27493 * graphite-isl-ast-to-gimple.c: Likewise.
27494 * graphite-optimize-isl.c: Likewise.
27495 * graphite-poly.c: Likewise.
27496 * graphite-sese-to-poly.c: Likewise.
27497 * graphite.h: Likewise.
27498 * toplev.c: Include isl/version.h and use isl_version () for
27499 printing the ISL version.
27500 * doc/install.texi: Update ISL requirement.
27502 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
27504 * doc/standards.texi (Standards): Update reference to
27507 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
27509 * doc/extend.texi (Named Address Spaces): sourceware.org now
27511 * doc/install.texi (Binaries): Ditto.
27514 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
27516 * doc/cpp.texi: Replace "stringify"/"stringification" with C
27517 standard terminology "stringize"/"stringizing" throughout.
27518 * doc/cppinternals.texi: Likewise.
27520 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
27522 * doc/extend.texi: Fix some spelling mistakes and typos.
27523 * doc/invoke.texi: Likewise.
27525 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
27528 * params.def (inline-min-speedup) Change from 10 to 8.
27530 2017-02-11 Jakub Jelinek <jakub@redhat.com>
27532 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
27535 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
27538 * ipa-inline-analysis.c (get_minimal_bb): New function.
27539 (record_modified): Use it.
27540 (remap_edge_change_prob): Handle also ancestor functions.
27542 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
27544 * doc/contrib.texi (Contributors): Remove broken link into
27545 the Mauve CVS repository.
27547 2017-02-11 Jakub Jelinek <jakub@redhat.com>
27549 PR middle-end/79454
27550 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
27551 result computation whenever lhs doesn't have vector mode, not
27552 just when it has BLKmode.
27554 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
27556 * doc/makefile.texi (profiledbootstrap): Refer to the
27557 installation instructions only in textual form.
27559 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27562 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
27564 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
27566 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
27567 (Specific): Update mingw-w64 reference.
27569 (Specific): Remove broken link to Renesas RX processor.
27571 2017-02-10 Richard Biener <rguenther@suse.de>
27573 * toplev.c (process_options): Do not mention obsolete graphite
27574 options when printing sorry message about missing graphite support.
27575 Mention -floop-nest-optimize.
27577 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
27579 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
27580 (vtst_p16): Likewise.
27581 (vtstq_p8): Likewise.
27582 (vtstq_p16): Likewise.
27584 (vtstq_p64): Likewise.
27585 * config/arm/arm_neon.h (vgetq_lane_p64): New.
27586 (vset_lane_p64): New.
27587 (vsetq_lane_p64): New.
27589 2017-02-10 Jakub Jelinek <jakub@redhat.com>
27591 PR tree-optimization/79411
27592 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
27593 stmt operands are SSA_NAMEs used in abnormal phis.
27594 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
27597 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
27600 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
27603 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
27605 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
27607 2017-02-09 Jakub Jelinek <jakub@redhat.com>
27609 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
27613 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
27614 not arbitrary TREE_CONSTANT.
27617 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
27618 "omp declare target link" attribute unless is_global_var.
27619 * omp-offload.c (find_link_var_op): Likewise.
27621 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
27622 Chung-Lin Tang <cltang@codesourcery.com>
27624 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
27626 (gimplify_adjust_omp_clauses): Don't delete TILE.
27627 (gimplify_omp_for): Deal with TILE.
27628 * internal-fn.c (expand_GOACC_TILE): New function.
27629 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
27631 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
27632 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
27634 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
27635 avoid DIV for outermost collapse var.
27636 (expand_oacc_for): Insert tile element loop as needed. Adjust.
27637 Remove out of date comments, fix whitespace.
27638 * omp-general.c (omp_extract_for_data): Deal with tiling.
27639 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
27640 adjust OLF_DIM_BASE value.
27641 (struct omp_for_data): Add tiling field.
27642 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
27643 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
27644 for auto loops. Remove default auto determining, moved to
27645 oacc_loop_fixed_partitions.
27646 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
27647 stmts, add e_mask field.
27648 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
27649 (oacc_thread_numbers): Use oacc_dim_call.
27650 (oacc_xform_tile): New.
27651 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
27652 (finish_oacc_loop): Adjust for ifns vector.
27653 (oacc_loop_discover_walk): Append loop abstraction sites to list,
27654 add case for GOACC_TILE fns.
27655 (oacc_loop_xform_loop): Delete.
27656 (oacc_loop_process): Iterate over call list directly, and add
27657 handling for GOACC_TILE fns.
27658 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
27660 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
27661 vector partitioning to outer loops. Assign 2 partitions to loops
27662 when available. Add TILE handling.
27663 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
27664 (execite_oacc_device_lower): Process GOACC_TILE fns,
27665 ignore unknown specs.
27666 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
27667 * tree.c (omp_clause_num_ops): Adjust TILE ops.
27668 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
27670 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
27672 * configure.ac (ACX_BUGURL): Update.
27673 * configure: Regenerate.
27675 2017-02-09 Richard Biener <rguenther@suse.de>
27677 PR tree-optimization/69823
27678 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
27679 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
27681 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
27683 * config/arc/arc-c.def: Add __NPS400__ definition.
27684 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
27685 (TARGET_NPS400): Define.
27687 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
27689 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
27691 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
27692 pointer, arch_info.
27693 (arc_cpu_types): Fill the arch_info field with a pointer into the
27694 arc_arch_types table.
27695 (arc_selected_cpu): Declare.
27696 * config/arc/arc.c (arc_selected_cpu): Make global.
27697 (arc_selected_arch): Delete.
27698 (arc_base_cpu): Delete.
27699 (arc_override_options): Remove references to deleted variables,
27700 update access to arch information.
27701 (ARC_OPT): Update access to arch information.
27702 (ARC_OPTX): Likewise.
27703 * config/arc/arc.h (arc_base_cpu): Remove declaration.
27704 (TARGET_ARC600): Update access to arch information.
27705 (TARGET_ARC601): Likewise.
27706 (TARGET_ARC700): Likewise.
27707 (TARGET_EM): Likewise.
27708 (TARGET_HS): Likewise.
27709 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
27712 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
27715 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
27716 condition/operands for integer GE/LE/GEU/LEU operations.
27718 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
27720 PR translation/79397
27721 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
27724 2017-02-08 Martin Jambor <mjambor@suse.cz>
27727 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
27728 whether allocation happened.
27729 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
27730 nothing was allocated.
27732 2017-02-08 Jakub Jelinek <jakub@redhat.com>
27734 PR tree-optimization/79408
27735 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
27736 constant, but SSA_NAME with a known integer range, use the minimum
27737 of that range instead of op1 to determine if modulo can be replaced
27738 with its first operand.
27740 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27742 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
27744 2017-02-08 Richard Biener <rguenther@suse.de>
27746 PR tree-optimization/71824
27747 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
27748 Check all loops contained in the merged region.
27750 2017-02-07 Andrew Pinski <apinski@cavium.com>
27752 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
27754 2017-02-07 Andrew Pinski <apinski@cavium.com>
27756 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
27757 (thunderxt88): Likewise.
27758 (thunderxt81): Disable LSE and change v8.1 to v8.
27759 (thunderxt83): Likewise.
27761 2017-02-07 Jakub Jelinek <jakub@redhat.com>
27762 Richard Biener <rguenther@suse.de>
27764 PR middle-end/79399
27765 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
27766 type from int to size_t.
27767 * ira-costs.c (struct_costs_size): Change type from int to size_t.
27769 2017-02-07 Jakub Jelinek <jakub@redhat.com>
27771 PR rtl-optimization/79386
27772 * cprop.c (bypass_conditional_jumps): Initialize
27773 bypass_last_basic_block already before splitting bbs after
27774 unconditional traps...
27775 (bypass_conditional_jumps): ... rather than here.
27778 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
27779 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
27780 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
27781 fix -masm=intel patterns.
27783 2017-02-07 Richard Biener <rguenther@suse.de>
27785 PR tree-optimization/79256
27786 PR middle-end/79278
27787 * builtins.c (get_object_alignment_2): Use min_align_of_type
27788 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
27789 and ADJUST_FIELD_ALIGN.
27791 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
27793 * doc/tm.texi: Regenerate.
27794 * stor-layout.c (layout_decl): Adjust.
27795 (update_alignment_for_field): Likewise.
27796 (place_field): Likewise.
27797 (min_align_of_type): Likewise.
27798 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
27799 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
27800 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
27801 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
27802 * config/frv/frv.c (frv_adjust_field_align): Likewise.
27803 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
27804 * config/i386/i386.c (x86_field_alignment): Likewise.
27805 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
27806 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
27807 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
27808 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
27809 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
27810 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
27814 2017-01-30 Richard Biener <rguenther@suse.de>
27816 PR tree-optimization/79256
27817 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
27818 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
27821 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
27823 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
27824 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
27825 builtins to SImode and emit a zero-extend, if necessary.
27827 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
27829 * docs/invoke.texi (RISC-V Options): Alphabetize.
27831 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
27833 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
27836 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
27838 * config/riscv/riscv.c: New file.
27839 * common/config/riscv/riscv-common.c: Likewise.
27840 * config.gcc: Likewise.
27841 * config/riscv/constraints.md: Likewise.
27842 * config/riscv/elf.h: Likewise.
27843 * config/riscv/generic.md: Likewise.
27844 * config/riscv/linux.h: Likewise.
27845 * config/riscv/multilib-generator: Likewise.
27846 * config/riscv/peephole.md: Likewise.
27847 * config/riscv/pic.md: Likewise.
27848 * config/riscv/predicates.md: Likewise.
27849 * config/riscv/riscv-builtins.c: Likewise.
27850 * config/riscv/riscv-c.c: Likewise.
27851 * config/riscv/riscv-ftypes.def: Likewise.
27852 * config/riscv/riscv-modes.def: Likewise.
27853 * config/riscv/riscv-opts.h: Likewise.
27854 * config/riscv/riscv-protos.h: Likewise.
27855 * config/riscv/riscv.h: Likewise.
27856 * config/riscv/riscv.md: Likewise.
27857 * config/riscv/riscv.opt: Likewise.
27858 * config/riscv/sync.md: Likewise.
27859 * config/riscv/t-elf-multilib: Likewise.
27860 * config/riscv/t-linux: Likewise.
27861 * config/riscv/t-linux-multilib: Likewise.
27862 * config/riscv/t-riscv: Likewise.
27863 * configure.ac: Likewise.
27864 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
27865 Waterman as RISC-V maintainers.
27866 * doc/install.texi: Add RISC-V entries.
27867 * doc/invoke.texi: Add RISC-V options section.
27868 * doc/md.texi: Add RISC-V constraints section.
27869 * configure: Regenerated.
27871 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
27874 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
27875 false values to be constant vectors with all 0 or all 1 bits set.
27876 (vcondu<mode><mode>): Likewise.
27877 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
27879 (fpmask_comparison_operator): Update comment.
27880 (vecint_comparison_operator): New predicate.
27881 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
27882 vector conditionals when the true and false values are constant
27883 vectors with all 0 bits or all 1 bits set.
27885 2017-02-06 Martin Sebor <msebor@redhat.com>
27887 PR tree-optimization/79376
27888 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
27890 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
27892 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
27893 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
27894 to simplify split condition.
27896 2017-02-06 Jakub Jelinek <jakub@redhat.com>
27898 * omp-expand.c (oxpand_omp_atomic_fetch_op,
27899 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
27902 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
27904 PR rtl-optimization/68664
27905 * target.def (can_speculate_insn): New hook.
27906 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
27907 * doc/tm.texi: Regenerate.
27908 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
27909 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
27910 (rs6000_sched_can_speculate_insn): New function.
27912 2017-02-06 Jakub Jelinek <jakub@redhat.com>
27914 PR tree-optimization/79284
27915 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
27916 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
27917 vectorizable_mask_load_store, vectorizable_operation,
27918 vect_is_simple_cond, get_same_sized_vectype): Use it instead
27919 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
27920 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
27921 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
27922 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
27923 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
27924 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
27925 is_gimple_assign (stmt). Replace another such test with
27926 is_gimple_assign (stmt).
27928 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
27931 * config/avr/avr.c (rtl-iter.h): Include it.
27932 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
27933 (avr_legitimate_combined_insn): ...and implementation.
27935 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
27937 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
27938 * config/s390/s390.c (s390_const_operand_ok)
27939 (s390_canonicalize_comparison, s390_extract_part)
27940 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
27941 (s390_contiguous_bitmask_p, s390_rtx_costs)
27942 (legitimize_pic_address): Likewise.
27943 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
27944 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
27945 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
27946 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
27947 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
27949 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
27951 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
27952 REGNO($0) == REGNO($1).
27954 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27956 * config/s390/linux.h(SIZE_TYPE): Add comment.
27958 2017-02-06 Julian Brown <julian@codesourcery.com>
27959 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27960 Virendra Pathak <virendra.pathak@broadcom.com>
27962 * config/aarch64/aarch64-cores.def: Change the scheduler
27964 * config/aarch64/aarch64.md: Include thunderx2t99.md.
27965 * config/aarch64/thunderx2t99.md: New file.
27967 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
27969 * doc/standards.texi (Go Language): Update link to language
27972 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
27974 * tree-eh.c (lower_resx): Sanitize profile.
27975 (cleanup_empty_eh_move_lp): Likewise.
27977 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
27980 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
27982 * cfgloopmanip.h (loop_version): Update prototype.
27983 * modulo-sched.c (sms_schedule): Update call of loop_version.
27984 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
27985 * tree-parloops.c (gen_parallel_loop): Likewise.
27986 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
27987 * tree-ssa-loop-split.c (split_loop): Likewise.
27988 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
27989 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
27991 2017-02-05 Martin Liska <mliska@suse.cz>
27994 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
27996 (print_operand_address): Initialize a struct to zero.
27998 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
28000 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
28001 garbage collector only in textual form.
28003 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
28005 * doc/extend.texi (x86 specific memory model extensions for
28006 transactional memory): Simplify a phrase.
28008 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
28011 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
28012 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
28013 (atomic_storedi_1): Likewise.
28015 2017-02-04 Jakub Jelinek <jakub@redhat.com>
28017 PR tree-optimization/79338
28018 * tree-parloops.c (gather_scalar_reductions): Don't call
28019 vect_analyze_loop_form for loop->inner before destroying loop's
28022 2017-02-03 Martin Sebor <msebor@redhat.com>
28024 PR tree-optimization/79327
28025 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
28026 when precision has resulted in leading zeros.
28027 (format_integer): Adjust the likely counter to assume an unknown
28028 argument that may be zero is non-zero.
28030 2017-02-03 Jason Merrill <jason@redhat.com>
28033 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
28034 avoid copying non-taken branch.
28036 2017-02-03 Jakub Jelinek <jakub@redhat.com>
28038 PR tree-optimization/79340
28039 * tree-vect-loop.c (vectorizable_reduction): Release
28040 vec_defs elements after safe_splicing them into other vectors.
28043 PR tree-optimization/79327
28044 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
28045 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
28047 (format_integer): Use wide_int_to_tree instead of build_int_cst
28048 + to_?hwi. If argmin is NULL, just set argmin and argmax to
28049 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
28050 of shortest and longest sequence.
28052 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
28054 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
28055 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
28057 2017-02-03 Walter Lee <walt@tilera.com>
28060 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
28061 after initial stackframe link reg save.
28062 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
28064 2017-02-03 Jakub Jelinek <jakub@redhat.com>
28067 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
28068 wu for stxssp alternative.
28070 2017-02-03 Martin Sebor <msebor@redhat.com>
28072 PR tree-optimization/79352
28073 * gimple-fold.c (get_range_strlen): Add argument.
28074 (get_range_strlen): Change return type to bool.
28075 (get_maxval_strlen): Pass in a dummy argument.
28076 * gimple-fold.h (get_range_strlen): Change return type to bool.
28077 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
28078 * tree.h (array_at_struct_end_p): Add argument.
28079 * tree.c (array_at_struct_end_p): Handle it.
28081 2017-02-03 Martin Liska <mliska@suse.cz>
28084 * multiple_target.c (create_dispatcher_calls): Redirect edge
28085 from a caller of a dispatcher.
28086 (expand_target_clones): Make the clones local.
28087 (ipa_target_clone): Do both target clones and resolvers.
28088 (ipa_dispatcher_calls): Remove the pass.
28089 (pass_dispatcher_calls::gate): Likewise.
28090 (make_pass_dispatcher_calls): Likewise.
28091 * passes.def (pass_target_clone): Put as very first IPA early
28094 2017-02-03 Martin Liska <mliska@suse.cz>
28096 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
28097 in case of a function with ifunc attribute.
28099 2017-02-03 Martin Liska <mliska@suse.cz>
28101 * cgraph.c (cgraph_node::dump): Dump function version info.
28102 * symtab.c (symtab_node::dump_base): Add missing new line.
28104 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
28106 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
28107 (ifcombine_ifandif): Use it.
28109 2017-02-03 Martin Liska <mliska@suse.cz>
28111 * doc/invoke.texi: Document default value for
28112 use-after-scope-direct-emission-threshold.
28114 2017-02-03 Martin Liska <mliska@suse.cz>
28116 PR tree-optimization/79339
28117 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
28118 (format_floating): Likewise.
28120 2017-02-03 Martin Liska <mliska@suse.cz>
28123 * ipa-prop.c (ipa_node_params_t::insert): Remove current
28125 (ipa_node_params_t::remove): Likewise.
28126 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
28127 initialization from removed ipa_node_params_t::insert.
28128 (ipa_node_params::~ipa_node_params): Move from removed
28129 ipa_node_params_t::release.
28130 * symbol-summary.h (symbol_summary::m_released): New member.
28131 Do not release a summary twice. Do not allow to call finalizer
28132 for types of a summary that live in GGC memory.
28134 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28136 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
28139 2017-02-02 Martin Sebor <msebor@redhat.com>
28141 PR middle-end/79275
28142 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
28143 (format_string): Tighten up the range of output for non-constant
28144 strings and correct the expected range for wide non-constant strings.
28146 2017-02-02 Martin Sebor <msebor@redhat.com>
28148 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
28150 PR middle-end/32003
28151 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
28153 (-fdump-tree-@var): Add to index and document how to come up
28154 with pass-specific option and dump file names.
28155 (-fdump-passes): Clarify where to look for output.
28157 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
28159 PR middle-end/77445
28160 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
28161 statistics of the analyzed path; allow threading for speed when
28162 any of BBs along the path are optimized for speed.
28164 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
28166 PR middle-end/78468
28167 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
28168 settings of the virtual registers.
28171 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
28173 * explow.c (get_dynamic_stack_size): Take known alignment of stack
28174 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
28177 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28179 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
28180 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
28182 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28184 * config/s390/s390.md: Add missing comments with the expanded
28186 * config/s390/vector.md: Likewise.
28187 * config/s390/vx-builtins.md: Likewise.
28189 2017-02-02 Jakub Jelinek <jakub@redhat.com>
28192 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
28193 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
28194 conditions on a single line.
28196 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28198 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
28199 __S390_VX__ to __VX__.
28201 2017-02-01 Andrew Pinski <apinski@cavium.com>
28203 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
28204 stmt_info to record_stmt_cost.
28205 (vect_get_known_peeling_cost): Pass stmt_info if known to
28207 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
28208 cpu_vector_cost field into
28209 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
28210 field into vec_int_stmt_cost and vec_fp_stmt_cost.
28211 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
28212 splitting of scalar_stmt_cost and vec_stmt_cost.
28213 (thunderx_vector_cost): Likewise.
28214 (cortexa57_vector_cost): LIkewise.
28215 (exynosm1_vector_cost): Likewise.
28216 (xgene1_vector_cost): Likewise.
28217 (thunderx2t99_vector_cost): Improve after the splitting of the two
28219 (aarch64_builtin_vectorization_cost): Update for the splitting of
28220 scalar_stmt_cost and vec_stmt_cost.
28222 2017-02-01 Torvald Riegel <triegel@redhat.com>
28223 Richard Henderson <rth@redhat.com>
28225 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
28226 conditional on existance of a fast atomic load.
28227 * optabs-query.c (can_atomic_load_p): New function.
28228 * optabs-query.h (can_atomic_load_p): Declare it.
28229 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
28230 no fast atomic load is available for the particular size of access.
28231 (expand_atomic_compare_and_swap): Likewise.
28232 (expand_atomic_load): Likewise.
28233 (expand_atomic_store): Likewise.
28234 (expand_atomic_fetch_op): Likewise.
28235 * testsuite/lib/target-supports.exp
28236 (check_effective_target_sync_int_128): Remove x86 because it provides
28237 no fast atomic load.
28238 (check_effective_target_sync_int_128_runtime): Likewise.
28240 2017-02-01 Richard Biener <rguenther@suse.de>
28242 * graphite.c: Include tree-vectorizer.h for find_loop_location.
28243 (graphite_transform_loops): Provide opt-info for optimized nests.
28244 * tree-parloop.c (parallelize_loops): Provide opt-info for
28245 parallelized loops.
28247 2017-02-01 Richard Biener <rguenther@suse.de>
28249 PR middle-end/79315
28250 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
28251 was not set before.
28253 2017-02-01 Richard Biener <rguenther@suse.de>
28255 PR tree-optimization/71824
28256 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
28257 Verify the loops are valid in the merged SESE region.
28258 (scop_detection::can_represent_loop_1): Check analyzing the
28259 evolution of the number of iterations in the region succeeds.
28261 2017-01-31 Ian Lance Taylor <iant@golang.org>
28263 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
28264 REG_ARGS_SIZE note to 32-bit push insns and call insn.
28266 2017-01-31 David Malcolm <dmalcolm@redhat.com>
28268 PR preprocessor/79210
28269 * input.c (get_substring_ranges_for_loc): Replace line_width
28270 assertion with error-handling.
28272 2017-01-31 Richard Biener <rguenther@suse.de>
28274 PR tree-optimization/77318
28275 * graphite-sese-to-poly.c (extract_affine): Fix assert.
28276 (create_pw_aff_from_tree): Take loop parameter.
28277 (add_condition_to_pbb): Pass loop of the condition to
28278 create_pw_aff_from_tree.
28280 2017-01-31 Jakub Jelinek <jakub@redhat.com>
28282 * config/s390/s390.c (s390_asan_shadow_offset): New function.
28283 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
28285 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
28289 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
28291 (convert_int_to_float128): Likewise.
28292 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
28293 (convert_int_to_float128): Likewise.
28294 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
28295 (UNSPEC_IEEE128_CONVERT): Likewise.
28296 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
28297 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
28298 Use local variables for IBM extended format.
28299 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
28300 (fix_trunc<mode>si2_fprs): Likewise.
28301 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
28302 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
28303 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
28304 to know that we can now have integers of all sizes in vector
28306 (fix<uns>_<mode>di2_hw): Likewise.
28307 (float<uns>_<mode>si2_hw): Likewise.
28308 (fix_<mode>si2_hw): Likewise.
28309 (fixuns_<mode>si2_hw): Likewise.
28310 (float<uns>_<mode>di2_hw): Likewise.
28311 (float_<mode>di2_hw): Likewise.
28312 (float_<mode>si2_hw): Likewise.
28313 (floatuns_<mode>di2_hw): Likewise.
28314 (floatuns_<mode>si2_hw): Likewise.
28315 (xscvqp<su>wz_<mode>): Delete, no longer used.
28316 (xscvqp<su>dz_<mode>): Likewise.
28317 (xscv<su>dqp_<mode>): Likewise.
28318 (ieee128_mfvsrd_64bit): Likewise.
28319 (ieee128_mfvsrd_32bit): Likewise.
28320 (ieee128_mfvsrwz): Likewise.
28321 (ieee128_mtvsrw): Likewise.
28322 (ieee128_mtvsrd_64bit): Likewise.
28323 (ieee128_mtvsrd_32bit): Likewise.
28325 2017-01-31 Martin Liska <mliska@suse.cz>
28328 * ipa-prop.c (ipa_free_all_node_params): Call release method
28329 instead of ~sumbol_summary to not to trigger double times
28332 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
28334 PR tree-optimization/71691
28335 * bitmap.h (class auto_bitmap): New.
28336 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
28337 is_maybe_undefined instead of ssa_undefined_value_p.
28339 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28341 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
28342 __S390_ARCH_LEVEL__ to __ARCH__.
28344 2017-01-31 Jakub Jelinek <jakub@redhat.com>
28346 PR tree-optimization/79267
28347 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
28348 if should_remove_lhs_p is true.
28350 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
28353 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
28354 (add_alignment_attribute): New.
28355 (base_type_die): Add alignment attribute.
28356 (subrange_type_die): Likewise.
28357 (modified_type_die): Likewise.
28358 (gen_array_type_die): Likewise.
28359 (gen_descr_array_type_die: Likewise.
28360 (gen_enumeration_type_die): Likewise.
28361 (gen_subprogram_die): Likewise.
28362 (gen_variable_die): Likewise.
28363 (gen_field_die): Likewise.
28364 (gen_ptr_to_mbr_type_die): Likewise.
28365 (gen_struct_or_union_type_die): Likewise.
28366 (gen_subroutine_type_die): Likewise.
28367 (gen_typedef_die): Likewise.
28368 (base_type_cmp): Compare alignment attribute.
28370 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28373 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
28374 (setb_unsigned) New pattern for setb with CCUNS.
28375 * config/rs6000/rs6000.c (expand_block_compare): Use a different
28376 subfc./subfe sequence to avoid overflow problems. Generate a
28377 shorter sequence with cmpld/setb for power9.
28378 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
28379 for generating subfc. instruction.
28380 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
28381 now uses this instruction.
28383 2017-01-30 Ian Lance Taylor <iant@google.com>
28386 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
28387 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
28389 2017-01-30 Martin Sebor <msebor@redhat.com>
28391 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
28392 Move constant to the right of a relational operator.
28393 (get_mpfr_format_length, format_character, format_string): Ditto.
28394 (should_warn_p, maybe_warn): Same.
28396 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
28398 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
28401 * asan.c (get_translation_unit_decl): Remove function.
28402 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
28404 2017-01-30 Martin Liska <mliska@suse.cz>
28406 PR gcov-profile/79259
28407 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
28408 -fprofile-generate.
28410 2017-01-30 Martin Liska <mliska@suse.cz>
28413 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
28414 Initialize variables with NULL value.
28416 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
28419 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
28421 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
28423 2017-01-30 Richard Biener <rguenther@suse.de>
28425 PR tree-optimization/79276
28426 * tree-vrp.c (process_assert_insertions): Properly adjust common
28427 when removing a duplicate.
28429 * gcc.dg/torture/pr79276.c: New testcase.
28431 2017-01-30 Richard Biener <rguenther@suse.de>
28433 PR tree-optimization/79256
28434 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
28435 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
28437 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
28439 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
28442 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
28443 ("*r<noxa>sbg_<mode>_sll_bitmask")
28444 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
28445 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
28446 Use contiguous_bitmask_nowrap_operand.
28448 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28451 * config/rs6000/altivec.h (vec_xl): Revise #define.
28452 (vec_xst): Likewise.
28454 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
28456 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
28458 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
28460 PR rtl-optimization/79194
28461 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
28462 traps before call to bypass_conditional_jumps.
28464 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
28466 PR tree-optimization/71374
28467 * lra-constraints.c (check_conflict_input_operands): New.
28468 (match_reload): Use it.
28470 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
28473 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
28474 account to calculate conflict_set.
28476 2017-01-27 Bin Cheng <bin.cheng@arm.com>
28478 PR rtl-optimization/78559
28479 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
28480 other_insn in combine.
28482 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
28484 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
28485 uint16_type_node for BT_UINT16.
28487 2017-01-27 David Malcolm <dmalcolm@redhat.com>
28489 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
28490 "RTL Tests" to menu.
28491 (GIMPLE Tests): New node.
28492 (RTL Tests): New node.
28494 2017-01-27 Richard Biener <rguenther@suse.de>
28496 PR tree-optimization/79245
28497 * tree-loop-distribution.c (distribute_loop): Apply cost
28498 modeling also to detected patterns.
28500 2017-01-27 Richard Biener <rguenther@suse.de>
28502 PR tree-optimization/71433
28503 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
28504 (compare_assert_loc): New function.
28505 (process_assert_insertions): Sort and optimize assert locations
28506 to remove duplicates and push down identical assertions on
28507 edges to their destination block.
28509 2017-01-27 Richard Biener <rguenther@suse.de>
28511 PR tree-optimization/79244
28512 * tree-vrp.c (remove_range_assertions): Forcefully propagate
28513 out SSA names even if abnormal.
28515 2017-01-27 Jakub Jelinek <jakub@redhat.com>
28517 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
28518 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
28519 instead of MPFR_RNDN.
28521 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
28524 * arm.c (arm_option_override): Don't call build_target_option_node
28525 until after doing all option overrides.
28526 (arm_valid_target_attribute_tree): Likewise.
28528 2017-01-27 Martin Liska <mliska@suse.cz>
28530 * doc/invoke.texi (-fprofile-arcs): Document profiling support
28531 for {cd}tors and C++ {cd}tors.
28533 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
28535 * config/s390/s390.md ("*setmem_long_and")
28536 ("*setmem_long_and_31z"): Use zero_extend instead of and.
28538 2017-01-26 Martin Sebor <msebor@redhat.com>
28540 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
28543 2017-01-26 Martin Sebor <msebor@redhat.com>
28545 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
28546 HAVE_DFmode before using XFmode or DFmode.
28547 (parse_directive): Avoid using the z length modifier to avoid
28548 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
28550 PR middle-end/78703
28551 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
28552 to accept adjustment as an array.
28553 (get_int_range): New function.
28554 (struct directive): Make width and prec arrays.
28555 (directive::set_width, directive::set_precision): Call get_int_range.
28556 (format_integer, format_floating): Handle width and precision ranges.
28557 (format_string, parse_directive): Same.
28559 2017-01-26 Jakub Jelinek <jakub@redhat.com>
28562 * dwarf2out.c (generate_skeleton_bottom_up): For children with
28563 comdat_type_p set, just clone them, but keep the children in the
28567 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
28568 which have direct callers with -fvar-tracking-assignments enabled
28570 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
28571 inside of type units.
28573 2017-01-26 Martin Sebor <msebor@redhat.com>
28575 PR middle-end/78703
28576 * gimple-ssa-sprintf.c (struct result_range): Add likely and
28578 (struct format_result): Replace number_chars, number_chars_min,
28579 and number_chars_max with a single member of struct result_range.
28581 (format_result::operator+=): Adjust.
28582 (struct fmtresult): Remove bounded. Handle likely and unlikely
28584 (fmtresult::adjust_for_width_or_precision): New function.
28585 (fmtresult:type_max_digits): New function.
28586 (bytes_remaining): Handle likely and unlikely counters.
28587 (min_bytes_remaining): Remove.
28588 (format_percent): Simplify.
28589 (format_integer, format_floating): Set likely and unlikely counters.
28590 (get_string_length, format_character, format_string): Same.
28591 (format_plain, should_warn_p): New function.
28592 (maybe_warn): Call should_warn_p. Update diagnostic messages
28593 and handle those for all directives, including plain strings.
28594 (format_directive): Handle likely and unlikely counters.
28595 Remove unnecessary quoting from diagnostics. Add an informational
28597 (add_bytes): Remove.
28598 (pass_sprintf_length::compute_format_length): Simplify.
28599 (try_substitute_return_value): Handle likely and unlikely counters.
28601 2017-01-26 Carl Love <cel@us.ibm.com>
28603 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
28604 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
28606 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
28609 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
28610 endianess for subregs into account.
28611 * lra-constraints.c (lra_constraints): Do risky transformations
28612 always on the first iteration.
28613 * lra-lives.c (check_pseudos_live_through_calls): Add arg
28614 last_call_used_reg_set.
28615 (process_bb_lives): Define and use last_call_used_reg_set.
28616 * lra.c (lra): Always continue after lra_constraints on the first
28619 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
28621 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
28623 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
28625 2017-01-26 Jakub Jelinek <jakub@redhat.com>
28627 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
28628 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
28629 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
28630 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
28631 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
28632 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
28633 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
28634 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
28635 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
28637 2017-01-26 Marek Polacek <polacek@redhat.com>
28640 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
28641 for the third operand.
28643 2017-01-26 Jakub Jelinek <jakub@redhat.com>
28645 PR middle-end/79236
28646 * omp-low.c (struct omp_context): Add simt_stmt field.
28647 (scan_omp_for): Return omp_context *.
28648 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
28649 context to the _simt_ SIMD stmt.
28650 (lower_omp_for): For combined SIMD with sibling _simt_
28651 SIMD, make sure to use the same decls in _looptemp_
28652 clauses as in the sibling.
28654 2017-01-26 David Sherwood <david.sherwood@arm.com>
28656 PR middle-end/79212
28657 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
28660 2017-01-26 Jakub Jelinek <jakub@redhat.com>
28663 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
28664 emit fld b; fld a; if possible.
28666 * brig-builtins.def: Update copyright years.
28667 * config/arm/arm_acle_builtins.def: Update copyright years.
28669 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
28672 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
28673 constraint instead of o for the stxsd instruction.
28675 2017-01-25 Carl Love <cel@us.ibm.com>
28677 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
28678 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
28680 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
28682 * doc/invoke.texi (C++ Dialect Options): Fix typo.
28684 2017-01-25 Richard Biener <rguenther@suse.de>
28686 PR tree-optimization/69264
28687 * target.def (vector_alignment_reachable): Improve documentation.
28688 * doc/tm.texi: Regenerate.
28689 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
28691 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
28692 earlier changes with respect to TYPE_USER_ALIGN.
28693 (vector_alignment_reachable_p): Likewise. Improve dumping.
28695 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28698 * config/arm/arm.md (xordi3): Force constant operand into a register
28701 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28703 * doc/invoke.texi (-fstore-merging): Correct default optimization
28704 levels at which it is enabled.
28705 (-O): Move -fstore-merging from list to...
28708 2017-01-25 Richard Biener <rguenther@suse.de>
28711 * omp-expand.c: Include debug.h.
28712 (expand_omp_taskreg): Make sure to generate early debug before
28713 outlining anything from a function.
28714 (expand_omp_target): Likewise.
28715 (grid_expand_target_grid_body): Likewise.
28717 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
28720 * asan.c (get_translation_unit_decl): New function.
28721 (asan_add_global): Extract modules file name from globals
28722 TRANSLATION_UNIT_DECL name.
28724 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
28727 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
28728 for long calls with APCS frame and VFP.
28730 2017-01-24 David Malcolm <dmalcolm@redhat.com>
28732 * cfg.c (original_copy_tables_initialized_p): New function.
28733 * cfg.h (original_copy_tables_initialized_p): New decl.
28734 * cfgrtl.c (relink_block_chain): Guard the call to
28735 free_original_copy_tables with a call to
28736 original_copy_tables_initialized_p.
28737 * cgraph.h (symtab_node::native_rtl_p): New decl.
28738 * cgraphunit.c (symtab_node::native_rtl_p): New function.
28739 (symtab_node::needed_p): Don't assert for early assembly output
28740 for __RTL functions.
28741 (cgraph_node::finalize_function): Set "force_output" for __RTL
28743 (cgraph_node::analyze): Bail out early for __RTL functions.
28744 (analyze_functions): Update assertion to support __RTL functions.
28745 (cgraph_node::expand): Bail out early for __RTL functions.
28746 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
28748 * function.h (struct function): Update comment for field
28750 * gimple-expr.c: Include "tree-pass.h".
28751 (gimple_has_body_p): Return false for __RTL functions.
28752 * Makefile.in (OBJS): Add run-rtl-passes.o.
28753 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
28755 (gcc::pass_manager::get_clean_slate): New accessor.
28756 * passes.c: Include "insn-addr.h".
28757 (should_skip_pass_p): Add logging. Update logic for running
28758 "expand" to be compatible with both __GIMPLE and __RTL. Guard
28759 property-provider override so it is only done for gimple passes.
28761 (skip_pass): New function.
28762 (execute_one_pass): Call skip_pass when skipping passes.
28763 * read-md.c (md_reader::read_char): Support filtering
28764 the input to a subset of line numbers.
28765 (md_reader::md_reader): Initialize fields
28766 m_first_line and m_last_line.
28767 (md_reader::read_file_fragment): New function.
28768 * read-md.h (md_reader::read_file_fragment): New decl.
28769 (md_reader::m_first_line): New field.
28770 (md_reader::m_last_line): New field.
28771 * read-rtl-function.c (function_reader::create_function): Only
28772 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
28773 curr_properties. Set DECL_INITIAL to a dummy block.
28774 (read_rtl_function_body_from_file_range): New function.
28775 * read-rtl-function.h (read_rtl_function_body_from_file_range):
28777 * run-rtl-passes.c: New file.
28778 * run-rtl-passes.h: New file.
28780 2017-01-24 Jeff Law <law@redhat.com>
28782 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
28785 2017-01-24 Bin Cheng <bin.cheng@arm.com>
28787 PR tree-optimization/79159
28788 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
28789 (record_nonwrapping_iv): Improve boundary using above function if no
28790 value range information.
28792 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
28793 Martin Jambor <mjambor@suse.cz>
28795 * brig-builtins.def: New file.
28796 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
28797 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
28798 (DEF_HSAIL_SAT_BUILTIN): Likewise.
28799 (DEF_HSAIL_INTR_BUILTIN): Likewise.
28800 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
28801 * builtin-types.def (BT_INT8): New.
28802 (BT_INT16): Likewise.
28803 (BT_UINT8): Likewise.
28804 (BT_UINT16): Likewise.
28805 (BT_FN_ULONG): Likewise.
28806 (BT_FN_UINT_INT): Likewise.
28807 (BT_FN_UINT_ULONG): Likewise.
28808 (BT_FN_UINT_LONG): Likewise.
28809 (BT_FN_UINT_PTR): Likewise.
28810 (BT_FN_ULONG_PTR): Likewise.
28811 (BT_FN_INT8_FLOAT): Likewise.
28812 (BT_FN_INT16_FLOAT): Likewise.
28813 (BT_FN_UINT32_FLOAT): Likewise.
28814 (BT_FN_UINT16_FLOAT): Likewise.
28815 (BT_FN_UINT8_FLOAT): Likewise.
28816 (BT_FN_UINT64_FLOAT): Likewise.
28817 (BT_FN_UINT16_UINT32): Likewise.
28818 (BT_FN_UINT32_UINT16): Likewise.
28819 (BT_FN_UINT16_UINT16_UINT16): Likewise.
28820 (BT_FN_INT_PTR_INT): Likewise.
28821 (BT_FN_UINT_PTR_UINT): Likewise.
28822 (BT_FN_LONG_PTR_LONG): Likewise.
28823 (BT_FN_ULONG_PTR_ULONG): Likewise.
28824 (BT_FN_VOID_UINT64_UINT64): Likewise.
28825 (BT_FN_UINT8_UINT8_UINT8): Likewise.
28826 (BT_FN_INT8_INT8_INT8): Likewise.
28827 (BT_FN_INT16_INT16_INT16): Likewise.
28828 (BT_FN_INT_INT_INT): Likewise.
28829 (BT_FN_UINT_FLOAT_UINT): Likewise.
28830 (BT_FN_FLOAT_UINT_UINT): Likewise.
28831 (BT_FN_ULONG_UINT_UINT): Likewise.
28832 (BT_FN_ULONG_UINT_PTR): Likewise.
28833 (BT_FN_ULONG_ULONG_ULONG): Likewise.
28834 (BT_FN_UINT_UINT_UINT): Likewise.
28835 (BT_FN_VOID_UINT_PTR): Likewise.
28836 (BT_FN_UINT_UINT_PTR: Likewise.
28837 (BT_FN_UINT32_UINT64_PTR): Likewise.
28838 (BT_FN_INT_INT_UINT_UINT): Likewise.
28839 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
28840 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
28841 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
28842 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
28843 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
28844 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
28845 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
28846 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
28847 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
28848 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
28849 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
28850 * doc/frontends.texi: List BRIG FE.
28851 * doc/install.texi (Testing): Add BRIG tesring requirements.
28852 * doc/invoke.texi (Overall Options): Mention BRIG.
28853 * doc/standards.texi (Standards): Doucment BRIG HSA version.
28855 2017-01-24 Richard Biener <rguenther@suse.de>
28857 PR translation/79208
28858 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
28860 2017-01-24 Martin Jambor <mjambor@suse.cz>
28863 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
28864 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
28865 and known_contexts.
28867 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
28869 PR middle-end/79123
28870 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
28871 casts from signed to unsigned really don't have a range.
28873 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
28875 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
28876 GMP_RNDx for compatiblity.
28878 2017-01-24 Martin Liska <mliska@suse.cz>
28881 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
28882 that would prevent us to call alloca with -1 as argument.
28884 2017-01-24 Jakub Jelinek <jakub@redhat.com>
28886 * dwarf2out.c (output_compilation_unit_header, output_file_names):
28887 Avoid -Wformat-security warning.
28889 2017-01-23 Andrew Pinski <apinski@cavium.com>
28891 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
28894 2017-01-23 Martin Sebor <msebor@redhat.com>
28896 PR middle-end/78703
28897 * gimple-ssa-sprintf.c (warn_level): New global.
28898 (format_integer): Use it here and throughout the rest of the file.
28899 Use the same switch to compute sign as base.
28900 (maybe_warn): New function.
28901 (format_directive): Factor out warnings into maybe_warn.
28902 Add debugging output. Use warn_level.
28903 (add_bytes): Use warn_level.
28904 (pass_sprintf_length::compute_format_length): Add debugging output.
28905 (try_substitute_return_value): Same.
28906 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
28908 PR middle-end/78703
28909 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
28910 (struct fmtresult, format_integer, format_floating): Adjust.
28911 (fmtresult::fmtresult): Set max correctly in two argument ctor.
28912 (get_string_length, format_string,format_directive): Same.
28913 (pass_sprintf_length::compute_format_length): Same.
28914 (try_substitute_return_value): Simplify slightly.
28916 PR middle-end/78703
28917 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
28918 (fmtresult::operator+=): Outlined.
28919 (struct fmtresult): Add ctors.
28920 (struct conversion_spec): Rename...
28921 (struct directive): ...to this. Add and remove data members.
28922 (directive::set_width, directive::set_precision): New functions.
28923 (format_percent): Use fmtresult ctor.
28924 (get_width_and_precision): Remove.
28925 (format_integer): Make naming changes. Avoid computing width and
28927 (format_floating): Same. Adjust indentation.
28928 (format_character, format_none): New functions.
28929 (format_string): Moved character handling to format_character.
28930 (format_directive): Remove arguments, change return type.
28931 (parse_directive): New function.
28932 (pass_sprintf_length::compute_format_length): Move directive
28933 parsing to parse_directive.
28935 2017-01-23 Jakub Jelinek <jakub@redhat.com>
28937 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
28938 (assign_assembler_name_if_needed): ... this.
28939 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
28940 (assign_assembler_name_if_needed): ... this.
28941 (free_lang_data_in_cgraph): Adjust callers.
28942 * cgraphunit.c (cgraph_node::analyze): Likewise.
28943 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
28946 2017-01-23 Richard Biener <rguenther@suse.de>
28948 PR tree-optimization/79088
28949 PR tree-optimization/79188
28950 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
28951 resetting loop bounds after last path deletion. Reset loop
28952 bounds of the target loop, make code match the comments.
28953 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
28954 Make sure loops need no fixups.
28956 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
28958 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
28959 exponent support with double type for first argument.
28960 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
28961 type returned by __builtin_vec_extract_sig,
28962 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
28963 functions from "vector int" to "vector unsigned int" or from
28964 "vector long long int" to "vector unsigned long long int".
28965 Changed type returned by __builtin_vec_extract_exp,
28966 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
28967 functions from "vector int" to "vector unsigned int" or from
28968 "vector long long int" to "vector unsigned long long int".
28969 Changed return type of __builtin_vec_test_data_class,
28970 __builtin_vec_test_data_class_sp, and
28971 __builtin_vec_test_data_class_dp from "vector int" to
28972 "vector bool int" or from "vector long long int" to "vector bool
28973 long long int" and changed second argument type from "unsigned
28974 int" to "int". Added new overloaded function forms "vector float
28975 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
28976 "vector float __builtin_vec_insert_exp_sp (vector float, vector
28977 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
28978 double, vector unsigned long long int)" and "vector double
28979 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
28980 long int)". Changed return type of
28981 __builtin_scalar_test_data_class and
28982 __builtin_scalar_test_data_class_sp and
28983 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
28984 int" and changed second argument from "unsigned int" to "int".
28985 Changed type returned by __builtin_scalar_test_neg,
28986 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
28987 from "int" to "bool int". Added new overloaded function form
28988 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
28989 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
28990 exponent double-precision with floating point first argument.
28991 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
28992 documentation of scalar_test_data_class, scalar_test_neg,
28993 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
28994 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
28995 vec_test_data_class built-in functions to reflect refinements in
28996 their type signatures.
28998 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
29000 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
29002 (aarch64_elf_asm_destructor): Likewise.
29004 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
29006 PR rtl-optimization/78634
29007 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
29008 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
29009 * ifcvt.c (noce_try_cmove): Add missing cost check.
29011 PR rtl-optimization/71724
29012 * combine.c (if_then_else_cond): Look for situations where it is
29013 beneficial to undo the work of one of the recursive calls.
29015 2017-01-23 Bin Cheng <bin.cheng@arm.com>
29017 PR tree-optimization/70754
29018 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
29019 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
29020 combined stmt before it if not NULL.
29021 (combine_chains): Process refs reversely and compute dominance point
29024 2017-01-23 Martin Liska <mliska@suse.cz>
29026 PR tree-optimization/79196
29027 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
29028 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
29030 (strlen_optimize_stmt): Call the renamed function.
29032 2017-01-23 Michael Matz <matz@suse.de>
29034 PR tree-optimization/78384
29035 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
29037 2017-01-23 Richard Biener <rguenther@suse.de>
29039 PR tree-optimization/79186
29040 * tree-vrp.c (register_new_assert_for): Make sure we've seen
29041 both incoming edges before moving an assert.
29043 2017-01-23 Martin Jambor <mjambor@suse.cz>
29045 * ipa-prop.c (load_from_param_1): Removed.
29046 (load_from_unmodified_param): Bits from load_from_param_1 put back
29048 (load_from_param): Removed.
29049 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
29050 with stmt. Reverted back to use of load_from_unmodified_param.
29052 2017-01-23 Martin Jambor <mjambor@suse.cz>
29055 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
29056 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
29057 field a pointer to garbage collected vector, mark lattices and
29058 ipcp_orig_node with GTY((skip)).
29059 (ipa_get_param_count): Adjust to descriptors being a pointer.
29060 (ipa_get_param): Likewise.
29061 (ipa_get_type): Likewise.
29062 (ipa_get_param_move_cost): Likewise.
29063 (ipa_set_param_used): Likewise.
29064 (ipa_get_controlled_uses): Likewise.
29065 (ipa_set_controlled_uses): Likewise.
29066 (ipa_is_param_used): Likewise.
29067 (ipa_node_params_t): Move into garbage collector. New methods insert
29069 (ipa_node_params_sum): Annotate wth GTY(()).
29070 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
29072 (ipa_load_from_parm_agg): Adjust declaration.
29073 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
29074 * ipa-profile.c (ipa_profile): Likewise.
29075 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
29076 (ipa_populate_param_decls): Make descriptors parameter garbage
29078 (ipa_dump_param): Adjust to descriptors being a pointer.
29079 (ipa_alloc_node_params): Likewise.
29080 (ipa_initialize_node_params): Likewise.
29081 (load_from_param_1): Make descriptors parameter garbage collected.
29082 (load_from_unmodified_param): Likewise.
29083 (load_from_param): Likewise.
29084 (ipa_load_from_parm_agg): Likewise.
29085 (ipa_node_params::~ipa_node_params): Removed.
29086 (ipa_free_all_node_params): Remove call to delete operator.
29087 (ipa_node_params_t::insert): New.
29088 (ipa_node_params_t::remove): Likewise.
29089 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
29090 copy known_csts and known_contexts vectors.
29091 (ipa_read_node_info): Adjust to descriptors being a pointer.
29092 (ipcp_modif_dom_walker): Make m_descriptors field garbage
29094 (ipcp_transform_function): Make descriptors variable garbage
29097 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
29099 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
29100 * config/i386/avx512dqintrin.h: Ditto.
29101 * config/i386/avx512fintrin.h: Ditto.
29102 * config/i386/i386.c: Handle new builtins.
29103 * config/i386/i386-builtin.def: Add new builtins.
29104 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
29105 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
29107 2017-01-23 Jakub Jelinek <jakub@redhat.com>
29108 Martin Liska <mliska@suse.cz>
29110 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
29111 * asan.c (asan_expand_poison_ifn): Support stores and use
29112 appropriate ASAN report function.
29113 * internal-fn.c (expand_ASAN_POISON_USE): New function.
29114 * internal-fn.def (ASAN_POISON_USE): Declare.
29115 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
29116 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
29117 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
29118 ASAN_POISON calls w/o LHS.
29119 * tree-ssa.c (execute_update_addresses_taken): Create clobber
29120 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
29121 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
29122 * gimplify.c (asan_poison_variables): Add attribute
29123 use_after_scope_memory to variables that really needs to live
29125 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
29126 having the attribute.
29128 2017-01-23 Martin Liska <mliska@suse.cz>
29130 * asan.c (create_asan_shadow_var): New function.
29131 (asan_expand_poison_ifn): Likewise.
29132 * asan.h (asan_expand_poison_ifn): New declaration.
29133 * internal-fn.c (expand_ASAN_POISON): Likewise.
29134 * internal-fn.def (ASAN_POISON): New builtin.
29135 * sanopt.c (pass_sanopt::execute): Expand
29136 asan_expand_poison_ifn.
29137 * tree-inline.c (copy_decl_for_dup_finish): Make function
29139 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
29140 * tree-ssa.c (is_asan_mark_p): New function.
29141 (execute_update_addresses_taken): Rewrite local variables
29142 (identified just by use-after-scope as addressable) into SSA.
29144 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
29146 * doc/install.texi (Specific): opensource.apple.com uses https
29147 now. Remove trailing slash.
29149 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
29151 * README.Portability: Remove note on an Irix compatibility issue.
29153 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
29155 * gcov.c (INCLUDE_ALGORITHM): Define.
29156 (INCLUDE_VECTOR): Define.
29157 No longer include <vector> and <algorithm> directly.
29159 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
29161 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
29163 * doc/invoke.texi (Code Gen Options): Ditto.
29165 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
29168 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
29170 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
29172 rtl-optimization/79125
29173 * cprop.c (local_cprop_pass): Handle cases where we make an
29174 unconditional trap.
29176 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
29180 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
29181 read from, for big endian.
29183 2017-01-20 Jiong Wang <jiong.wang@arm.com>
29185 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
29186 register pauth builtins for LP64 only.
29188 2017-01-20 Marek Polacek <polacek@redhat.com>
29191 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
29194 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
29196 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
29198 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
29199 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
29200 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
29202 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29205 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
29206 in big-endian mode when they are not a single duplicated value.
29208 2017-01-20 Richard Biener <rguenther@suse.de>
29210 * BASE-VER: Bump to 7.0.1.
29212 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
29214 * omp-low.c (omplow_simd_context): New struct. Use it...
29215 (lower_rec_simd_input_clauses): ...here and...
29216 (lower_rec_input_clauses): ...here to hold common data. Adjust all
29217 references to idx, lane, max_vf, is_simt.
29219 2017-01-20 Graham Markall <graham.markall@embecosm.com>
29221 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
29224 2017-01-20 Martin Jambor <mjambor@suse.cz>
29226 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
29227 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
29229 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
29230 (GTFILES): Rename hsa.c to hsa-common.c.
29231 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
29232 * hsa-dump.c: Likewise.
29233 * hsa-gen.c: Likewise.
29234 * hsa-regalloc.c: Likewise.
29235 * ipa-hsa.c: Likewise.
29236 * omp-expand.c: Likewise.
29237 * omp-low.c: Likewise.
29238 * toplev.c: Likewise.
29240 2017-01-20 Marek Polacek <polacek@redhat.com>
29243 * doc/invoke.texi: Document -Wduplicated-branches.
29244 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
29245 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
29246 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
29247 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
29248 return 0 only when not OEP_LEXICOGRAPHIC.
29249 (fold_build_cleanup_point_expr): Use the expression
29250 location when building CLEANUP_POINT_EXPR.
29251 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
29252 * tree.c (add_expr): Handle error_mark_node.
29254 2017-01-20 Martin Liska <mliska@suse.cz>
29257 * tree-profile.c (init_ic_make_global_vars): Do not call
29259 (gimple_init_gcov_profiler): Likewise.
29261 2017-01-20 Martin Liska <mliska@suse.cz>
29264 * cgraph.h (maybe_create_reference): Remove argument and
29266 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
29268 * ipa-cp.c (create_specialized_node): Likewise.
29269 * symtab.c (symtab_node::maybe_create_reference): Handle
29270 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
29272 2017-01-20 Martin Liska <mliska@suse.cz>
29274 * read-rtl-function.c (function_reader::create_function): Use
29275 build_decl instread of build_decl_stat.
29277 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
29279 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
29280 * config/i386/avx512dqintrin.h: Ditto.
29281 * config/i386/avx512fintrin.h: Ditto.
29282 * config/i386/i386-builtin-types.def: Add new types.
29283 * config/i386/i386.c: Handle new types.
29284 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
29285 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
29286 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
29287 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
29288 (__builtin_ia32_kshiftridi): New.
29289 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
29291 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
29295 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
29296 define to rs6000_init_stack_protect_guard.
29297 (rs6000_init_stack_protect_guard): New function.
29299 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
29300 Yunqiang Su <yunqiang.su@imgtec.com>
29302 * config.gcc (supported_defaults): Add madd4.
29303 (with_madd4): Add validation.
29304 (all_defaults): Add madd4.
29305 * config/mips/mips.opt (mmadd4): New option.
29306 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
29308 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
29310 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
29311 (ISA_HAS_FUSED_MADD4): Likewise.
29312 * doc/invoke.texi (-mmadd4): Document the new option.
29313 * doc/install.texi (--with-madd4): Document the new option.
29315 2017-01-19 Jiong Wang <jiong.wang@arm.com>
29317 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
29318 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
29319 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
29320 (aarch64_init_pauth_hint_builtins): New.
29321 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
29322 (aarch64_expand_builtin): Expand new builtins.
29324 2017-01-19 Jiong Wang <jiong.wang@arm.com>
29326 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
29327 * combine-stack-adj.c (no_unhandled_cfa): Handle
29328 REG_CFA_TOGGLE_RA_MANGLE.
29329 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
29330 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
29331 info for return address signing.
29332 (aarch64_expand_epilogue): Likewise.
29334 2017-01-19 Jiong Wang <jiong.wang@arm.com>
29336 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
29337 * config/aarch64/aarch64-protos.h
29338 (aarch64_return_address_signing_enabled): New declaration.
29339 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
29341 (aarch64_expand_prologue): Sign return address before it's pushed onto
29343 (aarch64_expand_epilogue): Authenticate return address fetched from
29345 (aarch64_override_options): Sanity check for ILP32 and ISA level.
29346 (aarch64_attributes): New function attributes for "sign-return-address".
29347 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
29348 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
29349 ("*do_return"): Generate combined instructions according to key index.
29350 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
29351 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
29353 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
29354 * config/aarch64/aarch64.opt (msign-return-address=): New.
29355 * doc/extend.texi (AArch64 Function Attributes): Documents
29356 "sign-return-address=".
29357 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
29359 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
29361 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
29362 overall option summary.
29364 2017-01-19 Jiong Wang <jiong.wang@arm.com>
29366 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
29367 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
29368 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
29369 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
29371 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
29373 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
29374 -mpower9-minmax by default for -mcpu=power9.
29375 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
29376 128-bit floating point.
29378 2017-01-20 Alan Modra <amodra@gmail.com>
29380 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
29381 optimizing for size.
29383 2017-01-20 Alan Modra <amodra@gmail.com>
29386 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
29387 for strcmp and strncmp from corresponding builtin decl.
29389 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
29391 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
29392 instead of i386/rtems-64.h.
29393 * config/i386/rtems-64.h: Remove.
29395 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
29399 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
29401 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
29403 2017-01-19 Tamar Christina <tamar.christina@arm.com>
29405 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
29406 Change int to HOST_WIDE_INT.
29407 * config/aarch64/aarch64-protos.h
29408 (aarch64_simd_gen_const_vector_dup): Likewise.
29409 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
29411 2017-01-19 David Malcolm <dmalcolm@redhat.com>
29413 * langhooks-def.h (lhd_type_for_size): New decl.
29414 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
29415 * langhooks.c (lhd_type_for_size): New function, taken from
29418 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
29420 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
29421 define_bypass for CR latency.
29422 (power9-cracked-alu): Update bypass latency and remove power9-branch.
29423 (power9-alu2): Add define_bypass for CR latency.
29425 (power9-mul): Update insn latency.
29426 (power9-mul-compare): Update insn latency, bypass latency and remove
29429 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29431 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
29433 * config/aarch64/aarch64.md
29434 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
29435 aarch64_nopcrelative_literal_loads.
29436 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
29438 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
29440 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
29441 TARGET_LOONGSON_3A.
29442 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
29444 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
29447 * config.gcc (supported_defaults): Add lxc1-sxc1.
29448 (with_lxc1_sxc1): Add validation.
29449 (all_defaults): Add lxc1-sxc1.
29450 * config/mips/mips.opt (mlxc1-sxc1): New option.
29451 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
29453 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
29454 __mips_no_lxc1_sxc1.
29455 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
29456 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
29457 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
29459 2017-01-19 Richard Biener <rguenther@suse.de>
29461 PR tree-optimization/72488
29462 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
29463 sure to restore SSA info.
29464 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
29466 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
29468 PR rtl-optimization/79121
29469 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
29470 of the inner type when shifting an extended value.
29472 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
29475 * symtab.c (symtab_node::equal_address_to): Fix comparing of
29476 interposable aliases.
29478 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
29481 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
29482 Use the evmergelohi instruction.
29483 (mov_si<mode>_e500_subreg4_2_le): Likewise.
29484 (mov_sitf_e500_subreg8_2_be): Likewise.
29485 (mov_sitf_e500_subreg12_2_le): Likewise.
29486 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
29487 (mov_si<mode>_e500_subreg4_2_be): Likewise.
29488 (mov_sitf_e500_subreg8_2_le): Likewise.
29489 (mov_sitf_e500_subreg12_2_be): Likewise.
29491 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29493 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
29494 attribute from vecsimple to vecperm.
29495 (altivec_vbpermq2): Likewise.
29497 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29500 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
29502 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29503 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
29504 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
29505 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
29506 case where N arg is SIZE_MAX.
29507 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
29508 (cmpstrsi): Add pattern.
29510 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
29512 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
29513 __builtin_vec_revb builtins.
29514 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
29515 built-in functions to support generation of the ISA 3.0 XXBR<x>
29516 vector byte reverse instructions.
29517 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
29518 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
29519 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
29520 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
29521 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
29522 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
29523 (P9V_BUILTIN_VEC_REVB): Likewise.
29524 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
29525 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
29526 (p9_xxbrq_v16qi): Likewise.
29527 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
29528 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
29529 (p9_xxbrh_v8hi): Likewise.
29530 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
29531 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
29532 vec_revb built-in functions.
29534 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
29536 PR rtl-optimization/78952
29537 * config/i386/i386.md (any_extract): New code iterator.
29538 (*insvqi_2): Use any_extract for source operand.
29539 (*insvqi_3): Use any_shiftrt for source operand.
29541 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
29543 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
29545 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
29547 2017-01-18 Matthias Klose <doko@ubuntu.com>
29549 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
29551 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29553 * config/rs6000/altivec.h (vec_bperm): Change #define.
29554 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
29555 (altivec_vbpermq2): New define_insn.
29556 (altivec_vbpermd): Likewise.
29557 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
29558 function interface.
29559 (VBPERMD): Likewise.
29560 (VBPERM): New polymorphic function interface.
29561 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
29562 Add entries for P9V_BUILTIN_VEC_VBPERM.
29563 * doc/extend.texi: Add interfaces for vec_bperm.
29565 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29567 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
29568 first letter of error messages.
29569 (s390_resolve_overloaded_builtin): Likewise.
29570 * config/s390/s390.c (s390_expand_builtin): Likewise.
29571 (s390_invalid_arg_for_unprototyped_fn): Likewise.
29572 (s390_valid_target_attribute_inner_p): Likewise.
29573 * config/s390/s390.md ("tabort"): Likewise.
29575 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
29577 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
29578 (ISA_AVOID_DIV_HILO): New macro.
29579 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
29580 (ISA_HAS_DDIV): Likewise.
29582 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
29584 * doc/invoke.texi (fabi-version): Correct number of occurrences.
29586 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
29588 * doc/invoke.texi (fabi-version): Spelling fix.
29590 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
29593 * doc/invoke.texi (fabi-version): Mention mangling fix for
29596 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
29599 * doc/invoke.texi (fabi-version): Document discriminator mangling.
29601 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
29604 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
29605 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
29607 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
29609 (stack_protect_test): Ditto.
29610 * config/rs6000/rs6000.opt (mstack-protector-guard=,
29611 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
29613 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
29614 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
29615 -mstack-protector-guard-offset=.
29616 (RS/6000 and PowerPC Options): Ditto.
29618 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
29620 * config/i386/i386.h (MASK_CLASS_P): New define.
29621 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
29622 there are no registers from different register sets also when
29623 mask registers are used. Update function comment.
29624 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
29625 to (*k/*r) and (*k/*km) alternatives.
29627 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
29629 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
29630 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
29631 (EH_RETURN_HANDLER_RTX): New define.
29632 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
29633 Force frame pointer in EH return functions.
29634 (aarch64_expand_epilogue): Add barrier for eh_return.
29635 (aarch64_final_eh_return_addr): Remove.
29636 (aarch64_eh_return_handler_rtx): New function.
29637 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
29639 (aarch64_eh_return_handler_rtx): New prototype.
29641 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29643 * config/rs6000/altivec.h (vec_rlmi): New #define.
29644 (vec_vrlnm): Likewise.
29645 (vec_rlnm): Likewise.
29646 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
29647 (UNSPEC_VRLNM): Likewise.
29648 (VIlong): New mode iterator.
29649 (altivec_vrl<VI_char>mi): New define_insn.
29650 (altivec_vrl<VI_char>nm): Likewise.
29651 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
29653 (VRLDNM): Likewise.
29654 (RLNM): New polymorphic function entry.
29655 (VRLWMI): New monomorphic function entry.
29656 (VRLDMI): Likewise.
29657 (RLMI): New polymorphic function entry.
29658 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
29659 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
29660 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
29663 2017-01-17 Jakub Jelinek <jakub@redhat.com>
29666 * dwarf2out.c (field_byte_offset): Restore the
29667 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
29668 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
29669 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
29672 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
29675 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
29677 2017-01-17 Jakub Jelinek <jakub@redhat.com>
29680 * dwarf2out.c (add_data_member_location_attribute): For constant
29681 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
29682 instead of DW_AT_data_member_location, DW_AT_bit_offset and
29683 DW_AT_byte_size attributes.
29685 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
29687 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
29688 after forcing to constant memory when the code model is medium.
29690 2017-01-17 Julia Koval <julia.koval@intel.com>
29693 * config/i386/avx512fintrin.h
29694 (_mm512_i32gather_ps): Change __addr type to void const*.
29695 (_mm512_mask_i32gather_ps): Ditto.
29696 (_mm512_i32gather_pd): Ditto.
29697 (_mm512_mask_i32gather_pd): Ditto.
29698 (_mm512_i64gather_ps): Ditto.
29699 (_mm512_mask_i64gather_ps): Ditto.
29700 (_mm512_i64gather_pd): Ditto.
29701 (_mm512_mask_i64gather_pd): Ditto.
29702 (_mm512_i32gather_epi32): Ditto.
29703 (_mm512_mask_i32gather_epi32): Ditto.
29704 (_mm512_i32gather_epi64): Ditto.
29705 (_mm512_mask_i32gather_epi64): Ditto.
29706 (_mm512_i64gather_epi32): Ditto.
29707 (_mm512_mask_i64gather_epi32): Ditto.
29708 (_mm512_i64gather_epi64): Ditto.
29709 (_mm512_mask_i64gather_epi64): Ditto.
29710 (_mm512_i32scatter_ps): Change __addr type to void*.
29711 (_mm512_mask_i32scatter_ps): Ditto.
29712 (_mm512_i32scatter_pd): Ditto.
29713 (_mm512_mask_i32scatter_pd): Ditto.
29714 (_mm512_i64scatter_ps): Ditto.
29715 (_mm512_mask_i64scatter_ps): Ditto.
29716 (_mm512_i64scatter_pd): Ditto.
29717 (_mm512_mask_i64scatter_pd): Ditto.
29718 (_mm512_i32scatter_epi32): Ditto.
29719 (_mm512_mask_i32scatter_epi32): Ditto.
29720 (_mm512_i32scatter_epi64): Ditto.
29721 (_mm512_mask_i32scatter_epi64): Ditto.
29722 (_mm512_i64scatter_epi32): Ditto.
29723 (_mm512_mask_i64scatter_epi32): Ditto.
29724 (_mm512_i64scatter_epi64): Ditto.
29725 (_mm512_mask_i64scatter_epi64): Ditto.
29726 * config/i386/avx512pfintrin.h
29727 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
29728 (_mm512_mask_prefetch_i32gather_ps): Ditto.
29729 (_mm512_mask_prefetch_i64gather_pd): Ditto.
29730 (_mm512_mask_prefetch_i64gather_ps): Ditto.
29731 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
29732 (_mm512_prefetch_i32scatter_ps): Ditto.
29733 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
29734 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
29735 (_mm512_prefetch_i64scatter_pd): Ditto.
29736 (_mm512_prefetch_i64scatter_ps): Ditto.
29737 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
29738 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
29739 * config/i386/avx512vlintrin.h
29740 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
29741 (_mm_mmask_i32gather_ps): Ditto.
29742 (_mm256_mmask_i32gather_pd): Ditto.
29743 (_mm_mmask_i32gather_pd): Ditto.
29744 (_mm256_mmask_i64gather_ps): Ditto.
29745 (_mm_mmask_i64gather_ps): Ditto.
29746 (_mm256_mmask_i64gather_pd): Ditto.
29747 (_mm_mmask_i64gather_pd): Ditto.
29748 (_mm256_mmask_i32gather_epi32): Ditto.
29749 (_mm_mmask_i32gather_epi32): Ditto.
29750 (_mm256_mmask_i32gather_epi64): Ditto.
29751 (_mm_mmask_i32gather_epi64): Ditto.
29752 (_mm256_mmask_i64gather_epi32): Ditto.
29753 (_mm_mmask_i64gather_epi32): Ditto.
29754 (_mm256_mmask_i64gather_epi64): Ditto.
29755 (_mm_mmask_i64gather_epi64): Ditto.
29756 (_mm256_i32scatter_ps): Change __addr type to void*.
29757 (_mm256_mask_i32scatter_ps): Ditto.
29758 (_mm_i32scatter_ps): Ditto.
29759 (_mm_mask_i32scatter_ps): Ditto.
29760 (_mm256_i32scatter_pd): Ditto.
29761 (_mm256_mask_i32scatter_pd): Ditto.
29762 (_mm_i32scatter_pd): Ditto.
29763 (_mm_mask_i32scatter_pd): Ditto.
29764 (_mm256_i64scatter_ps): Ditto.
29765 (_mm256_mask_i64scatter_ps): Ditto.
29766 (_mm_i64scatter_ps): Ditto.
29767 (_mm_mask_i64scatter_ps): Ditto.
29768 (_mm256_i64scatter_pd): Ditto.
29769 (_mm256_mask_i64scatter_pd): Ditto.
29770 (_mm_i64scatter_pd): Ditto.
29771 (_mm_mask_i64scatter_pd): Ditto.
29772 (_mm256_i32scatter_epi32): Ditto.
29773 (_mm256_mask_i32scatter_epi32): Ditto.
29774 (_mm_i32scatter_epi32): Ditto.
29775 (_mm_mask_i32scatter_epi32): Ditto.
29776 (_mm256_i32scatter_epi64): Ditto.
29777 (_mm256_mask_i32scatter_epi64): Ditto.
29778 (_mm_i32scatter_epi64): Ditto.
29779 (_mm_mask_i32scatter_epi64): Ditto.
29780 (_mm256_i64scatter_epi32): Ditto.
29781 (_mm256_mask_i64scatter_epi32): Ditto.
29782 (_mm_i64scatter_epi32): Ditto.
29783 (_mm_mask_i64scatter_epi32): Ditto.
29784 (_mm256_i64scatter_epi64): Ditto.
29785 (_mm256_mask_i64scatter_epi64): Ditto.
29786 (_mm_i64scatter_epi64): Ditto.
29787 (_mm_mask_i64scatter_epi64): Ditto.
29788 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
29789 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
29790 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
29791 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
29792 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
29793 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
29794 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
29795 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
29796 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
29797 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
29798 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
29799 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
29800 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
29801 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
29802 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
29803 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
29804 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
29805 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
29806 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
29807 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
29808 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
29809 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
29810 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
29811 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
29812 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
29813 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
29814 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
29815 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
29816 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
29817 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
29818 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
29819 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
29820 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
29821 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
29822 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
29823 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
29824 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
29825 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
29826 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
29827 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
29828 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
29829 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
29830 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
29831 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
29832 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
29833 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
29834 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
29835 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
29836 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
29837 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
29838 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
29839 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
29840 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
29841 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
29842 definitions accordingly.
29844 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
29845 Kuan-Lin Chen <kuanlinchentw@gmail.com>
29848 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
29851 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
29854 * ira-conflicts.c (ira_build_conflicts): Update total conflict
29855 hard regs for inner regno.
29857 2017-01-17 Martin Liska <mliska@suse.cz>
29860 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
29861 assumption and add comment.
29863 2017-01-17 Nathan Sidwell <nathan@acm.org>
29865 * ipa-visibility.c (localize_node): New function, broken out of ...
29866 (function_and_variable_visibility): ... here. Call it.
29868 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
29870 PR middle-end/77445
29871 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
29872 correctly set frequency of oudgoing edge.
29873 (duplicate_thread_path): Fix profile updating.
29875 2017-01-17 Jakub Jelinek <jakub@redhat.com>
29878 * configure.ac: Add GCC_BASE_VER.
29879 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
29880 version from BASE-VER file.
29881 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
29882 (gcc.o): Depend on $(BASEVER).
29883 * common.opt (dumpfullversion): New option.
29884 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
29885 * doc/invoke.texi: Document -dumpfullversion.
29886 * doc/install.texi: Document --with-gcc-major-version-only.
29887 * configure: Regenerated.
29889 2017-01-17 Richard Biener <rguenther@suse.de>
29891 PR tree-optimization/71433
29892 * tree-vrp.c (register_new_assert_for): Merge same asserts
29893 on all incoming edges.
29894 (process_assert_insertions_for): Handle insertions at the
29897 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
29899 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
29900 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
29902 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
29905 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
29908 2017-01-17 Alan Modra <amodra@gmail.com>
29911 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
29912 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
29913 symbolic stack limit when pic.
29915 2017-01-16 Martin Sebor <msebor@redhat.com>
29917 PR tree-optimization/78608
29918 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
29920 2017-01-16 Jeff Law <law@redhat.com>
29923 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
29924 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
29925 for several include directories that may be relative to sysroot.
29926 * config/i386/x-mingw32 (gplus_includedir): Define.
29927 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
29928 (native_system_includedir): Likewise.
29929 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
29930 override if TARGET_SYSTEM_ROOT is defined.
29931 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
29933 PR tree-optimization/79090
29934 PR tree-optimization/33562
29935 PR tree-optimization/61912
29936 PR tree-optimization/77485
29937 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
29938 and computed trims into the dump file.
29940 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
29942 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
29944 2017-01-16 Jakub Jelinek <jakub@redhat.com>
29947 * gimplify.c (gimplify_init_constructor): If want_value and
29948 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
29952 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
29953 sequence. Formatting fixes.
29954 (doloop_optimize): Formatting fixes.
29957 * gcc.c (debug_level_greater_than_spec_func): New function.
29958 (static_spec_functions): Add debug-level-gt spec function.
29959 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
29961 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
29962 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
29963 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
29964 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
29965 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
29966 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
29968 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
29970 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
29971 QImode fixups to general and mask registers only.
29973 2017-01-16 Carl Love <cel@us.ibm.com>
29975 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
29976 for built-in functions
29977 vector signed char vec_nabs (vector signed char)
29978 vector signed short vec_nabs (vector signed short)
29979 vector signed int vec_nabs (vector signed int)
29980 vector signed long long vec_nabs (vector signed long long)
29981 vector float vec_nabs (vector float)
29982 vector double vec_nabs (vector double)
29983 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
29985 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
29986 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
29987 * doc/extend.texi: Update the documentation file for the new built-in
29990 2017-01-16 Martin Sebor <msebor@redhat.com>
29992 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
29995 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29997 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
29998 UNSPEC_VSX__XXSPLTD to require special splat handling.
30000 2017-01-16 David Malcolm <dmalcolm@redhat.com>
30003 * system.h: Poison strndup.
30005 2017-01-16 Alan Modra <amodra@gmail.com>
30008 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
30011 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
30013 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
30015 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
30017 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
30018 call recog here. Assert that INSN_CODE (insn) is non-negative.
30020 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
30023 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
30025 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
30026 in the currently scheduled RTL fragment.
30028 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
30030 PR rtl-optimization/78751
30031 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
30034 2017-01-14 Jeff Law <law@redhat.com>
30036 PR tree-optimization/79090
30037 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
30038 variable length stores.
30039 (compute_trims): Delete dead assignment to *trim_tail.
30040 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
30043 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
30045 PR rtl-optimization/78626
30046 PR rtl-optimization/78727
30047 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
30048 of a block, and split such blocks after everything else is finished.
30050 2017-01-14 Alan Modra <amodra@gmail.com>
30053 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
30054 target legitimate_combined_insn.
30055 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
30056 (rs6000_legitimate_combined_insn): New function.
30057 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
30059 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
30060 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
30061 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
30063 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
30065 * doc/frontends.texi (G++ and GCC): Remove references to Java.
30067 2017-01-13 Jeff Law <law@redhat.com>
30069 PR tree-optimization/33562
30070 PR tree-optimization/61912
30071 PR tree-optimization/77485
30072 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
30074 (delete_dead_assignment): Likewise.
30075 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
30076 statement to delete_dead_call and delete_dead_assignment.
30078 2017-01-13 David Malcolm <dmalcolm@redhat.com>
30081 * substring-locations.c (format_warning_va): Strengthen case 1 so
30082 that both endpoints of the substring must be within the format
30083 range for just the substring to be printed.
30085 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
30087 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
30088 * config/i386/i386.c (ix86_target_string): Add missing options
30089 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
30090 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
30091 flags_other and ix86_target_other to flags2_other. Display unknown
30093 (ix86_valid_target_attribute_inner_p): Add missing options and
30094 reorder options by implied ISAs, as in ix86_target_string.
30096 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
30098 * hash-table.h (hash_table::too_empty_p): New function.
30099 (hash_table::expand): Use it.
30100 (hash_table::traverse): Likewise.
30101 (hash_table::empty_slot): Use sizeof (value_type) instead of
30102 sizeof (PTR) to convert bytes to elements. Shrink the table
30103 if the current size is excessive for the current number of
30106 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
30108 * ira-costs.c (record_reg_classes): Break from the inner loop
30109 early once alt_fail is known to be true. Update outer loop
30110 handling accordingly.
30112 2017-01-13 Jeff Law <law@redhat.com>
30114 * tree-ssa-dse.c (decrement_count): New function.
30115 (increment_start_addr, maybe_trim_memstar_call): Likewise.
30116 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
30117 when we know the partially dead statement is a mem* function.
30119 PR tree-optimization/61912
30120 PR tree-optimization/77485
30121 * tree-ssa-dse.c: Include expr.h.
30122 (maybe_trim_constructor_store): New function.
30123 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
30125 PR tree-optimization/33562
30126 PR tree-optimization/61912
30127 PR tree-optimization/77485
30128 * doc/invoke.texi: Document new dse-max-object-size param.
30129 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
30130 * tree-ssa-dse.c: Include params.h.
30131 (dse_store_status): New enum.
30132 (initialize_ao_ref_for_dse): New, partially extracted from
30134 (valid_ao_ref_for_dse, normalize_ref): New.
30135 (setup_live_bytes_from_ref, compute_trims): Likewise.
30136 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
30137 (maybe_trim_partially_dead_store): Likewise.
30138 (maybe_trim_complex_store): Likewise.
30139 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
30140 Track what bytes live from the original store. Return tri-state
30141 for dead, partially dead or live.
30142 (dse_dom_walker): Add constructor, destructor and new private members.
30143 (delete_dead_call, delete_dead_assignment): New extracted from
30145 (dse_optimize_stmt): Make a member of dse_dom_walker.
30146 Use initialize_ao_ref_for_dse.
30148 PR tree-optimization/33562
30149 PR tree-optimization/61912
30150 PR tree-optimization/77485
30151 * sbitmap.h (bitmap_count_bits): Prototype.
30152 (bitmap_clear_range, bitmap_set_range): Likewise.
30153 * sbitmap.c (bitmap_clear_range): New function.
30154 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
30156 2017-01-13 Martin Liska <mliska@suse.cz>
30159 * function.c (set_cfun): Add new argument force.
30160 * function.h (set_cfun): Likewise.
30161 * ipa-inline-transform.c (inline_call): Use the function when
30162 strict alising from is dropped for function we inline to.
30164 2017-01-13 Richard Biener <rguenther@suse.de>
30166 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
30167 for dumping GIMPLE INTEGER_CSTs.
30169 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
30171 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
30172 to 201112L since C++17.
30174 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
30177 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
30178 if -fsanitize=kernel-address is present.
30180 2017-01-13 Richard Biener <rguenther@suse.de>
30182 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
30183 as _Literal ( type ) number in case usual suffixes do not
30184 preserve all information.
30186 2017-01-13 Richard Biener <rguenther@suse.de>
30188 PR tree-optimization/77283
30189 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
30190 and ssa-iterators.h.
30191 (is_feasible_trace): Implement a cost model based on joiner
30194 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
30197 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
30198 char or short to __float128/_Float128 directly.
30200 2017-01-12 Martin Sebor <msebor@redhat.com>
30202 to -Wformat-overflow.
30203 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
30204 (min_bytes_remaining): Same.
30205 (get_string_length): Same.
30206 (format_string): Same.
30207 (format_directive): Same.
30209 (pass_sprintf_length::handle_gimple_call): Same.
30211 2017-01-12 Jakub Jelinek <jakub@redhat.com>
30213 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
30214 info.nowrite calls with no lhs that can't throw. Return bool
30215 whether gsi_remove has been called or not.
30216 (pass_sprintf_length::handle_gimple_call): Return bool whether
30217 try_substitute_return_value called gsi_remove. Formatting fix.
30218 (pass_sprintf_length::execute): Don't use gsi_remove if
30219 handle_gimple_call returned true.
30222 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
30223 be removed due to side-effects, don't remove following barrier nor
30224 turn the successor edge into fallthru edge.
30226 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30229 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
30230 element-reversing loads and stores as not swappable.
30232 2017-01-12 Nathan Sidwell <nathan@acm.org>
30233 Nicolai Stange <nicstange@gmail.com>
30235 * combine.c (try_combine): Don't ignore result of overlap checking
30236 loop. Combine overlap & asm check into single loop.
30238 2017-01-12 Richard Biener <rguenther@suse.de>
30240 * tree-pretty-print.c (dump_generic_node): Provide -gimple
30241 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
30243 2017-01-12 Richard Biener <rguenther@suse.de>
30245 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
30246 and TS_TARGET_OPTION directly derive from TS_BASE.
30247 * tree-core.h (tree_optimization_option): Derive from tree_base.
30248 (tree_target_option): Likewise.
30250 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
30252 * config/i386/i386.c (memory_address_length): Increase len
30253 only when rip_relative_addr_p returns false.
30255 2017-01-11 Julia Koval <julia.koval@intel.com>
30257 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
30258 (OPTION_MASK_ISA_SGX_SET): New.
30259 (ix86_handle_option): Handle OPT_msgx.
30260 * config.gcc: Added sgxintrin.h.
30261 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
30262 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
30263 * config/i386/i386.c (ix86_target_string): Add -msgx.
30265 (ix86_option_override_internal): Handle new options.
30266 (ix86_valid_target_attribute_inner_p): Add sgx.
30267 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
30268 * config/i386/i386.opt: Add msgx.
30269 * config/i386/sgxintrin.h: New file.
30270 * config/i386/x86intrin.h: Add sgxintrin.h.
30272 2017-01-11 Jakub Jelinek <jakub@redhat.com>
30275 * fold-const.c (maybe_nonzero_address): Return 1 for function
30277 (tree_single_nonzero_warnv_p): Don't handle function local objects
30281 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
30284 2017-01-11 David Malcolm <dmalcolm@redhat.com>
30287 * opts.c: Include "spellcheck.h"
30288 (struct string_fragment): New struct.
30289 (struct edit_distance_traits<const string_fragment &>): New
30291 (get_closest_sanitizer_option): New function.
30292 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
30294 2017-01-11 Jakub Jelinek <jakub@redhat.com>
30296 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
30298 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
30299 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
30300 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
30301 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
30302 for initial die_offset if dwarf_split_debug_info.
30303 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
30304 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
30306 (output_skeleton_debug_sections): Formatting fix. Use
30307 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
30308 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
30310 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
30312 * config/arm/cortex-a53.md: Add bypasses for
30313 cortex_a53_r2f_cvt.
30314 (cortex_a53_r2f): Only use for transfers.
30315 (cortex_a53_f2r): Likewise.
30316 (cortex_a53_r2f_cvt): Add reservation for conversions.
30317 (cortex_a53_f2r_cvt): Likewise.
30319 2017-01-11 Tamar Christina <tamar.christina@arm.com>
30321 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
30322 to all inlined functions, change static to extern.
30324 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
30327 * config/arm/arm.c (legitimize_pic_address): Handle reference to
30329 (arm_assemble_integer): Likewise.
30331 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
30333 * config.gcc: Use new awk script to check CPU, FPU and architecture
30334 parameters for --with-... options.
30335 * config/arm/parsecpu.awk: New file
30336 * config/arm/arm-cpus.in: New file.
30337 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
30339 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
30341 * config/arm/t-arm: Update dependency rules.
30342 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
30343 of processing .def files.
30344 * config/arm/genopt.sh: Deleted.
30345 * config/arm/gentune.sh: Deleted.
30346 * config/arm/arm-cores.def: Deleted.
30347 * config/arm/arm-arches.def: Deleted.
30348 * config/arm/arm-fpus.def: Deleted.
30349 * config/arm/arm-tune.md: Regenerated.
30350 * config/arm/arm-tables.opt: Regenerated.
30351 * config/arm/arm-cpu.h: New generated file.
30352 * config/arm/arm-cpu-data.h: New generated file.
30353 * config/arm/arm-cpu-cdata.h: New generated file.
30355 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
30358 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
30360 (input_varpool_node): Unpack dynamically_initialized bit.
30362 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
30364 PR rtl-optimization/79032
30365 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
30366 the alignment of the adjusted memory reference against that of MODE,
30367 instead of the alignment of the original memory reference.
30369 2017-01-11 Martin Jambor <mjambor@suse.cz>
30371 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
30373 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
30374 decorated functions.
30376 2017-01-11 Richard Biener <rguenther@suse.de>
30378 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
30379 set range/nonnull info for PHI results. Do not set it on
30380 stmts marked for removal.
30382 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
30384 * expr.c (store_field): In the bitfield case, fetch the return value
30385 from the registers before applying a single big-endian adjustment.
30386 Always do a final load for a BLKmode value not larger than a word.
30388 2017-01-10 David Malcolm <dmalcolm@redhat.com>
30391 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
30392 that we correctly handle column numbers greater than
30393 LINE_MAP_MAX_COLUMN_NUMBER.
30395 2017-01-10 Martin Sebor <msebor@redhat.com>
30397 PR middle-end/78245
30398 * gimple-ssa-sprintf.c (get_destination_size): Call
30399 {init,fini}object_sizes.
30400 * tree-object-size.c (addr_object_size): Adjust.
30401 (pass_through_call): Adjust.
30402 (pass_object_sizes::execute): Adjust.
30403 * tree-object-size.h (fini_object_sizes): Declare.
30405 2017-01-10 Martin Sebor <msebor@redhat.com>
30407 PR tree-optimization/78775
30408 * builtins.c (get_size_range): Move...
30409 * calls.c: ...to here.
30410 (alloc_max_size): Accept zero argument.
30411 (operand_signed_p): Remove.
30412 (maybe_warn_alloc_args_overflow): Call get_size_range.
30413 * calls.h (get_size_range): Declare.
30415 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
30417 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
30418 from TI's devices.csv file as of September 2016.
30419 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
30421 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
30423 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
30424 * doc/invoke.texi: Likewise.
30425 * doc/md.texi: Likewise.
30426 * doc/objc.texi: Likewise.
30428 2017-01-10 Joshua Conner <joshconner@google.com>
30430 * config/arm/fuchsia-elf.h: New file.
30431 * config/fuchsia.h: New file.
30432 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
30433 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
30435 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
30437 2016-01-10 Richard Biener <rguenther@suse.de>
30439 PR tree-optimization/79034
30440 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
30441 Propagate out degenerate PHIs in the joiner.
30443 2017-01-10 Martin Liska <mliska@suse.cz>
30445 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
30446 (sort_congruence_classes_by_decl_uid): Likewise.
30447 (sort_congruence_class_groups_by_decl_uid): Likewise.
30448 (sem_item_optimizer::merge_classes): Sort class, groups in these
30449 classes and members in the groups by DECL_UID of declarations.
30450 This would make merge operations stable.
30452 2017-01-10 Martin Liska <mliska@suse.cz>
30454 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
30455 usage of m_classes_vec.
30456 (sem_item_optimizer::~sem_item_optimizer): Likewise.
30457 (sem_item_optimizer::get_group_by_hash): Likewise.
30458 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
30459 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
30460 (sem_item_optimizer::verify_classes): Likewise.
30461 (sem_item_optimizer::process_cong_reduction): Likewise.
30462 (sem_item_optimizer::dump_cong_classes): Likewise.
30463 (sem_item_optimizer::merge_classes): Likewise.
30464 * ipa-icf.h (congruence_class_hash): Rename from
30465 congruence_class_group_hash. Remove declaration of m_classes_vec.
30467 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
30469 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
30470 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
30471 * config.gcc: Add avx512vpopcntdqintrin.h.
30472 * config/i386/avx512vpopcntdqintrin.h: New.
30473 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
30474 * config/i386/i386-builtin-types.def: Add new types.
30475 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
30476 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
30477 __builtin_ia32_vpopcountq_v8di_mask): New.
30478 * config/i386/i386-c.c (ix86_target_macros_internal): Define
30479 __AVX512VPOPCNTDQ__.
30480 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
30481 (PTA_AVX512VPOPCNTDQ): Define.
30482 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
30483 TARGET_AVX512VPOPCNTDQ_P): Define.
30484 * config/i386/i386.opt: Add mavx512vpopcntdq.
30485 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
30486 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
30488 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
30490 PR middle-end/77484
30491 * predict.def (PRED_CALL): Set to 67.
30493 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
30495 * expr.c (store_field): In the bitfield case, if the value comes from
30496 a function call and is of an aggregate type returned in registers, do
30497 not modify the field mode; extract the value in all cases if the mode
30498 is BLKmode and the size is not larger than a word.
30500 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
30503 * config/i386/cpuid.h: Fix undefined behavior.
30505 2017-01-04 Jeff Law <law@redhat.com>
30507 PR tree-optimization/79007
30508 PR tree-optimization/67955
30509 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
30510 conservative for pt.null when flag_non_call_exceptions is on.
30512 2017-01-09 Jakub Jelinek <jakub@redhat.com>
30514 PR translation/79019
30515 PR translation/79020
30516 * params.def (PARAM_INLINE_MIN_SPEEDUP,
30517 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
30518 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
30520 * config/avr/avr.opt (maccumulate-args): Likewise.
30521 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
30522 * common.opt (freport-bug): Likewise.
30523 * cif-code.def (CIF_FINAL_ERROR): Likewise.
30524 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
30525 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
30526 translatable string.
30527 * config/i386/i386.c (function_value_32): Likewise.
30528 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
30529 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
30531 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
30532 * common/config/msp430/msp430-common.c (msp430_handle_option):
30534 * symtab.c (symtab_node::verify_base): Likewise.
30535 * opts.c (set_debug_level): Likewise.
30536 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
30537 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
30538 missing whitespace to translatable strings.
30539 * config/avr/avr.md (bswapsi2): Fix typo in comment.
30540 * config/sh/superh.h: Likewise.
30541 * config/i386/xopintrin.h: Likewise.
30542 * config/i386/znver1.md: Likewise.
30543 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
30544 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
30545 * double-int.h (struct double_int): Likewise.
30546 * double-int.c (div_and_round_double): Likewise.
30547 * wide-int.cc: Likewise.
30548 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
30549 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
30550 * cfgcleanup.c (crossjumps_occured): Renamed to ...
30551 (crossjumps_occurred): ... this.
30552 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
30555 PR tree-optimization/78899
30556 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
30557 returning bool return struct loop *, NULL for failure and the new
30559 (versionable_outer_loop_p): Don't version outer loop if it has
30560 dont_vectorized bit set.
30561 (tree_if_conversion): When versioning outer loop, ensure
30562 tree_if_conversion is performed also on the inner loop of the
30563 non-vectorizable outer loop copy.
30564 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
30565 LOOP_VECTORIZED in inner loop of the scalar outer loop and
30566 prevent vectorization of it.
30567 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
30568 the outer loop vectorization of the non-scalar version is attempted
30569 before vectorization of the inner loop in scalar version. If
30570 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
30571 vectorization of its inner loop.
30572 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
30573 has 2 inner loops, rename also on edges from bb whose single pred
30574 is outer_loop->header. Fix typo in function comment.
30576 2017-01-09 Martin Sebor <msebor@redhat.com>
30579 * asan.c (asan_emit_stack_protection): Increase local buffer size
30580 to avoid snprintf truncation warning.
30582 2017-01-09 Andrew Pinski <apinski@cavium.com>
30584 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
30585 to reference thunderx2t99 for the tuning structure
30586 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
30588 (thunderx2t99_extra_costs): This.
30589 * config/aarch64/aarch64-tune.md: Regenerate.
30590 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
30591 (vulcan_addrcost_table): This.
30592 (vulcan_regmove_cost): Rename to ...
30593 (thunderx2t99_regmove_cost): This.
30594 (vulcan_vector_cost): Rename to ...
30595 (thunderx2t99_vector_cost): this.
30596 (vulcan_branch_cost): Rename to ...
30597 (thunderx2t99_branch_cost): This.
30598 (vulcan_tunings): Rename to ...
30599 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
30600 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
30602 2017-01-09 Martin Jambor <mjambor@suse.cz>
30606 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
30607 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
30608 (propagate_vr_accross_jump_function): Use the above function for all
30609 value range computations for pass-through jump functions and type
30610 converasion from explicit value range values.
30611 (ipcp_propagate_stage): Do not attempt to deduce types of formal
30612 parameters from TYPE_ARG_TYPES.
30613 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
30614 (ipa_write_node_info): Stream type of the actual argument.
30615 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
30617 2017-01-09 Martin Liska <mliska@suse.cz>
30620 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
30621 (lookup_compiler): Do not show error message with have_E.
30623 2017-01-09 Jakub Jelinek <jakub@redhat.com>
30625 PR tree-optimization/78938
30626 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
30627 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
30628 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
30629 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
30632 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30634 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
30637 2017-01-09 Richard Biener <rguenther@suse.de>
30639 PR tree-optimization/78997
30640 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
30641 name condition properly.
30643 2017-01-09 Richard Biener <rguenther@suse.de>
30646 * dwarf2out.c (is_cxx): New overload with context.
30647 (is_naming_typedef_decl): Use it.
30649 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
30651 * invoke.texi (Option Summary): Correct spacing in option lists
30652 and add line breaks to fix over-long lines.
30654 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
30656 PR middle-end/17660
30658 * extend.texi (Common Variable Attributes): Add xref to GCC
30659 Internals manual to explain mode attribute keywords.
30661 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
30664 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
30665 and Preprocessor Options.
30666 (Options for Linking): Document -pthread here....
30667 (RS/6000 and PowerPC Options): ...not here.
30668 (Solaris 2 Options): ...or here.
30669 * doc/cppopts.texi: Document -pthread.
30671 2017-01-08 Martin Sebor <msebor@redhat.com>
30673 PR middle-end/77708
30674 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
30675 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
30676 New member functions.
30677 (format_directive): Used them.
30679 (pass_sprintf_length::handle_gimple_call): Same.
30680 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
30681 to avoid truncation for any argument.
30682 (extract_affine_mul): Same.
30683 * tree.c (get_file_function_name): Same.
30685 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
30687 PR middle-end/77484
30688 * predict.def (PRED_INDIR_CALL): Set to 86.
30690 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
30692 PR preprocessor/54124
30693 * doc/cppopts.texi: Reformat -d subtable to list the full name
30694 of the options. Add cross-reference to the docs for the general
30695 compiler -d options.
30696 * doc/invoke.texi (Developer Options): Add cross-reference to the
30697 preprocessor-specific -d option documentation.
30699 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
30701 PR preprocessor/13498
30702 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
30703 redudant material, and reflect new command-line options.
30704 (System Headers): Likewise.
30706 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
30708 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
30709 -isystem, and -idirafter. Copy-edit.
30710 * doc/cppopts.texi: Copy-edit. Remove contradiction about
30711 default for -ftrack-macro-expansion. Delete obsolete and
30712 badly-formatted implementation details about -fdebug-cpp output.
30713 * doc/cppwarnopts.texi: Copy-edit.
30715 2017-01-07 David Malcolm <dmalcolm@redhat.com>
30718 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
30719 that the transition from a max line width >= 1<<10 to narrower
30720 lines works correctly.
30722 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
30724 * doc/options.texi (PerFunction): New.
30725 * opt-functions.awk (switch_flags): Map both Optimization and
30726 PerFunction to CL_OPTIMIZATION.
30727 * opth-gen.awk: Test for PerFunction flag along with
30729 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
30730 it only when the latter is present. Skip those that don't in
30731 the hash function generator.
30732 * common.opt (fvar-tracking): Mark as PerFunction instead of
30734 (fvar-tracking-assignments): Likewise.
30735 (fvar-tracking-assignments-toggle): Likewise.
30736 (fvar-tracking-uninit): Likewise.
30738 2017-01-07 Jakub Jelinek <jakub@redhat.com>
30740 PR translation/79018
30741 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
30744 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
30747 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
30748 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
30749 TARGET_LONG_JUMP_TABLE_OFFSETS.
30750 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
30751 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
30752 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
30753 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
30754 * config/m68k/m68k.md (tablejump expander): Likewise.
30755 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
30756 TARGET_LONG_JUMP_TABLE_OFFSETS.
30757 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
30758 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
30760 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
30761 David Holsgrove <david.holsgrove@xilinx.com>
30763 * common/config/microblaze/microblaze-common.c
30764 (TARGET_EXCEPT_UNWIND_INFO): Remove.
30765 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
30767 * config/microblaze/microblaze.c (microblaze_must_save_register)
30768 (microblaze_expand_epilogue, microblaze_return_addr): Handle
30770 (microblaze_eh_return): New function.
30771 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
30772 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
30773 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
30774 * config/microblaze/microblaze.md (eh_return): New pattern.
30776 2017-01-06 Jakub Jelinek <jakub@redhat.com>
30778 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
30779 GCC_DIAGNOSTIC_STRINGIFY): Define.
30781 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
30783 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30785 * config/arm/arm.md (<mcrr>): New.
30787 * config/arm/arm.c (arm_arch5te): New.
30788 (arm_option_override): Set arm_arch5te.
30789 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
30791 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
30792 (arm_mcrr_qualifiers): ... this. New.
30793 (MRRC_QUALIFIERS): Define to...
30794 (arm_mrrc_qualifiers): ... this. New.
30795 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
30797 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
30798 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
30799 (MRRCI, mrrc, MRRC): New.
30800 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
30801 VUNSPEC_MRRC2): New.
30803 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30805 * config/arm/arm.md (<mcr>): New.
30807 * config/arm/arm.c (arm_coproc_builtin_available): Add
30808 support for mcr, mrc, mcr2 and mrc2.
30809 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
30810 (arm_mcr_qualifiers): ... this. New.
30811 (MRC_QUALIFIERS): Define to ...
30812 (arm_mrc_qualifiers): ... this. New.
30813 (MCR_QUALIFIERS): Define to ...
30814 (arm_mcr_qualifiers): ... this. New.
30815 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
30817 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
30818 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
30819 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
30820 VUNSPEC_MRC2): New.
30822 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30824 * config/arm/arm.md (*ldc): New.
30828 * config/arm/arm.c (arm_coproc_builtin_available): Add
30829 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
30830 (arm_coproc_ldc_stc_legitimate_address): New.
30831 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
30832 'qualifier_const_pointer'.
30833 (LDC_QUALIFIERS): Define to...
30834 (arm_ldc_qualifiers): ... this. New.
30835 (STC_QUALIFIERS): Define to...
30836 (arm_stc_qualifiers): ... this. New.
30837 * config/arm/arm-protos.h
30838 (arm_coproc_ldc_stc_legitimate_address): New.
30839 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
30840 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
30841 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
30842 stc2, stcl, stc2l): New.
30843 * config/arm/constraints.md (Uz): New.
30844 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
30845 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
30846 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
30847 VUNSPEC_STC2L): New.
30849 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30851 * config/arm/arm.md (<cdp>): New.
30852 * config/arm/arm.c (neon_const_bounds): Rename this ...
30853 (arm_const_bounds): ... this.
30854 (arm_coproc_builtin_available): New.
30855 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
30856 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
30857 (CDP_QUALIFIERS): Define to...
30858 (arm_cdp_qualifiers): ... this. New.
30860 (arm_expand_builtin_args): Add case for 6 arguments.
30861 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
30862 (arm_const_bounds): ... this.
30863 (arm_coproc_builtin_available): New.
30864 * config/arm/arm_acle.h (__arm_cdp): New.
30866 * config/arm/arm_acle_builtins.def (cdp): New.
30868 * config/arm/iterators.md (CDPI,CDP,cdp): New.
30869 * config/arm/neon.md: Rename all 'neon_const_bounds' to
30870 'arm_const_bounds'.
30871 * config/arm/types.md (coproc): New.
30872 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
30873 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
30874 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
30875 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
30877 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30879 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
30880 (UBINOP_QUALIFIERS): New.
30882 (acle_builtin_data): New. Change comment.
30883 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
30884 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
30885 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
30886 arm_acle_builtins.def.
30887 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
30888 (arm_init_acle_builtins): New.
30889 (CRC32_BUILTIN): Remove.
30890 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
30891 crc32cb, crc32ch and crc32cw.
30892 (arm_init_crc32_builtins): Remove.
30893 (arm_init_builtins): Use arm_init_acle_builtins rather
30894 than arm_init_crc32_builtins.
30895 (arm_expand_acle_builtin): New.
30896 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
30897 * config/arm/arm_acle_builtins.def: New.
30899 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
30901 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
30902 (arm_builtin_datum): ... this.
30903 (arm_init_neon_builtin): Rename to ...
30904 (arm_init_builtin): ... this. Add a new parameters PREFIX
30905 and USE_SIG_IN_NAME.
30906 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
30907 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
30908 'arm_builtin_datum'.
30909 (arm_init_vfp_builtins): Likewise.
30910 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
30911 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
30912 (arm_expand_neon_args): Rename to ...
30913 (arm_expand_builtin_args): ... this. Rename builtin_arg
30914 enum values and differentiate between ARG_BUILTIN_MEMORY
30915 and ARG_BUILTIN_NEON_MEMORY.
30916 (arm_expand_neon_builtin_1): Rename to ...
30917 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
30918 values, arm_expand_builtin_args and add bool parameter NEON.
30919 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
30920 (arm_expand_vfp_builtin): Likewise.
30921 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
30923 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
30925 PR middle-end/77484
30926 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
30927 * predict.c (tree_estimate_probability_bb): Reverse direction of
30928 polymorphic call predictor.
30930 2017-01-06 David Malcolm <dmalcolm@redhat.com>
30932 * passes.c (execute_one_pass): Split out pass-skipping logic into...
30933 (determine_pass_name_match): ...this new function and...
30934 (should_skip_pass_p): ...this new function.
30936 2017-01-06 Nathan Sidwell <nathan@acm.org>
30938 * ipa-visibility.c (function_and_variable_visibility): Reformat
30939 comments and long lines. Remove extrneous if.
30940 * symtab.c (symtab_node::make_decl_local): Fix code format.
30941 (symtab_node::set_section_for_node): Fix comment typo.
30943 2017-01-06 Martin Liska <mliska@suse.cz>
30946 * lra-constraints.c: Rename invariant to lra_invariant.
30947 * predict.c (set_even_probabilities): Initialize e to NULL.
30949 2017-01-05 Martin Sebor <msebor@redhat.com>
30951 PR tree-optimization/78910
30952 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
30953 (format_integer): Correct off-by-one error in the handling
30954 of precision with negative numbers in signed conversions..
30956 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
30958 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
30960 2017-01-05 Jakub Jelinek <jakub@redhat.com>
30962 PR tree-optimization/71016
30963 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
30964 factor_out_conditional_conversion. Formatting fix.
30965 (factor_out_conditional_conversion): Add cond_stmt argument.
30966 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
30967 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
30970 2017-01-05 David Malcolm <dmalcolm@redhat.com>
30972 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
30973 read-rtl-function.o, and selftest-rtl.o.
30974 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
30975 (selftest::aarch64_test_loading_full_dump): New function.
30976 (selftest::aarch64_run_selftests): New function.
30977 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
30978 selftest::aarch64_run_selftests.
30979 * config/i386/i386.c
30980 (selftest::ix86_test_loading_dump_fragment_1): New function.
30981 (selftest::ix86_test_loading_call_insn): New function.
30982 (selftest::ix86_test_loading_full_dump): New function.
30983 (selftest::ix86_test_loading_unspec): New function.
30984 (selftest::ix86_run_selftests): Call the new functions.
30985 * emit-rtl.c (maybe_set_max_label_num): New function.
30986 * emit-rtl.h (maybe_set_max_label_num): New decl.
30987 * function.c (instantiate_decls): Guard call to
30988 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
30989 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
30991 * gensupport.c (gen_reader::gen_reader): Pass "false"
30992 for new "compact" param of rtx_reader.
30993 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
30994 rather than an empty string for NULL strings.
30995 * read-md.c: Potentially include config.h rather than bconfig.h.
30996 Wrap include of errors.h with #ifdef GENERATOR_FILE.
30997 (have_error): New global, copied from errors.c.
30998 (md_reader::read_name): Rename to...
30999 (md_reader::read_name_1): ...this, adding "out_loc" param,
31000 and converting "missing name or number" to returning false, rather
31002 (md_reader::read_name): Reimplement in terms of read_name_1.
31003 (md_reader::read_name_or_nil): New function.
31004 (md_reader::read_string): Handle "(nil)" by returning NULL.
31005 (md_reader::md_reader): Add new param "compact".
31006 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
31007 (md_reader::read_file): New method.
31008 * read-md.h (md_reader::md_reader): Add new param "compact".
31009 (md_reader::read_file): New method.
31010 (md_reader::is_compact): New accessor.
31011 (md_reader::read_name): Convert return type from void to file_location.
31012 (md_reader::read_name_or_nil): New decl.
31013 (md_reader::read_name_1): New decl.
31014 (md_reader::m_compact): New field.
31015 (noop_reader::noop_reader): Pass "false" for new "compact" param
31017 (rtx_reader::rtx_reader): Add new "compact" param.
31018 (rtx_reader::read_rtx_operand): Make virtual and convert return
31019 type from void to rtx.
31020 (rtx_reader::read_until): New decl.
31021 (rtx_reader::handle_any_trailing_information): New virtual function.
31022 (rtx_reader::postprocess): New virtual function.
31023 (rtx_reader::finalize_string): New virtual function.
31024 (rtx_reader::m_in_call_function_usage): New field.
31025 (rtx_reader::m_reuse_rtx_by_id): New field.
31026 * read-rtl-function.c: New file.
31027 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
31028 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
31029 (selftest::verify_three_block_rtl_cfg): New decl.
31030 * read-rtl-function.h: New file.
31031 * read-rtl.c: Potentially include config.h rather than bconfig.h.
31032 For host, include function.h, memmodel.h, and emit-rtl.h.
31033 (one_time_initialization): New function.
31034 (struct compact_insn_name): New struct.
31035 (compact_insn_names): New array.
31036 (find_code): Handle insn codes in compact dumps.
31037 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
31038 (bind_subst_iter_and_attr): Likewise.
31039 (add_condition_to_string): Likewise.
31040 (add_condition_to_rtx): Likewise.
31041 (apply_attribute_uses): Likewise.
31042 (add_current_iterators): Likewise.
31043 (apply_iterators): Likewise.
31044 (initialize_iterators): Guard usage of apply_subst_iterator with
31045 #ifdef GENERATOR_FILE.
31046 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
31047 (md_reader::read_mapping): Likewise.
31048 (add_define_attr_for_define_subst): Likewise.
31049 (add_define_subst_attr): Likewise.
31050 (read_subst_mapping): Likewise.
31051 (check_code_iterator): Likewise.
31052 (rtx_reader::read_rtx): Likewise. Move one-time initialization
31054 (one_time_initialization): New function.
31055 (rtx_reader::read_until): New method.
31056 (read_flags): New function.
31057 (parse_reg_note_name): New function.
31058 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
31059 Handle reuse_rtx ids.
31060 Wrap iterator lookup within #ifdef GENERATOR_FILE.
31061 Add parsing support for RTL dumps, mirroring the special-cases in
31062 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
31063 values, and calling handle_any_trailing_information.
31064 (rtx_reader::read_rtx_operand): Convert return type from void
31065 to rtx, returning return_rtx. Handle case 'e'. Call
31066 finalize_string on XSTR and XTMPL fields.
31067 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
31068 "(nil)" values were omitted. Call the postprocess vfunc on the
31070 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
31071 class ctor. Initialize m_in_call_function_usage. Call
31072 one_time_initialization.
31073 * rtl-tests.c (selftest::test_uncond_jump): Call
31074 set_new_first_and_last_insn.
31075 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
31076 * selftest-rtl.c: New file.
31077 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
31078 (selftest::get_insn_by_uid): New decl.
31079 * selftest-run-tests.c (selftest::run_tests): Call
31080 read_rtl_function_c_tests.
31081 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
31082 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
31085 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
31087 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
31088 operands in a special way. Assert that pos+len <= mode precision.
31090 2017-01-05 Jakub Jelinek <jakub@redhat.com>
31092 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
31093 3 argument Alias with unlimited for the negative form.
31094 (fno-vect-cost-model): Removed.
31096 2017-01-05 Martin Liska <mliska@suse.cz>
31098 * hsa-gen.c (gen_hsa_divmod): New function.
31099 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
31101 2017-01-05 Martin Liska <mliska@suse.cz>
31104 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
31107 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31109 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
31110 small constant length operands.
31112 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31114 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
31115 between loop iterations.
31117 2017-01-05 Martin Liska <mliska@suse.cz>
31120 * gimplify.c (gimplify_decl_expr): Compare to
31121 asan_poisoned_variables instread of checking flags.
31122 (gimplify_target_expr): Likewise.
31123 (gimplify_expr): Likewise.
31124 (gimplify_function_tree): Conditionally initialize
31125 asan_poisoned_variables.
31127 2017-01-04 Jeff Law <law@redhat.com>
31129 PR tree-optimizatin/78812
31130 * rtl.h (contains_mem_rtx_p): Prototype.
31131 * ifcvt.c (containts_mem_rtx_p): Move from here to...
31132 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
31133 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
31134 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
31135 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
31137 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
31139 * input.c (assert_char_at_range): Default-initialize actual_range.
31141 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
31143 * df-scan.c (df_ref_create_structure): Make regno unsigned,
31144 to match the caller.
31146 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
31148 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
31149 insns after final jump in test to emit dummy move.
31151 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
31153 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
31154 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
31156 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
31158 * multiple_target.c (create_dispatcher_calls): Init e_next.
31159 * tree-ssa-loop-split.c (split_loop): Init border.
31160 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
31163 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
31168 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
31169 (altivec_register_operand): Do not return true if the operand
31170 contains a SUBREG mixing SImode and SFmode.
31171 (vsx_register_operand): Likewise.
31172 (vsx_reg_sfsubreg_ok): New predicate.
31173 (vfloat_operand): Do not return true if the operand contains a
31174 SUBREG mixing SImode and SFmode.
31175 (vint_operand): Likewise.
31176 (vlogical_operand): Likewise.
31177 (gpc_reg_operand): Likewise.
31178 (int_reg_operand): Likewise.
31179 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
31180 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
31181 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
31183 (rs6000_emit_move_si_sf_subreg): New helper function.
31184 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
31185 fixup SUBREGs involving SImode and SFmode.
31186 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
31187 numbers for the new peephole2 optimization.
31188 (peephole2 for SFmode unions): New peephole2 to optimize cases in
31189 the GLIBC math library that do AND/IOR/XOR operations on single
31190 precision floating point.
31191 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
31192 target macros to say whether we need to avoid SUBREGs mixing
31194 (TARGET_ALLOW_SF_SUBREG): Likewise.
31195 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
31196 (UNSPEC_SI_FROM_SF): Likewise.
31197 (iorxor): Change spacing.
31198 (and_ior_xor): New iterator for AND, IOR, and XOR.
31199 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
31200 (movdi_from_sf_zero_ext): Likewise.
31201 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
31202 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
31203 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
31204 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
31205 (fms<mode>4): Likewise.
31206 (fnma<mode>4): Likewise.
31207 (fnms<mode>4): Likewise.
31208 (nfma<mode>4): Likewise.
31209 (nfms<mode>4): Likewise.
31211 2017-01-04 Marek Polacek <polacek@redhat.com>
31214 * doc/invoke.texi: Document -Wpointer-compare.
31216 2017-01-04 Jakub Jelinek <jakub@redhat.com>
31218 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
31221 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
31222 descriptions for -gdwarf-5 and emit them as uleb128 instead of
31225 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
31228 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
31229 documentation of the powerpc_popcntb_ok attribute.
31230 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
31231 code to issue warning messages if a requested CPU configuration is
31232 not supported by the binary (assembler and loader) toolchain.
31233 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
31234 made to define a built-in function that has been disabled.
31235 (paired_init_builtins): Add assertion to prevent ICE if attempt is
31236 made to define a built-in function that has been disabled.
31237 (altivec_init_builtins): Add comment explaining why definition
31238 of the DST built-in functions is not preceded by an assertion
31239 check. Add assertions to prevent ICE if attempts are made to
31240 define an altivec predicate or an abs* built-in function that has
31242 (htm_init_builtins): Add comment explaining why definition of the
31243 htm built-in functions is not preceded by an assertion check.
31245 2017-01-04 Jeff Law <law@redhat.com>
31247 PR tree-optimizatin/67955
31248 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
31249 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
31250 the points-to solution does not include pt_null. Use DECL_PT_UID
31253 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
31255 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
31256 Use gen_int_mode instead of gen_lopwart for const_int operands.
31258 2017-01-04 Jakub Jelinek <jakub@redhat.com>
31260 PR tree-optimization/71563
31261 * match.pd: Simplify X << Y into X if Y is known to be 0 or
31262 out of range value - has low bits known to be zero.
31264 2017-01-04 Alan Modra <amodra@gmail.com>
31266 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
31267 * configure: Regenerate.
31268 * config.in: Regenerate.
31270 2017-01-04 Jakub Jelinek <jakub@redhat.com>
31273 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
31274 a substring of the message, but strcmp with the whole message. Ifdef
31275 ENABLE_NLS, translate the message first using dgettext.
31277 2017-01-03 Jeff Law <law@redhat.com>
31279 PR tree-optimizatin/78856
31280 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
31281 (mark_threaded_blocks): Remove code to truncate thread paths that
31282 cross multiple loop headers. Instead invalidate the cached loop
31283 iteration information and handle case of a thread path walking
31284 into an irreducible region.
31286 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
31289 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
31290 assertions. Add support for doing the signbit if the IEEE 128-bit
31291 floating point value is in a GPR.
31292 * config/rs6000/rs6000.md (Fsignbit): Delete.
31293 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
31294 Update the length attribute if the value is in a GPR.
31295 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
31296 the sign or zero extension instruction, since the value is always 0/1.
31297 (signbit<mode>2_dm2): Delete using <Fsignbit>.
31300 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
31301 extracting SImode to a GPR register so that we can generate a
31302 store, limit the vector to be in a traditional Altivec register
31303 for the vextuwrx instruction.
31305 2017-01-03 Ian Lance Taylor <iant@google.com>
31307 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
31309 2017-01-03 Martin Sebor <msebor@redhat.com>
31311 PR tree-optimization/78696
31312 * gimple-ssa-sprintf.c (format_floating): Correct handling of
31313 precision. Use MPFR for %f for greater fidelity. Correct handling
31315 (pass_sprintf_length::compute_format_length): Set width and precision
31316 specified by asrerisk to void_node for vararg functions.
31317 (try_substitute_return_value): Adjust dump output.
31319 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
31321 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
31323 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
31325 * doc/invoke.texi (SPARC options): Document -mlra as the default.
31326 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
31327 -mlra/-mno-lra was passed to the compiler.
31329 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
31331 PR rtl-optimization/65618
31332 * emit-rtl.c (try_split): Move initialization of "before" and
31333 "after" to just before the call to emit_insn_after_setloc.
31335 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
31337 * doc/md.texi (Standard Names): Remove reference to Java frontend.
31339 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
31341 * dwarf2out.c (gen_enumeration_type_die): When
31342 -gno-strict-dwarf, add a DW_AT_encoding attribute.
31344 2017-01-03 Jakub Jelinek <jakub@redhat.com>
31346 PR tree-optimization/78965
31347 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
31348 Change first argument from const call_info & to call_info &. For %n
31349 set info.nowrite to false.
31351 PR middle-end/78901
31352 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
31353 possibly throwing calls.
31355 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
31356 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
31357 and fns handling, rather than in a separate case SSA_NAME.
31359 2017-01-02 Jeff Law <law@redhat.com>
31361 * config/darwin-driver.c (darwin_driver_init): Const-correctness
31362 fixes for first_period and second_period variables.
31364 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
31367 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
31368 (*insvqi_1): New insn pattern.
31369 (*insvqi_1_mem_rex64): Ditto.
31370 (*insvqi_2): Ditto.
31371 (*insvqi_3): Rename from *insvqi.
31373 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
31375 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
31377 * doc/cfg.texi (Edges): Remove reference to Java.
31378 (Maintaining the CFG): Ditto.
31380 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
31382 PR middle-end/77674
31383 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
31384 transparent aliases.
31386 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
31388 PR middle-end/77484
31389 * predict.def (PRED_CALL): Update hitrate.
31390 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
31391 * predict.c (tree_estimate_probability_bb): Split CALL predictor
31392 into direct/indirect/polymorphic variants.
31394 2017-01-01 Jakub Jelinek <jakub@redhat.com>
31396 Update copyright years.
31398 * gcc.c (process_command): Update copyright notice dates.
31399 * gcov-dump.c (print_version): Ditto.
31400 * gcov.c (print_version): Ditto.
31401 * gcov-tool.c (print_version): Ditto.
31402 * gengtype.c (create_file): Ditto.
31403 * doc/cpp.texi: Bump @copying's copyright year.
31404 * doc/cppinternals.texi: Ditto.
31405 * doc/gcc.texi: Ditto.
31406 * doc/gccint.texi: Ditto.
31407 * doc/gcov.texi: Ditto.
31408 * doc/install.texi: Ditto.
31409 * doc/invoke.texi: Ditto.
31411 Copyright (C) 2017 Free Software Foundation, Inc.
31413 Copying and distribution of this file, with or without modification,
31414 are permitted in any medium without royalty provided the copyright
31415 notice and this notice are preserved.