1 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
4 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
5 to consider in curr_insn_transform.
7 2017-03-24 Jakub Jelinek <jakub@redhat.com>
9 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
10 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
13 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
16 argument to the overloaded builtin variants. Use the new flag to
17 deprecate certain builtin variants.
18 * config/s390/s390-builtin-types.def: Add new builtin types.
19 * config/s390/s390-builtins.h: Support new flags field for
21 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
22 (s390_macro_to_expand): Enable vector float data type.
23 (s390_cpu_cpp_builtins_internal): Indicate support of the new
24 builtins by incrementing the __VEC__ version number.
25 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
27 (s390_resolve_overloaded_builtin): Emit error messages depending
29 * config/s390/s390.c (s390_expand_builtin): Support additional
30 flags argument. Change error message to match the messages
32 * config/s390/s390.md: New UNSPEC_* constants.
33 (op_type): Add new instruction types.
34 * config/s390/vecintrin.h: Add new builtins and test data class
36 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
37 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
38 (VEC_INEXACT, VEC_NOINEXACT): New constants.
39 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
40 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
41 ("vec_mergel<mode>"): V_HW -> VEC_HW.
43 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
44 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
45 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
46 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
48 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
49 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
50 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
51 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
53 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
54 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
55 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
56 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
57 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
58 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
59 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
61 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
62 ("vec_scatter_element<V_HW_4:mode>_DI")
63 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
64 ("vec_fpint<mode>", "vflls")
65 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
66 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
67 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
68 ("*vec_cmphe<mode>_cc"): ... these.
70 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
71 mode constant instead of magic value.
73 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
75 * config/s390/s390.c (s390_expand_vec_compare): Support other
76 vector floating point modes than just V2DF.
77 (s390_expand_vcond): Likewise.
78 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
79 (s390_cannot_change_mode_class): Prevent mode changes between TF
80 and V1TF in vector registers.
81 * config/s390/s390.md (DF, SF): New mode attributes.
82 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
83 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
84 SFmode support for VRs.
85 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
87 (VFT, VF_HW): New mode iterators.
88 (vw, sdx): New mode attributes.
89 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
90 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
91 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
92 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
93 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
94 also the new vector floating point modes. Renaming to ...
96 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
97 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
98 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
99 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
100 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
101 ("vec_unordered<mode>"): ... these.
103 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
104 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
105 ("*vec_extendv2df"): New insn definitions.
107 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
109 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
110 ("mulditi3_2", "*muldi3_sign"): New patterns.
111 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
112 rename the pattern definition.
114 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
116 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
118 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
120 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
122 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
123 instruction if possible.
124 * config/s390/vector.md (vec_halfnumelts): New mode
126 ("*vec_vllezlf<mode>"): New pattern.
128 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
130 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
131 ("popcountv4si2", "popcountv2di2"): Rename to ...
132 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
133 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
135 ("popcount<mode>2_vxe"): New pattern.
137 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
139 * common/config/s390/s390-common.c (processor_flags_table): Add
141 * config.gcc: Add arch12.
142 * config/s390/driver-native.c (s390_host_detect_local_cpu):
143 Default to arch12 for unknown CPU model numbers.
144 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
145 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
146 PROCESSOR_max sanity check.
147 * config/s390/s390-opts.h (enum processor_type): Add
149 * config/s390/s390.c (processor_table): Add arch12.
150 (s390_expand_builtin): Add check for B_VXE flag.
151 (s390_issue_rate): Add PROCESSOR_ARCH12.
152 (s390_get_sched_attrmask): Likewise.
153 (s390_get_unit_mask): Likewise.
154 (s390_sched_score): Enable z13 scheduling for arch12.
155 (s390_sched_reorder): Likewise.
156 (s390_sched_variable_issue): Likewise.
157 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
159 (s390_tune_attr): Use z13 scheduling also for arch12.
160 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
161 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
162 (TARGET_VXE_P): New macros.
163 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
164 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
165 * config/s390/s390.opt: Add arch12 as processor_type.
167 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
169 * config/s390/s390.md
170 ("fixuns_truncdddi2", "fixuns_trunctddi2")
171 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
172 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
174 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
175 Rename expanders to ...
177 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
178 ("fixuns_truncdddi2_emu"): ... these.
180 ("fixuns_trunc<mode>si2_emu"): New expander.
182 ("*fixuns_truncdfdi2_z13"): Rename to ...
183 ("*fixuns_truncdfdi2_vx"): ... this.
185 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
187 * config/s390/2964.md: Remove the single element vector compare
188 instructions which are no longer used.
189 * config/s390/s390.c (s390_select_ccmode): Remove handling of
191 (s390_canonicalize_comparison): Remove handling of DFmode
193 (s390_expand_vec_compare_scalar): Remove function.
194 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
195 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
197 ("*cmp<mode>_ccs"): Add wfcdb instruction.
199 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
201 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
203 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
204 will anyway by matched by mov<mode>_64dfp.
206 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
208 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
209 vlef/vstef. Add missing operand to vleif.
211 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
213 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
214 pair for all vector types with 64 bit elements.
215 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
216 * config/s390/vector.md (V_HW_64): ... here.
217 (V_128_NOSINGLE): New mode iterator.
218 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
219 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
220 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
221 ("*vec_load_pairv2di"): Change to ...
222 ("*vec_load_pair<mode>"): ... this one.
224 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
226 * config/s390/constraints.md: Add comments.
227 (jKK): Reject element sizes > 8 bytes.
228 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
230 * config/s390/s390.md: Add the s_operand checks formerly in
231 s390_split_ok_p to various splitters where they are still
233 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
234 for 128 bit vectors. Plus two splitters.
236 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
238 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
241 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
244 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
245 error if the boundary argument is not constant.
247 2017-03-24 Jakub Jelinek <jakub@redhat.com>
249 PR rtl-optimization/80112
250 * loop-doloop.c (doloop_condition_get): Don't check condition
251 if cmp isn't SET with IF_THEN_ELSE src.
253 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
255 PR tree-optimization/80158
256 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
257 replacing a candidate statement, also replace it for the
258 candidate's alternate interpretation.
259 (replace_rhs_if_not_dup): Likewise.
260 (replace_one_candidate): Likewise.
262 2017-03-24 Richard Biener <rguenther@suse.de>
264 PR tree-optimization/80167
265 * graphite-isl-ast-to-gimple.c
266 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
268 (translate_isl_ast_to_gimple::get_rename): Likewise.
270 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
272 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
273 handling of certain combinations of target options, including the
274 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
275 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
277 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
280 * config/arm/arm.md (*load_multiple): Add reload_completed to
283 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
284 Richard Biener <rguenth@suse.com>
286 PR tree-optimization/79908
287 PR tree-optimization/80136
288 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
289 been cast away, gimplify_and_add suffices.
291 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
293 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
295 2017-03-23 Richard Biener <rguenther@suse.de>
297 PR tree-optimization/80032
298 * gimplify.c (gimple_push_cleanup): Forced unconditional
299 cleanups still have to go to the conditional_cleanups
302 2017-03-22 Jakub Jelinek <jakub@redhat.com>
304 PR tree-optimization/80072
305 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
307 (next_operand_entry_id): Change type to unsigned int.
308 (sort_by_operand_rank): Make sure to return the right return value
309 even if unsigned fields are bigger than INT_MAX.
310 (struct oecount): Change cnt and id type to unsigned int.
311 (oecount_hasher::equal): Formatting fix.
312 (oecount_cmp): Make sure to return the right return value
313 even if unsigned fields are bigger than INT_MAX.
314 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
317 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
318 TREE_READONLY on result if writing it more than once.
321 * doc/invoke.texi (-fsanitize=thread): Document that with
322 -fnon-call-exceptions atomics are not able to throw
326 * tsan.c: Include tree-eh.h.
327 (instrument_builtin_call): Call maybe_clean_eh_stmt or
328 maybe_clean_or_replace_eh_stmt where needed.
329 (instrument_memory_accesses): Add cfg_changed argument.
330 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
332 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
334 PR rtl-optimization/63191
335 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
336 wrapper function, moved the whole old content into ...
337 (ix86_delegitimize_address_1): ... this. New inline function.
338 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
339 true as last argument instead of ix86_delegitimize_address.
341 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
343 * config/aarch64/aarch64.c (generic_branch_cost): Copy
344 cortexa57_branch_cost.
346 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
348 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
350 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
353 * doc/md.texi (Constraints): Document wA constraint.
354 * config/rs6000/constraints.md (wA): New.
355 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
356 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
357 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
358 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
360 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
363 * gimplify.c (is_oacc_declared): New function.
364 (oacc_default_clause): Use it to set default flags for acc declared
365 variables inside parallel regions.
366 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
368 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
369 declare attribute to any decl as necessary.
371 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
374 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
375 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
376 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
377 (arm_arch_lpae): This.
378 * config/arm/arm.c (arm_arch7ve): Rename into ...
379 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
380 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
383 2017-03-22 Martin Liska <mliska@suse.cz>
386 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
387 error message instead of an ICE.
389 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
391 * doc/extend.texi (6.11 Additional Floating Types): Revise.
393 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
395 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
397 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
400 2017-03-21 Martin Sebor <msebor@redhat.com>
402 * doc/extend.texi: Use "cannot" instead of "can't."
403 * doc/hostconfig.texi: Same.
404 * doc/install.texi: Same.
405 * doc/invoke.texi: Same.
406 * doc/loop.texi: Same.
408 * doc/objc.texi: Same.
409 * doc/rtl.texi: Same.
411 * doc/tm.texi.in: Same.
412 * doc/trouble.texi: Same.
414 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
417 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
418 (collect_checksum_attributes): Set it.
419 (die_checksum_ordered): Use it.
421 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
423 PR tree-optimization/79908
424 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
425 change: For a VA_ARG whose LHS has been cast away, use
426 force_gimple_operand to construct the side effects.
428 2017-03-21 David Malcolm <dmalcolm@redhat.com>
431 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
432 more amenable to translation.
433 (oacc_loop_auto_partitions): Likewise.
435 2017-03-21 Marek Polacek <polacek@redhat.com>
436 Martin Sebor <msebor@redhat.com>
438 PR tree-optimization/80109
439 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
442 2017-03-21 Jakub Jelinek <jakub@redhat.com>
443 Segher Boessenkool <segher@kernel.crashing.org>
446 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
447 check reg_used_between_p between insn and one of succ or succ2
448 depending on if succ is artificial insn not inserted into insn
451 2017-03-21 Martin Liska <mliska@suse.cz>
453 PR gcov-profile/80081
454 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
455 * doc/gcc.texi: Include gcov-dump stuff.
456 * doc/gcov-dump.texi: New file.
458 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
460 PR rtl-optimization/79150
461 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
462 conditional jump, if the jump is the last insn of the loop.
464 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
465 Richard Biener <rguenth@suse.com>
467 PR tree-optimization/79908
468 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
469 been cast away, use force_gimple_operand to construct the side
472 2017-03-21 Martin Liska <mliska@suse.cz>
475 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
478 2017-03-21 Brad Spengler <spender@grsecurity.net>
481 * plugin.c (htab_hash_plugin): New function.
482 (add_new_plugin): Use it and adjust.
483 (parse_plugin_arg_opt): Adjust.
484 (init_one_plugin): Likewise.
486 2017-03-21 Richard Biener <rguenther@suse.de>
488 PR tree-optimization/80032
489 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
490 if set force the cleanup to happen unconditionally.
491 (gimplify_target_expr): Push inserted clobbers with force_uncond
492 to avoid them being removed by control-dependent DCE.
494 2017-03-21 Richard Biener <rguenther@suse.de>
496 PR tree-optimization/80122
497 * tree-inline.c (copy_bb): Do not expans va-arg packs or
498 va_arg_pack_len when the inlined call stmt requires pack
500 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
502 2017-03-21 Jakub Jelinek <jakub@redhat.com>
505 * tsan.c (instrument_builtin_call): If the memory model argument
506 is not a constant, assume it is valid.
509 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
512 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
514 PR rtl-optimization/79910
515 * combine.c (can_combine_p): Do not allow combining an I0 or I1
516 if its dest is used by an insn before I2 (other than the combined
517 insns themselves, which are properly handled already).
519 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
522 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
524 * combine.c (record_used_regs): New static function.
525 (try_combine): Handle situations where there is an additional
526 instruction between I2 and I3 which needs to have a LOG_LINK
530 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
532 * combine.c (try_combine): Delete redundant i1 test. Call
533 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
535 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
538 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
541 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
543 PR tree-optimization/80054
544 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
545 the optimization if a PHI or any of its arguments is not dominated
546 by the candidate's basis. Use gphi* rather than gimple* as
548 (replace_profitable_candidates): Clean up a gimple* variable that
549 should be a gphi* variable.
551 2017-03-20 Martin Sebor <msebor@redhat.com>
554 * doc/extend.texi (attribute constructor): Document present limitation.
556 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
559 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
560 __POWER9_VECTOR__ #ifdef control, change template definition to
561 use Power9-specific built-in function.
562 (vec_any_eq): Likewise.
563 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
564 to control outcomes from this test.
565 (vector_ae_<mode>p): For VEC_F modes, likewise.
567 2017-03-20 Ian Lance Taylor <iant@google.com>
569 * config/i386/i386.c (ix86_function_regparm): Save an extra
570 register for -fsplit-stack with DECL_STATIC_CHAIN.
572 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
575 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
576 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
578 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
580 * config/riscv/riscv.c (riscv_print_operand): Use "fence
582 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
585 2017-03-20 Marek Polacek <polacek@redhat.com>
588 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
590 2017-03-20 Richard Biener <rguenther@suse.de>
592 PR tree-optimization/80113
593 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
594 allocate extra SSA name for PHI def.
595 (add_close_phis_to_outer_loops): Likewise.
596 (add_close_phis_to_merge_points): Likewise.
597 (copy_loop_close_phi_args): Likewise.
598 (copy_cond_phi_nodes): Likewise.
600 2017-03-20 Martin Liska <mliska@suse.cz>
603 * tree-chkp.c (chkp_build_returned_bound): Do not build
604 returned bounds for a LHS that's not a BOUNDED_P type.
606 2017-03-20 Martin Liska <mliska@suse.cz>
610 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
611 COMPLEX_CST and VECTOR_CST.
613 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
616 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
617 target operand. A new splitter adds the clobber statement in case
618 the target operand is dead anyway.
620 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
622 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
623 to age-old versions of binutils and glibc.
625 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
627 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
629 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
631 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
633 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
635 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
636 requirement for binutils 2.13.
638 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
640 * combine.c (try_combine): Delete redundant i1 test. Call
641 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
643 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
645 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
646 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
648 <riscv64-*-elf>: Re-arrange section
649 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
650 <riscv32-*-linux>: Likewise.
651 <riscv64-*-elf>: Likewise
652 <riscv64-*-linux>: Likewise.
654 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
657 * aarch64.opt(verbose-cost-dump): Fix typo.
659 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
662 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
663 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
665 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
667 * reload.c (find_reloads): When reloading a nonoffsettable address,
668 use RELOAD_OTHER for it and its address reloads.
670 PR rtl-optimization/79910
671 * combine.c (record_used_regs): New static function.
672 (try_combine): Handle situations where there is an additional
673 instruction between I2 and I3 which needs to have a LOG_LINK
676 2017-03-17 Jeff Law <law@redhat.com>
678 PR tree-optimization/71437
679 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
680 conditional in the hash table first.
681 (vrp_dom_walker::before_dom_children): Extract condition from
682 ASSERT_EXPR. Record condition, its inverion and any implied
685 2017-03-17 Marek Polacek <polacek@redhat.com>
686 Markus Trippelsdorf <markus@trippelsdorf.de>
688 PR tree-optimization/80079
689 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
692 2017-03-17 Richard Biener <rguenther@suse.de>
695 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
696 Properly verify the LHS before the RHS possibly claims to be
698 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
701 2017-03-17 Martin Jambor <mjambor@suse.cz>
703 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
704 (List of -O2 options): Likewise.
705 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
708 2017-03-17 Tom de Vries <tom@codesourcery.com>
710 * gcov-dump.c (print_usage): Print bug_report_url.
712 2017-03-17 Richard Biener <rguenther@suse.de>
715 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
716 (parser::peek): Likewise.
718 2017-03-17 Richard Biener <rguenther@suse.de>
720 PR tree-optimization/80048
721 * sese.c (free_sese_info): Properly release rename_map and
722 copied_bb_map elements.
724 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
726 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
727 Add linked-list forward and backlinks. Insert on
728 construction, remove on destruction.
729 (class pass_store_merging): Add m_stores_head field.
730 (pass_store_merging::terminate_and_process_all_chains):
731 Iterate over m_stores_head list.
732 (pass_store_merging::terminate_all_aliasing_chains):
734 (pass_store_merging::execute): Check for debug stmts first.
735 Push new chains onto the m_stores_head stack.
737 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
740 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
741 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
742 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
744 2017-03-16 Jeff Law <law@redhat.com>
746 PR tree-optimization/71437
747 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
748 member function. Implementation moved into after_dom_children
749 member function and into the threader's thread_outgoing_edges
751 (dom_opt_dom_walker::after_dom_children): Simplify by moving
752 some code into new thread_outgoing_edges.
753 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
754 definition. Simplify marker handling (do it here). Assume we always
755 have the available expression and the const/copies tables.
756 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
758 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
759 * tree-vrp.c (equiv_stack): No longer file scoped.
760 (vrp_dom_walker): New class.
761 (vrp_dom_walker::before_dom_children): New member function.
762 (vrp_dom_walker::after_dom_children): Likewise.
763 (identify_jump_threads): Setup domwalker. Use it rather than
764 walking edges in a random order by hand. Simplify setup/finalization.
765 (finalize_jump_threads): Remove.
766 (vrp_finalize): Do not call identify_jump_threads here.
767 (execute_vrp): Do it here instead and call thread_through_all_blocks
770 PR tree-optimization/71437
771 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
773 (simplify_stmt_for_jump_threading): Add basic_block argument. All
775 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
776 (dom_opt_dom_walker::thread_across_edge): Remove
777 handle_dominating_asserts argument. All callers changed.
778 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
779 changes. Remove calls to lhs_of_dominating_assert. Other
780 uses of handle_dominating_asserts turn into unconditional code
781 (simplify_control_stmt_condition_1): Likewise.
782 (simplify_control_stmt_condition): Likewise.
783 (thread_through_normal_block, thread_across_edge): Likewise.
784 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
785 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
786 object if it is not an SSA_NAME.
787 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
788 before calling into the VRP specific simplifiers.
789 (identify_jump_threads): Remove handle_dominating_asserts
792 2017-03-16 Jakub Jelinek <jakub@redhat.com>
795 * tree-diagnostic.c (default_tree_printer): No longer static.
796 * tree-diagnostic.h (default_tree_printer): New prototype.
798 2017-03-16 Tamar Christina <tamar.christina@arm.com>
800 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
801 Change ins into fmov.
803 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
805 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
806 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
807 Use h_con constraint for operand 1.
808 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
809 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
811 2017-03-15 Jeff Law <law@redhat.com>
813 PR tree-optimization/71437
814 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
815 (record_temporary_equivalences): Use it.
817 PR tree-optimization/71437
818 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
819 tree-ssa-scopedtables.
820 (lookup_avail_expr, build_and_record_new_cond): Likewise.
821 (record_conditions, record_cond, vuse_eq): Likewise.
822 (record_edge_info): Adjust to API tweak of record_conditions.
823 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
824 (record_temporary_equivalences, optimize_stmt): Likewise.
825 (eliminate_redundant_computations): Likewise.
826 (record_equivalences_from_stmt): Likewise.
827 * tree-ssa-scopedtables.c: Include options.h and params.h.
828 (vuse_eq): New function, moved from tree-ssa-dom.c
829 (build_and_record_new_cond): Likewise.
830 (record_conditions): Likewise. Accept vector of conditions rather
831 than edge_equivalence structure for first argument.
832 for the first argument.
833 (avail_exprs_stack::lookup_avail_expr): New member function, moved
835 (avail_exprs_stack::record_cond): Likewise.
836 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
838 (avail_exprs_stack): Add new member functions lookup_avail_expr
840 (record_conditions): Declare.
842 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
845 * lra-constraints.c (process_alt_operands): Increase reject for
846 reloading an input/output operand.
848 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
851 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
852 insns to convert from signed/unsigned char/short to IEEE 128-bit
854 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
856 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
859 * config/i386/i386.c (ix86_vector_duplicate_value): Create
860 subreg of inner mode for values already in registers.
862 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
864 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
865 iteration reg is used after the loop.
867 2017-03-14 Martin Sebor <msebor@redhat.com>
869 PR tree-optimization/79800
870 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
871 precision in negative-positive range.
872 (format_floating): Call non-const overload with adjusted precision.
874 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
877 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
880 2017-03-14 Martin Sebor <msebor@redhat.com>
883 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
884 * builtins.def (aligned_alloc): Use it.
887 * Makefile.in (GTFILES): Add calls.c.
888 * calls.c: Include "gt-calls.h".
890 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
892 PR rtl-optimization/79728
893 * regs.h (struct target_regs): New field
894 x_contains_allocatable_regs_of_mode.
895 (contains_allocatable_regs_of_mode): New macro.
896 * reginfo.c (init_reg_sets_1): Initialize it, and change
897 contains_reg_of_mode so it includes global regs as well.
898 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
899 rather than contains_regs_of_mode.
901 2017-03-14 Martin Liska <mliska@suse.cz>
903 * doc/invoke.texi: Document options that can't be combined with
904 -fcheck-pointer-bounds.
906 2017-03-14 Martin Liska <mliska@suse.cz>
909 * doc/invoke.texi (-Wchkp): Document the option.
911 2017-03-14 Martin Liska <mliska@suse.cz>
913 * Makefile.in: Install gcov-dump.
915 2017-03-14 Martin Liska <mliska@suse.cz>
917 * multiple_target.c (expand_target_clones): Bail out for
918 an invalid attribute.
920 2017-03-14 Richard Biener <rguenther@suse.de>
922 * alias.c (struct alias_set_entry): Pack properly.
923 * cfgloop.h (struct loop): Likewise.
924 * cse.c (struct set): Likewise.
925 * ipa-utils.c (struct searchc_env): Likewise.
926 * loop-invariant.c (struct invariant): Likewise.
927 * lra-remat.c (struct cand): Likewise.
928 * recog.c (struct change_t): Likewise.
929 * rtl.h (struct address_info): Likewise.
930 * symbol-summary.h (function_summary): Likewise.
931 * tree-loop-distribution.c (struct partition): Likewise.
932 * tree-object-size.c (struct object_size_info): Likewise.
933 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
934 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
935 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
936 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
937 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
938 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
939 (struct _stmt_vec_info): Likewise.
941 2017-03-14 Martin Liska <mliska@suse.cz>
944 * multiple_target.c (create_dispatcher_calls): Check that
945 a target can create a function dispatcher.
947 2017-03-14 Martin Liska <mliska@suse.cz>
950 * multiple_target.c (expand_target_clones): Drop local.local
951 flag for default implementation.
953 2017-03-14 Richard Biener <rguenther@suse.de>
955 PR tree-optimization/80030
956 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
958 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
960 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
961 gcc_fallthrough() instead of __attribute__((fallthrough));
963 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
965 * doc/gcc.texi: Remove "up" link to (DIR).
966 * doc/gccint.texi: Ditto.
968 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
970 * doc/install.texi (Specific) <avr>: Remove reference to
973 2017-03-13 Jeff Law <law@redhat.com>
975 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
976 attribute rather than comments.
978 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
979 match_scratch operand is highest.
981 2017-03-13 Martin Liska <mliska@suse.cz>
984 * ipa-pure-const.c (warn_function_noreturn): If the declarations
985 is a CHKP clone, use original declaration.
987 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
989 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
990 (arc_conditional_register_usage): Use a different allocation order
991 when optimizing for size.
992 * common/config/arc/arc-common.c (arc_option_optimization_table):
993 Section anchors default on when optimizing for size.
995 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
997 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
999 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
1001 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
1002 * config/arc/arc.md (cpu_facility): Add cd variant.
1003 (*movqi_insn): Add code density variant.
1004 (*movhi_insn): Likewise.
1005 (*movqi_insn): Likewise.
1006 (*addsi3_mixed): Likewise.
1007 (subsi3_insn): Likewise.
1009 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
1011 * config/arc/arc.md (movsi_cond_exec): Update constraint.
1013 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
1015 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
1016 expressions with MINUS and UNARY ops.
1018 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1021 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
1023 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
1024 between vec_select and vector argument.
1025 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
1026 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
1027 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
1028 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
1029 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
1030 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
1032 2017-03-13 Richard Biener <rguenther@suse.de>
1035 * params.def (vect-max-peeling-for-alignment): Fix typo.
1037 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
1039 * doc/install.texi (Specific) <mips-*-*>: Remove description of
1040 issue that only occurred with binutils below 2.18.
1042 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
1044 * doc/install.texi (Specific) <cris-axis-elf>: No longer
1045 refer to binutils 2.11/2.12 minimum.
1047 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
1049 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
1050 ftp.kernel.org and simplify binutils requirement.
1052 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
1054 * doc/invoke.texi (Warning Options): Fix spelling of link-time
1056 (Optimize Options): Ditto. Also remove redundancy.
1058 2017-03-10 David Malcolm <dmalcolm@redhat.com>
1060 PR translation/79848
1061 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
1063 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
1064 to G_ to avoid double translation.
1066 2017-03-10 David Malcolm <dmalcolm@redhat.com>
1068 PR translation/79923
1069 * auto-profile.c (get_combined_location): Convert leading
1070 character of diagnostics to lower case and remove trailing period.
1071 (read_profile): Likewise for various diagnostics.
1072 * config/arm/arm.c (arm_option_override): Remove trailing period
1073 from various diagnostics.
1074 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
1075 (msp430_expand_delay_cycles): Likewise.
1077 2017-03-10 David Malcolm <dmalcolm@redhat.com>
1080 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
1081 full command-line argument, rather than just "str".
1082 (aarch64_validate_march): Likewise.
1083 (aarch64_validate_mtune): Likewise.
1085 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
1087 PR rtl-optimization/78911
1088 * lra-assigns.c (must_not_spill_p): New function.
1089 (spill_for): Use it.
1091 2017-03-10 Jakub Jelinek <jakub@redhat.com>
1093 PR tree-optimization/79981
1094 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
1095 ATOMIC_COMPARE_EXCHANGE ifn result.
1096 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
1097 IFN_ATOMIC_COMPARE_EXCHANGE.
1099 2017-03-10 David Malcolm <dmalcolm@redhat.com>
1102 * opts.c (parse_sanitizer_options): Add missing question mark to
1103 "did you mean" message.
1105 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1107 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
1109 (VMULEUH_UNS): Likewise.
1110 (VMULOUB_UNS): Likewise.
1111 (VMULOUH_UNS): Likewise.
1112 * config/rs6000/rs6000.c (builtin_function_type): Remove
1113 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
1115 2017-03-10 David Malcolm <dmalcolm@redhat.com>
1118 * read-rtl-function.c (function_reader::read_rtx_operand): Update
1119 x with result of extra_parsing_for_operand_code_0.
1120 (function_reader::extra_parsing_for_operand_code_0): Convert
1121 return type from void to rtx, returning x. When reading
1122 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
1123 larger size containing struct block_symbol.
1125 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
1127 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
1128 -mfloat128-hardware without -m64.
1130 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
1133 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
1134 entries to the case statement that marks unsigned arguments to
1135 overloaded functions.
1137 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
1139 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
1140 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
1142 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
1145 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
1146 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
1148 2017-03-10 Martin Liska <mliska@suse.cz>
1152 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
1153 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
1156 2017-03-10 Olivier Hainque <hainque@adacore.com>
1158 * tree-switch-conversion (array_value_type): Start by resetting
1159 candidate type to it's main variant.
1161 2017-03-10 Jakub Jelinek <jakub@redhat.com>
1163 PR rtl-optimization/79909
1164 * combine.c (try_combine): Use simplify_replace_rtx on individual
1165 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
1166 of the whole CALL_INSN_FUNCTION_USAGE.
1168 PR tree-optimization/79972
1169 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
1170 get_range_info on SSA_NAMEs. Formatting fixes.
1172 2017-03-10 Richard Biener <rguenther@suse.de>
1173 Jakub Jelinek <jakub@redhat.com>
1175 PR tree-optimization/77975
1176 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
1177 edge to be constant.
1178 (get_val_for): For constant x return it. Formatting fix.
1179 (loop_niter_by_eval): Avoid pointless looping if the next iteration
1180 would use the same bases as the current one.
1182 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1184 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
1185 instead of vec_select for V1TImode.
1186 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
1188 (VSX_LE_128): Add V1TI to this mode iterator.
1189 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
1190 (*vsx_le_perm_store_<mode>): Likewise.
1191 (pre-reload splitter for VSX stores): Likewise.
1192 (post-reload splitter for VSX stores): Likewise.
1193 (*vsx_xxpermdi2_le_<mode>): Likewise.
1194 (*vsx_lxvd2x2_le_<mode>): Likewise.
1195 (*vsx_stxvd2x2_le_<mode>): Likewise.
1197 2017-03-09 Michael Eager <eager@eagercon.com>
1199 Correct failures with --enable-checking=yes,rtl.
1201 * config/microblaze/microblaze.c (microblaze_expand_shift):
1202 Replace GET_CODE test with CONST_INT_P and INTVAL test with
1203 test for const0_rtx.
1204 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
1205 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
1207 2017-03-09 Richard Biener <rguenther@suse.de>
1209 PR tree-optimization/79977
1210 * graphite-scop-detection.c (scop_detection::merge_sese):
1211 Handle the case of extra exits to blocks dominating the entry.
1213 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
1215 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
1218 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
1220 PR rtl-optimization/79949
1221 * lra-constraints.c (process_alt_operands): Check memory when
1222 trying to predict a cycle. Print about the overall increase.
1224 2017-03-09 Richard Biener <rguenther@suse.de>
1227 * gimple-expr.c (useless_type_conversion_p): Preserve
1228 TYPE_SATURATING for fixed-point types.
1230 2017-03-09 Richard Biener <rguenther@suse.de>
1233 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
1234 alignment of BLKmode params.
1236 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1239 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
1240 (VALL_NO_V2Q): Likewise.
1242 * config/aarch64/aarch64-simd.md
1243 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
1245 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
1246 VALL_NO_V2Q mode iterator.
1247 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
1249 2017-03-09 Martin Liska <mliska@suse.cz>
1251 PR tree-optimization/79631
1252 * tree-chkp-opt.c (chkp_is_constant_addr): Call
1253 tree_int_cst_sign_bit just for INTEGER constants.
1255 2017-03-09 Martin Liska <mliska@suse.cz>
1259 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
1262 2017-03-09 Marek Polacek <polacek@redhat.com>
1265 * tree.c (inchash::add_expr): Handle TREE_VEC.
1267 2017-03-09 Martin Liska <mliska@suse.cz>
1270 (chkp_narrow_size_and_offset): New function.
1271 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
1272 (void chkp_parse_bit_field_ref): New function.
1273 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
1274 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
1276 2017-03-09 Martin Liska <mliska@suse.cz>
1279 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
1280 (chkp_find_bounds_1): Remove gcc_unreachable.
1282 2017-03-09 Jakub Jelinek <jakub@redhat.com>
1285 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
1286 BUILT_IN_SYNC*, determine the access type from the size suffix and
1287 always build a MEM_REF with that type. Handle forgotten
1288 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
1291 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
1292 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
1293 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
1294 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
1295 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
1296 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
1297 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
1298 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
1299 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
1300 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
1301 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
1302 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
1303 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
1304 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
1305 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
1306 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
1307 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
1308 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
1309 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
1310 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
1311 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
1312 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
1313 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
1314 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
1315 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
1316 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
1317 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
1318 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
1319 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
1320 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
1321 definitions outside of __OPTIMIZE__ guarded section.
1324 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
1325 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
1326 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
1327 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
1330 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1332 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
1333 ("vfenez<mode>"): Add missing constraints.
1335 2017-03-08 Martin Sebor <msebor@redhat.com>
1338 * config/nds32/nds32.c (nds32_option_override):
1340 2017-03-08 Jakub Jelinek <jakub@redhat.com>
1343 * gimplify.c (gimplify_omp_for): Replace index var in outer
1344 taskloop statement with an artificial variable and add
1345 OMP_CLAUSE_PRIVATE clause for it.
1347 2017-03-08 Richard Biener <rguenther@suse.de>
1349 PR tree-optimization/79955
1350 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
1351 for accesses that are completely outside of the variable.
1353 2017-03-08 Andrew Haley <aph@redhat.com>
1355 PR tree-optimization/79943
1356 * tree-ssa-loop-split.c (compute_new_first_bound): When
1357 calculating the new upper bound, (END-BEG) should be added, not
1360 2017-03-08 Jakub Jelinek <jakub@redhat.com>
1362 * config/avr/avr.md (setmemhi): Make sure match_dup
1363 operand number comes before match_scratch.
1365 2017-03-08 Richard Biener <rguenther@suse.de>
1367 PR tree-optimization/79920
1368 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
1369 with ncopies == 1 to ...
1370 (vect_transform_slp_perm_load): ... here. Properly compute
1371 all element loads by iterating VF times over the group. Do
1372 not handle ncopies (computed in a broken way) in
1373 vect_create_mask_and_perm.
1375 2017-03-08 Jakub Jelinek <jakub@redhat.com>
1378 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
1379 is a uniform vector, use uniform_vector_p return value instead of
1380 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
1382 2017-03-07 Marek Polacek <polacek@redhat.com>
1385 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
1386 (alloca_call_type): Likewise.
1388 2017-03-07 Martin Liska <mliska@suse.cz>
1390 * gcov.c (process_args): Put comment to correct location.
1392 2017-03-07 Martin Liska <mliska@suse.cz>
1395 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
1396 Use array_at_struct_end_p instead of DECL_CHAIN (field).
1397 (chkp_narrow_bounds_for_field): Likewise.
1398 (chkp_parse_array_and_component_ref): Pass one more argument to
1401 2017-03-07 Richard Biener <rguenther@suse.de>
1403 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
1406 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
1408 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
1409 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
1411 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1414 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
1415 to end of description.
1416 (PARAM_MAX_STORES_TO_MERGE): Likewise.
1418 2017-03-07 Jakub Jelinek <jakub@redhat.com>
1420 PR rtl-optimization/79901
1421 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
1423 (*avx512f_<code><mode>3<mask_name>): ... this.
1424 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
1425 iterator instead of VI8_AVX2_AVX512BW.
1427 PR rtl-optimization/79901
1428 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
1429 min/max expander, expand it using expand_vec_cond_expr.
1432 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
1435 2017-03-06 Jakub Jelinek <jakub@redhat.com>
1438 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
1439 to void * for PCH reasons.
1440 * dwarf2out.c (output_loc_operands, output_die): Cast
1441 v.val_vec.array to unsigned char *.
1443 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
1446 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
1449 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
1451 PR rtl-optimization/79571
1452 * lra-constraints.c (process_alt_operands): Calculate static
1453 reject and subtract it from overall when only addresses will be
1456 2017-03-06 Julia Koval <julia.koval@intel.com>
1459 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
1460 incoming stack boundary to 128 for 64-bit targets.
1462 2017-03-06 Richard Biener <rguenther@suse.de>
1464 PR tree-optimization/79894
1465 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
1466 to NULL after folding it.
1468 2017-03-06 Richard Biener <rguenther@suse.de>
1470 PR tree-optimization/79824
1471 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
1472 check disabling peeling for gaps.
1474 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
1476 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
1477 attributes): Document gettimeofday.
1479 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
1481 * config/s390/s390.c (s390_option_override_internal): Set
1482 PARAM_MIN_VECT_LOOP_BOUND
1484 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
1486 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
1487 * config/s390/s390.md: Likewise.
1489 2017-03-06 Jakub Jelinek <jakub@redhat.com>
1492 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
1493 (<avx2_avx512>_perm<mode>): Rename to ...
1494 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
1496 (<avx512>_perm<mode>_mask): Rename to ...
1497 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
1499 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
1500 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
1501 instead of VI8F_256_512.
1502 (avx512f_perm<mode>): New define_expand.
1503 (avx512f_perm<mode>_mask): Likewise.
1504 (avx512f_perm<mode>_1<mask_name>): New define_insn.
1505 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
1507 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
1509 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
1510 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
1512 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
1514 2017-03-06 Martin Liska <mliska@suse.cz>
1517 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
1518 when having a SSA NAME w/o VAR_DECL assigned to it.
1520 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
1522 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
1523 msa_dpsub_<su>_d): Fix MODE for vec_select.
1525 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
1527 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
1529 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
1531 2017-03-06 Richard Biener <rguenther@suse.de>
1533 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
1534 * plugin.c (register_plugin_info): Likewise.
1535 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
1537 2017-03-05 Jakub Jelinek <jakub@redhat.com>
1539 * config/i386/sse.md (sse_storehps, sse_storelps,
1540 avx_<castmode><avxsizesuffix>_<castmode>,
1541 avx512f_<castmode><avxsizesuffix>_<castmode>,
1542 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
1543 in condition that at least one operand is not a MEM.
1545 2017-03-03 Jakub Jelinek <jakub@redhat.com>
1548 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
1549 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
1551 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
1552 gimple_call_nothrow_p flag based on whether original builtin can throw.
1553 If it can, emit following stmts on the fallthrough edge.
1554 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
1555 don't create new bb if inserting just debug stmts on the edge, try to
1556 insert them on the fallthru bb or just reset debug stmts.
1558 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
1561 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
1562 restore recog_data (including the operand rtxes inside it) around
1563 the call to get_insn_template.
1565 2017-03-03 Martin Sebor <msebor@redhat.com>
1567 PR tree-optimization/79699
1568 * context.c (context::~context): Free MPFR caches to avoid
1569 a memory leak on program exit.
1571 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1573 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
1574 Use wide_int::ulow () instead of .elt (0).
1576 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
1578 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
1579 (*pushxf): Limit oF constraint to 32bit targets and add oC
1580 constraint for 64bit targets.
1581 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
1582 (*pushdf): Change rmF constraint to rmC.
1584 2017-03-03 Martin Liska <mliska@suse.cz>
1586 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
1587 Remove unused variable.
1589 2017-03-03 Jakub Jelinek <jakub@redhat.com>
1592 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
1593 is a memory operand, increase num_memory.
1594 (ix86_expand_args_builtin): Likewise.
1596 2017-03-03 Jan Hubicka <jh@suse.cz>
1599 * ipa-devirt.c (maybe_record_node): Properly handle
1600 __cxa_pure_virtual visibility.
1602 2017-03-03 Martin Liska <mliska@suse.cz>
1604 PR tree-optimization/79803
1605 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
1607 (pass_loop_prefetch::execute): Disabled optimization if an
1608 assumption about L1 cache size is not met.
1610 2017-03-03 Martin Liska <mliska@suse.cz>
1612 PR rtl-optimization/79574
1613 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
1614 (hash_scan_set): Likewise.
1615 (dump_hash_table): Likewise.
1616 (hoist_code): Likewise.
1618 2017-03-03 Richard Biener <rguenther@suse.de>
1620 * fixed-value.c (fixed_from_string): Restore use of elt (1)
1621 in place of uhigh ().
1622 (fixed_convert_from_real): Likewise.
1624 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
1627 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
1629 2017-03-03 Richard Biener <rguenther@suse.de>
1632 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
1633 TYPE_OVERFLOW_UNDEFINED check.
1635 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1637 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
1639 (vector_ae_<mode>_p): Likewise.
1640 (vector_nez_<mode>_p): Likewise.
1641 (vector_ne_v2di_p): Likewise.
1642 (vector_ae_v2di_p): Likewise.
1643 (vector_ne_<mode>_p): Likewise.
1644 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
1646 (vsx_tsqrt<mode>2_fe): Likewise.
1648 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
1651 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
1653 2017-03-02 Jakub Jelinek <jakub@redhat.com>
1655 PR rtl-optimization/79780
1656 * cprop.c (one_cprop_pass): When second and further conditional trap
1657 in a single basic block is turned into an unconditional trap, turn it
1658 into a deleted note to avoid RTL verification failures.
1660 2017-03-02 Richard Biener <rguenther@suse.de>
1662 * fold-const.c (const_binop): Use ulow () instead of elt (0).
1664 2017-03-02 Richard Biener <rguenther@suse.de>
1666 PR tree-optimization/79345
1668 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
1669 param and abort the walk, returning -1 if it is hit.
1670 (walk_aliased_vdefs): Take a limit param and pass it on.
1671 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
1672 defaulting to 0 and return a signed int.
1673 * tree-ssa-uninit.c (struct check_defs_data): New struct.
1674 (check_defs): New helper.
1675 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
1676 about uninitialized memory.
1677 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
1678 bogus uninitialized warning.
1679 (fixed_convert_from_real): Likewise.
1681 2017-03-02 Bin Cheng <bin.cheng@arm.com>
1683 PR tree-optimization/66768
1684 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
1685 iv_use if base object can't be determined.
1687 2017-03-02 Jakub Jelinek <jakub@redhat.com>
1689 PR tree-optimization/79345
1690 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
1691 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
1692 (get_pattern_stats): Initialize it.
1693 * genemit.c (gen_expand): Verify match_scratch numbers come after
1694 match_operand/match_dup numbers.
1695 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
1696 match_scratch numbers.
1697 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
1699 * config/s390/s390.md (trunctdsd2): Likewise.
1701 2017-03-02 Richard Biener <rguenther@suse.de>
1703 * wide-int.h (wide_int_storage::operator=): Implement in terms
1706 2017-03-02 Richard Biener <rguenther@suse.de>
1708 PR tree-optimization/79777
1709 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
1710 the to insert expression to sth existing.
1712 2017-03-01 Martin Sebor <msebor@redhat.com>
1715 * gimple-ssa-sprintf.c
1716 (directive::known_width_and_precision): New function.
1717 (format_integer): Use it.
1718 (get_mpfr_format_length): Consider the full range of precision
1719 when computing %g output with the # flag. Set the likely byte
1720 count to 3 rather than 1 when precision is indeterminate.
1721 (format_floating): Correct the lower bound of precision.
1723 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1725 * doc/invoke.texi: Document default code model for 64-bit Linux.
1727 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
1730 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
1731 udiv rather than div since input pattern is unsigned.
1733 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
1735 * config/i386/i386.c (print_reg): Warn for values of
1736 unsupported size in integer register.
1738 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
1741 * config/rs6000/predicates.md (current_file_function_operand): Do
1742 not allow self calls to be local if the function is replaceable.
1744 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
1747 * config/rs6000/altivec.h (vec_ctz and others): Change the
1748 preprocessor macro that controls conditional compilation from
1749 _ARCH_PWR9 to __POWER9_VECTOR__.
1750 (vec_all_ne): Change parameterization of __altivec_scalar_pred
1751 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
1752 control (instead of _ARCH_PWR9 control) so that template
1753 definition uses power9-specific function.
1754 (vec_any_eq): Likewise.
1755 (vec_all_ne): Change macro definition to use a power9-specific
1756 expansion under #ifdef __POWER9_VECTOR__ control (instead of
1757 _ARCH_PWR9 control).
1758 (vec_any_eq) Likewise.
1759 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
1760 expansion for CMPNEF to remove support for xvcmpnesp instruction.
1761 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
1762 support for xvcmpnedp instruction.
1763 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
1764 macro expansion so that Power9 implementation of vec_all_ne does
1765 not use the AltiVec predicate framework.
1766 (VCMPNEH_P): Likewise.
1767 (VCMPNEW_P): Likewise.
1768 (VCMPNED_P): Likewise.
1769 (VCMPNEFP_P): Likewise.
1770 (VCMPNEDP_P): Likewise.
1771 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
1772 implementation of vec_any_eq to not use AltiVec predicate
1774 (VCMPAEH_P): Likewise.
1775 (VCMPAEW_P): Likewise.
1776 (VCMPAED_P): Likewise.
1777 (VCMPAEFP_P): Likewise.
1778 (VCMPAEDP_P): Likewise.
1779 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
1780 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
1781 not use the AltiVec predicate framework.
1782 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
1783 of vec_any_eq to not use AltiVec predicate framework.
1784 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
1785 support for predefined __POWER9_VECTOR__ macro to indicate that
1786 Power9 instruction selection is enabled.
1787 (altivec_overloaded_builtins): Remove extraneous
1788 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
1789 function argument types RS6000_BTI_bool_V16QI and
1790 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
1791 entry for overloaded function argument types RS6000_BTI_bool_V4SI
1792 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
1793 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
1794 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
1795 Power9 for implementations of vec_cmpne. Change the signature for
1796 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
1797 (representing vec_all_ne) to remove the previously described first
1798 argument of type RS6000_BTI_INTSI, as this was an artifact of
1799 reliance on the AltiVec predicate framework, which is no longer
1800 used in the implementation of these functions. Add
1801 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
1802 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
1803 since, unlike the AltiVec predicate framework implementation, we
1804 do not share function descriptors between vec_alle and vec_anyeq.
1805 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
1806 set of modes that receive special treatment even when
1807 TARGET_P9_VECTOR is true. The special treatment emits code that
1808 does not depend on Power9 instructions.
1809 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
1810 define_expand to not rely on AltiVec predicate framework.
1811 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
1813 (vector_ne_v2di_p): Change this define_expand to not rely on
1814 AltiVec predicate framework.
1815 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
1817 (vector_ne_<mode>_p): Change this define_expand to not rely on
1818 AltiVec predicate framework.
1819 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
1821 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
1822 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
1823 define_insn pattern.
1824 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
1825 define_insn pattern because the xvcmpne<VSs>. instruction is not
1827 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
1828 instruction is not supported.
1830 2017-03-01 Jakub Jelinek <jakub@redhat.com>
1832 * config/nvptx/nvptx.c: Include intl.h.
1834 2017-03-01 Martin Jambor <mjambor@suse.cz>
1837 * ipa-prop.h (ipa_bits): Removed field known.
1838 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
1839 to pointers. Adjusted their comments to warn about their sharing.
1840 (ipcp_transformation_summary): Change bits to a vector of pointers.
1841 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
1842 (ipa_get_ipa_bits_for_value): Declare.
1843 * tree-vrp.h (value_range): Mark as GTY((for_user)).
1844 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
1845 (ipa_bits_hash_table): Likewise.
1846 (ipa_vr_ggc_hash_traits): Likewise.
1847 (ipa_vr_hash_table): Likewise.
1848 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
1849 being pointers and vr_known being removed.
1850 (ipa_set_jf_unknown): Likewise.
1851 (ipa_get_ipa_bits_for_value): New function.
1852 (ipa_set_jfunc_bits): Likewise.
1853 (ipa_get_value_range): New overloaded functions.
1854 (ipa_set_jfunc_vr): Likewise.
1855 (ipa_compute_jump_functions_for_edge): Use the above functions to
1856 construct bits and vr parts of jump functions.
1857 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
1858 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
1860 (ipcp_grow_transformations_if_necessary): Also allocate
1861 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
1863 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
1864 them. Fix too long lines.
1865 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
1866 vr_known being removed.
1867 (ipa_read_jump_function): Use new setter functions to construct bits
1868 and vr parts of jump functions or set them to NULL.
1869 (write_ipcp_transformation_info): Adjust for bits being pointers.
1870 (read_ipcp_transformation_info): Likewise.
1871 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
1873 Include gt-ipa-prop.h.
1874 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
1876 (ipcp_store_bits_results): Likewise.
1877 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
1878 Do not write to existing jump functions but use a temporary instead.
1880 2017-03-01 Jakub Jelinek <jakub@redhat.com>
1883 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
1884 attempt to use its first operand as BIT_FIELD_REF base.
1886 2017-03-01 Richard Biener <rguenther@suse.de>
1889 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
1890 interpolating formula in wrapping arithmetic.
1891 (chrec_apply): Convert chrec_evaluate return value to wanted type.
1893 2017-03-01 Jakub Jelinek <jakub@redhat.com>
1895 PR tree-optimization/79734
1896 * tree-vect-generic.c (expand_vector_condition): Optimize
1897 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
1898 Handle VEC_COND_EXPR where comparison has different inner width from
1901 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
1903 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
1904 markup, and similar issues. Remove @opindex entries for things
1905 that aren't options. Add missing -mmpy-option entries.
1907 2017-02-28 Jakub Jelinek <jakub@redhat.com>
1909 PR tree-optimization/79737
1910 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
1911 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
1912 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
1913 instead of byte_size. Formatting fix.
1914 (shift_bytes_in_array_right): Formatting fix.
1916 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
1919 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
1920 condition on optimize for the leaf function test.
1922 2017-02-28 Martin Liska <mliska@suse.cz>
1925 * read-rtl-function.c (function_reader::handle_unknown_directive):
1926 Bail out when one uses -flto.
1928 2017-02-28 Martin Liska <mliska@suse.cz>
1930 * common.opt: Replace space with tabular for options of <number>
1932 * config/i386/i386.opt: Show <number> value for
1933 -mlarge-data-threshold.
1934 * opts.c (print_filtered_help): Do not display number in hexadecimal
1937 2017-02-28 Martin Liska <mliska@suse.cz>
1939 * common.opt: Fix --help=option -Q for options which are of
1942 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
1944 * config/i386/i386.c (print_reg): Error out for values
1945 of 8-bit size in invalid integer register.
1947 2017-02-28 Martin Sebor <msebor@redhat.com>
1949 PR tree-optimization/79691
1950 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
1952 2017-02-28 Jakub Jelinek <jakub@redhat.com>
1955 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
1956 gcc_unreachable with output_operand_lossage.
1958 2017-02-28 Richard Biener <rguenther@suse.de>
1960 PR tree-optimization/79740
1961 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
1963 (visit_nary_op): Insert the nary into the hashtable if we
1964 pattern-matched sth.
1965 * tree-ssa-pre.c (eliminate_insert): Robustify.
1967 2017-02-28 Richard Biener <rguenther@suse.de>
1970 * fold-const.c (decode_field_reference): Reject out-of-bound
1973 2017-02-28 Jakub Jelinek <jakub@redhat.com>
1975 * config/i386/i386.c: Include intl.h.
1976 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
1977 instead of just cond ? "..." : "...".
1978 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
1979 * coverage.c (read_counts_file): Likewise.
1980 * omp-offload.c: Include intl.h.
1981 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
1982 of just cond ? "..." : "...".
1983 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
1984 of just cond ? "..." : "...".
1986 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
1989 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
1991 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
1992 'tune for' CPU name.
1993 * config/arm/arm-cpu-data.h: Regenerated.
1995 2017-02-28 Richard Biener <rguenther@suse.de>
1997 PR tree-optimization/79732
1998 * tree-inline.c (expand_call_inline): Do not shadow var.
2000 2017-02-28 Richard Biener <rguenther@suse.de>
2002 PR tree-optimization/79723
2003 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
2004 address-space properly.
2006 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
2008 * doc/optinfo.texi (Optimization groups): Fix option used for
2010 * doc/invoke.texi (-fopt-info): Document "omp".
2011 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
2012 (OPTGROUP_ALL): Add OPTGROUP_OMP.
2013 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
2014 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
2015 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
2017 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
2019 * dumpfile.c (optgroup_options): Instead of "openmp", associate
2020 OPTGROUP_OMP with "omp".
2022 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
2025 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
2026 for arithmetic shift of unsigned V2DI.
2028 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
2030 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
2031 arc/linux.h headers.
2032 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
2033 (LINK_SPEC): Likewise.
2034 (ARC_TLS_EXTRA_START_SPEC): Likewise.
2035 (EXTRA_SPECS): Likewise.
2036 (STARTFILE_SPEC): Likewise.
2037 (ENDFILE_SPEC): Likewise.
2038 (LIB_SPEC): Likewise.
2039 (TARGET_SDATA_DEFAULT): Likewise.
2040 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
2041 (MULTILIB_DEFAULTS): Likewise.
2042 (DWARF2_UNWIND_INFO): Likewise.
2043 * config/arc/big.h: New file.
2044 * config/arc/elf.h: Likewise.
2045 * config/arc/linux.h: Likewise.
2046 * config/arc/t-uClibc: Remove.
2048 2017-02-27 Bin Cheng <bin.cheng@arm.com>
2050 PR tree-optimization/77536
2051 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
2052 (tree_transform_and_unroll_loop): Use above function to compute the
2053 estimated niter of unrolled loop and use it when scaling profile.
2054 Also use count info rather than frequency if it's non-zero.
2055 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
2056 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
2057 (vect_transform_loop): Call above function.
2059 2017-02-27 Richard Biener <rguenther@suse.de>
2061 PR tree-optimization/45397
2062 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
2063 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
2064 (visit_nary_op): Add pattern matching for CSEing sign-changed
2065 or truncated operations with wider ones.
2067 2017-02-27 Richard Biener <rguenther@suse.de>
2069 PR tree-optimization/79690
2070 * tree-vect-stmts.c (vectorizable_store): Use vector type
2071 built from the DR with address-space.
2073 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
2075 * doc/invoke.texi (Optimize Options): Refine the description
2076 of asan-use-after-return.
2078 2017-02-25 Alan Modra <amodra@gmail.com>
2080 PR rtl-optimization/79584
2081 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
2082 base, not ad->base_term, the reg within base. Remove assertion
2083 that ad->base == ad->base_term. Replace gen_int_mode using
2084 bogus mode with const0_rtx.
2086 2017-02-25 Jakub Jelinek <jakub@redhat.com>
2089 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
2090 FMA_EXPR like tcc_binary or tcc_unary.
2092 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
2095 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
2097 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
2098 (output_loc_operands): Handle DW_OP_call_ref and
2099 DW_OP_GNU_variable_value.
2100 (struct variable_value_struct): New type.
2101 (struct variable_value_hasher): Likewise.
2102 (variable_value_hash): New variable.
2103 (string_types): Remove.
2104 (copy_loc_descr): New function.
2105 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
2106 (prepend_loc_descr_to_each): New function.
2107 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
2108 instead of add_loc_descr_to_each if the first argument is single
2109 location list and the second has multiple.
2110 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
2111 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
2112 when looking for variable value which doesn't have other location info.
2113 (loc_list_from_tree): Formatting fix.
2114 (gen_array_type_die): Simplify DW_AT_string_length handling.
2115 (adjust_string_types): Remove.
2116 (gen_subprogram_die): Don't call adjust_string_types nor test/set
2117 string_types. Call resolve_variable_values.
2118 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
2119 (resolve_addr_in_expr): Likewise. Add A argument.
2120 (copy_deref_exprloc): Remove deref argument. Adjust for the
2121 original expression being DW_OP_GNU_variable_value with optionally
2122 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
2123 optionally after it.
2124 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
2125 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
2126 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
2127 (variable_value_hasher::hash, variable_value_hasher::equal): New
2129 (resolve_variable_value_in_expr, resolve_variable_value,
2130 resolve_variable_values, note_variable_value_in_expr,
2131 note_variable_value): New functions.
2132 (dwarf2out_early_finish): Call note_variable_value on all toplevel
2135 2017-02-24 Jakub Jelinek <jakub@redhat.com>
2138 * opts.h (handle_generated_option): Add GENERATED_P argument.
2139 * opts-common.c (handle_option): Adjust function comment.
2140 (handle_generated_option): Add GENERATED_P argument, pass it to
2142 (control_warning_option): Pass false to handle_generated_option
2144 * opts.c (maybe_default_option): Pass true to handle_generated_option
2146 * optc-gen.awk: Likewise.
2148 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
2150 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
2151 a REG, look at the REG it is a SUBREG of.
2152 (splitter for cmpeqsi_t): Ditto.
2154 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
2156 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
2157 the special USEs with the pattern of the insn, not the insn itself.
2159 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
2162 * doc/invoke.texi: Document -mload-store-pairs.
2164 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
2165 Sandra Loosemore <sandra@codesourcery.com>
2167 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
2168 argument isn't a CONST_INT.
2169 (nios2_alternate_compare_const): Assert op is a CONST_INT.
2170 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
2171 (nios2_validate_compare): Bypass alternate compare logic if *op2
2173 (ldstwm_operation_p): Return false if first_base is not a REG or
2174 if first_offset is not a CONST_INT.
2176 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
2178 * config/cris/cris.md: Use correct operand in a define_peephole2.
2180 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
2182 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
2184 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
2186 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
2187 this_insn if it is an INSN or JUMP_INSN.
2188 (force_offsettable): Look at base, not at addr.
2189 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
2190 on things that aren't necessarily CONST_INTs.
2192 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
2194 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
2195 -mfpmath=sse is the default also for x86-32 targets with SSE2
2196 instruction set when @option{-ffast-math} is enabled
2198 2017-02-24 Jeff Law <law@redhat.com>
2200 PR rtl-optimizatoin/79286
2201 * ira.c (update_equiv_regs): Drop may_trap_p exception to
2204 2017-02-24 Richard Biener <rguenther@suse.de>
2206 PR tree-optimization/79389
2207 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
2210 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
2212 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
2213 function comment to reflect reality.
2214 (loop_exits_before_overflow): Fix typo in function description.
2216 2017-02-24 Richard Biener <rguenther@suse.de>
2218 PR tree-optimization/79389
2219 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
2220 properly that a threading opportunity exists. Detect conditional
2221 copy/constant propagation opportunities.
2223 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
2225 * config/visium/visium.md (type): Add trap.
2226 (b): New mode attribute.
2227 (*btst): Rename into...
2228 (*btst<mode>): ...this and adjust.
2229 (*cbranchsi4_btst_insn): Rename into...
2230 (*cbranch<mode>4_btst_insn): ...this and adjust.
2231 (trap): New define_insn.
2233 2017-02-23 Jakub Jelinek <jakub@redhat.com>
2235 PR tree-optimization/79389
2236 * ifcvt.c (struct noce_if_info): Add rev_cond field.
2237 (noce_reversed_cond_code): New function.
2238 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
2239 reversed_comparison_code. Formatting fix.
2240 (noce_try_store_flag): Test rev_cond != NULL in addition to
2241 reversed_comparison_code.
2242 (noce_try_store_flag_constants): Likewise.
2243 (noce_try_store_flag_mask): Likewise.
2244 (noce_try_addcc): Use rev_cond if non-NULL instead of
2245 reversed_comparison_code.
2246 (noce_try_cmove_arith): Likewise. Formatting fixes.
2247 (noce_try_minmax, noce_try_abs): Clear rev_cond.
2248 (noce_find_if_block): Initialize rev_cond.
2249 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
2250 instead of false as last argument never attempt to reverse it
2253 2017-02-23 Bin Cheng <bin.cheng@arm.com>
2255 PR tree-optimization/79663
2256 * tree-predcom.c (combine_chains): Process refs in reverse order
2257 only for ZERO length chains, and add explaining comment.
2259 2017-02-23 Jeff Law <law@redhat.com>
2261 PR tree-optimization/79578
2262 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
2263 in call to operand_equal_p.
2265 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
2268 * config/i386/cpuid.h: Fix another undefined behavior.
2270 2017-02-23 Richard Biener <rguenther@suse.de>
2272 PR tree-optimization/79683
2273 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
2274 vector types for data-refs.
2276 2017-02-23 Martin Liska <mliska@suse.cz>
2278 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
2280 2017-02-23 Jakub Jelinek <jakub@redhat.com>
2283 * internal-fn.c (get_range_pos_neg): Moved to ...
2284 * tree.c (get_range_pos_neg): ... here. No longer static.
2285 * tree.h (get_range_pos_neg): New prototype.
2286 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
2287 are known to be in between 0 and signed maximum inclusive, try to
2288 expand both unsigned and signed divmod and use the cheaper one from
2291 2017-02-22 Jeff Law <law@redhat.com>
2293 PR tree-optimization/79578
2294 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
2295 to compare base operands.
2297 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
2300 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
2301 gpc_reg_operand instead of fpr_reg_operand.
2303 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
2305 * config/mips/mips.c (mips_return_in_memory): Force FP
2306 vector types to be returned in memory for o32 ABI.
2308 2017-02-22 Jakub Jelinek <jakub@redhat.com>
2310 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
2311 instead of DW_TAG_member for static data member declarations and don't
2312 set no_linkage_name for static inline data members.
2313 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
2316 2017-02-22 Martin Liska <mliska@suse.cz>
2318 * doc/invoke.texi: Replace inequality signs with square brackets
2321 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2323 PR tree-optimization/68644
2324 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
2326 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
2329 * lra-constraints.c (simplify_operand_subreg): Handle
2330 WORD_REGISTER_OPERATIONS targets.
2332 2017-02-22 Jakub Jelinek <jakub@redhat.com>
2335 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
2336 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
2337 elimination by swapping fld*.
2339 2017-02-22 Richard Biener <rguenther@suse.de>
2341 PR tree-optimization/79673
2342 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
2343 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
2344 irrelevant address-space qualifiers and avoiding a
2345 ADDR_SPACE_CONVERT_EXPR from fold_convert.
2347 2017-02-22 Richard Biener <rguenther@suse.de>
2349 PR tree-optimization/79666
2350 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
2351 to not symbolically negate if that may introduce undefined
2354 2017-02-22 Martin Liska <mliska@suse.cz>
2357 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
2358 * data-streamer-out.c (streamer_write_gcov_count_stream):
2360 * value-prof.c (stream_out_histogram_value): Make assert more
2361 precise based on type of counter.
2363 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
2366 * config/i386/i386.md (standard_x87sse_constant_load splitter):
2367 Use nonimmediate_operand instead of memory_operand for operand 1.
2368 (float-extend standard_x87sse_constant_load splitter): Ditto.
2370 2017-02-21 Jeff Law <law@redhat.com>
2372 PR tree-optimization/79621
2373 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
2374 blocks with edges to themselves.
2376 2017-02-21 Jakub Jelinek <jakub@redhat.com>
2379 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
2380 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
2381 Use gimple_call_builtin_p.
2384 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
2385 on temporarily removed DEBUG_INSNs.
2387 PR tree-optimization/79649
2388 * tree-loop-distribution.c (classify_partition): Give up on
2389 non-generic address space loads/stores.
2391 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
2393 * doc/loop.texi (Loop manipulation): Remove nonexistent
2394 tree_ssa_loop_version from the documentation.
2395 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
2397 2017-02-21 Jakub Jelinek <jakub@redhat.com>
2400 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
2401 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
2402 * config/rs6000/rs6000.c: Include except.h.
2403 (rs6000_expand_split_stack_prologue): Call
2404 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
2406 2017-02-21 Martin Jambor <mjambor@suse.cz>
2409 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
2412 2017-02-21 Martin Jambor <mjambor@suse.cz>
2414 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
2415 for backward compatibility only.
2416 * doc/invoke.texi (Option Summary): Remove all references to
2419 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
2423 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
2425 * lra-constraints.c (curr_insn_transform): Handle
2426 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
2428 2017-02-21 Martin Liska <mliska@suse.cz>
2430 * config/i386/i386.opt: Replace -masm-dialect with -masm.
2432 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
2434 PR translation/79638
2435 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
2437 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
2440 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
2441 (arm_function_ok_for_sibcall): Return false for an indirect call by
2442 descriptor if all the argument registers are used.
2443 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
2444 alignment of the function.
2446 2017-02-21 Jakub Jelinek <jakub@redhat.com>
2448 PR tree-optimization/61441
2449 * simplify-rtx.c (simplify_const_unary_operation): For
2450 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
2451 the sNaN unmodified.
2453 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
2455 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
2456 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
2457 instead of SYSTEM_HEADER_DIR.
2459 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
2460 Martin Liška <mliska@suse.cz>
2462 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
2463 Fix typos and grammar, use active voice, and clarify.
2465 2017-02-20 Marek Polacek <polacek@redhat.com>
2468 * gimplify.c (gimplify_expr): Handle unused *&&L;.
2471 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
2473 2017-02-20 Jakub Jelinek <jakub@redhat.com>
2476 * config/i386/i386.c (ix86_expand_builtin): Handle
2477 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
2478 ix86_builtins_isa[fcode].isa as a requirement of those
2479 flags and any other flag in the bitmask.
2480 (ix86_init_mmx_sse_builtins): Use 0 instead of
2481 ~OPTION_MASK_ISA_64BIT as mask.
2482 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
2483 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
2484 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
2485 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
2487 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
2490 * lra-constraints.c (split_reg): Check requested split mode
2491 is supported by the register.
2493 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
2495 * lra-constraints.c (simplify_operand_subreg): Remove early
2498 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
2501 * lra-constraints.c (curr_insn_transform): Tighten condition
2502 for converting SUBREG reloads from OP_OUT to OP_INOUT.
2504 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
2507 * lra-constraints.c (curr_insn_transform): Handle
2508 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
2510 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
2513 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
2515 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
2517 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
2520 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
2523 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2525 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
2528 2017-02-18 Jakub Jelinek <jakub@redhat.com>
2530 * final.c (last_columnnum, override_columnnum): New variables.
2531 (final_start_function): Set last_columnnum, pass it to begin_prologue
2532 hook and pass 0 to dwarf2out_begin_prologue.
2533 (final_scan_insn): Update override_columnnum. Pass last_columnnum
2534 to source_line debug hook.
2535 (notice_source_line): Compute last_columnnum and for debug_column_info
2536 return true on column changes.
2537 * debug.h (struct gcc_debug_hooks): Add column argument to
2538 source_line and begin_prologue hooks.
2539 (debug_nothing_int_charstar_int_bool): Remove prototype.
2540 (debug_nothing_int_int_charstar,
2541 debug_nothing_int_int_charstar_int_bool): New prototypes.
2542 (dwarf2out_begin_prologue): Add column argument.
2543 * debug.c (do_nothing_debug_hooks): Adjust source_line and
2544 begin_prologue hooks.
2545 (debug_nothing_int_charstar_int_bool): Remove.
2546 (debug_nothing_int_int_charstar,
2547 debug_nothing_int_int_charstar_int_bool): New functions.
2548 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
2549 through to dwarf2out_source_line.
2550 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
2551 (dwarf2out_source_line): Add column argument, emit it if requested.
2552 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
2554 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
2555 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
2556 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
2557 through to dwarf2out_begin_prologue.
2558 (vmsdbgout_source_line): Add column argument, pass it through to
2559 dwarf2out_source_line.
2560 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
2561 dbxout_source_line caller.
2562 (dbxout_source_line): Add column argument.
2564 * common.opt (gno-column-info, gcolumn-info): New options.
2565 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
2566 (check_die): Also test for multiple DW_AT_decl_column attributes.
2567 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
2568 DW_AT_decl_column if requested.
2569 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
2571 (gen_variable_die): Likewise.
2572 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
2573 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
2576 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
2577 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
2578 (ix86_handle_option): Handle OPT_m3dnowa.
2579 * doc/invoke.texi (-m3dnowa): Document.
2580 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
2581 -m3dnowa instead of -m3dnow -march=athlon.
2584 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
2585 instead of gcc_assert for K, r and R code checks. Formatting fixes.
2587 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2590 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
2591 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
2592 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
2593 generator for vsx_xxpermdi_<mode>_be.
2594 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
2595 force big-endian semantics.
2596 (vsx_xxpermdi_<mode>_be): New define_expand with same
2597 implementation as previous version of vsx_xxpermdi_<mode>.
2599 2017-02-17 Jakub Jelinek <jakub@redhat.com>
2601 PR tree-optimization/79327
2602 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
2603 variable, its initialization and use.
2605 2017-02-17 Julia Koval <julia.koval@intel.com>
2607 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
2608 (OPTION_MASK_ISA_PKU_UNSET): New.
2609 (ix86_handle_option): Handle -mrdpid.
2610 * config/i386/cpuid.h (bit_RDPID): New.
2611 * config/i386/driver-i386.c (host_detect_local_cpu):
2612 Detect RDPID feature.
2613 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
2614 * config/i386/i386-c.c (ix86_target_macros_internal):
2616 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
2617 (ix86_valid_target_attribute_inner_p): Add "rdpid".
2618 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
2619 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
2620 * config/i386/i386.md (define_insn "rdpid"): New.
2621 * config/i386/i386.opt Add -mrdpid.
2622 * config/i386/immintrin.h (_rdpid_u32): New.
2624 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
2626 PR rtl-optimization/79541
2627 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
2628 instead of transforming it into USE.
2630 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
2632 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
2633 If HONOR_SNANS (SFmode) force the input to a register.
2634 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
2635 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
2636 an frsp or similar insn.
2638 2017-02-17 Martin Liska <mliska@suse.cz>
2640 PR rtl-optimization/79577
2641 * params.def (selsched-max-sched-times): Increase minimum to 1.
2643 2017-02-17 Martin Liska <mliska@suse.cz>
2645 PR rtl-optimization/79574
2646 * gcse.c (want_to_gcse_p): Prevent integer overflow.
2648 2017-02-17 Martin Liska <mliska@suse.cz>
2650 PR tree-optimization/79529
2651 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
2652 ssa_defined_default_def_p to handle cases which are implicitly
2654 * tree-ssa.c (ssa_defined_default_def_p): New function.
2655 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
2656 which are implicitly defined.
2657 * tree-ssa.h (ssa_defined_default_def_p): Declare.
2659 2017-02-17 Richard Biener <rguenther@suse.de>
2662 * params.def (max-ssa-name-query-depth): Limit to 10.
2664 2017-02-17 Richard Biener <rguenther@suse.de>
2666 PR tree-optimization/79552
2667 * tree-ssa-structalias.c (visit_loadstore): Properly verify
2670 2017-02-17 Richard Biener <rguenther@suse.de>
2673 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
2675 2017-02-17 Marek Polacek <polacek@redhat.com>
2678 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
2679 (fold_negate_expr): New wrapper.
2681 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
2683 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
2684 Correct terminology and de-emphasize pre-standard behavior.
2686 2017-02-16 Alan Modra <amodra@gmail.com>
2688 PR rtl-optimization/79286
2689 * ira.c (def_dominates_uses): New function.
2690 (update_equiv_regs): Don't create an equivalence for insns that
2691 may trap where the register def does not dominate the use.
2693 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
2695 PR rtl-optimization/78127
2696 * lra.c (lra): Call lra_eliminate before finish the loop after
2699 2017-02-16 Richard Biener <rguenther@suse.de>
2701 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
2703 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
2704 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
2705 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
2706 (isl_val_int_from_wi): New function.
2707 (extract_affine_gmp): Rename to ...
2708 (extract_affine_wi): ... this, take a widest_int.
2709 (extract_affine_int): Just wrap extract_affine_wi.
2710 (add_param_constraints): Use isl_val_int_from_wi.
2711 (add_loop_constraints): Likewise, and extract_affine_wi.
2713 2017-02-15 Jeff Law <law@redhat.com>
2716 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
2717 ira_init_register_move_cost_if_necessary.
2719 2017-02-15 Martin Sebor <msebor@redhat.com>
2722 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
2723 removed in a prior commit.
2725 2017-02-15 Bin Cheng <bin.cheng@arm.com>
2727 PR tree-optimization/79347
2728 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
2729 counters during peeling.
2731 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
2733 * Makefile.in (site.exp): Remove "set ISLVER".
2735 2017-02-15 Jakub Jelinek <jakub@redhat.com>
2738 * real.c (real_from_integer): Call real_convert even for decimal.
2740 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
2743 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
2745 2017-02-14 Andrew Pinski <apinski@cavium.com>
2747 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
2748 cores and change the partno/implementer to be correct.
2749 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
2750 the 'B" as the implementer.
2751 * config/aarch64/aarch64-tune.md: Regenerate.
2753 2017-02-14 Carl Love <cel@us.ibm.com>
2755 * config/rs6000/rs6000.c: Add case statement entry to make the
2756 xvcvuxdsp built-in argument unsigned.
2757 * config/rs6000/vsx.md: Fix the source and return operand types so they
2758 match the instruction definitions from the ISA document. Fix typo
2759 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
2762 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
2765 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
2766 member early_clobber_alts.
2767 * lra-lives.c (reg_early_clobber_p): New.
2768 (process_bb_lives): Use it.
2769 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
2770 (debug_operand_data): Initialize early_clobber_alts.
2771 (setup_operand_alternative): Set up early_clobber_alts.
2772 (collect_non_operand_hard_regs): Ditto. Pass early clobber
2773 alternatives to new_insn_reg.
2774 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
2776 (lra_update_insn_regno_info): Pass the new arg.
2778 2017-02-14 Jakub Jelinek <jakub@redhat.com>
2781 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
2782 (new_oacc_loop_raw): Don't clear already cleared fields.
2785 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
2786 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
2787 _mm512_prefetch_i64gather_ps): New inline functions and macros.
2789 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
2792 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
2794 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
2797 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
2798 the extra instruction to the right place to store 128-bit constant
2801 2017-02-14 Martin Sebor <msebor@redhat.com>
2804 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
2805 warning for strings of unknown length.
2807 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
2809 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
2811 2017-02-14 Jeff Law <law@redhat.com>
2814 * ira-costs.c (scan_one_insn): Initialize register move costs
2815 for pseudos seen in USE/CLOBBER insns.
2817 PR tree-optimization/79095
2818 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
2819 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
2820 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
2821 if the operands are known to be not equal, then the resulting range
2823 (intersect_ranges): If the new range is ~[0,0] and the old range is
2824 wide, then prefer ~[0,0].
2825 * tree-vrp.c (overflow_comparison_p_1): New function.
2826 (overflow_comparison_p): New function.
2827 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
2828 if NAME is used in an overflow test.
2829 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
2830 overflow check that can be expressed as an equality test, then adjust
2831 ops to be that equality test.
2833 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2835 * config/s390/s390-builtin-types.def: Remove flags argument.
2836 * config/s390/s390.c (s390_init_builtins): Likewise.
2838 2017-02-14 Martin Liska <mliska@suse.cz>
2840 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
2841 vector. Fix trailing white spaces.
2843 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
2845 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
2848 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2850 PR rtl-optimization/68664
2851 * config/arm/arm.c (arm_sched_can_speculate_insn):
2852 New function. Declare prototype.
2853 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
2855 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2857 PR rtl-optimization/68664
2858 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
2860 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
2862 2017-02-14 Amit Pawar <amit.pawar@amd.com>
2864 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
2865 max skip bytes for function, loop and jump.
2867 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2869 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
2870 ABS_EXPR for gimple dump.
2872 2017-02-14 Jakub Jelinek <jakub@redhat.com>
2875 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
2877 PR tree-optimization/79408
2878 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
2879 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
2880 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
2881 also if rhs1 is INTEGER_CST.
2883 2017-02-14 Richard Biener <rguenther@suse.de>
2886 * tree-into-ssa.c (insert_phi_nodes): When the function can
2887 have abnormal edges rewrite SSA names with broken use-def
2888 dominance out of SSA and register them for PHI insertion.
2890 2017-02-13 Martin Sebor <msebor@redhat.com>
2893 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
2894 clearing info.nowrite flag when snprintf size argument is a range.
2896 2017-02-13 Jakub Jelinek <jakub@redhat.com>
2898 * cprop.c (cprop_jump): Add missing space in string literal.
2899 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
2900 (get_constraint_for_component_ref): Likewise.
2901 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
2902 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
2903 * lra-constraints.c (process_alt_operands): Likewise.
2904 * ipa-inline.c (inline_small_functions): Likewise.
2905 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
2906 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
2907 * trans-mem.c (diagnose_tm_1_op): Likewise.
2908 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
2909 (grid_parallel_clauses_gridifiable): Likewise.
2911 * config/nvptx/mkoffload.c (process): Add space in between
2914 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
2915 "MOD4_SSE_REGS" and "ALL_REGS".
2917 * spellcheck.c (test_data): Add , in between "foo" and "food".
2919 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2922 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
2923 boundary crossing check and subsequent code generation agree.
2925 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2927 * config/aarch64/aarch64.c (has_memory_op): Delete.
2928 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
2931 2017-02-13 Jakub Jelinek <jakub@redhat.com>
2933 PR rtl-optimization/79388
2934 PR rtl-optimization/79450
2935 * combine.c (distribute_notes): When removing TEM_INSN for which
2936 corresponding dest has last value recorded, invalidate that last
2939 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2941 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
2942 of explicit '@'. Add missing assembly comment marker on branch costs
2945 2017-02-13 Nathan Sidwell <nathan@acm.org>
2947 * gengtype-lex.l (<in_struct>): Add '/'.
2949 2017-02-13 Martin Liska <mliska@suse.cz>
2952 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
2954 2017-02-13 Richard Biener <rguenther@suse.de>
2956 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
2958 * configure: Re-generate.
2959 * config.in: Likewise.
2960 * graphite-dependences.c: Simplify as if
2961 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
2962 * graphite-isl-ast-to-gimple.c: Likewise.
2963 * graphite-optimize-isl.c: Likewise.
2964 * graphite-poly.c: Likewise.
2965 * graphite-sese-to-poly.c: Likewise.
2966 * graphite.h: Likewise.
2967 * toplev.c: Include isl/version.h and use isl_version () for
2968 printing the ISL version.
2969 * doc/install.texi: Update ISL requirement.
2971 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
2973 * doc/standards.texi (Standards): Update reference to
2976 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
2978 * doc/extend.texi (Named Address Spaces): sourceware.org now
2980 * doc/install.texi (Binaries): Ditto.
2983 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
2985 * doc/cpp.texi: Replace "stringify"/"stringification" with C
2986 standard terminology "stringize"/"stringizing" throughout.
2987 * doc/cppinternals.texi: Likewise.
2989 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
2991 * doc/extend.texi: Fix some spelling mistakes and typos.
2992 * doc/invoke.texi: Likewise.
2994 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
2997 * params.def (inline-min-speedup) Change from 10 to 8.
2999 2017-02-11 Jakub Jelinek <jakub@redhat.com>
3001 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
3004 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
3007 * ipa-inline-analysis.c (get_minimal_bb): New function.
3008 (record_modified): Use it.
3009 (remap_edge_change_prob): Handle also ancestor functions.
3011 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
3013 * doc/contrib.texi (Contributors): Remove broken link into
3014 the Mauve CVS repository.
3016 2017-02-11 Jakub Jelinek <jakub@redhat.com>
3019 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
3020 result computation whenever lhs doesn't have vector mode, not
3021 just when it has BLKmode.
3023 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
3025 * doc/makefile.texi (profiledbootstrap): Refer to the
3026 installation instructions only in textual form.
3028 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3031 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
3033 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
3035 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
3036 (Specific): Update mingw-w64 reference.
3038 (Specific): Remove broken link to Renesas RX processor.
3040 2017-02-10 Richard Biener <rguenther@suse.de>
3042 * toplev.c (process_options): Do not mention obsolete graphite
3043 options when printing sorry message about missing graphite support.
3044 Mention -floop-nest-optimize.
3046 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
3048 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
3049 (vtst_p16): Likewise.
3050 (vtstq_p8): Likewise.
3051 (vtstq_p16): Likewise.
3053 (vtstq_p64): Likewise.
3054 * config/arm/arm_neon.h (vgetq_lane_p64): New.
3055 (vset_lane_p64): New.
3056 (vsetq_lane_p64): New.
3058 2017-02-10 Jakub Jelinek <jakub@redhat.com>
3060 PR tree-optimization/79411
3061 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
3062 stmt operands are SSA_NAMEs used in abnormal phis.
3063 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
3066 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
3069 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
3072 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
3074 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
3076 2017-02-09 Jakub Jelinek <jakub@redhat.com>
3078 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
3082 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
3083 not arbitrary TREE_CONSTANT.
3086 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
3087 "omp declare target link" attribute unless is_global_var.
3088 * omp-offload.c (find_link_var_op): Likewise.
3090 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
3091 Chung-Lin Tang <cltang@codesourcery.com>
3093 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
3095 (gimplify_adjust_omp_clauses): Don't delete TILE.
3096 (gimplify_omp_for): Deal with TILE.
3097 * internal-fn.c (expand_GOACC_TILE): New function.
3098 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
3100 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
3101 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
3103 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
3104 avoid DIV for outermost collapse var.
3105 (expand_oacc_for): Insert tile element loop as needed. Adjust.
3106 Remove out of date comments, fix whitespace.
3107 * omp-general.c (omp_extract_for_data): Deal with tiling.
3108 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
3109 adjust OLF_DIM_BASE value.
3110 (struct omp_for_data): Add tiling field.
3111 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
3112 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
3113 for auto loops. Remove default auto determining, moved to
3114 oacc_loop_fixed_partitions.
3115 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
3116 stmts, add e_mask field.
3117 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
3118 (oacc_thread_numbers): Use oacc_dim_call.
3119 (oacc_xform_tile): New.
3120 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
3121 (finish_oacc_loop): Adjust for ifns vector.
3122 (oacc_loop_discover_walk): Append loop abstraction sites to list,
3123 add case for GOACC_TILE fns.
3124 (oacc_loop_xform_loop): Delete.
3125 (oacc_loop_process): Iterate over call list directly, and add
3126 handling for GOACC_TILE fns.
3127 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
3129 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
3130 vector partitioning to outer loops. Assign 2 partitions to loops
3131 when available. Add TILE handling.
3132 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
3133 (execite_oacc_device_lower): Process GOACC_TILE fns, ignore unknown specs.
3134 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
3135 * tree.c (omp_clause_num_ops): Adjust TILE ops.
3136 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
3138 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
3140 * configure.ac (ACX_BUGURL): Update.
3141 * configure: Regenerate.
3143 2017-02-09 Richard Biener <rguenther@suse.de>
3145 PR tree-optimization/69823
3146 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
3147 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
3149 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
3151 * config/arc/arc-c.def: Add __NPS400__ definition.
3152 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
3153 (TARGET_NPS400): Define.
3155 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
3157 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
3159 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
3161 (arc_cpu_types): Fill the arch_info field with a pointer into the
3162 arc_arch_types table.
3163 (arc_selected_cpu): Declare.
3164 * config/arc/arc.c (arc_selected_cpu): Make global.
3165 (arc_selected_arch): Delete.
3166 (arc_base_cpu): Delete.
3167 (arc_override_options): Remove references to deleted variables,
3168 update access to arch information.
3169 (ARC_OPT): Update access to arch information.
3170 (ARC_OPTX): Likewise.
3171 * config/arc/arc.h (arc_base_cpu): Remove declaration.
3172 (TARGET_ARC600): Update access to arch information.
3173 (TARGET_ARC601): Likewise.
3174 (TARGET_ARC700): Likewise.
3175 (TARGET_EM): Likewise.
3176 (TARGET_HS): Likewise.
3177 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
3180 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
3183 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
3184 condition/operands for integer GE/LE/GEU/LEU operations.
3186 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
3188 PR translation/79397
3189 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
3192 2017-02-08 Martin Jambor <mjambor@suse.cz>
3195 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
3196 whether allocation happened.
3197 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
3198 nothing was allocated.
3200 2017-02-08 Jakub Jelinek <jakub@redhat.com>
3202 PR tree-optimization/79408
3203 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
3204 constant, but SSA_NAME with a known integer range, use the minimum
3205 of that range instead of op1 to determine if modulo can be replaced
3206 with its first operand.
3208 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3210 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
3212 2017-02-08 Richard Biener <rguenther@suse.de>
3214 PR tree-optimization/71824
3215 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
3216 Check all loops contained in the merged region.
3218 2017-02-07 Andrew Pinski <apinski@cavium.com>
3220 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
3222 2017-02-07 Andrew Pinski <apinski@cavium.com>
3224 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
3225 (thunderxt88): Likewise.
3226 (thunderxt81): Disable LSE and change v8.1 to v8.
3227 (thunderxt83): Likewise.
3229 2017-02-07 Jakub Jelinek <jakub@redhat.com>
3230 Richard Biener <rguenther@suse.de>
3233 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
3234 type from int to size_t.
3235 * ira-costs.c (struct_costs_size): Change type from int to size_t.
3237 2017-02-07 Jakub Jelinek <jakub@redhat.com>
3239 PR rtl-optimization/79386
3240 * cprop.c (bypass_conditional_jumps): Initialize
3241 bypass_last_basic_block already before splitting bbs after
3242 unconditional traps...
3243 (bypass_conditional_jumps): ... rather than here.
3246 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
3247 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
3248 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
3249 fix -masm=intel patterns.
3251 2017-02-07 Richard Biener <rguenther@suse.de>
3253 PR tree-optimization/79256
3255 * builtins.c (get_object_alignment_2): Use min_align_of_type
3256 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
3257 and ADJUST_FIELD_ALIGN.
3259 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
3261 * doc/tm.texi: Regenerate.
3262 * stor-layout.c (layout_decl): Adjust.
3263 (update_alignment_for_field): Likewise.
3264 (place_field): Likewise.
3265 (min_align_of_type): Likewise.
3266 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
3267 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
3268 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
3269 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
3270 * config/frv/frv.c (frv_adjust_field_align): Likewise.
3271 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
3272 * config/i386/i386.c (x86_field_alignment): Likewise.
3273 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
3274 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
3275 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
3276 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
3277 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
3278 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
3282 2017-01-30 Richard Biener <rguenther@suse.de>
3284 PR tree-optimization/79256
3285 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
3286 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
3289 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
3291 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
3292 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
3293 builtins to SImode and emit a zero-extend, if necessary.
3295 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
3297 * docs/invoke.texi (RISC-V Options): Alphabetize.
3299 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
3301 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
3304 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
3306 * config/riscv/riscv.c: New file.
3307 * gcc/common/config/riscv/riscv-common.c: Likewise.
3308 * config.gcc: Likewise.
3309 * config/riscv/constraints.md: Likewise.
3310 * config/riscv/elf.h: Likewise.
3311 * config/riscv/generic.md: Likewise.
3312 * config/riscv/linux.h: Likewise.
3313 * config/riscv/multilib-generator: Likewise.
3314 * config/riscv/peephole.md: Likewise.
3315 * config/riscv/pic.md: Likewise.
3316 * config/riscv/predicates.md: Likewise.
3317 * config/riscv/riscv-builtins.c: Likewise.
3318 * config/riscv/riscv-c.c: Likewise.
3319 * config/riscv/riscv-ftypes.def: Likewise.
3320 * config/riscv/riscv-modes.def: Likewise.
3321 * config/riscv/riscv-opts.h: Likewise.
3322 * config/riscv/riscv-protos.h: Likewise.
3323 * config/riscv/riscv.h: Likewise.
3324 * config/riscv/riscv.md: Likewise.
3325 * config/riscv/riscv.opt: Likewise.
3326 * config/riscv/sync.md: Likewise.
3327 * config/riscv/t-elf-multilib: Likewise.
3328 * config/riscv/t-linux: Likewise.
3329 * config/riscv/t-linux-multilib: Likewise.
3330 * config/riscv/t-riscv: Likewise.
3331 * configure.ac: Likewise.
3332 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
3333 Waterman as RISC-V maintainers.
3334 * doc/install.texi: Add RISC-V entries.
3335 * doc/invoke.texi: Add RISC-V options section.
3336 * doc/md.texi: Add RISC-V constraints section.
3337 * configure: Regenerated.
3339 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
3342 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
3343 false values to be constant vectors with all 0 or all 1 bits set.
3344 (vcondu<mode><mode>): Likewise.
3345 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
3347 (fpmask_comparison_operator): Update comment.
3348 (vecint_comparison_operator): New predicate.
3349 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
3350 vector conditionals when the true and false values are constant
3351 vectors with all 0 bits or all 1 bits set.
3353 2017-02-06 Martin Sebor <msebor@redhat.com>
3355 PR tree-optimization/79376
3356 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
3358 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
3360 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
3361 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
3362 to simplify split condition.
3364 2017-02-06 Jakub Jelinek <jakub@redhat.com>
3366 * omp-expand.c (oxpand_omp_atomic_fetch_op,
3367 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
3370 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
3372 PR rtl-optimization/68664
3373 * target.def (can_speculate_insn): New hook.
3374 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
3375 * doc/tm.texi: Regenerate.
3376 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
3377 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
3378 (rs6000_sched_can_speculate_insn): New function.
3380 2017-02-06 Jakub Jelinek <jakub@redhat.com>
3382 PR tree-optimization/79284
3383 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
3384 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
3385 vectorizable_mask_load_store, vectorizable_operation,
3386 vect_is_simple_cond, get_same_sized_vectype): Use it instead
3387 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
3388 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
3389 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
3390 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
3391 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
3392 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
3393 is_gimple_assign (stmt). Replace another such test with
3394 is_gimple_assign (stmt).
3396 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
3399 * config/avr/avr.c (rtl-iter.h): Include it.
3400 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
3401 (avr_legitimate_combined_insn): ...and implementation.
3403 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
3405 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
3406 * config/s390/s390.c (s390_const_operand_ok)
3407 (s390_canonicalize_comparison, s390_extract_part)
3408 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
3409 (s390_contiguous_bitmask_p, s390_rtx_costs)
3410 (legitimize_pic_address): Likewise.
3411 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
3412 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
3413 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
3414 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
3415 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
3417 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
3419 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
3420 REGNO($0) == REGNO($1).
3422 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3424 * config/s390/linux.h(SIZE_TYPE): Add comment.
3426 2017-02-06 Julian Brown <julian@codesourcery.com>
3427 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
3428 Virendra Pathak <virendra.pathak@broadcom.com>
3430 * config/aarch64/aarch64-cores.def: Change the scheduler
3432 * config/aarch64/aarch64.md: Include thunderx2t99.md.
3433 * config/aarch64/thunderx2t99.md: New file.
3435 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
3437 * doc/standards.texi (Go Language): Update link to language
3440 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
3442 * tree-eh.c (lower_resx): Sanitize profile.
3443 (cleanup_empty_eh_move_lp): Likewise.
3445 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
3448 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
3450 * cfgloopmanip.h (loop_version): Update prototype.
3451 * modulo-sched.c (sms_schedule): Update call of loop_version.
3452 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
3453 * tree-parloops.c (gen_parallel_loop): Likewise.
3454 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
3455 * tree-ssa-loop-split.c (split_loop): Likewise.
3456 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
3457 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
3459 2017-02-05 Martin Liska <mliska@suse.cz>
3462 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
3464 (print_operand_address): Initialize a struct to zero.
3466 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
3468 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
3469 garbage collector only in textual form.
3471 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
3473 * doc/extend.texi (x86 specific memory model extensions for
3474 transactional memory): Simplify a phrase.
3476 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
3479 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
3480 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
3481 (atomic_storedi_1): Likewise.
3483 2017-02-04 Jakub Jelinek <jakub@redhat.com>
3485 PR tree-optimization/79338
3486 * tree-parloops.c (gather_scalar_reductions): Don't call
3487 vect_analyze_loop_form for loop->inner before destroying loop's
3490 2017-02-03 Martin Sebor <msebor@redhat.com>
3492 PR tree-optimization/79327
3493 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
3494 when precision has resulted in leading zeros.
3495 (format_integer): Adjust the likely counter to assume an unknown
3496 argument that may be zero is non-zero.
3498 2017-02-03 Jason Merrill <jason@redhat.com>
3501 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
3502 avoid copying non-taken branch.
3504 2017-02-03 Jakub Jelinek <jakub@redhat.com>
3506 PR tree-optimization/79340
3507 * tree-vect-loop.c (vectorizable_reduction): Release
3508 vec_defs elements after safe_splicing them into other vectors.
3511 PR tree-optimization/79327
3512 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
3513 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
3515 (format_integer): Use wide_int_to_tree instead of build_int_cst
3516 + to_?hwi. If argmin is NULL, just set argmin and argmax to
3517 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
3518 of shortest and longest sequence.
3520 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
3522 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
3523 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
3525 2017-02-03 Walter Lee <walt@tilera.com>
3528 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
3529 after initial stackframe link reg save.
3530 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
3532 2017-02-03 Jakub Jelinek <jakub@redhat.com>
3535 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
3536 wu for stxssp alternative.
3538 2017-02-03 Martin Sebor <msebor@redhat.com>
3540 PR tree-optimization/79352
3541 * gimple-fold.c (get_range_strlen): Add argument.
3542 (get_range_strlen): Change return type to bool.
3543 (get_maxval_strlen): Pass in a dummy argument.
3544 * gimple-fold.h (get_range_strlen): Change return type to bool.
3545 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
3546 * tree.h (array_at_struct_end_p): Add argument.
3547 * tree.c (array_at_struct_end_p): Handle it.
3549 2017-02-03 Martin Liska <mliska@suse.cz>
3552 * multiple_target.c (create_dispatcher_calls): Redirect edge
3553 from a caller of a dispatcher.
3554 (expand_target_clones): Make the clones local.
3555 (ipa_target_clone): Do both target clones and resolvers.
3556 (ipa_dispatcher_calls): Remove the pass.
3557 (pass_dispatcher_calls::gate): Likewise.
3558 (make_pass_dispatcher_calls): Likewise.
3559 * passes.def (pass_target_clone): Put as very first IPA early
3562 2017-02-03 Martin Liska <mliska@suse.cz>
3564 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
3565 in case of a function with ifunc attribute.
3567 2017-02-03 Martin Liska <mliska@suse.cz>
3569 * cgraph.c (cgraph_node::dump): Dump function version info.
3570 * symtab.c (symtab_node::dump_base): Add missing new line.
3572 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
3574 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
3575 (ifcombine_ifandif): Use it.
3577 2017-02-03 Martin Liska <mliska@suse.cz>
3579 * doc/invoke.texi: Document default value for
3580 use-after-scope-direct-emission-threshold.
3582 2017-02-03 Martin Liska <mliska@suse.cz>
3584 PR tree-optimization/79339
3585 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
3586 (format_floating): Likewise.
3588 2017-02-03 Martin Liska <mliska@suse.cz>
3591 * ipa-prop.c (ipa_node_params_t::insert): Remove current
3593 (ipa_node_params_t::remove): Likewise.
3594 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
3595 initialization from removed ipa_node_params_t::insert.
3596 (ipa_node_params::~ipa_node_params): Move from removed
3597 ipa_node_params_t::release.
3598 * symbol-summary.h (symbol_summary::m_released): New member.
3599 Do not release a summary twice. Do not allow to call finalizer
3600 for types of a summary that live in GGC memory.
3602 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
3604 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
3607 2017-02-02 Martin Sebor <msebor@redhat.com>
3610 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
3611 (format_string): Tighten up the range of output for non-constant
3612 strings and correct the expected range for wide non-constant strings.
3614 2017-02-02 Martin Sebor <msebor@redhat.com>
3616 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
3619 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
3621 (-fdump-tree-@var): Add to index and document how to come up
3622 with pass-specific option and dump file names.
3623 (-fdump-passes): Clarify where to look for output.
3625 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
3628 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
3629 statistics of the analyzed path; allow threading for speed when
3630 any of BBs along the path are optimized for speed.
3632 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
3635 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
3636 settings of the virtual registers.
3639 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
3641 * explow.c (get_dynamic_stack_size): Take known alignment of stack
3642 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
3645 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3647 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
3648 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
3650 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3652 * config/s390/s390.md: Add missing comments with the expanded
3654 * config/s390/vector.md: Likewise.
3655 * config/s390/vx-builtins.md: Likewise.
3657 2017-02-02 Jakub Jelinek <jakub@redhat.com>
3660 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
3661 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
3662 conditions on a single line.
3664 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3666 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
3667 __S390_VX__ to __VX__.
3669 2017-02-01 Andrew Pinski <apinski@cavium.com>
3671 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
3672 stmt_info to record_stmt_cost.
3673 (vect_get_known_peeling_cost): Pass stmt_info if known to
3675 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
3676 cpu_vector_cost field into
3677 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
3678 field into vec_int_stmt_cost and vec_fp_stmt_cost.
3679 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
3680 splitting of scalar_stmt_cost and vec_stmt_cost.
3681 (thunderx_vector_cost): Likewise.
3682 (cortexa57_vector_cost): LIkewise.
3683 (exynosm1_vector_cost): Likewise.
3684 (xgene1_vector_cost): Likewise.
3685 (thunderx2t99_vector_cost): Improve after the splitting of the two
3687 (aarch64_builtin_vectorization_cost): Update for the splitting of
3688 scalar_stmt_cost and vec_stmt_cost.
3690 2017-02-01 Torvald Riegel <triegel@redhat.com>
3691 Richard Henderson <rth@redhat.com>
3693 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
3694 conditional on existance of a fast atomic load.
3695 * optabs-query.c (can_atomic_load_p): New function.
3696 * optabs-query.h (can_atomic_load_p): Declare it.
3697 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
3698 no fast atomic load is available for the particular size of access.
3699 (expand_atomic_compare_and_swap): Likewise.
3700 (expand_atomic_load): Likewise.
3701 (expand_atomic_store): Likewise.
3702 (expand_atomic_fetch_op): Likewise.
3703 * testsuite/lib/target-supports.exp
3704 (check_effective_target_sync_int_128): Remove x86 because it provides
3705 no fast atomic load.
3706 (check_effective_target_sync_int_128_runtime): Likewise.
3708 2017-02-01 Richard Biener <rguenther@suse.de>
3710 * graphite.c: Include tree-vectorizer.h for find_loop_location.
3711 (graphite_transform_loops): Provide opt-info for optimized nests.
3712 * tree-parloop.c (parallelize_loops): Provide opt-info for
3715 2017-02-01 Richard Biener <rguenther@suse.de>
3718 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
3721 2017-02-01 Richard Biener <rguenther@suse.de>
3723 PR tree-optimization/71824
3724 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
3725 Verify the loops are valid in the merged SESE region.
3726 (scop_detection::can_represent_loop_1): Check analyzing the
3727 evolution of the number of iterations in the region succeeds.
3729 2017-01-31 Ian Lance Taylor <iant@golang.org>
3731 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
3732 REG_ARGS_SIZE note to 32-bit push insns and call insn.
3734 2017-01-31 David Malcolm <dmalcolm@redhat.com>
3736 PR preprocessor/79210
3737 * input.c (get_substring_ranges_for_loc): Replace line_width
3738 assertion with error-handling.
3740 2017-01-31 Richard Biener <rguenther@suse.de>
3742 PR tree-optimization/77318
3743 * graphite-sese-to-poly.c (extract_affine): Fix assert.
3744 (create_pw_aff_from_tree): Take loop parameter.
3745 (add_condition_to_pbb): Pass loop of the condition to
3746 create_pw_aff_from_tree.
3748 2017-01-31 Jakub Jelinek <jakub@redhat.com>
3750 * config/s390/s390.c (s390_asan_shadow_offset): New function.
3751 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
3753 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
3757 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
3759 (convert_int_to_float128): Likewise.
3760 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
3761 (convert_int_to_float128): Likewise.
3762 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
3763 (UNSPEC_IEEE128_CONVERT): Likewise.
3764 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
3765 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
3766 Use local variables for IBM extended format.
3767 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
3768 (fix_trunc<mode>si2_fprs): Likewise.
3769 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
3770 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
3771 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
3772 to know that we can now have integers of all sizes in vector
3774 (fix<uns>_<mode>di2_hw): Likewise.
3775 (float<uns>_<mode>si2_hw): Likewise.
3776 (fix_<mode>si2_hw): Likewise.
3777 (fixuns_<mode>si2_hw): Likewise.
3778 (float<uns>_<mode>di2_hw): Likewise.
3779 (float_<mode>di2_hw): Likewise.
3780 (float_<mode>si2_hw): Likewise.
3781 (floatuns_<mode>di2_hw): Likewise.
3782 (floatuns_<mode>si2_hw): Likewise.
3783 (xscvqp<su>wz_<mode>): Delete, no longer used.
3784 (xscvqp<su>dz_<mode>): Likewise.
3785 (xscv<su>dqp_<mode>): Likewise.
3786 (ieee128_mfvsrd_64bit): Likewise.
3787 (ieee128_mfvsrd_32bit): Likewise.
3788 (ieee128_mfvsrwz): Likewise.
3789 (ieee128_mtvsrw): Likewise.
3790 (ieee128_mtvsrd_64bit): Likewise.
3791 (ieee128_mtvsrd_32bit): Likewise.
3793 2017-01-31 Martin Liska <mliska@suse.cz>
3796 * ipa-prop.c (ipa_free_all_node_params): Call release method
3797 instead of ~sumbol_summary to not to trigger double times
3800 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
3802 PR tree-optimization/71691
3803 * bitmap.h (class auto_bitmap): New.
3804 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
3805 is_maybe_undefined instead of ssa_undefined_value_p.
3807 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3809 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
3810 __S390_ARCH_LEVEL__ to __ARCH__.
3812 2017-01-31 Jakub Jelinek <jakub@redhat.com>
3814 PR tree-optimization/79267
3815 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
3816 if should_remove_lhs_p is true.
3818 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
3821 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
3822 (add_alignment_attribute): New.
3823 (base_type_die): Add alignment attribute.
3824 (subrange_type_die): Likewise.
3825 (modified_type_die): Likewise.
3826 (gen_array_type_die): Likewise.
3827 (gen_descr_array_type_die: Likewise.
3828 (gen_enumeration_type_die): Likewise.
3829 (gen_subprogram_die): Likewise.
3830 (gen_variable_die): Likewise.
3831 (gen_field_die): Likewise.
3832 (gen_ptr_to_mbr_type_die): Likewise.
3833 (gen_struct_or_union_type_die): Likewise.
3834 (gen_subroutine_type_die): Likewise.
3835 (gen_typedef_die): Likewise.
3836 (base_type_cmp): Compare alignment attribute.
3838 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3841 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
3842 (setb_unsigned) New pattern for setb with CCUNS.
3843 * config/rs6000/rs6000.c (expand_block_compare): Use a different
3844 subfc./subfe sequence to avoid overflow problems. Generate a
3845 shorter sequence with cmpld/setb for power9.
3846 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
3847 for generating subfc. instruction.
3848 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
3849 now uses this instruction.
3851 2017-01-30 Ian Lance Taylor <iant@google.com>
3854 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
3855 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
3857 2017-01-30 Martin Sebor <msebor@redhat.com>
3859 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
3860 Move constant to the right of a relational operator.
3861 (get_mpfr_format_length, format_character, format_string): Ditto.
3862 (should_warn_p, maybe_warn): Same.
3864 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
3866 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
3869 * asan.c (get_translation_unit_decl): Remove function.
3870 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
3872 2017-01-30 Martin Liska <mliska@suse.cz>
3874 PR gcov-profile/79259
3875 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
3878 2017-01-30 Martin Liska <mliska@suse.cz>
3881 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
3882 Initialize variables with NULL value.
3884 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
3887 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
3889 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
3891 2017-01-30 Richard Biener <rguenther@suse.de>
3893 PR tree-optimization/79276
3894 * tree-vrp.c (process_assert_insertions): Properly adjust common
3895 when removing a duplicate.
3897 * gcc.dg/torture/pr79276.c: New testcase.
3899 2017-01-30 Richard Biener <rguenther@suse.de>
3901 PR tree-optimization/79256
3902 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
3903 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
3905 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
3907 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
3910 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
3911 ("*r<noxa>sbg_<mode>_sll_bitmask")
3912 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
3913 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
3914 Use contiguous_bitmask_nowrap_operand.
3916 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3919 * config/rs6000/altivec.h (vec_xl): Revise #define.
3920 (vec_xst): Likewise.
3922 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
3924 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
3926 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
3928 PR rtl-optimization/79194
3929 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
3930 traps before call to bypass_conditional_jumps.
3932 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
3934 PR tree-optimization/71374
3935 * lra-constraints.c (check_conflict_input_operands): New.
3936 (match_reload): Use it.
3938 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
3941 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
3942 account to calculate conflict_set.
3944 2017-01-27 Bin Cheng <bin.cheng@arm.com>
3946 PR rtl-optimization/78559
3947 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
3948 other_insn in combine.
3950 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
3952 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
3953 uint16_type_node for BT_UINT16.
3955 2017-01-27 David Malcolm <dmalcolm@redhat.com>
3957 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
3958 "RTL Tests" to menu.
3959 (GIMPLE Tests): New node.
3960 (RTL Tests): New node.
3962 2017-01-27 Richard Biener <rguenther@suse.de>
3964 PR tree-optimization/79245
3965 * tree-loop-distribution.c (distribute_loop): Apply cost
3966 modeling also to detected patterns.
3968 2017-01-27 Richard Biener <rguenther@suse.de>
3970 PR tree-optimization/71433
3971 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
3972 (compare_assert_loc): New function.
3973 (process_assert_insertions): Sort and optimize assert locations
3974 to remove duplicates and push down identical assertions on
3975 edges to their destination block.
3977 2017-01-27 Richard Biener <rguenther@suse.de>
3979 PR tree-optimization/79244
3980 * tree-vrp.c (remove_range_assertions): Forcefully propagate
3981 out SSA names even if abnormal.
3983 2017-01-27 Jakub Jelinek <jakub@redhat.com>
3985 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
3986 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
3987 instead of MPFR_RNDN.
3989 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
3992 * arm.c (arm_option_override): Don't call build_target_option_node
3993 until after doing all option overrides.
3994 (arm_valid_target_attribute_tree): Likewise.
3996 2017-01-27 Martin Liska <mliska@suse.cz>
3998 * doc/invoke.texi (-fprofile-arcs): Document profiling support
3999 for {cd}tors and C++ {cd}tors.
4001 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
4003 * config/s390/s390.md ("*setmem_long_and")
4004 ("*setmem_long_and_31z"): Use zero_extend instead of and.
4006 2017-01-26 Martin Sebor <msebor@redhat.com>
4008 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
4011 2017-01-26 Martin Sebor <msebor@redhat.com>
4013 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
4014 HAVE_DFmode before using XFmode or DFmode.
4015 (parse_directive): Avoid using the z length modifier to avoid
4016 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
4019 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
4020 to accept adjustment as an array.
4021 (get_int_range): New function.
4022 (struct directive): Make width and prec arrays.
4023 (directive::set_width, directive::set_precision): Call get_int_range.
4024 (format_integer, format_floating): Handle width and precision ranges.
4025 (format_string, parse_directive): Same.
4027 2017-01-26 Jakub Jelinek <jakub@redhat.com>
4030 * dwarf2out.c (generate_skeleton_bottom_up): For children with
4031 comdat_type_p set, just clone them, but keep the children in the
4035 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
4036 which have direct callers with -fvar-tracking-assignments enabled
4038 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
4039 inside of type units.
4041 2017-01-26 Martin Sebor <msebor@redhat.com>
4044 * gimple-ssa-sprintf.c (struct result_range): Add likely and
4046 (struct format_result): Replace number_chars, number_chars_min,
4047 and number_chars_max with a single member of struct result_range.
4049 (format_result::operator+=): Adjust.
4050 (struct fmtresult): Remove bounded. Handle likely and unlikely
4052 (fmtresult::adjust_for_width_or_precision): New function.
4053 (fmtresult:type_max_digits): New function.
4054 (bytes_remaining): Handle likely and unlikely counters.
4055 (min_bytes_remaining): Remove.
4056 (format_percent): Simplify.
4057 (format_integer, format_floating): Set likely and unlikely counters.
4058 (get_string_length, format_character, format_string): Same.
4059 (format_plain, should_warn_p): New function.
4060 (maybe_warn): Call should_warn_p. Update diagnostic messages
4061 and handle those for all directives, including plain strings.
4062 (format_directive): Handle likely and unlikely counters.
4063 Remove unnecessary quoting from diagnostics. Add an informational
4065 (add_bytes): Remove.
4066 (pass_sprintf_length::compute_format_length): Simplify.
4067 (try_substitute_return_value): Handle likely and unlikely counters.
4069 2017-01-26 Carl Love <cel@us.ibm.com>
4071 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
4072 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
4074 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
4077 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
4078 endianess for subregs into account.
4079 * lra-constraints.c (lra_constraints): Do risky transformations
4080 always on the first iteration.
4081 * lra-lives.c (check_pseudos_live_through_calls): Add arg
4082 last_call_used_reg_set.
4083 (process_bb_lives): Define and use last_call_used_reg_set.
4084 * lra.c (lra): Always continue after lra_constraints on the first
4087 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
4089 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
4091 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
4093 2017-01-26 Jakub Jelinek <jakub@redhat.com>
4095 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
4096 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
4097 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
4098 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
4099 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
4100 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
4101 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
4102 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
4103 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
4105 2017-01-26 Marek Polacek <polacek@redhat.com>
4108 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
4109 for the third operand.
4111 2017-01-26 Jakub Jelinek <jakub@redhat.com>
4114 * omp-low.c (struct omp_context): Add simt_stmt field.
4115 (scan_omp_for): Return omp_context *.
4116 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
4117 context to the _simt_ SIMD stmt.
4118 (lower_omp_for): For combined SIMD with sibling _simt_
4119 SIMD, make sure to use the same decls in _looptemp_
4120 clauses as in the sibling.
4122 2017-01-26 David Sherwood <david.sherwood@arm.com>
4125 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
4128 2017-01-26 Jakub Jelinek <jakub@redhat.com>
4131 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
4132 emit fld b; fld a; if possible.
4134 * brig-builtins.def: Update copyright years.
4135 * config/arm/arm_acle_builtins.def: Update copyright years.
4137 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
4140 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
4141 constraint instead of o for the stxsd instruction.
4143 2017-01-25 Carl Love <cel@us.ibm.com>
4145 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
4146 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
4148 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
4150 * doc/invoke.texi (C++ Dialect Options): Fix typo.
4152 2017-01-25 Richard Biener <rguenther@suse.de>
4154 PR tree-optimization/69264
4155 * target.def (vector_alignment_reachable): Improve documentation.
4156 * doc/tm.texi: Regenerate.
4157 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
4159 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
4160 earlier changes with respect to TYPE_USER_ALIGN.
4161 (vector_alignment_reachable_p): Likewise. Improve dumping.
4163 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4166 * config/arm/arm.md (xordi3): Force constant operand into a register
4169 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4171 * doc/invoke.texi (-fstore-merging): Correct default optimization
4172 levels at which it is enabled.
4173 (-O): Move -fstore-merging from list to...
4176 2017-01-25 Richard Biener <rguenther@suse.de>
4179 * omp-expand.c: Include debug.h.
4180 (expand_omp_taskreg): Make sure to generate early debug before
4181 outlining anything from a function.
4182 (expand_omp_target): Likewise.
4183 (grid_expand_target_grid_body): Likewise.
4185 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
4188 * asan.c (get_translation_unit_decl): New function.
4189 (asan_add_global): Extract modules file name from globals
4190 TRANSLATION_UNIT_DECL name.
4192 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
4195 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
4196 for long calls with APCS frame and VFP.
4198 2017-01-24 David Malcolm <dmalcolm@redhat.com>
4200 * cfg.c (original_copy_tables_initialized_p): New function.
4201 * cfg.h (original_copy_tables_initialized_p): New decl.
4202 * cfgrtl.c (relink_block_chain): Guard the call to
4203 free_original_copy_tables with a call to
4204 original_copy_tables_initialized_p.
4205 * cgraph.h (symtab_node::native_rtl_p): New decl.
4206 * cgraphunit.c (symtab_node::native_rtl_p): New function.
4207 (symtab_node::needed_p): Don't assert for early assembly output
4208 for __RTL functions.
4209 (cgraph_node::finalize_function): Set "force_output" for __RTL
4211 (cgraph_node::analyze): Bail out early for __RTL functions.
4212 (analyze_functions): Update assertion to support __RTL functions.
4213 (cgraph_node::expand): Bail out early for __RTL functions.
4214 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
4216 * function.h (struct function): Update comment for field
4218 * gimple-expr.c: Include "tree-pass.h".
4219 (gimple_has_body_p): Return false for __RTL functions.
4220 * Makefile.in (OBJS): Add run-rtl-passes.o.
4221 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
4223 (gcc::pass_manager::get_clean_slate): New accessor.
4224 * passes.c: Include "insn-addr.h".
4225 (should_skip_pass_p): Add logging. Update logic for running
4226 "expand" to be compatible with both __GIMPLE and __RTL. Guard
4227 property-provider override so it is only done for gimple passes.
4229 (skip_pass): New function.
4230 (execute_one_pass): Call skip_pass when skipping passes.
4231 * read-md.c (md_reader::read_char): Support filtering
4232 the input to a subset of line numbers.
4233 (md_reader::md_reader): Initialize fields
4234 m_first_line and m_last_line.
4235 (md_reader::read_file_fragment): New function.
4236 * read-md.h (md_reader::read_file_fragment): New decl.
4237 (md_reader::m_first_line): New field.
4238 (md_reader::m_last_line): New field.
4239 * read-rtl-function.c (function_reader::create_function): Only
4240 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
4241 curr_properties. Set DECL_INITIAL to a dummy block.
4242 (read_rtl_function_body_from_file_range): New function.
4243 * read-rtl-function.h (read_rtl_function_body_from_file_range):
4245 * run-rtl-passes.c: New file.
4246 * run-rtl-passes.h: New file.
4248 2017-01-24 Jeff Law <law@redhat.com>
4250 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
4253 2017-01-24 Bin Cheng <bin.cheng@arm.com>
4255 PR tree-optimization/79159
4256 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
4257 (record_nonwrapping_iv): Improve boundary using above function if no
4258 value range information.
4260 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
4261 Martin Jambor <mjambor@suse.cz>
4263 * brig-builtins.def: New file.
4264 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
4265 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
4266 (DEF_HSAIL_SAT_BUILTIN): Likewise.
4267 (DEF_HSAIL_INTR_BUILTIN): Likewise.
4268 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
4269 * builtin-types.def (BT_INT8): New.
4270 (BT_INT16): Likewise.
4271 (BT_UINT8): Likewise.
4272 (BT_UINT16): Likewise.
4273 (BT_FN_ULONG): Likewise.
4274 (BT_FN_UINT_INT): Likewise.
4275 (BT_FN_UINT_ULONG): Likewise.
4276 (BT_FN_UINT_LONG): Likewise.
4277 (BT_FN_UINT_PTR): Likewise.
4278 (BT_FN_ULONG_PTR): Likewise.
4279 (BT_FN_INT8_FLOAT): Likewise.
4280 (BT_FN_INT16_FLOAT): Likewise.
4281 (BT_FN_UINT32_FLOAT): Likewise.
4282 (BT_FN_UINT16_FLOAT): Likewise.
4283 (BT_FN_UINT8_FLOAT): Likewise.
4284 (BT_FN_UINT64_FLOAT): Likewise.
4285 (BT_FN_UINT16_UINT32): Likewise.
4286 (BT_FN_UINT32_UINT16): Likewise.
4287 (BT_FN_UINT16_UINT16_UINT16): Likewise.
4288 (BT_FN_INT_PTR_INT): Likewise.
4289 (BT_FN_UINT_PTR_UINT): Likewise.
4290 (BT_FN_LONG_PTR_LONG): Likewise.
4291 (BT_FN_ULONG_PTR_ULONG): Likewise.
4292 (BT_FN_VOID_UINT64_UINT64): Likewise.
4293 (BT_FN_UINT8_UINT8_UINT8): Likewise.
4294 (BT_FN_INT8_INT8_INT8): Likewise.
4295 (BT_FN_INT16_INT16_INT16): Likewise.
4296 (BT_FN_INT_INT_INT): Likewise.
4297 (BT_FN_UINT_FLOAT_UINT): Likewise.
4298 (BT_FN_FLOAT_UINT_UINT): Likewise.
4299 (BT_FN_ULONG_UINT_UINT): Likewise.
4300 (BT_FN_ULONG_UINT_PTR): Likewise.
4301 (BT_FN_ULONG_ULONG_ULONG): Likewise.
4302 (BT_FN_UINT_UINT_UINT): Likewise.
4303 (BT_FN_VOID_UINT_PTR): Likewise.
4304 (BT_FN_UINT_UINT_PTR: Likewise.
4305 (BT_FN_UINT32_UINT64_PTR): Likewise.
4306 (BT_FN_INT_INT_UINT_UINT): Likewise.
4307 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
4308 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
4309 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
4310 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
4311 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
4312 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
4313 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
4314 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
4315 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
4316 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
4317 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
4318 * doc/frontends.texi: List BRIG FE.
4319 * doc/install.texi (Testing): Add BRIG tesring requirements.
4320 * doc/invoke.texi (Overall Options): Mention BRIG.
4321 * doc/standards.texi (Standards): Doucment BRIG HSA version.
4323 2017-01-24 Richard Biener <rguenther@suse.de>
4325 PR translation/79208
4326 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
4328 2017-01-24 Martin Jambor <mjambor@suse.cz>
4331 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
4332 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
4335 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
4338 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
4339 casts from signed to unsigned really don't have a range.
4341 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
4343 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
4344 GMP_RNDx for compatiblity.
4346 2017-01-24 Martin Liska <mliska@suse.cz>
4349 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
4350 that would prevent us to call alloca with -1 as argument.
4352 2017-01-24 Jakub Jelinek <jakub@redhat.com>
4354 * dwarf2out.c (output_compilation_unit_header, output_file_names):
4355 Avoid -Wformat-security warning.
4357 2017-01-23 Andrew Pinski <apinski@cavium.com>
4359 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
4362 2017-01-23 Martin Sebor <msebor@redhat.com>
4365 * gimple-ssa-sprintf.c (warn_level): New global.
4366 (format_integer): Use it here and throughout the rest of the file.
4367 Use the same switch to compute sign as base.
4368 (maybe_warn): New function.
4369 (format_directive): Factor out warnings into maybe_warn.
4370 Add debugging output. Use warn_level.
4371 (add_bytes): Use warn_level.
4372 (pass_sprintf_length::compute_format_length): Add debugging output.
4373 (try_substitute_return_value): Same.
4374 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
4377 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
4378 (struct fmtresult, format_integer, format_floating): Adjust.
4379 (fmtresult::fmtresult): Set max correctly in two argument ctor.
4380 (get_string_length, format_string,format_directive): Same.
4381 (pass_sprintf_length::compute_format_length): Same.
4382 (try_substitute_return_value): Simplify slightly.
4385 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
4386 (fmtresult::operator+=): Outlined.
4387 (struct fmtresult): Add ctors.
4388 (struct conversion_spec): Rename...
4389 (struct directive): ...to this. Add and remove data members.
4390 (directive::set_width, directive::set_precision): New functions.
4391 (format_percent): Use fmtresult ctor.
4392 (get_width_and_precision): Remove.
4393 (format_integer): Make naming changes. Avoid computing width and
4395 (format_floating): Same. Adjust indentation.
4396 (format_character, format_none): New functions.
4397 (format_string): Moved character handling to format_character.
4398 (format_directive): Remove arguments, change return type.
4399 (parse_directive): New function.
4400 (pass_sprintf_length::compute_format_length): Move directive
4401 parsing to parse_directive.
4403 2017-01-23 Jakub Jelinek <jakub@redhat.com>
4405 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
4406 (assign_assembler_name_if_needed): ... this.
4407 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
4408 (assign_assembler_name_if_needed): ... this.
4409 (free_lang_data_in_cgraph): Adjust callers.
4410 * cgraphunit.c (cgraph_node::analyze): Likewise.
4411 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
4414 2017-01-23 Richard Biener <rguenther@suse.de>
4416 PR tree-optimization/79088
4417 PR tree-optimization/79188
4418 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
4419 resetting loop bounds after last path deletion. Reset loop
4420 bounds of the target loop, make code match the comments.
4421 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
4422 Make sure loops need no fixups.
4424 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
4426 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
4427 exponent support with double type for first argument.
4428 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
4429 type returned by __builtin_vec_extract_sig,
4430 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
4431 functions from "vector int" to "vector unsigned int" or from
4432 "vector long long int" to "vector unsigned long long int".
4433 Changed type returned by __builtin_vec_extract_exp,
4434 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
4435 functions from "vector int" to "vector unsigned int" or from
4436 "vector long long int" to "vector unsigned long long int".
4437 Changed return type of __builtin_vec_test_data_class,
4438 __builtin_vec_test_data_class_sp, and
4439 __builtin_vec_test_data_class_dp from "vector int" to
4440 "vector bool int" or from "vector long long int" to "vector bool
4441 long long int" and changed second argument type from "unsigned
4442 int" to "int". Added new overloaded function forms "vector float
4443 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
4444 "vector float __builtin_vec_insert_exp_sp (vector float, vector
4445 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
4446 double, vector unsigned long long int)" and "vector double
4447 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
4448 long int)". Changed return type of
4449 __builtin_scalar_test_data_class and
4450 __builtin_scalar_test_data_class_sp and
4451 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
4452 int" and changed second argument from "unsigned int" to "int".
4453 Changed type returned by __builtin_scalar_test_neg,
4454 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
4455 from "int" to "bool int". Added new overloaded function form
4456 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
4457 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
4458 exponent double-precision with floating point first argument.
4459 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
4460 documentation of scalar_test_data_class, scalar_test_neg,
4461 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
4462 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
4463 vec_test_data_class built-in functions to reflect refinements in
4464 their type signatures.
4466 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
4468 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
4470 (aarch64_elf_asm_destructor): Likewise.
4472 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
4474 PR rtl-optimization/78634
4475 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
4476 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
4477 * ifcvt.c (noce_try_cmove): Add missing cost check.
4479 PR rtl-optimization/71724
4480 * combine.c (if_then_else_cond): Look for situations where it is
4481 beneficial to undo the work of one of the recursive calls.
4483 2017-01-23 Bin Cheng <bin.cheng@arm.com>
4485 PR tree-optimization/70754
4486 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
4487 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
4488 combined stmt before it if not NULL.
4489 (combine_chains): Process refs reversely and compute dominance point
4492 2017-01-23 Martin Liska <mliska@suse.cz>
4494 PR tree-optimization/79196
4495 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
4496 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
4498 (strlen_optimize_stmt): Call the renamed function.
4500 2017-01-23 Michael Matz <matz@suse.de>
4502 PR tree-optimization/78384
4503 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
4505 2017-01-23 Richard Biener <rguenther@suse.de>
4507 PR tree-optimization/79186
4508 * tree-vrp.c (register_new_assert_for): Make sure we've seen
4509 both incoming edges before moving an assert.
4511 2017-01-23 Martin Jambor <mjambor@suse.cz>
4513 * ipa-prop.c (load_from_param_1): Removed.
4514 (load_from_unmodified_param): Bits from load_from_param_1 put back
4516 (load_from_param): Removed.
4517 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
4518 with stmt. Reverted back to use of load_from_unmodified_param.
4520 2017-01-23 Martin Jambor <mjambor@suse.cz>
4523 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
4524 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
4525 field a pointer to garbage collected vector, mark lattices and
4526 ipcp_orig_node with GTY((skip)).
4527 (ipa_get_param_count): Adjust to descriptors being a pointer.
4528 (ipa_get_param): Likewise.
4529 (ipa_get_type): Likewise.
4530 (ipa_get_param_move_cost): Likewise.
4531 (ipa_set_param_used): Likewise.
4532 (ipa_get_controlled_uses): Likewise.
4533 (ipa_set_controlled_uses): Likewise.
4534 (ipa_is_param_used): Likewise.
4535 (ipa_node_params_t): Move into garbage collector. New methods insert
4537 (ipa_node_params_sum): Annotate wth GTY(()).
4538 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
4540 (ipa_load_from_parm_agg): Adjust declaration.
4541 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
4542 * ipa-profile.c (ipa_profile): Likewise.
4543 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
4544 (ipa_populate_param_decls): Make descriptors parameter garbage
4546 (ipa_dump_param): Adjust to descriptors being a pointer.
4547 (ipa_alloc_node_params): Likewise.
4548 (ipa_initialize_node_params): Likewise.
4549 (load_from_param_1): Make descriptors parameter garbage collected.
4550 (load_from_unmodified_param): Likewise.
4551 (load_from_param): Likewise.
4552 (ipa_load_from_parm_agg): Likewise.
4553 (ipa_node_params::~ipa_node_params): Removed.
4554 (ipa_free_all_node_params): Remove call to delete operator.
4555 (ipa_node_params_t::insert): New.
4556 (ipa_node_params_t::remove): Likewise.
4557 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
4558 copy known_csts and known_contexts vectors.
4559 (ipa_read_node_info): Adjust to descriptors being a pointer.
4560 (ipcp_modif_dom_walker): Make m_descriptors field garbage
4562 (ipcp_transform_function): Make descriptors variable garbage
4565 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
4567 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
4568 * config/i386/avx512dqintrin.h: Ditto.
4569 * config/i386/avx512fintrin.h: Ditto.
4570 * gcc/config/i386/i386.c: Handle new builtins.
4571 * config/i386/i386-builtin.def: Add new builtins.
4572 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
4573 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
4575 2017-01-23 Jakub Jelinek <jakub@redhat.com>
4576 Martin Liska <mliska@suse.cz>
4578 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
4579 * asan.c (asan_expand_poison_ifn): Support stores and use
4580 appropriate ASAN report function.
4581 * internal-fn.c (expand_ASAN_POISON_USE): New function.
4582 * internal-fn.def (ASAN_POISON_USE): Declare.
4583 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
4584 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
4585 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
4586 ASAN_POISON calls w/o LHS.
4587 * tree-ssa.c (execute_update_addresses_taken): Create clobber
4588 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
4589 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
4590 * gimplify.c (asan_poison_variables): Add attribute
4591 use_after_scope_memory to variables that really needs to live
4593 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
4594 having the attribute.
4596 2017-01-23 Martin Liska <mliska@suse.cz>
4598 * asan.c (create_asan_shadow_var): New function.
4599 (asan_expand_poison_ifn): Likewise.
4600 * asan.h (asan_expand_poison_ifn): New declaration.
4601 * internal-fn.c (expand_ASAN_POISON): Likewise.
4602 * internal-fn.def (ASAN_POISON): New builtin.
4603 * sanopt.c (pass_sanopt::execute): Expand
4604 asan_expand_poison_ifn.
4605 * tree-inline.c (copy_decl_for_dup_finish): Make function
4607 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
4608 * tree-ssa.c (is_asan_mark_p): New function.
4609 (execute_update_addresses_taken): Rewrite local variables
4610 (identified just by use-after-scope as addressable) into SSA.
4612 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
4614 * doc/install.texi (Specific): opensource.apple.com uses https
4615 now. Remove trailing slash.
4617 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
4619 * README.Portability: Remove note on an Irix compatibility issue.
4621 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
4623 * gcov.c (INCLUDE_ALGORITHM): Define.
4624 (INCLUDE_VECTOR): Define.
4625 No longer include <vector> and <algorithm> directly.
4627 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
4629 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
4631 * doc/invoke.texi (Code Gen Options): Ditto.
4633 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
4636 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
4638 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
4640 rtl-optimization/79125
4641 * cprop.c (local_cprop_pass): Handle cases where we make an
4644 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
4648 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
4649 read from, for big endian.
4651 2017-01-20 Jiong Wang <jiong.wang@arm.com>
4653 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
4654 register pauth builtins for LP64 only.
4656 2017-01-20 Marek Polacek <polacek@redhat.com>
4659 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
4662 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
4664 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
4666 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
4667 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
4668 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
4670 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4673 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
4674 in big-endian mode when they are not a single duplicated value.
4676 2017-01-20 Richard Biener <rguenther@suse.de>
4678 * BASE-VER: Bump to 7.0.1.
4680 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
4682 * omp-low.c (omplow_simd_context): New struct. Use it...
4683 (lower_rec_simd_input_clauses): ...here and...
4684 (lower_rec_input_clauses): ...here to hold common data. Adjust all
4685 references to idx, lane, max_vf, is_simt.
4687 2017-01-20 Graham Markall <graham.markall@embecosm.com>
4689 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
4692 2017-01-20 Martin Jambor <mjambor@suse.cz>
4694 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
4695 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
4697 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
4698 (GTFILES): Rename hsa.c to hsa-common.c.
4699 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
4700 * hsa-dump.c: Likewise.
4701 * hsa-gen.c: Likewise.
4702 * hsa-regalloc.c: Likewise.
4703 * ipa-hsa.c: Likewise.
4704 * omp-expand.c: Likewise.
4705 * omp-low.c: Likewise.
4706 * toplev.c: Likewise.
4708 2017-01-20 Marek Polacek <polacek@redhat.com>
4711 * doc/invoke.texi: Document -Wduplicated-branches.
4712 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
4713 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
4714 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
4715 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
4716 return 0 only when not OEP_LEXICOGRAPHIC.
4717 (fold_build_cleanup_point_expr): Use the expression
4718 location when building CLEANUP_POINT_EXPR.
4719 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
4720 * tree.c (add_expr): Handle error_mark_node.
4722 2017-01-20 Martin Liska <mliska@suse.cz>
4725 * tree-profile.c (init_ic_make_global_vars): Do not call
4727 (gimple_init_gcov_profiler): Likewise.
4729 2017-01-20 Martin Liska <mliska@suse.cz>
4732 * cgraph.h (maybe_create_reference): Remove argument and
4734 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
4736 * ipa-cp.c (create_specialized_node): Likewise.
4737 * symtab.c (symtab_node::maybe_create_reference): Handle
4738 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
4740 2017-01-20 Martin Liska <mliska@suse.cz>
4742 * read-rtl-function.c (function_reader::create_function): Use
4743 build_decl instread of build_decl_stat.
4745 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
4747 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
4748 * config/i386/avx512dqintrin.h: Ditto.
4749 * config/i386/avx512fintrin.h: Ditto.
4750 * config/i386/i386-builtin-types.def: Add new types.
4751 * gcc/config/i386/i386.c: Handle new types.
4752 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
4753 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
4754 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
4755 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
4756 (__builtin_ia32_kshiftridi): New.
4757 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
4759 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
4763 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
4764 define to rs6000_init_stack_protect_guard.
4765 (rs6000_init_stack_protect_guard): New function.
4767 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
4768 Yunqiang Su <yunqiang.su@imgtec.com>
4770 * config.gcc (supported_defaults): Add madd4.
4771 (with_madd4): Add validation.
4772 (all_defaults): Add madd4.
4773 * config/mips/mips.opt (mmadd4): New option.
4774 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
4776 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
4778 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
4779 (ISA_HAS_FUSED_MADD4): Likewise.
4780 * gcc/doc/invoke.texi (-mmadd4): Document the new option.
4781 * gcc/doc/install.texi (--with-madd4): Document the new option.
4783 2017-01-19 Jiong Wang <jiong.wang@arm.com>
4785 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
4786 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
4787 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
4788 (aarch64_init_pauth_hint_builtins): New.
4789 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
4790 (aarch64_expand_builtin): Expand new builtins.
4792 2017-01-19 Jiong Wang <jiong.wang@arm.com>
4794 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
4795 * combine-stack-adj.c (no_unhandled_cfa): Handle
4796 REG_CFA_TOGGLE_RA_MANGLE.
4797 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
4798 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
4799 info for return address signing.
4800 (aarch64_expand_epilogue): Likewise.
4802 2017-01-19 Jiong Wang <jiong.wang@arm.com>
4804 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
4805 * config/aarch64/aarch64-protos.h
4806 (aarch64_return_address_signing_enabled): New declaration.
4807 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
4809 (aarch64_expand_prologue): Sign return address before it's pushed onto
4811 (aarch64_expand_epilogue): Authenticate return address fetched from
4813 (aarch64_override_options): Sanity check for ILP32 and ISA level.
4814 (aarch64_attributes): New function attributes for "sign-return-address".
4815 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
4816 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
4817 ("*do_return"): Generate combined instructions according to key index.
4818 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
4819 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
4821 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
4822 * config/aarch64/aarch64.opt (msign-return-address=): New.
4823 * doc/extend.texi (AArch64 Function Attributes): Documents
4824 "sign-return-address=".
4825 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
4827 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
4829 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
4830 overall option summary.
4832 2017-01-19 Jiong Wang <jiong.wang@arm.com>
4834 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
4835 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
4836 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
4837 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
4839 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
4841 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
4842 -mpower9-minmax by default for -mcpu=power9.
4843 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
4844 128-bit floating point.
4846 2017-01-20 Alan Modra <amodra@gmail.com>
4848 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
4849 optimizing for size.
4851 2017-01-20 Alan Modra <amodra@gmail.com>
4854 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
4855 for strcmp and strncmp from corresponding builtin decl.
4857 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
4859 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
4860 instead of i386/rtems-64.h.
4861 * config/i386/rtems-64.h: Remove.
4863 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
4867 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
4869 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
4871 2017-01-19 Tamar Christina <tamar.christina@arm.com>
4873 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
4874 Change int to HOST_WIDE_INT.
4875 * config/aarch64/aarch64-protos.h
4876 (aarch64_simd_gen_const_vector_dup): Likewise.
4877 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
4879 2017-01-19 David Malcolm <dmalcolm@redhat.com>
4881 * langhooks-def.h (lhd_type_for_size): New decl.
4882 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
4883 * langhooks.c (lhd_type_for_size): New function, taken from
4886 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
4888 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
4889 define_bypass for CR latency.
4890 (power9-cracked-alu): Update bypass latency and remove power9-branch.
4891 (power9-alu2): Add define_bypass for CR latency.
4893 (power9-mul): Update insn latency.
4894 (power9-mul-compare): Update insn latency, bypass latency and remove
4897 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4899 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
4901 * config/aarch64/aarch64.md
4902 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
4903 aarch64_nopcrelative_literal_loads.
4904 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
4906 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
4908 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
4910 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
4912 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
4915 * config.gcc (supported_defaults): Add lxc1-sxc1.
4916 (with_lxc1_sxc1): Add validation.
4917 (all_defaults): Add lxc1-sxc1.
4918 * config/mips/mips.opt (mlxc1-sxc1): New option.
4919 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
4921 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
4922 __mips_no_lxc1_sxc1.
4923 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
4924 * gcc/doc/invoke.texi (-mlxc1-sxc1): Document the new option.
4925 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
4927 2017-01-19 Richard Biener <rguenther@suse.de>
4929 PR tree-optimization/72488
4930 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
4931 sure to restore SSA info.
4932 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
4934 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
4936 PR rtl-optimization/79121
4937 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
4938 of the inner type when shifting an extended value.
4940 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
4943 * symtab.c (symtab_node::equal_address_to): Fix comparing of
4944 interposable aliases.
4946 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
4949 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
4950 Use the evmergelohi instruction.
4951 (mov_si<mode>_e500_subreg4_2_le): Likewise.
4952 (mov_sitf_e500_subreg8_2_be): Likewise.
4953 (mov_sitf_e500_subreg12_2_le): Likewise.
4954 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
4955 (mov_si<mode>_e500_subreg4_2_be): Likewise.
4956 (mov_sitf_e500_subreg8_2_le): Likewise.
4957 (mov_sitf_e500_subreg12_2_be): Likewise.
4959 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4961 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
4962 attribute from vecsimple to vecperm.
4963 (altivec_vbpermq2): Likewise.
4965 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4968 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
4970 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
4971 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
4972 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
4973 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
4974 case where N arg is SIZE_MAX.
4975 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
4976 (cmpstrsi): Add pattern.
4978 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
4980 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4981 __builtin_vec_revb builtins.
4982 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
4983 built-in functions to support generation of the ISA 3.0 XXBR<x>
4984 vector byte reverse instructions.
4985 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
4986 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
4987 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
4988 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
4989 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
4990 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
4991 (P9V_BUILTIN_VEC_REVB): Likewise.
4992 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
4993 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
4994 (p9_xxbrq_v16qi): Likewise.
4995 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
4996 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
4997 (p9_xxbrh_v8hi): Likewise.
4998 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
4999 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
5000 vec_revb built-in functions.
5002 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
5004 PR rtl-optimization/78952
5005 * config/i386/i386.md (any_extract): New code iterator.
5006 (*insvqi_2): Use any_extract for source operand.
5007 (*insvqi_3): Use any_shiftrt for source operand.
5009 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
5011 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
5013 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
5015 2017-01-18 Matthias Klose <doko@ubuntu.com>
5017 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
5019 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5021 * config/rs6000/altivec.h (vec_bperm): Change #define.
5022 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
5023 (altivec_vbpermq2): New define_insn.
5024 (altivec_vbpermd): Likewise.
5025 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
5027 (VBPERMD): Likewise.
5028 (VBPERM): New polymorphic function interface.
5029 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
5030 Add entries for P9V_BUILTIN_VEC_VBPERM.
5031 * doc/extend.texi: Add interfaces for vec_bperm.
5033 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5035 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
5036 first letter of error messages.
5037 (s390_resolve_overloaded_builtin): Likewise.
5038 * config/s390/s390.c (s390_expand_builtin): Likewise.
5039 (s390_invalid_arg_for_unprototyped_fn): Likewise.
5040 (s390_valid_target_attribute_inner_p): Likewise.
5041 * config/s390/s390.md ("tabort"): Likewise.
5043 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
5045 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
5046 (ISA_AVOID_DIV_HILO): New macro.
5047 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
5048 (ISA_HAS_DDIV): Likewise.
5050 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
5052 * doc/invoke.texi (fabi-version): Correct number of occurrences.
5054 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
5056 * doc/invoke.texi (fabi-version): Spelling fix.
5058 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
5061 * doc/invoke.texi (fabi-version): Mention mangling fix for
5064 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
5067 * doc/invoke.texi (fabi-version): Document discriminator mangling.
5069 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
5072 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
5073 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
5075 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
5077 (stack_protect_test): Ditto.
5078 * config/rs6000/rs6000.opt (mstack-protector-guard=,
5079 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
5081 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
5082 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
5083 -mstack-protector-guard-offset=.
5084 (RS/6000 and PowerPC Options): Ditto.
5086 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
5088 * config/i386/i386.h (MASK_CLASS_P): New define.
5089 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
5090 there are no registers from different register sets also when
5091 mask registers are used. Update function comment.
5092 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
5093 to (*k/*r) and (*k/*km) alternatives.
5095 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
5097 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
5098 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
5099 (EH_RETURN_HANDLER_RTX): New define.
5100 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
5101 Force frame pointer in EH return functions.
5102 (aarch64_expand_epilogue): Add barrier for eh_return.
5103 (aarch64_final_eh_return_addr): Remove.
5104 (aarch64_eh_return_handler_rtx): New function.
5105 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
5107 (aarch64_eh_return_handler_rtx): New prototype.
5109 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5111 * config/rs6000/altivec.h (vec_rlmi): New #define.
5112 (vec_vrlnm): Likewise.
5113 (vec_rlnm): Likewise.
5114 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
5115 (UNSPEC_VRLNM): Likewise.
5116 (VIlong): New mode iterator.
5117 (altivec_vrl<VI_char>mi): New define_insn.
5118 (altivec_vrl<VI_char>nm): Likewise.
5119 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
5122 (RLNM): New polymorphic function entry.
5123 (VRLWMI): New monomorphic function entry.
5125 (RLMI): New polymorphic function entry.
5126 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
5127 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
5128 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
5131 2017-01-17 Jakub Jelinek <jakub@redhat.com>
5134 * dwarf2out.c (field_byte_offset): Restore the
5135 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
5136 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
5137 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
5140 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
5143 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
5145 2017-01-17 Jakub Jelinek <jakub@redhat.com>
5148 * dwarf2out.c (add_data_member_location_attribute): For constant
5149 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
5150 instead of DW_AT_data_member_location, DW_AT_bit_offset and
5151 DW_AT_byte_size attributes.
5153 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
5155 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
5156 after forcing to constant memory when the code model is medium.
5158 2017-01-17 Julia Koval <julia.koval@intel.com>
5161 * config/i386/avx512fintrin.h
5162 (_mm512_i32gather_ps): Change __addr type to void const*.
5163 (_mm512_mask_i32gather_ps): Ditto.
5164 (_mm512_i32gather_pd): Ditto.
5165 (_mm512_mask_i32gather_pd): Ditto.
5166 (_mm512_i64gather_ps): Ditto.
5167 (_mm512_mask_i64gather_ps): Ditto.
5168 (_mm512_i64gather_pd): Ditto.
5169 (_mm512_mask_i64gather_pd): Ditto.
5170 (_mm512_i32gather_epi32): Ditto.
5171 (_mm512_mask_i32gather_epi32): Ditto.
5172 (_mm512_i32gather_epi64): Ditto.
5173 (_mm512_mask_i32gather_epi64): Ditto.
5174 (_mm512_i64gather_epi32): Ditto.
5175 (_mm512_mask_i64gather_epi32): Ditto.
5176 (_mm512_i64gather_epi64): Ditto.
5177 (_mm512_mask_i64gather_epi64): Ditto.
5178 (_mm512_i32scatter_ps): Change __addr type to void*.
5179 (_mm512_mask_i32scatter_ps): Ditto.
5180 (_mm512_i32scatter_pd): Ditto.
5181 (_mm512_mask_i32scatter_pd): Ditto.
5182 (_mm512_i64scatter_ps): Ditto.
5183 (_mm512_mask_i64scatter_ps): Ditto.
5184 (_mm512_i64scatter_pd): Ditto.
5185 (_mm512_mask_i64scatter_pd): Ditto.
5186 (_mm512_i32scatter_epi32): Ditto.
5187 (_mm512_mask_i32scatter_epi32): Ditto.
5188 (_mm512_i32scatter_epi64): Ditto.
5189 (_mm512_mask_i32scatter_epi64): Ditto.
5190 (_mm512_i64scatter_epi32): Ditto.
5191 (_mm512_mask_i64scatter_epi32): Ditto.
5192 (_mm512_i64scatter_epi64): Ditto.
5193 (_mm512_mask_i64scatter_epi64): Ditto.
5194 * config/i386/avx512pfintrin.h
5195 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
5196 (_mm512_mask_prefetch_i32gather_ps): Ditto.
5197 (_mm512_mask_prefetch_i64gather_pd): Ditto.
5198 (_mm512_mask_prefetch_i64gather_ps): Ditto.
5199 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
5200 (_mm512_prefetch_i32scatter_ps): Ditto.
5201 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
5202 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
5203 (_mm512_prefetch_i64scatter_pd): Ditto.
5204 (_mm512_prefetch_i64scatter_ps): Ditto.
5205 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
5206 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
5207 * config/i386/avx512vlintrin.h
5208 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
5209 (_mm_mmask_i32gather_ps): Ditto.
5210 (_mm256_mmask_i32gather_pd): Ditto.
5211 (_mm_mmask_i32gather_pd): Ditto.
5212 (_mm256_mmask_i64gather_ps): Ditto.
5213 (_mm_mmask_i64gather_ps): Ditto.
5214 (_mm256_mmask_i64gather_pd): Ditto.
5215 (_mm_mmask_i64gather_pd): Ditto.
5216 (_mm256_mmask_i32gather_epi32): Ditto.
5217 (_mm_mmask_i32gather_epi32): Ditto.
5218 (_mm256_mmask_i32gather_epi64): Ditto.
5219 (_mm_mmask_i32gather_epi64): Ditto.
5220 (_mm256_mmask_i64gather_epi32): Ditto.
5221 (_mm_mmask_i64gather_epi32): Ditto.
5222 (_mm256_mmask_i64gather_epi64): Ditto.
5223 (_mm_mmask_i64gather_epi64): Ditto.
5224 (_mm256_i32scatter_ps): Change __addr type to void*.
5225 (_mm256_mask_i32scatter_ps): Ditto.
5226 (_mm_i32scatter_ps): Ditto.
5227 (_mm_mask_i32scatter_ps): Ditto.
5228 (_mm256_i32scatter_pd): Ditto.
5229 (_mm256_mask_i32scatter_pd): Ditto.
5230 (_mm_i32scatter_pd): Ditto.
5231 (_mm_mask_i32scatter_pd): Ditto.
5232 (_mm256_i64scatter_ps): Ditto.
5233 (_mm256_mask_i64scatter_ps): Ditto.
5234 (_mm_i64scatter_ps): Ditto.
5235 (_mm_mask_i64scatter_ps): Ditto.
5236 (_mm256_i64scatter_pd): Ditto.
5237 (_mm256_mask_i64scatter_pd): Ditto.
5238 (_mm_i64scatter_pd): Ditto.
5239 (_mm_mask_i64scatter_pd): Ditto.
5240 (_mm256_i32scatter_epi32): Ditto.
5241 (_mm256_mask_i32scatter_epi32): Ditto.
5242 (_mm_i32scatter_epi32): Ditto.
5243 (_mm_mask_i32scatter_epi32): Ditto.
5244 (_mm256_i32scatter_epi64): Ditto.
5245 (_mm256_mask_i32scatter_epi64): Ditto.
5246 (_mm_i32scatter_epi64): Ditto.
5247 (_mm_mask_i32scatter_epi64): Ditto.
5248 (_mm256_i64scatter_epi32): Ditto.
5249 (_mm256_mask_i64scatter_epi32): Ditto.
5250 (_mm_i64scatter_epi32): Ditto.
5251 (_mm_mask_i64scatter_epi32): Ditto.
5252 (_mm256_i64scatter_epi64): Ditto.
5253 (_mm256_mask_i64scatter_epi64): Ditto.
5254 (_mm_i64scatter_epi64): Ditto.
5255 (_mm_mask_i64scatter_epi64): Ditto.
5256 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
5257 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
5258 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
5259 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
5260 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
5261 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
5262 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
5263 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
5264 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
5265 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
5266 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
5267 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
5268 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
5269 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
5270 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
5271 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
5272 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
5273 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
5274 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
5275 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
5276 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
5277 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
5278 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
5279 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
5280 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
5281 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
5282 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
5283 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
5284 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
5285 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
5286 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
5287 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
5288 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
5289 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
5290 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
5291 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
5292 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
5293 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
5294 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
5295 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
5296 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
5297 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
5298 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
5299 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
5300 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
5301 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
5302 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
5303 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
5304 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
5305 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
5306 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
5307 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
5308 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
5309 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
5310 definitions accordingly.
5312 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
5313 Kuan-Lin Chen <kuanlinchentw@gmail.com>
5316 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
5319 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
5322 * ira-conflicts.c (ira_build_conflicts): Update total conflict
5323 hard regs for inner regno.
5325 2017-01-17 Martin Liska <mliska@suse.cz>
5328 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
5329 assumption and add comment.
5331 2017-01-17 Nathan Sidwell <nathan@acm.org>
5333 * ipa-visibility.c (localize_node): New function, broken out of ...
5334 (function_and_variable_visibility): ... here. Call it.
5336 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
5339 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
5340 correctly set frequency of oudgoing edge.
5341 (duplicate_thread_path): Fix profile updating.
5343 2017-01-17 Jakub Jelinek <jakub@redhat.com>
5346 * configure.ac: Add GCC_BASE_VER.
5347 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
5348 version from BASE-VER file.
5349 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
5350 (gcc.o): Depend on $(BASEVER).
5351 * common.opt (dumpfullversion): New option.
5352 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
5353 * doc/invoke.texi: Document -dumpfullversion.
5354 * doc/install.texi: Document --with-gcc-major-version-only.
5355 * configure: Regenerated.
5357 2017-01-17 Richard Biener <rguenther@suse.de>
5359 PR tree-optimization/71433
5360 * tree-vrp.c (register_new_assert_for): Merge same asserts
5361 on all incoming edges.
5362 (process_assert_insertions_for): Handle insertions at the
5365 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
5367 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
5368 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
5370 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
5373 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
5376 2017-01-17 Alan Modra <amodra@gmail.com>
5379 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
5380 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
5381 symbolic stack limit when pic.
5383 2017-01-16 Martin Sebor <msebor@redhat.com>
5385 PR tree-optimization/78608
5386 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
5388 2017-01-16 Jeff Law <law@redhat.com>
5391 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
5392 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
5393 for several include directories that may be relative to sysroot.
5394 * config/i386/x-mingw32 (gplus_includedir): Define.
5395 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
5396 (native_system_includedir): Likewise.
5397 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
5398 override if TARGET_SYSTEM_ROOT is defined.
5399 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
5401 PR tree-optimization/79090
5402 PR tree-optimization/33562
5403 PR tree-optimization/61912
5404 PR tree-optimization/77485
5405 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
5406 and computed trims into the dump file.
5408 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
5410 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
5412 2017-01-16 Jakub Jelinek <jakub@redhat.com>
5415 * gimplify.c (gimplify_init_constructor): If want_value and
5416 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
5420 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
5421 sequence. Formatting fixes.
5422 (doloop_optimize): Formatting fixes.
5425 * gcc.c (debug_level_greater_than_spec_func): New function.
5426 (static_spec_functions): Add debug-level-gt spec function.
5427 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
5429 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
5430 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
5431 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
5432 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
5433 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
5434 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
5436 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
5438 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
5439 QImode fixups to general and mask registers only.
5441 2017-01-16 Carl Love <cel@us.ibm.com>
5443 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
5444 for built-in functions
5445 vector signed char vec_nabs (vector signed char)
5446 vector signed short vec_nabs (vector signed short)
5447 vector signed int vec_nabs (vector signed int)
5448 vector signed long long vec_nabs (vector signed long long)
5449 vector float vec_nabs (vector float)
5450 vector double vec_nabs (vector double)
5451 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
5453 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
5454 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
5455 * doc/extend.texi: Update the documentation file for the new built-in
5458 2017-01-16 Martin Sebor <msebor@redhat.com>
5460 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
5463 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5465 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
5466 UNSPEC_VSX__XXSPLTD to require special splat handling.
5468 2017-01-16 David Malcolm <dmalcolm@redhat.com>
5471 * system.h: Poison strndup.
5473 2017-01-16 Alan Modra <amodra@gmail.com>
5476 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
5479 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
5481 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
5483 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
5485 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
5486 call recog here. Assert that INSN_CODE (insn) is non-negative.
5488 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
5491 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
5493 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
5494 in the currently scheduled RTL fragment.
5496 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
5498 PR rtl-optimization/78751
5499 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
5502 2017-01-14 Jeff Law <law@redhat.com>
5504 PR tree-optimization/79090
5505 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
5506 variable length stores.
5507 (compute_trims): Delete dead assignment to *trim_tail.
5508 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
5511 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
5513 PR rtl-optimization/78626
5514 PR rtl-optimization/78727
5515 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
5516 of a block, and split such blocks after everything else is finished.
5518 2017-01-14 Alan Modra <amodra@gmail.com>
5521 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
5522 target legitimate_combined_insn.
5523 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
5524 (rs6000_legitimate_combined_insn): New function.
5525 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
5527 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
5528 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
5529 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
5531 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
5533 * doc/frontends.texi (G++ and GCC): Remove references to Java.
5535 2017-01-13 Jeff Law <law@redhat.com>
5537 PR tree-optimization/33562
5538 PR tree-optimization/61912
5539 PR tree-optimization/77485
5540 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
5542 (delete_dead_assignment): Likewise.
5543 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
5544 statement to delete_dead_call and delete_dead_assignment.
5546 2017-01-13 David Malcolm <dmalcolm@redhat.com>
5549 * substring-locations.c (format_warning_va): Strengthen case 1 so
5550 that both endpoints of the substring must be within the format
5551 range for just the substring to be printed.
5553 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
5555 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
5556 * config/i386/i386.c (ix86_target_string): Add missing options
5557 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
5558 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
5559 flags_other and ix86_target_other to flags2_other. Display unknown
5561 (ix86_valid_target_attribute_inner_p): Add missing options and
5562 reorder options by implied ISAs, as in ix86_target_string.
5564 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
5566 * hash-table.h (hash_table::too_empty_p): New function.
5567 (hash_table::expand): Use it.
5568 (hash_table::traverse): Likewise.
5569 (hash_table::empty_slot): Use sizeof (value_type) instead of
5570 sizeof (PTR) to convert bytes to elements. Shrink the table
5571 if the current size is excessive for the current number of
5574 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
5576 * ira-costs.c (record_reg_classes): Break from the inner loop
5577 early once alt_fail is known to be true. Update outer loop
5578 handling accordingly.
5580 2017-01-13 Jeff Law <law@redhat.com>
5582 * tree-ssa-dse.c (decrement_count): New function.
5583 (increment_start_addr, maybe_trim_memstar_call): Likewise.
5584 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
5585 when we know the partially dead statement is a mem* function.
5587 PR tree-optimization/61912
5588 PR tree-optimization/77485
5589 * tree-ssa-dse.c: Include expr.h.
5590 (maybe_trim_constructor_store): New function.
5591 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
5593 PR tree-optimization/33562
5594 PR tree-optimization/61912
5595 PR tree-optimization/77485
5596 * doc/invoke.texi: Document new dse-max-object-size param.
5597 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
5598 * tree-ssa-dse.c: Include params.h.
5599 (dse_store_status): New enum.
5600 (initialize_ao_ref_for_dse): New, partially extracted from
5602 (valid_ao_ref_for_dse, normalize_ref): New.
5603 (setup_live_bytes_from_ref, compute_trims): Likewise.
5604 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
5605 (maybe_trim_partially_dead_store): Likewise.
5606 (maybe_trim_complex_store): Likewise.
5607 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
5608 Track what bytes live from the original store. Return tri-state
5609 for dead, partially dead or live.
5610 (dse_dom_walker): Add constructor, destructor and new private members.
5611 (delete_dead_call, delete_dead_assignment): New extracted from
5613 (dse_optimize_stmt): Make a member of dse_dom_walker.
5614 Use initialize_ao_ref_for_dse.
5616 PR tree-optimization/33562
5617 PR tree-optimization/61912
5618 PR tree-optimization/77485
5619 * sbitmap.h (bitmap_count_bits): Prototype.
5620 (bitmap_clear_range, bitmap_set_range): Likewise.
5621 * sbitmap.c (bitmap_clear_range): New function.
5622 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
5624 2017-01-13 Martin Liska <mliska@suse.cz>
5627 * function.c (set_cfun): Add new argument force.
5628 * function.h (set_cfun): Likewise.
5629 * ipa-inline-transform.c (inline_call): Use the function when
5630 strict alising from is dropped for function we inline to.
5632 2017-01-13 Richard Biener <rguenther@suse.de>
5634 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
5635 for dumping GIMPLE INTEGER_CSTs.
5637 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5639 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
5640 to 201112L since C++17.
5642 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
5645 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
5646 if -fsanitize=kernel-address is present.
5648 2017-01-13 Richard Biener <rguenther@suse.de>
5650 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
5651 as _Literal ( type ) number in case usual suffixes do not
5652 preserve all information.
5654 2017-01-13 Richard Biener <rguenther@suse.de>
5656 PR tree-optimization/77283
5657 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
5658 and ssa-iterators.h.
5659 (is_feasible_trace): Implement a cost model based on joiner
5662 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
5665 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
5666 char or short to __float128/_Float128 directly.
5668 2017-01-12 Martin Sebor <msebor@redhat.com>
5670 to -Wformat-overflow.
5671 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
5672 (min_bytes_remaining): Same.
5673 (get_string_length): Same.
5674 (format_string): Same.
5675 (format_directive): Same.
5677 (pass_sprintf_length::handle_gimple_call): Same.
5679 2017-01-12 Jakub Jelinek <jakub@redhat.com>
5681 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
5682 info.nowrite calls with no lhs that can't throw. Return bool
5683 whether gsi_remove has been called or not.
5684 (pass_sprintf_length::handle_gimple_call): Return bool whether
5685 try_substitute_return_value called gsi_remove. Formatting fix.
5686 (pass_sprintf_length::execute): Don't use gsi_remove if
5687 handle_gimple_call returned true.
5690 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
5691 be removed due to side-effects, don't remove following barrier nor
5692 turn the successor edge into fallthru edge.
5694 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5697 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
5698 element-reversing loads and stores as not swappable.
5700 2017-01-12 Nathan Sidwell <nathan@acm.org>
5701 Nicolai Stange <nicstange@gmail.com>
5703 * combine.c (try_combine): Don't ignore result of overlap checking
5704 loop. Combine overlap & asm check into single loop.
5706 2017-01-12 Richard Biener <rguenther@suse.de>
5708 * tree-pretty-print.c (dump_generic_node): Provide -gimple
5709 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
5711 2017-01-12 Richard Biener <rguenther@suse.de>
5713 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
5714 and TS_TARGET_OPTION directly derive from TS_BASE.
5715 * tree-core.h (tree_optimization_option): Derive from tree_base.
5716 (tree_target_option): Likewise.
5718 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
5720 * config/i386/i386.c (memory_address_length): Increase len
5721 only when rip_relative_addr_p returns false.
5723 2017-01-11 Julia Koval <julia.koval@intel.com>
5725 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
5726 (OPTION_MASK_ISA_SGX_SET): New.
5727 (ix86_handle_option): Handle OPT_msgx.
5728 * config.gcc: Added sgxintrin.h.
5729 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
5730 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
5731 * config/i386/i386.c (ix86_target_string): Add -msgx.
5733 (ix86_option_override_internal): Handle new options.
5734 (ix86_valid_target_attribute_inner_p): Add sgx.
5735 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
5736 * config/i386/i386.opt: Add msgx.
5737 * config/i386/sgxintrin.h: New file.
5738 * config/i386/x86intrin.h: Add sgxintrin.h.
5740 2017-01-11 Jakub Jelinek <jakub@redhat.com>
5743 * fold-const.c (maybe_nonzero_address): Return 1 for function
5745 (tree_single_nonzero_warnv_p): Don't handle function local objects
5749 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
5752 2017-01-11 David Malcolm <dmalcolm@redhat.com>
5755 * opts.c: Include "spellcheck.h"
5756 (struct string_fragment): New struct.
5757 (struct edit_distance_traits<const string_fragment &>): New
5759 (get_closest_sanitizer_option): New function.
5760 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
5762 2017-01-11 Jakub Jelinek <jakub@redhat.com>
5764 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
5766 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
5767 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
5768 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
5769 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
5770 for initial die_offset if dwarf_split_debug_info.
5771 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
5772 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
5774 (output_skeleton_debug_sections): Formatting fix. Use
5775 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
5776 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
5778 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
5780 * config/arm/cortex-a53.md: Add bypasses for
5782 (cortex_a53_r2f): Only use for transfers.
5783 (cortex_a53_f2r): Likewise.
5784 (cortex_a53_r2f_cvt): Add reservation for conversions.
5785 (cortex_a53_f2r_cvt): Likewise.
5787 2017-01-11 Tamar Christina <tamar.christina@arm.com>
5789 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
5790 to all inlined functions, change static to extern.
5792 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
5795 * config/arm/arm.c (legitimize_pic_address): Handle reference to
5797 (arm_assemble_integer): Likewise.
5799 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
5801 * config.gcc: Use new awk script to check CPU, FPU and architecture
5802 parameters for --with-... options.
5803 * config/arm/parsecpu.awk: New file
5804 * config/arm/arm-cpus.in: New file.
5805 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
5807 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
5809 * config/arm/t-arm: Update dependency rules.
5810 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
5811 of processing .def files.
5812 * config/arm/genopt.sh: Deleted.
5813 * config/arm/gentune.sh: Deleted.
5814 * config/arm/arm-cores.def: Deleted.
5815 * config/arm/arm-arches.def: Deleted.
5816 * config/arm/arm-fpus.def: Deleted.
5817 * config/arm/arm-tune.md: Regenerated.
5818 * config/arm/arm-tables.opt: Regenerated.
5819 * config/arm/arm-cpu.h: New generated file.
5820 * config/arm/arm-cpu-data.h: New generated file.
5821 * config/arm/arm-cpu-cdata.h: New generated file.
5823 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
5826 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
5828 (input_varpool_node): Unpack dynamically_initialized bit.
5830 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
5832 PR rtl-optimization/79032
5833 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
5834 the alignment of the adjusted memory reference against that of MODE,
5835 instead of the alignment of the original memory reference.
5837 2017-01-11 Martin Jambor <mjambor@suse.cz>
5839 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
5841 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
5842 decorated functions.
5844 2017-01-11 Richard Biener <rguenther@suse.de>
5846 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
5847 set range/nonnull info for PHI results. Do not set it on
5848 stmts marked for removal.
5850 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
5852 * expr.c (store_field): In the bitfield case, fetch the return value
5853 from the registers before applying a single big-endian adjustment.
5854 Always do a final load for a BLKmode value not larger than a word.
5856 2017-01-10 David Malcolm <dmalcolm@redhat.com>
5859 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
5860 that we correctly handle column numbers greater than
5861 LINE_MAP_MAX_COLUMN_NUMBER.
5863 2017-01-10 Martin Sebor <msebor@redhat.com>
5866 * gimple-ssa-sprintf.c (get_destination_size): Call
5867 {init,fini}object_sizes.
5868 * tree-object-size.c (addr_object_size): Adjust.
5869 (pass_through_call): Adjust.
5870 (pass_object_sizes::execute): Adjust.
5871 * tree-object-size.h (fini_object_sizes): Declare.
5873 2017-01-10 Martin Sebor <msebor@redhat.com>
5875 PR tree-optimization/78775
5876 * builtins.c (get_size_range): Move...
5877 * calls.c: ...to here.
5878 (alloc_max_size): Accept zero argument.
5879 (operand_signed_p): Remove.
5880 (maybe_warn_alloc_args_overflow): Call get_size_range.
5881 * calls.h (get_size_range): Declare.
5883 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
5885 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
5886 from TI's devices.csv file as of September 2016.
5887 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
5889 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
5891 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
5892 * doc/invoke.texi: Likewise.
5893 * doc/md.texi: Likewise.
5894 * doc/objc.texi: Likewise.
5896 2017-01-10 Joshua Conner <joshconner@google.com>
5898 * config/arm/fuchsia-elf.h: New file.
5899 * config/fuchsia.h: New file.
5900 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
5901 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
5903 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
5905 2016-01-10 Richard Biener <rguenther@suse.de>
5907 PR tree-optimization/79034
5908 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
5909 Propagate out degenerate PHIs in the joiner.
5911 2017-01-10 Martin Liska <mliska@suse.cz>
5913 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
5914 (sort_congruence_classes_by_decl_uid): Likewise.
5915 (sort_congruence_class_groups_by_decl_uid): Likewise.
5916 (sem_item_optimizer::merge_classes): Sort class, groups in these
5917 classes and members in the groups by DECL_UID of declarations.
5918 This would make merge operations stable.
5920 2017-01-10 Martin Liska <mliska@suse.cz>
5922 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
5923 usage of m_classes_vec.
5924 (sem_item_optimizer::~sem_item_optimizer): Likewise.
5925 (sem_item_optimizer::get_group_by_hash): Likewise.
5926 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
5927 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
5928 (sem_item_optimizer::verify_classes): Likewise.
5929 (sem_item_optimizer::process_cong_reduction): Likewise.
5930 (sem_item_optimizer::dump_cong_classes): Likewise.
5931 (sem_item_optimizer::merge_classes): Likewise.
5932 * ipa-icf.h (congruence_class_hash): Rename from
5933 congruence_class_group_hash. Remove declaration of m_classes_vec.
5935 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
5937 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
5938 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
5939 * config.gcc: Add avx512vpopcntdqintrin.h.
5940 * config/i386/avx512vpopcntdqintrin.h: New.
5941 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
5942 * config/i386/i386-builtin-types.def: Add new types.
5943 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
5944 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
5945 __builtin_ia32_vpopcountq_v8di_mask): New.
5946 * config/i386/i386-c.c (ix86_target_macros_internal): Define
5947 __AVX512VPOPCNTDQ__.
5948 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
5949 (PTA_AVX512VPOPCNTDQ): Define.
5950 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
5951 TARGET_AVX512VPOPCNTDQ_P): Define.
5952 * config/i386/i386.opt: Add mavx512vpopcntdq.
5953 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
5954 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
5956 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
5959 * predict.def (PRED_CALL): Set to 67.
5961 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
5963 * expr.c (store_field): In the bitfield case, if the value comes from
5964 a function call and is of an aggregate type returned in registers, do
5965 not modify the field mode; extract the value in all cases if the mode
5966 is BLKmode and the size is not larger than a word.
5968 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
5971 * config/i386/cpuid.h: Fix undefined behavior.
5973 2017-01-04 Jeff Law <law@redhat.com>
5975 PR tree-optimization/79007
5976 PR tree-optimization/67955
5977 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
5978 conservative for pt.null when flag_non_call_exceptions is on.
5980 2017-01-09 Jakub Jelinek <jakub@redhat.com>
5982 PR translation/79019
5983 PR translation/79020
5984 * params.def (PARAM_INLINE_MIN_SPEEDUP,
5985 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
5986 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
5988 * config/avr/avr.opt (maccumulate-args): Likewise.
5989 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
5990 * common.opt (freport-bug): Likewise.
5991 * cif-code.def (CIF_FINAL_ERROR): Likewise.
5992 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
5993 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
5994 translatable string.
5995 * config/i386/i386.c (function_value_32): Likewise.
5996 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
5997 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
5999 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
6000 * common/config/msp430/msp430-common.c (msp430_handle_option):
6002 * symtab.c (symtab_node::verify_base): Likewise.
6003 * opts.c (set_debug_level): Likewise.
6004 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
6005 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
6006 missing whitespace to translatable strings.
6007 * config/avr/avr.md (bswapsi2): Fix typo in comment.
6008 * config/sh/superh.h: Likewise.
6009 * config/i386/xopintrin.h: Likewise.
6010 * config/i386/znver1.md: Likewise.
6011 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
6012 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
6013 * double-int.h (struct double_int): Likewise.
6014 * double-int.c (div_and_round_double): Likewise.
6015 * wide-int.cc: Likewise.
6016 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
6017 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
6018 * cfgcleanup.c (crossjumps_occured): Renamed to ...
6019 (crossjumps_occurred): ... this.
6020 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
6023 PR tree-optimization/78899
6024 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
6025 returning bool return struct loop *, NULL for failure and the new
6027 (versionable_outer_loop_p): Don't version outer loop if it has
6028 dont_vectorized bit set.
6029 (tree_if_conversion): When versioning outer loop, ensure
6030 tree_if_conversion is performed also on the inner loop of the
6031 non-vectorizable outer loop copy.
6032 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
6033 LOOP_VECTORIZED in inner loop of the scalar outer loop and
6034 prevent vectorization of it.
6035 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
6036 the outer loop vectorization of the non-scalar version is attempted
6037 before vectorization of the inner loop in scalar version. If
6038 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
6039 vectorization of its inner loop.
6040 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
6041 has 2 inner loops, rename also on edges from bb whose single pred
6042 is outer_loop->header. Fix typo in function comment.
6044 2017-01-09 Martin Sebor <msebor@redhat.com>
6047 * asan.c (asan_emit_stack_protection): Increase local buffer size
6048 to avoid snprintf truncation warning.
6050 2017-01-09 Andrew Pinski <apinski@cavium.com>
6052 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
6053 to reference thunderx2t99 for the tuning structure
6054 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
6056 (thunderx2t99_extra_costs): This.
6057 * config/aarch64/aarch64-tune.md: Regenerate.
6058 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
6059 (vulcan_addrcost_table): This.
6060 (vulcan_regmove_cost): Rename to ...
6061 (thunderx2t99_regmove_cost): This.
6062 (vulcan_vector_cost): Rename to ...
6063 (thunderx2t99_vector_cost): this.
6064 (vulcan_branch_cost): Rename to ...
6065 (thunderx2t99_branch_cost): This.
6066 (vulcan_tunings): Rename to ...
6067 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
6068 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
6070 2017-01-09 Martin Jambor <mjambor@suse.cz>
6074 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
6075 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
6076 (propagate_vr_accross_jump_function): Use the above function for all
6077 value range computations for pass-through jump functions and type
6078 converasion from explicit value range values.
6079 (ipcp_propagate_stage): Do not attempt to deduce types of formal
6080 parameters from TYPE_ARG_TYPES.
6081 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
6082 (ipa_write_node_info): Stream type of the actual argument.
6083 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
6085 2017-01-09 Martin Liska <mliska@suse.cz>
6088 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
6089 (lookup_compiler): Do not show error message with have_E.
6091 2017-01-09 Jakub Jelinek <jakub@redhat.com>
6093 PR tree-optimization/78938
6094 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
6095 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
6096 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
6097 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
6100 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6102 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
6105 2017-01-09 Richard Biener <rguenther@suse.de>
6107 PR tree-optimization/78997
6108 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
6109 name condition properly.
6111 2017-01-09 Richard Biener <rguenther@suse.de>
6114 * dwarf2out.c (is_cxx): New overload with context.
6115 (is_naming_typedef_decl): Use it.
6117 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
6119 * invoke.texi (Option Summary): Correct spacing in option lists
6120 and add line breaks to fix over-long lines.
6122 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
6126 * extend.texi (Common Variable Attributes): Add xref to GCC
6127 Internals manual to explain mode attribute keywords.
6129 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
6132 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
6133 and Preprocessor Options.
6134 (Options for Linking): Document -pthread here....
6135 (RS/6000 and PowerPC Options): ...not here.
6136 (Solaris 2 Options): ...or here.
6137 * doc/cppopts.texi: Document -pthread.
6139 2017-01-08 Martin Sebor <msebor@redhat.com>
6142 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
6143 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
6144 New member functions.
6145 (format_directive): Used them.
6147 (pass_sprintf_length::handle_gimple_call): Same.
6148 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
6149 to avoid truncation for any argument.
6150 (extract_affine_mul): Same.
6151 * tree.c (get_file_function_name): Same.
6153 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
6156 * predict.def (PRED_INDIR_CALL): Set to 86.
6158 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
6160 PR preprocessor/54124
6161 * doc/cppopts.texi: Reformat -d subtable to list the full name
6162 of the options. Add cross-reference to the docs for the general
6163 compiler -d options.
6164 * doc/invoke.texi (Developer Options): Add cross-reference to the
6165 preprocessor-specific -d option documentation.
6167 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
6169 PR preprocessor/13498
6170 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
6171 redudant material, and reflect new command-line options.
6172 (System Headers): Likewise.
6174 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
6176 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
6177 -isystem, and -idirafter. Copy-edit.
6178 * doc/cppopts.texi: Copy-edit. Remove contradiction about
6179 default for -ftrack-macro-expansion. Delete obsolete and
6180 badly-formatted implementation details about -fdebug-cpp output.
6181 * doc/cppwarnopts.texi: Copy-edit.
6183 2017-01-07 David Malcolm <dmalcolm@redhat.com>
6186 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
6187 that the transition from a max line width >= 1<<10 to narrower
6188 lines works correctly.
6190 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
6192 * doc/options.texi (PerFunction): New.
6193 * opt-functions.awk (switch_flags): Map both Optimization and
6194 PerFunction to CL_OPTIMIZATION.
6195 * opth-gen.awk: Test for PerFunction flag along with
6197 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
6198 it only when the latter is present. Skip those that don't in
6199 the hash function generator.
6200 * common.opt (fvar-tracking): Mark as PerFunction instead of
6202 (fvar-tracking-assignments): Likewise.
6203 (fvar-tracking-assignments-toggle): Likewise.
6204 (fvar-tracking-uninit): Likewise.
6206 2017-01-07 Jakub Jelinek <jakub@redhat.com>
6208 PR translation/79018
6209 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
6212 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
6215 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
6216 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
6217 TARGET_LONG_JUMP_TABLE_OFFSETS.
6218 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
6219 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
6220 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
6221 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
6222 * config/m68k/m68k.md (tablejump expander): Likewise.
6223 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
6224 TARGET_LONG_JUMP_TABLE_OFFSETS.
6225 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
6226 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
6228 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
6229 David Holsgrove <david.holsgrove@xilinx.com>
6231 * common/config/microblaze/microblaze-common.c
6232 (TARGET_EXCEPT_UNWIND_INFO): Remove.
6233 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
6235 * config/microblaze/microblaze.c (microblaze_must_save_register)
6236 (microblaze_expand_epilogue, microblaze_return_addr): Handle
6238 (microblaze_eh_return): New function.
6239 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
6240 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
6241 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
6242 * config/microblaze/microblaze.md (eh_return): New pattern.
6244 2017-01-06 Jakub Jelinek <jakub@redhat.com>
6246 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
6247 GCC_DIAGNOSTIC_STRINGIFY): Define.
6249 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
6251 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
6253 * config/arm/arm.md (<mcrr>): New.
6255 * config/arm/arm.c (arm_arch5te): New.
6256 (arm_option_override): Set arm_arch5te.
6257 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
6259 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
6260 (arm_mcrr_qualifiers): ... this. New.
6261 (MRRC_QUALIFIERS): Define to...
6262 (arm_mrrc_qualifiers): ... this. New.
6263 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
6265 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
6266 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
6267 (MRRCI, mrrc, MRRC): New.
6268 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
6269 VUNSPEC_MRRC2): New.
6271 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
6273 * config/arm/arm.md (<mcr>): New.
6275 * config/arm/arm.c (arm_coproc_builtin_available): Add
6276 support for mcr, mrc, mcr2 and mrc2.
6277 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
6278 (arm_mcr_qualifiers): ... this. New.
6279 (MRC_QUALIFIERS): Define to ...
6280 (arm_mrc_qualifiers): ... this. New.
6281 (MCR_QUALIFIERS): Define to ...
6282 (arm_mcr_qualifiers): ... this. New.
6283 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
6285 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
6286 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
6287 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
6290 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
6292 * config/arm/arm.md (*ldc): New.
6296 * config/arm/arm.c (arm_coproc_builtin_available): Add
6297 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
6298 (arm_coproc_ldc_stc_legitimate_address): New.
6299 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
6300 'qualifier_const_pointer'.
6301 (LDC_QUALIFIERS): Define to...
6302 (arm_ldc_qualifiers): ... this. New.
6303 (STC_QUALIFIERS): Define to...
6304 (arm_stc_qualifiers): ... this. New.
6305 * config/arm/arm-protos.h
6306 (arm_coproc_ldc_stc_legitimate_address): New.
6307 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
6308 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
6309 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
6310 stc2, stcl, stc2l): New.
6311 * config/arm/constraints.md (Uz): New.
6312 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
6313 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
6314 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
6315 VUNSPEC_STC2L): New.
6317 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
6319 * config/arm/arm.md (<cdp>): New.
6320 * config/arm/arm.c (neon_const_bounds): Rename this ...
6321 (arm_const_bounds): ... this.
6322 (arm_coproc_builtin_available): New.
6323 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
6324 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
6325 (CDP_QUALIFIERS): Define to...
6326 (arm_cdp_qualifiers): ... this. New.
6328 (arm_expand_builtin_args): Add case for 6 arguments.
6329 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
6330 (arm_const_bounds): ... this.
6331 (arm_coproc_builtin_available): New.
6332 * config/arm/arm_acle.h (__arm_cdp): New.
6334 * config/arm/arm_acle_builtins.def (cdp): New.
6336 * config/arm/iterators.md (CDPI,CDP,cdp): New.
6337 * config/arm/neon.md: Rename all 'neon_const_bounds' to
6339 * config/arm/types.md (coproc): New.
6340 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
6341 * gcc/doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
6342 * gcc/doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
6343 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
6345 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
6347 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
6348 (UBINOP_QUALIFIERS): New.
6350 (acle_builtin_data): New. Change comment.
6351 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
6352 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
6353 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
6354 arm_acle_builtins.def.
6355 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
6356 (arm_init_acle_builtins): New.
6357 (CRC32_BUILTIN): Remove.
6358 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
6359 crc32cb, crc32ch and crc32cw.
6360 (arm_init_crc32_builtins): Remove.
6361 (arm_init_builtins): Use arm_init_acle_builtins rather
6362 than arm_init_crc32_builtins.
6363 (arm_expand_acle_builtin): New.
6364 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
6365 * config/arm/arm_acle_builtins.def: New.
6367 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
6369 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
6370 (arm_builtin_datum): ... this.
6371 (arm_init_neon_builtin): Rename to ...
6372 (arm_init_builtin): ... this. Add a new parameters PREFIX
6373 and USE_SIG_IN_NAME.
6374 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
6375 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
6376 'arm_builtin_datum'.
6377 (arm_init_vfp_builtins): Likewise.
6378 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
6379 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
6380 (arm_expand_neon_args): Rename to ...
6381 (arm_expand_builtin_args): ... this. Rename builtin_arg
6382 enum values and differentiate between ARG_BUILTIN_MEMORY
6383 and ARG_BUILTIN_NEON_MEMORY.
6384 (arm_expand_neon_builtin_1): Rename to ...
6385 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
6386 values, arm_expand_builtin_args and add bool parameter NEON.
6387 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
6388 (arm_expand_vfp_builtin): Likewise.
6389 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
6391 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
6394 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
6395 * predict.c (tree_estimate_probability_bb): Reverse direction of
6396 polymorphic call predictor.
6398 2017-01-06 David Malcolm <dmalcolm@redhat.com>
6400 * passes.c (execute_one_pass): Split out pass-skipping logic into...
6401 (determine_pass_name_match): ...this new function and...
6402 (should_skip_pass_p): ...this new function.
6404 2017-01-06 Nathan Sidwell <nathan@acm.org>
6406 * ipa-visibility.c (function_and_variable_visibility): Reformat
6407 comments and long lines. Remove extrneous if.
6408 * symtab.c (symtab_node::make_decl_local): Fix code format.
6409 (symtab_node::set_section_for_node): Fix comment typo.
6411 2017-01-06 Martin Liska <mliska@suse.cz>
6414 * lra-constraints.c: Rename invariant to lra_invariant.
6415 * predict.c (set_even_probabilities): Initialize e to NULL.
6417 2017-01-05 Martin Sebor <msebor@redhat.com>
6419 PR tree-optimization/78910
6420 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
6421 (format_integer): Correct off-by-one error in the handling
6422 of precision with negative numbers in signed conversions..
6424 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
6426 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
6428 2017-01-05 Jakub Jelinek <jakub@redhat.com>
6430 PR tree-optimization/71016
6431 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
6432 factor_out_conditional_conversion. Formatting fix.
6433 (factor_out_conditional_conversion): Add cond_stmt argument.
6434 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
6435 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
6438 2017-01-05 David Malcolm <dmalcolm@redhat.com>
6440 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
6441 read-rtl-function.o, and selftest-rtl.o.
6442 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
6443 (selftest::aarch64_test_loading_full_dump): New function.
6444 (selftest::aarch64_run_selftests): New function.
6445 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
6446 selftest::aarch64_run_selftests.
6447 * config/i386/i386.c
6448 (selftest::ix86_test_loading_dump_fragment_1): New function.
6449 (selftest::ix86_test_loading_call_insn): New function.
6450 (selftest::ix86_test_loading_full_dump): New function.
6451 (selftest::ix86_test_loading_unspec): New function.
6452 (selftest::ix86_run_selftests): Call the new functions.
6453 * emit-rtl.c (maybe_set_max_label_num): New function.
6454 * emit-rtl.h (maybe_set_max_label_num): New decl.
6455 * function.c (instantiate_decls): Guard call to
6456 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
6457 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
6459 * gensupport.c (gen_reader::gen_reader): Pass "false"
6460 for new "compact" param of rtx_reader.
6461 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
6462 rather than an empty string for NULL strings.
6463 * read-md.c: Potentially include config.h rather than bconfig.h.
6464 Wrap include of errors.h with #ifdef GENERATOR_FILE.
6465 (have_error): New global, copied from errors.c.
6466 (md_reader::read_name): Rename to...
6467 (md_reader::read_name_1): ...this, adding "out_loc" param,
6468 and converting "missing name or number" to returning false, rather
6470 (md_reader::read_name): Reimplement in terms of read_name_1.
6471 (md_reader::read_name_or_nil): New function.
6472 (md_reader::read_string): Handle "(nil)" by returning NULL.
6473 (md_reader::md_reader): Add new param "compact".
6474 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
6475 (md_reader::read_file): New method.
6476 * read-md.h (md_reader::md_reader): Add new param "compact".
6477 (md_reader::read_file): New method.
6478 (md_reader::is_compact): New accessor.
6479 (md_reader::read_name): Convert return type from void to file_location.
6480 (md_reader::read_name_or_nil): New decl.
6481 (md_reader::read_name_1): New decl.
6482 (md_reader::m_compact): New field.
6483 (noop_reader::noop_reader): Pass "false" for new "compact" param
6485 (rtx_reader::rtx_reader): Add new "compact" param.
6486 (rtx_reader::read_rtx_operand): Make virtual and convert return
6487 type from void to rtx.
6488 (rtx_reader::read_until): New decl.
6489 (rtx_reader::handle_any_trailing_information): New virtual function.
6490 (rtx_reader::postprocess): New virtual function.
6491 (rtx_reader::finalize_string): New virtual function.
6492 (rtx_reader::m_in_call_function_usage): New field.
6493 (rtx_reader::m_reuse_rtx_by_id): New field.
6494 * read-rtl-function.c: New file.
6495 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
6496 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
6497 (selftest::verify_three_block_rtl_cfg): New decl.
6498 * read-rtl-function.h: New file.
6499 * read-rtl.c: Potentially include config.h rather than bconfig.h.
6500 For host, include function.h, memmodel.h, and emit-rtl.h.
6501 (one_time_initialization): New function.
6502 (struct compact_insn_name): New struct.
6503 (compact_insn_names): New array.
6504 (find_code): Handle insn codes in compact dumps.
6505 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
6506 (bind_subst_iter_and_attr): Likewise.
6507 (add_condition_to_string): Likewise.
6508 (add_condition_to_rtx): Likewise.
6509 (apply_attribute_uses): Likewise.
6510 (add_current_iterators): Likewise.
6511 (apply_iterators): Likewise.
6512 (initialize_iterators): Guard usage of apply_subst_iterator with
6513 #ifdef GENERATOR_FILE.
6514 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
6515 (md_reader::read_mapping): Likewise.
6516 (add_define_attr_for_define_subst): Likewise.
6517 (add_define_subst_attr): Likewise.
6518 (read_subst_mapping): Likewise.
6519 (check_code_iterator): Likewise.
6520 (rtx_reader::read_rtx): Likewise. Move one-time initialization
6522 (one_time_initialization): New function.
6523 (rtx_reader::read_until): New method.
6524 (read_flags): New function.
6525 (parse_reg_note_name): New function.
6526 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
6527 Handle reuse_rtx ids.
6528 Wrap iterator lookup within #ifdef GENERATOR_FILE.
6529 Add parsing support for RTL dumps, mirroring the special-cases in
6530 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
6531 values, and calling handle_any_trailing_information.
6532 (rtx_reader::read_rtx_operand): Convert return type from void
6533 to rtx, returning return_rtx. Handle case 'e'. Call
6534 finalize_string on XSTR and XTMPL fields.
6535 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
6536 "(nil)" values were omitted. Call the postprocess vfunc on the
6538 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
6539 class ctor. Initialize m_in_call_function_usage. Call
6540 one_time_initialization.
6541 * rtl-tests.c (selftest::test_uncond_jump): Call
6542 set_new_first_and_last_insn.
6543 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
6544 * selftest-rtl.c: New file.
6545 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
6546 (selftest::get_insn_by_uid): New decl.
6547 * selftest-run-tests.c (selftest::run_tests): Call
6548 read_rtl_function_c_tests.
6549 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
6550 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
6553 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
6555 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
6556 operands in a special way. Assert that pos+len <= mode precision.
6558 2017-01-05 Jakub Jelinek <jakub@redhat.com>
6560 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
6561 3 argument Alias with unlimited for the negative form.
6562 (fno-vect-cost-model): Removed.
6564 2017-01-05 Martin Liska <mliska@suse.cz>
6566 * hsa-gen.c (gen_hsa_divmod): New function.
6567 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
6569 2017-01-05 Martin Liska <mliska@suse.cz>
6572 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
6575 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6577 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
6578 small constant length operands.
6580 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6582 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
6583 between loop iterations.
6585 2017-01-05 Martin Liska <mliska@suse.cz>
6588 * gimplify.c (gimplify_decl_expr): Compare to
6589 asan_poisoned_variables instread of checking flags.
6590 (gimplify_target_expr): Likewise.
6591 (gimplify_expr): Likewise.
6592 (gimplify_function_tree): Conditionally initialize
6593 asan_poisoned_variables.
6595 2017-01-04 Jeff Law <law@redhat.com>
6597 PR tree-optimizatin/78812
6598 * rtl.h (contains_mem_rtx_p): Prototype.
6599 * ifcvt.c (containts_mem_rtx_p): Move from here to...
6600 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
6601 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
6602 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
6603 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
6605 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
6607 * input.c (assert_char_at_range): Default-initialize actual_range.
6609 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
6611 * df-scan.c (df_ref_create_structure): Make regno unsigned,
6612 to match the caller.
6614 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
6616 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
6617 insns after final jump in test to emit dummy move.
6619 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
6621 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
6622 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
6624 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
6626 * multiple_target.c (create_dispatcher_calls): Init e_next.
6627 * tree-ssa-loop-split.c (split_loop): Init border.
6628 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
6631 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
6636 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
6637 (altivec_register_operand): Do not return true if the operand
6638 contains a SUBREG mixing SImode and SFmode.
6639 (vsx_register_operand): Likewise.
6640 (vsx_reg_sfsubreg_ok): New predicate.
6641 (vfloat_operand): Do not return true if the operand contains a
6642 SUBREG mixing SImode and SFmode.
6643 (vint_operand): Likewise.
6644 (vlogical_operand): Likewise.
6645 (gpc_reg_operand): Likewise.
6646 (int_reg_operand): Likewise.
6647 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
6648 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
6649 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
6651 (rs6000_emit_move_si_sf_subreg): New helper function.
6652 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
6653 fixup SUBREGs involving SImode and SFmode.
6654 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
6655 numbers for the new peephole2 optimization.
6656 (peephole2 for SFmode unions): New peephole2 to optimize cases in
6657 the GLIBC math library that do AND/IOR/XOR operations on single
6658 precision floating point.
6659 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
6660 target macros to say whether we need to avoid SUBREGs mixing
6662 (TARGET_ALLOW_SF_SUBREG): Likewise.
6663 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
6664 (UNSPEC_SI_FROM_SF): Likewise.
6665 (iorxor): Change spacing.
6666 (and_ior_xor): New iterator for AND, IOR, and XOR.
6667 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
6668 (movdi_from_sf_zero_ext): Likewise.
6669 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
6670 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
6671 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
6672 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
6673 (fms<mode>4): Likewise.
6674 (fnma<mode>4): Likewise.
6675 (fnms<mode>4): Likewise.
6676 (nfma<mode>4): Likewise.
6677 (nfms<mode>4): Likewise.
6679 2017-01-04 Marek Polacek <polacek@redhat.com>
6682 * doc/invoke.texi: Document -Wpointer-compare.
6684 2017-01-04 Jakub Jelinek <jakub@redhat.com>
6686 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
6689 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
6690 descriptions for -gdwarf-5 and emit them as uleb128 instead of
6693 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
6696 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
6697 documentation of the powerpc_popcntb_ok attribute.
6698 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
6699 code to issue warning messages if a requested CPU configuration is
6700 not supported by the binary (assembler and loader) toolchain.
6701 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
6702 made to define a built-in function that has been disabled.
6703 (paired_init_builtins): Add assertion to prevent ICE if attempt is
6704 made to define a built-in function that has been disabled.
6705 (altivec_init_builtins): Add comment explaining why definition
6706 of the DST built-in functions is not preceded by an assertion
6707 check. Add assertions to prevent ICE if attempts are made to
6708 define an altivec predicate or an abs* built-in function that has
6710 (htm_init_builtins): Add comment explaining why definition of the
6711 htm built-in functions is not preceded by an assertion check.
6713 2017-01-04 Jeff Law <law@redhat.com>
6715 PR tree-optimizatin/67955
6716 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
6717 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
6718 the points-to solution does not include pt_null. Use DECL_PT_UID
6721 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
6723 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
6724 Use gen_int_mode instead of gen_lopwart for const_int operands.
6726 2017-01-04 Jakub Jelinek <jakub@redhat.com>
6728 PR tree-optimization/71563
6729 * match.pd: Simplify X << Y into X if Y is known to be 0 or
6730 out of range value - has low bits known to be zero.
6732 2017-01-04 Alan Modra <amodra@gmail.com>
6734 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
6735 * configure: Regenerate.
6736 * config.in: Regenerate.
6738 2017-01-04 Jakub Jelinek <jakub@redhat.com>
6741 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
6742 a substring of the message, but strcmp with the whole message. Ifdef
6743 ENABLE_NLS, translate the message first using dgettext.
6745 2017-01-03 Jeff Law <law@redhat.com>
6747 PR tree-optimizatin/78856
6748 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
6749 (mark_threaded_blocks): Remove code to truncate thread paths that
6750 cross multiple loop headers. Instead invalidate the cached loop
6751 iteration information and handle case of a thread path walking
6752 into an irreducible region.
6754 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
6757 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
6758 assertions. Add support for doing the signbit if the IEEE 128-bit
6759 floating point value is in a GPR.
6760 * config/rs6000/rs6000.md (Fsignbit): Delete.
6761 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
6762 Update the length attribute if the value is in a GPR.
6763 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
6764 the sign or zero extension instruction, since the value is always 0/1.
6765 (signbit<mode>2_dm2): Delete using <Fsignbit>.
6768 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
6769 extracting SImode to a GPR register so that we can generate a
6770 store, limit the vector to be in a traditional Altivec register
6771 for the vextuwrx instruction.
6773 2017-01-03 Ian Lance Taylor <iant@google.com>
6775 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
6777 2017-01-03 Martin Sebor <msebor@redhat.com>
6779 PR tree-optimization/78696
6780 * gimple-ssa-sprintf.c (format_floating): Correct handling of
6781 precision. Use MPFR for %f for greater fidelity. Correct handling
6783 (pass_sprintf_length::compute_format_length): Set width and precision
6784 specified by asrerisk to void_node for vararg functions.
6785 (try_substitute_return_value): Adjust dump output.
6787 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
6789 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
6791 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
6793 * doc/invoke.texi (SPARC options): Document -mlra as the default.
6794 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
6795 -mlra/-mno-lra was passed to the compiler.
6797 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
6799 PR rtl-optimization/65618
6800 * emit-rtl.c (try_split): Move initialization of "before" and
6801 "after" to just before the call to emit_insn_after_setloc.
6803 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
6805 * doc/md.texi (Standard Names): Remove reference to Java frontend.
6807 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
6809 * dwarf2out.c (gen_enumeration_type_die): When
6810 -gno-strict-dwarf, add a DW_AT_encoding attribute.
6812 2017-01-03 Jakub Jelinek <jakub@redhat.com>
6814 PR tree-optimization/78965
6815 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
6816 Change first argument from const call_info & to call_info &. For %n
6817 set info.nowrite to false.
6820 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
6821 possibly throwing calls.
6823 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
6824 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
6825 and fns handling, rather than in a separate case SSA_NAME.
6827 2017-01-02 Jeff Law <law@redhat.com>
6829 * config/darwin-driver.c (darwin_driver_init): Const-correctness
6830 fixes for first_period and second_period variables.
6832 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
6835 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
6836 (*insvqi_1): New insn pattern.
6837 (*insvqi_1_mem_rex64): Ditto.
6839 (*insvqi_3): Rename from *insvqi.
6841 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
6843 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
6845 * doc/cfg.texi (Edges): Remove reference to Java.
6846 (Maintaining the CFG): Ditto.
6848 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
6851 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
6852 transparent aliases.
6854 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
6857 * predict.def (PRED_CALL): Update hitrate.
6858 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
6859 * predict.c (tree_estimate_probability_bb): Split CALL predictor
6860 into direct/indirect/polymorphic variants.
6862 2017-01-01 Jakub Jelinek <jakub@redhat.com>
6864 Update copyright years.
6866 * gcc.c (process_command): Update copyright notice dates.
6867 * gcov-dump.c (print_version): Ditto.
6868 * gcov.c (print_version): Ditto.
6869 * gcov-tool.c (print_version): Ditto.
6870 * gengtype.c (create_file): Ditto.
6871 * doc/cpp.texi: Bump @copying's copyright year.
6872 * doc/cppinternals.texi: Ditto.
6873 * doc/gcc.texi: Ditto.
6874 * doc/gccint.texi: Ditto.
6875 * doc/gcov.texi: Ditto.
6876 * doc/install.texi: Ditto.
6877 * doc/invoke.texi: Ditto.
6879 Copyright (C) 2017 Free Software Foundation, Inc.
6881 Copying and distribution of this file, with or without modification,
6882 are permitted in any medium without royalty provided the copyright
6883 notice and this notice are preserved.