1 2017-03-28 Bin Cheng <bin.cheng@arm.com>
3 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right
6 2017-03-28 Bin Cheng <bin.cheng@arm.com>
8 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
9 mark new edge's irreducible flag accordign to it.
10 (vect_do_peeling): Check loop preheader edge's irreducible flag
11 and pass it to function slpeel_add_loop_guard.
13 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
15 PR tree-optimization/80218
16 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
17 Update block frequencies and counts.
19 2017-03-28 Richard Biener <rguenther@suse.de>
21 PR tree-optimization/78644
22 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
23 of a simplification result we may not use it at all.
25 2017-03-28 Richard Biener <rguenther@suse.de>
28 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
29 without arguments, generate default definition of a SSA name.
31 2017-03-28 Richard Biener <rguenther@suse.de>
34 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
35 TYPE_REF_CAN_ALIAS_ALL references.
36 * fold-const.c (fold_indirect_ref_1): Likewise.
38 2017-03-28 Martin Liska <mliska@suse.cz>
41 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
42 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
44 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
45 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
47 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
48 (EXTRA_SPECS): Define.
49 (SUBTARGET_EXTRA_SPECS): Likewise.
50 (SUBTARGET_CPP_SPEC): Likewise.
51 * config/arc/elf.h (EXTRA_SPECS): Renamed to
52 SUBTARGET_EXTRA_SPECS.
53 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
55 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
57 * config/arc/simdext.md (vst64_insn): Update pattern.
58 (vld32wh_insn): Likewise.
59 (vld32wl_insn): Likewise.
60 (vld64_insn): Likewise.
61 (vld32_insn): Likewise.
63 2017-03-28 Marek Polacek <polacek@redhat.com>
66 * fold-const.c (fold_comparison): Use protected_set_expr_location
67 instead of SET_EXPR_LOCATION.
69 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
71 * tree.c (add_expr): Avoid name lookup warning.
73 2017-03-27 Jeff Law <law@redhat.com>
75 PR tree-optimization/80216
76 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
77 function name. Limit recursion depth.
78 (record_temporary_equivalences): Corresponding changes.
80 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
82 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
85 2017-03-27 Jakub Jelinek <jakub@redhat.com>
88 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
90 * cfgcleanup.c (reg_note_cfa_p): New array.
91 (insns_have_identical_cfa_notes): New function.
92 (old_insns_match_p): Don't cross-jump in between /f
93 and non-/f instructions. If both i1 and i2 are frame related,
94 verify all CFA notes, their order and content.
96 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
99 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
100 HImode and SImode with zero extend to DImode to one insn.
101 (bswap<mode>2_extenddi): Likewise.
102 (bswapsi2_extenddi): Likewise.
103 (bswaphi2_extendsi): Likewise.
104 (bswaphi2): Combine bswap HImode and SImode into one insn.
105 Separate memory insns from swapping register.
106 (bswapsi2): Likewise.
107 (bswap<mode>2): Likewise.
108 (bswaphi2_internal): Delete, no longer used.
109 (bswapsi2_internal): Likewise.
110 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
111 store, and gpr<-gpr swap insns.
112 (bswap<mode>2_store): Likewise.
113 (bswaphi2_reg): Register only splitter, combine with the splitter.
114 (bswaphi2 splitter): Likewise.
115 (bswapsi2_reg): Likewise.
116 (bswapsi2 splitter): Likewise.
117 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
118 the insns into load, store, and register/register insns.
119 (bswapdi2_ldbrx): Likewise.
120 (bswapdi2_load): Likewise.
121 (bswapdi2_store): Likewise.
122 (bswapdi2_reg): Likewise.
124 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
126 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
127 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
129 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
132 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
134 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
135 special handling for target option conflicts between dform
136 options (-mpower9-dform, -mpower9-dform-vector,
137 -mpower9-dform-scalar) and -mno-direct-move.
139 2017-03-27 Richard Biener <rguenther@suse.de>
141 PR tree-optimization/80181
142 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
144 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
146 * config/arc/predicates.md (move_double_src_operand): Replace the
147 call to move_double_src_operand with a call to address_operand.
149 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
151 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
152 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
153 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
155 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
157 * config/arc/predicates.md (long_immediate_loadstore_operand):
158 Consider scaled addresses cases.
160 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
162 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
163 restored when in interrupt.
164 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
165 doesn't have delay slot.
167 2017-03-27 Richard Biener <rguenther@suse.de>
170 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
171 inlined thunk clones.
173 2017-03-27 Jakub Jelinek <jakub@redhat.com>
176 * asan.c (instrument_derefs): Copy over last operand from
177 original COMPONENT_REF to the new COMPONENT_REF with
178 DECL_BIT_FIELD_REPRESENTATIVE.
179 * ubsan.c (instrument_object_size): Likewise.
181 2017-03-27 Richard Biener <rguenther@suse.de>
183 PR tree-optimization/80170
184 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
185 sure DR/SCEV didnt fold in constants we do not see when looking
186 at the reference base alignment.
188 2017-03-27 Richard Biener <rguenther@suse.de>
191 * gimple-fold.c (fold_ctor_reference): Properly guard against
192 NULL return value from canonicalize_constructor_val.
194 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
197 * config/i386/i386.c (ix86_expand_builtin)
198 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
199 flags reg setting and flags reg using instructions.
200 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
201 clobbering instructions to zero extend op2.
203 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
205 * doc/install.texi (Configuration) <--with-aix-soname>:
206 Update link to AIX ld.
208 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
210 PR rtl-optimization/80160
211 PR rtl-optimization/80159
212 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
213 reg_alternate_class into account.
215 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
218 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
219 to consider in curr_insn_transform.
221 2017-03-24 Jakub Jelinek <jakub@redhat.com>
223 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
224 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
227 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
229 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
230 argument to the overloaded builtin variants. Use the new flag to
231 deprecate certain builtin variants.
232 * config/s390/s390-builtin-types.def: Add new builtin types.
233 * config/s390/s390-builtins.h: Support new flags field for
235 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
236 (s390_macro_to_expand): Enable vector float data type.
237 (s390_cpu_cpp_builtins_internal): Indicate support of the new
238 builtins by incrementing the __VEC__ version number.
239 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
241 (s390_resolve_overloaded_builtin): Emit error messages depending
242 on the builtin flags.
243 * config/s390/s390.c (s390_expand_builtin): Support additional
244 flags argument. Change error message to match the messages
246 * config/s390/s390.md: New UNSPEC_* constants.
247 (op_type): Add new instruction types.
248 * config/s390/vecintrin.h: Add new builtins and test data class
250 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
251 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
252 (VEC_INEXACT, VEC_NOINEXACT): New constants.
253 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
254 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
255 ("vec_mergel<mode>"): V_HW -> VEC_HW.
257 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
258 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
259 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
260 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
262 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
263 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
264 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
265 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
267 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
268 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
269 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
270 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
271 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
272 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
273 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
275 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
276 ("vec_scatter_element<V_HW_4:mode>_DI")
277 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
278 ("vec_fpint<mode>", "vflls")
279 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
280 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
281 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
282 ("*vec_cmphe<mode>_cc"): ... these.
284 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
285 mode constant instead of magic value.
287 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
289 * config/s390/s390.c (s390_expand_vec_compare): Support other
290 vector floating point modes than just V2DF.
291 (s390_expand_vcond): Likewise.
292 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
293 (s390_cannot_change_mode_class): Prevent mode changes between TF
294 and V1TF in vector registers.
295 * config/s390/s390.md (DF, SF): New mode attributes.
296 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
297 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
298 SFmode support for VRs.
299 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
301 (VFT, VF_HW): New mode iterators.
302 (vw, sdx): New mode attributes.
303 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
304 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
305 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
306 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
307 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
308 also the new vector floating point modes. Renaming to ...
310 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
311 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
312 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
313 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
314 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
315 ("vec_unordered<mode>"): ... these.
317 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
318 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
319 ("*vec_extendv2df"): New insn definitions.
321 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
323 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
324 ("mulditi3_2", "*muldi3_sign"): New patterns.
325 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
326 rename the pattern definition.
328 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
330 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
332 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
334 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
336 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
337 instruction if possible.
338 * config/s390/vector.md (vec_halfnumelts): New mode
340 ("*vec_vllezlf<mode>"): New pattern.
342 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
344 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
345 ("popcountv4si2", "popcountv2di2"): Rename to ...
346 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
347 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
349 ("popcount<mode>2_vxe"): New pattern.
351 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
353 * common/config/s390/s390-common.c (processor_flags_table): Add
355 * config.gcc: Add arch12.
356 * config/s390/driver-native.c (s390_host_detect_local_cpu):
357 Default to arch12 for unknown CPU model numbers.
358 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
359 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
360 PROCESSOR_max sanity check.
361 * config/s390/s390-opts.h (enum processor_type): Add
363 * config/s390/s390.c (processor_table): Add arch12.
364 (s390_expand_builtin): Add check for B_VXE flag.
365 (s390_issue_rate): Add PROCESSOR_ARCH12.
366 (s390_get_sched_attrmask): Likewise.
367 (s390_get_unit_mask): Likewise.
368 (s390_sched_score): Enable z13 scheduling for arch12.
369 (s390_sched_reorder): Likewise.
370 (s390_sched_variable_issue): Likewise.
371 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
373 (s390_tune_attr): Use z13 scheduling also for arch12.
374 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
375 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
376 (TARGET_VXE_P): New macros.
377 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
378 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
379 * config/s390/s390.opt: Add arch12 as processor_type.
381 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
383 * config/s390/s390.md
384 ("fixuns_truncdddi2", "fixuns_trunctddi2")
385 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
386 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
388 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
389 Rename expanders to ...
391 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
392 ("fixuns_truncdddi2_emu"): ... these.
394 ("fixuns_trunc<mode>si2_emu"): New expander.
396 ("*fixuns_truncdfdi2_z13"): Rename to ...
397 ("*fixuns_truncdfdi2_vx"): ... this.
399 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
401 * config/s390/2964.md: Remove the single element vector compare
402 instructions which are no longer used.
403 * config/s390/s390.c (s390_select_ccmode): Remove handling of
405 (s390_canonicalize_comparison): Remove handling of DFmode
407 (s390_expand_vec_compare_scalar): Remove function.
408 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
409 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
411 ("*cmp<mode>_ccs"): Add wfcdb instruction.
413 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
415 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
417 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
418 will anyway by matched by mov<mode>_64dfp.
420 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
422 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
423 vlef/vstef. Add missing operand to vleif.
425 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
427 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
428 pair for all vector types with 64 bit elements.
429 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
430 * config/s390/vector.md (V_HW_64): ... here.
431 (V_128_NOSINGLE): New mode iterator.
432 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
433 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
434 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
435 ("*vec_load_pairv2di"): Change to ...
436 ("*vec_load_pair<mode>"): ... this one.
438 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
440 * config/s390/constraints.md: Add comments.
441 (jKK): Reject element sizes > 8 bytes.
442 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
444 * config/s390/s390.md: Add the s_operand checks formerly in
445 s390_split_ok_p to various splitters where they are still
447 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
448 for 128 bit vectors. Plus two splitters.
450 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
452 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
455 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
458 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
459 error if the boundary argument is not constant.
461 2017-03-24 Jakub Jelinek <jakub@redhat.com>
463 PR rtl-optimization/80112
464 * loop-doloop.c (doloop_condition_get): Don't check condition
465 if cmp isn't SET with IF_THEN_ELSE src.
467 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
469 PR tree-optimization/80158
470 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
471 replacing a candidate statement, also replace it for the
472 candidate's alternate interpretation.
473 (replace_rhs_if_not_dup): Likewise.
474 (replace_one_candidate): Likewise.
476 2017-03-24 Richard Biener <rguenther@suse.de>
478 PR tree-optimization/80167
479 * graphite-isl-ast-to-gimple.c
480 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
482 (translate_isl_ast_to_gimple::get_rename): Likewise.
484 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
486 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
487 handling of certain combinations of target options, including the
488 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
489 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
491 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
494 * config/arm/arm.md (*load_multiple): Add reload_completed to
497 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
498 Richard Biener <rguenth@suse.com>
500 PR tree-optimization/79908
501 PR tree-optimization/80136
502 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
503 been cast away, gimplify_and_add suffices.
505 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
507 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
509 2017-03-23 Richard Biener <rguenther@suse.de>
511 PR tree-optimization/80032
512 * gimplify.c (gimple_push_cleanup): Forced unconditional
513 cleanups still have to go to the conditional_cleanups
516 2017-03-22 Jakub Jelinek <jakub@redhat.com>
518 PR tree-optimization/80072
519 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
521 (next_operand_entry_id): Change type to unsigned int.
522 (sort_by_operand_rank): Make sure to return the right return value
523 even if unsigned fields are bigger than INT_MAX.
524 (struct oecount): Change cnt and id type to unsigned int.
525 (oecount_hasher::equal): Formatting fix.
526 (oecount_cmp): Make sure to return the right return value
527 even if unsigned fields are bigger than INT_MAX.
528 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
531 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
532 TREE_READONLY on result if writing it more than once.
535 * doc/invoke.texi (-fsanitize=thread): Document that with
536 -fnon-call-exceptions atomics are not able to throw
540 * tsan.c: Include tree-eh.h.
541 (instrument_builtin_call): Call maybe_clean_eh_stmt or
542 maybe_clean_or_replace_eh_stmt where needed.
543 (instrument_memory_accesses): Add cfg_changed argument.
544 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
546 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
548 PR rtl-optimization/63191
549 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
550 wrapper function, moved the whole old content into ...
551 (ix86_delegitimize_address_1): ... this. New inline function.
552 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
553 true as last argument instead of ix86_delegitimize_address.
555 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
557 * config/aarch64/aarch64.c (generic_branch_cost): Copy
558 cortexa57_branch_cost.
560 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
562 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
564 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
567 * doc/md.texi (Constraints): Document wA constraint.
568 * config/rs6000/constraints.md (wA): New.
569 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
570 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
571 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
572 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
574 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
577 * gimplify.c (is_oacc_declared): New function.
578 (oacc_default_clause): Use it to set default flags for acc declared
579 variables inside parallel regions.
580 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
582 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
583 declare attribute to any decl as necessary.
585 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
588 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
589 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
590 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
591 (arm_arch_lpae): This.
592 * config/arm/arm.c (arm_arch7ve): Rename into ...
593 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
594 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
597 2017-03-22 Martin Liska <mliska@suse.cz>
600 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
601 error message instead of an ICE.
603 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
605 * doc/extend.texi (6.11 Additional Floating Types): Revise.
607 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
609 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
611 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
614 2017-03-21 Martin Sebor <msebor@redhat.com>
616 * doc/extend.texi: Use "cannot" instead of "can't."
617 * doc/hostconfig.texi: Same.
618 * doc/install.texi: Same.
619 * doc/invoke.texi: Same.
620 * doc/loop.texi: Same.
622 * doc/objc.texi: Same.
623 * doc/rtl.texi: Same.
625 * doc/tm.texi.in: Same.
626 * doc/trouble.texi: Same.
628 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
631 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
632 (collect_checksum_attributes): Set it.
633 (die_checksum_ordered): Use it.
635 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
637 PR tree-optimization/79908
638 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
639 change: For a VA_ARG whose LHS has been cast away, use
640 force_gimple_operand to construct the side effects.
642 2017-03-21 David Malcolm <dmalcolm@redhat.com>
645 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
646 more amenable to translation.
647 (oacc_loop_auto_partitions): Likewise.
649 2017-03-21 Marek Polacek <polacek@redhat.com>
650 Martin Sebor <msebor@redhat.com>
652 PR tree-optimization/80109
653 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
656 2017-03-21 Jakub Jelinek <jakub@redhat.com>
657 Segher Boessenkool <segher@kernel.crashing.org>
660 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
661 check reg_used_between_p between insn and one of succ or succ2
662 depending on if succ is artificial insn not inserted into insn
665 2017-03-21 Martin Liska <mliska@suse.cz>
667 PR gcov-profile/80081
668 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
669 * doc/gcc.texi: Include gcov-dump stuff.
670 * doc/gcov-dump.texi: New file.
672 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
674 PR rtl-optimization/79150
675 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
676 conditional jump, if the jump is the last insn of the loop.
678 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
679 Richard Biener <rguenth@suse.com>
681 PR tree-optimization/79908
682 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
683 been cast away, use force_gimple_operand to construct the side
686 2017-03-21 Martin Liska <mliska@suse.cz>
689 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
692 2017-03-21 Brad Spengler <spender@grsecurity.net>
695 * plugin.c (htab_hash_plugin): New function.
696 (add_new_plugin): Use it and adjust.
697 (parse_plugin_arg_opt): Adjust.
698 (init_one_plugin): Likewise.
700 2017-03-21 Richard Biener <rguenther@suse.de>
702 PR tree-optimization/80032
703 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
704 if set force the cleanup to happen unconditionally.
705 (gimplify_target_expr): Push inserted clobbers with force_uncond
706 to avoid them being removed by control-dependent DCE.
708 2017-03-21 Richard Biener <rguenther@suse.de>
710 PR tree-optimization/80122
711 * tree-inline.c (copy_bb): Do not expans va-arg packs or
712 va_arg_pack_len when the inlined call stmt requires pack
714 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
716 2017-03-21 Jakub Jelinek <jakub@redhat.com>
719 * tsan.c (instrument_builtin_call): If the memory model argument
720 is not a constant, assume it is valid.
723 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
726 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
728 PR rtl-optimization/79910
729 * combine.c (can_combine_p): Do not allow combining an I0 or I1
730 if its dest is used by an insn before I2 (other than the combined
731 insns themselves, which are properly handled already).
733 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
736 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
738 * combine.c (record_used_regs): New static function.
739 (try_combine): Handle situations where there is an additional
740 instruction between I2 and I3 which needs to have a LOG_LINK
744 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
746 * combine.c (try_combine): Delete redundant i1 test. Call
747 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
749 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
752 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
755 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
757 PR tree-optimization/80054
758 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
759 the optimization if a PHI or any of its arguments is not dominated
760 by the candidate's basis. Use gphi* rather than gimple* as
762 (replace_profitable_candidates): Clean up a gimple* variable that
763 should be a gphi* variable.
765 2017-03-20 Martin Sebor <msebor@redhat.com>
768 * doc/extend.texi (attribute constructor): Document present limitation.
770 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
773 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
774 __POWER9_VECTOR__ #ifdef control, change template definition to
775 use Power9-specific built-in function.
776 (vec_any_eq): Likewise.
777 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
778 to control outcomes from this test.
779 (vector_ae_<mode>p): For VEC_F modes, likewise.
781 2017-03-20 Ian Lance Taylor <iant@google.com>
783 * config/i386/i386.c (ix86_function_regparm): Save an extra
784 register for -fsplit-stack with DECL_STATIC_CHAIN.
786 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
789 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
790 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
792 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
794 * config/riscv/riscv.c (riscv_print_operand): Use "fence
796 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
799 2017-03-20 Marek Polacek <polacek@redhat.com>
802 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
804 2017-03-20 Richard Biener <rguenther@suse.de>
806 PR tree-optimization/80113
807 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
808 allocate extra SSA name for PHI def.
809 (add_close_phis_to_outer_loops): Likewise.
810 (add_close_phis_to_merge_points): Likewise.
811 (copy_loop_close_phi_args): Likewise.
812 (copy_cond_phi_nodes): Likewise.
814 2017-03-20 Martin Liska <mliska@suse.cz>
817 * tree-chkp.c (chkp_build_returned_bound): Do not build
818 returned bounds for a LHS that's not a BOUNDED_P type.
820 2017-03-20 Martin Liska <mliska@suse.cz>
824 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
825 COMPLEX_CST and VECTOR_CST.
827 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
830 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
831 target operand. A new splitter adds the clobber statement in case
832 the target operand is dead anyway.
834 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
836 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
837 to age-old versions of binutils and glibc.
839 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
841 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
843 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
845 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
847 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
849 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
850 requirement for binutils 2.13.
852 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
854 * combine.c (try_combine): Delete redundant i1 test. Call
855 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
857 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
859 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
860 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
862 <riscv64-*-elf>: Re-arrange section
863 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
864 <riscv32-*-linux>: Likewise.
865 <riscv64-*-elf>: Likewise
866 <riscv64-*-linux>: Likewise.
868 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
871 * aarch64.opt(verbose-cost-dump): Fix typo.
873 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
876 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
877 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
879 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
881 * reload.c (find_reloads): When reloading a nonoffsettable address,
882 use RELOAD_OTHER for it and its address reloads.
884 PR rtl-optimization/79910
885 * combine.c (record_used_regs): New static function.
886 (try_combine): Handle situations where there is an additional
887 instruction between I2 and I3 which needs to have a LOG_LINK
890 2017-03-17 Jeff Law <law@redhat.com>
892 PR tree-optimization/71437
893 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
894 conditional in the hash table first.
895 (vrp_dom_walker::before_dom_children): Extract condition from
896 ASSERT_EXPR. Record condition, its inverion and any implied
899 2017-03-17 Marek Polacek <polacek@redhat.com>
900 Markus Trippelsdorf <markus@trippelsdorf.de>
902 PR tree-optimization/80079
903 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
906 2017-03-17 Richard Biener <rguenther@suse.de>
909 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
910 Properly verify the LHS before the RHS possibly claims to be
912 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
915 2017-03-17 Martin Jambor <mjambor@suse.cz>
917 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
918 (List of -O2 options): Likewise.
919 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
922 2017-03-17 Tom de Vries <tom@codesourcery.com>
924 * gcov-dump.c (print_usage): Print bug_report_url.
926 2017-03-17 Richard Biener <rguenther@suse.de>
929 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
930 (parser::peek): Likewise.
932 2017-03-17 Richard Biener <rguenther@suse.de>
934 PR tree-optimization/80048
935 * sese.c (free_sese_info): Properly release rename_map and
936 copied_bb_map elements.
938 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
940 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
941 Add linked-list forward and backlinks. Insert on
942 construction, remove on destruction.
943 (class pass_store_merging): Add m_stores_head field.
944 (pass_store_merging::terminate_and_process_all_chains):
945 Iterate over m_stores_head list.
946 (pass_store_merging::terminate_all_aliasing_chains):
948 (pass_store_merging::execute): Check for debug stmts first.
949 Push new chains onto the m_stores_head stack.
951 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
954 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
955 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
956 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
958 2017-03-16 Jeff Law <law@redhat.com>
960 PR tree-optimization/71437
961 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
962 member function. Implementation moved into after_dom_children
963 member function and into the threader's thread_outgoing_edges
965 (dom_opt_dom_walker::after_dom_children): Simplify by moving
966 some code into new thread_outgoing_edges.
967 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
968 definition. Simplify marker handling (do it here). Assume we always
969 have the available expression and the const/copies tables.
970 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
972 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
973 * tree-vrp.c (equiv_stack): No longer file scoped.
974 (vrp_dom_walker): New class.
975 (vrp_dom_walker::before_dom_children): New member function.
976 (vrp_dom_walker::after_dom_children): Likewise.
977 (identify_jump_threads): Setup domwalker. Use it rather than
978 walking edges in a random order by hand. Simplify setup/finalization.
979 (finalize_jump_threads): Remove.
980 (vrp_finalize): Do not call identify_jump_threads here.
981 (execute_vrp): Do it here instead and call thread_through_all_blocks
984 PR tree-optimization/71437
985 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
987 (simplify_stmt_for_jump_threading): Add basic_block argument. All
989 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
990 (dom_opt_dom_walker::thread_across_edge): Remove
991 handle_dominating_asserts argument. All callers changed.
992 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
993 changes. Remove calls to lhs_of_dominating_assert. Other
994 uses of handle_dominating_asserts turn into unconditional code
995 (simplify_control_stmt_condition_1): Likewise.
996 (simplify_control_stmt_condition): Likewise.
997 (thread_through_normal_block, thread_across_edge): Likewise.
998 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
999 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
1000 object if it is not an SSA_NAME.
1001 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
1002 before calling into the VRP specific simplifiers.
1003 (identify_jump_threads): Remove handle_dominating_asserts
1006 2017-03-16 Jakub Jelinek <jakub@redhat.com>
1009 * tree-diagnostic.c (default_tree_printer): No longer static.
1010 * tree-diagnostic.h (default_tree_printer): New prototype.
1012 2017-03-16 Tamar Christina <tamar.christina@arm.com>
1014 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
1015 Change ins into fmov.
1017 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1019 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
1020 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
1021 Use h_con constraint for operand 1.
1022 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
1023 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
1025 2017-03-15 Jeff Law <law@redhat.com>
1027 PR tree-optimization/71437
1028 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
1029 (record_temporary_equivalences): Use it.
1031 PR tree-optimization/71437
1032 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
1033 tree-ssa-scopedtables.
1034 (lookup_avail_expr, build_and_record_new_cond): Likewise.
1035 (record_conditions, record_cond, vuse_eq): Likewise.
1036 (record_edge_info): Adjust to API tweak of record_conditions.
1037 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
1038 (record_temporary_equivalences, optimize_stmt): Likewise.
1039 (eliminate_redundant_computations): Likewise.
1040 (record_equivalences_from_stmt): Likewise.
1041 * tree-ssa-scopedtables.c: Include options.h and params.h.
1042 (vuse_eq): New function, moved from tree-ssa-dom.c
1043 (build_and_record_new_cond): Likewise.
1044 (record_conditions): Likewise. Accept vector of conditions rather
1045 than edge_equivalence structure for first argument.
1046 for the first argument.
1047 (avail_exprs_stack::lookup_avail_expr): New member function, moved
1048 from tree-ssa-dom.c.
1049 (avail_exprs_stack::record_cond): Likewise.
1050 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
1051 from tree-ssa-dom.c.
1052 (avail_exprs_stack): Add new member functions lookup_avail_expr
1054 (record_conditions): Declare.
1056 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
1059 * lra-constraints.c (process_alt_operands): Increase reject for
1060 reloading an input/output operand.
1062 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
1065 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
1066 insns to convert from signed/unsigned char/short to IEEE 128-bit
1068 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
1070 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
1073 * config/i386/i386.c (ix86_vector_duplicate_value): Create
1074 subreg of inner mode for values already in registers.
1076 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
1078 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
1079 iteration reg is used after the loop.
1081 2017-03-14 Martin Sebor <msebor@redhat.com>
1083 PR tree-optimization/79800
1084 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
1085 precision in negative-positive range.
1086 (format_floating): Call non-const overload with adjusted precision.
1088 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
1091 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
1094 2017-03-14 Martin Sebor <msebor@redhat.com>
1097 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
1098 * builtins.def (aligned_alloc): Use it.
1101 * Makefile.in (GTFILES): Add calls.c.
1102 * calls.c: Include "gt-calls.h".
1104 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
1106 PR rtl-optimization/79728
1107 * regs.h (struct target_regs): New field
1108 x_contains_allocatable_regs_of_mode.
1109 (contains_allocatable_regs_of_mode): New macro.
1110 * reginfo.c (init_reg_sets_1): Initialize it, and change
1111 contains_reg_of_mode so it includes global regs as well.
1112 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
1113 rather than contains_regs_of_mode.
1115 2017-03-14 Martin Liska <mliska@suse.cz>
1117 * doc/invoke.texi: Document options that can't be combined with
1118 -fcheck-pointer-bounds.
1120 2017-03-14 Martin Liska <mliska@suse.cz>
1123 * doc/invoke.texi (-Wchkp): Document the option.
1125 2017-03-14 Martin Liska <mliska@suse.cz>
1127 * Makefile.in: Install gcov-dump.
1129 2017-03-14 Martin Liska <mliska@suse.cz>
1131 * multiple_target.c (expand_target_clones): Bail out for
1132 an invalid attribute.
1134 2017-03-14 Richard Biener <rguenther@suse.de>
1136 * alias.c (struct alias_set_entry): Pack properly.
1137 * cfgloop.h (struct loop): Likewise.
1138 * cse.c (struct set): Likewise.
1139 * ipa-utils.c (struct searchc_env): Likewise.
1140 * loop-invariant.c (struct invariant): Likewise.
1141 * lra-remat.c (struct cand): Likewise.
1142 * recog.c (struct change_t): Likewise.
1143 * rtl.h (struct address_info): Likewise.
1144 * symbol-summary.h (function_summary): Likewise.
1145 * tree-loop-distribution.c (struct partition): Likewise.
1146 * tree-object-size.c (struct object_size_info): Likewise.
1147 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
1148 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
1149 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
1150 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
1151 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
1152 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
1153 (struct _stmt_vec_info): Likewise.
1155 2017-03-14 Martin Liska <mliska@suse.cz>
1158 * multiple_target.c (create_dispatcher_calls): Check that
1159 a target can create a function dispatcher.
1161 2017-03-14 Martin Liska <mliska@suse.cz>
1164 * multiple_target.c (expand_target_clones): Drop local.local
1165 flag for default implementation.
1167 2017-03-14 Richard Biener <rguenther@suse.de>
1169 PR tree-optimization/80030
1170 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
1172 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
1174 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
1175 gcc_fallthrough() instead of __attribute__((fallthrough));
1177 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
1179 * doc/gcc.texi: Remove "up" link to (DIR).
1180 * doc/gccint.texi: Ditto.
1182 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
1184 * doc/install.texi (Specific) <avr>: Remove reference to
1187 2017-03-13 Jeff Law <law@redhat.com>
1189 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
1190 attribute rather than comments.
1192 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
1193 match_scratch operand is highest.
1195 2017-03-13 Martin Liska <mliska@suse.cz>
1198 * ipa-pure-const.c (warn_function_noreturn): If the declarations
1199 is a CHKP clone, use original declaration.
1201 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
1203 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
1204 (arc_conditional_register_usage): Use a different allocation order
1205 when optimizing for size.
1206 * common/config/arc/arc-common.c (arc_option_optimization_table):
1207 Section anchors default on when optimizing for size.
1209 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
1211 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
1213 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
1215 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
1216 * config/arc/arc.md (cpu_facility): Add cd variant.
1217 (*movqi_insn): Add code density variant.
1218 (*movhi_insn): Likewise.
1219 (*movqi_insn): Likewise.
1220 (*addsi3_mixed): Likewise.
1221 (subsi3_insn): Likewise.
1223 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
1225 * config/arc/arc.md (movsi_cond_exec): Update constraint.
1227 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
1229 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
1230 expressions with MINUS and UNARY ops.
1232 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1235 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
1237 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
1238 between vec_select and vector argument.
1239 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
1240 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
1241 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
1242 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
1243 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
1244 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
1246 2017-03-13 Richard Biener <rguenther@suse.de>
1249 * params.def (vect-max-peeling-for-alignment): Fix typo.
1251 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
1253 * doc/install.texi (Specific) <mips-*-*>: Remove description of
1254 issue that only occurred with binutils below 2.18.
1256 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
1258 * doc/install.texi (Specific) <cris-axis-elf>: No longer
1259 refer to binutils 2.11/2.12 minimum.
1261 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
1263 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
1264 ftp.kernel.org and simplify binutils requirement.
1266 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
1268 * doc/invoke.texi (Warning Options): Fix spelling of link-time
1270 (Optimize Options): Ditto. Also remove redundancy.
1272 2017-03-10 David Malcolm <dmalcolm@redhat.com>
1274 PR translation/79848
1275 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
1277 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
1278 to G_ to avoid double translation.
1280 2017-03-10 David Malcolm <dmalcolm@redhat.com>
1282 PR translation/79923
1283 * auto-profile.c (get_combined_location): Convert leading
1284 character of diagnostics to lower case and remove trailing period.
1285 (read_profile): Likewise for various diagnostics.
1286 * config/arm/arm.c (arm_option_override): Remove trailing period
1287 from various diagnostics.
1288 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
1289 (msp430_expand_delay_cycles): Likewise.
1291 2017-03-10 David Malcolm <dmalcolm@redhat.com>
1294 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
1295 full command-line argument, rather than just "str".
1296 (aarch64_validate_march): Likewise.
1297 (aarch64_validate_mtune): Likewise.
1299 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
1301 PR rtl-optimization/78911
1302 * lra-assigns.c (must_not_spill_p): New function.
1303 (spill_for): Use it.
1305 2017-03-10 Jakub Jelinek <jakub@redhat.com>
1307 PR tree-optimization/79981
1308 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
1309 ATOMIC_COMPARE_EXCHANGE ifn result.
1310 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
1311 IFN_ATOMIC_COMPARE_EXCHANGE.
1313 2017-03-10 David Malcolm <dmalcolm@redhat.com>
1316 * opts.c (parse_sanitizer_options): Add missing question mark to
1317 "did you mean" message.
1319 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1321 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
1323 (VMULEUH_UNS): Likewise.
1324 (VMULOUB_UNS): Likewise.
1325 (VMULOUH_UNS): Likewise.
1326 * config/rs6000/rs6000.c (builtin_function_type): Remove
1327 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
1329 2017-03-10 David Malcolm <dmalcolm@redhat.com>
1332 * read-rtl-function.c (function_reader::read_rtx_operand): Update
1333 x with result of extra_parsing_for_operand_code_0.
1334 (function_reader::extra_parsing_for_operand_code_0): Convert
1335 return type from void to rtx, returning x. When reading
1336 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
1337 larger size containing struct block_symbol.
1339 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
1341 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
1342 -mfloat128-hardware without -m64.
1344 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
1347 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
1348 entries to the case statement that marks unsigned arguments to
1349 overloaded functions.
1351 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
1353 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
1354 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
1356 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
1359 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
1360 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
1362 2017-03-10 Martin Liska <mliska@suse.cz>
1366 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
1367 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
1370 2017-03-10 Olivier Hainque <hainque@adacore.com>
1372 * tree-switch-conversion (array_value_type): Start by resetting
1373 candidate type to it's main variant.
1375 2017-03-10 Jakub Jelinek <jakub@redhat.com>
1377 PR rtl-optimization/79909
1378 * combine.c (try_combine): Use simplify_replace_rtx on individual
1379 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
1380 of the whole CALL_INSN_FUNCTION_USAGE.
1382 PR tree-optimization/79972
1383 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
1384 get_range_info on SSA_NAMEs. Formatting fixes.
1386 2017-03-10 Richard Biener <rguenther@suse.de>
1387 Jakub Jelinek <jakub@redhat.com>
1389 PR tree-optimization/77975
1390 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
1391 edge to be constant.
1392 (get_val_for): For constant x return it. Formatting fix.
1393 (loop_niter_by_eval): Avoid pointless looping if the next iteration
1394 would use the same bases as the current one.
1396 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1398 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
1399 instead of vec_select for V1TImode.
1400 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
1402 (VSX_LE_128): Add V1TI to this mode iterator.
1403 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
1404 (*vsx_le_perm_store_<mode>): Likewise.
1405 (pre-reload splitter for VSX stores): Likewise.
1406 (post-reload splitter for VSX stores): Likewise.
1407 (*vsx_xxpermdi2_le_<mode>): Likewise.
1408 (*vsx_lxvd2x2_le_<mode>): Likewise.
1409 (*vsx_stxvd2x2_le_<mode>): Likewise.
1411 2017-03-09 Michael Eager <eager@eagercon.com>
1413 Correct failures with --enable-checking=yes,rtl.
1415 * config/microblaze/microblaze.c (microblaze_expand_shift):
1416 Replace GET_CODE test with CONST_INT_P and INTVAL test with
1417 test for const0_rtx.
1418 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
1419 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
1421 2017-03-09 Richard Biener <rguenther@suse.de>
1423 PR tree-optimization/79977
1424 * graphite-scop-detection.c (scop_detection::merge_sese):
1425 Handle the case of extra exits to blocks dominating the entry.
1427 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
1429 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
1432 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
1434 PR rtl-optimization/79949
1435 * lra-constraints.c (process_alt_operands): Check memory when
1436 trying to predict a cycle. Print about the overall increase.
1438 2017-03-09 Richard Biener <rguenther@suse.de>
1441 * gimple-expr.c (useless_type_conversion_p): Preserve
1442 TYPE_SATURATING for fixed-point types.
1444 2017-03-09 Richard Biener <rguenther@suse.de>
1447 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
1448 alignment of BLKmode params.
1450 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1453 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
1454 (VALL_NO_V2Q): Likewise.
1456 * config/aarch64/aarch64-simd.md
1457 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
1459 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
1460 VALL_NO_V2Q mode iterator.
1461 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
1463 2017-03-09 Martin Liska <mliska@suse.cz>
1465 PR tree-optimization/79631
1466 * tree-chkp-opt.c (chkp_is_constant_addr): Call
1467 tree_int_cst_sign_bit just for INTEGER constants.
1469 2017-03-09 Martin Liska <mliska@suse.cz>
1473 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
1476 2017-03-09 Marek Polacek <polacek@redhat.com>
1479 * tree.c (inchash::add_expr): Handle TREE_VEC.
1481 2017-03-09 Martin Liska <mliska@suse.cz>
1484 (chkp_narrow_size_and_offset): New function.
1485 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
1486 (void chkp_parse_bit_field_ref): New function.
1487 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
1488 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
1490 2017-03-09 Martin Liska <mliska@suse.cz>
1493 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
1494 (chkp_find_bounds_1): Remove gcc_unreachable.
1496 2017-03-09 Jakub Jelinek <jakub@redhat.com>
1499 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
1500 BUILT_IN_SYNC*, determine the access type from the size suffix and
1501 always build a MEM_REF with that type. Handle forgotten
1502 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
1505 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
1506 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
1507 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
1508 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
1509 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
1510 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
1511 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
1512 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
1513 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
1514 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
1515 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
1516 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
1517 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
1518 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
1519 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
1520 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
1521 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
1522 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
1523 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
1524 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
1525 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
1526 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
1527 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
1528 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
1529 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
1530 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
1531 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
1532 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
1533 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
1534 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
1535 definitions outside of __OPTIMIZE__ guarded section.
1538 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
1539 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
1540 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
1541 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
1544 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1546 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
1547 ("vfenez<mode>"): Add missing constraints.
1549 2017-03-08 Martin Sebor <msebor@redhat.com>
1552 * config/nds32/nds32.c (nds32_option_override):
1554 2017-03-08 Jakub Jelinek <jakub@redhat.com>
1557 * gimplify.c (gimplify_omp_for): Replace index var in outer
1558 taskloop statement with an artificial variable and add
1559 OMP_CLAUSE_PRIVATE clause for it.
1561 2017-03-08 Richard Biener <rguenther@suse.de>
1563 PR tree-optimization/79955
1564 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
1565 for accesses that are completely outside of the variable.
1567 2017-03-08 Andrew Haley <aph@redhat.com>
1569 PR tree-optimization/79943
1570 * tree-ssa-loop-split.c (compute_new_first_bound): When
1571 calculating the new upper bound, (END-BEG) should be added, not
1574 2017-03-08 Jakub Jelinek <jakub@redhat.com>
1576 * config/avr/avr.md (setmemhi): Make sure match_dup
1577 operand number comes before match_scratch.
1579 2017-03-08 Richard Biener <rguenther@suse.de>
1581 PR tree-optimization/79920
1582 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
1583 with ncopies == 1 to ...
1584 (vect_transform_slp_perm_load): ... here. Properly compute
1585 all element loads by iterating VF times over the group. Do
1586 not handle ncopies (computed in a broken way) in
1587 vect_create_mask_and_perm.
1589 2017-03-08 Jakub Jelinek <jakub@redhat.com>
1592 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
1593 is a uniform vector, use uniform_vector_p return value instead of
1594 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
1596 2017-03-07 Marek Polacek <polacek@redhat.com>
1599 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
1600 (alloca_call_type): Likewise.
1602 2017-03-07 Martin Liska <mliska@suse.cz>
1604 * gcov.c (process_args): Put comment to correct location.
1606 2017-03-07 Martin Liska <mliska@suse.cz>
1609 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
1610 Use array_at_struct_end_p instead of DECL_CHAIN (field).
1611 (chkp_narrow_bounds_for_field): Likewise.
1612 (chkp_parse_array_and_component_ref): Pass one more argument to
1615 2017-03-07 Richard Biener <rguenther@suse.de>
1617 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
1620 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
1622 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
1623 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
1625 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1628 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
1629 to end of description.
1630 (PARAM_MAX_STORES_TO_MERGE): Likewise.
1632 2017-03-07 Jakub Jelinek <jakub@redhat.com>
1634 PR rtl-optimization/79901
1635 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
1637 (*avx512f_<code><mode>3<mask_name>): ... this.
1638 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
1639 iterator instead of VI8_AVX2_AVX512BW.
1641 PR rtl-optimization/79901
1642 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
1643 min/max expander, expand it using expand_vec_cond_expr.
1646 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
1649 2017-03-06 Jakub Jelinek <jakub@redhat.com>
1652 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
1653 to void * for PCH reasons.
1654 * dwarf2out.c (output_loc_operands, output_die): Cast
1655 v.val_vec.array to unsigned char *.
1657 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
1660 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
1663 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
1665 PR rtl-optimization/79571
1666 * lra-constraints.c (process_alt_operands): Calculate static
1667 reject and subtract it from overall when only addresses will be
1670 2017-03-06 Julia Koval <julia.koval@intel.com>
1673 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
1674 incoming stack boundary to 128 for 64-bit targets.
1676 2017-03-06 Richard Biener <rguenther@suse.de>
1678 PR tree-optimization/79894
1679 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
1680 to NULL after folding it.
1682 2017-03-06 Richard Biener <rguenther@suse.de>
1684 PR tree-optimization/79824
1685 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
1686 check disabling peeling for gaps.
1688 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
1690 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
1691 attributes): Document gettimeofday.
1693 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
1695 * config/s390/s390.c (s390_option_override_internal): Set
1696 PARAM_MIN_VECT_LOOP_BOUND
1698 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
1700 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
1701 * config/s390/s390.md: Likewise.
1703 2017-03-06 Jakub Jelinek <jakub@redhat.com>
1706 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
1707 (<avx2_avx512>_perm<mode>): Rename to ...
1708 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
1710 (<avx512>_perm<mode>_mask): Rename to ...
1711 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
1713 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
1714 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
1715 instead of VI8F_256_512.
1716 (avx512f_perm<mode>): New define_expand.
1717 (avx512f_perm<mode>_mask): Likewise.
1718 (avx512f_perm<mode>_1<mask_name>): New define_insn.
1719 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
1721 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
1723 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
1724 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
1726 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
1728 2017-03-06 Martin Liska <mliska@suse.cz>
1731 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
1732 when having a SSA NAME w/o VAR_DECL assigned to it.
1734 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
1736 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
1737 msa_dpsub_<su>_d): Fix MODE for vec_select.
1739 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
1741 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
1743 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
1745 2017-03-06 Richard Biener <rguenther@suse.de>
1747 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
1748 * plugin.c (register_plugin_info): Likewise.
1749 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
1751 2017-03-05 Jakub Jelinek <jakub@redhat.com>
1753 * config/i386/sse.md (sse_storehps, sse_storelps,
1754 avx_<castmode><avxsizesuffix>_<castmode>,
1755 avx512f_<castmode><avxsizesuffix>_<castmode>,
1756 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
1757 in condition that at least one operand is not a MEM.
1759 2017-03-03 Jakub Jelinek <jakub@redhat.com>
1762 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
1763 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
1765 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
1766 gimple_call_nothrow_p flag based on whether original builtin can throw.
1767 If it can, emit following stmts on the fallthrough edge.
1768 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
1769 don't create new bb if inserting just debug stmts on the edge, try to
1770 insert them on the fallthru bb or just reset debug stmts.
1772 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
1775 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
1776 restore recog_data (including the operand rtxes inside it) around
1777 the call to get_insn_template.
1779 2017-03-03 Martin Sebor <msebor@redhat.com>
1781 PR tree-optimization/79699
1782 * context.c (context::~context): Free MPFR caches to avoid
1783 a memory leak on program exit.
1785 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1787 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
1788 Use wide_int::ulow () instead of .elt (0).
1790 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
1792 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
1793 (*pushxf): Limit oF constraint to 32bit targets and add oC
1794 constraint for 64bit targets.
1795 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
1796 (*pushdf): Change rmF constraint to rmC.
1798 2017-03-03 Martin Liska <mliska@suse.cz>
1800 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
1801 Remove unused variable.
1803 2017-03-03 Jakub Jelinek <jakub@redhat.com>
1806 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
1807 is a memory operand, increase num_memory.
1808 (ix86_expand_args_builtin): Likewise.
1810 2017-03-03 Jan Hubicka <jh@suse.cz>
1813 * ipa-devirt.c (maybe_record_node): Properly handle
1814 __cxa_pure_virtual visibility.
1816 2017-03-03 Martin Liska <mliska@suse.cz>
1818 PR tree-optimization/79803
1819 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
1821 (pass_loop_prefetch::execute): Disabled optimization if an
1822 assumption about L1 cache size is not met.
1824 2017-03-03 Martin Liska <mliska@suse.cz>
1826 PR rtl-optimization/79574
1827 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
1828 (hash_scan_set): Likewise.
1829 (dump_hash_table): Likewise.
1830 (hoist_code): Likewise.
1832 2017-03-03 Richard Biener <rguenther@suse.de>
1834 * fixed-value.c (fixed_from_string): Restore use of elt (1)
1835 in place of uhigh ().
1836 (fixed_convert_from_real): Likewise.
1838 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
1841 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
1843 2017-03-03 Richard Biener <rguenther@suse.de>
1846 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
1847 TYPE_OVERFLOW_UNDEFINED check.
1849 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1851 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
1853 (vector_ae_<mode>_p): Likewise.
1854 (vector_nez_<mode>_p): Likewise.
1855 (vector_ne_v2di_p): Likewise.
1856 (vector_ae_v2di_p): Likewise.
1857 (vector_ne_<mode>_p): Likewise.
1858 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
1860 (vsx_tsqrt<mode>2_fe): Likewise.
1862 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
1865 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
1867 2017-03-02 Jakub Jelinek <jakub@redhat.com>
1869 PR rtl-optimization/79780
1870 * cprop.c (one_cprop_pass): When second and further conditional trap
1871 in a single basic block is turned into an unconditional trap, turn it
1872 into a deleted note to avoid RTL verification failures.
1874 2017-03-02 Richard Biener <rguenther@suse.de>
1876 * fold-const.c (const_binop): Use ulow () instead of elt (0).
1878 2017-03-02 Richard Biener <rguenther@suse.de>
1880 PR tree-optimization/79345
1882 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
1883 param and abort the walk, returning -1 if it is hit.
1884 (walk_aliased_vdefs): Take a limit param and pass it on.
1885 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
1886 defaulting to 0 and return a signed int.
1887 * tree-ssa-uninit.c (struct check_defs_data): New struct.
1888 (check_defs): New helper.
1889 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
1890 about uninitialized memory.
1891 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
1892 bogus uninitialized warning.
1893 (fixed_convert_from_real): Likewise.
1895 2017-03-02 Bin Cheng <bin.cheng@arm.com>
1897 PR tree-optimization/66768
1898 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
1899 iv_use if base object can't be determined.
1901 2017-03-02 Jakub Jelinek <jakub@redhat.com>
1903 PR tree-optimization/79345
1904 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
1905 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
1906 (get_pattern_stats): Initialize it.
1907 * genemit.c (gen_expand): Verify match_scratch numbers come after
1908 match_operand/match_dup numbers.
1909 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
1910 match_scratch numbers.
1911 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
1913 * config/s390/s390.md (trunctdsd2): Likewise.
1915 2017-03-02 Richard Biener <rguenther@suse.de>
1917 * wide-int.h (wide_int_storage::operator=): Implement in terms
1920 2017-03-02 Richard Biener <rguenther@suse.de>
1922 PR tree-optimization/79777
1923 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
1924 the to insert expression to sth existing.
1926 2017-03-01 Martin Sebor <msebor@redhat.com>
1929 * gimple-ssa-sprintf.c
1930 (directive::known_width_and_precision): New function.
1931 (format_integer): Use it.
1932 (get_mpfr_format_length): Consider the full range of precision
1933 when computing %g output with the # flag. Set the likely byte
1934 count to 3 rather than 1 when precision is indeterminate.
1935 (format_floating): Correct the lower bound of precision.
1937 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1939 * doc/invoke.texi: Document default code model for 64-bit Linux.
1941 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
1944 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
1945 udiv rather than div since input pattern is unsigned.
1947 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
1949 * config/i386/i386.c (print_reg): Warn for values of
1950 unsupported size in integer register.
1952 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
1955 * config/rs6000/predicates.md (current_file_function_operand): Do
1956 not allow self calls to be local if the function is replaceable.
1958 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
1961 * config/rs6000/altivec.h (vec_ctz and others): Change the
1962 preprocessor macro that controls conditional compilation from
1963 _ARCH_PWR9 to __POWER9_VECTOR__.
1964 (vec_all_ne): Change parameterization of __altivec_scalar_pred
1965 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
1966 control (instead of _ARCH_PWR9 control) so that template
1967 definition uses power9-specific function.
1968 (vec_any_eq): Likewise.
1969 (vec_all_ne): Change macro definition to use a power9-specific
1970 expansion under #ifdef __POWER9_VECTOR__ control (instead of
1971 _ARCH_PWR9 control).
1972 (vec_any_eq) Likewise.
1973 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
1974 expansion for CMPNEF to remove support for xvcmpnesp instruction.
1975 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
1976 support for xvcmpnedp instruction.
1977 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
1978 macro expansion so that Power9 implementation of vec_all_ne does
1979 not use the AltiVec predicate framework.
1980 (VCMPNEH_P): Likewise.
1981 (VCMPNEW_P): Likewise.
1982 (VCMPNED_P): Likewise.
1983 (VCMPNEFP_P): Likewise.
1984 (VCMPNEDP_P): Likewise.
1985 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
1986 implementation of vec_any_eq to not use AltiVec predicate
1988 (VCMPAEH_P): Likewise.
1989 (VCMPAEW_P): Likewise.
1990 (VCMPAED_P): Likewise.
1991 (VCMPAEFP_P): Likewise.
1992 (VCMPAEDP_P): Likewise.
1993 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
1994 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
1995 not use the AltiVec predicate framework.
1996 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
1997 of vec_any_eq to not use AltiVec predicate framework.
1998 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
1999 support for predefined __POWER9_VECTOR__ macro to indicate that
2000 Power9 instruction selection is enabled.
2001 (altivec_overloaded_builtins): Remove extraneous
2002 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
2003 function argument types RS6000_BTI_bool_V16QI and
2004 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
2005 entry for overloaded function argument types RS6000_BTI_bool_V4SI
2006 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
2007 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
2008 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
2009 Power9 for implementations of vec_cmpne. Change the signature for
2010 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
2011 (representing vec_all_ne) to remove the previously described first
2012 argument of type RS6000_BTI_INTSI, as this was an artifact of
2013 reliance on the AltiVec predicate framework, which is no longer
2014 used in the implementation of these functions. Add
2015 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
2016 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
2017 since, unlike the AltiVec predicate framework implementation, we
2018 do not share function descriptors between vec_alle and vec_anyeq.
2019 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
2020 set of modes that receive special treatment even when
2021 TARGET_P9_VECTOR is true. The special treatment emits code that
2022 does not depend on Power9 instructions.
2023 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
2024 define_expand to not rely on AltiVec predicate framework.
2025 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
2027 (vector_ne_v2di_p): Change this define_expand to not rely on
2028 AltiVec predicate framework.
2029 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
2031 (vector_ne_<mode>_p): Change this define_expand to not rely on
2032 AltiVec predicate framework.
2033 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
2035 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
2036 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
2037 define_insn pattern.
2038 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
2039 define_insn pattern because the xvcmpne<VSs>. instruction is not
2041 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
2042 instruction is not supported.
2044 2017-03-01 Jakub Jelinek <jakub@redhat.com>
2046 * config/nvptx/nvptx.c: Include intl.h.
2048 2017-03-01 Martin Jambor <mjambor@suse.cz>
2051 * ipa-prop.h (ipa_bits): Removed field known.
2052 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
2053 to pointers. Adjusted their comments to warn about their sharing.
2054 (ipcp_transformation_summary): Change bits to a vector of pointers.
2055 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
2056 (ipa_get_ipa_bits_for_value): Declare.
2057 * tree-vrp.h (value_range): Mark as GTY((for_user)).
2058 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
2059 (ipa_bits_hash_table): Likewise.
2060 (ipa_vr_ggc_hash_traits): Likewise.
2061 (ipa_vr_hash_table): Likewise.
2062 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
2063 being pointers and vr_known being removed.
2064 (ipa_set_jf_unknown): Likewise.
2065 (ipa_get_ipa_bits_for_value): New function.
2066 (ipa_set_jfunc_bits): Likewise.
2067 (ipa_get_value_range): New overloaded functions.
2068 (ipa_set_jfunc_vr): Likewise.
2069 (ipa_compute_jump_functions_for_edge): Use the above functions to
2070 construct bits and vr parts of jump functions.
2071 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
2072 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
2074 (ipcp_grow_transformations_if_necessary): Also allocate
2075 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
2077 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
2078 them. Fix too long lines.
2079 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
2080 vr_known being removed.
2081 (ipa_read_jump_function): Use new setter functions to construct bits
2082 and vr parts of jump functions or set them to NULL.
2083 (write_ipcp_transformation_info): Adjust for bits being pointers.
2084 (read_ipcp_transformation_info): Likewise.
2085 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
2087 Include gt-ipa-prop.h.
2088 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
2090 (ipcp_store_bits_results): Likewise.
2091 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
2092 Do not write to existing jump functions but use a temporary instead.
2094 2017-03-01 Jakub Jelinek <jakub@redhat.com>
2097 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
2098 attempt to use its first operand as BIT_FIELD_REF base.
2100 2017-03-01 Richard Biener <rguenther@suse.de>
2103 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
2104 interpolating formula in wrapping arithmetic.
2105 (chrec_apply): Convert chrec_evaluate return value to wanted type.
2107 2017-03-01 Jakub Jelinek <jakub@redhat.com>
2109 PR tree-optimization/79734
2110 * tree-vect-generic.c (expand_vector_condition): Optimize
2111 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
2112 Handle VEC_COND_EXPR where comparison has different inner width from
2115 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
2117 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
2118 markup, and similar issues. Remove @opindex entries for things
2119 that aren't options. Add missing -mmpy-option entries.
2121 2017-02-28 Jakub Jelinek <jakub@redhat.com>
2123 PR tree-optimization/79737
2124 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
2125 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
2126 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
2127 instead of byte_size. Formatting fix.
2128 (shift_bytes_in_array_right): Formatting fix.
2130 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
2133 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
2134 condition on optimize for the leaf function test.
2136 2017-02-28 Martin Liska <mliska@suse.cz>
2139 * read-rtl-function.c (function_reader::handle_unknown_directive):
2140 Bail out when one uses -flto.
2142 2017-02-28 Martin Liska <mliska@suse.cz>
2144 * common.opt: Replace space with tabular for options of <number>
2146 * config/i386/i386.opt: Show <number> value for
2147 -mlarge-data-threshold.
2148 * opts.c (print_filtered_help): Do not display number in hexadecimal
2151 2017-02-28 Martin Liska <mliska@suse.cz>
2153 * common.opt: Fix --help=option -Q for options which are of
2156 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
2158 * config/i386/i386.c (print_reg): Error out for values
2159 of 8-bit size in invalid integer register.
2161 2017-02-28 Martin Sebor <msebor@redhat.com>
2163 PR tree-optimization/79691
2164 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
2166 2017-02-28 Jakub Jelinek <jakub@redhat.com>
2169 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
2170 gcc_unreachable with output_operand_lossage.
2172 2017-02-28 Richard Biener <rguenther@suse.de>
2174 PR tree-optimization/79740
2175 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
2177 (visit_nary_op): Insert the nary into the hashtable if we
2178 pattern-matched sth.
2179 * tree-ssa-pre.c (eliminate_insert): Robustify.
2181 2017-02-28 Richard Biener <rguenther@suse.de>
2184 * fold-const.c (decode_field_reference): Reject out-of-bound
2187 2017-02-28 Jakub Jelinek <jakub@redhat.com>
2189 * config/i386/i386.c: Include intl.h.
2190 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
2191 instead of just cond ? "..." : "...".
2192 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
2193 * coverage.c (read_counts_file): Likewise.
2194 * omp-offload.c: Include intl.h.
2195 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
2196 of just cond ? "..." : "...".
2197 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
2198 of just cond ? "..." : "...".
2200 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
2203 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
2205 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
2206 'tune for' CPU name.
2207 * config/arm/arm-cpu-data.h: Regenerated.
2209 2017-02-28 Richard Biener <rguenther@suse.de>
2211 PR tree-optimization/79732
2212 * tree-inline.c (expand_call_inline): Do not shadow var.
2214 2017-02-28 Richard Biener <rguenther@suse.de>
2216 PR tree-optimization/79723
2217 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
2218 address-space properly.
2220 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
2222 * doc/optinfo.texi (Optimization groups): Fix option used for
2224 * doc/invoke.texi (-fopt-info): Document "omp".
2225 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
2226 (OPTGROUP_ALL): Add OPTGROUP_OMP.
2227 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
2228 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
2229 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
2231 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
2233 * dumpfile.c (optgroup_options): Instead of "openmp", associate
2234 OPTGROUP_OMP with "omp".
2236 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
2239 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
2240 for arithmetic shift of unsigned V2DI.
2242 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
2244 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
2245 arc/linux.h headers.
2246 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
2247 (LINK_SPEC): Likewise.
2248 (ARC_TLS_EXTRA_START_SPEC): Likewise.
2249 (EXTRA_SPECS): Likewise.
2250 (STARTFILE_SPEC): Likewise.
2251 (ENDFILE_SPEC): Likewise.
2252 (LIB_SPEC): Likewise.
2253 (TARGET_SDATA_DEFAULT): Likewise.
2254 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
2255 (MULTILIB_DEFAULTS): Likewise.
2256 (DWARF2_UNWIND_INFO): Likewise.
2257 * config/arc/big.h: New file.
2258 * config/arc/elf.h: Likewise.
2259 * config/arc/linux.h: Likewise.
2260 * config/arc/t-uClibc: Remove.
2262 2017-02-27 Bin Cheng <bin.cheng@arm.com>
2264 PR tree-optimization/77536
2265 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
2266 (tree_transform_and_unroll_loop): Use above function to compute the
2267 estimated niter of unrolled loop and use it when scaling profile.
2268 Also use count info rather than frequency if it's non-zero.
2269 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
2270 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
2271 (vect_transform_loop): Call above function.
2273 2017-02-27 Richard Biener <rguenther@suse.de>
2275 PR tree-optimization/45397
2276 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
2277 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
2278 (visit_nary_op): Add pattern matching for CSEing sign-changed
2279 or truncated operations with wider ones.
2281 2017-02-27 Richard Biener <rguenther@suse.de>
2283 PR tree-optimization/79690
2284 * tree-vect-stmts.c (vectorizable_store): Use vector type
2285 built from the DR with address-space.
2287 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
2289 * doc/invoke.texi (Optimize Options): Refine the description
2290 of asan-use-after-return.
2292 2017-02-25 Alan Modra <amodra@gmail.com>
2294 PR rtl-optimization/79584
2295 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
2296 base, not ad->base_term, the reg within base. Remove assertion
2297 that ad->base == ad->base_term. Replace gen_int_mode using
2298 bogus mode with const0_rtx.
2300 2017-02-25 Jakub Jelinek <jakub@redhat.com>
2303 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
2304 FMA_EXPR like tcc_binary or tcc_unary.
2306 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
2309 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
2311 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
2312 (output_loc_operands): Handle DW_OP_call_ref and
2313 DW_OP_GNU_variable_value.
2314 (struct variable_value_struct): New type.
2315 (struct variable_value_hasher): Likewise.
2316 (variable_value_hash): New variable.
2317 (string_types): Remove.
2318 (copy_loc_descr): New function.
2319 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
2320 (prepend_loc_descr_to_each): New function.
2321 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
2322 instead of add_loc_descr_to_each if the first argument is single
2323 location list and the second has multiple.
2324 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
2325 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
2326 when looking for variable value which doesn't have other location info.
2327 (loc_list_from_tree): Formatting fix.
2328 (gen_array_type_die): Simplify DW_AT_string_length handling.
2329 (adjust_string_types): Remove.
2330 (gen_subprogram_die): Don't call adjust_string_types nor test/set
2331 string_types. Call resolve_variable_values.
2332 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
2333 (resolve_addr_in_expr): Likewise. Add A argument.
2334 (copy_deref_exprloc): Remove deref argument. Adjust for the
2335 original expression being DW_OP_GNU_variable_value with optionally
2336 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
2337 optionally after it.
2338 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
2339 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
2340 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
2341 (variable_value_hasher::hash, variable_value_hasher::equal): New
2343 (resolve_variable_value_in_expr, resolve_variable_value,
2344 resolve_variable_values, note_variable_value_in_expr,
2345 note_variable_value): New functions.
2346 (dwarf2out_early_finish): Call note_variable_value on all toplevel
2349 2017-02-24 Jakub Jelinek <jakub@redhat.com>
2352 * opts.h (handle_generated_option): Add GENERATED_P argument.
2353 * opts-common.c (handle_option): Adjust function comment.
2354 (handle_generated_option): Add GENERATED_P argument, pass it to
2356 (control_warning_option): Pass false to handle_generated_option
2358 * opts.c (maybe_default_option): Pass true to handle_generated_option
2360 * optc-gen.awk: Likewise.
2362 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
2364 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
2365 a REG, look at the REG it is a SUBREG of.
2366 (splitter for cmpeqsi_t): Ditto.
2368 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
2370 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
2371 the special USEs with the pattern of the insn, not the insn itself.
2373 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
2376 * doc/invoke.texi: Document -mload-store-pairs.
2378 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
2379 Sandra Loosemore <sandra@codesourcery.com>
2381 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
2382 argument isn't a CONST_INT.
2383 (nios2_alternate_compare_const): Assert op is a CONST_INT.
2384 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
2385 (nios2_validate_compare): Bypass alternate compare logic if *op2
2387 (ldstwm_operation_p): Return false if first_base is not a REG or
2388 if first_offset is not a CONST_INT.
2390 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
2392 * config/cris/cris.md: Use correct operand in a define_peephole2.
2394 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
2396 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
2398 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
2400 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
2401 this_insn if it is an INSN or JUMP_INSN.
2402 (force_offsettable): Look at base, not at addr.
2403 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
2404 on things that aren't necessarily CONST_INTs.
2406 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
2408 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
2409 -mfpmath=sse is the default also for x86-32 targets with SSE2
2410 instruction set when @option{-ffast-math} is enabled
2412 2017-02-24 Jeff Law <law@redhat.com>
2414 PR rtl-optimizatoin/79286
2415 * ira.c (update_equiv_regs): Drop may_trap_p exception to
2418 2017-02-24 Richard Biener <rguenther@suse.de>
2420 PR tree-optimization/79389
2421 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
2424 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
2426 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
2427 function comment to reflect reality.
2428 (loop_exits_before_overflow): Fix typo in function description.
2430 2017-02-24 Richard Biener <rguenther@suse.de>
2432 PR tree-optimization/79389
2433 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
2434 properly that a threading opportunity exists. Detect conditional
2435 copy/constant propagation opportunities.
2437 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
2439 * config/visium/visium.md (type): Add trap.
2440 (b): New mode attribute.
2441 (*btst): Rename into...
2442 (*btst<mode>): ...this and adjust.
2443 (*cbranchsi4_btst_insn): Rename into...
2444 (*cbranch<mode>4_btst_insn): ...this and adjust.
2445 (trap): New define_insn.
2447 2017-02-23 Jakub Jelinek <jakub@redhat.com>
2449 PR tree-optimization/79389
2450 * ifcvt.c (struct noce_if_info): Add rev_cond field.
2451 (noce_reversed_cond_code): New function.
2452 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
2453 reversed_comparison_code. Formatting fix.
2454 (noce_try_store_flag): Test rev_cond != NULL in addition to
2455 reversed_comparison_code.
2456 (noce_try_store_flag_constants): Likewise.
2457 (noce_try_store_flag_mask): Likewise.
2458 (noce_try_addcc): Use rev_cond if non-NULL instead of
2459 reversed_comparison_code.
2460 (noce_try_cmove_arith): Likewise. Formatting fixes.
2461 (noce_try_minmax, noce_try_abs): Clear rev_cond.
2462 (noce_find_if_block): Initialize rev_cond.
2463 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
2464 instead of false as last argument never attempt to reverse it
2467 2017-02-23 Bin Cheng <bin.cheng@arm.com>
2469 PR tree-optimization/79663
2470 * tree-predcom.c (combine_chains): Process refs in reverse order
2471 only for ZERO length chains, and add explaining comment.
2473 2017-02-23 Jeff Law <law@redhat.com>
2475 PR tree-optimization/79578
2476 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
2477 in call to operand_equal_p.
2479 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
2482 * config/i386/cpuid.h: Fix another undefined behavior.
2484 2017-02-23 Richard Biener <rguenther@suse.de>
2486 PR tree-optimization/79683
2487 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
2488 vector types for data-refs.
2490 2017-02-23 Martin Liska <mliska@suse.cz>
2492 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
2494 2017-02-23 Jakub Jelinek <jakub@redhat.com>
2497 * internal-fn.c (get_range_pos_neg): Moved to ...
2498 * tree.c (get_range_pos_neg): ... here. No longer static.
2499 * tree.h (get_range_pos_neg): New prototype.
2500 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
2501 are known to be in between 0 and signed maximum inclusive, try to
2502 expand both unsigned and signed divmod and use the cheaper one from
2505 2017-02-22 Jeff Law <law@redhat.com>
2507 PR tree-optimization/79578
2508 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
2509 to compare base operands.
2511 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
2514 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
2515 gpc_reg_operand instead of fpr_reg_operand.
2517 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
2519 * config/mips/mips.c (mips_return_in_memory): Force FP
2520 vector types to be returned in memory for o32 ABI.
2522 2017-02-22 Jakub Jelinek <jakub@redhat.com>
2524 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
2525 instead of DW_TAG_member for static data member declarations and don't
2526 set no_linkage_name for static inline data members.
2527 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
2530 2017-02-22 Martin Liska <mliska@suse.cz>
2532 * doc/invoke.texi: Replace inequality signs with square brackets
2535 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2537 PR tree-optimization/68644
2538 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
2540 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
2543 * lra-constraints.c (simplify_operand_subreg): Handle
2544 WORD_REGISTER_OPERATIONS targets.
2546 2017-02-22 Jakub Jelinek <jakub@redhat.com>
2549 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
2550 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
2551 elimination by swapping fld*.
2553 2017-02-22 Richard Biener <rguenther@suse.de>
2555 PR tree-optimization/79673
2556 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
2557 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
2558 irrelevant address-space qualifiers and avoiding a
2559 ADDR_SPACE_CONVERT_EXPR from fold_convert.
2561 2017-02-22 Richard Biener <rguenther@suse.de>
2563 PR tree-optimization/79666
2564 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
2565 to not symbolically negate if that may introduce undefined
2568 2017-02-22 Martin Liska <mliska@suse.cz>
2571 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
2572 * data-streamer-out.c (streamer_write_gcov_count_stream):
2574 * value-prof.c (stream_out_histogram_value): Make assert more
2575 precise based on type of counter.
2577 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
2580 * config/i386/i386.md (standard_x87sse_constant_load splitter):
2581 Use nonimmediate_operand instead of memory_operand for operand 1.
2582 (float-extend standard_x87sse_constant_load splitter): Ditto.
2584 2017-02-21 Jeff Law <law@redhat.com>
2586 PR tree-optimization/79621
2587 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
2588 blocks with edges to themselves.
2590 2017-02-21 Jakub Jelinek <jakub@redhat.com>
2593 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
2594 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
2595 Use gimple_call_builtin_p.
2598 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
2599 on temporarily removed DEBUG_INSNs.
2601 PR tree-optimization/79649
2602 * tree-loop-distribution.c (classify_partition): Give up on
2603 non-generic address space loads/stores.
2605 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
2607 * doc/loop.texi (Loop manipulation): Remove nonexistent
2608 tree_ssa_loop_version from the documentation.
2609 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
2611 2017-02-21 Jakub Jelinek <jakub@redhat.com>
2614 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
2615 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
2616 * config/rs6000/rs6000.c: Include except.h.
2617 (rs6000_expand_split_stack_prologue): Call
2618 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
2620 2017-02-21 Martin Jambor <mjambor@suse.cz>
2623 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
2626 2017-02-21 Martin Jambor <mjambor@suse.cz>
2628 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
2629 for backward compatibility only.
2630 * doc/invoke.texi (Option Summary): Remove all references to
2633 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
2637 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
2639 * lra-constraints.c (curr_insn_transform): Handle
2640 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
2642 2017-02-21 Martin Liska <mliska@suse.cz>
2644 * config/i386/i386.opt: Replace -masm-dialect with -masm.
2646 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
2648 PR translation/79638
2649 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
2651 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
2654 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
2655 (arm_function_ok_for_sibcall): Return false for an indirect call by
2656 descriptor if all the argument registers are used.
2657 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
2658 alignment of the function.
2660 2017-02-21 Jakub Jelinek <jakub@redhat.com>
2662 PR tree-optimization/61441
2663 * simplify-rtx.c (simplify_const_unary_operation): For
2664 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
2665 the sNaN unmodified.
2667 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
2669 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
2670 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
2671 instead of SYSTEM_HEADER_DIR.
2673 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
2674 Martin Liška <mliska@suse.cz>
2676 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
2677 Fix typos and grammar, use active voice, and clarify.
2679 2017-02-20 Marek Polacek <polacek@redhat.com>
2682 * gimplify.c (gimplify_expr): Handle unused *&&L;.
2685 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
2687 2017-02-20 Jakub Jelinek <jakub@redhat.com>
2690 * config/i386/i386.c (ix86_expand_builtin): Handle
2691 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
2692 ix86_builtins_isa[fcode].isa as a requirement of those
2693 flags and any other flag in the bitmask.
2694 (ix86_init_mmx_sse_builtins): Use 0 instead of
2695 ~OPTION_MASK_ISA_64BIT as mask.
2696 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
2697 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
2698 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
2699 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
2701 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
2704 * lra-constraints.c (split_reg): Check requested split mode
2705 is supported by the register.
2707 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
2709 * lra-constraints.c (simplify_operand_subreg): Remove early
2712 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
2715 * lra-constraints.c (curr_insn_transform): Tighten condition
2716 for converting SUBREG reloads from OP_OUT to OP_INOUT.
2718 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
2721 * lra-constraints.c (curr_insn_transform): Handle
2722 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
2724 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
2727 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
2729 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
2731 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
2734 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
2737 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2739 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
2742 2017-02-18 Jakub Jelinek <jakub@redhat.com>
2744 * final.c (last_columnnum, override_columnnum): New variables.
2745 (final_start_function): Set last_columnnum, pass it to begin_prologue
2746 hook and pass 0 to dwarf2out_begin_prologue.
2747 (final_scan_insn): Update override_columnnum. Pass last_columnnum
2748 to source_line debug hook.
2749 (notice_source_line): Compute last_columnnum and for debug_column_info
2750 return true on column changes.
2751 * debug.h (struct gcc_debug_hooks): Add column argument to
2752 source_line and begin_prologue hooks.
2753 (debug_nothing_int_charstar_int_bool): Remove prototype.
2754 (debug_nothing_int_int_charstar,
2755 debug_nothing_int_int_charstar_int_bool): New prototypes.
2756 (dwarf2out_begin_prologue): Add column argument.
2757 * debug.c (do_nothing_debug_hooks): Adjust source_line and
2758 begin_prologue hooks.
2759 (debug_nothing_int_charstar_int_bool): Remove.
2760 (debug_nothing_int_int_charstar,
2761 debug_nothing_int_int_charstar_int_bool): New functions.
2762 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
2763 through to dwarf2out_source_line.
2764 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
2765 (dwarf2out_source_line): Add column argument, emit it if requested.
2766 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
2768 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
2769 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
2770 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
2771 through to dwarf2out_begin_prologue.
2772 (vmsdbgout_source_line): Add column argument, pass it through to
2773 dwarf2out_source_line.
2774 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
2775 dbxout_source_line caller.
2776 (dbxout_source_line): Add column argument.
2778 * common.opt (gno-column-info, gcolumn-info): New options.
2779 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
2780 (check_die): Also test for multiple DW_AT_decl_column attributes.
2781 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
2782 DW_AT_decl_column if requested.
2783 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
2785 (gen_variable_die): Likewise.
2786 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
2787 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
2790 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
2791 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
2792 (ix86_handle_option): Handle OPT_m3dnowa.
2793 * doc/invoke.texi (-m3dnowa): Document.
2794 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
2795 -m3dnowa instead of -m3dnow -march=athlon.
2798 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
2799 instead of gcc_assert for K, r and R code checks. Formatting fixes.
2801 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2804 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
2805 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
2806 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
2807 generator for vsx_xxpermdi_<mode>_be.
2808 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
2809 force big-endian semantics.
2810 (vsx_xxpermdi_<mode>_be): New define_expand with same
2811 implementation as previous version of vsx_xxpermdi_<mode>.
2813 2017-02-17 Jakub Jelinek <jakub@redhat.com>
2815 PR tree-optimization/79327
2816 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
2817 variable, its initialization and use.
2819 2017-02-17 Julia Koval <julia.koval@intel.com>
2821 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
2822 (OPTION_MASK_ISA_PKU_UNSET): New.
2823 (ix86_handle_option): Handle -mrdpid.
2824 * config/i386/cpuid.h (bit_RDPID): New.
2825 * config/i386/driver-i386.c (host_detect_local_cpu):
2826 Detect RDPID feature.
2827 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
2828 * config/i386/i386-c.c (ix86_target_macros_internal):
2830 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
2831 (ix86_valid_target_attribute_inner_p): Add "rdpid".
2832 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
2833 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
2834 * config/i386/i386.md (define_insn "rdpid"): New.
2835 * config/i386/i386.opt Add -mrdpid.
2836 * config/i386/immintrin.h (_rdpid_u32): New.
2838 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
2840 PR rtl-optimization/79541
2841 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
2842 instead of transforming it into USE.
2844 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
2846 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
2847 If HONOR_SNANS (SFmode) force the input to a register.
2848 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
2849 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
2850 an frsp or similar insn.
2852 2017-02-17 Martin Liska <mliska@suse.cz>
2854 PR rtl-optimization/79577
2855 * params.def (selsched-max-sched-times): Increase minimum to 1.
2857 2017-02-17 Martin Liska <mliska@suse.cz>
2859 PR rtl-optimization/79574
2860 * gcse.c (want_to_gcse_p): Prevent integer overflow.
2862 2017-02-17 Martin Liska <mliska@suse.cz>
2864 PR tree-optimization/79529
2865 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
2866 ssa_defined_default_def_p to handle cases which are implicitly
2868 * tree-ssa.c (ssa_defined_default_def_p): New function.
2869 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
2870 which are implicitly defined.
2871 * tree-ssa.h (ssa_defined_default_def_p): Declare.
2873 2017-02-17 Richard Biener <rguenther@suse.de>
2876 * params.def (max-ssa-name-query-depth): Limit to 10.
2878 2017-02-17 Richard Biener <rguenther@suse.de>
2880 PR tree-optimization/79552
2881 * tree-ssa-structalias.c (visit_loadstore): Properly verify
2884 2017-02-17 Richard Biener <rguenther@suse.de>
2887 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
2889 2017-02-17 Marek Polacek <polacek@redhat.com>
2892 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
2893 (fold_negate_expr): New wrapper.
2895 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
2897 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
2898 Correct terminology and de-emphasize pre-standard behavior.
2900 2017-02-16 Alan Modra <amodra@gmail.com>
2902 PR rtl-optimization/79286
2903 * ira.c (def_dominates_uses): New function.
2904 (update_equiv_regs): Don't create an equivalence for insns that
2905 may trap where the register def does not dominate the use.
2907 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
2909 PR rtl-optimization/78127
2910 * lra.c (lra): Call lra_eliminate before finish the loop after
2913 2017-02-16 Richard Biener <rguenther@suse.de>
2915 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
2917 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
2918 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
2919 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
2920 (isl_val_int_from_wi): New function.
2921 (extract_affine_gmp): Rename to ...
2922 (extract_affine_wi): ... this, take a widest_int.
2923 (extract_affine_int): Just wrap extract_affine_wi.
2924 (add_param_constraints): Use isl_val_int_from_wi.
2925 (add_loop_constraints): Likewise, and extract_affine_wi.
2927 2017-02-15 Jeff Law <law@redhat.com>
2930 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
2931 ira_init_register_move_cost_if_necessary.
2933 2017-02-15 Martin Sebor <msebor@redhat.com>
2936 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
2937 removed in a prior commit.
2939 2017-02-15 Bin Cheng <bin.cheng@arm.com>
2941 PR tree-optimization/79347
2942 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
2943 counters during peeling.
2945 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
2947 * Makefile.in (site.exp): Remove "set ISLVER".
2949 2017-02-15 Jakub Jelinek <jakub@redhat.com>
2952 * real.c (real_from_integer): Call real_convert even for decimal.
2954 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
2957 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
2959 2017-02-14 Andrew Pinski <apinski@cavium.com>
2961 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
2962 cores and change the partno/implementer to be correct.
2963 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
2964 the 'B" as the implementer.
2965 * config/aarch64/aarch64-tune.md: Regenerate.
2967 2017-02-14 Carl Love <cel@us.ibm.com>
2969 * config/rs6000/rs6000.c: Add case statement entry to make the
2970 xvcvuxdsp built-in argument unsigned.
2971 * config/rs6000/vsx.md: Fix the source and return operand types so they
2972 match the instruction definitions from the ISA document. Fix typo
2973 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
2976 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
2979 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
2980 member early_clobber_alts.
2981 * lra-lives.c (reg_early_clobber_p): New.
2982 (process_bb_lives): Use it.
2983 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
2984 (debug_operand_data): Initialize early_clobber_alts.
2985 (setup_operand_alternative): Set up early_clobber_alts.
2986 (collect_non_operand_hard_regs): Ditto. Pass early clobber
2987 alternatives to new_insn_reg.
2988 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
2990 (lra_update_insn_regno_info): Pass the new arg.
2992 2017-02-14 Jakub Jelinek <jakub@redhat.com>
2995 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
2996 (new_oacc_loop_raw): Don't clear already cleared fields.
2999 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
3000 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
3001 _mm512_prefetch_i64gather_ps): New inline functions and macros.
3003 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
3006 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
3008 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
3011 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
3012 the extra instruction to the right place to store 128-bit constant
3015 2017-02-14 Martin Sebor <msebor@redhat.com>
3018 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
3019 warning for strings of unknown length.
3021 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
3023 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
3025 2017-02-14 Jeff Law <law@redhat.com>
3028 * ira-costs.c (scan_one_insn): Initialize register move costs
3029 for pseudos seen in USE/CLOBBER insns.
3031 PR tree-optimization/79095
3032 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
3033 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
3034 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
3035 if the operands are known to be not equal, then the resulting range
3037 (intersect_ranges): If the new range is ~[0,0] and the old range is
3038 wide, then prefer ~[0,0].
3039 * tree-vrp.c (overflow_comparison_p_1): New function.
3040 (overflow_comparison_p): New function.
3041 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
3042 if NAME is used in an overflow test.
3043 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
3044 overflow check that can be expressed as an equality test, then adjust
3045 ops to be that equality test.
3047 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3049 * config/s390/s390-builtin-types.def: Remove flags argument.
3050 * config/s390/s390.c (s390_init_builtins): Likewise.
3052 2017-02-14 Martin Liska <mliska@suse.cz>
3054 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
3055 vector. Fix trailing white spaces.
3057 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
3059 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
3062 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3064 PR rtl-optimization/68664
3065 * config/arm/arm.c (arm_sched_can_speculate_insn):
3066 New function. Declare prototype.
3067 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
3069 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3071 PR rtl-optimization/68664
3072 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
3074 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
3076 2017-02-14 Amit Pawar <amit.pawar@amd.com>
3078 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
3079 max skip bytes for function, loop and jump.
3081 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3083 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
3084 ABS_EXPR for gimple dump.
3086 2017-02-14 Jakub Jelinek <jakub@redhat.com>
3089 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
3091 PR tree-optimization/79408
3092 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
3093 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
3094 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
3095 also if rhs1 is INTEGER_CST.
3097 2017-02-14 Richard Biener <rguenther@suse.de>
3100 * tree-into-ssa.c (insert_phi_nodes): When the function can
3101 have abnormal edges rewrite SSA names with broken use-def
3102 dominance out of SSA and register them for PHI insertion.
3104 2017-02-13 Martin Sebor <msebor@redhat.com>
3107 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
3108 clearing info.nowrite flag when snprintf size argument is a range.
3110 2017-02-13 Jakub Jelinek <jakub@redhat.com>
3112 * cprop.c (cprop_jump): Add missing space in string literal.
3113 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
3114 (get_constraint_for_component_ref): Likewise.
3115 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
3116 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
3117 * lra-constraints.c (process_alt_operands): Likewise.
3118 * ipa-inline.c (inline_small_functions): Likewise.
3119 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
3120 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
3121 * trans-mem.c (diagnose_tm_1_op): Likewise.
3122 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
3123 (grid_parallel_clauses_gridifiable): Likewise.
3125 * config/nvptx/mkoffload.c (process): Add space in between
3128 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
3129 "MOD4_SSE_REGS" and "ALL_REGS".
3131 * spellcheck.c (test_data): Add , in between "foo" and "food".
3133 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3136 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
3137 boundary crossing check and subsequent code generation agree.
3139 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3141 * config/aarch64/aarch64.c (has_memory_op): Delete.
3142 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
3145 2017-02-13 Jakub Jelinek <jakub@redhat.com>
3147 PR rtl-optimization/79388
3148 PR rtl-optimization/79450
3149 * combine.c (distribute_notes): When removing TEM_INSN for which
3150 corresponding dest has last value recorded, invalidate that last
3153 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3155 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
3156 of explicit '@'. Add missing assembly comment marker on branch costs
3159 2017-02-13 Nathan Sidwell <nathan@acm.org>
3161 * gengtype-lex.l (<in_struct>): Add '/'.
3163 2017-02-13 Martin Liska <mliska@suse.cz>
3166 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
3168 2017-02-13 Richard Biener <rguenther@suse.de>
3170 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
3172 * configure: Re-generate.
3173 * config.in: Likewise.
3174 * graphite-dependences.c: Simplify as if
3175 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
3176 * graphite-isl-ast-to-gimple.c: Likewise.
3177 * graphite-optimize-isl.c: Likewise.
3178 * graphite-poly.c: Likewise.
3179 * graphite-sese-to-poly.c: Likewise.
3180 * graphite.h: Likewise.
3181 * toplev.c: Include isl/version.h and use isl_version () for
3182 printing the ISL version.
3183 * doc/install.texi: Update ISL requirement.
3185 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
3187 * doc/standards.texi (Standards): Update reference to
3190 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
3192 * doc/extend.texi (Named Address Spaces): sourceware.org now
3194 * doc/install.texi (Binaries): Ditto.
3197 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
3199 * doc/cpp.texi: Replace "stringify"/"stringification" with C
3200 standard terminology "stringize"/"stringizing" throughout.
3201 * doc/cppinternals.texi: Likewise.
3203 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
3205 * doc/extend.texi: Fix some spelling mistakes and typos.
3206 * doc/invoke.texi: Likewise.
3208 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
3211 * params.def (inline-min-speedup) Change from 10 to 8.
3213 2017-02-11 Jakub Jelinek <jakub@redhat.com>
3215 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
3218 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
3221 * ipa-inline-analysis.c (get_minimal_bb): New function.
3222 (record_modified): Use it.
3223 (remap_edge_change_prob): Handle also ancestor functions.
3225 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
3227 * doc/contrib.texi (Contributors): Remove broken link into
3228 the Mauve CVS repository.
3230 2017-02-11 Jakub Jelinek <jakub@redhat.com>
3233 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
3234 result computation whenever lhs doesn't have vector mode, not
3235 just when it has BLKmode.
3237 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
3239 * doc/makefile.texi (profiledbootstrap): Refer to the
3240 installation instructions only in textual form.
3242 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3245 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
3247 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
3249 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
3250 (Specific): Update mingw-w64 reference.
3252 (Specific): Remove broken link to Renesas RX processor.
3254 2017-02-10 Richard Biener <rguenther@suse.de>
3256 * toplev.c (process_options): Do not mention obsolete graphite
3257 options when printing sorry message about missing graphite support.
3258 Mention -floop-nest-optimize.
3260 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
3262 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
3263 (vtst_p16): Likewise.
3264 (vtstq_p8): Likewise.
3265 (vtstq_p16): Likewise.
3267 (vtstq_p64): Likewise.
3268 * config/arm/arm_neon.h (vgetq_lane_p64): New.
3269 (vset_lane_p64): New.
3270 (vsetq_lane_p64): New.
3272 2017-02-10 Jakub Jelinek <jakub@redhat.com>
3274 PR tree-optimization/79411
3275 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
3276 stmt operands are SSA_NAMEs used in abnormal phis.
3277 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
3280 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
3283 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
3286 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
3288 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
3290 2017-02-09 Jakub Jelinek <jakub@redhat.com>
3292 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
3296 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
3297 not arbitrary TREE_CONSTANT.
3300 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
3301 "omp declare target link" attribute unless is_global_var.
3302 * omp-offload.c (find_link_var_op): Likewise.
3304 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
3305 Chung-Lin Tang <cltang@codesourcery.com>
3307 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
3309 (gimplify_adjust_omp_clauses): Don't delete TILE.
3310 (gimplify_omp_for): Deal with TILE.
3311 * internal-fn.c (expand_GOACC_TILE): New function.
3312 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
3314 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
3315 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
3317 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
3318 avoid DIV for outermost collapse var.
3319 (expand_oacc_for): Insert tile element loop as needed. Adjust.
3320 Remove out of date comments, fix whitespace.
3321 * omp-general.c (omp_extract_for_data): Deal with tiling.
3322 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
3323 adjust OLF_DIM_BASE value.
3324 (struct omp_for_data): Add tiling field.
3325 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
3326 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
3327 for auto loops. Remove default auto determining, moved to
3328 oacc_loop_fixed_partitions.
3329 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
3330 stmts, add e_mask field.
3331 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
3332 (oacc_thread_numbers): Use oacc_dim_call.
3333 (oacc_xform_tile): New.
3334 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
3335 (finish_oacc_loop): Adjust for ifns vector.
3336 (oacc_loop_discover_walk): Append loop abstraction sites to list,
3337 add case for GOACC_TILE fns.
3338 (oacc_loop_xform_loop): Delete.
3339 (oacc_loop_process): Iterate over call list directly, and add
3340 handling for GOACC_TILE fns.
3341 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
3343 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
3344 vector partitioning to outer loops. Assign 2 partitions to loops
3345 when available. Add TILE handling.
3346 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
3347 (execite_oacc_device_lower): Process GOACC_TILE fns, ignore unknown specs.
3348 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
3349 * tree.c (omp_clause_num_ops): Adjust TILE ops.
3350 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
3352 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
3354 * configure.ac (ACX_BUGURL): Update.
3355 * configure: Regenerate.
3357 2017-02-09 Richard Biener <rguenther@suse.de>
3359 PR tree-optimization/69823
3360 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
3361 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
3363 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
3365 * config/arc/arc-c.def: Add __NPS400__ definition.
3366 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
3367 (TARGET_NPS400): Define.
3369 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
3371 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
3373 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
3375 (arc_cpu_types): Fill the arch_info field with a pointer into the
3376 arc_arch_types table.
3377 (arc_selected_cpu): Declare.
3378 * config/arc/arc.c (arc_selected_cpu): Make global.
3379 (arc_selected_arch): Delete.
3380 (arc_base_cpu): Delete.
3381 (arc_override_options): Remove references to deleted variables,
3382 update access to arch information.
3383 (ARC_OPT): Update access to arch information.
3384 (ARC_OPTX): Likewise.
3385 * config/arc/arc.h (arc_base_cpu): Remove declaration.
3386 (TARGET_ARC600): Update access to arch information.
3387 (TARGET_ARC601): Likewise.
3388 (TARGET_ARC700): Likewise.
3389 (TARGET_EM): Likewise.
3390 (TARGET_HS): Likewise.
3391 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
3394 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
3397 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
3398 condition/operands for integer GE/LE/GEU/LEU operations.
3400 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
3402 PR translation/79397
3403 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
3406 2017-02-08 Martin Jambor <mjambor@suse.cz>
3409 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
3410 whether allocation happened.
3411 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
3412 nothing was allocated.
3414 2017-02-08 Jakub Jelinek <jakub@redhat.com>
3416 PR tree-optimization/79408
3417 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
3418 constant, but SSA_NAME with a known integer range, use the minimum
3419 of that range instead of op1 to determine if modulo can be replaced
3420 with its first operand.
3422 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3424 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
3426 2017-02-08 Richard Biener <rguenther@suse.de>
3428 PR tree-optimization/71824
3429 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
3430 Check all loops contained in the merged region.
3432 2017-02-07 Andrew Pinski <apinski@cavium.com>
3434 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
3436 2017-02-07 Andrew Pinski <apinski@cavium.com>
3438 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
3439 (thunderxt88): Likewise.
3440 (thunderxt81): Disable LSE and change v8.1 to v8.
3441 (thunderxt83): Likewise.
3443 2017-02-07 Jakub Jelinek <jakub@redhat.com>
3444 Richard Biener <rguenther@suse.de>
3447 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
3448 type from int to size_t.
3449 * ira-costs.c (struct_costs_size): Change type from int to size_t.
3451 2017-02-07 Jakub Jelinek <jakub@redhat.com>
3453 PR rtl-optimization/79386
3454 * cprop.c (bypass_conditional_jumps): Initialize
3455 bypass_last_basic_block already before splitting bbs after
3456 unconditional traps...
3457 (bypass_conditional_jumps): ... rather than here.
3460 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
3461 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
3462 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
3463 fix -masm=intel patterns.
3465 2017-02-07 Richard Biener <rguenther@suse.de>
3467 PR tree-optimization/79256
3469 * builtins.c (get_object_alignment_2): Use min_align_of_type
3470 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
3471 and ADJUST_FIELD_ALIGN.
3473 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
3475 * doc/tm.texi: Regenerate.
3476 * stor-layout.c (layout_decl): Adjust.
3477 (update_alignment_for_field): Likewise.
3478 (place_field): Likewise.
3479 (min_align_of_type): Likewise.
3480 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
3481 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
3482 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
3483 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
3484 * config/frv/frv.c (frv_adjust_field_align): Likewise.
3485 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
3486 * config/i386/i386.c (x86_field_alignment): Likewise.
3487 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
3488 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
3489 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
3490 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
3491 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
3492 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
3496 2017-01-30 Richard Biener <rguenther@suse.de>
3498 PR tree-optimization/79256
3499 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
3500 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
3503 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
3505 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
3506 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
3507 builtins to SImode and emit a zero-extend, if necessary.
3509 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
3511 * docs/invoke.texi (RISC-V Options): Alphabetize.
3513 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
3515 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
3518 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
3520 * config/riscv/riscv.c: New file.
3521 * gcc/common/config/riscv/riscv-common.c: Likewise.
3522 * config.gcc: Likewise.
3523 * config/riscv/constraints.md: Likewise.
3524 * config/riscv/elf.h: Likewise.
3525 * config/riscv/generic.md: Likewise.
3526 * config/riscv/linux.h: Likewise.
3527 * config/riscv/multilib-generator: Likewise.
3528 * config/riscv/peephole.md: Likewise.
3529 * config/riscv/pic.md: Likewise.
3530 * config/riscv/predicates.md: Likewise.
3531 * config/riscv/riscv-builtins.c: Likewise.
3532 * config/riscv/riscv-c.c: Likewise.
3533 * config/riscv/riscv-ftypes.def: Likewise.
3534 * config/riscv/riscv-modes.def: Likewise.
3535 * config/riscv/riscv-opts.h: Likewise.
3536 * config/riscv/riscv-protos.h: Likewise.
3537 * config/riscv/riscv.h: Likewise.
3538 * config/riscv/riscv.md: Likewise.
3539 * config/riscv/riscv.opt: Likewise.
3540 * config/riscv/sync.md: Likewise.
3541 * config/riscv/t-elf-multilib: Likewise.
3542 * config/riscv/t-linux: Likewise.
3543 * config/riscv/t-linux-multilib: Likewise.
3544 * config/riscv/t-riscv: Likewise.
3545 * configure.ac: Likewise.
3546 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
3547 Waterman as RISC-V maintainers.
3548 * doc/install.texi: Add RISC-V entries.
3549 * doc/invoke.texi: Add RISC-V options section.
3550 * doc/md.texi: Add RISC-V constraints section.
3551 * configure: Regenerated.
3553 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
3556 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
3557 false values to be constant vectors with all 0 or all 1 bits set.
3558 (vcondu<mode><mode>): Likewise.
3559 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
3561 (fpmask_comparison_operator): Update comment.
3562 (vecint_comparison_operator): New predicate.
3563 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
3564 vector conditionals when the true and false values are constant
3565 vectors with all 0 bits or all 1 bits set.
3567 2017-02-06 Martin Sebor <msebor@redhat.com>
3569 PR tree-optimization/79376
3570 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
3572 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
3574 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
3575 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
3576 to simplify split condition.
3578 2017-02-06 Jakub Jelinek <jakub@redhat.com>
3580 * omp-expand.c (oxpand_omp_atomic_fetch_op,
3581 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
3584 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
3586 PR rtl-optimization/68664
3587 * target.def (can_speculate_insn): New hook.
3588 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
3589 * doc/tm.texi: Regenerate.
3590 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
3591 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
3592 (rs6000_sched_can_speculate_insn): New function.
3594 2017-02-06 Jakub Jelinek <jakub@redhat.com>
3596 PR tree-optimization/79284
3597 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
3598 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
3599 vectorizable_mask_load_store, vectorizable_operation,
3600 vect_is_simple_cond, get_same_sized_vectype): Use it instead
3601 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
3602 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
3603 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
3604 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
3605 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
3606 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
3607 is_gimple_assign (stmt). Replace another such test with
3608 is_gimple_assign (stmt).
3610 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
3613 * config/avr/avr.c (rtl-iter.h): Include it.
3614 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
3615 (avr_legitimate_combined_insn): ...and implementation.
3617 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
3619 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
3620 * config/s390/s390.c (s390_const_operand_ok)
3621 (s390_canonicalize_comparison, s390_extract_part)
3622 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
3623 (s390_contiguous_bitmask_p, s390_rtx_costs)
3624 (legitimize_pic_address): Likewise.
3625 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
3626 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
3627 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
3628 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
3629 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
3631 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
3633 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
3634 REGNO($0) == REGNO($1).
3636 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3638 * config/s390/linux.h(SIZE_TYPE): Add comment.
3640 2017-02-06 Julian Brown <julian@codesourcery.com>
3641 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
3642 Virendra Pathak <virendra.pathak@broadcom.com>
3644 * config/aarch64/aarch64-cores.def: Change the scheduler
3646 * config/aarch64/aarch64.md: Include thunderx2t99.md.
3647 * config/aarch64/thunderx2t99.md: New file.
3649 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
3651 * doc/standards.texi (Go Language): Update link to language
3654 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
3656 * tree-eh.c (lower_resx): Sanitize profile.
3657 (cleanup_empty_eh_move_lp): Likewise.
3659 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
3662 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
3664 * cfgloopmanip.h (loop_version): Update prototype.
3665 * modulo-sched.c (sms_schedule): Update call of loop_version.
3666 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
3667 * tree-parloops.c (gen_parallel_loop): Likewise.
3668 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
3669 * tree-ssa-loop-split.c (split_loop): Likewise.
3670 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
3671 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
3673 2017-02-05 Martin Liska <mliska@suse.cz>
3676 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
3678 (print_operand_address): Initialize a struct to zero.
3680 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
3682 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
3683 garbage collector only in textual form.
3685 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
3687 * doc/extend.texi (x86 specific memory model extensions for
3688 transactional memory): Simplify a phrase.
3690 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
3693 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
3694 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
3695 (atomic_storedi_1): Likewise.
3697 2017-02-04 Jakub Jelinek <jakub@redhat.com>
3699 PR tree-optimization/79338
3700 * tree-parloops.c (gather_scalar_reductions): Don't call
3701 vect_analyze_loop_form for loop->inner before destroying loop's
3704 2017-02-03 Martin Sebor <msebor@redhat.com>
3706 PR tree-optimization/79327
3707 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
3708 when precision has resulted in leading zeros.
3709 (format_integer): Adjust the likely counter to assume an unknown
3710 argument that may be zero is non-zero.
3712 2017-02-03 Jason Merrill <jason@redhat.com>
3715 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
3716 avoid copying non-taken branch.
3718 2017-02-03 Jakub Jelinek <jakub@redhat.com>
3720 PR tree-optimization/79340
3721 * tree-vect-loop.c (vectorizable_reduction): Release
3722 vec_defs elements after safe_splicing them into other vectors.
3725 PR tree-optimization/79327
3726 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
3727 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
3729 (format_integer): Use wide_int_to_tree instead of build_int_cst
3730 + to_?hwi. If argmin is NULL, just set argmin and argmax to
3731 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
3732 of shortest and longest sequence.
3734 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
3736 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
3737 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
3739 2017-02-03 Walter Lee <walt@tilera.com>
3742 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
3743 after initial stackframe link reg save.
3744 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
3746 2017-02-03 Jakub Jelinek <jakub@redhat.com>
3749 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
3750 wu for stxssp alternative.
3752 2017-02-03 Martin Sebor <msebor@redhat.com>
3754 PR tree-optimization/79352
3755 * gimple-fold.c (get_range_strlen): Add argument.
3756 (get_range_strlen): Change return type to bool.
3757 (get_maxval_strlen): Pass in a dummy argument.
3758 * gimple-fold.h (get_range_strlen): Change return type to bool.
3759 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
3760 * tree.h (array_at_struct_end_p): Add argument.
3761 * tree.c (array_at_struct_end_p): Handle it.
3763 2017-02-03 Martin Liska <mliska@suse.cz>
3766 * multiple_target.c (create_dispatcher_calls): Redirect edge
3767 from a caller of a dispatcher.
3768 (expand_target_clones): Make the clones local.
3769 (ipa_target_clone): Do both target clones and resolvers.
3770 (ipa_dispatcher_calls): Remove the pass.
3771 (pass_dispatcher_calls::gate): Likewise.
3772 (make_pass_dispatcher_calls): Likewise.
3773 * passes.def (pass_target_clone): Put as very first IPA early
3776 2017-02-03 Martin Liska <mliska@suse.cz>
3778 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
3779 in case of a function with ifunc attribute.
3781 2017-02-03 Martin Liska <mliska@suse.cz>
3783 * cgraph.c (cgraph_node::dump): Dump function version info.
3784 * symtab.c (symtab_node::dump_base): Add missing new line.
3786 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
3788 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
3789 (ifcombine_ifandif): Use it.
3791 2017-02-03 Martin Liska <mliska@suse.cz>
3793 * doc/invoke.texi: Document default value for
3794 use-after-scope-direct-emission-threshold.
3796 2017-02-03 Martin Liska <mliska@suse.cz>
3798 PR tree-optimization/79339
3799 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
3800 (format_floating): Likewise.
3802 2017-02-03 Martin Liska <mliska@suse.cz>
3805 * ipa-prop.c (ipa_node_params_t::insert): Remove current
3807 (ipa_node_params_t::remove): Likewise.
3808 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
3809 initialization from removed ipa_node_params_t::insert.
3810 (ipa_node_params::~ipa_node_params): Move from removed
3811 ipa_node_params_t::release.
3812 * symbol-summary.h (symbol_summary::m_released): New member.
3813 Do not release a summary twice. Do not allow to call finalizer
3814 for types of a summary that live in GGC memory.
3816 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
3818 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
3821 2017-02-02 Martin Sebor <msebor@redhat.com>
3824 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
3825 (format_string): Tighten up the range of output for non-constant
3826 strings and correct the expected range for wide non-constant strings.
3828 2017-02-02 Martin Sebor <msebor@redhat.com>
3830 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
3833 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
3835 (-fdump-tree-@var): Add to index and document how to come up
3836 with pass-specific option and dump file names.
3837 (-fdump-passes): Clarify where to look for output.
3839 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
3842 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
3843 statistics of the analyzed path; allow threading for speed when
3844 any of BBs along the path are optimized for speed.
3846 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
3849 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
3850 settings of the virtual registers.
3853 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
3855 * explow.c (get_dynamic_stack_size): Take known alignment of stack
3856 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
3859 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3861 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
3862 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
3864 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3866 * config/s390/s390.md: Add missing comments with the expanded
3868 * config/s390/vector.md: Likewise.
3869 * config/s390/vx-builtins.md: Likewise.
3871 2017-02-02 Jakub Jelinek <jakub@redhat.com>
3874 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
3875 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
3876 conditions on a single line.
3878 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3880 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
3881 __S390_VX__ to __VX__.
3883 2017-02-01 Andrew Pinski <apinski@cavium.com>
3885 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
3886 stmt_info to record_stmt_cost.
3887 (vect_get_known_peeling_cost): Pass stmt_info if known to
3889 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
3890 cpu_vector_cost field into
3891 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
3892 field into vec_int_stmt_cost and vec_fp_stmt_cost.
3893 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
3894 splitting of scalar_stmt_cost and vec_stmt_cost.
3895 (thunderx_vector_cost): Likewise.
3896 (cortexa57_vector_cost): LIkewise.
3897 (exynosm1_vector_cost): Likewise.
3898 (xgene1_vector_cost): Likewise.
3899 (thunderx2t99_vector_cost): Improve after the splitting of the two
3901 (aarch64_builtin_vectorization_cost): Update for the splitting of
3902 scalar_stmt_cost and vec_stmt_cost.
3904 2017-02-01 Torvald Riegel <triegel@redhat.com>
3905 Richard Henderson <rth@redhat.com>
3907 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
3908 conditional on existance of a fast atomic load.
3909 * optabs-query.c (can_atomic_load_p): New function.
3910 * optabs-query.h (can_atomic_load_p): Declare it.
3911 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
3912 no fast atomic load is available for the particular size of access.
3913 (expand_atomic_compare_and_swap): Likewise.
3914 (expand_atomic_load): Likewise.
3915 (expand_atomic_store): Likewise.
3916 (expand_atomic_fetch_op): Likewise.
3917 * testsuite/lib/target-supports.exp
3918 (check_effective_target_sync_int_128): Remove x86 because it provides
3919 no fast atomic load.
3920 (check_effective_target_sync_int_128_runtime): Likewise.
3922 2017-02-01 Richard Biener <rguenther@suse.de>
3924 * graphite.c: Include tree-vectorizer.h for find_loop_location.
3925 (graphite_transform_loops): Provide opt-info for optimized nests.
3926 * tree-parloop.c (parallelize_loops): Provide opt-info for
3929 2017-02-01 Richard Biener <rguenther@suse.de>
3932 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
3935 2017-02-01 Richard Biener <rguenther@suse.de>
3937 PR tree-optimization/71824
3938 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
3939 Verify the loops are valid in the merged SESE region.
3940 (scop_detection::can_represent_loop_1): Check analyzing the
3941 evolution of the number of iterations in the region succeeds.
3943 2017-01-31 Ian Lance Taylor <iant@golang.org>
3945 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
3946 REG_ARGS_SIZE note to 32-bit push insns and call insn.
3948 2017-01-31 David Malcolm <dmalcolm@redhat.com>
3950 PR preprocessor/79210
3951 * input.c (get_substring_ranges_for_loc): Replace line_width
3952 assertion with error-handling.
3954 2017-01-31 Richard Biener <rguenther@suse.de>
3956 PR tree-optimization/77318
3957 * graphite-sese-to-poly.c (extract_affine): Fix assert.
3958 (create_pw_aff_from_tree): Take loop parameter.
3959 (add_condition_to_pbb): Pass loop of the condition to
3960 create_pw_aff_from_tree.
3962 2017-01-31 Jakub Jelinek <jakub@redhat.com>
3964 * config/s390/s390.c (s390_asan_shadow_offset): New function.
3965 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
3967 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
3971 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
3973 (convert_int_to_float128): Likewise.
3974 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
3975 (convert_int_to_float128): Likewise.
3976 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
3977 (UNSPEC_IEEE128_CONVERT): Likewise.
3978 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
3979 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
3980 Use local variables for IBM extended format.
3981 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
3982 (fix_trunc<mode>si2_fprs): Likewise.
3983 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
3984 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
3985 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
3986 to know that we can now have integers of all sizes in vector
3988 (fix<uns>_<mode>di2_hw): Likewise.
3989 (float<uns>_<mode>si2_hw): Likewise.
3990 (fix_<mode>si2_hw): Likewise.
3991 (fixuns_<mode>si2_hw): Likewise.
3992 (float<uns>_<mode>di2_hw): Likewise.
3993 (float_<mode>di2_hw): Likewise.
3994 (float_<mode>si2_hw): Likewise.
3995 (floatuns_<mode>di2_hw): Likewise.
3996 (floatuns_<mode>si2_hw): Likewise.
3997 (xscvqp<su>wz_<mode>): Delete, no longer used.
3998 (xscvqp<su>dz_<mode>): Likewise.
3999 (xscv<su>dqp_<mode>): Likewise.
4000 (ieee128_mfvsrd_64bit): Likewise.
4001 (ieee128_mfvsrd_32bit): Likewise.
4002 (ieee128_mfvsrwz): Likewise.
4003 (ieee128_mtvsrw): Likewise.
4004 (ieee128_mtvsrd_64bit): Likewise.
4005 (ieee128_mtvsrd_32bit): Likewise.
4007 2017-01-31 Martin Liska <mliska@suse.cz>
4010 * ipa-prop.c (ipa_free_all_node_params): Call release method
4011 instead of ~sumbol_summary to not to trigger double times
4014 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
4016 PR tree-optimization/71691
4017 * bitmap.h (class auto_bitmap): New.
4018 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
4019 is_maybe_undefined instead of ssa_undefined_value_p.
4021 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4023 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
4024 __S390_ARCH_LEVEL__ to __ARCH__.
4026 2017-01-31 Jakub Jelinek <jakub@redhat.com>
4028 PR tree-optimization/79267
4029 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
4030 if should_remove_lhs_p is true.
4032 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
4035 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
4036 (add_alignment_attribute): New.
4037 (base_type_die): Add alignment attribute.
4038 (subrange_type_die): Likewise.
4039 (modified_type_die): Likewise.
4040 (gen_array_type_die): Likewise.
4041 (gen_descr_array_type_die: Likewise.
4042 (gen_enumeration_type_die): Likewise.
4043 (gen_subprogram_die): Likewise.
4044 (gen_variable_die): Likewise.
4045 (gen_field_die): Likewise.
4046 (gen_ptr_to_mbr_type_die): Likewise.
4047 (gen_struct_or_union_type_die): Likewise.
4048 (gen_subroutine_type_die): Likewise.
4049 (gen_typedef_die): Likewise.
4050 (base_type_cmp): Compare alignment attribute.
4052 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
4055 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
4056 (setb_unsigned) New pattern for setb with CCUNS.
4057 * config/rs6000/rs6000.c (expand_block_compare): Use a different
4058 subfc./subfe sequence to avoid overflow problems. Generate a
4059 shorter sequence with cmpld/setb for power9.
4060 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
4061 for generating subfc. instruction.
4062 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
4063 now uses this instruction.
4065 2017-01-30 Ian Lance Taylor <iant@google.com>
4068 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
4069 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
4071 2017-01-30 Martin Sebor <msebor@redhat.com>
4073 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
4074 Move constant to the right of a relational operator.
4075 (get_mpfr_format_length, format_character, format_string): Ditto.
4076 (should_warn_p, maybe_warn): Same.
4078 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
4080 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
4083 * asan.c (get_translation_unit_decl): Remove function.
4084 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
4086 2017-01-30 Martin Liska <mliska@suse.cz>
4088 PR gcov-profile/79259
4089 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
4092 2017-01-30 Martin Liska <mliska@suse.cz>
4095 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
4096 Initialize variables with NULL value.
4098 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
4101 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
4103 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
4105 2017-01-30 Richard Biener <rguenther@suse.de>
4107 PR tree-optimization/79276
4108 * tree-vrp.c (process_assert_insertions): Properly adjust common
4109 when removing a duplicate.
4111 * gcc.dg/torture/pr79276.c: New testcase.
4113 2017-01-30 Richard Biener <rguenther@suse.de>
4115 PR tree-optimization/79256
4116 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
4117 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
4119 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
4121 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
4124 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
4125 ("*r<noxa>sbg_<mode>_sll_bitmask")
4126 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
4127 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
4128 Use contiguous_bitmask_nowrap_operand.
4130 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4133 * config/rs6000/altivec.h (vec_xl): Revise #define.
4134 (vec_xst): Likewise.
4136 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
4138 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
4140 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
4142 PR rtl-optimization/79194
4143 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
4144 traps before call to bypass_conditional_jumps.
4146 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
4148 PR tree-optimization/71374
4149 * lra-constraints.c (check_conflict_input_operands): New.
4150 (match_reload): Use it.
4152 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
4155 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
4156 account to calculate conflict_set.
4158 2017-01-27 Bin Cheng <bin.cheng@arm.com>
4160 PR rtl-optimization/78559
4161 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
4162 other_insn in combine.
4164 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
4166 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
4167 uint16_type_node for BT_UINT16.
4169 2017-01-27 David Malcolm <dmalcolm@redhat.com>
4171 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
4172 "RTL Tests" to menu.
4173 (GIMPLE Tests): New node.
4174 (RTL Tests): New node.
4176 2017-01-27 Richard Biener <rguenther@suse.de>
4178 PR tree-optimization/79245
4179 * tree-loop-distribution.c (distribute_loop): Apply cost
4180 modeling also to detected patterns.
4182 2017-01-27 Richard Biener <rguenther@suse.de>
4184 PR tree-optimization/71433
4185 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
4186 (compare_assert_loc): New function.
4187 (process_assert_insertions): Sort and optimize assert locations
4188 to remove duplicates and push down identical assertions on
4189 edges to their destination block.
4191 2017-01-27 Richard Biener <rguenther@suse.de>
4193 PR tree-optimization/79244
4194 * tree-vrp.c (remove_range_assertions): Forcefully propagate
4195 out SSA names even if abnormal.
4197 2017-01-27 Jakub Jelinek <jakub@redhat.com>
4199 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
4200 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
4201 instead of MPFR_RNDN.
4203 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
4206 * arm.c (arm_option_override): Don't call build_target_option_node
4207 until after doing all option overrides.
4208 (arm_valid_target_attribute_tree): Likewise.
4210 2017-01-27 Martin Liska <mliska@suse.cz>
4212 * doc/invoke.texi (-fprofile-arcs): Document profiling support
4213 for {cd}tors and C++ {cd}tors.
4215 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
4217 * config/s390/s390.md ("*setmem_long_and")
4218 ("*setmem_long_and_31z"): Use zero_extend instead of and.
4220 2017-01-26 Martin Sebor <msebor@redhat.com>
4222 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
4225 2017-01-26 Martin Sebor <msebor@redhat.com>
4227 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
4228 HAVE_DFmode before using XFmode or DFmode.
4229 (parse_directive): Avoid using the z length modifier to avoid
4230 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
4233 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
4234 to accept adjustment as an array.
4235 (get_int_range): New function.
4236 (struct directive): Make width and prec arrays.
4237 (directive::set_width, directive::set_precision): Call get_int_range.
4238 (format_integer, format_floating): Handle width and precision ranges.
4239 (format_string, parse_directive): Same.
4241 2017-01-26 Jakub Jelinek <jakub@redhat.com>
4244 * dwarf2out.c (generate_skeleton_bottom_up): For children with
4245 comdat_type_p set, just clone them, but keep the children in the
4249 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
4250 which have direct callers with -fvar-tracking-assignments enabled
4252 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
4253 inside of type units.
4255 2017-01-26 Martin Sebor <msebor@redhat.com>
4258 * gimple-ssa-sprintf.c (struct result_range): Add likely and
4260 (struct format_result): Replace number_chars, number_chars_min,
4261 and number_chars_max with a single member of struct result_range.
4263 (format_result::operator+=): Adjust.
4264 (struct fmtresult): Remove bounded. Handle likely and unlikely
4266 (fmtresult::adjust_for_width_or_precision): New function.
4267 (fmtresult:type_max_digits): New function.
4268 (bytes_remaining): Handle likely and unlikely counters.
4269 (min_bytes_remaining): Remove.
4270 (format_percent): Simplify.
4271 (format_integer, format_floating): Set likely and unlikely counters.
4272 (get_string_length, format_character, format_string): Same.
4273 (format_plain, should_warn_p): New function.
4274 (maybe_warn): Call should_warn_p. Update diagnostic messages
4275 and handle those for all directives, including plain strings.
4276 (format_directive): Handle likely and unlikely counters.
4277 Remove unnecessary quoting from diagnostics. Add an informational
4279 (add_bytes): Remove.
4280 (pass_sprintf_length::compute_format_length): Simplify.
4281 (try_substitute_return_value): Handle likely and unlikely counters.
4283 2017-01-26 Carl Love <cel@us.ibm.com>
4285 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
4286 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
4288 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
4291 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
4292 endianess for subregs into account.
4293 * lra-constraints.c (lra_constraints): Do risky transformations
4294 always on the first iteration.
4295 * lra-lives.c (check_pseudos_live_through_calls): Add arg
4296 last_call_used_reg_set.
4297 (process_bb_lives): Define and use last_call_used_reg_set.
4298 * lra.c (lra): Always continue after lra_constraints on the first
4301 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
4303 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
4305 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
4307 2017-01-26 Jakub Jelinek <jakub@redhat.com>
4309 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
4310 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
4311 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
4312 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
4313 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
4314 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
4315 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
4316 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
4317 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
4319 2017-01-26 Marek Polacek <polacek@redhat.com>
4322 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
4323 for the third operand.
4325 2017-01-26 Jakub Jelinek <jakub@redhat.com>
4328 * omp-low.c (struct omp_context): Add simt_stmt field.
4329 (scan_omp_for): Return omp_context *.
4330 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
4331 context to the _simt_ SIMD stmt.
4332 (lower_omp_for): For combined SIMD with sibling _simt_
4333 SIMD, make sure to use the same decls in _looptemp_
4334 clauses as in the sibling.
4336 2017-01-26 David Sherwood <david.sherwood@arm.com>
4339 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
4342 2017-01-26 Jakub Jelinek <jakub@redhat.com>
4345 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
4346 emit fld b; fld a; if possible.
4348 * brig-builtins.def: Update copyright years.
4349 * config/arm/arm_acle_builtins.def: Update copyright years.
4351 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
4354 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
4355 constraint instead of o for the stxsd instruction.
4357 2017-01-25 Carl Love <cel@us.ibm.com>
4359 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
4360 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
4362 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
4364 * doc/invoke.texi (C++ Dialect Options): Fix typo.
4366 2017-01-25 Richard Biener <rguenther@suse.de>
4368 PR tree-optimization/69264
4369 * target.def (vector_alignment_reachable): Improve documentation.
4370 * doc/tm.texi: Regenerate.
4371 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
4373 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
4374 earlier changes with respect to TYPE_USER_ALIGN.
4375 (vector_alignment_reachable_p): Likewise. Improve dumping.
4377 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4380 * config/arm/arm.md (xordi3): Force constant operand into a register
4383 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4385 * doc/invoke.texi (-fstore-merging): Correct default optimization
4386 levels at which it is enabled.
4387 (-O): Move -fstore-merging from list to...
4390 2017-01-25 Richard Biener <rguenther@suse.de>
4393 * omp-expand.c: Include debug.h.
4394 (expand_omp_taskreg): Make sure to generate early debug before
4395 outlining anything from a function.
4396 (expand_omp_target): Likewise.
4397 (grid_expand_target_grid_body): Likewise.
4399 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
4402 * asan.c (get_translation_unit_decl): New function.
4403 (asan_add_global): Extract modules file name from globals
4404 TRANSLATION_UNIT_DECL name.
4406 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
4409 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
4410 for long calls with APCS frame and VFP.
4412 2017-01-24 David Malcolm <dmalcolm@redhat.com>
4414 * cfg.c (original_copy_tables_initialized_p): New function.
4415 * cfg.h (original_copy_tables_initialized_p): New decl.
4416 * cfgrtl.c (relink_block_chain): Guard the call to
4417 free_original_copy_tables with a call to
4418 original_copy_tables_initialized_p.
4419 * cgraph.h (symtab_node::native_rtl_p): New decl.
4420 * cgraphunit.c (symtab_node::native_rtl_p): New function.
4421 (symtab_node::needed_p): Don't assert for early assembly output
4422 for __RTL functions.
4423 (cgraph_node::finalize_function): Set "force_output" for __RTL
4425 (cgraph_node::analyze): Bail out early for __RTL functions.
4426 (analyze_functions): Update assertion to support __RTL functions.
4427 (cgraph_node::expand): Bail out early for __RTL functions.
4428 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
4430 * function.h (struct function): Update comment for field
4432 * gimple-expr.c: Include "tree-pass.h".
4433 (gimple_has_body_p): Return false for __RTL functions.
4434 * Makefile.in (OBJS): Add run-rtl-passes.o.
4435 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
4437 (gcc::pass_manager::get_clean_slate): New accessor.
4438 * passes.c: Include "insn-addr.h".
4439 (should_skip_pass_p): Add logging. Update logic for running
4440 "expand" to be compatible with both __GIMPLE and __RTL. Guard
4441 property-provider override so it is only done for gimple passes.
4443 (skip_pass): New function.
4444 (execute_one_pass): Call skip_pass when skipping passes.
4445 * read-md.c (md_reader::read_char): Support filtering
4446 the input to a subset of line numbers.
4447 (md_reader::md_reader): Initialize fields
4448 m_first_line and m_last_line.
4449 (md_reader::read_file_fragment): New function.
4450 * read-md.h (md_reader::read_file_fragment): New decl.
4451 (md_reader::m_first_line): New field.
4452 (md_reader::m_last_line): New field.
4453 * read-rtl-function.c (function_reader::create_function): Only
4454 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
4455 curr_properties. Set DECL_INITIAL to a dummy block.
4456 (read_rtl_function_body_from_file_range): New function.
4457 * read-rtl-function.h (read_rtl_function_body_from_file_range):
4459 * run-rtl-passes.c: New file.
4460 * run-rtl-passes.h: New file.
4462 2017-01-24 Jeff Law <law@redhat.com>
4464 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
4467 2017-01-24 Bin Cheng <bin.cheng@arm.com>
4469 PR tree-optimization/79159
4470 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
4471 (record_nonwrapping_iv): Improve boundary using above function if no
4472 value range information.
4474 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
4475 Martin Jambor <mjambor@suse.cz>
4477 * brig-builtins.def: New file.
4478 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
4479 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
4480 (DEF_HSAIL_SAT_BUILTIN): Likewise.
4481 (DEF_HSAIL_INTR_BUILTIN): Likewise.
4482 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
4483 * builtin-types.def (BT_INT8): New.
4484 (BT_INT16): Likewise.
4485 (BT_UINT8): Likewise.
4486 (BT_UINT16): Likewise.
4487 (BT_FN_ULONG): Likewise.
4488 (BT_FN_UINT_INT): Likewise.
4489 (BT_FN_UINT_ULONG): Likewise.
4490 (BT_FN_UINT_LONG): Likewise.
4491 (BT_FN_UINT_PTR): Likewise.
4492 (BT_FN_ULONG_PTR): Likewise.
4493 (BT_FN_INT8_FLOAT): Likewise.
4494 (BT_FN_INT16_FLOAT): Likewise.
4495 (BT_FN_UINT32_FLOAT): Likewise.
4496 (BT_FN_UINT16_FLOAT): Likewise.
4497 (BT_FN_UINT8_FLOAT): Likewise.
4498 (BT_FN_UINT64_FLOAT): Likewise.
4499 (BT_FN_UINT16_UINT32): Likewise.
4500 (BT_FN_UINT32_UINT16): Likewise.
4501 (BT_FN_UINT16_UINT16_UINT16): Likewise.
4502 (BT_FN_INT_PTR_INT): Likewise.
4503 (BT_FN_UINT_PTR_UINT): Likewise.
4504 (BT_FN_LONG_PTR_LONG): Likewise.
4505 (BT_FN_ULONG_PTR_ULONG): Likewise.
4506 (BT_FN_VOID_UINT64_UINT64): Likewise.
4507 (BT_FN_UINT8_UINT8_UINT8): Likewise.
4508 (BT_FN_INT8_INT8_INT8): Likewise.
4509 (BT_FN_INT16_INT16_INT16): Likewise.
4510 (BT_FN_INT_INT_INT): Likewise.
4511 (BT_FN_UINT_FLOAT_UINT): Likewise.
4512 (BT_FN_FLOAT_UINT_UINT): Likewise.
4513 (BT_FN_ULONG_UINT_UINT): Likewise.
4514 (BT_FN_ULONG_UINT_PTR): Likewise.
4515 (BT_FN_ULONG_ULONG_ULONG): Likewise.
4516 (BT_FN_UINT_UINT_UINT): Likewise.
4517 (BT_FN_VOID_UINT_PTR): Likewise.
4518 (BT_FN_UINT_UINT_PTR: Likewise.
4519 (BT_FN_UINT32_UINT64_PTR): Likewise.
4520 (BT_FN_INT_INT_UINT_UINT): Likewise.
4521 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
4522 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
4523 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
4524 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
4525 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
4526 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
4527 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
4528 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
4529 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
4530 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
4531 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
4532 * doc/frontends.texi: List BRIG FE.
4533 * doc/install.texi (Testing): Add BRIG tesring requirements.
4534 * doc/invoke.texi (Overall Options): Mention BRIG.
4535 * doc/standards.texi (Standards): Doucment BRIG HSA version.
4537 2017-01-24 Richard Biener <rguenther@suse.de>
4539 PR translation/79208
4540 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
4542 2017-01-24 Martin Jambor <mjambor@suse.cz>
4545 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
4546 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
4549 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
4552 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
4553 casts from signed to unsigned really don't have a range.
4555 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
4557 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
4558 GMP_RNDx for compatiblity.
4560 2017-01-24 Martin Liska <mliska@suse.cz>
4563 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
4564 that would prevent us to call alloca with -1 as argument.
4566 2017-01-24 Jakub Jelinek <jakub@redhat.com>
4568 * dwarf2out.c (output_compilation_unit_header, output_file_names):
4569 Avoid -Wformat-security warning.
4571 2017-01-23 Andrew Pinski <apinski@cavium.com>
4573 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
4576 2017-01-23 Martin Sebor <msebor@redhat.com>
4579 * gimple-ssa-sprintf.c (warn_level): New global.
4580 (format_integer): Use it here and throughout the rest of the file.
4581 Use the same switch to compute sign as base.
4582 (maybe_warn): New function.
4583 (format_directive): Factor out warnings into maybe_warn.
4584 Add debugging output. Use warn_level.
4585 (add_bytes): Use warn_level.
4586 (pass_sprintf_length::compute_format_length): Add debugging output.
4587 (try_substitute_return_value): Same.
4588 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
4591 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
4592 (struct fmtresult, format_integer, format_floating): Adjust.
4593 (fmtresult::fmtresult): Set max correctly in two argument ctor.
4594 (get_string_length, format_string,format_directive): Same.
4595 (pass_sprintf_length::compute_format_length): Same.
4596 (try_substitute_return_value): Simplify slightly.
4599 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
4600 (fmtresult::operator+=): Outlined.
4601 (struct fmtresult): Add ctors.
4602 (struct conversion_spec): Rename...
4603 (struct directive): ...to this. Add and remove data members.
4604 (directive::set_width, directive::set_precision): New functions.
4605 (format_percent): Use fmtresult ctor.
4606 (get_width_and_precision): Remove.
4607 (format_integer): Make naming changes. Avoid computing width and
4609 (format_floating): Same. Adjust indentation.
4610 (format_character, format_none): New functions.
4611 (format_string): Moved character handling to format_character.
4612 (format_directive): Remove arguments, change return type.
4613 (parse_directive): New function.
4614 (pass_sprintf_length::compute_format_length): Move directive
4615 parsing to parse_directive.
4617 2017-01-23 Jakub Jelinek <jakub@redhat.com>
4619 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
4620 (assign_assembler_name_if_needed): ... this.
4621 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
4622 (assign_assembler_name_if_needed): ... this.
4623 (free_lang_data_in_cgraph): Adjust callers.
4624 * cgraphunit.c (cgraph_node::analyze): Likewise.
4625 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
4628 2017-01-23 Richard Biener <rguenther@suse.de>
4630 PR tree-optimization/79088
4631 PR tree-optimization/79188
4632 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
4633 resetting loop bounds after last path deletion. Reset loop
4634 bounds of the target loop, make code match the comments.
4635 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
4636 Make sure loops need no fixups.
4638 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
4640 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
4641 exponent support with double type for first argument.
4642 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
4643 type returned by __builtin_vec_extract_sig,
4644 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
4645 functions from "vector int" to "vector unsigned int" or from
4646 "vector long long int" to "vector unsigned long long int".
4647 Changed type returned by __builtin_vec_extract_exp,
4648 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
4649 functions from "vector int" to "vector unsigned int" or from
4650 "vector long long int" to "vector unsigned long long int".
4651 Changed return type of __builtin_vec_test_data_class,
4652 __builtin_vec_test_data_class_sp, and
4653 __builtin_vec_test_data_class_dp from "vector int" to
4654 "vector bool int" or from "vector long long int" to "vector bool
4655 long long int" and changed second argument type from "unsigned
4656 int" to "int". Added new overloaded function forms "vector float
4657 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
4658 "vector float __builtin_vec_insert_exp_sp (vector float, vector
4659 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
4660 double, vector unsigned long long int)" and "vector double
4661 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
4662 long int)". Changed return type of
4663 __builtin_scalar_test_data_class and
4664 __builtin_scalar_test_data_class_sp and
4665 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
4666 int" and changed second argument from "unsigned int" to "int".
4667 Changed type returned by __builtin_scalar_test_neg,
4668 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
4669 from "int" to "bool int". Added new overloaded function form
4670 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
4671 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
4672 exponent double-precision with floating point first argument.
4673 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
4674 documentation of scalar_test_data_class, scalar_test_neg,
4675 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
4676 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
4677 vec_test_data_class built-in functions to reflect refinements in
4678 their type signatures.
4680 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
4682 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
4684 (aarch64_elf_asm_destructor): Likewise.
4686 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
4688 PR rtl-optimization/78634
4689 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
4690 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
4691 * ifcvt.c (noce_try_cmove): Add missing cost check.
4693 PR rtl-optimization/71724
4694 * combine.c (if_then_else_cond): Look for situations where it is
4695 beneficial to undo the work of one of the recursive calls.
4697 2017-01-23 Bin Cheng <bin.cheng@arm.com>
4699 PR tree-optimization/70754
4700 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
4701 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
4702 combined stmt before it if not NULL.
4703 (combine_chains): Process refs reversely and compute dominance point
4706 2017-01-23 Martin Liska <mliska@suse.cz>
4708 PR tree-optimization/79196
4709 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
4710 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
4712 (strlen_optimize_stmt): Call the renamed function.
4714 2017-01-23 Michael Matz <matz@suse.de>
4716 PR tree-optimization/78384
4717 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
4719 2017-01-23 Richard Biener <rguenther@suse.de>
4721 PR tree-optimization/79186
4722 * tree-vrp.c (register_new_assert_for): Make sure we've seen
4723 both incoming edges before moving an assert.
4725 2017-01-23 Martin Jambor <mjambor@suse.cz>
4727 * ipa-prop.c (load_from_param_1): Removed.
4728 (load_from_unmodified_param): Bits from load_from_param_1 put back
4730 (load_from_param): Removed.
4731 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
4732 with stmt. Reverted back to use of load_from_unmodified_param.
4734 2017-01-23 Martin Jambor <mjambor@suse.cz>
4737 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
4738 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
4739 field a pointer to garbage collected vector, mark lattices and
4740 ipcp_orig_node with GTY((skip)).
4741 (ipa_get_param_count): Adjust to descriptors being a pointer.
4742 (ipa_get_param): Likewise.
4743 (ipa_get_type): Likewise.
4744 (ipa_get_param_move_cost): Likewise.
4745 (ipa_set_param_used): Likewise.
4746 (ipa_get_controlled_uses): Likewise.
4747 (ipa_set_controlled_uses): Likewise.
4748 (ipa_is_param_used): Likewise.
4749 (ipa_node_params_t): Move into garbage collector. New methods insert
4751 (ipa_node_params_sum): Annotate wth GTY(()).
4752 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
4754 (ipa_load_from_parm_agg): Adjust declaration.
4755 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
4756 * ipa-profile.c (ipa_profile): Likewise.
4757 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
4758 (ipa_populate_param_decls): Make descriptors parameter garbage
4760 (ipa_dump_param): Adjust to descriptors being a pointer.
4761 (ipa_alloc_node_params): Likewise.
4762 (ipa_initialize_node_params): Likewise.
4763 (load_from_param_1): Make descriptors parameter garbage collected.
4764 (load_from_unmodified_param): Likewise.
4765 (load_from_param): Likewise.
4766 (ipa_load_from_parm_agg): Likewise.
4767 (ipa_node_params::~ipa_node_params): Removed.
4768 (ipa_free_all_node_params): Remove call to delete operator.
4769 (ipa_node_params_t::insert): New.
4770 (ipa_node_params_t::remove): Likewise.
4771 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
4772 copy known_csts and known_contexts vectors.
4773 (ipa_read_node_info): Adjust to descriptors being a pointer.
4774 (ipcp_modif_dom_walker): Make m_descriptors field garbage
4776 (ipcp_transform_function): Make descriptors variable garbage
4779 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
4781 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
4782 * config/i386/avx512dqintrin.h: Ditto.
4783 * config/i386/avx512fintrin.h: Ditto.
4784 * gcc/config/i386/i386.c: Handle new builtins.
4785 * config/i386/i386-builtin.def: Add new builtins.
4786 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
4787 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
4789 2017-01-23 Jakub Jelinek <jakub@redhat.com>
4790 Martin Liska <mliska@suse.cz>
4792 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
4793 * asan.c (asan_expand_poison_ifn): Support stores and use
4794 appropriate ASAN report function.
4795 * internal-fn.c (expand_ASAN_POISON_USE): New function.
4796 * internal-fn.def (ASAN_POISON_USE): Declare.
4797 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
4798 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
4799 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
4800 ASAN_POISON calls w/o LHS.
4801 * tree-ssa.c (execute_update_addresses_taken): Create clobber
4802 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
4803 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
4804 * gimplify.c (asan_poison_variables): Add attribute
4805 use_after_scope_memory to variables that really needs to live
4807 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
4808 having the attribute.
4810 2017-01-23 Martin Liska <mliska@suse.cz>
4812 * asan.c (create_asan_shadow_var): New function.
4813 (asan_expand_poison_ifn): Likewise.
4814 * asan.h (asan_expand_poison_ifn): New declaration.
4815 * internal-fn.c (expand_ASAN_POISON): Likewise.
4816 * internal-fn.def (ASAN_POISON): New builtin.
4817 * sanopt.c (pass_sanopt::execute): Expand
4818 asan_expand_poison_ifn.
4819 * tree-inline.c (copy_decl_for_dup_finish): Make function
4821 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
4822 * tree-ssa.c (is_asan_mark_p): New function.
4823 (execute_update_addresses_taken): Rewrite local variables
4824 (identified just by use-after-scope as addressable) into SSA.
4826 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
4828 * doc/install.texi (Specific): opensource.apple.com uses https
4829 now. Remove trailing slash.
4831 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
4833 * README.Portability: Remove note on an Irix compatibility issue.
4835 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
4837 * gcov.c (INCLUDE_ALGORITHM): Define.
4838 (INCLUDE_VECTOR): Define.
4839 No longer include <vector> and <algorithm> directly.
4841 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
4843 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
4845 * doc/invoke.texi (Code Gen Options): Ditto.
4847 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
4850 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
4852 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
4854 rtl-optimization/79125
4855 * cprop.c (local_cprop_pass): Handle cases where we make an
4858 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
4862 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
4863 read from, for big endian.
4865 2017-01-20 Jiong Wang <jiong.wang@arm.com>
4867 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
4868 register pauth builtins for LP64 only.
4870 2017-01-20 Marek Polacek <polacek@redhat.com>
4873 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
4876 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
4878 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
4880 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
4881 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
4882 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
4884 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4887 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
4888 in big-endian mode when they are not a single duplicated value.
4890 2017-01-20 Richard Biener <rguenther@suse.de>
4892 * BASE-VER: Bump to 7.0.1.
4894 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
4896 * omp-low.c (omplow_simd_context): New struct. Use it...
4897 (lower_rec_simd_input_clauses): ...here and...
4898 (lower_rec_input_clauses): ...here to hold common data. Adjust all
4899 references to idx, lane, max_vf, is_simt.
4901 2017-01-20 Graham Markall <graham.markall@embecosm.com>
4903 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
4906 2017-01-20 Martin Jambor <mjambor@suse.cz>
4908 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
4909 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
4911 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
4912 (GTFILES): Rename hsa.c to hsa-common.c.
4913 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
4914 * hsa-dump.c: Likewise.
4915 * hsa-gen.c: Likewise.
4916 * hsa-regalloc.c: Likewise.
4917 * ipa-hsa.c: Likewise.
4918 * omp-expand.c: Likewise.
4919 * omp-low.c: Likewise.
4920 * toplev.c: Likewise.
4922 2017-01-20 Marek Polacek <polacek@redhat.com>
4925 * doc/invoke.texi: Document -Wduplicated-branches.
4926 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
4927 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
4928 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
4929 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
4930 return 0 only when not OEP_LEXICOGRAPHIC.
4931 (fold_build_cleanup_point_expr): Use the expression
4932 location when building CLEANUP_POINT_EXPR.
4933 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
4934 * tree.c (add_expr): Handle error_mark_node.
4936 2017-01-20 Martin Liska <mliska@suse.cz>
4939 * tree-profile.c (init_ic_make_global_vars): Do not call
4941 (gimple_init_gcov_profiler): Likewise.
4943 2017-01-20 Martin Liska <mliska@suse.cz>
4946 * cgraph.h (maybe_create_reference): Remove argument and
4948 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
4950 * ipa-cp.c (create_specialized_node): Likewise.
4951 * symtab.c (symtab_node::maybe_create_reference): Handle
4952 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
4954 2017-01-20 Martin Liska <mliska@suse.cz>
4956 * read-rtl-function.c (function_reader::create_function): Use
4957 build_decl instread of build_decl_stat.
4959 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
4961 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
4962 * config/i386/avx512dqintrin.h: Ditto.
4963 * config/i386/avx512fintrin.h: Ditto.
4964 * config/i386/i386-builtin-types.def: Add new types.
4965 * gcc/config/i386/i386.c: Handle new types.
4966 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
4967 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
4968 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
4969 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
4970 (__builtin_ia32_kshiftridi): New.
4971 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
4973 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
4977 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
4978 define to rs6000_init_stack_protect_guard.
4979 (rs6000_init_stack_protect_guard): New function.
4981 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
4982 Yunqiang Su <yunqiang.su@imgtec.com>
4984 * config.gcc (supported_defaults): Add madd4.
4985 (with_madd4): Add validation.
4986 (all_defaults): Add madd4.
4987 * config/mips/mips.opt (mmadd4): New option.
4988 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
4990 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
4992 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
4993 (ISA_HAS_FUSED_MADD4): Likewise.
4994 * gcc/doc/invoke.texi (-mmadd4): Document the new option.
4995 * gcc/doc/install.texi (--with-madd4): Document the new option.
4997 2017-01-19 Jiong Wang <jiong.wang@arm.com>
4999 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
5000 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
5001 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
5002 (aarch64_init_pauth_hint_builtins): New.
5003 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
5004 (aarch64_expand_builtin): Expand new builtins.
5006 2017-01-19 Jiong Wang <jiong.wang@arm.com>
5008 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
5009 * combine-stack-adj.c (no_unhandled_cfa): Handle
5010 REG_CFA_TOGGLE_RA_MANGLE.
5011 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
5012 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
5013 info for return address signing.
5014 (aarch64_expand_epilogue): Likewise.
5016 2017-01-19 Jiong Wang <jiong.wang@arm.com>
5018 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
5019 * config/aarch64/aarch64-protos.h
5020 (aarch64_return_address_signing_enabled): New declaration.
5021 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
5023 (aarch64_expand_prologue): Sign return address before it's pushed onto
5025 (aarch64_expand_epilogue): Authenticate return address fetched from
5027 (aarch64_override_options): Sanity check for ILP32 and ISA level.
5028 (aarch64_attributes): New function attributes for "sign-return-address".
5029 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
5030 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
5031 ("*do_return"): Generate combined instructions according to key index.
5032 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
5033 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
5035 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
5036 * config/aarch64/aarch64.opt (msign-return-address=): New.
5037 * doc/extend.texi (AArch64 Function Attributes): Documents
5038 "sign-return-address=".
5039 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
5041 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
5043 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
5044 overall option summary.
5046 2017-01-19 Jiong Wang <jiong.wang@arm.com>
5048 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
5049 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
5050 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
5051 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
5053 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
5055 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
5056 -mpower9-minmax by default for -mcpu=power9.
5057 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
5058 128-bit floating point.
5060 2017-01-20 Alan Modra <amodra@gmail.com>
5062 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
5063 optimizing for size.
5065 2017-01-20 Alan Modra <amodra@gmail.com>
5068 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
5069 for strcmp and strncmp from corresponding builtin decl.
5071 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
5073 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
5074 instead of i386/rtems-64.h.
5075 * config/i386/rtems-64.h: Remove.
5077 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
5081 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
5083 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
5085 2017-01-19 Tamar Christina <tamar.christina@arm.com>
5087 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
5088 Change int to HOST_WIDE_INT.
5089 * config/aarch64/aarch64-protos.h
5090 (aarch64_simd_gen_const_vector_dup): Likewise.
5091 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
5093 2017-01-19 David Malcolm <dmalcolm@redhat.com>
5095 * langhooks-def.h (lhd_type_for_size): New decl.
5096 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
5097 * langhooks.c (lhd_type_for_size): New function, taken from
5100 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
5102 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
5103 define_bypass for CR latency.
5104 (power9-cracked-alu): Update bypass latency and remove power9-branch.
5105 (power9-alu2): Add define_bypass for CR latency.
5107 (power9-mul): Update insn latency.
5108 (power9-mul-compare): Update insn latency, bypass latency and remove
5111 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5113 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
5115 * config/aarch64/aarch64.md
5116 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
5117 aarch64_nopcrelative_literal_loads.
5118 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
5120 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
5122 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
5124 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
5126 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
5129 * config.gcc (supported_defaults): Add lxc1-sxc1.
5130 (with_lxc1_sxc1): Add validation.
5131 (all_defaults): Add lxc1-sxc1.
5132 * config/mips/mips.opt (mlxc1-sxc1): New option.
5133 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
5135 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
5136 __mips_no_lxc1_sxc1.
5137 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
5138 * gcc/doc/invoke.texi (-mlxc1-sxc1): Document the new option.
5139 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
5141 2017-01-19 Richard Biener <rguenther@suse.de>
5143 PR tree-optimization/72488
5144 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
5145 sure to restore SSA info.
5146 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
5148 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
5150 PR rtl-optimization/79121
5151 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
5152 of the inner type when shifting an extended value.
5154 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
5157 * symtab.c (symtab_node::equal_address_to): Fix comparing of
5158 interposable aliases.
5160 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
5163 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
5164 Use the evmergelohi instruction.
5165 (mov_si<mode>_e500_subreg4_2_le): Likewise.
5166 (mov_sitf_e500_subreg8_2_be): Likewise.
5167 (mov_sitf_e500_subreg12_2_le): Likewise.
5168 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
5169 (mov_si<mode>_e500_subreg4_2_be): Likewise.
5170 (mov_sitf_e500_subreg8_2_le): Likewise.
5171 (mov_sitf_e500_subreg12_2_be): Likewise.
5173 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5175 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
5176 attribute from vecsimple to vecperm.
5177 (altivec_vbpermq2): Likewise.
5179 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5182 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
5184 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
5185 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
5186 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
5187 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
5188 case where N arg is SIZE_MAX.
5189 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
5190 (cmpstrsi): Add pattern.
5192 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
5194 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5195 __builtin_vec_revb builtins.
5196 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
5197 built-in functions to support generation of the ISA 3.0 XXBR<x>
5198 vector byte reverse instructions.
5199 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
5200 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
5201 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
5202 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
5203 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
5204 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
5205 (P9V_BUILTIN_VEC_REVB): Likewise.
5206 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
5207 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
5208 (p9_xxbrq_v16qi): Likewise.
5209 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
5210 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
5211 (p9_xxbrh_v8hi): Likewise.
5212 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
5213 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
5214 vec_revb built-in functions.
5216 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
5218 PR rtl-optimization/78952
5219 * config/i386/i386.md (any_extract): New code iterator.
5220 (*insvqi_2): Use any_extract for source operand.
5221 (*insvqi_3): Use any_shiftrt for source operand.
5223 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
5225 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
5227 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
5229 2017-01-18 Matthias Klose <doko@ubuntu.com>
5231 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
5233 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5235 * config/rs6000/altivec.h (vec_bperm): Change #define.
5236 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
5237 (altivec_vbpermq2): New define_insn.
5238 (altivec_vbpermd): Likewise.
5239 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
5241 (VBPERMD): Likewise.
5242 (VBPERM): New polymorphic function interface.
5243 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
5244 Add entries for P9V_BUILTIN_VEC_VBPERM.
5245 * doc/extend.texi: Add interfaces for vec_bperm.
5247 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5249 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
5250 first letter of error messages.
5251 (s390_resolve_overloaded_builtin): Likewise.
5252 * config/s390/s390.c (s390_expand_builtin): Likewise.
5253 (s390_invalid_arg_for_unprototyped_fn): Likewise.
5254 (s390_valid_target_attribute_inner_p): Likewise.
5255 * config/s390/s390.md ("tabort"): Likewise.
5257 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
5259 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
5260 (ISA_AVOID_DIV_HILO): New macro.
5261 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
5262 (ISA_HAS_DDIV): Likewise.
5264 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
5266 * doc/invoke.texi (fabi-version): Correct number of occurrences.
5268 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
5270 * doc/invoke.texi (fabi-version): Spelling fix.
5272 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
5275 * doc/invoke.texi (fabi-version): Mention mangling fix for
5278 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
5281 * doc/invoke.texi (fabi-version): Document discriminator mangling.
5283 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
5286 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
5287 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
5289 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
5291 (stack_protect_test): Ditto.
5292 * config/rs6000/rs6000.opt (mstack-protector-guard=,
5293 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
5295 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
5296 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
5297 -mstack-protector-guard-offset=.
5298 (RS/6000 and PowerPC Options): Ditto.
5300 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
5302 * config/i386/i386.h (MASK_CLASS_P): New define.
5303 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
5304 there are no registers from different register sets also when
5305 mask registers are used. Update function comment.
5306 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
5307 to (*k/*r) and (*k/*km) alternatives.
5309 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
5311 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
5312 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
5313 (EH_RETURN_HANDLER_RTX): New define.
5314 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
5315 Force frame pointer in EH return functions.
5316 (aarch64_expand_epilogue): Add barrier for eh_return.
5317 (aarch64_final_eh_return_addr): Remove.
5318 (aarch64_eh_return_handler_rtx): New function.
5319 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
5321 (aarch64_eh_return_handler_rtx): New prototype.
5323 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5325 * config/rs6000/altivec.h (vec_rlmi): New #define.
5326 (vec_vrlnm): Likewise.
5327 (vec_rlnm): Likewise.
5328 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
5329 (UNSPEC_VRLNM): Likewise.
5330 (VIlong): New mode iterator.
5331 (altivec_vrl<VI_char>mi): New define_insn.
5332 (altivec_vrl<VI_char>nm): Likewise.
5333 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
5336 (RLNM): New polymorphic function entry.
5337 (VRLWMI): New monomorphic function entry.
5339 (RLMI): New polymorphic function entry.
5340 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
5341 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
5342 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
5345 2017-01-17 Jakub Jelinek <jakub@redhat.com>
5348 * dwarf2out.c (field_byte_offset): Restore the
5349 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
5350 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
5351 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
5354 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
5357 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
5359 2017-01-17 Jakub Jelinek <jakub@redhat.com>
5362 * dwarf2out.c (add_data_member_location_attribute): For constant
5363 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
5364 instead of DW_AT_data_member_location, DW_AT_bit_offset and
5365 DW_AT_byte_size attributes.
5367 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
5369 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
5370 after forcing to constant memory when the code model is medium.
5372 2017-01-17 Julia Koval <julia.koval@intel.com>
5375 * config/i386/avx512fintrin.h
5376 (_mm512_i32gather_ps): Change __addr type to void const*.
5377 (_mm512_mask_i32gather_ps): Ditto.
5378 (_mm512_i32gather_pd): Ditto.
5379 (_mm512_mask_i32gather_pd): Ditto.
5380 (_mm512_i64gather_ps): Ditto.
5381 (_mm512_mask_i64gather_ps): Ditto.
5382 (_mm512_i64gather_pd): Ditto.
5383 (_mm512_mask_i64gather_pd): Ditto.
5384 (_mm512_i32gather_epi32): Ditto.
5385 (_mm512_mask_i32gather_epi32): Ditto.
5386 (_mm512_i32gather_epi64): Ditto.
5387 (_mm512_mask_i32gather_epi64): Ditto.
5388 (_mm512_i64gather_epi32): Ditto.
5389 (_mm512_mask_i64gather_epi32): Ditto.
5390 (_mm512_i64gather_epi64): Ditto.
5391 (_mm512_mask_i64gather_epi64): Ditto.
5392 (_mm512_i32scatter_ps): Change __addr type to void*.
5393 (_mm512_mask_i32scatter_ps): Ditto.
5394 (_mm512_i32scatter_pd): Ditto.
5395 (_mm512_mask_i32scatter_pd): Ditto.
5396 (_mm512_i64scatter_ps): Ditto.
5397 (_mm512_mask_i64scatter_ps): Ditto.
5398 (_mm512_i64scatter_pd): Ditto.
5399 (_mm512_mask_i64scatter_pd): Ditto.
5400 (_mm512_i32scatter_epi32): Ditto.
5401 (_mm512_mask_i32scatter_epi32): Ditto.
5402 (_mm512_i32scatter_epi64): Ditto.
5403 (_mm512_mask_i32scatter_epi64): Ditto.
5404 (_mm512_i64scatter_epi32): Ditto.
5405 (_mm512_mask_i64scatter_epi32): Ditto.
5406 (_mm512_i64scatter_epi64): Ditto.
5407 (_mm512_mask_i64scatter_epi64): Ditto.
5408 * config/i386/avx512pfintrin.h
5409 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
5410 (_mm512_mask_prefetch_i32gather_ps): Ditto.
5411 (_mm512_mask_prefetch_i64gather_pd): Ditto.
5412 (_mm512_mask_prefetch_i64gather_ps): Ditto.
5413 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
5414 (_mm512_prefetch_i32scatter_ps): Ditto.
5415 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
5416 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
5417 (_mm512_prefetch_i64scatter_pd): Ditto.
5418 (_mm512_prefetch_i64scatter_ps): Ditto.
5419 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
5420 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
5421 * config/i386/avx512vlintrin.h
5422 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
5423 (_mm_mmask_i32gather_ps): Ditto.
5424 (_mm256_mmask_i32gather_pd): Ditto.
5425 (_mm_mmask_i32gather_pd): Ditto.
5426 (_mm256_mmask_i64gather_ps): Ditto.
5427 (_mm_mmask_i64gather_ps): Ditto.
5428 (_mm256_mmask_i64gather_pd): Ditto.
5429 (_mm_mmask_i64gather_pd): Ditto.
5430 (_mm256_mmask_i32gather_epi32): Ditto.
5431 (_mm_mmask_i32gather_epi32): Ditto.
5432 (_mm256_mmask_i32gather_epi64): Ditto.
5433 (_mm_mmask_i32gather_epi64): Ditto.
5434 (_mm256_mmask_i64gather_epi32): Ditto.
5435 (_mm_mmask_i64gather_epi32): Ditto.
5436 (_mm256_mmask_i64gather_epi64): Ditto.
5437 (_mm_mmask_i64gather_epi64): Ditto.
5438 (_mm256_i32scatter_ps): Change __addr type to void*.
5439 (_mm256_mask_i32scatter_ps): Ditto.
5440 (_mm_i32scatter_ps): Ditto.
5441 (_mm_mask_i32scatter_ps): Ditto.
5442 (_mm256_i32scatter_pd): Ditto.
5443 (_mm256_mask_i32scatter_pd): Ditto.
5444 (_mm_i32scatter_pd): Ditto.
5445 (_mm_mask_i32scatter_pd): Ditto.
5446 (_mm256_i64scatter_ps): Ditto.
5447 (_mm256_mask_i64scatter_ps): Ditto.
5448 (_mm_i64scatter_ps): Ditto.
5449 (_mm_mask_i64scatter_ps): Ditto.
5450 (_mm256_i64scatter_pd): Ditto.
5451 (_mm256_mask_i64scatter_pd): Ditto.
5452 (_mm_i64scatter_pd): Ditto.
5453 (_mm_mask_i64scatter_pd): Ditto.
5454 (_mm256_i32scatter_epi32): Ditto.
5455 (_mm256_mask_i32scatter_epi32): Ditto.
5456 (_mm_i32scatter_epi32): Ditto.
5457 (_mm_mask_i32scatter_epi32): Ditto.
5458 (_mm256_i32scatter_epi64): Ditto.
5459 (_mm256_mask_i32scatter_epi64): Ditto.
5460 (_mm_i32scatter_epi64): Ditto.
5461 (_mm_mask_i32scatter_epi64): Ditto.
5462 (_mm256_i64scatter_epi32): Ditto.
5463 (_mm256_mask_i64scatter_epi32): Ditto.
5464 (_mm_i64scatter_epi32): Ditto.
5465 (_mm_mask_i64scatter_epi32): Ditto.
5466 (_mm256_i64scatter_epi64): Ditto.
5467 (_mm256_mask_i64scatter_epi64): Ditto.
5468 (_mm_i64scatter_epi64): Ditto.
5469 (_mm_mask_i64scatter_epi64): Ditto.
5470 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
5471 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
5472 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
5473 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
5474 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
5475 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
5476 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
5477 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
5478 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
5479 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
5480 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
5481 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
5482 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
5483 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
5484 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
5485 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
5486 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
5487 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
5488 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
5489 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
5490 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
5491 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
5492 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
5493 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
5494 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
5495 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
5496 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
5497 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
5498 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
5499 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
5500 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
5501 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
5502 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
5503 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
5504 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
5505 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
5506 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
5507 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
5508 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
5509 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
5510 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
5511 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
5512 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
5513 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
5514 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
5515 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
5516 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
5517 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
5518 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
5519 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
5520 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
5521 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
5522 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
5523 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
5524 definitions accordingly.
5526 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
5527 Kuan-Lin Chen <kuanlinchentw@gmail.com>
5530 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
5533 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
5536 * ira-conflicts.c (ira_build_conflicts): Update total conflict
5537 hard regs for inner regno.
5539 2017-01-17 Martin Liska <mliska@suse.cz>
5542 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
5543 assumption and add comment.
5545 2017-01-17 Nathan Sidwell <nathan@acm.org>
5547 * ipa-visibility.c (localize_node): New function, broken out of ...
5548 (function_and_variable_visibility): ... here. Call it.
5550 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
5553 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
5554 correctly set frequency of oudgoing edge.
5555 (duplicate_thread_path): Fix profile updating.
5557 2017-01-17 Jakub Jelinek <jakub@redhat.com>
5560 * configure.ac: Add GCC_BASE_VER.
5561 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
5562 version from BASE-VER file.
5563 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
5564 (gcc.o): Depend on $(BASEVER).
5565 * common.opt (dumpfullversion): New option.
5566 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
5567 * doc/invoke.texi: Document -dumpfullversion.
5568 * doc/install.texi: Document --with-gcc-major-version-only.
5569 * configure: Regenerated.
5571 2017-01-17 Richard Biener <rguenther@suse.de>
5573 PR tree-optimization/71433
5574 * tree-vrp.c (register_new_assert_for): Merge same asserts
5575 on all incoming edges.
5576 (process_assert_insertions_for): Handle insertions at the
5579 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
5581 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
5582 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
5584 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
5587 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
5590 2017-01-17 Alan Modra <amodra@gmail.com>
5593 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
5594 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
5595 symbolic stack limit when pic.
5597 2017-01-16 Martin Sebor <msebor@redhat.com>
5599 PR tree-optimization/78608
5600 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
5602 2017-01-16 Jeff Law <law@redhat.com>
5605 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
5606 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
5607 for several include directories that may be relative to sysroot.
5608 * config/i386/x-mingw32 (gplus_includedir): Define.
5609 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
5610 (native_system_includedir): Likewise.
5611 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
5612 override if TARGET_SYSTEM_ROOT is defined.
5613 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
5615 PR tree-optimization/79090
5616 PR tree-optimization/33562
5617 PR tree-optimization/61912
5618 PR tree-optimization/77485
5619 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
5620 and computed trims into the dump file.
5622 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
5624 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
5626 2017-01-16 Jakub Jelinek <jakub@redhat.com>
5629 * gimplify.c (gimplify_init_constructor): If want_value and
5630 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
5634 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
5635 sequence. Formatting fixes.
5636 (doloop_optimize): Formatting fixes.
5639 * gcc.c (debug_level_greater_than_spec_func): New function.
5640 (static_spec_functions): Add debug-level-gt spec function.
5641 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
5643 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
5644 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
5645 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
5646 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
5647 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
5648 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
5650 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
5652 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
5653 QImode fixups to general and mask registers only.
5655 2017-01-16 Carl Love <cel@us.ibm.com>
5657 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
5658 for built-in functions
5659 vector signed char vec_nabs (vector signed char)
5660 vector signed short vec_nabs (vector signed short)
5661 vector signed int vec_nabs (vector signed int)
5662 vector signed long long vec_nabs (vector signed long long)
5663 vector float vec_nabs (vector float)
5664 vector double vec_nabs (vector double)
5665 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
5667 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
5668 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
5669 * doc/extend.texi: Update the documentation file for the new built-in
5672 2017-01-16 Martin Sebor <msebor@redhat.com>
5674 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
5677 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5679 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
5680 UNSPEC_VSX__XXSPLTD to require special splat handling.
5682 2017-01-16 David Malcolm <dmalcolm@redhat.com>
5685 * system.h: Poison strndup.
5687 2017-01-16 Alan Modra <amodra@gmail.com>
5690 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
5693 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
5695 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
5697 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
5699 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
5700 call recog here. Assert that INSN_CODE (insn) is non-negative.
5702 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
5705 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
5707 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
5708 in the currently scheduled RTL fragment.
5710 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
5712 PR rtl-optimization/78751
5713 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
5716 2017-01-14 Jeff Law <law@redhat.com>
5718 PR tree-optimization/79090
5719 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
5720 variable length stores.
5721 (compute_trims): Delete dead assignment to *trim_tail.
5722 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
5725 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
5727 PR rtl-optimization/78626
5728 PR rtl-optimization/78727
5729 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
5730 of a block, and split such blocks after everything else is finished.
5732 2017-01-14 Alan Modra <amodra@gmail.com>
5735 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
5736 target legitimate_combined_insn.
5737 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
5738 (rs6000_legitimate_combined_insn): New function.
5739 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
5741 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
5742 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
5743 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
5745 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
5747 * doc/frontends.texi (G++ and GCC): Remove references to Java.
5749 2017-01-13 Jeff Law <law@redhat.com>
5751 PR tree-optimization/33562
5752 PR tree-optimization/61912
5753 PR tree-optimization/77485
5754 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
5756 (delete_dead_assignment): Likewise.
5757 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
5758 statement to delete_dead_call and delete_dead_assignment.
5760 2017-01-13 David Malcolm <dmalcolm@redhat.com>
5763 * substring-locations.c (format_warning_va): Strengthen case 1 so
5764 that both endpoints of the substring must be within the format
5765 range for just the substring to be printed.
5767 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
5769 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
5770 * config/i386/i386.c (ix86_target_string): Add missing options
5771 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
5772 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
5773 flags_other and ix86_target_other to flags2_other. Display unknown
5775 (ix86_valid_target_attribute_inner_p): Add missing options and
5776 reorder options by implied ISAs, as in ix86_target_string.
5778 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
5780 * hash-table.h (hash_table::too_empty_p): New function.
5781 (hash_table::expand): Use it.
5782 (hash_table::traverse): Likewise.
5783 (hash_table::empty_slot): Use sizeof (value_type) instead of
5784 sizeof (PTR) to convert bytes to elements. Shrink the table
5785 if the current size is excessive for the current number of
5788 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
5790 * ira-costs.c (record_reg_classes): Break from the inner loop
5791 early once alt_fail is known to be true. Update outer loop
5792 handling accordingly.
5794 2017-01-13 Jeff Law <law@redhat.com>
5796 * tree-ssa-dse.c (decrement_count): New function.
5797 (increment_start_addr, maybe_trim_memstar_call): Likewise.
5798 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
5799 when we know the partially dead statement is a mem* function.
5801 PR tree-optimization/61912
5802 PR tree-optimization/77485
5803 * tree-ssa-dse.c: Include expr.h.
5804 (maybe_trim_constructor_store): New function.
5805 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
5807 PR tree-optimization/33562
5808 PR tree-optimization/61912
5809 PR tree-optimization/77485
5810 * doc/invoke.texi: Document new dse-max-object-size param.
5811 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
5812 * tree-ssa-dse.c: Include params.h.
5813 (dse_store_status): New enum.
5814 (initialize_ao_ref_for_dse): New, partially extracted from
5816 (valid_ao_ref_for_dse, normalize_ref): New.
5817 (setup_live_bytes_from_ref, compute_trims): Likewise.
5818 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
5819 (maybe_trim_partially_dead_store): Likewise.
5820 (maybe_trim_complex_store): Likewise.
5821 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
5822 Track what bytes live from the original store. Return tri-state
5823 for dead, partially dead or live.
5824 (dse_dom_walker): Add constructor, destructor and new private members.
5825 (delete_dead_call, delete_dead_assignment): New extracted from
5827 (dse_optimize_stmt): Make a member of dse_dom_walker.
5828 Use initialize_ao_ref_for_dse.
5830 PR tree-optimization/33562
5831 PR tree-optimization/61912
5832 PR tree-optimization/77485
5833 * sbitmap.h (bitmap_count_bits): Prototype.
5834 (bitmap_clear_range, bitmap_set_range): Likewise.
5835 * sbitmap.c (bitmap_clear_range): New function.
5836 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
5838 2017-01-13 Martin Liska <mliska@suse.cz>
5841 * function.c (set_cfun): Add new argument force.
5842 * function.h (set_cfun): Likewise.
5843 * ipa-inline-transform.c (inline_call): Use the function when
5844 strict alising from is dropped for function we inline to.
5846 2017-01-13 Richard Biener <rguenther@suse.de>
5848 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
5849 for dumping GIMPLE INTEGER_CSTs.
5851 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5853 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
5854 to 201112L since C++17.
5856 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
5859 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
5860 if -fsanitize=kernel-address is present.
5862 2017-01-13 Richard Biener <rguenther@suse.de>
5864 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
5865 as _Literal ( type ) number in case usual suffixes do not
5866 preserve all information.
5868 2017-01-13 Richard Biener <rguenther@suse.de>
5870 PR tree-optimization/77283
5871 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
5872 and ssa-iterators.h.
5873 (is_feasible_trace): Implement a cost model based on joiner
5876 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
5879 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
5880 char or short to __float128/_Float128 directly.
5882 2017-01-12 Martin Sebor <msebor@redhat.com>
5884 to -Wformat-overflow.
5885 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
5886 (min_bytes_remaining): Same.
5887 (get_string_length): Same.
5888 (format_string): Same.
5889 (format_directive): Same.
5891 (pass_sprintf_length::handle_gimple_call): Same.
5893 2017-01-12 Jakub Jelinek <jakub@redhat.com>
5895 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
5896 info.nowrite calls with no lhs that can't throw. Return bool
5897 whether gsi_remove has been called or not.
5898 (pass_sprintf_length::handle_gimple_call): Return bool whether
5899 try_substitute_return_value called gsi_remove. Formatting fix.
5900 (pass_sprintf_length::execute): Don't use gsi_remove if
5901 handle_gimple_call returned true.
5904 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
5905 be removed due to side-effects, don't remove following barrier nor
5906 turn the successor edge into fallthru edge.
5908 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5911 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
5912 element-reversing loads and stores as not swappable.
5914 2017-01-12 Nathan Sidwell <nathan@acm.org>
5915 Nicolai Stange <nicstange@gmail.com>
5917 * combine.c (try_combine): Don't ignore result of overlap checking
5918 loop. Combine overlap & asm check into single loop.
5920 2017-01-12 Richard Biener <rguenther@suse.de>
5922 * tree-pretty-print.c (dump_generic_node): Provide -gimple
5923 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
5925 2017-01-12 Richard Biener <rguenther@suse.de>
5927 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
5928 and TS_TARGET_OPTION directly derive from TS_BASE.
5929 * tree-core.h (tree_optimization_option): Derive from tree_base.
5930 (tree_target_option): Likewise.
5932 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
5934 * config/i386/i386.c (memory_address_length): Increase len
5935 only when rip_relative_addr_p returns false.
5937 2017-01-11 Julia Koval <julia.koval@intel.com>
5939 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
5940 (OPTION_MASK_ISA_SGX_SET): New.
5941 (ix86_handle_option): Handle OPT_msgx.
5942 * config.gcc: Added sgxintrin.h.
5943 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
5944 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
5945 * config/i386/i386.c (ix86_target_string): Add -msgx.
5947 (ix86_option_override_internal): Handle new options.
5948 (ix86_valid_target_attribute_inner_p): Add sgx.
5949 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
5950 * config/i386/i386.opt: Add msgx.
5951 * config/i386/sgxintrin.h: New file.
5952 * config/i386/x86intrin.h: Add sgxintrin.h.
5954 2017-01-11 Jakub Jelinek <jakub@redhat.com>
5957 * fold-const.c (maybe_nonzero_address): Return 1 for function
5959 (tree_single_nonzero_warnv_p): Don't handle function local objects
5963 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
5966 2017-01-11 David Malcolm <dmalcolm@redhat.com>
5969 * opts.c: Include "spellcheck.h"
5970 (struct string_fragment): New struct.
5971 (struct edit_distance_traits<const string_fragment &>): New
5973 (get_closest_sanitizer_option): New function.
5974 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
5976 2017-01-11 Jakub Jelinek <jakub@redhat.com>
5978 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
5980 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
5981 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
5982 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
5983 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
5984 for initial die_offset if dwarf_split_debug_info.
5985 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
5986 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
5988 (output_skeleton_debug_sections): Formatting fix. Use
5989 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
5990 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
5992 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
5994 * config/arm/cortex-a53.md: Add bypasses for
5996 (cortex_a53_r2f): Only use for transfers.
5997 (cortex_a53_f2r): Likewise.
5998 (cortex_a53_r2f_cvt): Add reservation for conversions.
5999 (cortex_a53_f2r_cvt): Likewise.
6001 2017-01-11 Tamar Christina <tamar.christina@arm.com>
6003 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
6004 to all inlined functions, change static to extern.
6006 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
6009 * config/arm/arm.c (legitimize_pic_address): Handle reference to
6011 (arm_assemble_integer): Likewise.
6013 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
6015 * config.gcc: Use new awk script to check CPU, FPU and architecture
6016 parameters for --with-... options.
6017 * config/arm/parsecpu.awk: New file
6018 * config/arm/arm-cpus.in: New file.
6019 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
6021 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
6023 * config/arm/t-arm: Update dependency rules.
6024 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
6025 of processing .def files.
6026 * config/arm/genopt.sh: Deleted.
6027 * config/arm/gentune.sh: Deleted.
6028 * config/arm/arm-cores.def: Deleted.
6029 * config/arm/arm-arches.def: Deleted.
6030 * config/arm/arm-fpus.def: Deleted.
6031 * config/arm/arm-tune.md: Regenerated.
6032 * config/arm/arm-tables.opt: Regenerated.
6033 * config/arm/arm-cpu.h: New generated file.
6034 * config/arm/arm-cpu-data.h: New generated file.
6035 * config/arm/arm-cpu-cdata.h: New generated file.
6037 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
6040 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
6042 (input_varpool_node): Unpack dynamically_initialized bit.
6044 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
6046 PR rtl-optimization/79032
6047 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
6048 the alignment of the adjusted memory reference against that of MODE,
6049 instead of the alignment of the original memory reference.
6051 2017-01-11 Martin Jambor <mjambor@suse.cz>
6053 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
6055 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
6056 decorated functions.
6058 2017-01-11 Richard Biener <rguenther@suse.de>
6060 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
6061 set range/nonnull info for PHI results. Do not set it on
6062 stmts marked for removal.
6064 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
6066 * expr.c (store_field): In the bitfield case, fetch the return value
6067 from the registers before applying a single big-endian adjustment.
6068 Always do a final load for a BLKmode value not larger than a word.
6070 2017-01-10 David Malcolm <dmalcolm@redhat.com>
6073 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
6074 that we correctly handle column numbers greater than
6075 LINE_MAP_MAX_COLUMN_NUMBER.
6077 2017-01-10 Martin Sebor <msebor@redhat.com>
6080 * gimple-ssa-sprintf.c (get_destination_size): Call
6081 {init,fini}object_sizes.
6082 * tree-object-size.c (addr_object_size): Adjust.
6083 (pass_through_call): Adjust.
6084 (pass_object_sizes::execute): Adjust.
6085 * tree-object-size.h (fini_object_sizes): Declare.
6087 2017-01-10 Martin Sebor <msebor@redhat.com>
6089 PR tree-optimization/78775
6090 * builtins.c (get_size_range): Move...
6091 * calls.c: ...to here.
6092 (alloc_max_size): Accept zero argument.
6093 (operand_signed_p): Remove.
6094 (maybe_warn_alloc_args_overflow): Call get_size_range.
6095 * calls.h (get_size_range): Declare.
6097 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
6099 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
6100 from TI's devices.csv file as of September 2016.
6101 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
6103 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
6105 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
6106 * doc/invoke.texi: Likewise.
6107 * doc/md.texi: Likewise.
6108 * doc/objc.texi: Likewise.
6110 2017-01-10 Joshua Conner <joshconner@google.com>
6112 * config/arm/fuchsia-elf.h: New file.
6113 * config/fuchsia.h: New file.
6114 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
6115 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
6117 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
6119 2016-01-10 Richard Biener <rguenther@suse.de>
6121 PR tree-optimization/79034
6122 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
6123 Propagate out degenerate PHIs in the joiner.
6125 2017-01-10 Martin Liska <mliska@suse.cz>
6127 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
6128 (sort_congruence_classes_by_decl_uid): Likewise.
6129 (sort_congruence_class_groups_by_decl_uid): Likewise.
6130 (sem_item_optimizer::merge_classes): Sort class, groups in these
6131 classes and members in the groups by DECL_UID of declarations.
6132 This would make merge operations stable.
6134 2017-01-10 Martin Liska <mliska@suse.cz>
6136 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
6137 usage of m_classes_vec.
6138 (sem_item_optimizer::~sem_item_optimizer): Likewise.
6139 (sem_item_optimizer::get_group_by_hash): Likewise.
6140 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
6141 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
6142 (sem_item_optimizer::verify_classes): Likewise.
6143 (sem_item_optimizer::process_cong_reduction): Likewise.
6144 (sem_item_optimizer::dump_cong_classes): Likewise.
6145 (sem_item_optimizer::merge_classes): Likewise.
6146 * ipa-icf.h (congruence_class_hash): Rename from
6147 congruence_class_group_hash. Remove declaration of m_classes_vec.
6149 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
6151 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
6152 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
6153 * config.gcc: Add avx512vpopcntdqintrin.h.
6154 * config/i386/avx512vpopcntdqintrin.h: New.
6155 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
6156 * config/i386/i386-builtin-types.def: Add new types.
6157 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
6158 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
6159 __builtin_ia32_vpopcountq_v8di_mask): New.
6160 * config/i386/i386-c.c (ix86_target_macros_internal): Define
6161 __AVX512VPOPCNTDQ__.
6162 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
6163 (PTA_AVX512VPOPCNTDQ): Define.
6164 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
6165 TARGET_AVX512VPOPCNTDQ_P): Define.
6166 * config/i386/i386.opt: Add mavx512vpopcntdq.
6167 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
6168 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
6170 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
6173 * predict.def (PRED_CALL): Set to 67.
6175 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
6177 * expr.c (store_field): In the bitfield case, if the value comes from
6178 a function call and is of an aggregate type returned in registers, do
6179 not modify the field mode; extract the value in all cases if the mode
6180 is BLKmode and the size is not larger than a word.
6182 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
6185 * config/i386/cpuid.h: Fix undefined behavior.
6187 2017-01-04 Jeff Law <law@redhat.com>
6189 PR tree-optimization/79007
6190 PR tree-optimization/67955
6191 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
6192 conservative for pt.null when flag_non_call_exceptions is on.
6194 2017-01-09 Jakub Jelinek <jakub@redhat.com>
6196 PR translation/79019
6197 PR translation/79020
6198 * params.def (PARAM_INLINE_MIN_SPEEDUP,
6199 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
6200 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
6202 * config/avr/avr.opt (maccumulate-args): Likewise.
6203 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
6204 * common.opt (freport-bug): Likewise.
6205 * cif-code.def (CIF_FINAL_ERROR): Likewise.
6206 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
6207 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
6208 translatable string.
6209 * config/i386/i386.c (function_value_32): Likewise.
6210 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
6211 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
6213 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
6214 * common/config/msp430/msp430-common.c (msp430_handle_option):
6216 * symtab.c (symtab_node::verify_base): Likewise.
6217 * opts.c (set_debug_level): Likewise.
6218 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
6219 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
6220 missing whitespace to translatable strings.
6221 * config/avr/avr.md (bswapsi2): Fix typo in comment.
6222 * config/sh/superh.h: Likewise.
6223 * config/i386/xopintrin.h: Likewise.
6224 * config/i386/znver1.md: Likewise.
6225 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
6226 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
6227 * double-int.h (struct double_int): Likewise.
6228 * double-int.c (div_and_round_double): Likewise.
6229 * wide-int.cc: Likewise.
6230 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
6231 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
6232 * cfgcleanup.c (crossjumps_occured): Renamed to ...
6233 (crossjumps_occurred): ... this.
6234 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
6237 PR tree-optimization/78899
6238 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
6239 returning bool return struct loop *, NULL for failure and the new
6241 (versionable_outer_loop_p): Don't version outer loop if it has
6242 dont_vectorized bit set.
6243 (tree_if_conversion): When versioning outer loop, ensure
6244 tree_if_conversion is performed also on the inner loop of the
6245 non-vectorizable outer loop copy.
6246 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
6247 LOOP_VECTORIZED in inner loop of the scalar outer loop and
6248 prevent vectorization of it.
6249 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
6250 the outer loop vectorization of the non-scalar version is attempted
6251 before vectorization of the inner loop in scalar version. If
6252 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
6253 vectorization of its inner loop.
6254 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
6255 has 2 inner loops, rename also on edges from bb whose single pred
6256 is outer_loop->header. Fix typo in function comment.
6258 2017-01-09 Martin Sebor <msebor@redhat.com>
6261 * asan.c (asan_emit_stack_protection): Increase local buffer size
6262 to avoid snprintf truncation warning.
6264 2017-01-09 Andrew Pinski <apinski@cavium.com>
6266 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
6267 to reference thunderx2t99 for the tuning structure
6268 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
6270 (thunderx2t99_extra_costs): This.
6271 * config/aarch64/aarch64-tune.md: Regenerate.
6272 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
6273 (vulcan_addrcost_table): This.
6274 (vulcan_regmove_cost): Rename to ...
6275 (thunderx2t99_regmove_cost): This.
6276 (vulcan_vector_cost): Rename to ...
6277 (thunderx2t99_vector_cost): this.
6278 (vulcan_branch_cost): Rename to ...
6279 (thunderx2t99_branch_cost): This.
6280 (vulcan_tunings): Rename to ...
6281 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
6282 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
6284 2017-01-09 Martin Jambor <mjambor@suse.cz>
6288 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
6289 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
6290 (propagate_vr_accross_jump_function): Use the above function for all
6291 value range computations for pass-through jump functions and type
6292 converasion from explicit value range values.
6293 (ipcp_propagate_stage): Do not attempt to deduce types of formal
6294 parameters from TYPE_ARG_TYPES.
6295 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
6296 (ipa_write_node_info): Stream type of the actual argument.
6297 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
6299 2017-01-09 Martin Liska <mliska@suse.cz>
6302 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
6303 (lookup_compiler): Do not show error message with have_E.
6305 2017-01-09 Jakub Jelinek <jakub@redhat.com>
6307 PR tree-optimization/78938
6308 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
6309 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
6310 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
6311 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
6314 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6316 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
6319 2017-01-09 Richard Biener <rguenther@suse.de>
6321 PR tree-optimization/78997
6322 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
6323 name condition properly.
6325 2017-01-09 Richard Biener <rguenther@suse.de>
6328 * dwarf2out.c (is_cxx): New overload with context.
6329 (is_naming_typedef_decl): Use it.
6331 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
6333 * invoke.texi (Option Summary): Correct spacing in option lists
6334 and add line breaks to fix over-long lines.
6336 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
6340 * extend.texi (Common Variable Attributes): Add xref to GCC
6341 Internals manual to explain mode attribute keywords.
6343 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
6346 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
6347 and Preprocessor Options.
6348 (Options for Linking): Document -pthread here....
6349 (RS/6000 and PowerPC Options): ...not here.
6350 (Solaris 2 Options): ...or here.
6351 * doc/cppopts.texi: Document -pthread.
6353 2017-01-08 Martin Sebor <msebor@redhat.com>
6356 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
6357 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
6358 New member functions.
6359 (format_directive): Used them.
6361 (pass_sprintf_length::handle_gimple_call): Same.
6362 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
6363 to avoid truncation for any argument.
6364 (extract_affine_mul): Same.
6365 * tree.c (get_file_function_name): Same.
6367 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
6370 * predict.def (PRED_INDIR_CALL): Set to 86.
6372 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
6374 PR preprocessor/54124
6375 * doc/cppopts.texi: Reformat -d subtable to list the full name
6376 of the options. Add cross-reference to the docs for the general
6377 compiler -d options.
6378 * doc/invoke.texi (Developer Options): Add cross-reference to the
6379 preprocessor-specific -d option documentation.
6381 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
6383 PR preprocessor/13498
6384 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
6385 redudant material, and reflect new command-line options.
6386 (System Headers): Likewise.
6388 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
6390 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
6391 -isystem, and -idirafter. Copy-edit.
6392 * doc/cppopts.texi: Copy-edit. Remove contradiction about
6393 default for -ftrack-macro-expansion. Delete obsolete and
6394 badly-formatted implementation details about -fdebug-cpp output.
6395 * doc/cppwarnopts.texi: Copy-edit.
6397 2017-01-07 David Malcolm <dmalcolm@redhat.com>
6400 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
6401 that the transition from a max line width >= 1<<10 to narrower
6402 lines works correctly.
6404 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
6406 * doc/options.texi (PerFunction): New.
6407 * opt-functions.awk (switch_flags): Map both Optimization and
6408 PerFunction to CL_OPTIMIZATION.
6409 * opth-gen.awk: Test for PerFunction flag along with
6411 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
6412 it only when the latter is present. Skip those that don't in
6413 the hash function generator.
6414 * common.opt (fvar-tracking): Mark as PerFunction instead of
6416 (fvar-tracking-assignments): Likewise.
6417 (fvar-tracking-assignments-toggle): Likewise.
6418 (fvar-tracking-uninit): Likewise.
6420 2017-01-07 Jakub Jelinek <jakub@redhat.com>
6422 PR translation/79018
6423 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
6426 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
6429 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
6430 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
6431 TARGET_LONG_JUMP_TABLE_OFFSETS.
6432 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
6433 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
6434 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
6435 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
6436 * config/m68k/m68k.md (tablejump expander): Likewise.
6437 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
6438 TARGET_LONG_JUMP_TABLE_OFFSETS.
6439 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
6440 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
6442 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
6443 David Holsgrove <david.holsgrove@xilinx.com>
6445 * common/config/microblaze/microblaze-common.c
6446 (TARGET_EXCEPT_UNWIND_INFO): Remove.
6447 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
6449 * config/microblaze/microblaze.c (microblaze_must_save_register)
6450 (microblaze_expand_epilogue, microblaze_return_addr): Handle
6452 (microblaze_eh_return): New function.
6453 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
6454 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
6455 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
6456 * config/microblaze/microblaze.md (eh_return): New pattern.
6458 2017-01-06 Jakub Jelinek <jakub@redhat.com>
6460 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
6461 GCC_DIAGNOSTIC_STRINGIFY): Define.
6463 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
6465 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
6467 * config/arm/arm.md (<mcrr>): New.
6469 * config/arm/arm.c (arm_arch5te): New.
6470 (arm_option_override): Set arm_arch5te.
6471 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
6473 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
6474 (arm_mcrr_qualifiers): ... this. New.
6475 (MRRC_QUALIFIERS): Define to...
6476 (arm_mrrc_qualifiers): ... this. New.
6477 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
6479 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
6480 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
6481 (MRRCI, mrrc, MRRC): New.
6482 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
6483 VUNSPEC_MRRC2): New.
6485 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
6487 * config/arm/arm.md (<mcr>): New.
6489 * config/arm/arm.c (arm_coproc_builtin_available): Add
6490 support for mcr, mrc, mcr2 and mrc2.
6491 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
6492 (arm_mcr_qualifiers): ... this. New.
6493 (MRC_QUALIFIERS): Define to ...
6494 (arm_mrc_qualifiers): ... this. New.
6495 (MCR_QUALIFIERS): Define to ...
6496 (arm_mcr_qualifiers): ... this. New.
6497 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
6499 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
6500 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
6501 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
6504 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
6506 * config/arm/arm.md (*ldc): New.
6510 * config/arm/arm.c (arm_coproc_builtin_available): Add
6511 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
6512 (arm_coproc_ldc_stc_legitimate_address): New.
6513 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
6514 'qualifier_const_pointer'.
6515 (LDC_QUALIFIERS): Define to...
6516 (arm_ldc_qualifiers): ... this. New.
6517 (STC_QUALIFIERS): Define to...
6518 (arm_stc_qualifiers): ... this. New.
6519 * config/arm/arm-protos.h
6520 (arm_coproc_ldc_stc_legitimate_address): New.
6521 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
6522 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
6523 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
6524 stc2, stcl, stc2l): New.
6525 * config/arm/constraints.md (Uz): New.
6526 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
6527 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
6528 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
6529 VUNSPEC_STC2L): New.
6531 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
6533 * config/arm/arm.md (<cdp>): New.
6534 * config/arm/arm.c (neon_const_bounds): Rename this ...
6535 (arm_const_bounds): ... this.
6536 (arm_coproc_builtin_available): New.
6537 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
6538 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
6539 (CDP_QUALIFIERS): Define to...
6540 (arm_cdp_qualifiers): ... this. New.
6542 (arm_expand_builtin_args): Add case for 6 arguments.
6543 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
6544 (arm_const_bounds): ... this.
6545 (arm_coproc_builtin_available): New.
6546 * config/arm/arm_acle.h (__arm_cdp): New.
6548 * config/arm/arm_acle_builtins.def (cdp): New.
6550 * config/arm/iterators.md (CDPI,CDP,cdp): New.
6551 * config/arm/neon.md: Rename all 'neon_const_bounds' to
6553 * config/arm/types.md (coproc): New.
6554 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
6555 * gcc/doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
6556 * gcc/doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
6557 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
6559 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
6561 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
6562 (UBINOP_QUALIFIERS): New.
6564 (acle_builtin_data): New. Change comment.
6565 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
6566 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
6567 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
6568 arm_acle_builtins.def.
6569 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
6570 (arm_init_acle_builtins): New.
6571 (CRC32_BUILTIN): Remove.
6572 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
6573 crc32cb, crc32ch and crc32cw.
6574 (arm_init_crc32_builtins): Remove.
6575 (arm_init_builtins): Use arm_init_acle_builtins rather
6576 than arm_init_crc32_builtins.
6577 (arm_expand_acle_builtin): New.
6578 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
6579 * config/arm/arm_acle_builtins.def: New.
6581 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
6583 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
6584 (arm_builtin_datum): ... this.
6585 (arm_init_neon_builtin): Rename to ...
6586 (arm_init_builtin): ... this. Add a new parameters PREFIX
6587 and USE_SIG_IN_NAME.
6588 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
6589 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
6590 'arm_builtin_datum'.
6591 (arm_init_vfp_builtins): Likewise.
6592 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
6593 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
6594 (arm_expand_neon_args): Rename to ...
6595 (arm_expand_builtin_args): ... this. Rename builtin_arg
6596 enum values and differentiate between ARG_BUILTIN_MEMORY
6597 and ARG_BUILTIN_NEON_MEMORY.
6598 (arm_expand_neon_builtin_1): Rename to ...
6599 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
6600 values, arm_expand_builtin_args and add bool parameter NEON.
6601 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
6602 (arm_expand_vfp_builtin): Likewise.
6603 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
6605 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
6608 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
6609 * predict.c (tree_estimate_probability_bb): Reverse direction of
6610 polymorphic call predictor.
6612 2017-01-06 David Malcolm <dmalcolm@redhat.com>
6614 * passes.c (execute_one_pass): Split out pass-skipping logic into...
6615 (determine_pass_name_match): ...this new function and...
6616 (should_skip_pass_p): ...this new function.
6618 2017-01-06 Nathan Sidwell <nathan@acm.org>
6620 * ipa-visibility.c (function_and_variable_visibility): Reformat
6621 comments and long lines. Remove extrneous if.
6622 * symtab.c (symtab_node::make_decl_local): Fix code format.
6623 (symtab_node::set_section_for_node): Fix comment typo.
6625 2017-01-06 Martin Liska <mliska@suse.cz>
6628 * lra-constraints.c: Rename invariant to lra_invariant.
6629 * predict.c (set_even_probabilities): Initialize e to NULL.
6631 2017-01-05 Martin Sebor <msebor@redhat.com>
6633 PR tree-optimization/78910
6634 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
6635 (format_integer): Correct off-by-one error in the handling
6636 of precision with negative numbers in signed conversions..
6638 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
6640 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
6642 2017-01-05 Jakub Jelinek <jakub@redhat.com>
6644 PR tree-optimization/71016
6645 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
6646 factor_out_conditional_conversion. Formatting fix.
6647 (factor_out_conditional_conversion): Add cond_stmt argument.
6648 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
6649 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
6652 2017-01-05 David Malcolm <dmalcolm@redhat.com>
6654 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
6655 read-rtl-function.o, and selftest-rtl.o.
6656 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
6657 (selftest::aarch64_test_loading_full_dump): New function.
6658 (selftest::aarch64_run_selftests): New function.
6659 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
6660 selftest::aarch64_run_selftests.
6661 * config/i386/i386.c
6662 (selftest::ix86_test_loading_dump_fragment_1): New function.
6663 (selftest::ix86_test_loading_call_insn): New function.
6664 (selftest::ix86_test_loading_full_dump): New function.
6665 (selftest::ix86_test_loading_unspec): New function.
6666 (selftest::ix86_run_selftests): Call the new functions.
6667 * emit-rtl.c (maybe_set_max_label_num): New function.
6668 * emit-rtl.h (maybe_set_max_label_num): New decl.
6669 * function.c (instantiate_decls): Guard call to
6670 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
6671 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
6673 * gensupport.c (gen_reader::gen_reader): Pass "false"
6674 for new "compact" param of rtx_reader.
6675 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
6676 rather than an empty string for NULL strings.
6677 * read-md.c: Potentially include config.h rather than bconfig.h.
6678 Wrap include of errors.h with #ifdef GENERATOR_FILE.
6679 (have_error): New global, copied from errors.c.
6680 (md_reader::read_name): Rename to...
6681 (md_reader::read_name_1): ...this, adding "out_loc" param,
6682 and converting "missing name or number" to returning false, rather
6684 (md_reader::read_name): Reimplement in terms of read_name_1.
6685 (md_reader::read_name_or_nil): New function.
6686 (md_reader::read_string): Handle "(nil)" by returning NULL.
6687 (md_reader::md_reader): Add new param "compact".
6688 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
6689 (md_reader::read_file): New method.
6690 * read-md.h (md_reader::md_reader): Add new param "compact".
6691 (md_reader::read_file): New method.
6692 (md_reader::is_compact): New accessor.
6693 (md_reader::read_name): Convert return type from void to file_location.
6694 (md_reader::read_name_or_nil): New decl.
6695 (md_reader::read_name_1): New decl.
6696 (md_reader::m_compact): New field.
6697 (noop_reader::noop_reader): Pass "false" for new "compact" param
6699 (rtx_reader::rtx_reader): Add new "compact" param.
6700 (rtx_reader::read_rtx_operand): Make virtual and convert return
6701 type from void to rtx.
6702 (rtx_reader::read_until): New decl.
6703 (rtx_reader::handle_any_trailing_information): New virtual function.
6704 (rtx_reader::postprocess): New virtual function.
6705 (rtx_reader::finalize_string): New virtual function.
6706 (rtx_reader::m_in_call_function_usage): New field.
6707 (rtx_reader::m_reuse_rtx_by_id): New field.
6708 * read-rtl-function.c: New file.
6709 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
6710 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
6711 (selftest::verify_three_block_rtl_cfg): New decl.
6712 * read-rtl-function.h: New file.
6713 * read-rtl.c: Potentially include config.h rather than bconfig.h.
6714 For host, include function.h, memmodel.h, and emit-rtl.h.
6715 (one_time_initialization): New function.
6716 (struct compact_insn_name): New struct.
6717 (compact_insn_names): New array.
6718 (find_code): Handle insn codes in compact dumps.
6719 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
6720 (bind_subst_iter_and_attr): Likewise.
6721 (add_condition_to_string): Likewise.
6722 (add_condition_to_rtx): Likewise.
6723 (apply_attribute_uses): Likewise.
6724 (add_current_iterators): Likewise.
6725 (apply_iterators): Likewise.
6726 (initialize_iterators): Guard usage of apply_subst_iterator with
6727 #ifdef GENERATOR_FILE.
6728 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
6729 (md_reader::read_mapping): Likewise.
6730 (add_define_attr_for_define_subst): Likewise.
6731 (add_define_subst_attr): Likewise.
6732 (read_subst_mapping): Likewise.
6733 (check_code_iterator): Likewise.
6734 (rtx_reader::read_rtx): Likewise. Move one-time initialization
6736 (one_time_initialization): New function.
6737 (rtx_reader::read_until): New method.
6738 (read_flags): New function.
6739 (parse_reg_note_name): New function.
6740 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
6741 Handle reuse_rtx ids.
6742 Wrap iterator lookup within #ifdef GENERATOR_FILE.
6743 Add parsing support for RTL dumps, mirroring the special-cases in
6744 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
6745 values, and calling handle_any_trailing_information.
6746 (rtx_reader::read_rtx_operand): Convert return type from void
6747 to rtx, returning return_rtx. Handle case 'e'. Call
6748 finalize_string on XSTR and XTMPL fields.
6749 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
6750 "(nil)" values were omitted. Call the postprocess vfunc on the
6752 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
6753 class ctor. Initialize m_in_call_function_usage. Call
6754 one_time_initialization.
6755 * rtl-tests.c (selftest::test_uncond_jump): Call
6756 set_new_first_and_last_insn.
6757 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
6758 * selftest-rtl.c: New file.
6759 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
6760 (selftest::get_insn_by_uid): New decl.
6761 * selftest-run-tests.c (selftest::run_tests): Call
6762 read_rtl_function_c_tests.
6763 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
6764 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
6767 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
6769 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
6770 operands in a special way. Assert that pos+len <= mode precision.
6772 2017-01-05 Jakub Jelinek <jakub@redhat.com>
6774 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
6775 3 argument Alias with unlimited for the negative form.
6776 (fno-vect-cost-model): Removed.
6778 2017-01-05 Martin Liska <mliska@suse.cz>
6780 * hsa-gen.c (gen_hsa_divmod): New function.
6781 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
6783 2017-01-05 Martin Liska <mliska@suse.cz>
6786 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
6789 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6791 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
6792 small constant length operands.
6794 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6796 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
6797 between loop iterations.
6799 2017-01-05 Martin Liska <mliska@suse.cz>
6802 * gimplify.c (gimplify_decl_expr): Compare to
6803 asan_poisoned_variables instread of checking flags.
6804 (gimplify_target_expr): Likewise.
6805 (gimplify_expr): Likewise.
6806 (gimplify_function_tree): Conditionally initialize
6807 asan_poisoned_variables.
6809 2017-01-04 Jeff Law <law@redhat.com>
6811 PR tree-optimizatin/78812
6812 * rtl.h (contains_mem_rtx_p): Prototype.
6813 * ifcvt.c (containts_mem_rtx_p): Move from here to...
6814 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
6815 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
6816 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
6817 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
6819 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
6821 * input.c (assert_char_at_range): Default-initialize actual_range.
6823 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
6825 * df-scan.c (df_ref_create_structure): Make regno unsigned,
6826 to match the caller.
6828 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
6830 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
6831 insns after final jump in test to emit dummy move.
6833 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
6835 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
6836 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
6838 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
6840 * multiple_target.c (create_dispatcher_calls): Init e_next.
6841 * tree-ssa-loop-split.c (split_loop): Init border.
6842 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
6845 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
6850 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
6851 (altivec_register_operand): Do not return true if the operand
6852 contains a SUBREG mixing SImode and SFmode.
6853 (vsx_register_operand): Likewise.
6854 (vsx_reg_sfsubreg_ok): New predicate.
6855 (vfloat_operand): Do not return true if the operand contains a
6856 SUBREG mixing SImode and SFmode.
6857 (vint_operand): Likewise.
6858 (vlogical_operand): Likewise.
6859 (gpc_reg_operand): Likewise.
6860 (int_reg_operand): Likewise.
6861 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
6862 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
6863 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
6865 (rs6000_emit_move_si_sf_subreg): New helper function.
6866 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
6867 fixup SUBREGs involving SImode and SFmode.
6868 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
6869 numbers for the new peephole2 optimization.
6870 (peephole2 for SFmode unions): New peephole2 to optimize cases in
6871 the GLIBC math library that do AND/IOR/XOR operations on single
6872 precision floating point.
6873 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
6874 target macros to say whether we need to avoid SUBREGs mixing
6876 (TARGET_ALLOW_SF_SUBREG): Likewise.
6877 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
6878 (UNSPEC_SI_FROM_SF): Likewise.
6879 (iorxor): Change spacing.
6880 (and_ior_xor): New iterator for AND, IOR, and XOR.
6881 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
6882 (movdi_from_sf_zero_ext): Likewise.
6883 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
6884 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
6885 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
6886 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
6887 (fms<mode>4): Likewise.
6888 (fnma<mode>4): Likewise.
6889 (fnms<mode>4): Likewise.
6890 (nfma<mode>4): Likewise.
6891 (nfms<mode>4): Likewise.
6893 2017-01-04 Marek Polacek <polacek@redhat.com>
6896 * doc/invoke.texi: Document -Wpointer-compare.
6898 2017-01-04 Jakub Jelinek <jakub@redhat.com>
6900 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
6903 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
6904 descriptions for -gdwarf-5 and emit them as uleb128 instead of
6907 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
6910 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
6911 documentation of the powerpc_popcntb_ok attribute.
6912 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
6913 code to issue warning messages if a requested CPU configuration is
6914 not supported by the binary (assembler and loader) toolchain.
6915 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
6916 made to define a built-in function that has been disabled.
6917 (paired_init_builtins): Add assertion to prevent ICE if attempt is
6918 made to define a built-in function that has been disabled.
6919 (altivec_init_builtins): Add comment explaining why definition
6920 of the DST built-in functions is not preceded by an assertion
6921 check. Add assertions to prevent ICE if attempts are made to
6922 define an altivec predicate or an abs* built-in function that has
6924 (htm_init_builtins): Add comment explaining why definition of the
6925 htm built-in functions is not preceded by an assertion check.
6927 2017-01-04 Jeff Law <law@redhat.com>
6929 PR tree-optimizatin/67955
6930 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
6931 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
6932 the points-to solution does not include pt_null. Use DECL_PT_UID
6935 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
6937 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
6938 Use gen_int_mode instead of gen_lopwart for const_int operands.
6940 2017-01-04 Jakub Jelinek <jakub@redhat.com>
6942 PR tree-optimization/71563
6943 * match.pd: Simplify X << Y into X if Y is known to be 0 or
6944 out of range value - has low bits known to be zero.
6946 2017-01-04 Alan Modra <amodra@gmail.com>
6948 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
6949 * configure: Regenerate.
6950 * config.in: Regenerate.
6952 2017-01-04 Jakub Jelinek <jakub@redhat.com>
6955 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
6956 a substring of the message, but strcmp with the whole message. Ifdef
6957 ENABLE_NLS, translate the message first using dgettext.
6959 2017-01-03 Jeff Law <law@redhat.com>
6961 PR tree-optimizatin/78856
6962 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
6963 (mark_threaded_blocks): Remove code to truncate thread paths that
6964 cross multiple loop headers. Instead invalidate the cached loop
6965 iteration information and handle case of a thread path walking
6966 into an irreducible region.
6968 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
6971 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
6972 assertions. Add support for doing the signbit if the IEEE 128-bit
6973 floating point value is in a GPR.
6974 * config/rs6000/rs6000.md (Fsignbit): Delete.
6975 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
6976 Update the length attribute if the value is in a GPR.
6977 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
6978 the sign or zero extension instruction, since the value is always 0/1.
6979 (signbit<mode>2_dm2): Delete using <Fsignbit>.
6982 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
6983 extracting SImode to a GPR register so that we can generate a
6984 store, limit the vector to be in a traditional Altivec register
6985 for the vextuwrx instruction.
6987 2017-01-03 Ian Lance Taylor <iant@google.com>
6989 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
6991 2017-01-03 Martin Sebor <msebor@redhat.com>
6993 PR tree-optimization/78696
6994 * gimple-ssa-sprintf.c (format_floating): Correct handling of
6995 precision. Use MPFR for %f for greater fidelity. Correct handling
6997 (pass_sprintf_length::compute_format_length): Set width and precision
6998 specified by asrerisk to void_node for vararg functions.
6999 (try_substitute_return_value): Adjust dump output.
7001 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
7003 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
7005 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
7007 * doc/invoke.texi (SPARC options): Document -mlra as the default.
7008 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
7009 -mlra/-mno-lra was passed to the compiler.
7011 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
7013 PR rtl-optimization/65618
7014 * emit-rtl.c (try_split): Move initialization of "before" and
7015 "after" to just before the call to emit_insn_after_setloc.
7017 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
7019 * doc/md.texi (Standard Names): Remove reference to Java frontend.
7021 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
7023 * dwarf2out.c (gen_enumeration_type_die): When
7024 -gno-strict-dwarf, add a DW_AT_encoding attribute.
7026 2017-01-03 Jakub Jelinek <jakub@redhat.com>
7028 PR tree-optimization/78965
7029 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
7030 Change first argument from const call_info & to call_info &. For %n
7031 set info.nowrite to false.
7034 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
7035 possibly throwing calls.
7037 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
7038 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
7039 and fns handling, rather than in a separate case SSA_NAME.
7041 2017-01-02 Jeff Law <law@redhat.com>
7043 * config/darwin-driver.c (darwin_driver_init): Const-correctness
7044 fixes for first_period and second_period variables.
7046 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
7049 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
7050 (*insvqi_1): New insn pattern.
7051 (*insvqi_1_mem_rex64): Ditto.
7053 (*insvqi_3): Rename from *insvqi.
7055 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
7057 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
7059 * doc/cfg.texi (Edges): Remove reference to Java.
7060 (Maintaining the CFG): Ditto.
7062 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
7065 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
7066 transparent aliases.
7068 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
7071 * predict.def (PRED_CALL): Update hitrate.
7072 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
7073 * predict.c (tree_estimate_probability_bb): Split CALL predictor
7074 into direct/indirect/polymorphic variants.
7076 2017-01-01 Jakub Jelinek <jakub@redhat.com>
7078 Update copyright years.
7080 * gcc.c (process_command): Update copyright notice dates.
7081 * gcov-dump.c (print_version): Ditto.
7082 * gcov.c (print_version): Ditto.
7083 * gcov-tool.c (print_version): Ditto.
7084 * gengtype.c (create_file): Ditto.
7085 * doc/cpp.texi: Bump @copying's copyright year.
7086 * doc/cppinternals.texi: Ditto.
7087 * doc/gcc.texi: Ditto.
7088 * doc/gccint.texi: Ditto.
7089 * doc/gcov.texi: Ditto.
7090 * doc/install.texi: Ditto.
7091 * doc/invoke.texi: Ditto.
7093 Copyright (C) 2017 Free Software Foundation, Inc.
7095 Copying and distribution of this file, with or without modification,
7096 are permitted in any medium without royalty provided the copyright
7097 notice and this notice are preserved.