1 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3 PR tree-optimization/80158
4 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
5 Handle possible future case of more than one alternate
7 (replace_rhs_if_not_dup): Likewise.
8 (replace_one_candidate): Likewise.
10 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
12 PR rtl-optimization/80193
13 * ira.c (ira): Do not check allocation for LRA.
15 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
17 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
18 (nvptx_output_simt_exit): Declare.
19 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
20 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
21 (init_softstack_frame): Move initialization of crtl->is_leaf to...
22 (nvptx_declare_function_name): ...here. Emit declaration of local
23 memory space buffer for omp_simt_enter insn.
24 (nvptx_output_unisimt_switch): New.
25 (nvptx_output_softstack_switch): New.
26 (nvptx_output_simt_enter): New.
27 (nvptx_output_simt_exit): New.
28 * config/nvptx/nvptx.h (struct machine_function): New fields
29 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
30 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
31 (UNSPECV_SIMT_EXIT): Ditto.
32 (omp_simt_enter_insn): New insn.
33 (omp_simt_enter): New expansion.
34 (omp_simt_exit): New insn.
35 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
37 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
38 (expand_GOMP_SIMT_ENTER_ALLOC): New.
39 (expand_GOMP_SIMT_EXIT): New.
40 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
41 (GOMP_SIMT_ENTER_ALLOC): Ditto.
42 (GOMP_SIMT_EXIT): Ditto.
43 * target-insns.def (omp_simt_enter): New insn.
44 (omp_simt_exit): Ditto.
45 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
47 (lower_rec_simd_input_clauses): Implement SIMT privatization.
48 (lower_rec_input_clauses): Likewise.
49 (lower_lastprivate_clauses): Handle SIMT privatization.
51 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
52 (ompdevlow_adjust_simt_enter): New.
53 (find_simtpriv_var_op): New.
54 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
55 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
57 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
58 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
59 (copy_decl_for_dup_finish): Ditto.
61 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
63 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
66 * config/i386/i386.c (ix86_option_override_internal): Always
67 allow -mpreferred-stack-boundary=3 for 64-bit targets.
69 2017-03-28 Bin Cheng <bin.cheng@arm.com>
71 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
73 2017-03-28 Bin Cheng <bin.cheng@arm.com>
75 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
76 mark new edge's irreducible flag accordign to it.
77 (vect_do_peeling): Check loop preheader edge's irreducible flag
78 and pass it to function slpeel_add_loop_guard.
80 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
82 PR tree-optimization/80218
83 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
84 Update block frequencies and counts.
86 2017-03-28 Richard Biener <rguenther@suse.de>
88 PR tree-optimization/78644
89 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
90 of a simplification result we may not use it at all.
92 2017-03-28 Richard Biener <rguenther@suse.de>
95 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
96 without arguments, generate default definition of a SSA name.
98 2017-03-28 Richard Biener <rguenther@suse.de>
101 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
102 TYPE_REF_CAN_ALIAS_ALL references.
103 * fold-const.c (fold_indirect_ref_1): Likewise.
105 2017-03-28 Martin Liska <mliska@suse.cz>
108 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
109 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
111 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
112 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
114 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
115 (EXTRA_SPECS): Define.
116 (SUBTARGET_EXTRA_SPECS): Likewise.
117 (SUBTARGET_CPP_SPEC): Likewise.
118 * config/arc/elf.h (EXTRA_SPECS): Renamed to
119 SUBTARGET_EXTRA_SPECS.
120 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
122 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
124 * config/arc/simdext.md (vst64_insn): Update pattern.
125 (vld32wh_insn): Likewise.
126 (vld32wl_insn): Likewise.
127 (vld64_insn): Likewise.
128 (vld32_insn): Likewise.
130 2017-03-28 Marek Polacek <polacek@redhat.com>
133 * fold-const.c (fold_comparison): Use protected_set_expr_location
134 instead of SET_EXPR_LOCATION.
136 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
138 * tree.c (add_expr): Avoid name lookup warning.
140 2017-03-27 Jeff Law <law@redhat.com>
142 PR tree-optimization/80216
143 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
144 function name. Limit recursion depth.
145 (record_temporary_equivalences): Corresponding changes.
147 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
149 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
152 2017-03-27 Jakub Jelinek <jakub@redhat.com>
155 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
157 * cfgcleanup.c (reg_note_cfa_p): New array.
158 (insns_have_identical_cfa_notes): New function.
159 (old_insns_match_p): Don't cross-jump in between /f
160 and non-/f instructions. If both i1 and i2 are frame related,
161 verify all CFA notes, their order and content.
163 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
166 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
167 HImode and SImode with zero extend to DImode to one insn.
168 (bswap<mode>2_extenddi): Likewise.
169 (bswapsi2_extenddi): Likewise.
170 (bswaphi2_extendsi): Likewise.
171 (bswaphi2): Combine bswap HImode and SImode into one insn.
172 Separate memory insns from swapping register.
173 (bswapsi2): Likewise.
174 (bswap<mode>2): Likewise.
175 (bswaphi2_internal): Delete, no longer used.
176 (bswapsi2_internal): Likewise.
177 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
178 store, and gpr<-gpr swap insns.
179 (bswap<mode>2_store): Likewise.
180 (bswaphi2_reg): Register only splitter, combine with the splitter.
181 (bswaphi2 splitter): Likewise.
182 (bswapsi2_reg): Likewise.
183 (bswapsi2 splitter): Likewise.
184 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
185 the insns into load, store, and register/register insns.
186 (bswapdi2_ldbrx): Likewise.
187 (bswapdi2_load): Likewise.
188 (bswapdi2_store): Likewise.
189 (bswapdi2_reg): Likewise.
191 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
193 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
194 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
196 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
199 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
201 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
202 special handling for target option conflicts between dform
203 options (-mpower9-dform, -mpower9-dform-vector,
204 -mpower9-dform-scalar) and -mno-direct-move.
206 2017-03-27 Richard Biener <rguenther@suse.de>
208 PR tree-optimization/80181
209 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
211 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
213 * config/arc/predicates.md (move_double_src_operand): Replace the
214 call to move_double_src_operand with a call to address_operand.
216 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
218 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
219 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
220 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
222 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
224 * config/arc/predicates.md (long_immediate_loadstore_operand):
225 Consider scaled addresses cases.
227 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
229 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
230 restored when in interrupt.
231 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
232 doesn't have delay slot.
234 2017-03-27 Richard Biener <rguenther@suse.de>
237 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
238 inlined thunk clones.
240 2017-03-27 Jakub Jelinek <jakub@redhat.com>
243 * asan.c (instrument_derefs): Copy over last operand from
244 original COMPONENT_REF to the new COMPONENT_REF with
245 DECL_BIT_FIELD_REPRESENTATIVE.
246 * ubsan.c (instrument_object_size): Likewise.
248 2017-03-27 Richard Biener <rguenther@suse.de>
250 PR tree-optimization/80170
251 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
252 sure DR/SCEV didnt fold in constants we do not see when looking
253 at the reference base alignment.
255 2017-03-27 Richard Biener <rguenther@suse.de>
258 * gimple-fold.c (fold_ctor_reference): Properly guard against
259 NULL return value from canonicalize_constructor_val.
261 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
264 * config/i386/i386.c (ix86_expand_builtin)
265 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
266 flags reg setting and flags reg using instructions.
267 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
268 clobbering instructions to zero extend op2.
270 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
272 * doc/install.texi (Configuration) <--with-aix-soname>:
273 Update link to AIX ld.
275 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
277 PR rtl-optimization/80160
278 PR rtl-optimization/80159
279 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
280 reg_alternate_class into account.
282 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
285 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
286 to consider in curr_insn_transform.
288 2017-03-24 Jakub Jelinek <jakub@redhat.com>
290 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
291 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
294 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
296 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
297 argument to the overloaded builtin variants. Use the new flag to
298 deprecate certain builtin variants.
299 * config/s390/s390-builtin-types.def: Add new builtin types.
300 * config/s390/s390-builtins.h: Support new flags field for
302 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
303 (s390_macro_to_expand): Enable vector float data type.
304 (s390_cpu_cpp_builtins_internal): Indicate support of the new
305 builtins by incrementing the __VEC__ version number.
306 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
308 (s390_resolve_overloaded_builtin): Emit error messages depending
309 on the builtin flags.
310 * config/s390/s390.c (s390_expand_builtin): Support additional
311 flags argument. Change error message to match the messages
313 * config/s390/s390.md: New UNSPEC_* constants.
314 (op_type): Add new instruction types.
315 * config/s390/vecintrin.h: Add new builtins and test data class
317 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
318 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
319 (VEC_INEXACT, VEC_NOINEXACT): New constants.
320 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
321 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
322 ("vec_mergel<mode>"): V_HW -> VEC_HW.
324 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
325 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
326 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
327 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
329 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
330 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
331 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
332 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
334 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
335 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
336 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
337 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
338 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
339 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
340 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
342 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
343 ("vec_scatter_element<V_HW_4:mode>_DI")
344 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
345 ("vec_fpint<mode>", "vflls")
346 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
347 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
348 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
349 ("*vec_cmphe<mode>_cc"): ... these.
351 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
352 mode constant instead of magic value.
354 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
356 * config/s390/s390.c (s390_expand_vec_compare): Support other
357 vector floating point modes than just V2DF.
358 (s390_expand_vcond): Likewise.
359 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
360 (s390_cannot_change_mode_class): Prevent mode changes between TF
361 and V1TF in vector registers.
362 * config/s390/s390.md (DF, SF): New mode attributes.
363 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
364 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
365 SFmode support for VRs.
366 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
368 (VFT, VF_HW): New mode iterators.
369 (vw, sdx): New mode attributes.
370 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
371 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
372 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
373 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
374 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
375 also the new vector floating point modes. Renaming to ...
377 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
378 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
379 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
380 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
381 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
382 ("vec_unordered<mode>"): ... these.
384 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
385 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
386 ("*vec_extendv2df"): New insn definitions.
388 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
390 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
391 ("mulditi3_2", "*muldi3_sign"): New patterns.
392 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
393 rename the pattern definition.
395 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
397 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
399 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
401 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
403 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
404 instruction if possible.
405 * config/s390/vector.md (vec_halfnumelts): New mode
407 ("*vec_vllezlf<mode>"): New pattern.
409 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
411 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
412 ("popcountv4si2", "popcountv2di2"): Rename to ...
413 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
414 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
416 ("popcount<mode>2_vxe"): New pattern.
418 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
420 * common/config/s390/s390-common.c (processor_flags_table): Add
422 * config.gcc: Add arch12.
423 * config/s390/driver-native.c (s390_host_detect_local_cpu):
424 Default to arch12 for unknown CPU model numbers.
425 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
426 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
427 PROCESSOR_max sanity check.
428 * config/s390/s390-opts.h (enum processor_type): Add
430 * config/s390/s390.c (processor_table): Add arch12.
431 (s390_expand_builtin): Add check for B_VXE flag.
432 (s390_issue_rate): Add PROCESSOR_ARCH12.
433 (s390_get_sched_attrmask): Likewise.
434 (s390_get_unit_mask): Likewise.
435 (s390_sched_score): Enable z13 scheduling for arch12.
436 (s390_sched_reorder): Likewise.
437 (s390_sched_variable_issue): Likewise.
438 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
440 (s390_tune_attr): Use z13 scheduling also for arch12.
441 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
442 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
443 (TARGET_VXE_P): New macros.
444 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
445 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
446 * config/s390/s390.opt: Add arch12 as processor_type.
448 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
450 * config/s390/s390.md
451 ("fixuns_truncdddi2", "fixuns_trunctddi2")
452 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
453 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
455 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
456 Rename expanders to ...
458 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
459 ("fixuns_truncdddi2_emu"): ... these.
461 ("fixuns_trunc<mode>si2_emu"): New expander.
463 ("*fixuns_truncdfdi2_z13"): Rename to ...
464 ("*fixuns_truncdfdi2_vx"): ... this.
466 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
468 * config/s390/2964.md: Remove the single element vector compare
469 instructions which are no longer used.
470 * config/s390/s390.c (s390_select_ccmode): Remove handling of
472 (s390_canonicalize_comparison): Remove handling of DFmode
474 (s390_expand_vec_compare_scalar): Remove function.
475 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
476 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
478 ("*cmp<mode>_ccs"): Add wfcdb instruction.
480 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
482 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
484 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
485 will anyway by matched by mov<mode>_64dfp.
487 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
489 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
490 vlef/vstef. Add missing operand to vleif.
492 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
494 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
495 pair for all vector types with 64 bit elements.
496 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
497 * config/s390/vector.md (V_HW_64): ... here.
498 (V_128_NOSINGLE): New mode iterator.
499 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
500 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
501 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
502 ("*vec_load_pairv2di"): Change to ...
503 ("*vec_load_pair<mode>"): ... this one.
505 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
507 * config/s390/constraints.md: Add comments.
508 (jKK): Reject element sizes > 8 bytes.
509 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
511 * config/s390/s390.md: Add the s_operand checks formerly in
512 s390_split_ok_p to various splitters where they are still
514 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
515 for 128 bit vectors. Plus two splitters.
517 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
519 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
522 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
525 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
526 error if the boundary argument is not constant.
528 2017-03-24 Jakub Jelinek <jakub@redhat.com>
530 PR rtl-optimization/80112
531 * loop-doloop.c (doloop_condition_get): Don't check condition
532 if cmp isn't SET with IF_THEN_ELSE src.
534 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
536 PR tree-optimization/80158
537 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
538 replacing a candidate statement, also replace it for the
539 candidate's alternate interpretation.
540 (replace_rhs_if_not_dup): Likewise.
541 (replace_one_candidate): Likewise.
543 2017-03-24 Richard Biener <rguenther@suse.de>
545 PR tree-optimization/80167
546 * graphite-isl-ast-to-gimple.c
547 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
549 (translate_isl_ast_to_gimple::get_rename): Likewise.
551 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
553 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
554 handling of certain combinations of target options, including the
555 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
556 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
558 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
561 * config/arm/arm.md (*load_multiple): Add reload_completed to
564 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
565 Richard Biener <rguenth@suse.de>
567 PR tree-optimization/79908
568 PR tree-optimization/80136
569 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
570 been cast away, gimplify_and_add suffices.
572 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
574 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
576 2017-03-23 Richard Biener <rguenther@suse.de>
578 PR tree-optimization/80032
579 * gimplify.c (gimple_push_cleanup): Forced unconditional
580 cleanups still have to go to the conditional_cleanups
583 2017-03-22 Jakub Jelinek <jakub@redhat.com>
585 PR tree-optimization/80072
586 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
588 (next_operand_entry_id): Change type to unsigned int.
589 (sort_by_operand_rank): Make sure to return the right return value
590 even if unsigned fields are bigger than INT_MAX.
591 (struct oecount): Change cnt and id type to unsigned int.
592 (oecount_hasher::equal): Formatting fix.
593 (oecount_cmp): Make sure to return the right return value
594 even if unsigned fields are bigger than INT_MAX.
595 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
598 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
599 TREE_READONLY on result if writing it more than once.
602 * doc/invoke.texi (-fsanitize=thread): Document that with
603 -fnon-call-exceptions atomics are not able to throw
607 * tsan.c: Include tree-eh.h.
608 (instrument_builtin_call): Call maybe_clean_eh_stmt or
609 maybe_clean_or_replace_eh_stmt where needed.
610 (instrument_memory_accesses): Add cfg_changed argument.
611 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
613 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
615 PR rtl-optimization/63191
616 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
617 wrapper function, moved the whole old content into ...
618 (ix86_delegitimize_address_1): ... this. New inline function.
619 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
620 true as last argument instead of ix86_delegitimize_address.
622 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
624 * config/aarch64/aarch64.c (generic_branch_cost): Copy
625 cortexa57_branch_cost.
627 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
629 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
631 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
634 * doc/md.texi (Constraints): Document wA constraint.
635 * config/rs6000/constraints.md (wA): New.
636 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
637 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
638 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
639 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
641 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
644 * gimplify.c (is_oacc_declared): New function.
645 (oacc_default_clause): Use it to set default flags for acc declared
646 variables inside parallel regions.
647 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
649 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
650 declare attribute to any decl as necessary.
652 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
655 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
656 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
657 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
658 (arm_arch_lpae): This.
659 * config/arm/arm.c (arm_arch7ve): Rename into ...
660 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
661 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
664 2017-03-22 Martin Liska <mliska@suse.cz>
667 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
668 error message instead of an ICE.
670 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
672 * doc/extend.texi (6.11 Additional Floating Types): Revise.
674 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
676 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
678 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
681 2017-03-21 Martin Sebor <msebor@redhat.com>
683 * doc/extend.texi: Use "cannot" instead of "can't."
684 * doc/hostconfig.texi: Same.
685 * doc/install.texi: Same.
686 * doc/invoke.texi: Same.
687 * doc/loop.texi: Same.
689 * doc/objc.texi: Same.
690 * doc/rtl.texi: Same.
692 * doc/tm.texi.in: Same.
693 * doc/trouble.texi: Same.
695 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
698 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
699 (collect_checksum_attributes): Set it.
700 (die_checksum_ordered): Use it.
702 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
704 PR tree-optimization/79908
705 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
706 change: For a VA_ARG whose LHS has been cast away, use
707 force_gimple_operand to construct the side effects.
709 2017-03-21 David Malcolm <dmalcolm@redhat.com>
712 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
713 more amenable to translation.
714 (oacc_loop_auto_partitions): Likewise.
716 2017-03-21 Marek Polacek <polacek@redhat.com>
717 Martin Sebor <msebor@redhat.com>
719 PR tree-optimization/80109
720 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
723 2017-03-21 Jakub Jelinek <jakub@redhat.com>
724 Segher Boessenkool <segher@kernel.crashing.org>
727 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
728 check reg_used_between_p between insn and one of succ or succ2
729 depending on if succ is artificial insn not inserted into insn
732 2017-03-21 Martin Liska <mliska@suse.cz>
734 PR gcov-profile/80081
735 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
736 * doc/gcc.texi: Include gcov-dump stuff.
737 * doc/gcov-dump.texi: New file.
739 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
741 PR rtl-optimization/79150
742 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
743 conditional jump, if the jump is the last insn of the loop.
745 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
746 Richard Biener <rguenth@suse.com>
748 PR tree-optimization/79908
749 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
750 been cast away, use force_gimple_operand to construct the side
753 2017-03-21 Martin Liska <mliska@suse.cz>
756 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
759 2017-03-21 Brad Spengler <spender@grsecurity.net>
762 * plugin.c (htab_hash_plugin): New function.
763 (add_new_plugin): Use it and adjust.
764 (parse_plugin_arg_opt): Adjust.
765 (init_one_plugin): Likewise.
767 2017-03-21 Richard Biener <rguenther@suse.de>
769 PR tree-optimization/80032
770 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
771 if set force the cleanup to happen unconditionally.
772 (gimplify_target_expr): Push inserted clobbers with force_uncond
773 to avoid them being removed by control-dependent DCE.
775 2017-03-21 Richard Biener <rguenther@suse.de>
777 PR tree-optimization/80122
778 * tree-inline.c (copy_bb): Do not expans va-arg packs or
779 va_arg_pack_len when the inlined call stmt requires pack
781 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
783 2017-03-21 Jakub Jelinek <jakub@redhat.com>
786 * tsan.c (instrument_builtin_call): If the memory model argument
787 is not a constant, assume it is valid.
790 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
793 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
795 PR rtl-optimization/79910
796 * combine.c (can_combine_p): Do not allow combining an I0 or I1
797 if its dest is used by an insn before I2 (other than the combined
798 insns themselves, which are properly handled already).
800 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
803 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
805 * combine.c (record_used_regs): New static function.
806 (try_combine): Handle situations where there is an additional
807 instruction between I2 and I3 which needs to have a LOG_LINK
811 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
813 * combine.c (try_combine): Delete redundant i1 test. Call
814 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
816 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
819 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
822 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
824 PR tree-optimization/80054
825 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
826 the optimization if a PHI or any of its arguments is not dominated
827 by the candidate's basis. Use gphi* rather than gimple* as
829 (replace_profitable_candidates): Clean up a gimple* variable that
830 should be a gphi* variable.
832 2017-03-20 Martin Sebor <msebor@redhat.com>
835 * doc/extend.texi (attribute constructor): Document present limitation.
837 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
840 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
841 __POWER9_VECTOR__ #ifdef control, change template definition to
842 use Power9-specific built-in function.
843 (vec_any_eq): Likewise.
844 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
845 to control outcomes from this test.
846 (vector_ae_<mode>p): For VEC_F modes, likewise.
848 2017-03-20 Ian Lance Taylor <iant@google.com>
850 * config/i386/i386.c (ix86_function_regparm): Save an extra
851 register for -fsplit-stack with DECL_STATIC_CHAIN.
853 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
856 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
857 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
859 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
861 * config/riscv/riscv.c (riscv_print_operand): Use "fence
863 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
866 2017-03-20 Marek Polacek <polacek@redhat.com>
869 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
871 2017-03-20 Richard Biener <rguenther@suse.de>
873 PR tree-optimization/80113
874 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
875 allocate extra SSA name for PHI def.
876 (add_close_phis_to_outer_loops): Likewise.
877 (add_close_phis_to_merge_points): Likewise.
878 (copy_loop_close_phi_args): Likewise.
879 (copy_cond_phi_nodes): Likewise.
881 2017-03-20 Martin Liska <mliska@suse.cz>
884 * tree-chkp.c (chkp_build_returned_bound): Do not build
885 returned bounds for a LHS that's not a BOUNDED_P type.
887 2017-03-20 Martin Liska <mliska@suse.cz>
891 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
892 COMPLEX_CST and VECTOR_CST.
894 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
897 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
898 target operand. A new splitter adds the clobber statement in case
899 the target operand is dead anyway.
901 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
903 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
904 to age-old versions of binutils and glibc.
906 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
908 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
910 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
912 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
914 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
916 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
917 requirement for binutils 2.13.
919 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
921 * combine.c (try_combine): Delete redundant i1 test. Call
922 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
924 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
926 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
927 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
929 <riscv64-*-elf>: Re-arrange section
930 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
931 <riscv32-*-linux>: Likewise.
932 <riscv64-*-elf>: Likewise
933 <riscv64-*-linux>: Likewise.
935 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
938 * aarch64.opt(verbose-cost-dump): Fix typo.
940 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
943 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
944 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
946 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
948 * reload.c (find_reloads): When reloading a nonoffsettable address,
949 use RELOAD_OTHER for it and its address reloads.
951 PR rtl-optimization/79910
952 * combine.c (record_used_regs): New static function.
953 (try_combine): Handle situations where there is an additional
954 instruction between I2 and I3 which needs to have a LOG_LINK
957 2017-03-17 Jeff Law <law@redhat.com>
959 PR tree-optimization/71437
960 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
961 conditional in the hash table first.
962 (vrp_dom_walker::before_dom_children): Extract condition from
963 ASSERT_EXPR. Record condition, its inverion and any implied
966 2017-03-17 Marek Polacek <polacek@redhat.com>
967 Markus Trippelsdorf <markus@trippelsdorf.de>
969 PR tree-optimization/80079
970 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
973 2017-03-17 Richard Biener <rguenther@suse.de>
976 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
977 Properly verify the LHS before the RHS possibly claims to be
979 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
982 2017-03-17 Martin Jambor <mjambor@suse.cz>
984 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
985 (List of -O2 options): Likewise.
986 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
989 2017-03-17 Tom de Vries <tom@codesourcery.com>
991 * gcov-dump.c (print_usage): Print bug_report_url.
993 2017-03-17 Richard Biener <rguenther@suse.de>
996 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
997 (parser::peek): Likewise.
999 2017-03-17 Richard Biener <rguenther@suse.de>
1001 PR tree-optimization/80048
1002 * sese.c (free_sese_info): Properly release rename_map and
1003 copied_bb_map elements.
1005 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
1007 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
1008 Add linked-list forward and backlinks. Insert on
1009 construction, remove on destruction.
1010 (class pass_store_merging): Add m_stores_head field.
1011 (pass_store_merging::terminate_and_process_all_chains):
1012 Iterate over m_stores_head list.
1013 (pass_store_merging::terminate_all_aliasing_chains):
1015 (pass_store_merging::execute): Check for debug stmts first.
1016 Push new chains onto the m_stores_head stack.
1018 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
1021 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
1022 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
1023 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
1025 2017-03-16 Jeff Law <law@redhat.com>
1027 PR tree-optimization/71437
1028 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
1029 member function. Implementation moved into after_dom_children
1030 member function and into the threader's thread_outgoing_edges
1032 (dom_opt_dom_walker::after_dom_children): Simplify by moving
1033 some code into new thread_outgoing_edges.
1034 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
1035 definition. Simplify marker handling (do it here). Assume we always
1036 have the available expression and the const/copies tables.
1037 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
1039 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
1040 * tree-vrp.c (equiv_stack): No longer file scoped.
1041 (vrp_dom_walker): New class.
1042 (vrp_dom_walker::before_dom_children): New member function.
1043 (vrp_dom_walker::after_dom_children): Likewise.
1044 (identify_jump_threads): Setup domwalker. Use it rather than
1045 walking edges in a random order by hand. Simplify setup/finalization.
1046 (finalize_jump_threads): Remove.
1047 (vrp_finalize): Do not call identify_jump_threads here.
1048 (execute_vrp): Do it here instead and call thread_through_all_blocks
1051 PR tree-optimization/71437
1052 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
1054 (simplify_stmt_for_jump_threading): Add basic_block argument. All
1056 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
1057 (dom_opt_dom_walker::thread_across_edge): Remove
1058 handle_dominating_asserts argument. All callers changed.
1059 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
1060 changes. Remove calls to lhs_of_dominating_assert. Other
1061 uses of handle_dominating_asserts turn into unconditional code
1062 (simplify_control_stmt_condition_1): Likewise.
1063 (simplify_control_stmt_condition): Likewise.
1064 (thread_through_normal_block, thread_across_edge): Likewise.
1065 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
1066 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
1067 object if it is not an SSA_NAME.
1068 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
1069 before calling into the VRP specific simplifiers.
1070 (identify_jump_threads): Remove handle_dominating_asserts
1073 2017-03-16 Jakub Jelinek <jakub@redhat.com>
1076 * tree-diagnostic.c (default_tree_printer): No longer static.
1077 * tree-diagnostic.h (default_tree_printer): New prototype.
1079 2017-03-16 Tamar Christina <tamar.christina@arm.com>
1081 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
1082 Change ins into fmov.
1084 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1086 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
1087 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
1088 Use h_con constraint for operand 1.
1089 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
1090 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
1092 2017-03-15 Jeff Law <law@redhat.com>
1094 PR tree-optimization/71437
1095 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
1096 (record_temporary_equivalences): Use it.
1098 PR tree-optimization/71437
1099 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
1100 tree-ssa-scopedtables.
1101 (lookup_avail_expr, build_and_record_new_cond): Likewise.
1102 (record_conditions, record_cond, vuse_eq): Likewise.
1103 (record_edge_info): Adjust to API tweak of record_conditions.
1104 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
1105 (record_temporary_equivalences, optimize_stmt): Likewise.
1106 (eliminate_redundant_computations): Likewise.
1107 (record_equivalences_from_stmt): Likewise.
1108 * tree-ssa-scopedtables.c: Include options.h and params.h.
1109 (vuse_eq): New function, moved from tree-ssa-dom.c
1110 (build_and_record_new_cond): Likewise.
1111 (record_conditions): Likewise. Accept vector of conditions rather
1112 than edge_equivalence structure for first argument.
1113 for the first argument.
1114 (avail_exprs_stack::lookup_avail_expr): New member function, moved
1115 from tree-ssa-dom.c.
1116 (avail_exprs_stack::record_cond): Likewise.
1117 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
1118 from tree-ssa-dom.c.
1119 (avail_exprs_stack): Add new member functions lookup_avail_expr
1121 (record_conditions): Declare.
1123 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
1126 * lra-constraints.c (process_alt_operands): Increase reject for
1127 reloading an input/output operand.
1129 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
1132 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
1133 insns to convert from signed/unsigned char/short to IEEE 128-bit
1135 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
1137 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
1140 * config/i386/i386.c (ix86_vector_duplicate_value): Create
1141 subreg of inner mode for values already in registers.
1143 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
1145 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
1146 iteration reg is used after the loop.
1148 2017-03-14 Martin Sebor <msebor@redhat.com>
1150 PR tree-optimization/79800
1151 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
1152 precision in negative-positive range.
1153 (format_floating): Call non-const overload with adjusted precision.
1155 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
1158 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
1161 2017-03-14 Martin Sebor <msebor@redhat.com>
1164 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
1165 * builtins.def (aligned_alloc): Use it.
1168 * Makefile.in (GTFILES): Add calls.c.
1169 * calls.c: Include "gt-calls.h".
1171 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
1173 PR rtl-optimization/79728
1174 * regs.h (struct target_regs): New field
1175 x_contains_allocatable_regs_of_mode.
1176 (contains_allocatable_regs_of_mode): New macro.
1177 * reginfo.c (init_reg_sets_1): Initialize it, and change
1178 contains_reg_of_mode so it includes global regs as well.
1179 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
1180 rather than contains_regs_of_mode.
1182 2017-03-14 Martin Liska <mliska@suse.cz>
1184 * doc/invoke.texi: Document options that can't be combined with
1185 -fcheck-pointer-bounds.
1187 2017-03-14 Martin Liska <mliska@suse.cz>
1190 * doc/invoke.texi (-Wchkp): Document the option.
1192 2017-03-14 Martin Liska <mliska@suse.cz>
1194 * Makefile.in: Install gcov-dump.
1196 2017-03-14 Martin Liska <mliska@suse.cz>
1198 * multiple_target.c (expand_target_clones): Bail out for
1199 an invalid attribute.
1201 2017-03-14 Richard Biener <rguenther@suse.de>
1203 * alias.c (struct alias_set_entry): Pack properly.
1204 * cfgloop.h (struct loop): Likewise.
1205 * cse.c (struct set): Likewise.
1206 * ipa-utils.c (struct searchc_env): Likewise.
1207 * loop-invariant.c (struct invariant): Likewise.
1208 * lra-remat.c (struct cand): Likewise.
1209 * recog.c (struct change_t): Likewise.
1210 * rtl.h (struct address_info): Likewise.
1211 * symbol-summary.h (function_summary): Likewise.
1212 * tree-loop-distribution.c (struct partition): Likewise.
1213 * tree-object-size.c (struct object_size_info): Likewise.
1214 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
1215 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
1216 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
1217 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
1218 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
1219 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
1220 (struct _stmt_vec_info): Likewise.
1222 2017-03-14 Martin Liska <mliska@suse.cz>
1225 * multiple_target.c (create_dispatcher_calls): Check that
1226 a target can create a function dispatcher.
1228 2017-03-14 Martin Liska <mliska@suse.cz>
1231 * multiple_target.c (expand_target_clones): Drop local.local
1232 flag for default implementation.
1234 2017-03-14 Richard Biener <rguenther@suse.de>
1236 PR tree-optimization/80030
1237 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
1239 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
1241 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
1242 gcc_fallthrough() instead of __attribute__((fallthrough));
1244 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
1246 * doc/gcc.texi: Remove "up" link to (DIR).
1247 * doc/gccint.texi: Ditto.
1249 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
1251 * doc/install.texi (Specific) <avr>: Remove reference to
1254 2017-03-13 Jeff Law <law@redhat.com>
1256 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
1257 attribute rather than comments.
1259 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
1260 match_scratch operand is highest.
1262 2017-03-13 Martin Liska <mliska@suse.cz>
1265 * ipa-pure-const.c (warn_function_noreturn): If the declarations
1266 is a CHKP clone, use original declaration.
1268 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
1270 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
1271 (arc_conditional_register_usage): Use a different allocation order
1272 when optimizing for size.
1273 * common/config/arc/arc-common.c (arc_option_optimization_table):
1274 Section anchors default on when optimizing for size.
1276 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
1278 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
1280 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
1282 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
1283 * config/arc/arc.md (cpu_facility): Add cd variant.
1284 (*movqi_insn): Add code density variant.
1285 (*movhi_insn): Likewise.
1286 (*movqi_insn): Likewise.
1287 (*addsi3_mixed): Likewise.
1288 (subsi3_insn): Likewise.
1290 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
1292 * config/arc/arc.md (movsi_cond_exec): Update constraint.
1294 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
1296 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
1297 expressions with MINUS and UNARY ops.
1299 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1302 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
1304 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
1305 between vec_select and vector argument.
1306 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
1307 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
1308 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
1309 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
1310 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
1311 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
1313 2017-03-13 Richard Biener <rguenther@suse.de>
1316 * params.def (vect-max-peeling-for-alignment): Fix typo.
1318 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
1320 * doc/install.texi (Specific) <mips-*-*>: Remove description of
1321 issue that only occurred with binutils below 2.18.
1323 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
1325 * doc/install.texi (Specific) <cris-axis-elf>: No longer
1326 refer to binutils 2.11/2.12 minimum.
1328 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
1330 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
1331 ftp.kernel.org and simplify binutils requirement.
1333 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
1335 * doc/invoke.texi (Warning Options): Fix spelling of link-time
1337 (Optimize Options): Ditto. Also remove redundancy.
1339 2017-03-10 David Malcolm <dmalcolm@redhat.com>
1341 PR translation/79848
1342 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
1344 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
1345 to G_ to avoid double translation.
1347 2017-03-10 David Malcolm <dmalcolm@redhat.com>
1349 PR translation/79923
1350 * auto-profile.c (get_combined_location): Convert leading
1351 character of diagnostics to lower case and remove trailing period.
1352 (read_profile): Likewise for various diagnostics.
1353 * config/arm/arm.c (arm_option_override): Remove trailing period
1354 from various diagnostics.
1355 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
1356 (msp430_expand_delay_cycles): Likewise.
1358 2017-03-10 David Malcolm <dmalcolm@redhat.com>
1361 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
1362 full command-line argument, rather than just "str".
1363 (aarch64_validate_march): Likewise.
1364 (aarch64_validate_mtune): Likewise.
1366 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
1368 PR rtl-optimization/78911
1369 * lra-assigns.c (must_not_spill_p): New function.
1370 (spill_for): Use it.
1372 2017-03-10 Jakub Jelinek <jakub@redhat.com>
1374 PR tree-optimization/79981
1375 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
1376 ATOMIC_COMPARE_EXCHANGE ifn result.
1377 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
1378 IFN_ATOMIC_COMPARE_EXCHANGE.
1380 2017-03-10 David Malcolm <dmalcolm@redhat.com>
1383 * opts.c (parse_sanitizer_options): Add missing question mark to
1384 "did you mean" message.
1386 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1388 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
1390 (VMULEUH_UNS): Likewise.
1391 (VMULOUB_UNS): Likewise.
1392 (VMULOUH_UNS): Likewise.
1393 * config/rs6000/rs6000.c (builtin_function_type): Remove
1394 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
1396 2017-03-10 David Malcolm <dmalcolm@redhat.com>
1399 * read-rtl-function.c (function_reader::read_rtx_operand): Update
1400 x with result of extra_parsing_for_operand_code_0.
1401 (function_reader::extra_parsing_for_operand_code_0): Convert
1402 return type from void to rtx, returning x. When reading
1403 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
1404 larger size containing struct block_symbol.
1406 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
1408 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
1409 -mfloat128-hardware without -m64.
1411 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
1414 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
1415 entries to the case statement that marks unsigned arguments to
1416 overloaded functions.
1418 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
1420 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
1421 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
1423 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
1426 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
1427 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
1429 2017-03-10 Martin Liska <mliska@suse.cz>
1433 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
1434 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
1437 2017-03-10 Olivier Hainque <hainque@adacore.com>
1439 * tree-switch-conversion (array_value_type): Start by resetting
1440 candidate type to it's main variant.
1442 2017-03-10 Jakub Jelinek <jakub@redhat.com>
1444 PR rtl-optimization/79909
1445 * combine.c (try_combine): Use simplify_replace_rtx on individual
1446 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
1447 of the whole CALL_INSN_FUNCTION_USAGE.
1449 PR tree-optimization/79972
1450 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
1451 get_range_info on SSA_NAMEs. Formatting fixes.
1453 2017-03-10 Richard Biener <rguenther@suse.de>
1454 Jakub Jelinek <jakub@redhat.com>
1456 PR tree-optimization/77975
1457 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
1458 edge to be constant.
1459 (get_val_for): For constant x return it. Formatting fix.
1460 (loop_niter_by_eval): Avoid pointless looping if the next iteration
1461 would use the same bases as the current one.
1463 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1465 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
1466 instead of vec_select for V1TImode.
1467 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
1469 (VSX_LE_128): Add V1TI to this mode iterator.
1470 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
1471 (*vsx_le_perm_store_<mode>): Likewise.
1472 (pre-reload splitter for VSX stores): Likewise.
1473 (post-reload splitter for VSX stores): Likewise.
1474 (*vsx_xxpermdi2_le_<mode>): Likewise.
1475 (*vsx_lxvd2x2_le_<mode>): Likewise.
1476 (*vsx_stxvd2x2_le_<mode>): Likewise.
1478 2017-03-09 Michael Eager <eager@eagercon.com>
1480 Correct failures with --enable-checking=yes,rtl.
1482 * config/microblaze/microblaze.c (microblaze_expand_shift):
1483 Replace GET_CODE test with CONST_INT_P and INTVAL test with
1484 test for const0_rtx.
1485 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
1486 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
1488 2017-03-09 Richard Biener <rguenther@suse.de>
1490 PR tree-optimization/79977
1491 * graphite-scop-detection.c (scop_detection::merge_sese):
1492 Handle the case of extra exits to blocks dominating the entry.
1494 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
1496 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
1499 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
1501 PR rtl-optimization/79949
1502 * lra-constraints.c (process_alt_operands): Check memory when
1503 trying to predict a cycle. Print about the overall increase.
1505 2017-03-09 Richard Biener <rguenther@suse.de>
1508 * gimple-expr.c (useless_type_conversion_p): Preserve
1509 TYPE_SATURATING for fixed-point types.
1511 2017-03-09 Richard Biener <rguenther@suse.de>
1514 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
1515 alignment of BLKmode params.
1517 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1520 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
1521 (VALL_NO_V2Q): Likewise.
1523 * config/aarch64/aarch64-simd.md
1524 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
1526 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
1527 VALL_NO_V2Q mode iterator.
1528 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
1530 2017-03-09 Martin Liska <mliska@suse.cz>
1532 PR tree-optimization/79631
1533 * tree-chkp-opt.c (chkp_is_constant_addr): Call
1534 tree_int_cst_sign_bit just for INTEGER constants.
1536 2017-03-09 Martin Liska <mliska@suse.cz>
1540 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
1543 2017-03-09 Marek Polacek <polacek@redhat.com>
1546 * tree.c (inchash::add_expr): Handle TREE_VEC.
1548 2017-03-09 Martin Liska <mliska@suse.cz>
1551 (chkp_narrow_size_and_offset): New function.
1552 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
1553 (void chkp_parse_bit_field_ref): New function.
1554 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
1555 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
1557 2017-03-09 Martin Liska <mliska@suse.cz>
1560 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
1561 (chkp_find_bounds_1): Remove gcc_unreachable.
1563 2017-03-09 Jakub Jelinek <jakub@redhat.com>
1566 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
1567 BUILT_IN_SYNC*, determine the access type from the size suffix and
1568 always build a MEM_REF with that type. Handle forgotten
1569 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
1572 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
1573 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
1574 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
1575 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
1576 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
1577 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
1578 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
1579 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
1580 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
1581 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
1582 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
1583 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
1584 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
1585 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
1586 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
1587 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
1588 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
1589 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
1590 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
1591 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
1592 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
1593 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
1594 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
1595 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
1596 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
1597 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
1598 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
1599 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
1600 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
1601 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
1602 definitions outside of __OPTIMIZE__ guarded section.
1605 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
1606 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
1607 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
1608 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
1611 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1613 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
1614 ("vfenez<mode>"): Add missing constraints.
1616 2017-03-08 Martin Sebor <msebor@redhat.com>
1619 * config/nds32/nds32.c (nds32_option_override):
1621 2017-03-08 Jakub Jelinek <jakub@redhat.com>
1624 * gimplify.c (gimplify_omp_for): Replace index var in outer
1625 taskloop statement with an artificial variable and add
1626 OMP_CLAUSE_PRIVATE clause for it.
1628 2017-03-08 Richard Biener <rguenther@suse.de>
1630 PR tree-optimization/79955
1631 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
1632 for accesses that are completely outside of the variable.
1634 2017-03-08 Andrew Haley <aph@redhat.com>
1636 PR tree-optimization/79943
1637 * tree-ssa-loop-split.c (compute_new_first_bound): When
1638 calculating the new upper bound, (END-BEG) should be added, not
1641 2017-03-08 Jakub Jelinek <jakub@redhat.com>
1643 * config/avr/avr.md (setmemhi): Make sure match_dup
1644 operand number comes before match_scratch.
1646 2017-03-08 Richard Biener <rguenther@suse.de>
1648 PR tree-optimization/79920
1649 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
1650 with ncopies == 1 to ...
1651 (vect_transform_slp_perm_load): ... here. Properly compute
1652 all element loads by iterating VF times over the group. Do
1653 not handle ncopies (computed in a broken way) in
1654 vect_create_mask_and_perm.
1656 2017-03-08 Jakub Jelinek <jakub@redhat.com>
1659 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
1660 is a uniform vector, use uniform_vector_p return value instead of
1661 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
1663 2017-03-07 Marek Polacek <polacek@redhat.com>
1666 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
1667 (alloca_call_type): Likewise.
1669 2017-03-07 Martin Liska <mliska@suse.cz>
1671 * gcov.c (process_args): Put comment to correct location.
1673 2017-03-07 Martin Liska <mliska@suse.cz>
1676 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
1677 Use array_at_struct_end_p instead of DECL_CHAIN (field).
1678 (chkp_narrow_bounds_for_field): Likewise.
1679 (chkp_parse_array_and_component_ref): Pass one more argument to
1682 2017-03-07 Richard Biener <rguenther@suse.de>
1684 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
1687 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
1689 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
1690 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
1692 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1695 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
1696 to end of description.
1697 (PARAM_MAX_STORES_TO_MERGE): Likewise.
1699 2017-03-07 Jakub Jelinek <jakub@redhat.com>
1701 PR rtl-optimization/79901
1702 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
1704 (*avx512f_<code><mode>3<mask_name>): ... this.
1705 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
1706 iterator instead of VI8_AVX2_AVX512BW.
1708 PR rtl-optimization/79901
1709 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
1710 min/max expander, expand it using expand_vec_cond_expr.
1713 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
1716 2017-03-06 Jakub Jelinek <jakub@redhat.com>
1719 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
1720 to void * for PCH reasons.
1721 * dwarf2out.c (output_loc_operands, output_die): Cast
1722 v.val_vec.array to unsigned char *.
1724 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
1727 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
1730 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
1732 PR rtl-optimization/79571
1733 * lra-constraints.c (process_alt_operands): Calculate static
1734 reject and subtract it from overall when only addresses will be
1737 2017-03-06 Julia Koval <julia.koval@intel.com>
1740 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
1741 incoming stack boundary to 128 for 64-bit targets.
1743 2017-03-06 Richard Biener <rguenther@suse.de>
1745 PR tree-optimization/79894
1746 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
1747 to NULL after folding it.
1749 2017-03-06 Richard Biener <rguenther@suse.de>
1751 PR tree-optimization/79824
1752 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
1753 check disabling peeling for gaps.
1755 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
1757 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
1758 attributes): Document gettimeofday.
1760 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
1762 * config/s390/s390.c (s390_option_override_internal): Set
1763 PARAM_MIN_VECT_LOOP_BOUND
1765 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
1767 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
1768 * config/s390/s390.md: Likewise.
1770 2017-03-06 Jakub Jelinek <jakub@redhat.com>
1773 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
1774 (<avx2_avx512>_perm<mode>): Rename to ...
1775 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
1777 (<avx512>_perm<mode>_mask): Rename to ...
1778 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
1780 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
1781 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
1782 instead of VI8F_256_512.
1783 (avx512f_perm<mode>): New define_expand.
1784 (avx512f_perm<mode>_mask): Likewise.
1785 (avx512f_perm<mode>_1<mask_name>): New define_insn.
1786 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
1788 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
1790 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
1791 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
1793 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
1795 2017-03-06 Martin Liska <mliska@suse.cz>
1798 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
1799 when having a SSA NAME w/o VAR_DECL assigned to it.
1801 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
1803 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
1804 msa_dpsub_<su>_d): Fix MODE for vec_select.
1806 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
1808 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
1810 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
1812 2017-03-06 Richard Biener <rguenther@suse.de>
1814 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
1815 * plugin.c (register_plugin_info): Likewise.
1816 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
1818 2017-03-05 Jakub Jelinek <jakub@redhat.com>
1820 * config/i386/sse.md (sse_storehps, sse_storelps,
1821 avx_<castmode><avxsizesuffix>_<castmode>,
1822 avx512f_<castmode><avxsizesuffix>_<castmode>,
1823 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
1824 in condition that at least one operand is not a MEM.
1826 2017-03-03 Jakub Jelinek <jakub@redhat.com>
1829 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
1830 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
1832 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
1833 gimple_call_nothrow_p flag based on whether original builtin can throw.
1834 If it can, emit following stmts on the fallthrough edge.
1835 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
1836 don't create new bb if inserting just debug stmts on the edge, try to
1837 insert them on the fallthru bb or just reset debug stmts.
1839 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
1842 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
1843 restore recog_data (including the operand rtxes inside it) around
1844 the call to get_insn_template.
1846 2017-03-03 Martin Sebor <msebor@redhat.com>
1848 PR tree-optimization/79699
1849 * context.c (context::~context): Free MPFR caches to avoid
1850 a memory leak on program exit.
1852 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1854 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
1855 Use wide_int::ulow () instead of .elt (0).
1857 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
1859 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
1860 (*pushxf): Limit oF constraint to 32bit targets and add oC
1861 constraint for 64bit targets.
1862 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
1863 (*pushdf): Change rmF constraint to rmC.
1865 2017-03-03 Martin Liska <mliska@suse.cz>
1867 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
1868 Remove unused variable.
1870 2017-03-03 Jakub Jelinek <jakub@redhat.com>
1873 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
1874 is a memory operand, increase num_memory.
1875 (ix86_expand_args_builtin): Likewise.
1877 2017-03-03 Jan Hubicka <jh@suse.cz>
1880 * ipa-devirt.c (maybe_record_node): Properly handle
1881 __cxa_pure_virtual visibility.
1883 2017-03-03 Martin Liska <mliska@suse.cz>
1885 PR tree-optimization/79803
1886 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
1888 (pass_loop_prefetch::execute): Disabled optimization if an
1889 assumption about L1 cache size is not met.
1891 2017-03-03 Martin Liska <mliska@suse.cz>
1893 PR rtl-optimization/79574
1894 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
1895 (hash_scan_set): Likewise.
1896 (dump_hash_table): Likewise.
1897 (hoist_code): Likewise.
1899 2017-03-03 Richard Biener <rguenther@suse.de>
1901 * fixed-value.c (fixed_from_string): Restore use of elt (1)
1902 in place of uhigh ().
1903 (fixed_convert_from_real): Likewise.
1905 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
1908 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
1910 2017-03-03 Richard Biener <rguenther@suse.de>
1913 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
1914 TYPE_OVERFLOW_UNDEFINED check.
1916 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1918 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
1920 (vector_ae_<mode>_p): Likewise.
1921 (vector_nez_<mode>_p): Likewise.
1922 (vector_ne_v2di_p): Likewise.
1923 (vector_ae_v2di_p): Likewise.
1924 (vector_ne_<mode>_p): Likewise.
1925 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
1927 (vsx_tsqrt<mode>2_fe): Likewise.
1929 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
1932 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
1934 2017-03-02 Jakub Jelinek <jakub@redhat.com>
1936 PR rtl-optimization/79780
1937 * cprop.c (one_cprop_pass): When second and further conditional trap
1938 in a single basic block is turned into an unconditional trap, turn it
1939 into a deleted note to avoid RTL verification failures.
1941 2017-03-02 Richard Biener <rguenther@suse.de>
1943 * fold-const.c (const_binop): Use ulow () instead of elt (0).
1945 2017-03-02 Richard Biener <rguenther@suse.de>
1947 PR tree-optimization/79345
1949 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
1950 param and abort the walk, returning -1 if it is hit.
1951 (walk_aliased_vdefs): Take a limit param and pass it on.
1952 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
1953 defaulting to 0 and return a signed int.
1954 * tree-ssa-uninit.c (struct check_defs_data): New struct.
1955 (check_defs): New helper.
1956 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
1957 about uninitialized memory.
1958 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
1959 bogus uninitialized warning.
1960 (fixed_convert_from_real): Likewise.
1962 2017-03-02 Bin Cheng <bin.cheng@arm.com>
1964 PR tree-optimization/66768
1965 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
1966 iv_use if base object can't be determined.
1968 2017-03-02 Jakub Jelinek <jakub@redhat.com>
1970 PR tree-optimization/79345
1971 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
1972 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
1973 (get_pattern_stats): Initialize it.
1974 * genemit.c (gen_expand): Verify match_scratch numbers come after
1975 match_operand/match_dup numbers.
1976 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
1977 match_scratch numbers.
1978 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
1980 * config/s390/s390.md (trunctdsd2): Likewise.
1982 2017-03-02 Richard Biener <rguenther@suse.de>
1984 * wide-int.h (wide_int_storage::operator=): Implement in terms
1987 2017-03-02 Richard Biener <rguenther@suse.de>
1989 PR tree-optimization/79777
1990 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
1991 the to insert expression to sth existing.
1993 2017-03-01 Martin Sebor <msebor@redhat.com>
1996 * gimple-ssa-sprintf.c
1997 (directive::known_width_and_precision): New function.
1998 (format_integer): Use it.
1999 (get_mpfr_format_length): Consider the full range of precision
2000 when computing %g output with the # flag. Set the likely byte
2001 count to 3 rather than 1 when precision is indeterminate.
2002 (format_floating): Correct the lower bound of precision.
2004 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2006 * doc/invoke.texi: Document default code model for 64-bit Linux.
2008 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2011 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
2012 udiv rather than div since input pattern is unsigned.
2014 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
2016 * config/i386/i386.c (print_reg): Warn for values of
2017 unsupported size in integer register.
2019 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
2022 * config/rs6000/predicates.md (current_file_function_operand): Do
2023 not allow self calls to be local if the function is replaceable.
2025 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
2028 * config/rs6000/altivec.h (vec_ctz and others): Change the
2029 preprocessor macro that controls conditional compilation from
2030 _ARCH_PWR9 to __POWER9_VECTOR__.
2031 (vec_all_ne): Change parameterization of __altivec_scalar_pred
2032 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
2033 control (instead of _ARCH_PWR9 control) so that template
2034 definition uses power9-specific function.
2035 (vec_any_eq): Likewise.
2036 (vec_all_ne): Change macro definition to use a power9-specific
2037 expansion under #ifdef __POWER9_VECTOR__ control (instead of
2038 _ARCH_PWR9 control).
2039 (vec_any_eq) Likewise.
2040 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
2041 expansion for CMPNEF to remove support for xvcmpnesp instruction.
2042 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
2043 support for xvcmpnedp instruction.
2044 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
2045 macro expansion so that Power9 implementation of vec_all_ne does
2046 not use the AltiVec predicate framework.
2047 (VCMPNEH_P): Likewise.
2048 (VCMPNEW_P): Likewise.
2049 (VCMPNED_P): Likewise.
2050 (VCMPNEFP_P): Likewise.
2051 (VCMPNEDP_P): Likewise.
2052 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
2053 implementation of vec_any_eq to not use AltiVec predicate
2055 (VCMPAEH_P): Likewise.
2056 (VCMPAEW_P): Likewise.
2057 (VCMPAED_P): Likewise.
2058 (VCMPAEFP_P): Likewise.
2059 (VCMPAEDP_P): Likewise.
2060 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
2061 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
2062 not use the AltiVec predicate framework.
2063 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
2064 of vec_any_eq to not use AltiVec predicate framework.
2065 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
2066 support for predefined __POWER9_VECTOR__ macro to indicate that
2067 Power9 instruction selection is enabled.
2068 (altivec_overloaded_builtins): Remove extraneous
2069 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
2070 function argument types RS6000_BTI_bool_V16QI and
2071 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
2072 entry for overloaded function argument types RS6000_BTI_bool_V4SI
2073 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
2074 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
2075 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
2076 Power9 for implementations of vec_cmpne. Change the signature for
2077 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
2078 (representing vec_all_ne) to remove the previously described first
2079 argument of type RS6000_BTI_INTSI, as this was an artifact of
2080 reliance on the AltiVec predicate framework, which is no longer
2081 used in the implementation of these functions. Add
2082 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
2083 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
2084 since, unlike the AltiVec predicate framework implementation, we
2085 do not share function descriptors between vec_alle and vec_anyeq.
2086 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
2087 set of modes that receive special treatment even when
2088 TARGET_P9_VECTOR is true. The special treatment emits code that
2089 does not depend on Power9 instructions.
2090 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
2091 define_expand to not rely on AltiVec predicate framework.
2092 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
2094 (vector_ne_v2di_p): Change this define_expand to not rely on
2095 AltiVec predicate framework.
2096 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
2098 (vector_ne_<mode>_p): Change this define_expand to not rely on
2099 AltiVec predicate framework.
2100 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
2102 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
2103 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
2104 define_insn pattern.
2105 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
2106 define_insn pattern because the xvcmpne<VSs>. instruction is not
2108 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
2109 instruction is not supported.
2111 2017-03-01 Jakub Jelinek <jakub@redhat.com>
2113 * config/nvptx/nvptx.c: Include intl.h.
2115 2017-03-01 Martin Jambor <mjambor@suse.cz>
2118 * ipa-prop.h (ipa_bits): Removed field known.
2119 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
2120 to pointers. Adjusted their comments to warn about their sharing.
2121 (ipcp_transformation_summary): Change bits to a vector of pointers.
2122 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
2123 (ipa_get_ipa_bits_for_value): Declare.
2124 * tree-vrp.h (value_range): Mark as GTY((for_user)).
2125 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
2126 (ipa_bits_hash_table): Likewise.
2127 (ipa_vr_ggc_hash_traits): Likewise.
2128 (ipa_vr_hash_table): Likewise.
2129 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
2130 being pointers and vr_known being removed.
2131 (ipa_set_jf_unknown): Likewise.
2132 (ipa_get_ipa_bits_for_value): New function.
2133 (ipa_set_jfunc_bits): Likewise.
2134 (ipa_get_value_range): New overloaded functions.
2135 (ipa_set_jfunc_vr): Likewise.
2136 (ipa_compute_jump_functions_for_edge): Use the above functions to
2137 construct bits and vr parts of jump functions.
2138 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
2139 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
2141 (ipcp_grow_transformations_if_necessary): Also allocate
2142 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
2144 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
2145 them. Fix too long lines.
2146 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
2147 vr_known being removed.
2148 (ipa_read_jump_function): Use new setter functions to construct bits
2149 and vr parts of jump functions or set them to NULL.
2150 (write_ipcp_transformation_info): Adjust for bits being pointers.
2151 (read_ipcp_transformation_info): Likewise.
2152 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
2154 Include gt-ipa-prop.h.
2155 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
2157 (ipcp_store_bits_results): Likewise.
2158 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
2159 Do not write to existing jump functions but use a temporary instead.
2161 2017-03-01 Jakub Jelinek <jakub@redhat.com>
2164 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
2165 attempt to use its first operand as BIT_FIELD_REF base.
2167 2017-03-01 Richard Biener <rguenther@suse.de>
2170 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
2171 interpolating formula in wrapping arithmetic.
2172 (chrec_apply): Convert chrec_evaluate return value to wanted type.
2174 2017-03-01 Jakub Jelinek <jakub@redhat.com>
2176 PR tree-optimization/79734
2177 * tree-vect-generic.c (expand_vector_condition): Optimize
2178 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
2179 Handle VEC_COND_EXPR where comparison has different inner width from
2182 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
2184 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
2185 markup, and similar issues. Remove @opindex entries for things
2186 that aren't options. Add missing -mmpy-option entries.
2188 2017-02-28 Jakub Jelinek <jakub@redhat.com>
2190 PR tree-optimization/79737
2191 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
2192 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
2193 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
2194 instead of byte_size. Formatting fix.
2195 (shift_bytes_in_array_right): Formatting fix.
2197 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
2200 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
2201 condition on optimize for the leaf function test.
2203 2017-02-28 Martin Liska <mliska@suse.cz>
2206 * read-rtl-function.c (function_reader::handle_unknown_directive):
2207 Bail out when one uses -flto.
2209 2017-02-28 Martin Liska <mliska@suse.cz>
2211 * common.opt: Replace space with tabular for options of <number>
2213 * config/i386/i386.opt: Show <number> value for
2214 -mlarge-data-threshold.
2215 * opts.c (print_filtered_help): Do not display number in hexadecimal
2218 2017-02-28 Martin Liska <mliska@suse.cz>
2220 * common.opt: Fix --help=option -Q for options which are of
2223 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
2225 * config/i386/i386.c (print_reg): Error out for values
2226 of 8-bit size in invalid integer register.
2228 2017-02-28 Martin Sebor <msebor@redhat.com>
2230 PR tree-optimization/79691
2231 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
2233 2017-02-28 Jakub Jelinek <jakub@redhat.com>
2236 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
2237 gcc_unreachable with output_operand_lossage.
2239 2017-02-28 Richard Biener <rguenther@suse.de>
2241 PR tree-optimization/79740
2242 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
2244 (visit_nary_op): Insert the nary into the hashtable if we
2245 pattern-matched sth.
2246 * tree-ssa-pre.c (eliminate_insert): Robustify.
2248 2017-02-28 Richard Biener <rguenther@suse.de>
2251 * fold-const.c (decode_field_reference): Reject out-of-bound
2254 2017-02-28 Jakub Jelinek <jakub@redhat.com>
2256 * config/i386/i386.c: Include intl.h.
2257 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
2258 instead of just cond ? "..." : "...".
2259 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
2260 * coverage.c (read_counts_file): Likewise.
2261 * omp-offload.c: Include intl.h.
2262 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
2263 of just cond ? "..." : "...".
2264 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
2265 of just cond ? "..." : "...".
2267 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
2270 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
2272 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
2273 'tune for' CPU name.
2274 * config/arm/arm-cpu-data.h: Regenerated.
2276 2017-02-28 Richard Biener <rguenther@suse.de>
2278 PR tree-optimization/79732
2279 * tree-inline.c (expand_call_inline): Do not shadow var.
2281 2017-02-28 Richard Biener <rguenther@suse.de>
2283 PR tree-optimization/79723
2284 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
2285 address-space properly.
2287 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
2289 * doc/optinfo.texi (Optimization groups): Fix option used for
2291 * doc/invoke.texi (-fopt-info): Document "omp".
2292 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
2293 (OPTGROUP_ALL): Add OPTGROUP_OMP.
2294 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
2295 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
2296 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
2298 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
2300 * dumpfile.c (optgroup_options): Instead of "openmp", associate
2301 OPTGROUP_OMP with "omp".
2303 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
2306 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
2307 for arithmetic shift of unsigned V2DI.
2309 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
2311 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
2312 arc/linux.h headers.
2313 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
2314 (LINK_SPEC): Likewise.
2315 (ARC_TLS_EXTRA_START_SPEC): Likewise.
2316 (EXTRA_SPECS): Likewise.
2317 (STARTFILE_SPEC): Likewise.
2318 (ENDFILE_SPEC): Likewise.
2319 (LIB_SPEC): Likewise.
2320 (TARGET_SDATA_DEFAULT): Likewise.
2321 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
2322 (MULTILIB_DEFAULTS): Likewise.
2323 (DWARF2_UNWIND_INFO): Likewise.
2324 * config/arc/big.h: New file.
2325 * config/arc/elf.h: Likewise.
2326 * config/arc/linux.h: Likewise.
2327 * config/arc/t-uClibc: Remove.
2329 2017-02-27 Bin Cheng <bin.cheng@arm.com>
2331 PR tree-optimization/77536
2332 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
2333 (tree_transform_and_unroll_loop): Use above function to compute the
2334 estimated niter of unrolled loop and use it when scaling profile.
2335 Also use count info rather than frequency if it's non-zero.
2336 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
2337 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
2338 (vect_transform_loop): Call above function.
2340 2017-02-27 Richard Biener <rguenther@suse.de>
2342 PR tree-optimization/45397
2343 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
2344 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
2345 (visit_nary_op): Add pattern matching for CSEing sign-changed
2346 or truncated operations with wider ones.
2348 2017-02-27 Richard Biener <rguenther@suse.de>
2350 PR tree-optimization/79690
2351 * tree-vect-stmts.c (vectorizable_store): Use vector type
2352 built from the DR with address-space.
2354 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
2356 * doc/invoke.texi (Optimize Options): Refine the description
2357 of asan-use-after-return.
2359 2017-02-25 Alan Modra <amodra@gmail.com>
2361 PR rtl-optimization/79584
2362 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
2363 base, not ad->base_term, the reg within base. Remove assertion
2364 that ad->base == ad->base_term. Replace gen_int_mode using
2365 bogus mode with const0_rtx.
2367 2017-02-25 Jakub Jelinek <jakub@redhat.com>
2370 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
2371 FMA_EXPR like tcc_binary or tcc_unary.
2373 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
2376 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
2378 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
2379 (output_loc_operands): Handle DW_OP_call_ref and
2380 DW_OP_GNU_variable_value.
2381 (struct variable_value_struct): New type.
2382 (struct variable_value_hasher): Likewise.
2383 (variable_value_hash): New variable.
2384 (string_types): Remove.
2385 (copy_loc_descr): New function.
2386 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
2387 (prepend_loc_descr_to_each): New function.
2388 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
2389 instead of add_loc_descr_to_each if the first argument is single
2390 location list and the second has multiple.
2391 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
2392 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
2393 when looking for variable value which doesn't have other location info.
2394 (loc_list_from_tree): Formatting fix.
2395 (gen_array_type_die): Simplify DW_AT_string_length handling.
2396 (adjust_string_types): Remove.
2397 (gen_subprogram_die): Don't call adjust_string_types nor test/set
2398 string_types. Call resolve_variable_values.
2399 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
2400 (resolve_addr_in_expr): Likewise. Add A argument.
2401 (copy_deref_exprloc): Remove deref argument. Adjust for the
2402 original expression being DW_OP_GNU_variable_value with optionally
2403 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
2404 optionally after it.
2405 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
2406 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
2407 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
2408 (variable_value_hasher::hash, variable_value_hasher::equal): New
2410 (resolve_variable_value_in_expr, resolve_variable_value,
2411 resolve_variable_values, note_variable_value_in_expr,
2412 note_variable_value): New functions.
2413 (dwarf2out_early_finish): Call note_variable_value on all toplevel
2416 2017-02-24 Jakub Jelinek <jakub@redhat.com>
2419 * opts.h (handle_generated_option): Add GENERATED_P argument.
2420 * opts-common.c (handle_option): Adjust function comment.
2421 (handle_generated_option): Add GENERATED_P argument, pass it to
2423 (control_warning_option): Pass false to handle_generated_option
2425 * opts.c (maybe_default_option): Pass true to handle_generated_option
2427 * optc-gen.awk: Likewise.
2429 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
2431 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
2432 a REG, look at the REG it is a SUBREG of.
2433 (splitter for cmpeqsi_t): Ditto.
2435 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
2437 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
2438 the special USEs with the pattern of the insn, not the insn itself.
2440 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
2443 * doc/invoke.texi: Document -mload-store-pairs.
2445 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
2446 Sandra Loosemore <sandra@codesourcery.com>
2448 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
2449 argument isn't a CONST_INT.
2450 (nios2_alternate_compare_const): Assert op is a CONST_INT.
2451 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
2452 (nios2_validate_compare): Bypass alternate compare logic if *op2
2454 (ldstwm_operation_p): Return false if first_base is not a REG or
2455 if first_offset is not a CONST_INT.
2457 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
2459 * config/cris/cris.md: Use correct operand in a define_peephole2.
2461 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
2463 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
2465 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
2467 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
2468 this_insn if it is an INSN or JUMP_INSN.
2469 (force_offsettable): Look at base, not at addr.
2470 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
2471 on things that aren't necessarily CONST_INTs.
2473 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
2475 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
2476 -mfpmath=sse is the default also for x86-32 targets with SSE2
2477 instruction set when @option{-ffast-math} is enabled
2479 2017-02-24 Jeff Law <law@redhat.com>
2481 PR rtl-optimizatoin/79286
2482 * ira.c (update_equiv_regs): Drop may_trap_p exception to
2485 2017-02-24 Richard Biener <rguenther@suse.de>
2487 PR tree-optimization/79389
2488 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
2491 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
2493 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
2494 function comment to reflect reality.
2495 (loop_exits_before_overflow): Fix typo in function description.
2497 2017-02-24 Richard Biener <rguenther@suse.de>
2499 PR tree-optimization/79389
2500 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
2501 properly that a threading opportunity exists. Detect conditional
2502 copy/constant propagation opportunities.
2504 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
2506 * config/visium/visium.md (type): Add trap.
2507 (b): New mode attribute.
2508 (*btst): Rename into...
2509 (*btst<mode>): ...this and adjust.
2510 (*cbranchsi4_btst_insn): Rename into...
2511 (*cbranch<mode>4_btst_insn): ...this and adjust.
2512 (trap): New define_insn.
2514 2017-02-23 Jakub Jelinek <jakub@redhat.com>
2516 PR tree-optimization/79389
2517 * ifcvt.c (struct noce_if_info): Add rev_cond field.
2518 (noce_reversed_cond_code): New function.
2519 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
2520 reversed_comparison_code. Formatting fix.
2521 (noce_try_store_flag): Test rev_cond != NULL in addition to
2522 reversed_comparison_code.
2523 (noce_try_store_flag_constants): Likewise.
2524 (noce_try_store_flag_mask): Likewise.
2525 (noce_try_addcc): Use rev_cond if non-NULL instead of
2526 reversed_comparison_code.
2527 (noce_try_cmove_arith): Likewise. Formatting fixes.
2528 (noce_try_minmax, noce_try_abs): Clear rev_cond.
2529 (noce_find_if_block): Initialize rev_cond.
2530 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
2531 instead of false as last argument never attempt to reverse it
2534 2017-02-23 Bin Cheng <bin.cheng@arm.com>
2536 PR tree-optimization/79663
2537 * tree-predcom.c (combine_chains): Process refs in reverse order
2538 only for ZERO length chains, and add explaining comment.
2540 2017-02-23 Jeff Law <law@redhat.com>
2542 PR tree-optimization/79578
2543 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
2544 in call to operand_equal_p.
2546 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
2549 * config/i386/cpuid.h: Fix another undefined behavior.
2551 2017-02-23 Richard Biener <rguenther@suse.de>
2553 PR tree-optimization/79683
2554 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
2555 vector types for data-refs.
2557 2017-02-23 Martin Liska <mliska@suse.cz>
2559 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
2561 2017-02-23 Jakub Jelinek <jakub@redhat.com>
2564 * internal-fn.c (get_range_pos_neg): Moved to ...
2565 * tree.c (get_range_pos_neg): ... here. No longer static.
2566 * tree.h (get_range_pos_neg): New prototype.
2567 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
2568 are known to be in between 0 and signed maximum inclusive, try to
2569 expand both unsigned and signed divmod and use the cheaper one from
2572 2017-02-22 Jeff Law <law@redhat.com>
2574 PR tree-optimization/79578
2575 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
2576 to compare base operands.
2578 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
2581 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
2582 gpc_reg_operand instead of fpr_reg_operand.
2584 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
2586 * config/mips/mips.c (mips_return_in_memory): Force FP
2587 vector types to be returned in memory for o32 ABI.
2589 2017-02-22 Jakub Jelinek <jakub@redhat.com>
2591 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
2592 instead of DW_TAG_member for static data member declarations and don't
2593 set no_linkage_name for static inline data members.
2594 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
2597 2017-02-22 Martin Liska <mliska@suse.cz>
2599 * doc/invoke.texi: Replace inequality signs with square brackets
2602 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2604 PR tree-optimization/68644
2605 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
2607 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
2610 * lra-constraints.c (simplify_operand_subreg): Handle
2611 WORD_REGISTER_OPERATIONS targets.
2613 2017-02-22 Jakub Jelinek <jakub@redhat.com>
2616 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
2617 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
2618 elimination by swapping fld*.
2620 2017-02-22 Richard Biener <rguenther@suse.de>
2622 PR tree-optimization/79673
2623 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
2624 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
2625 irrelevant address-space qualifiers and avoiding a
2626 ADDR_SPACE_CONVERT_EXPR from fold_convert.
2628 2017-02-22 Richard Biener <rguenther@suse.de>
2630 PR tree-optimization/79666
2631 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
2632 to not symbolically negate if that may introduce undefined
2635 2017-02-22 Martin Liska <mliska@suse.cz>
2638 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
2639 * data-streamer-out.c (streamer_write_gcov_count_stream):
2641 * value-prof.c (stream_out_histogram_value): Make assert more
2642 precise based on type of counter.
2644 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
2647 * config/i386/i386.md (standard_x87sse_constant_load splitter):
2648 Use nonimmediate_operand instead of memory_operand for operand 1.
2649 (float-extend standard_x87sse_constant_load splitter): Ditto.
2651 2017-02-21 Jeff Law <law@redhat.com>
2653 PR tree-optimization/79621
2654 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
2655 blocks with edges to themselves.
2657 2017-02-21 Jakub Jelinek <jakub@redhat.com>
2660 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
2661 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
2662 Use gimple_call_builtin_p.
2665 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
2666 on temporarily removed DEBUG_INSNs.
2668 PR tree-optimization/79649
2669 * tree-loop-distribution.c (classify_partition): Give up on
2670 non-generic address space loads/stores.
2672 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
2674 * doc/loop.texi (Loop manipulation): Remove nonexistent
2675 tree_ssa_loop_version from the documentation.
2676 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
2678 2017-02-21 Jakub Jelinek <jakub@redhat.com>
2681 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
2682 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
2683 * config/rs6000/rs6000.c: Include except.h.
2684 (rs6000_expand_split_stack_prologue): Call
2685 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
2687 2017-02-21 Martin Jambor <mjambor@suse.cz>
2690 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
2693 2017-02-21 Martin Jambor <mjambor@suse.cz>
2695 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
2696 for backward compatibility only.
2697 * doc/invoke.texi (Option Summary): Remove all references to
2700 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
2704 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
2706 * lra-constraints.c (curr_insn_transform): Handle
2707 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
2709 2017-02-21 Martin Liska <mliska@suse.cz>
2711 * config/i386/i386.opt: Replace -masm-dialect with -masm.
2713 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
2715 PR translation/79638
2716 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
2718 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
2721 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
2722 (arm_function_ok_for_sibcall): Return false for an indirect call by
2723 descriptor if all the argument registers are used.
2724 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
2725 alignment of the function.
2727 2017-02-21 Jakub Jelinek <jakub@redhat.com>
2729 PR tree-optimization/61441
2730 * simplify-rtx.c (simplify_const_unary_operation): For
2731 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
2732 the sNaN unmodified.
2734 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
2736 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
2737 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
2738 instead of SYSTEM_HEADER_DIR.
2740 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
2741 Martin Liška <mliska@suse.cz>
2743 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
2744 Fix typos and grammar, use active voice, and clarify.
2746 2017-02-20 Marek Polacek <polacek@redhat.com>
2749 * gimplify.c (gimplify_expr): Handle unused *&&L;.
2752 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
2754 2017-02-20 Jakub Jelinek <jakub@redhat.com>
2757 * config/i386/i386.c (ix86_expand_builtin): Handle
2758 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
2759 ix86_builtins_isa[fcode].isa as a requirement of those
2760 flags and any other flag in the bitmask.
2761 (ix86_init_mmx_sse_builtins): Use 0 instead of
2762 ~OPTION_MASK_ISA_64BIT as mask.
2763 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
2764 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
2765 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
2766 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
2768 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
2771 * lra-constraints.c (split_reg): Check requested split mode
2772 is supported by the register.
2774 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
2776 * lra-constraints.c (simplify_operand_subreg): Remove early
2779 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
2782 * lra-constraints.c (curr_insn_transform): Tighten condition
2783 for converting SUBREG reloads from OP_OUT to OP_INOUT.
2785 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
2788 * lra-constraints.c (curr_insn_transform): Handle
2789 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
2791 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
2794 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
2796 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
2798 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
2801 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
2804 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2806 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
2809 2017-02-18 Jakub Jelinek <jakub@redhat.com>
2811 * final.c (last_columnnum, override_columnnum): New variables.
2812 (final_start_function): Set last_columnnum, pass it to begin_prologue
2813 hook and pass 0 to dwarf2out_begin_prologue.
2814 (final_scan_insn): Update override_columnnum. Pass last_columnnum
2815 to source_line debug hook.
2816 (notice_source_line): Compute last_columnnum and for debug_column_info
2817 return true on column changes.
2818 * debug.h (struct gcc_debug_hooks): Add column argument to
2819 source_line and begin_prologue hooks.
2820 (debug_nothing_int_charstar_int_bool): Remove prototype.
2821 (debug_nothing_int_int_charstar,
2822 debug_nothing_int_int_charstar_int_bool): New prototypes.
2823 (dwarf2out_begin_prologue): Add column argument.
2824 * debug.c (do_nothing_debug_hooks): Adjust source_line and
2825 begin_prologue hooks.
2826 (debug_nothing_int_charstar_int_bool): Remove.
2827 (debug_nothing_int_int_charstar,
2828 debug_nothing_int_int_charstar_int_bool): New functions.
2829 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
2830 through to dwarf2out_source_line.
2831 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
2832 (dwarf2out_source_line): Add column argument, emit it if requested.
2833 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
2835 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
2836 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
2837 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
2838 through to dwarf2out_begin_prologue.
2839 (vmsdbgout_source_line): Add column argument, pass it through to
2840 dwarf2out_source_line.
2841 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
2842 dbxout_source_line caller.
2843 (dbxout_source_line): Add column argument.
2845 * common.opt (gno-column-info, gcolumn-info): New options.
2846 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
2847 (check_die): Also test for multiple DW_AT_decl_column attributes.
2848 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
2849 DW_AT_decl_column if requested.
2850 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
2852 (gen_variable_die): Likewise.
2853 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
2854 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
2857 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
2858 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
2859 (ix86_handle_option): Handle OPT_m3dnowa.
2860 * doc/invoke.texi (-m3dnowa): Document.
2861 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
2862 -m3dnowa instead of -m3dnow -march=athlon.
2865 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
2866 instead of gcc_assert for K, r and R code checks. Formatting fixes.
2868 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2871 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
2872 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
2873 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
2874 generator for vsx_xxpermdi_<mode>_be.
2875 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
2876 force big-endian semantics.
2877 (vsx_xxpermdi_<mode>_be): New define_expand with same
2878 implementation as previous version of vsx_xxpermdi_<mode>.
2880 2017-02-17 Jakub Jelinek <jakub@redhat.com>
2882 PR tree-optimization/79327
2883 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
2884 variable, its initialization and use.
2886 2017-02-17 Julia Koval <julia.koval@intel.com>
2888 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
2889 (OPTION_MASK_ISA_PKU_UNSET): New.
2890 (ix86_handle_option): Handle -mrdpid.
2891 * config/i386/cpuid.h (bit_RDPID): New.
2892 * config/i386/driver-i386.c (host_detect_local_cpu):
2893 Detect RDPID feature.
2894 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
2895 * config/i386/i386-c.c (ix86_target_macros_internal):
2897 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
2898 (ix86_valid_target_attribute_inner_p): Add "rdpid".
2899 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
2900 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
2901 * config/i386/i386.md (define_insn "rdpid"): New.
2902 * config/i386/i386.opt Add -mrdpid.
2903 * config/i386/immintrin.h (_rdpid_u32): New.
2905 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
2907 PR rtl-optimization/79541
2908 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
2909 instead of transforming it into USE.
2911 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
2913 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
2914 If HONOR_SNANS (SFmode) force the input to a register.
2915 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
2916 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
2917 an frsp or similar insn.
2919 2017-02-17 Martin Liska <mliska@suse.cz>
2921 PR rtl-optimization/79577
2922 * params.def (selsched-max-sched-times): Increase minimum to 1.
2924 2017-02-17 Martin Liska <mliska@suse.cz>
2926 PR rtl-optimization/79574
2927 * gcse.c (want_to_gcse_p): Prevent integer overflow.
2929 2017-02-17 Martin Liska <mliska@suse.cz>
2931 PR tree-optimization/79529
2932 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
2933 ssa_defined_default_def_p to handle cases which are implicitly
2935 * tree-ssa.c (ssa_defined_default_def_p): New function.
2936 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
2937 which are implicitly defined.
2938 * tree-ssa.h (ssa_defined_default_def_p): Declare.
2940 2017-02-17 Richard Biener <rguenther@suse.de>
2943 * params.def (max-ssa-name-query-depth): Limit to 10.
2945 2017-02-17 Richard Biener <rguenther@suse.de>
2947 PR tree-optimization/79552
2948 * tree-ssa-structalias.c (visit_loadstore): Properly verify
2951 2017-02-17 Richard Biener <rguenther@suse.de>
2954 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
2956 2017-02-17 Marek Polacek <polacek@redhat.com>
2959 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
2960 (fold_negate_expr): New wrapper.
2962 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
2964 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
2965 Correct terminology and de-emphasize pre-standard behavior.
2967 2017-02-16 Alan Modra <amodra@gmail.com>
2969 PR rtl-optimization/79286
2970 * ira.c (def_dominates_uses): New function.
2971 (update_equiv_regs): Don't create an equivalence for insns that
2972 may trap where the register def does not dominate the use.
2974 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
2976 PR rtl-optimization/78127
2977 * lra.c (lra): Call lra_eliminate before finish the loop after
2980 2017-02-16 Richard Biener <rguenther@suse.de>
2982 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
2984 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
2985 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
2986 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
2987 (isl_val_int_from_wi): New function.
2988 (extract_affine_gmp): Rename to ...
2989 (extract_affine_wi): ... this, take a widest_int.
2990 (extract_affine_int): Just wrap extract_affine_wi.
2991 (add_param_constraints): Use isl_val_int_from_wi.
2992 (add_loop_constraints): Likewise, and extract_affine_wi.
2994 2017-02-15 Jeff Law <law@redhat.com>
2997 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
2998 ira_init_register_move_cost_if_necessary.
3000 2017-02-15 Martin Sebor <msebor@redhat.com>
3003 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
3004 removed in a prior commit.
3006 2017-02-15 Bin Cheng <bin.cheng@arm.com>
3008 PR tree-optimization/79347
3009 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
3010 counters during peeling.
3012 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
3014 * Makefile.in (site.exp): Remove "set ISLVER".
3016 2017-02-15 Jakub Jelinek <jakub@redhat.com>
3019 * real.c (real_from_integer): Call real_convert even for decimal.
3021 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
3024 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
3026 2017-02-14 Andrew Pinski <apinski@cavium.com>
3028 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
3029 cores and change the partno/implementer to be correct.
3030 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
3031 the 'B" as the implementer.
3032 * config/aarch64/aarch64-tune.md: Regenerate.
3034 2017-02-14 Carl Love <cel@us.ibm.com>
3036 * config/rs6000/rs6000.c: Add case statement entry to make the
3037 xvcvuxdsp built-in argument unsigned.
3038 * config/rs6000/vsx.md: Fix the source and return operand types so they
3039 match the instruction definitions from the ISA document. Fix typo
3040 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
3043 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
3046 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
3047 member early_clobber_alts.
3048 * lra-lives.c (reg_early_clobber_p): New.
3049 (process_bb_lives): Use it.
3050 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
3051 (debug_operand_data): Initialize early_clobber_alts.
3052 (setup_operand_alternative): Set up early_clobber_alts.
3053 (collect_non_operand_hard_regs): Ditto. Pass early clobber
3054 alternatives to new_insn_reg.
3055 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
3057 (lra_update_insn_regno_info): Pass the new arg.
3059 2017-02-14 Jakub Jelinek <jakub@redhat.com>
3062 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
3063 (new_oacc_loop_raw): Don't clear already cleared fields.
3066 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
3067 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
3068 _mm512_prefetch_i64gather_ps): New inline functions and macros.
3070 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
3073 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
3075 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
3078 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
3079 the extra instruction to the right place to store 128-bit constant
3082 2017-02-14 Martin Sebor <msebor@redhat.com>
3085 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
3086 warning for strings of unknown length.
3088 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
3090 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
3092 2017-02-14 Jeff Law <law@redhat.com>
3095 * ira-costs.c (scan_one_insn): Initialize register move costs
3096 for pseudos seen in USE/CLOBBER insns.
3098 PR tree-optimization/79095
3099 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
3100 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
3101 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
3102 if the operands are known to be not equal, then the resulting range
3104 (intersect_ranges): If the new range is ~[0,0] and the old range is
3105 wide, then prefer ~[0,0].
3106 * tree-vrp.c (overflow_comparison_p_1): New function.
3107 (overflow_comparison_p): New function.
3108 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
3109 if NAME is used in an overflow test.
3110 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
3111 overflow check that can be expressed as an equality test, then adjust
3112 ops to be that equality test.
3114 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3116 * config/s390/s390-builtin-types.def: Remove flags argument.
3117 * config/s390/s390.c (s390_init_builtins): Likewise.
3119 2017-02-14 Martin Liska <mliska@suse.cz>
3121 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
3122 vector. Fix trailing white spaces.
3124 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
3126 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
3129 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3131 PR rtl-optimization/68664
3132 * config/arm/arm.c (arm_sched_can_speculate_insn):
3133 New function. Declare prototype.
3134 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
3136 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3138 PR rtl-optimization/68664
3139 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
3141 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
3143 2017-02-14 Amit Pawar <amit.pawar@amd.com>
3145 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
3146 max skip bytes for function, loop and jump.
3148 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3150 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
3151 ABS_EXPR for gimple dump.
3153 2017-02-14 Jakub Jelinek <jakub@redhat.com>
3156 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
3158 PR tree-optimization/79408
3159 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
3160 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
3161 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
3162 also if rhs1 is INTEGER_CST.
3164 2017-02-14 Richard Biener <rguenther@suse.de>
3167 * tree-into-ssa.c (insert_phi_nodes): When the function can
3168 have abnormal edges rewrite SSA names with broken use-def
3169 dominance out of SSA and register them for PHI insertion.
3171 2017-02-13 Martin Sebor <msebor@redhat.com>
3174 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
3175 clearing info.nowrite flag when snprintf size argument is a range.
3177 2017-02-13 Jakub Jelinek <jakub@redhat.com>
3179 * cprop.c (cprop_jump): Add missing space in string literal.
3180 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
3181 (get_constraint_for_component_ref): Likewise.
3182 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
3183 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
3184 * lra-constraints.c (process_alt_operands): Likewise.
3185 * ipa-inline.c (inline_small_functions): Likewise.
3186 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
3187 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
3188 * trans-mem.c (diagnose_tm_1_op): Likewise.
3189 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
3190 (grid_parallel_clauses_gridifiable): Likewise.
3192 * config/nvptx/mkoffload.c (process): Add space in between
3195 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
3196 "MOD4_SSE_REGS" and "ALL_REGS".
3198 * spellcheck.c (test_data): Add , in between "foo" and "food".
3200 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3203 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
3204 boundary crossing check and subsequent code generation agree.
3206 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3208 * config/aarch64/aarch64.c (has_memory_op): Delete.
3209 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
3212 2017-02-13 Jakub Jelinek <jakub@redhat.com>
3214 PR rtl-optimization/79388
3215 PR rtl-optimization/79450
3216 * combine.c (distribute_notes): When removing TEM_INSN for which
3217 corresponding dest has last value recorded, invalidate that last
3220 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3222 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
3223 of explicit '@'. Add missing assembly comment marker on branch costs
3226 2017-02-13 Nathan Sidwell <nathan@acm.org>
3228 * gengtype-lex.l (<in_struct>): Add '/'.
3230 2017-02-13 Martin Liska <mliska@suse.cz>
3233 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
3235 2017-02-13 Richard Biener <rguenther@suse.de>
3237 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
3239 * configure: Re-generate.
3240 * config.in: Likewise.
3241 * graphite-dependences.c: Simplify as if
3242 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
3243 * graphite-isl-ast-to-gimple.c: Likewise.
3244 * graphite-optimize-isl.c: Likewise.
3245 * graphite-poly.c: Likewise.
3246 * graphite-sese-to-poly.c: Likewise.
3247 * graphite.h: Likewise.
3248 * toplev.c: Include isl/version.h and use isl_version () for
3249 printing the ISL version.
3250 * doc/install.texi: Update ISL requirement.
3252 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
3254 * doc/standards.texi (Standards): Update reference to
3257 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
3259 * doc/extend.texi (Named Address Spaces): sourceware.org now
3261 * doc/install.texi (Binaries): Ditto.
3264 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
3266 * doc/cpp.texi: Replace "stringify"/"stringification" with C
3267 standard terminology "stringize"/"stringizing" throughout.
3268 * doc/cppinternals.texi: Likewise.
3270 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
3272 * doc/extend.texi: Fix some spelling mistakes and typos.
3273 * doc/invoke.texi: Likewise.
3275 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
3278 * params.def (inline-min-speedup) Change from 10 to 8.
3280 2017-02-11 Jakub Jelinek <jakub@redhat.com>
3282 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
3285 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
3288 * ipa-inline-analysis.c (get_minimal_bb): New function.
3289 (record_modified): Use it.
3290 (remap_edge_change_prob): Handle also ancestor functions.
3292 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
3294 * doc/contrib.texi (Contributors): Remove broken link into
3295 the Mauve CVS repository.
3297 2017-02-11 Jakub Jelinek <jakub@redhat.com>
3300 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
3301 result computation whenever lhs doesn't have vector mode, not
3302 just when it has BLKmode.
3304 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
3306 * doc/makefile.texi (profiledbootstrap): Refer to the
3307 installation instructions only in textual form.
3309 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3312 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
3314 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
3316 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
3317 (Specific): Update mingw-w64 reference.
3319 (Specific): Remove broken link to Renesas RX processor.
3321 2017-02-10 Richard Biener <rguenther@suse.de>
3323 * toplev.c (process_options): Do not mention obsolete graphite
3324 options when printing sorry message about missing graphite support.
3325 Mention -floop-nest-optimize.
3327 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
3329 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
3330 (vtst_p16): Likewise.
3331 (vtstq_p8): Likewise.
3332 (vtstq_p16): Likewise.
3334 (vtstq_p64): Likewise.
3335 * config/arm/arm_neon.h (vgetq_lane_p64): New.
3336 (vset_lane_p64): New.
3337 (vsetq_lane_p64): New.
3339 2017-02-10 Jakub Jelinek <jakub@redhat.com>
3341 PR tree-optimization/79411
3342 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
3343 stmt operands are SSA_NAMEs used in abnormal phis.
3344 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
3347 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
3350 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
3353 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
3355 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
3357 2017-02-09 Jakub Jelinek <jakub@redhat.com>
3359 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
3363 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
3364 not arbitrary TREE_CONSTANT.
3367 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
3368 "omp declare target link" attribute unless is_global_var.
3369 * omp-offload.c (find_link_var_op): Likewise.
3371 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
3372 Chung-Lin Tang <cltang@codesourcery.com>
3374 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
3376 (gimplify_adjust_omp_clauses): Don't delete TILE.
3377 (gimplify_omp_for): Deal with TILE.
3378 * internal-fn.c (expand_GOACC_TILE): New function.
3379 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
3381 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
3382 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
3384 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
3385 avoid DIV for outermost collapse var.
3386 (expand_oacc_for): Insert tile element loop as needed. Adjust.
3387 Remove out of date comments, fix whitespace.
3388 * omp-general.c (omp_extract_for_data): Deal with tiling.
3389 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
3390 adjust OLF_DIM_BASE value.
3391 (struct omp_for_data): Add tiling field.
3392 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
3393 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
3394 for auto loops. Remove default auto determining, moved to
3395 oacc_loop_fixed_partitions.
3396 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
3397 stmts, add e_mask field.
3398 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
3399 (oacc_thread_numbers): Use oacc_dim_call.
3400 (oacc_xform_tile): New.
3401 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
3402 (finish_oacc_loop): Adjust for ifns vector.
3403 (oacc_loop_discover_walk): Append loop abstraction sites to list,
3404 add case for GOACC_TILE fns.
3405 (oacc_loop_xform_loop): Delete.
3406 (oacc_loop_process): Iterate over call list directly, and add
3407 handling for GOACC_TILE fns.
3408 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
3410 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
3411 vector partitioning to outer loops. Assign 2 partitions to loops
3412 when available. Add TILE handling.
3413 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
3414 (execite_oacc_device_lower): Process GOACC_TILE fns, ignore unknown specs.
3415 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
3416 * tree.c (omp_clause_num_ops): Adjust TILE ops.
3417 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
3419 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
3421 * configure.ac (ACX_BUGURL): Update.
3422 * configure: Regenerate.
3424 2017-02-09 Richard Biener <rguenther@suse.de>
3426 PR tree-optimization/69823
3427 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
3428 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
3430 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
3432 * config/arc/arc-c.def: Add __NPS400__ definition.
3433 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
3434 (TARGET_NPS400): Define.
3436 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
3438 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
3440 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
3442 (arc_cpu_types): Fill the arch_info field with a pointer into the
3443 arc_arch_types table.
3444 (arc_selected_cpu): Declare.
3445 * config/arc/arc.c (arc_selected_cpu): Make global.
3446 (arc_selected_arch): Delete.
3447 (arc_base_cpu): Delete.
3448 (arc_override_options): Remove references to deleted variables,
3449 update access to arch information.
3450 (ARC_OPT): Update access to arch information.
3451 (ARC_OPTX): Likewise.
3452 * config/arc/arc.h (arc_base_cpu): Remove declaration.
3453 (TARGET_ARC600): Update access to arch information.
3454 (TARGET_ARC601): Likewise.
3455 (TARGET_ARC700): Likewise.
3456 (TARGET_EM): Likewise.
3457 (TARGET_HS): Likewise.
3458 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
3461 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
3464 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
3465 condition/operands for integer GE/LE/GEU/LEU operations.
3467 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
3469 PR translation/79397
3470 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
3473 2017-02-08 Martin Jambor <mjambor@suse.cz>
3476 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
3477 whether allocation happened.
3478 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
3479 nothing was allocated.
3481 2017-02-08 Jakub Jelinek <jakub@redhat.com>
3483 PR tree-optimization/79408
3484 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
3485 constant, but SSA_NAME with a known integer range, use the minimum
3486 of that range instead of op1 to determine if modulo can be replaced
3487 with its first operand.
3489 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3491 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
3493 2017-02-08 Richard Biener <rguenther@suse.de>
3495 PR tree-optimization/71824
3496 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
3497 Check all loops contained in the merged region.
3499 2017-02-07 Andrew Pinski <apinski@cavium.com>
3501 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
3503 2017-02-07 Andrew Pinski <apinski@cavium.com>
3505 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
3506 (thunderxt88): Likewise.
3507 (thunderxt81): Disable LSE and change v8.1 to v8.
3508 (thunderxt83): Likewise.
3510 2017-02-07 Jakub Jelinek <jakub@redhat.com>
3511 Richard Biener <rguenther@suse.de>
3514 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
3515 type from int to size_t.
3516 * ira-costs.c (struct_costs_size): Change type from int to size_t.
3518 2017-02-07 Jakub Jelinek <jakub@redhat.com>
3520 PR rtl-optimization/79386
3521 * cprop.c (bypass_conditional_jumps): Initialize
3522 bypass_last_basic_block already before splitting bbs after
3523 unconditional traps...
3524 (bypass_conditional_jumps): ... rather than here.
3527 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
3528 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
3529 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
3530 fix -masm=intel patterns.
3532 2017-02-07 Richard Biener <rguenther@suse.de>
3534 PR tree-optimization/79256
3536 * builtins.c (get_object_alignment_2): Use min_align_of_type
3537 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
3538 and ADJUST_FIELD_ALIGN.
3540 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
3542 * doc/tm.texi: Regenerate.
3543 * stor-layout.c (layout_decl): Adjust.
3544 (update_alignment_for_field): Likewise.
3545 (place_field): Likewise.
3546 (min_align_of_type): Likewise.
3547 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
3548 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
3549 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
3550 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
3551 * config/frv/frv.c (frv_adjust_field_align): Likewise.
3552 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
3553 * config/i386/i386.c (x86_field_alignment): Likewise.
3554 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
3555 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
3556 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
3557 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
3558 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
3559 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
3563 2017-01-30 Richard Biener <rguenther@suse.de>
3565 PR tree-optimization/79256
3566 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
3567 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
3570 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
3572 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
3573 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
3574 builtins to SImode and emit a zero-extend, if necessary.
3576 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
3578 * docs/invoke.texi (RISC-V Options): Alphabetize.
3580 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
3582 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
3585 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
3587 * config/riscv/riscv.c: New file.
3588 * gcc/common/config/riscv/riscv-common.c: Likewise.
3589 * config.gcc: Likewise.
3590 * config/riscv/constraints.md: Likewise.
3591 * config/riscv/elf.h: Likewise.
3592 * config/riscv/generic.md: Likewise.
3593 * config/riscv/linux.h: Likewise.
3594 * config/riscv/multilib-generator: Likewise.
3595 * config/riscv/peephole.md: Likewise.
3596 * config/riscv/pic.md: Likewise.
3597 * config/riscv/predicates.md: Likewise.
3598 * config/riscv/riscv-builtins.c: Likewise.
3599 * config/riscv/riscv-c.c: Likewise.
3600 * config/riscv/riscv-ftypes.def: Likewise.
3601 * config/riscv/riscv-modes.def: Likewise.
3602 * config/riscv/riscv-opts.h: Likewise.
3603 * config/riscv/riscv-protos.h: Likewise.
3604 * config/riscv/riscv.h: Likewise.
3605 * config/riscv/riscv.md: Likewise.
3606 * config/riscv/riscv.opt: Likewise.
3607 * config/riscv/sync.md: Likewise.
3608 * config/riscv/t-elf-multilib: Likewise.
3609 * config/riscv/t-linux: Likewise.
3610 * config/riscv/t-linux-multilib: Likewise.
3611 * config/riscv/t-riscv: Likewise.
3612 * configure.ac: Likewise.
3613 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
3614 Waterman as RISC-V maintainers.
3615 * doc/install.texi: Add RISC-V entries.
3616 * doc/invoke.texi: Add RISC-V options section.
3617 * doc/md.texi: Add RISC-V constraints section.
3618 * configure: Regenerated.
3620 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
3623 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
3624 false values to be constant vectors with all 0 or all 1 bits set.
3625 (vcondu<mode><mode>): Likewise.
3626 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
3628 (fpmask_comparison_operator): Update comment.
3629 (vecint_comparison_operator): New predicate.
3630 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
3631 vector conditionals when the true and false values are constant
3632 vectors with all 0 bits or all 1 bits set.
3634 2017-02-06 Martin Sebor <msebor@redhat.com>
3636 PR tree-optimization/79376
3637 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
3639 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
3641 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
3642 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
3643 to simplify split condition.
3645 2017-02-06 Jakub Jelinek <jakub@redhat.com>
3647 * omp-expand.c (oxpand_omp_atomic_fetch_op,
3648 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
3651 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
3653 PR rtl-optimization/68664
3654 * target.def (can_speculate_insn): New hook.
3655 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
3656 * doc/tm.texi: Regenerate.
3657 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
3658 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
3659 (rs6000_sched_can_speculate_insn): New function.
3661 2017-02-06 Jakub Jelinek <jakub@redhat.com>
3663 PR tree-optimization/79284
3664 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
3665 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
3666 vectorizable_mask_load_store, vectorizable_operation,
3667 vect_is_simple_cond, get_same_sized_vectype): Use it instead
3668 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
3669 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
3670 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
3671 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
3672 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
3673 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
3674 is_gimple_assign (stmt). Replace another such test with
3675 is_gimple_assign (stmt).
3677 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
3680 * config/avr/avr.c (rtl-iter.h): Include it.
3681 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
3682 (avr_legitimate_combined_insn): ...and implementation.
3684 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
3686 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
3687 * config/s390/s390.c (s390_const_operand_ok)
3688 (s390_canonicalize_comparison, s390_extract_part)
3689 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
3690 (s390_contiguous_bitmask_p, s390_rtx_costs)
3691 (legitimize_pic_address): Likewise.
3692 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
3693 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
3694 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
3695 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
3696 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
3698 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
3700 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
3701 REGNO($0) == REGNO($1).
3703 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3705 * config/s390/linux.h(SIZE_TYPE): Add comment.
3707 2017-02-06 Julian Brown <julian@codesourcery.com>
3708 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
3709 Virendra Pathak <virendra.pathak@broadcom.com>
3711 * config/aarch64/aarch64-cores.def: Change the scheduler
3713 * config/aarch64/aarch64.md: Include thunderx2t99.md.
3714 * config/aarch64/thunderx2t99.md: New file.
3716 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
3718 * doc/standards.texi (Go Language): Update link to language
3721 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
3723 * tree-eh.c (lower_resx): Sanitize profile.
3724 (cleanup_empty_eh_move_lp): Likewise.
3726 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
3729 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
3731 * cfgloopmanip.h (loop_version): Update prototype.
3732 * modulo-sched.c (sms_schedule): Update call of loop_version.
3733 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
3734 * tree-parloops.c (gen_parallel_loop): Likewise.
3735 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
3736 * tree-ssa-loop-split.c (split_loop): Likewise.
3737 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
3738 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
3740 2017-02-05 Martin Liska <mliska@suse.cz>
3743 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
3745 (print_operand_address): Initialize a struct to zero.
3747 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
3749 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
3750 garbage collector only in textual form.
3752 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
3754 * doc/extend.texi (x86 specific memory model extensions for
3755 transactional memory): Simplify a phrase.
3757 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
3760 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
3761 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
3762 (atomic_storedi_1): Likewise.
3764 2017-02-04 Jakub Jelinek <jakub@redhat.com>
3766 PR tree-optimization/79338
3767 * tree-parloops.c (gather_scalar_reductions): Don't call
3768 vect_analyze_loop_form for loop->inner before destroying loop's
3771 2017-02-03 Martin Sebor <msebor@redhat.com>
3773 PR tree-optimization/79327
3774 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
3775 when precision has resulted in leading zeros.
3776 (format_integer): Adjust the likely counter to assume an unknown
3777 argument that may be zero is non-zero.
3779 2017-02-03 Jason Merrill <jason@redhat.com>
3782 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
3783 avoid copying non-taken branch.
3785 2017-02-03 Jakub Jelinek <jakub@redhat.com>
3787 PR tree-optimization/79340
3788 * tree-vect-loop.c (vectorizable_reduction): Release
3789 vec_defs elements after safe_splicing them into other vectors.
3792 PR tree-optimization/79327
3793 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
3794 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
3796 (format_integer): Use wide_int_to_tree instead of build_int_cst
3797 + to_?hwi. If argmin is NULL, just set argmin and argmax to
3798 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
3799 of shortest and longest sequence.
3801 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
3803 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
3804 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
3806 2017-02-03 Walter Lee <walt@tilera.com>
3809 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
3810 after initial stackframe link reg save.
3811 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
3813 2017-02-03 Jakub Jelinek <jakub@redhat.com>
3816 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
3817 wu for stxssp alternative.
3819 2017-02-03 Martin Sebor <msebor@redhat.com>
3821 PR tree-optimization/79352
3822 * gimple-fold.c (get_range_strlen): Add argument.
3823 (get_range_strlen): Change return type to bool.
3824 (get_maxval_strlen): Pass in a dummy argument.
3825 * gimple-fold.h (get_range_strlen): Change return type to bool.
3826 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
3827 * tree.h (array_at_struct_end_p): Add argument.
3828 * tree.c (array_at_struct_end_p): Handle it.
3830 2017-02-03 Martin Liska <mliska@suse.cz>
3833 * multiple_target.c (create_dispatcher_calls): Redirect edge
3834 from a caller of a dispatcher.
3835 (expand_target_clones): Make the clones local.
3836 (ipa_target_clone): Do both target clones and resolvers.
3837 (ipa_dispatcher_calls): Remove the pass.
3838 (pass_dispatcher_calls::gate): Likewise.
3839 (make_pass_dispatcher_calls): Likewise.
3840 * passes.def (pass_target_clone): Put as very first IPA early
3843 2017-02-03 Martin Liska <mliska@suse.cz>
3845 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
3846 in case of a function with ifunc attribute.
3848 2017-02-03 Martin Liska <mliska@suse.cz>
3850 * cgraph.c (cgraph_node::dump): Dump function version info.
3851 * symtab.c (symtab_node::dump_base): Add missing new line.
3853 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
3855 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
3856 (ifcombine_ifandif): Use it.
3858 2017-02-03 Martin Liska <mliska@suse.cz>
3860 * doc/invoke.texi: Document default value for
3861 use-after-scope-direct-emission-threshold.
3863 2017-02-03 Martin Liska <mliska@suse.cz>
3865 PR tree-optimization/79339
3866 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
3867 (format_floating): Likewise.
3869 2017-02-03 Martin Liska <mliska@suse.cz>
3872 * ipa-prop.c (ipa_node_params_t::insert): Remove current
3874 (ipa_node_params_t::remove): Likewise.
3875 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
3876 initialization from removed ipa_node_params_t::insert.
3877 (ipa_node_params::~ipa_node_params): Move from removed
3878 ipa_node_params_t::release.
3879 * symbol-summary.h (symbol_summary::m_released): New member.
3880 Do not release a summary twice. Do not allow to call finalizer
3881 for types of a summary that live in GGC memory.
3883 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
3885 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
3888 2017-02-02 Martin Sebor <msebor@redhat.com>
3891 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
3892 (format_string): Tighten up the range of output for non-constant
3893 strings and correct the expected range for wide non-constant strings.
3895 2017-02-02 Martin Sebor <msebor@redhat.com>
3897 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
3900 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
3902 (-fdump-tree-@var): Add to index and document how to come up
3903 with pass-specific option and dump file names.
3904 (-fdump-passes): Clarify where to look for output.
3906 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
3909 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
3910 statistics of the analyzed path; allow threading for speed when
3911 any of BBs along the path are optimized for speed.
3913 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
3916 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
3917 settings of the virtual registers.
3920 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
3922 * explow.c (get_dynamic_stack_size): Take known alignment of stack
3923 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
3926 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3928 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
3929 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
3931 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3933 * config/s390/s390.md: Add missing comments with the expanded
3935 * config/s390/vector.md: Likewise.
3936 * config/s390/vx-builtins.md: Likewise.
3938 2017-02-02 Jakub Jelinek <jakub@redhat.com>
3941 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
3942 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
3943 conditions on a single line.
3945 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3947 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
3948 __S390_VX__ to __VX__.
3950 2017-02-01 Andrew Pinski <apinski@cavium.com>
3952 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
3953 stmt_info to record_stmt_cost.
3954 (vect_get_known_peeling_cost): Pass stmt_info if known to
3956 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
3957 cpu_vector_cost field into
3958 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
3959 field into vec_int_stmt_cost and vec_fp_stmt_cost.
3960 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
3961 splitting of scalar_stmt_cost and vec_stmt_cost.
3962 (thunderx_vector_cost): Likewise.
3963 (cortexa57_vector_cost): LIkewise.
3964 (exynosm1_vector_cost): Likewise.
3965 (xgene1_vector_cost): Likewise.
3966 (thunderx2t99_vector_cost): Improve after the splitting of the two
3968 (aarch64_builtin_vectorization_cost): Update for the splitting of
3969 scalar_stmt_cost and vec_stmt_cost.
3971 2017-02-01 Torvald Riegel <triegel@redhat.com>
3972 Richard Henderson <rth@redhat.com>
3974 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
3975 conditional on existance of a fast atomic load.
3976 * optabs-query.c (can_atomic_load_p): New function.
3977 * optabs-query.h (can_atomic_load_p): Declare it.
3978 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
3979 no fast atomic load is available for the particular size of access.
3980 (expand_atomic_compare_and_swap): Likewise.
3981 (expand_atomic_load): Likewise.
3982 (expand_atomic_store): Likewise.
3983 (expand_atomic_fetch_op): Likewise.
3984 * testsuite/lib/target-supports.exp
3985 (check_effective_target_sync_int_128): Remove x86 because it provides
3986 no fast atomic load.
3987 (check_effective_target_sync_int_128_runtime): Likewise.
3989 2017-02-01 Richard Biener <rguenther@suse.de>
3991 * graphite.c: Include tree-vectorizer.h for find_loop_location.
3992 (graphite_transform_loops): Provide opt-info for optimized nests.
3993 * tree-parloop.c (parallelize_loops): Provide opt-info for
3996 2017-02-01 Richard Biener <rguenther@suse.de>
3999 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
4002 2017-02-01 Richard Biener <rguenther@suse.de>
4004 PR tree-optimization/71824
4005 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
4006 Verify the loops are valid in the merged SESE region.
4007 (scop_detection::can_represent_loop_1): Check analyzing the
4008 evolution of the number of iterations in the region succeeds.
4010 2017-01-31 Ian Lance Taylor <iant@golang.org>
4012 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
4013 REG_ARGS_SIZE note to 32-bit push insns and call insn.
4015 2017-01-31 David Malcolm <dmalcolm@redhat.com>
4017 PR preprocessor/79210
4018 * input.c (get_substring_ranges_for_loc): Replace line_width
4019 assertion with error-handling.
4021 2017-01-31 Richard Biener <rguenther@suse.de>
4023 PR tree-optimization/77318
4024 * graphite-sese-to-poly.c (extract_affine): Fix assert.
4025 (create_pw_aff_from_tree): Take loop parameter.
4026 (add_condition_to_pbb): Pass loop of the condition to
4027 create_pw_aff_from_tree.
4029 2017-01-31 Jakub Jelinek <jakub@redhat.com>
4031 * config/s390/s390.c (s390_asan_shadow_offset): New function.
4032 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
4034 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
4038 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
4040 (convert_int_to_float128): Likewise.
4041 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
4042 (convert_int_to_float128): Likewise.
4043 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
4044 (UNSPEC_IEEE128_CONVERT): Likewise.
4045 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
4046 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
4047 Use local variables for IBM extended format.
4048 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
4049 (fix_trunc<mode>si2_fprs): Likewise.
4050 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
4051 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
4052 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
4053 to know that we can now have integers of all sizes in vector
4055 (fix<uns>_<mode>di2_hw): Likewise.
4056 (float<uns>_<mode>si2_hw): Likewise.
4057 (fix_<mode>si2_hw): Likewise.
4058 (fixuns_<mode>si2_hw): Likewise.
4059 (float<uns>_<mode>di2_hw): Likewise.
4060 (float_<mode>di2_hw): Likewise.
4061 (float_<mode>si2_hw): Likewise.
4062 (floatuns_<mode>di2_hw): Likewise.
4063 (floatuns_<mode>si2_hw): Likewise.
4064 (xscvqp<su>wz_<mode>): Delete, no longer used.
4065 (xscvqp<su>dz_<mode>): Likewise.
4066 (xscv<su>dqp_<mode>): Likewise.
4067 (ieee128_mfvsrd_64bit): Likewise.
4068 (ieee128_mfvsrd_32bit): Likewise.
4069 (ieee128_mfvsrwz): Likewise.
4070 (ieee128_mtvsrw): Likewise.
4071 (ieee128_mtvsrd_64bit): Likewise.
4072 (ieee128_mtvsrd_32bit): Likewise.
4074 2017-01-31 Martin Liska <mliska@suse.cz>
4077 * ipa-prop.c (ipa_free_all_node_params): Call release method
4078 instead of ~sumbol_summary to not to trigger double times
4081 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
4083 PR tree-optimization/71691
4084 * bitmap.h (class auto_bitmap): New.
4085 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
4086 is_maybe_undefined instead of ssa_undefined_value_p.
4088 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4090 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
4091 __S390_ARCH_LEVEL__ to __ARCH__.
4093 2017-01-31 Jakub Jelinek <jakub@redhat.com>
4095 PR tree-optimization/79267
4096 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
4097 if should_remove_lhs_p is true.
4099 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
4102 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
4103 (add_alignment_attribute): New.
4104 (base_type_die): Add alignment attribute.
4105 (subrange_type_die): Likewise.
4106 (modified_type_die): Likewise.
4107 (gen_array_type_die): Likewise.
4108 (gen_descr_array_type_die: Likewise.
4109 (gen_enumeration_type_die): Likewise.
4110 (gen_subprogram_die): Likewise.
4111 (gen_variable_die): Likewise.
4112 (gen_field_die): Likewise.
4113 (gen_ptr_to_mbr_type_die): Likewise.
4114 (gen_struct_or_union_type_die): Likewise.
4115 (gen_subroutine_type_die): Likewise.
4116 (gen_typedef_die): Likewise.
4117 (base_type_cmp): Compare alignment attribute.
4119 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
4122 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
4123 (setb_unsigned) New pattern for setb with CCUNS.
4124 * config/rs6000/rs6000.c (expand_block_compare): Use a different
4125 subfc./subfe sequence to avoid overflow problems. Generate a
4126 shorter sequence with cmpld/setb for power9.
4127 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
4128 for generating subfc. instruction.
4129 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
4130 now uses this instruction.
4132 2017-01-30 Ian Lance Taylor <iant@google.com>
4135 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
4136 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
4138 2017-01-30 Martin Sebor <msebor@redhat.com>
4140 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
4141 Move constant to the right of a relational operator.
4142 (get_mpfr_format_length, format_character, format_string): Ditto.
4143 (should_warn_p, maybe_warn): Same.
4145 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
4147 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
4150 * asan.c (get_translation_unit_decl): Remove function.
4151 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
4153 2017-01-30 Martin Liska <mliska@suse.cz>
4155 PR gcov-profile/79259
4156 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
4159 2017-01-30 Martin Liska <mliska@suse.cz>
4162 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
4163 Initialize variables with NULL value.
4165 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
4168 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
4170 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
4172 2017-01-30 Richard Biener <rguenther@suse.de>
4174 PR tree-optimization/79276
4175 * tree-vrp.c (process_assert_insertions): Properly adjust common
4176 when removing a duplicate.
4178 * gcc.dg/torture/pr79276.c: New testcase.
4180 2017-01-30 Richard Biener <rguenther@suse.de>
4182 PR tree-optimization/79256
4183 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
4184 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
4186 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
4188 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
4191 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
4192 ("*r<noxa>sbg_<mode>_sll_bitmask")
4193 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
4194 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
4195 Use contiguous_bitmask_nowrap_operand.
4197 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4200 * config/rs6000/altivec.h (vec_xl): Revise #define.
4201 (vec_xst): Likewise.
4203 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
4205 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
4207 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
4209 PR rtl-optimization/79194
4210 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
4211 traps before call to bypass_conditional_jumps.
4213 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
4215 PR tree-optimization/71374
4216 * lra-constraints.c (check_conflict_input_operands): New.
4217 (match_reload): Use it.
4219 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
4222 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
4223 account to calculate conflict_set.
4225 2017-01-27 Bin Cheng <bin.cheng@arm.com>
4227 PR rtl-optimization/78559
4228 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
4229 other_insn in combine.
4231 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
4233 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
4234 uint16_type_node for BT_UINT16.
4236 2017-01-27 David Malcolm <dmalcolm@redhat.com>
4238 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
4239 "RTL Tests" to menu.
4240 (GIMPLE Tests): New node.
4241 (RTL Tests): New node.
4243 2017-01-27 Richard Biener <rguenther@suse.de>
4245 PR tree-optimization/79245
4246 * tree-loop-distribution.c (distribute_loop): Apply cost
4247 modeling also to detected patterns.
4249 2017-01-27 Richard Biener <rguenther@suse.de>
4251 PR tree-optimization/71433
4252 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
4253 (compare_assert_loc): New function.
4254 (process_assert_insertions): Sort and optimize assert locations
4255 to remove duplicates and push down identical assertions on
4256 edges to their destination block.
4258 2017-01-27 Richard Biener <rguenther@suse.de>
4260 PR tree-optimization/79244
4261 * tree-vrp.c (remove_range_assertions): Forcefully propagate
4262 out SSA names even if abnormal.
4264 2017-01-27 Jakub Jelinek <jakub@redhat.com>
4266 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
4267 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
4268 instead of MPFR_RNDN.
4270 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
4273 * arm.c (arm_option_override): Don't call build_target_option_node
4274 until after doing all option overrides.
4275 (arm_valid_target_attribute_tree): Likewise.
4277 2017-01-27 Martin Liska <mliska@suse.cz>
4279 * doc/invoke.texi (-fprofile-arcs): Document profiling support
4280 for {cd}tors and C++ {cd}tors.
4282 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
4284 * config/s390/s390.md ("*setmem_long_and")
4285 ("*setmem_long_and_31z"): Use zero_extend instead of and.
4287 2017-01-26 Martin Sebor <msebor@redhat.com>
4289 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
4292 2017-01-26 Martin Sebor <msebor@redhat.com>
4294 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
4295 HAVE_DFmode before using XFmode or DFmode.
4296 (parse_directive): Avoid using the z length modifier to avoid
4297 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
4300 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
4301 to accept adjustment as an array.
4302 (get_int_range): New function.
4303 (struct directive): Make width and prec arrays.
4304 (directive::set_width, directive::set_precision): Call get_int_range.
4305 (format_integer, format_floating): Handle width and precision ranges.
4306 (format_string, parse_directive): Same.
4308 2017-01-26 Jakub Jelinek <jakub@redhat.com>
4311 * dwarf2out.c (generate_skeleton_bottom_up): For children with
4312 comdat_type_p set, just clone them, but keep the children in the
4316 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
4317 which have direct callers with -fvar-tracking-assignments enabled
4319 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
4320 inside of type units.
4322 2017-01-26 Martin Sebor <msebor@redhat.com>
4325 * gimple-ssa-sprintf.c (struct result_range): Add likely and
4327 (struct format_result): Replace number_chars, number_chars_min,
4328 and number_chars_max with a single member of struct result_range.
4330 (format_result::operator+=): Adjust.
4331 (struct fmtresult): Remove bounded. Handle likely and unlikely
4333 (fmtresult::adjust_for_width_or_precision): New function.
4334 (fmtresult:type_max_digits): New function.
4335 (bytes_remaining): Handle likely and unlikely counters.
4336 (min_bytes_remaining): Remove.
4337 (format_percent): Simplify.
4338 (format_integer, format_floating): Set likely and unlikely counters.
4339 (get_string_length, format_character, format_string): Same.
4340 (format_plain, should_warn_p): New function.
4341 (maybe_warn): Call should_warn_p. Update diagnostic messages
4342 and handle those for all directives, including plain strings.
4343 (format_directive): Handle likely and unlikely counters.
4344 Remove unnecessary quoting from diagnostics. Add an informational
4346 (add_bytes): Remove.
4347 (pass_sprintf_length::compute_format_length): Simplify.
4348 (try_substitute_return_value): Handle likely and unlikely counters.
4350 2017-01-26 Carl Love <cel@us.ibm.com>
4352 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
4353 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
4355 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
4358 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
4359 endianess for subregs into account.
4360 * lra-constraints.c (lra_constraints): Do risky transformations
4361 always on the first iteration.
4362 * lra-lives.c (check_pseudos_live_through_calls): Add arg
4363 last_call_used_reg_set.
4364 (process_bb_lives): Define and use last_call_used_reg_set.
4365 * lra.c (lra): Always continue after lra_constraints on the first
4368 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
4370 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
4372 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
4374 2017-01-26 Jakub Jelinek <jakub@redhat.com>
4376 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
4377 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
4378 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
4379 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
4380 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
4381 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
4382 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
4383 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
4384 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
4386 2017-01-26 Marek Polacek <polacek@redhat.com>
4389 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
4390 for the third operand.
4392 2017-01-26 Jakub Jelinek <jakub@redhat.com>
4395 * omp-low.c (struct omp_context): Add simt_stmt field.
4396 (scan_omp_for): Return omp_context *.
4397 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
4398 context to the _simt_ SIMD stmt.
4399 (lower_omp_for): For combined SIMD with sibling _simt_
4400 SIMD, make sure to use the same decls in _looptemp_
4401 clauses as in the sibling.
4403 2017-01-26 David Sherwood <david.sherwood@arm.com>
4406 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
4409 2017-01-26 Jakub Jelinek <jakub@redhat.com>
4412 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
4413 emit fld b; fld a; if possible.
4415 * brig-builtins.def: Update copyright years.
4416 * config/arm/arm_acle_builtins.def: Update copyright years.
4418 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
4421 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
4422 constraint instead of o for the stxsd instruction.
4424 2017-01-25 Carl Love <cel@us.ibm.com>
4426 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
4427 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
4429 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
4431 * doc/invoke.texi (C++ Dialect Options): Fix typo.
4433 2017-01-25 Richard Biener <rguenther@suse.de>
4435 PR tree-optimization/69264
4436 * target.def (vector_alignment_reachable): Improve documentation.
4437 * doc/tm.texi: Regenerate.
4438 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
4440 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
4441 earlier changes with respect to TYPE_USER_ALIGN.
4442 (vector_alignment_reachable_p): Likewise. Improve dumping.
4444 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4447 * config/arm/arm.md (xordi3): Force constant operand into a register
4450 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4452 * doc/invoke.texi (-fstore-merging): Correct default optimization
4453 levels at which it is enabled.
4454 (-O): Move -fstore-merging from list to...
4457 2017-01-25 Richard Biener <rguenther@suse.de>
4460 * omp-expand.c: Include debug.h.
4461 (expand_omp_taskreg): Make sure to generate early debug before
4462 outlining anything from a function.
4463 (expand_omp_target): Likewise.
4464 (grid_expand_target_grid_body): Likewise.
4466 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
4469 * asan.c (get_translation_unit_decl): New function.
4470 (asan_add_global): Extract modules file name from globals
4471 TRANSLATION_UNIT_DECL name.
4473 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
4476 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
4477 for long calls with APCS frame and VFP.
4479 2017-01-24 David Malcolm <dmalcolm@redhat.com>
4481 * cfg.c (original_copy_tables_initialized_p): New function.
4482 * cfg.h (original_copy_tables_initialized_p): New decl.
4483 * cfgrtl.c (relink_block_chain): Guard the call to
4484 free_original_copy_tables with a call to
4485 original_copy_tables_initialized_p.
4486 * cgraph.h (symtab_node::native_rtl_p): New decl.
4487 * cgraphunit.c (symtab_node::native_rtl_p): New function.
4488 (symtab_node::needed_p): Don't assert for early assembly output
4489 for __RTL functions.
4490 (cgraph_node::finalize_function): Set "force_output" for __RTL
4492 (cgraph_node::analyze): Bail out early for __RTL functions.
4493 (analyze_functions): Update assertion to support __RTL functions.
4494 (cgraph_node::expand): Bail out early for __RTL functions.
4495 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
4497 * function.h (struct function): Update comment for field
4499 * gimple-expr.c: Include "tree-pass.h".
4500 (gimple_has_body_p): Return false for __RTL functions.
4501 * Makefile.in (OBJS): Add run-rtl-passes.o.
4502 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
4504 (gcc::pass_manager::get_clean_slate): New accessor.
4505 * passes.c: Include "insn-addr.h".
4506 (should_skip_pass_p): Add logging. Update logic for running
4507 "expand" to be compatible with both __GIMPLE and __RTL. Guard
4508 property-provider override so it is only done for gimple passes.
4510 (skip_pass): New function.
4511 (execute_one_pass): Call skip_pass when skipping passes.
4512 * read-md.c (md_reader::read_char): Support filtering
4513 the input to a subset of line numbers.
4514 (md_reader::md_reader): Initialize fields
4515 m_first_line and m_last_line.
4516 (md_reader::read_file_fragment): New function.
4517 * read-md.h (md_reader::read_file_fragment): New decl.
4518 (md_reader::m_first_line): New field.
4519 (md_reader::m_last_line): New field.
4520 * read-rtl-function.c (function_reader::create_function): Only
4521 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
4522 curr_properties. Set DECL_INITIAL to a dummy block.
4523 (read_rtl_function_body_from_file_range): New function.
4524 * read-rtl-function.h (read_rtl_function_body_from_file_range):
4526 * run-rtl-passes.c: New file.
4527 * run-rtl-passes.h: New file.
4529 2017-01-24 Jeff Law <law@redhat.com>
4531 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
4534 2017-01-24 Bin Cheng <bin.cheng@arm.com>
4536 PR tree-optimization/79159
4537 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
4538 (record_nonwrapping_iv): Improve boundary using above function if no
4539 value range information.
4541 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
4542 Martin Jambor <mjambor@suse.cz>
4544 * brig-builtins.def: New file.
4545 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
4546 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
4547 (DEF_HSAIL_SAT_BUILTIN): Likewise.
4548 (DEF_HSAIL_INTR_BUILTIN): Likewise.
4549 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
4550 * builtin-types.def (BT_INT8): New.
4551 (BT_INT16): Likewise.
4552 (BT_UINT8): Likewise.
4553 (BT_UINT16): Likewise.
4554 (BT_FN_ULONG): Likewise.
4555 (BT_FN_UINT_INT): Likewise.
4556 (BT_FN_UINT_ULONG): Likewise.
4557 (BT_FN_UINT_LONG): Likewise.
4558 (BT_FN_UINT_PTR): Likewise.
4559 (BT_FN_ULONG_PTR): Likewise.
4560 (BT_FN_INT8_FLOAT): Likewise.
4561 (BT_FN_INT16_FLOAT): Likewise.
4562 (BT_FN_UINT32_FLOAT): Likewise.
4563 (BT_FN_UINT16_FLOAT): Likewise.
4564 (BT_FN_UINT8_FLOAT): Likewise.
4565 (BT_FN_UINT64_FLOAT): Likewise.
4566 (BT_FN_UINT16_UINT32): Likewise.
4567 (BT_FN_UINT32_UINT16): Likewise.
4568 (BT_FN_UINT16_UINT16_UINT16): Likewise.
4569 (BT_FN_INT_PTR_INT): Likewise.
4570 (BT_FN_UINT_PTR_UINT): Likewise.
4571 (BT_FN_LONG_PTR_LONG): Likewise.
4572 (BT_FN_ULONG_PTR_ULONG): Likewise.
4573 (BT_FN_VOID_UINT64_UINT64): Likewise.
4574 (BT_FN_UINT8_UINT8_UINT8): Likewise.
4575 (BT_FN_INT8_INT8_INT8): Likewise.
4576 (BT_FN_INT16_INT16_INT16): Likewise.
4577 (BT_FN_INT_INT_INT): Likewise.
4578 (BT_FN_UINT_FLOAT_UINT): Likewise.
4579 (BT_FN_FLOAT_UINT_UINT): Likewise.
4580 (BT_FN_ULONG_UINT_UINT): Likewise.
4581 (BT_FN_ULONG_UINT_PTR): Likewise.
4582 (BT_FN_ULONG_ULONG_ULONG): Likewise.
4583 (BT_FN_UINT_UINT_UINT): Likewise.
4584 (BT_FN_VOID_UINT_PTR): Likewise.
4585 (BT_FN_UINT_UINT_PTR: Likewise.
4586 (BT_FN_UINT32_UINT64_PTR): Likewise.
4587 (BT_FN_INT_INT_UINT_UINT): Likewise.
4588 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
4589 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
4590 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
4591 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
4592 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
4593 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
4594 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
4595 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
4596 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
4597 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
4598 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
4599 * doc/frontends.texi: List BRIG FE.
4600 * doc/install.texi (Testing): Add BRIG tesring requirements.
4601 * doc/invoke.texi (Overall Options): Mention BRIG.
4602 * doc/standards.texi (Standards): Doucment BRIG HSA version.
4604 2017-01-24 Richard Biener <rguenther@suse.de>
4606 PR translation/79208
4607 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
4609 2017-01-24 Martin Jambor <mjambor@suse.cz>
4612 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
4613 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
4616 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
4619 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
4620 casts from signed to unsigned really don't have a range.
4622 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
4624 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
4625 GMP_RNDx for compatiblity.
4627 2017-01-24 Martin Liska <mliska@suse.cz>
4630 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
4631 that would prevent us to call alloca with -1 as argument.
4633 2017-01-24 Jakub Jelinek <jakub@redhat.com>
4635 * dwarf2out.c (output_compilation_unit_header, output_file_names):
4636 Avoid -Wformat-security warning.
4638 2017-01-23 Andrew Pinski <apinski@cavium.com>
4640 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
4643 2017-01-23 Martin Sebor <msebor@redhat.com>
4646 * gimple-ssa-sprintf.c (warn_level): New global.
4647 (format_integer): Use it here and throughout the rest of the file.
4648 Use the same switch to compute sign as base.
4649 (maybe_warn): New function.
4650 (format_directive): Factor out warnings into maybe_warn.
4651 Add debugging output. Use warn_level.
4652 (add_bytes): Use warn_level.
4653 (pass_sprintf_length::compute_format_length): Add debugging output.
4654 (try_substitute_return_value): Same.
4655 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
4658 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
4659 (struct fmtresult, format_integer, format_floating): Adjust.
4660 (fmtresult::fmtresult): Set max correctly in two argument ctor.
4661 (get_string_length, format_string,format_directive): Same.
4662 (pass_sprintf_length::compute_format_length): Same.
4663 (try_substitute_return_value): Simplify slightly.
4666 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
4667 (fmtresult::operator+=): Outlined.
4668 (struct fmtresult): Add ctors.
4669 (struct conversion_spec): Rename...
4670 (struct directive): ...to this. Add and remove data members.
4671 (directive::set_width, directive::set_precision): New functions.
4672 (format_percent): Use fmtresult ctor.
4673 (get_width_and_precision): Remove.
4674 (format_integer): Make naming changes. Avoid computing width and
4676 (format_floating): Same. Adjust indentation.
4677 (format_character, format_none): New functions.
4678 (format_string): Moved character handling to format_character.
4679 (format_directive): Remove arguments, change return type.
4680 (parse_directive): New function.
4681 (pass_sprintf_length::compute_format_length): Move directive
4682 parsing to parse_directive.
4684 2017-01-23 Jakub Jelinek <jakub@redhat.com>
4686 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
4687 (assign_assembler_name_if_needed): ... this.
4688 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
4689 (assign_assembler_name_if_needed): ... this.
4690 (free_lang_data_in_cgraph): Adjust callers.
4691 * cgraphunit.c (cgraph_node::analyze): Likewise.
4692 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
4695 2017-01-23 Richard Biener <rguenther@suse.de>
4697 PR tree-optimization/79088
4698 PR tree-optimization/79188
4699 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
4700 resetting loop bounds after last path deletion. Reset loop
4701 bounds of the target loop, make code match the comments.
4702 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
4703 Make sure loops need no fixups.
4705 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
4707 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
4708 exponent support with double type for first argument.
4709 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
4710 type returned by __builtin_vec_extract_sig,
4711 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
4712 functions from "vector int" to "vector unsigned int" or from
4713 "vector long long int" to "vector unsigned long long int".
4714 Changed type returned by __builtin_vec_extract_exp,
4715 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
4716 functions from "vector int" to "vector unsigned int" or from
4717 "vector long long int" to "vector unsigned long long int".
4718 Changed return type of __builtin_vec_test_data_class,
4719 __builtin_vec_test_data_class_sp, and
4720 __builtin_vec_test_data_class_dp from "vector int" to
4721 "vector bool int" or from "vector long long int" to "vector bool
4722 long long int" and changed second argument type from "unsigned
4723 int" to "int". Added new overloaded function forms "vector float
4724 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
4725 "vector float __builtin_vec_insert_exp_sp (vector float, vector
4726 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
4727 double, vector unsigned long long int)" and "vector double
4728 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
4729 long int)". Changed return type of
4730 __builtin_scalar_test_data_class and
4731 __builtin_scalar_test_data_class_sp and
4732 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
4733 int" and changed second argument from "unsigned int" to "int".
4734 Changed type returned by __builtin_scalar_test_neg,
4735 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
4736 from "int" to "bool int". Added new overloaded function form
4737 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
4738 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
4739 exponent double-precision with floating point first argument.
4740 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
4741 documentation of scalar_test_data_class, scalar_test_neg,
4742 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
4743 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
4744 vec_test_data_class built-in functions to reflect refinements in
4745 their type signatures.
4747 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
4749 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
4751 (aarch64_elf_asm_destructor): Likewise.
4753 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
4755 PR rtl-optimization/78634
4756 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
4757 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
4758 * ifcvt.c (noce_try_cmove): Add missing cost check.
4760 PR rtl-optimization/71724
4761 * combine.c (if_then_else_cond): Look for situations where it is
4762 beneficial to undo the work of one of the recursive calls.
4764 2017-01-23 Bin Cheng <bin.cheng@arm.com>
4766 PR tree-optimization/70754
4767 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
4768 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
4769 combined stmt before it if not NULL.
4770 (combine_chains): Process refs reversely and compute dominance point
4773 2017-01-23 Martin Liska <mliska@suse.cz>
4775 PR tree-optimization/79196
4776 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
4777 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
4779 (strlen_optimize_stmt): Call the renamed function.
4781 2017-01-23 Michael Matz <matz@suse.de>
4783 PR tree-optimization/78384
4784 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
4786 2017-01-23 Richard Biener <rguenther@suse.de>
4788 PR tree-optimization/79186
4789 * tree-vrp.c (register_new_assert_for): Make sure we've seen
4790 both incoming edges before moving an assert.
4792 2017-01-23 Martin Jambor <mjambor@suse.cz>
4794 * ipa-prop.c (load_from_param_1): Removed.
4795 (load_from_unmodified_param): Bits from load_from_param_1 put back
4797 (load_from_param): Removed.
4798 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
4799 with stmt. Reverted back to use of load_from_unmodified_param.
4801 2017-01-23 Martin Jambor <mjambor@suse.cz>
4804 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
4805 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
4806 field a pointer to garbage collected vector, mark lattices and
4807 ipcp_orig_node with GTY((skip)).
4808 (ipa_get_param_count): Adjust to descriptors being a pointer.
4809 (ipa_get_param): Likewise.
4810 (ipa_get_type): Likewise.
4811 (ipa_get_param_move_cost): Likewise.
4812 (ipa_set_param_used): Likewise.
4813 (ipa_get_controlled_uses): Likewise.
4814 (ipa_set_controlled_uses): Likewise.
4815 (ipa_is_param_used): Likewise.
4816 (ipa_node_params_t): Move into garbage collector. New methods insert
4818 (ipa_node_params_sum): Annotate wth GTY(()).
4819 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
4821 (ipa_load_from_parm_agg): Adjust declaration.
4822 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
4823 * ipa-profile.c (ipa_profile): Likewise.
4824 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
4825 (ipa_populate_param_decls): Make descriptors parameter garbage
4827 (ipa_dump_param): Adjust to descriptors being a pointer.
4828 (ipa_alloc_node_params): Likewise.
4829 (ipa_initialize_node_params): Likewise.
4830 (load_from_param_1): Make descriptors parameter garbage collected.
4831 (load_from_unmodified_param): Likewise.
4832 (load_from_param): Likewise.
4833 (ipa_load_from_parm_agg): Likewise.
4834 (ipa_node_params::~ipa_node_params): Removed.
4835 (ipa_free_all_node_params): Remove call to delete operator.
4836 (ipa_node_params_t::insert): New.
4837 (ipa_node_params_t::remove): Likewise.
4838 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
4839 copy known_csts and known_contexts vectors.
4840 (ipa_read_node_info): Adjust to descriptors being a pointer.
4841 (ipcp_modif_dom_walker): Make m_descriptors field garbage
4843 (ipcp_transform_function): Make descriptors variable garbage
4846 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
4848 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
4849 * config/i386/avx512dqintrin.h: Ditto.
4850 * config/i386/avx512fintrin.h: Ditto.
4851 * gcc/config/i386/i386.c: Handle new builtins.
4852 * config/i386/i386-builtin.def: Add new builtins.
4853 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
4854 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
4856 2017-01-23 Jakub Jelinek <jakub@redhat.com>
4857 Martin Liska <mliska@suse.cz>
4859 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
4860 * asan.c (asan_expand_poison_ifn): Support stores and use
4861 appropriate ASAN report function.
4862 * internal-fn.c (expand_ASAN_POISON_USE): New function.
4863 * internal-fn.def (ASAN_POISON_USE): Declare.
4864 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
4865 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
4866 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
4867 ASAN_POISON calls w/o LHS.
4868 * tree-ssa.c (execute_update_addresses_taken): Create clobber
4869 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
4870 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
4871 * gimplify.c (asan_poison_variables): Add attribute
4872 use_after_scope_memory to variables that really needs to live
4874 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
4875 having the attribute.
4877 2017-01-23 Martin Liska <mliska@suse.cz>
4879 * asan.c (create_asan_shadow_var): New function.
4880 (asan_expand_poison_ifn): Likewise.
4881 * asan.h (asan_expand_poison_ifn): New declaration.
4882 * internal-fn.c (expand_ASAN_POISON): Likewise.
4883 * internal-fn.def (ASAN_POISON): New builtin.
4884 * sanopt.c (pass_sanopt::execute): Expand
4885 asan_expand_poison_ifn.
4886 * tree-inline.c (copy_decl_for_dup_finish): Make function
4888 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
4889 * tree-ssa.c (is_asan_mark_p): New function.
4890 (execute_update_addresses_taken): Rewrite local variables
4891 (identified just by use-after-scope as addressable) into SSA.
4893 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
4895 * doc/install.texi (Specific): opensource.apple.com uses https
4896 now. Remove trailing slash.
4898 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
4900 * README.Portability: Remove note on an Irix compatibility issue.
4902 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
4904 * gcov.c (INCLUDE_ALGORITHM): Define.
4905 (INCLUDE_VECTOR): Define.
4906 No longer include <vector> and <algorithm> directly.
4908 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
4910 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
4912 * doc/invoke.texi (Code Gen Options): Ditto.
4914 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
4917 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
4919 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
4921 rtl-optimization/79125
4922 * cprop.c (local_cprop_pass): Handle cases where we make an
4925 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
4929 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
4930 read from, for big endian.
4932 2017-01-20 Jiong Wang <jiong.wang@arm.com>
4934 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
4935 register pauth builtins for LP64 only.
4937 2017-01-20 Marek Polacek <polacek@redhat.com>
4940 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
4943 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
4945 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
4947 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
4948 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
4949 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
4951 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4954 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
4955 in big-endian mode when they are not a single duplicated value.
4957 2017-01-20 Richard Biener <rguenther@suse.de>
4959 * BASE-VER: Bump to 7.0.1.
4961 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
4963 * omp-low.c (omplow_simd_context): New struct. Use it...
4964 (lower_rec_simd_input_clauses): ...here and...
4965 (lower_rec_input_clauses): ...here to hold common data. Adjust all
4966 references to idx, lane, max_vf, is_simt.
4968 2017-01-20 Graham Markall <graham.markall@embecosm.com>
4970 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
4973 2017-01-20 Martin Jambor <mjambor@suse.cz>
4975 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
4976 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
4978 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
4979 (GTFILES): Rename hsa.c to hsa-common.c.
4980 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
4981 * hsa-dump.c: Likewise.
4982 * hsa-gen.c: Likewise.
4983 * hsa-regalloc.c: Likewise.
4984 * ipa-hsa.c: Likewise.
4985 * omp-expand.c: Likewise.
4986 * omp-low.c: Likewise.
4987 * toplev.c: Likewise.
4989 2017-01-20 Marek Polacek <polacek@redhat.com>
4992 * doc/invoke.texi: Document -Wduplicated-branches.
4993 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
4994 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
4995 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
4996 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
4997 return 0 only when not OEP_LEXICOGRAPHIC.
4998 (fold_build_cleanup_point_expr): Use the expression
4999 location when building CLEANUP_POINT_EXPR.
5000 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
5001 * tree.c (add_expr): Handle error_mark_node.
5003 2017-01-20 Martin Liska <mliska@suse.cz>
5006 * tree-profile.c (init_ic_make_global_vars): Do not call
5008 (gimple_init_gcov_profiler): Likewise.
5010 2017-01-20 Martin Liska <mliska@suse.cz>
5013 * cgraph.h (maybe_create_reference): Remove argument and
5015 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
5017 * ipa-cp.c (create_specialized_node): Likewise.
5018 * symtab.c (symtab_node::maybe_create_reference): Handle
5019 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
5021 2017-01-20 Martin Liska <mliska@suse.cz>
5023 * read-rtl-function.c (function_reader::create_function): Use
5024 build_decl instread of build_decl_stat.
5026 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
5028 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
5029 * config/i386/avx512dqintrin.h: Ditto.
5030 * config/i386/avx512fintrin.h: Ditto.
5031 * config/i386/i386-builtin-types.def: Add new types.
5032 * gcc/config/i386/i386.c: Handle new types.
5033 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
5034 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
5035 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
5036 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
5037 (__builtin_ia32_kshiftridi): New.
5038 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
5040 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
5044 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
5045 define to rs6000_init_stack_protect_guard.
5046 (rs6000_init_stack_protect_guard): New function.
5048 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
5049 Yunqiang Su <yunqiang.su@imgtec.com>
5051 * config.gcc (supported_defaults): Add madd4.
5052 (with_madd4): Add validation.
5053 (all_defaults): Add madd4.
5054 * config/mips/mips.opt (mmadd4): New option.
5055 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
5057 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
5059 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
5060 (ISA_HAS_FUSED_MADD4): Likewise.
5061 * gcc/doc/invoke.texi (-mmadd4): Document the new option.
5062 * gcc/doc/install.texi (--with-madd4): Document the new option.
5064 2017-01-19 Jiong Wang <jiong.wang@arm.com>
5066 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
5067 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
5068 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
5069 (aarch64_init_pauth_hint_builtins): New.
5070 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
5071 (aarch64_expand_builtin): Expand new builtins.
5073 2017-01-19 Jiong Wang <jiong.wang@arm.com>
5075 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
5076 * combine-stack-adj.c (no_unhandled_cfa): Handle
5077 REG_CFA_TOGGLE_RA_MANGLE.
5078 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
5079 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
5080 info for return address signing.
5081 (aarch64_expand_epilogue): Likewise.
5083 2017-01-19 Jiong Wang <jiong.wang@arm.com>
5085 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
5086 * config/aarch64/aarch64-protos.h
5087 (aarch64_return_address_signing_enabled): New declaration.
5088 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
5090 (aarch64_expand_prologue): Sign return address before it's pushed onto
5092 (aarch64_expand_epilogue): Authenticate return address fetched from
5094 (aarch64_override_options): Sanity check for ILP32 and ISA level.
5095 (aarch64_attributes): New function attributes for "sign-return-address".
5096 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
5097 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
5098 ("*do_return"): Generate combined instructions according to key index.
5099 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
5100 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
5102 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
5103 * config/aarch64/aarch64.opt (msign-return-address=): New.
5104 * doc/extend.texi (AArch64 Function Attributes): Documents
5105 "sign-return-address=".
5106 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
5108 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
5110 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
5111 overall option summary.
5113 2017-01-19 Jiong Wang <jiong.wang@arm.com>
5115 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
5116 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
5117 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
5118 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
5120 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
5122 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
5123 -mpower9-minmax by default for -mcpu=power9.
5124 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
5125 128-bit floating point.
5127 2017-01-20 Alan Modra <amodra@gmail.com>
5129 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
5130 optimizing for size.
5132 2017-01-20 Alan Modra <amodra@gmail.com>
5135 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
5136 for strcmp and strncmp from corresponding builtin decl.
5138 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
5140 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
5141 instead of i386/rtems-64.h.
5142 * config/i386/rtems-64.h: Remove.
5144 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
5148 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
5150 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
5152 2017-01-19 Tamar Christina <tamar.christina@arm.com>
5154 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
5155 Change int to HOST_WIDE_INT.
5156 * config/aarch64/aarch64-protos.h
5157 (aarch64_simd_gen_const_vector_dup): Likewise.
5158 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
5160 2017-01-19 David Malcolm <dmalcolm@redhat.com>
5162 * langhooks-def.h (lhd_type_for_size): New decl.
5163 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
5164 * langhooks.c (lhd_type_for_size): New function, taken from
5167 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
5169 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
5170 define_bypass for CR latency.
5171 (power9-cracked-alu): Update bypass latency and remove power9-branch.
5172 (power9-alu2): Add define_bypass for CR latency.
5174 (power9-mul): Update insn latency.
5175 (power9-mul-compare): Update insn latency, bypass latency and remove
5178 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5180 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
5182 * config/aarch64/aarch64.md
5183 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
5184 aarch64_nopcrelative_literal_loads.
5185 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
5187 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
5189 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
5191 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
5193 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
5196 * config.gcc (supported_defaults): Add lxc1-sxc1.
5197 (with_lxc1_sxc1): Add validation.
5198 (all_defaults): Add lxc1-sxc1.
5199 * config/mips/mips.opt (mlxc1-sxc1): New option.
5200 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
5202 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
5203 __mips_no_lxc1_sxc1.
5204 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
5205 * gcc/doc/invoke.texi (-mlxc1-sxc1): Document the new option.
5206 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
5208 2017-01-19 Richard Biener <rguenther@suse.de>
5210 PR tree-optimization/72488
5211 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
5212 sure to restore SSA info.
5213 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
5215 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
5217 PR rtl-optimization/79121
5218 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
5219 of the inner type when shifting an extended value.
5221 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
5224 * symtab.c (symtab_node::equal_address_to): Fix comparing of
5225 interposable aliases.
5227 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
5230 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
5231 Use the evmergelohi instruction.
5232 (mov_si<mode>_e500_subreg4_2_le): Likewise.
5233 (mov_sitf_e500_subreg8_2_be): Likewise.
5234 (mov_sitf_e500_subreg12_2_le): Likewise.
5235 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
5236 (mov_si<mode>_e500_subreg4_2_be): Likewise.
5237 (mov_sitf_e500_subreg8_2_le): Likewise.
5238 (mov_sitf_e500_subreg12_2_be): Likewise.
5240 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5242 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
5243 attribute from vecsimple to vecperm.
5244 (altivec_vbpermq2): Likewise.
5246 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5249 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
5251 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
5252 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
5253 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
5254 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
5255 case where N arg is SIZE_MAX.
5256 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
5257 (cmpstrsi): Add pattern.
5259 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
5261 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5262 __builtin_vec_revb builtins.
5263 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
5264 built-in functions to support generation of the ISA 3.0 XXBR<x>
5265 vector byte reverse instructions.
5266 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
5267 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
5268 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
5269 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
5270 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
5271 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
5272 (P9V_BUILTIN_VEC_REVB): Likewise.
5273 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
5274 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
5275 (p9_xxbrq_v16qi): Likewise.
5276 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
5277 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
5278 (p9_xxbrh_v8hi): Likewise.
5279 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
5280 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
5281 vec_revb built-in functions.
5283 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
5285 PR rtl-optimization/78952
5286 * config/i386/i386.md (any_extract): New code iterator.
5287 (*insvqi_2): Use any_extract for source operand.
5288 (*insvqi_3): Use any_shiftrt for source operand.
5290 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
5292 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
5294 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
5296 2017-01-18 Matthias Klose <doko@ubuntu.com>
5298 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
5300 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5302 * config/rs6000/altivec.h (vec_bperm): Change #define.
5303 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
5304 (altivec_vbpermq2): New define_insn.
5305 (altivec_vbpermd): Likewise.
5306 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
5308 (VBPERMD): Likewise.
5309 (VBPERM): New polymorphic function interface.
5310 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
5311 Add entries for P9V_BUILTIN_VEC_VBPERM.
5312 * doc/extend.texi: Add interfaces for vec_bperm.
5314 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5316 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
5317 first letter of error messages.
5318 (s390_resolve_overloaded_builtin): Likewise.
5319 * config/s390/s390.c (s390_expand_builtin): Likewise.
5320 (s390_invalid_arg_for_unprototyped_fn): Likewise.
5321 (s390_valid_target_attribute_inner_p): Likewise.
5322 * config/s390/s390.md ("tabort"): Likewise.
5324 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
5326 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
5327 (ISA_AVOID_DIV_HILO): New macro.
5328 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
5329 (ISA_HAS_DDIV): Likewise.
5331 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
5333 * doc/invoke.texi (fabi-version): Correct number of occurrences.
5335 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
5337 * doc/invoke.texi (fabi-version): Spelling fix.
5339 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
5342 * doc/invoke.texi (fabi-version): Mention mangling fix for
5345 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
5348 * doc/invoke.texi (fabi-version): Document discriminator mangling.
5350 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
5353 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
5354 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
5356 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
5358 (stack_protect_test): Ditto.
5359 * config/rs6000/rs6000.opt (mstack-protector-guard=,
5360 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
5362 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
5363 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
5364 -mstack-protector-guard-offset=.
5365 (RS/6000 and PowerPC Options): Ditto.
5367 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
5369 * config/i386/i386.h (MASK_CLASS_P): New define.
5370 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
5371 there are no registers from different register sets also when
5372 mask registers are used. Update function comment.
5373 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
5374 to (*k/*r) and (*k/*km) alternatives.
5376 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
5378 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
5379 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
5380 (EH_RETURN_HANDLER_RTX): New define.
5381 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
5382 Force frame pointer in EH return functions.
5383 (aarch64_expand_epilogue): Add barrier for eh_return.
5384 (aarch64_final_eh_return_addr): Remove.
5385 (aarch64_eh_return_handler_rtx): New function.
5386 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
5388 (aarch64_eh_return_handler_rtx): New prototype.
5390 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5392 * config/rs6000/altivec.h (vec_rlmi): New #define.
5393 (vec_vrlnm): Likewise.
5394 (vec_rlnm): Likewise.
5395 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
5396 (UNSPEC_VRLNM): Likewise.
5397 (VIlong): New mode iterator.
5398 (altivec_vrl<VI_char>mi): New define_insn.
5399 (altivec_vrl<VI_char>nm): Likewise.
5400 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
5403 (RLNM): New polymorphic function entry.
5404 (VRLWMI): New monomorphic function entry.
5406 (RLMI): New polymorphic function entry.
5407 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
5408 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
5409 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
5412 2017-01-17 Jakub Jelinek <jakub@redhat.com>
5415 * dwarf2out.c (field_byte_offset): Restore the
5416 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
5417 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
5418 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
5421 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
5424 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
5426 2017-01-17 Jakub Jelinek <jakub@redhat.com>
5429 * dwarf2out.c (add_data_member_location_attribute): For constant
5430 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
5431 instead of DW_AT_data_member_location, DW_AT_bit_offset and
5432 DW_AT_byte_size attributes.
5434 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
5436 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
5437 after forcing to constant memory when the code model is medium.
5439 2017-01-17 Julia Koval <julia.koval@intel.com>
5442 * config/i386/avx512fintrin.h
5443 (_mm512_i32gather_ps): Change __addr type to void const*.
5444 (_mm512_mask_i32gather_ps): Ditto.
5445 (_mm512_i32gather_pd): Ditto.
5446 (_mm512_mask_i32gather_pd): Ditto.
5447 (_mm512_i64gather_ps): Ditto.
5448 (_mm512_mask_i64gather_ps): Ditto.
5449 (_mm512_i64gather_pd): Ditto.
5450 (_mm512_mask_i64gather_pd): Ditto.
5451 (_mm512_i32gather_epi32): Ditto.
5452 (_mm512_mask_i32gather_epi32): Ditto.
5453 (_mm512_i32gather_epi64): Ditto.
5454 (_mm512_mask_i32gather_epi64): Ditto.
5455 (_mm512_i64gather_epi32): Ditto.
5456 (_mm512_mask_i64gather_epi32): Ditto.
5457 (_mm512_i64gather_epi64): Ditto.
5458 (_mm512_mask_i64gather_epi64): Ditto.
5459 (_mm512_i32scatter_ps): Change __addr type to void*.
5460 (_mm512_mask_i32scatter_ps): Ditto.
5461 (_mm512_i32scatter_pd): Ditto.
5462 (_mm512_mask_i32scatter_pd): Ditto.
5463 (_mm512_i64scatter_ps): Ditto.
5464 (_mm512_mask_i64scatter_ps): Ditto.
5465 (_mm512_i64scatter_pd): Ditto.
5466 (_mm512_mask_i64scatter_pd): Ditto.
5467 (_mm512_i32scatter_epi32): Ditto.
5468 (_mm512_mask_i32scatter_epi32): Ditto.
5469 (_mm512_i32scatter_epi64): Ditto.
5470 (_mm512_mask_i32scatter_epi64): Ditto.
5471 (_mm512_i64scatter_epi32): Ditto.
5472 (_mm512_mask_i64scatter_epi32): Ditto.
5473 (_mm512_i64scatter_epi64): Ditto.
5474 (_mm512_mask_i64scatter_epi64): Ditto.
5475 * config/i386/avx512pfintrin.h
5476 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
5477 (_mm512_mask_prefetch_i32gather_ps): Ditto.
5478 (_mm512_mask_prefetch_i64gather_pd): Ditto.
5479 (_mm512_mask_prefetch_i64gather_ps): Ditto.
5480 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
5481 (_mm512_prefetch_i32scatter_ps): Ditto.
5482 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
5483 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
5484 (_mm512_prefetch_i64scatter_pd): Ditto.
5485 (_mm512_prefetch_i64scatter_ps): Ditto.
5486 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
5487 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
5488 * config/i386/avx512vlintrin.h
5489 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
5490 (_mm_mmask_i32gather_ps): Ditto.
5491 (_mm256_mmask_i32gather_pd): Ditto.
5492 (_mm_mmask_i32gather_pd): Ditto.
5493 (_mm256_mmask_i64gather_ps): Ditto.
5494 (_mm_mmask_i64gather_ps): Ditto.
5495 (_mm256_mmask_i64gather_pd): Ditto.
5496 (_mm_mmask_i64gather_pd): Ditto.
5497 (_mm256_mmask_i32gather_epi32): Ditto.
5498 (_mm_mmask_i32gather_epi32): Ditto.
5499 (_mm256_mmask_i32gather_epi64): Ditto.
5500 (_mm_mmask_i32gather_epi64): Ditto.
5501 (_mm256_mmask_i64gather_epi32): Ditto.
5502 (_mm_mmask_i64gather_epi32): Ditto.
5503 (_mm256_mmask_i64gather_epi64): Ditto.
5504 (_mm_mmask_i64gather_epi64): Ditto.
5505 (_mm256_i32scatter_ps): Change __addr type to void*.
5506 (_mm256_mask_i32scatter_ps): Ditto.
5507 (_mm_i32scatter_ps): Ditto.
5508 (_mm_mask_i32scatter_ps): Ditto.
5509 (_mm256_i32scatter_pd): Ditto.
5510 (_mm256_mask_i32scatter_pd): Ditto.
5511 (_mm_i32scatter_pd): Ditto.
5512 (_mm_mask_i32scatter_pd): Ditto.
5513 (_mm256_i64scatter_ps): Ditto.
5514 (_mm256_mask_i64scatter_ps): Ditto.
5515 (_mm_i64scatter_ps): Ditto.
5516 (_mm_mask_i64scatter_ps): Ditto.
5517 (_mm256_i64scatter_pd): Ditto.
5518 (_mm256_mask_i64scatter_pd): Ditto.
5519 (_mm_i64scatter_pd): Ditto.
5520 (_mm_mask_i64scatter_pd): Ditto.
5521 (_mm256_i32scatter_epi32): Ditto.
5522 (_mm256_mask_i32scatter_epi32): Ditto.
5523 (_mm_i32scatter_epi32): Ditto.
5524 (_mm_mask_i32scatter_epi32): Ditto.
5525 (_mm256_i32scatter_epi64): Ditto.
5526 (_mm256_mask_i32scatter_epi64): Ditto.
5527 (_mm_i32scatter_epi64): Ditto.
5528 (_mm_mask_i32scatter_epi64): Ditto.
5529 (_mm256_i64scatter_epi32): Ditto.
5530 (_mm256_mask_i64scatter_epi32): Ditto.
5531 (_mm_i64scatter_epi32): Ditto.
5532 (_mm_mask_i64scatter_epi32): Ditto.
5533 (_mm256_i64scatter_epi64): Ditto.
5534 (_mm256_mask_i64scatter_epi64): Ditto.
5535 (_mm_i64scatter_epi64): Ditto.
5536 (_mm_mask_i64scatter_epi64): Ditto.
5537 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
5538 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
5539 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
5540 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
5541 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
5542 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
5543 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
5544 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
5545 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
5546 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
5547 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
5548 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
5549 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
5550 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
5551 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
5552 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
5553 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
5554 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
5555 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
5556 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
5557 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
5558 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
5559 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
5560 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
5561 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
5562 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
5563 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
5564 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
5565 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
5566 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
5567 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
5568 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
5569 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
5570 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
5571 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
5572 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
5573 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
5574 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
5575 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
5576 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
5577 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
5578 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
5579 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
5580 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
5581 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
5582 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
5583 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
5584 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
5585 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
5586 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
5587 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
5588 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
5589 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
5590 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
5591 definitions accordingly.
5593 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
5594 Kuan-Lin Chen <kuanlinchentw@gmail.com>
5597 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
5600 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
5603 * ira-conflicts.c (ira_build_conflicts): Update total conflict
5604 hard regs for inner regno.
5606 2017-01-17 Martin Liska <mliska@suse.cz>
5609 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
5610 assumption and add comment.
5612 2017-01-17 Nathan Sidwell <nathan@acm.org>
5614 * ipa-visibility.c (localize_node): New function, broken out of ...
5615 (function_and_variable_visibility): ... here. Call it.
5617 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
5620 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
5621 correctly set frequency of oudgoing edge.
5622 (duplicate_thread_path): Fix profile updating.
5624 2017-01-17 Jakub Jelinek <jakub@redhat.com>
5627 * configure.ac: Add GCC_BASE_VER.
5628 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
5629 version from BASE-VER file.
5630 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
5631 (gcc.o): Depend on $(BASEVER).
5632 * common.opt (dumpfullversion): New option.
5633 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
5634 * doc/invoke.texi: Document -dumpfullversion.
5635 * doc/install.texi: Document --with-gcc-major-version-only.
5636 * configure: Regenerated.
5638 2017-01-17 Richard Biener <rguenther@suse.de>
5640 PR tree-optimization/71433
5641 * tree-vrp.c (register_new_assert_for): Merge same asserts
5642 on all incoming edges.
5643 (process_assert_insertions_for): Handle insertions at the
5646 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
5648 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
5649 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
5651 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
5654 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
5657 2017-01-17 Alan Modra <amodra@gmail.com>
5660 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
5661 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
5662 symbolic stack limit when pic.
5664 2017-01-16 Martin Sebor <msebor@redhat.com>
5666 PR tree-optimization/78608
5667 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
5669 2017-01-16 Jeff Law <law@redhat.com>
5672 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
5673 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
5674 for several include directories that may be relative to sysroot.
5675 * config/i386/x-mingw32 (gplus_includedir): Define.
5676 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
5677 (native_system_includedir): Likewise.
5678 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
5679 override if TARGET_SYSTEM_ROOT is defined.
5680 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
5682 PR tree-optimization/79090
5683 PR tree-optimization/33562
5684 PR tree-optimization/61912
5685 PR tree-optimization/77485
5686 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
5687 and computed trims into the dump file.
5689 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
5691 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
5693 2017-01-16 Jakub Jelinek <jakub@redhat.com>
5696 * gimplify.c (gimplify_init_constructor): If want_value and
5697 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
5701 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
5702 sequence. Formatting fixes.
5703 (doloop_optimize): Formatting fixes.
5706 * gcc.c (debug_level_greater_than_spec_func): New function.
5707 (static_spec_functions): Add debug-level-gt spec function.
5708 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
5710 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
5711 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
5712 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
5713 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
5714 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
5715 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
5717 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
5719 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
5720 QImode fixups to general and mask registers only.
5722 2017-01-16 Carl Love <cel@us.ibm.com>
5724 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
5725 for built-in functions
5726 vector signed char vec_nabs (vector signed char)
5727 vector signed short vec_nabs (vector signed short)
5728 vector signed int vec_nabs (vector signed int)
5729 vector signed long long vec_nabs (vector signed long long)
5730 vector float vec_nabs (vector float)
5731 vector double vec_nabs (vector double)
5732 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
5734 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
5735 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
5736 * doc/extend.texi: Update the documentation file for the new built-in
5739 2017-01-16 Martin Sebor <msebor@redhat.com>
5741 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
5744 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5746 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
5747 UNSPEC_VSX__XXSPLTD to require special splat handling.
5749 2017-01-16 David Malcolm <dmalcolm@redhat.com>
5752 * system.h: Poison strndup.
5754 2017-01-16 Alan Modra <amodra@gmail.com>
5757 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
5760 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
5762 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
5764 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
5766 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
5767 call recog here. Assert that INSN_CODE (insn) is non-negative.
5769 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
5772 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
5774 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
5775 in the currently scheduled RTL fragment.
5777 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
5779 PR rtl-optimization/78751
5780 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
5783 2017-01-14 Jeff Law <law@redhat.com>
5785 PR tree-optimization/79090
5786 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
5787 variable length stores.
5788 (compute_trims): Delete dead assignment to *trim_tail.
5789 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
5792 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
5794 PR rtl-optimization/78626
5795 PR rtl-optimization/78727
5796 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
5797 of a block, and split such blocks after everything else is finished.
5799 2017-01-14 Alan Modra <amodra@gmail.com>
5802 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
5803 target legitimate_combined_insn.
5804 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
5805 (rs6000_legitimate_combined_insn): New function.
5806 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
5808 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
5809 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
5810 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
5812 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
5814 * doc/frontends.texi (G++ and GCC): Remove references to Java.
5816 2017-01-13 Jeff Law <law@redhat.com>
5818 PR tree-optimization/33562
5819 PR tree-optimization/61912
5820 PR tree-optimization/77485
5821 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
5823 (delete_dead_assignment): Likewise.
5824 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
5825 statement to delete_dead_call and delete_dead_assignment.
5827 2017-01-13 David Malcolm <dmalcolm@redhat.com>
5830 * substring-locations.c (format_warning_va): Strengthen case 1 so
5831 that both endpoints of the substring must be within the format
5832 range for just the substring to be printed.
5834 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
5836 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
5837 * config/i386/i386.c (ix86_target_string): Add missing options
5838 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
5839 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
5840 flags_other and ix86_target_other to flags2_other. Display unknown
5842 (ix86_valid_target_attribute_inner_p): Add missing options and
5843 reorder options by implied ISAs, as in ix86_target_string.
5845 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
5847 * hash-table.h (hash_table::too_empty_p): New function.
5848 (hash_table::expand): Use it.
5849 (hash_table::traverse): Likewise.
5850 (hash_table::empty_slot): Use sizeof (value_type) instead of
5851 sizeof (PTR) to convert bytes to elements. Shrink the table
5852 if the current size is excessive for the current number of
5855 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
5857 * ira-costs.c (record_reg_classes): Break from the inner loop
5858 early once alt_fail is known to be true. Update outer loop
5859 handling accordingly.
5861 2017-01-13 Jeff Law <law@redhat.com>
5863 * tree-ssa-dse.c (decrement_count): New function.
5864 (increment_start_addr, maybe_trim_memstar_call): Likewise.
5865 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
5866 when we know the partially dead statement is a mem* function.
5868 PR tree-optimization/61912
5869 PR tree-optimization/77485
5870 * tree-ssa-dse.c: Include expr.h.
5871 (maybe_trim_constructor_store): New function.
5872 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
5874 PR tree-optimization/33562
5875 PR tree-optimization/61912
5876 PR tree-optimization/77485
5877 * doc/invoke.texi: Document new dse-max-object-size param.
5878 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
5879 * tree-ssa-dse.c: Include params.h.
5880 (dse_store_status): New enum.
5881 (initialize_ao_ref_for_dse): New, partially extracted from
5883 (valid_ao_ref_for_dse, normalize_ref): New.
5884 (setup_live_bytes_from_ref, compute_trims): Likewise.
5885 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
5886 (maybe_trim_partially_dead_store): Likewise.
5887 (maybe_trim_complex_store): Likewise.
5888 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
5889 Track what bytes live from the original store. Return tri-state
5890 for dead, partially dead or live.
5891 (dse_dom_walker): Add constructor, destructor and new private members.
5892 (delete_dead_call, delete_dead_assignment): New extracted from
5894 (dse_optimize_stmt): Make a member of dse_dom_walker.
5895 Use initialize_ao_ref_for_dse.
5897 PR tree-optimization/33562
5898 PR tree-optimization/61912
5899 PR tree-optimization/77485
5900 * sbitmap.h (bitmap_count_bits): Prototype.
5901 (bitmap_clear_range, bitmap_set_range): Likewise.
5902 * sbitmap.c (bitmap_clear_range): New function.
5903 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
5905 2017-01-13 Martin Liska <mliska@suse.cz>
5908 * function.c (set_cfun): Add new argument force.
5909 * function.h (set_cfun): Likewise.
5910 * ipa-inline-transform.c (inline_call): Use the function when
5911 strict alising from is dropped for function we inline to.
5913 2017-01-13 Richard Biener <rguenther@suse.de>
5915 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
5916 for dumping GIMPLE INTEGER_CSTs.
5918 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5920 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
5921 to 201112L since C++17.
5923 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
5926 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
5927 if -fsanitize=kernel-address is present.
5929 2017-01-13 Richard Biener <rguenther@suse.de>
5931 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
5932 as _Literal ( type ) number in case usual suffixes do not
5933 preserve all information.
5935 2017-01-13 Richard Biener <rguenther@suse.de>
5937 PR tree-optimization/77283
5938 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
5939 and ssa-iterators.h.
5940 (is_feasible_trace): Implement a cost model based on joiner
5943 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
5946 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
5947 char or short to __float128/_Float128 directly.
5949 2017-01-12 Martin Sebor <msebor@redhat.com>
5951 to -Wformat-overflow.
5952 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
5953 (min_bytes_remaining): Same.
5954 (get_string_length): Same.
5955 (format_string): Same.
5956 (format_directive): Same.
5958 (pass_sprintf_length::handle_gimple_call): Same.
5960 2017-01-12 Jakub Jelinek <jakub@redhat.com>
5962 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
5963 info.nowrite calls with no lhs that can't throw. Return bool
5964 whether gsi_remove has been called or not.
5965 (pass_sprintf_length::handle_gimple_call): Return bool whether
5966 try_substitute_return_value called gsi_remove. Formatting fix.
5967 (pass_sprintf_length::execute): Don't use gsi_remove if
5968 handle_gimple_call returned true.
5971 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
5972 be removed due to side-effects, don't remove following barrier nor
5973 turn the successor edge into fallthru edge.
5975 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5978 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
5979 element-reversing loads and stores as not swappable.
5981 2017-01-12 Nathan Sidwell <nathan@acm.org>
5982 Nicolai Stange <nicstange@gmail.com>
5984 * combine.c (try_combine): Don't ignore result of overlap checking
5985 loop. Combine overlap & asm check into single loop.
5987 2017-01-12 Richard Biener <rguenther@suse.de>
5989 * tree-pretty-print.c (dump_generic_node): Provide -gimple
5990 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
5992 2017-01-12 Richard Biener <rguenther@suse.de>
5994 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
5995 and TS_TARGET_OPTION directly derive from TS_BASE.
5996 * tree-core.h (tree_optimization_option): Derive from tree_base.
5997 (tree_target_option): Likewise.
5999 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
6001 * config/i386/i386.c (memory_address_length): Increase len
6002 only when rip_relative_addr_p returns false.
6004 2017-01-11 Julia Koval <julia.koval@intel.com>
6006 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
6007 (OPTION_MASK_ISA_SGX_SET): New.
6008 (ix86_handle_option): Handle OPT_msgx.
6009 * config.gcc: Added sgxintrin.h.
6010 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
6011 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
6012 * config/i386/i386.c (ix86_target_string): Add -msgx.
6014 (ix86_option_override_internal): Handle new options.
6015 (ix86_valid_target_attribute_inner_p): Add sgx.
6016 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
6017 * config/i386/i386.opt: Add msgx.
6018 * config/i386/sgxintrin.h: New file.
6019 * config/i386/x86intrin.h: Add sgxintrin.h.
6021 2017-01-11 Jakub Jelinek <jakub@redhat.com>
6024 * fold-const.c (maybe_nonzero_address): Return 1 for function
6026 (tree_single_nonzero_warnv_p): Don't handle function local objects
6030 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
6033 2017-01-11 David Malcolm <dmalcolm@redhat.com>
6036 * opts.c: Include "spellcheck.h"
6037 (struct string_fragment): New struct.
6038 (struct edit_distance_traits<const string_fragment &>): New
6040 (get_closest_sanitizer_option): New function.
6041 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
6043 2017-01-11 Jakub Jelinek <jakub@redhat.com>
6045 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
6047 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
6048 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
6049 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
6050 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
6051 for initial die_offset if dwarf_split_debug_info.
6052 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
6053 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
6055 (output_skeleton_debug_sections): Formatting fix. Use
6056 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
6057 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
6059 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
6061 * config/arm/cortex-a53.md: Add bypasses for
6063 (cortex_a53_r2f): Only use for transfers.
6064 (cortex_a53_f2r): Likewise.
6065 (cortex_a53_r2f_cvt): Add reservation for conversions.
6066 (cortex_a53_f2r_cvt): Likewise.
6068 2017-01-11 Tamar Christina <tamar.christina@arm.com>
6070 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
6071 to all inlined functions, change static to extern.
6073 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
6076 * config/arm/arm.c (legitimize_pic_address): Handle reference to
6078 (arm_assemble_integer): Likewise.
6080 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
6082 * config.gcc: Use new awk script to check CPU, FPU and architecture
6083 parameters for --with-... options.
6084 * config/arm/parsecpu.awk: New file
6085 * config/arm/arm-cpus.in: New file.
6086 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
6088 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
6090 * config/arm/t-arm: Update dependency rules.
6091 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
6092 of processing .def files.
6093 * config/arm/genopt.sh: Deleted.
6094 * config/arm/gentune.sh: Deleted.
6095 * config/arm/arm-cores.def: Deleted.
6096 * config/arm/arm-arches.def: Deleted.
6097 * config/arm/arm-fpus.def: Deleted.
6098 * config/arm/arm-tune.md: Regenerated.
6099 * config/arm/arm-tables.opt: Regenerated.
6100 * config/arm/arm-cpu.h: New generated file.
6101 * config/arm/arm-cpu-data.h: New generated file.
6102 * config/arm/arm-cpu-cdata.h: New generated file.
6104 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
6107 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
6109 (input_varpool_node): Unpack dynamically_initialized bit.
6111 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
6113 PR rtl-optimization/79032
6114 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
6115 the alignment of the adjusted memory reference against that of MODE,
6116 instead of the alignment of the original memory reference.
6118 2017-01-11 Martin Jambor <mjambor@suse.cz>
6120 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
6122 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
6123 decorated functions.
6125 2017-01-11 Richard Biener <rguenther@suse.de>
6127 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
6128 set range/nonnull info for PHI results. Do not set it on
6129 stmts marked for removal.
6131 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
6133 * expr.c (store_field): In the bitfield case, fetch the return value
6134 from the registers before applying a single big-endian adjustment.
6135 Always do a final load for a BLKmode value not larger than a word.
6137 2017-01-10 David Malcolm <dmalcolm@redhat.com>
6140 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
6141 that we correctly handle column numbers greater than
6142 LINE_MAP_MAX_COLUMN_NUMBER.
6144 2017-01-10 Martin Sebor <msebor@redhat.com>
6147 * gimple-ssa-sprintf.c (get_destination_size): Call
6148 {init,fini}object_sizes.
6149 * tree-object-size.c (addr_object_size): Adjust.
6150 (pass_through_call): Adjust.
6151 (pass_object_sizes::execute): Adjust.
6152 * tree-object-size.h (fini_object_sizes): Declare.
6154 2017-01-10 Martin Sebor <msebor@redhat.com>
6156 PR tree-optimization/78775
6157 * builtins.c (get_size_range): Move...
6158 * calls.c: ...to here.
6159 (alloc_max_size): Accept zero argument.
6160 (operand_signed_p): Remove.
6161 (maybe_warn_alloc_args_overflow): Call get_size_range.
6162 * calls.h (get_size_range): Declare.
6164 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
6166 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
6167 from TI's devices.csv file as of September 2016.
6168 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
6170 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
6172 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
6173 * doc/invoke.texi: Likewise.
6174 * doc/md.texi: Likewise.
6175 * doc/objc.texi: Likewise.
6177 2017-01-10 Joshua Conner <joshconner@google.com>
6179 * config/arm/fuchsia-elf.h: New file.
6180 * config/fuchsia.h: New file.
6181 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
6182 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
6184 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
6186 2016-01-10 Richard Biener <rguenther@suse.de>
6188 PR tree-optimization/79034
6189 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
6190 Propagate out degenerate PHIs in the joiner.
6192 2017-01-10 Martin Liska <mliska@suse.cz>
6194 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
6195 (sort_congruence_classes_by_decl_uid): Likewise.
6196 (sort_congruence_class_groups_by_decl_uid): Likewise.
6197 (sem_item_optimizer::merge_classes): Sort class, groups in these
6198 classes and members in the groups by DECL_UID of declarations.
6199 This would make merge operations stable.
6201 2017-01-10 Martin Liska <mliska@suse.cz>
6203 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
6204 usage of m_classes_vec.
6205 (sem_item_optimizer::~sem_item_optimizer): Likewise.
6206 (sem_item_optimizer::get_group_by_hash): Likewise.
6207 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
6208 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
6209 (sem_item_optimizer::verify_classes): Likewise.
6210 (sem_item_optimizer::process_cong_reduction): Likewise.
6211 (sem_item_optimizer::dump_cong_classes): Likewise.
6212 (sem_item_optimizer::merge_classes): Likewise.
6213 * ipa-icf.h (congruence_class_hash): Rename from
6214 congruence_class_group_hash. Remove declaration of m_classes_vec.
6216 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
6218 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
6219 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
6220 * config.gcc: Add avx512vpopcntdqintrin.h.
6221 * config/i386/avx512vpopcntdqintrin.h: New.
6222 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
6223 * config/i386/i386-builtin-types.def: Add new types.
6224 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
6225 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
6226 __builtin_ia32_vpopcountq_v8di_mask): New.
6227 * config/i386/i386-c.c (ix86_target_macros_internal): Define
6228 __AVX512VPOPCNTDQ__.
6229 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
6230 (PTA_AVX512VPOPCNTDQ): Define.
6231 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
6232 TARGET_AVX512VPOPCNTDQ_P): Define.
6233 * config/i386/i386.opt: Add mavx512vpopcntdq.
6234 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
6235 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
6237 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
6240 * predict.def (PRED_CALL): Set to 67.
6242 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
6244 * expr.c (store_field): In the bitfield case, if the value comes from
6245 a function call and is of an aggregate type returned in registers, do
6246 not modify the field mode; extract the value in all cases if the mode
6247 is BLKmode and the size is not larger than a word.
6249 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
6252 * config/i386/cpuid.h: Fix undefined behavior.
6254 2017-01-04 Jeff Law <law@redhat.com>
6256 PR tree-optimization/79007
6257 PR tree-optimization/67955
6258 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
6259 conservative for pt.null when flag_non_call_exceptions is on.
6261 2017-01-09 Jakub Jelinek <jakub@redhat.com>
6263 PR translation/79019
6264 PR translation/79020
6265 * params.def (PARAM_INLINE_MIN_SPEEDUP,
6266 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
6267 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
6269 * config/avr/avr.opt (maccumulate-args): Likewise.
6270 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
6271 * common.opt (freport-bug): Likewise.
6272 * cif-code.def (CIF_FINAL_ERROR): Likewise.
6273 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
6274 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
6275 translatable string.
6276 * config/i386/i386.c (function_value_32): Likewise.
6277 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
6278 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
6280 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
6281 * common/config/msp430/msp430-common.c (msp430_handle_option):
6283 * symtab.c (symtab_node::verify_base): Likewise.
6284 * opts.c (set_debug_level): Likewise.
6285 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
6286 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
6287 missing whitespace to translatable strings.
6288 * config/avr/avr.md (bswapsi2): Fix typo in comment.
6289 * config/sh/superh.h: Likewise.
6290 * config/i386/xopintrin.h: Likewise.
6291 * config/i386/znver1.md: Likewise.
6292 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
6293 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
6294 * double-int.h (struct double_int): Likewise.
6295 * double-int.c (div_and_round_double): Likewise.
6296 * wide-int.cc: Likewise.
6297 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
6298 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
6299 * cfgcleanup.c (crossjumps_occured): Renamed to ...
6300 (crossjumps_occurred): ... this.
6301 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
6304 PR tree-optimization/78899
6305 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
6306 returning bool return struct loop *, NULL for failure and the new
6308 (versionable_outer_loop_p): Don't version outer loop if it has
6309 dont_vectorized bit set.
6310 (tree_if_conversion): When versioning outer loop, ensure
6311 tree_if_conversion is performed also on the inner loop of the
6312 non-vectorizable outer loop copy.
6313 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
6314 LOOP_VECTORIZED in inner loop of the scalar outer loop and
6315 prevent vectorization of it.
6316 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
6317 the outer loop vectorization of the non-scalar version is attempted
6318 before vectorization of the inner loop in scalar version. If
6319 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
6320 vectorization of its inner loop.
6321 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
6322 has 2 inner loops, rename also on edges from bb whose single pred
6323 is outer_loop->header. Fix typo in function comment.
6325 2017-01-09 Martin Sebor <msebor@redhat.com>
6328 * asan.c (asan_emit_stack_protection): Increase local buffer size
6329 to avoid snprintf truncation warning.
6331 2017-01-09 Andrew Pinski <apinski@cavium.com>
6333 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
6334 to reference thunderx2t99 for the tuning structure
6335 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
6337 (thunderx2t99_extra_costs): This.
6338 * config/aarch64/aarch64-tune.md: Regenerate.
6339 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
6340 (vulcan_addrcost_table): This.
6341 (vulcan_regmove_cost): Rename to ...
6342 (thunderx2t99_regmove_cost): This.
6343 (vulcan_vector_cost): Rename to ...
6344 (thunderx2t99_vector_cost): this.
6345 (vulcan_branch_cost): Rename to ...
6346 (thunderx2t99_branch_cost): This.
6347 (vulcan_tunings): Rename to ...
6348 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
6349 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
6351 2017-01-09 Martin Jambor <mjambor@suse.cz>
6355 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
6356 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
6357 (propagate_vr_accross_jump_function): Use the above function for all
6358 value range computations for pass-through jump functions and type
6359 converasion from explicit value range values.
6360 (ipcp_propagate_stage): Do not attempt to deduce types of formal
6361 parameters from TYPE_ARG_TYPES.
6362 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
6363 (ipa_write_node_info): Stream type of the actual argument.
6364 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
6366 2017-01-09 Martin Liska <mliska@suse.cz>
6369 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
6370 (lookup_compiler): Do not show error message with have_E.
6372 2017-01-09 Jakub Jelinek <jakub@redhat.com>
6374 PR tree-optimization/78938
6375 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
6376 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
6377 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
6378 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
6381 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6383 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
6386 2017-01-09 Richard Biener <rguenther@suse.de>
6388 PR tree-optimization/78997
6389 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
6390 name condition properly.
6392 2017-01-09 Richard Biener <rguenther@suse.de>
6395 * dwarf2out.c (is_cxx): New overload with context.
6396 (is_naming_typedef_decl): Use it.
6398 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
6400 * invoke.texi (Option Summary): Correct spacing in option lists
6401 and add line breaks to fix over-long lines.
6403 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
6407 * extend.texi (Common Variable Attributes): Add xref to GCC
6408 Internals manual to explain mode attribute keywords.
6410 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
6413 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
6414 and Preprocessor Options.
6415 (Options for Linking): Document -pthread here....
6416 (RS/6000 and PowerPC Options): ...not here.
6417 (Solaris 2 Options): ...or here.
6418 * doc/cppopts.texi: Document -pthread.
6420 2017-01-08 Martin Sebor <msebor@redhat.com>
6423 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
6424 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
6425 New member functions.
6426 (format_directive): Used them.
6428 (pass_sprintf_length::handle_gimple_call): Same.
6429 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
6430 to avoid truncation for any argument.
6431 (extract_affine_mul): Same.
6432 * tree.c (get_file_function_name): Same.
6434 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
6437 * predict.def (PRED_INDIR_CALL): Set to 86.
6439 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
6441 PR preprocessor/54124
6442 * doc/cppopts.texi: Reformat -d subtable to list the full name
6443 of the options. Add cross-reference to the docs for the general
6444 compiler -d options.
6445 * doc/invoke.texi (Developer Options): Add cross-reference to the
6446 preprocessor-specific -d option documentation.
6448 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
6450 PR preprocessor/13498
6451 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
6452 redudant material, and reflect new command-line options.
6453 (System Headers): Likewise.
6455 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
6457 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
6458 -isystem, and -idirafter. Copy-edit.
6459 * doc/cppopts.texi: Copy-edit. Remove contradiction about
6460 default for -ftrack-macro-expansion. Delete obsolete and
6461 badly-formatted implementation details about -fdebug-cpp output.
6462 * doc/cppwarnopts.texi: Copy-edit.
6464 2017-01-07 David Malcolm <dmalcolm@redhat.com>
6467 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
6468 that the transition from a max line width >= 1<<10 to narrower
6469 lines works correctly.
6471 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
6473 * doc/options.texi (PerFunction): New.
6474 * opt-functions.awk (switch_flags): Map both Optimization and
6475 PerFunction to CL_OPTIMIZATION.
6476 * opth-gen.awk: Test for PerFunction flag along with
6478 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
6479 it only when the latter is present. Skip those that don't in
6480 the hash function generator.
6481 * common.opt (fvar-tracking): Mark as PerFunction instead of
6483 (fvar-tracking-assignments): Likewise.
6484 (fvar-tracking-assignments-toggle): Likewise.
6485 (fvar-tracking-uninit): Likewise.
6487 2017-01-07 Jakub Jelinek <jakub@redhat.com>
6489 PR translation/79018
6490 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
6493 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
6496 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
6497 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
6498 TARGET_LONG_JUMP_TABLE_OFFSETS.
6499 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
6500 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
6501 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
6502 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
6503 * config/m68k/m68k.md (tablejump expander): Likewise.
6504 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
6505 TARGET_LONG_JUMP_TABLE_OFFSETS.
6506 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
6507 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
6509 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
6510 David Holsgrove <david.holsgrove@xilinx.com>
6512 * common/config/microblaze/microblaze-common.c
6513 (TARGET_EXCEPT_UNWIND_INFO): Remove.
6514 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
6516 * config/microblaze/microblaze.c (microblaze_must_save_register)
6517 (microblaze_expand_epilogue, microblaze_return_addr): Handle
6519 (microblaze_eh_return): New function.
6520 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
6521 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
6522 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
6523 * config/microblaze/microblaze.md (eh_return): New pattern.
6525 2017-01-06 Jakub Jelinek <jakub@redhat.com>
6527 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
6528 GCC_DIAGNOSTIC_STRINGIFY): Define.
6530 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
6532 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
6534 * config/arm/arm.md (<mcrr>): New.
6536 * config/arm/arm.c (arm_arch5te): New.
6537 (arm_option_override): Set arm_arch5te.
6538 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
6540 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
6541 (arm_mcrr_qualifiers): ... this. New.
6542 (MRRC_QUALIFIERS): Define to...
6543 (arm_mrrc_qualifiers): ... this. New.
6544 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
6546 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
6547 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
6548 (MRRCI, mrrc, MRRC): New.
6549 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
6550 VUNSPEC_MRRC2): New.
6552 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
6554 * config/arm/arm.md (<mcr>): New.
6556 * config/arm/arm.c (arm_coproc_builtin_available): Add
6557 support for mcr, mrc, mcr2 and mrc2.
6558 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
6559 (arm_mcr_qualifiers): ... this. New.
6560 (MRC_QUALIFIERS): Define to ...
6561 (arm_mrc_qualifiers): ... this. New.
6562 (MCR_QUALIFIERS): Define to ...
6563 (arm_mcr_qualifiers): ... this. New.
6564 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
6566 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
6567 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
6568 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
6571 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
6573 * config/arm/arm.md (*ldc): New.
6577 * config/arm/arm.c (arm_coproc_builtin_available): Add
6578 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
6579 (arm_coproc_ldc_stc_legitimate_address): New.
6580 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
6581 'qualifier_const_pointer'.
6582 (LDC_QUALIFIERS): Define to...
6583 (arm_ldc_qualifiers): ... this. New.
6584 (STC_QUALIFIERS): Define to...
6585 (arm_stc_qualifiers): ... this. New.
6586 * config/arm/arm-protos.h
6587 (arm_coproc_ldc_stc_legitimate_address): New.
6588 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
6589 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
6590 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
6591 stc2, stcl, stc2l): New.
6592 * config/arm/constraints.md (Uz): New.
6593 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
6594 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
6595 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
6596 VUNSPEC_STC2L): New.
6598 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
6600 * config/arm/arm.md (<cdp>): New.
6601 * config/arm/arm.c (neon_const_bounds): Rename this ...
6602 (arm_const_bounds): ... this.
6603 (arm_coproc_builtin_available): New.
6604 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
6605 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
6606 (CDP_QUALIFIERS): Define to...
6607 (arm_cdp_qualifiers): ... this. New.
6609 (arm_expand_builtin_args): Add case for 6 arguments.
6610 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
6611 (arm_const_bounds): ... this.
6612 (arm_coproc_builtin_available): New.
6613 * config/arm/arm_acle.h (__arm_cdp): New.
6615 * config/arm/arm_acle_builtins.def (cdp): New.
6617 * config/arm/iterators.md (CDPI,CDP,cdp): New.
6618 * config/arm/neon.md: Rename all 'neon_const_bounds' to
6620 * config/arm/types.md (coproc): New.
6621 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
6622 * gcc/doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
6623 * gcc/doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
6624 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
6626 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
6628 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
6629 (UBINOP_QUALIFIERS): New.
6631 (acle_builtin_data): New. Change comment.
6632 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
6633 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
6634 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
6635 arm_acle_builtins.def.
6636 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
6637 (arm_init_acle_builtins): New.
6638 (CRC32_BUILTIN): Remove.
6639 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
6640 crc32cb, crc32ch and crc32cw.
6641 (arm_init_crc32_builtins): Remove.
6642 (arm_init_builtins): Use arm_init_acle_builtins rather
6643 than arm_init_crc32_builtins.
6644 (arm_expand_acle_builtin): New.
6645 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
6646 * config/arm/arm_acle_builtins.def: New.
6648 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
6650 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
6651 (arm_builtin_datum): ... this.
6652 (arm_init_neon_builtin): Rename to ...
6653 (arm_init_builtin): ... this. Add a new parameters PREFIX
6654 and USE_SIG_IN_NAME.
6655 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
6656 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
6657 'arm_builtin_datum'.
6658 (arm_init_vfp_builtins): Likewise.
6659 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
6660 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
6661 (arm_expand_neon_args): Rename to ...
6662 (arm_expand_builtin_args): ... this. Rename builtin_arg
6663 enum values and differentiate between ARG_BUILTIN_MEMORY
6664 and ARG_BUILTIN_NEON_MEMORY.
6665 (arm_expand_neon_builtin_1): Rename to ...
6666 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
6667 values, arm_expand_builtin_args and add bool parameter NEON.
6668 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
6669 (arm_expand_vfp_builtin): Likewise.
6670 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
6672 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
6675 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
6676 * predict.c (tree_estimate_probability_bb): Reverse direction of
6677 polymorphic call predictor.
6679 2017-01-06 David Malcolm <dmalcolm@redhat.com>
6681 * passes.c (execute_one_pass): Split out pass-skipping logic into...
6682 (determine_pass_name_match): ...this new function and...
6683 (should_skip_pass_p): ...this new function.
6685 2017-01-06 Nathan Sidwell <nathan@acm.org>
6687 * ipa-visibility.c (function_and_variable_visibility): Reformat
6688 comments and long lines. Remove extrneous if.
6689 * symtab.c (symtab_node::make_decl_local): Fix code format.
6690 (symtab_node::set_section_for_node): Fix comment typo.
6692 2017-01-06 Martin Liska <mliska@suse.cz>
6695 * lra-constraints.c: Rename invariant to lra_invariant.
6696 * predict.c (set_even_probabilities): Initialize e to NULL.
6698 2017-01-05 Martin Sebor <msebor@redhat.com>
6700 PR tree-optimization/78910
6701 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
6702 (format_integer): Correct off-by-one error in the handling
6703 of precision with negative numbers in signed conversions..
6705 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
6707 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
6709 2017-01-05 Jakub Jelinek <jakub@redhat.com>
6711 PR tree-optimization/71016
6712 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
6713 factor_out_conditional_conversion. Formatting fix.
6714 (factor_out_conditional_conversion): Add cond_stmt argument.
6715 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
6716 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
6719 2017-01-05 David Malcolm <dmalcolm@redhat.com>
6721 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
6722 read-rtl-function.o, and selftest-rtl.o.
6723 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
6724 (selftest::aarch64_test_loading_full_dump): New function.
6725 (selftest::aarch64_run_selftests): New function.
6726 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
6727 selftest::aarch64_run_selftests.
6728 * config/i386/i386.c
6729 (selftest::ix86_test_loading_dump_fragment_1): New function.
6730 (selftest::ix86_test_loading_call_insn): New function.
6731 (selftest::ix86_test_loading_full_dump): New function.
6732 (selftest::ix86_test_loading_unspec): New function.
6733 (selftest::ix86_run_selftests): Call the new functions.
6734 * emit-rtl.c (maybe_set_max_label_num): New function.
6735 * emit-rtl.h (maybe_set_max_label_num): New decl.
6736 * function.c (instantiate_decls): Guard call to
6737 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
6738 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
6740 * gensupport.c (gen_reader::gen_reader): Pass "false"
6741 for new "compact" param of rtx_reader.
6742 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
6743 rather than an empty string for NULL strings.
6744 * read-md.c: Potentially include config.h rather than bconfig.h.
6745 Wrap include of errors.h with #ifdef GENERATOR_FILE.
6746 (have_error): New global, copied from errors.c.
6747 (md_reader::read_name): Rename to...
6748 (md_reader::read_name_1): ...this, adding "out_loc" param,
6749 and converting "missing name or number" to returning false, rather
6751 (md_reader::read_name): Reimplement in terms of read_name_1.
6752 (md_reader::read_name_or_nil): New function.
6753 (md_reader::read_string): Handle "(nil)" by returning NULL.
6754 (md_reader::md_reader): Add new param "compact".
6755 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
6756 (md_reader::read_file): New method.
6757 * read-md.h (md_reader::md_reader): Add new param "compact".
6758 (md_reader::read_file): New method.
6759 (md_reader::is_compact): New accessor.
6760 (md_reader::read_name): Convert return type from void to file_location.
6761 (md_reader::read_name_or_nil): New decl.
6762 (md_reader::read_name_1): New decl.
6763 (md_reader::m_compact): New field.
6764 (noop_reader::noop_reader): Pass "false" for new "compact" param
6766 (rtx_reader::rtx_reader): Add new "compact" param.
6767 (rtx_reader::read_rtx_operand): Make virtual and convert return
6768 type from void to rtx.
6769 (rtx_reader::read_until): New decl.
6770 (rtx_reader::handle_any_trailing_information): New virtual function.
6771 (rtx_reader::postprocess): New virtual function.
6772 (rtx_reader::finalize_string): New virtual function.
6773 (rtx_reader::m_in_call_function_usage): New field.
6774 (rtx_reader::m_reuse_rtx_by_id): New field.
6775 * read-rtl-function.c: New file.
6776 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
6777 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
6778 (selftest::verify_three_block_rtl_cfg): New decl.
6779 * read-rtl-function.h: New file.
6780 * read-rtl.c: Potentially include config.h rather than bconfig.h.
6781 For host, include function.h, memmodel.h, and emit-rtl.h.
6782 (one_time_initialization): New function.
6783 (struct compact_insn_name): New struct.
6784 (compact_insn_names): New array.
6785 (find_code): Handle insn codes in compact dumps.
6786 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
6787 (bind_subst_iter_and_attr): Likewise.
6788 (add_condition_to_string): Likewise.
6789 (add_condition_to_rtx): Likewise.
6790 (apply_attribute_uses): Likewise.
6791 (add_current_iterators): Likewise.
6792 (apply_iterators): Likewise.
6793 (initialize_iterators): Guard usage of apply_subst_iterator with
6794 #ifdef GENERATOR_FILE.
6795 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
6796 (md_reader::read_mapping): Likewise.
6797 (add_define_attr_for_define_subst): Likewise.
6798 (add_define_subst_attr): Likewise.
6799 (read_subst_mapping): Likewise.
6800 (check_code_iterator): Likewise.
6801 (rtx_reader::read_rtx): Likewise. Move one-time initialization
6803 (one_time_initialization): New function.
6804 (rtx_reader::read_until): New method.
6805 (read_flags): New function.
6806 (parse_reg_note_name): New function.
6807 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
6808 Handle reuse_rtx ids.
6809 Wrap iterator lookup within #ifdef GENERATOR_FILE.
6810 Add parsing support for RTL dumps, mirroring the special-cases in
6811 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
6812 values, and calling handle_any_trailing_information.
6813 (rtx_reader::read_rtx_operand): Convert return type from void
6814 to rtx, returning return_rtx. Handle case 'e'. Call
6815 finalize_string on XSTR and XTMPL fields.
6816 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
6817 "(nil)" values were omitted. Call the postprocess vfunc on the
6819 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
6820 class ctor. Initialize m_in_call_function_usage. Call
6821 one_time_initialization.
6822 * rtl-tests.c (selftest::test_uncond_jump): Call
6823 set_new_first_and_last_insn.
6824 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
6825 * selftest-rtl.c: New file.
6826 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
6827 (selftest::get_insn_by_uid): New decl.
6828 * selftest-run-tests.c (selftest::run_tests): Call
6829 read_rtl_function_c_tests.
6830 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
6831 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
6834 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
6836 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
6837 operands in a special way. Assert that pos+len <= mode precision.
6839 2017-01-05 Jakub Jelinek <jakub@redhat.com>
6841 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
6842 3 argument Alias with unlimited for the negative form.
6843 (fno-vect-cost-model): Removed.
6845 2017-01-05 Martin Liska <mliska@suse.cz>
6847 * hsa-gen.c (gen_hsa_divmod): New function.
6848 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
6850 2017-01-05 Martin Liska <mliska@suse.cz>
6853 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
6856 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6858 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
6859 small constant length operands.
6861 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6863 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
6864 between loop iterations.
6866 2017-01-05 Martin Liska <mliska@suse.cz>
6869 * gimplify.c (gimplify_decl_expr): Compare to
6870 asan_poisoned_variables instread of checking flags.
6871 (gimplify_target_expr): Likewise.
6872 (gimplify_expr): Likewise.
6873 (gimplify_function_tree): Conditionally initialize
6874 asan_poisoned_variables.
6876 2017-01-04 Jeff Law <law@redhat.com>
6878 PR tree-optimizatin/78812
6879 * rtl.h (contains_mem_rtx_p): Prototype.
6880 * ifcvt.c (containts_mem_rtx_p): Move from here to...
6881 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
6882 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
6883 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
6884 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
6886 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
6888 * input.c (assert_char_at_range): Default-initialize actual_range.
6890 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
6892 * df-scan.c (df_ref_create_structure): Make regno unsigned,
6893 to match the caller.
6895 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
6897 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
6898 insns after final jump in test to emit dummy move.
6900 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
6902 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
6903 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
6905 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
6907 * multiple_target.c (create_dispatcher_calls): Init e_next.
6908 * tree-ssa-loop-split.c (split_loop): Init border.
6909 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
6912 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
6917 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
6918 (altivec_register_operand): Do not return true if the operand
6919 contains a SUBREG mixing SImode and SFmode.
6920 (vsx_register_operand): Likewise.
6921 (vsx_reg_sfsubreg_ok): New predicate.
6922 (vfloat_operand): Do not return true if the operand contains a
6923 SUBREG mixing SImode and SFmode.
6924 (vint_operand): Likewise.
6925 (vlogical_operand): Likewise.
6926 (gpc_reg_operand): Likewise.
6927 (int_reg_operand): Likewise.
6928 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
6929 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
6930 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
6932 (rs6000_emit_move_si_sf_subreg): New helper function.
6933 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
6934 fixup SUBREGs involving SImode and SFmode.
6935 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
6936 numbers for the new peephole2 optimization.
6937 (peephole2 for SFmode unions): New peephole2 to optimize cases in
6938 the GLIBC math library that do AND/IOR/XOR operations on single
6939 precision floating point.
6940 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
6941 target macros to say whether we need to avoid SUBREGs mixing
6943 (TARGET_ALLOW_SF_SUBREG): Likewise.
6944 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
6945 (UNSPEC_SI_FROM_SF): Likewise.
6946 (iorxor): Change spacing.
6947 (and_ior_xor): New iterator for AND, IOR, and XOR.
6948 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
6949 (movdi_from_sf_zero_ext): Likewise.
6950 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
6951 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
6952 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
6953 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
6954 (fms<mode>4): Likewise.
6955 (fnma<mode>4): Likewise.
6956 (fnms<mode>4): Likewise.
6957 (nfma<mode>4): Likewise.
6958 (nfms<mode>4): Likewise.
6960 2017-01-04 Marek Polacek <polacek@redhat.com>
6963 * doc/invoke.texi: Document -Wpointer-compare.
6965 2017-01-04 Jakub Jelinek <jakub@redhat.com>
6967 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
6970 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
6971 descriptions for -gdwarf-5 and emit them as uleb128 instead of
6974 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
6977 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
6978 documentation of the powerpc_popcntb_ok attribute.
6979 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
6980 code to issue warning messages if a requested CPU configuration is
6981 not supported by the binary (assembler and loader) toolchain.
6982 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
6983 made to define a built-in function that has been disabled.
6984 (paired_init_builtins): Add assertion to prevent ICE if attempt is
6985 made to define a built-in function that has been disabled.
6986 (altivec_init_builtins): Add comment explaining why definition
6987 of the DST built-in functions is not preceded by an assertion
6988 check. Add assertions to prevent ICE if attempts are made to
6989 define an altivec predicate or an abs* built-in function that has
6991 (htm_init_builtins): Add comment explaining why definition of the
6992 htm built-in functions is not preceded by an assertion check.
6994 2017-01-04 Jeff Law <law@redhat.com>
6996 PR tree-optimizatin/67955
6997 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
6998 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
6999 the points-to solution does not include pt_null. Use DECL_PT_UID
7002 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
7004 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
7005 Use gen_int_mode instead of gen_lopwart for const_int operands.
7007 2017-01-04 Jakub Jelinek <jakub@redhat.com>
7009 PR tree-optimization/71563
7010 * match.pd: Simplify X << Y into X if Y is known to be 0 or
7011 out of range value - has low bits known to be zero.
7013 2017-01-04 Alan Modra <amodra@gmail.com>
7015 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
7016 * configure: Regenerate.
7017 * config.in: Regenerate.
7019 2017-01-04 Jakub Jelinek <jakub@redhat.com>
7022 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
7023 a substring of the message, but strcmp with the whole message. Ifdef
7024 ENABLE_NLS, translate the message first using dgettext.
7026 2017-01-03 Jeff Law <law@redhat.com>
7028 PR tree-optimizatin/78856
7029 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
7030 (mark_threaded_blocks): Remove code to truncate thread paths that
7031 cross multiple loop headers. Instead invalidate the cached loop
7032 iteration information and handle case of a thread path walking
7033 into an irreducible region.
7035 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
7038 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
7039 assertions. Add support for doing the signbit if the IEEE 128-bit
7040 floating point value is in a GPR.
7041 * config/rs6000/rs6000.md (Fsignbit): Delete.
7042 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
7043 Update the length attribute if the value is in a GPR.
7044 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
7045 the sign or zero extension instruction, since the value is always 0/1.
7046 (signbit<mode>2_dm2): Delete using <Fsignbit>.
7049 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
7050 extracting SImode to a GPR register so that we can generate a
7051 store, limit the vector to be in a traditional Altivec register
7052 for the vextuwrx instruction.
7054 2017-01-03 Ian Lance Taylor <iant@google.com>
7056 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
7058 2017-01-03 Martin Sebor <msebor@redhat.com>
7060 PR tree-optimization/78696
7061 * gimple-ssa-sprintf.c (format_floating): Correct handling of
7062 precision. Use MPFR for %f for greater fidelity. Correct handling
7064 (pass_sprintf_length::compute_format_length): Set width and precision
7065 specified by asrerisk to void_node for vararg functions.
7066 (try_substitute_return_value): Adjust dump output.
7068 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
7070 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
7072 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
7074 * doc/invoke.texi (SPARC options): Document -mlra as the default.
7075 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
7076 -mlra/-mno-lra was passed to the compiler.
7078 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
7080 PR rtl-optimization/65618
7081 * emit-rtl.c (try_split): Move initialization of "before" and
7082 "after" to just before the call to emit_insn_after_setloc.
7084 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
7086 * doc/md.texi (Standard Names): Remove reference to Java frontend.
7088 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
7090 * dwarf2out.c (gen_enumeration_type_die): When
7091 -gno-strict-dwarf, add a DW_AT_encoding attribute.
7093 2017-01-03 Jakub Jelinek <jakub@redhat.com>
7095 PR tree-optimization/78965
7096 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
7097 Change first argument from const call_info & to call_info &. For %n
7098 set info.nowrite to false.
7101 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
7102 possibly throwing calls.
7104 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
7105 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
7106 and fns handling, rather than in a separate case SSA_NAME.
7108 2017-01-02 Jeff Law <law@redhat.com>
7110 * config/darwin-driver.c (darwin_driver_init): Const-correctness
7111 fixes for first_period and second_period variables.
7113 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
7116 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
7117 (*insvqi_1): New insn pattern.
7118 (*insvqi_1_mem_rex64): Ditto.
7120 (*insvqi_3): Rename from *insvqi.
7122 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
7124 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
7126 * doc/cfg.texi (Edges): Remove reference to Java.
7127 (Maintaining the CFG): Ditto.
7129 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
7132 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
7133 transparent aliases.
7135 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
7138 * predict.def (PRED_CALL): Update hitrate.
7139 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
7140 * predict.c (tree_estimate_probability_bb): Split CALL predictor
7141 into direct/indirect/polymorphic variants.
7143 2017-01-01 Jakub Jelinek <jakub@redhat.com>
7145 Update copyright years.
7147 * gcc.c (process_command): Update copyright notice dates.
7148 * gcov-dump.c (print_version): Ditto.
7149 * gcov.c (print_version): Ditto.
7150 * gcov-tool.c (print_version): Ditto.
7151 * gengtype.c (create_file): Ditto.
7152 * doc/cpp.texi: Bump @copying's copyright year.
7153 * doc/cppinternals.texi: Ditto.
7154 * doc/gcc.texi: Ditto.
7155 * doc/gccint.texi: Ditto.
7156 * doc/gcov.texi: Ditto.
7157 * doc/install.texi: Ditto.
7158 * doc/invoke.texi: Ditto.
7160 Copyright (C) 2017 Free Software Foundation, Inc.
7162 Copying and distribution of this file, with or without modification,
7163 are permitted in any medium without royalty provided the copyright
7164 notice and this notice are preserved.