1 2017-10-30 Jakub Jelinek <jakub@redhat.com>
4 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
5 (struct store_immediate_info): Add bitregion_start and bitregion_end
7 (store_immediate_info::store_immediate_info): Add brs and bre
8 arguments and initialize bitregion_{start,end} from those.
9 (struct merged_store_group): Add bitregion_start, bitregion_end,
10 align_base and mask fields. Drop unnecessary struct keyword from
11 struct store_immediate_info. Add do_merge method.
12 (clear_bit_region_be): Use memset instead of loop storing zeros.
13 (merged_store_group::do_merge): New method.
14 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
15 stores as long as the surrounding bitregions have no gaps.
16 (merged_store_group::merge_overlapping): Use do_merge.
17 (merged_store_group::apply_stores): Test that bitregion_{start,end}
18 is byte aligned, rather than requiring that start and width are
19 byte aligned. Drop unnecessary struct keyword from
20 struct store_immediate_info. Allocate and populate also mask array.
21 Make start of the arrays relative to bitregion_start rather than
22 start and size them according to bitregion_{end,start} difference.
23 (struct imm_store_chain_info): Drop unnecessary struct keyword from
24 struct store_immediate_info.
25 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
26 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
27 struct keyword from struct store_immediate_info.
28 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
29 between stores as long as the surrounding bitregions have no gaps.
31 (struct split_store): Add orig non-static data member.
32 (split_store::split_store): Initialize orig to false.
33 (find_constituent_stmts): Return store_immediate_info *, non-NULL
34 if there is exactly a single original stmt. Change stmts argument
35 to pointer from reference, if NULL, don't push anything to it. Add
36 first argument, use it to optimize skipping over orig stmts that
37 are known to be before bitpos already. Simplify.
38 (split_group): Return unsigned int count how many stores are or
39 would be needed rather than a bool. Add allow_unaligned argument.
40 Change split_stores argument from reference to pointer, if NULL,
41 only do a dry run computing how many stores would be produced.
42 Rewritten algorithm to use both alignment and misalign if
43 !allow_unaligned and handle bitfield stores with gaps.
44 (imm_store_chain_info::output_merged_store): Set start_byte_pos
45 from bitregion_start instead of start. Compute allow_unaligned
46 here, if true, do 2 split_group dry runs to compute which one
47 produces fewer stores and prefer aligned if equal. Punt if
48 new count is bigger or equal than original before emitting any
49 statements, rather than during that. Remove no longer needed
50 new_ssa_names tracking. Replace num_stmts with
51 split_stores.length (). Use 32-bit stack allocated entries
52 in split_stores auto_vec. Try to reuse original store lhs/rhs1
53 if possible. Handle bitfields with gaps.
54 (pass_store_merging::execute): Ignore bitsize == 0 stores.
55 Compute bitregion_{start,end} for the stores and construct
56 store_immediate_info with that. Formatting fixes.
58 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
61 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
62 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
64 2017-10-29 Jim Wilson <wilson@tuliptree.org>
66 * gimplify.c: Include tm_p.h.
68 * common.opt (gcoff): Re-add as ignored option.
69 (gcoff1, gcoff2, gcoff3): Likewise.
71 * Makefile.in (OBJS): Delete sdbout.o.
72 (GTFILES): Delete $(srcdir)/sdbout.c.
73 * debug.h: Delete sdb_debug_hooks.
74 * final.c: Delete sdbout.h include.
75 (final_scan_insn): Delete SDB_DEBUG check.
76 (rest_of_clean_state): Likewise.
77 * output.h: Delete sdb_begin_function_line.
80 * toplev.c: Delete sdbout.h include.
81 (process_options): Delete SDB_DEBUG check.
82 * tree-core.h (tree_type_common): Delete pointer field of
84 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
86 * tree.h (TYPE_SYMTAB_POINTER): Delete.
87 (TYPE_SYMTAB_IS_POINTER): Delete.
88 (TYPE_SYMTAB_IS_DIE): Renumber.
89 * xcoffout.c: Refer to former sdbout.c file.
90 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
92 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
93 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
94 (Debugging Options): Delete -gcoff.
95 (-gstabs): Delete SDB reference.
97 (-gcoff@var{level}): Delete.
98 * doc/passes.texi (Debugging information output): Delete SDB and
100 * doc/tm.texi: Regenerate.
101 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
102 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
104 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
105 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
106 (SDB_DEBUGGING_INFO): Delete.
107 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
108 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
109 * target.def (output_source_filename): Delete COFF reference.
111 * common.opt (gcoff): Delete.
112 (gxcoff+): Update Negative chain.
113 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
115 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
116 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
117 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
119 (expand_function_start): Change sdb reference to past tense.
120 (expand_function_end): Change sdb reference to past tense.
121 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
122 * opts.c (debug_type_names): Delete coff entry.
123 (common_handle_option): Delete OPT_gcoff case.
124 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
126 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
127 * config/cris/cris.h: Delete SDB reference in comment.
128 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
129 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
130 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
131 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
133 (ix86_expand_prologue): Likewise.
134 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
135 * config/ia64/ia64.h: Likewise.
136 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
137 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
139 * config/mmix/mmix.h: Likewise.
140 * config/nds32/nds32.c: Likewise.
141 * config/stormy/storym16.h: Likewise.
142 * config/visium/visium.h: Likewise.
143 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
145 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
147 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
148 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
149 FRAME_POINTER_REGNUM point at high end of local var area.
151 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
153 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
154 Move comment around. Do not reset best_edge for a copiable
155 destination if the copy would cause a partition change.
156 (better_edge_p): Remove redundant check.
158 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
160 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
162 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
164 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
165 for math functions that have _Float<N> and _Float<N>X variants.
166 (mathfn_built_in_2): Add support for math functions that have
167 _Float<N> and _Float<N>X variants.
168 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
169 (expand_builtin_mathfn_ternary): Add support for fma with
170 _Float<N> and _Float<N>X variants.
171 (expand_builtin): Likewise.
172 (fold_builtin_3): Likewise.
173 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
174 create math function _Float<N> and _Float<N>X variants as external
176 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
177 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
178 the __builtin_ prefix and if not strict ansi, without the prefix.
179 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
180 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
181 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
182 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
183 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
184 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
185 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
186 function signatures for fma _Float<N> and _Float<N>X variants.
187 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
188 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
189 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
190 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
191 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
192 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
193 * gencfn-macros.c (print_case_cfn): Add support for math functions
194 that have _Float<N> and _Float<N>X variants.
195 (print_define_operator_list): Likewise.
196 (fltfn_suffixes): Likewise.
198 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
199 for math functions that have _Float<N> and _Float<N>X variants.
200 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
201 and _Float<N>X variants.
202 (COPYSIGN): Likewise.
205 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
206 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
208 (integer_valued_read_call_p): Likewise.
209 * fold-const-call.c (fold_const_call_ss): Likewise.
210 (fold_const_call_sss): Add support for copysign, fmin, and fmax
211 _Float<N> and _Float<N>X variants.
212 (fold_const_call_ssss): Add support for fma _Float<N> and
214 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
215 support for copysign and fma _Float<N> and _Float<N>X variants.
216 (backprop::process_builtin_call_use): Likewise.
217 * tree-call-cdce.c (can_test_argument_range); Add support for
218 sqrt _Float<N> and _Float<N>X variants.
219 (edom_only_function): Likewise.
220 (get_no_error_domain): Likewise.
221 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
222 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
223 copysign _Float<N> and _Float<N>X variants.
224 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
225 handled by machine independent code.
227 * doc/cpp.texi (Common Predefined Macros): Document defining
228 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
229 fma _Float<N> and _Float<N>X variants.
231 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
234 * config/i386/i386-modes.def (CCFPU): Remove definition.
235 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
236 (ix86_cc_modes_compatible): Ditto.
237 (ix86_expand_carry_flag_compare): Ditto.
238 (ix86_expand_int_movcc): Ditto.
239 (ix86_expand_int_addcc): Ditto.
240 (ix86_reverse_condition): Ditto.
241 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
242 Return true/false for unordered/ordered fp comparisons.
243 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
244 (ix86_prepare_fp_compare_args): Update for rename.
245 (ix86_expand_fp_compare): Update for rename. Generate unordered
246 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
247 (ix86_expand_sse_compare_and_jump): Ditto.
248 * config/i386/predicates.md (fcmov_comparison_operator):
249 Remove CCFPU mode handling.
250 (ix86_comparison_operator): Ditto.
251 (ix86_carry_flag_operator): Ditto.
252 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
253 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
254 (*cmpu<mode>_cc_i387): Ditto.
255 (FPCMP): Remove mode iterator.
256 (unord): Remove mode attribute.
257 (unord_subst): New define_subst transformation
258 (unord): New define_subst attribute.
260 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
261 (*cmpi<unord>xf_i387): Ditto.
262 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
263 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
264 using unord_subst transformation.
265 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
266 (round_saeonly): Also handle CCFP mode.
267 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
268 Remove UNSPEC_SAHF unspec handling.
270 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
272 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
274 2017-10-27 Jeff Law <law@redhat.com>
276 * gimple-ssa-sprintf.c: Include domwalk.h.
277 (class sprintf_dom_walker): New class, derived from dom_walker.
278 (sprintf_dom_walker::before_dom_children): New function.
279 (struct call_info): Moved into sprintf_dom_walker class
280 (compute_formath_length, handle_gimple_call): Likewise.
281 (sprintf_length::execute): Call the dominator walker rather
282 than walking the statements.
284 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
285 gimple statement locations.
286 (check_array_bounds): Corresponding changes. Get the statement's
287 location directly from wi->stmt.
289 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
292 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
294 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
296 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
297 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
299 2017-10-27 Jakub Jelinek <jakub@redhat.com>
302 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
303 * config/i386/i386.c (maybe_get_pool_constant): Removed.
304 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
305 maybe_get_pool_constant.
306 * config/i386/predicates.md (zero_extended_scalar_load_operand):
309 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
311 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
312 2.26 caveat. Update gas and gld versions.
313 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
316 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
318 * cgraph.h (set_malloc_flag): Declare.
319 * cgraph.c (set_malloc_flag_1): New function.
320 (set_malloc_flag): Likewise.
321 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
322 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
324 (read_ipa_call_summary): Add support for reading is_return_callee.
325 (write_ipa_call_summary): Stream is_return_callee.
326 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
327 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
328 ipa-prop.h, ipa-fnsummary.h.
329 (pure_const_names): Change to static.
330 (malloc_state_e): Define.
331 (malloc_state_names): Define.
332 (funct_state_d): Add field malloc_state.
333 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
334 (check_retval_uses): New function.
335 (malloc_candidate_p): Likewise.
336 (analyze_function): Add support for malloc attribute.
337 (pure_const_write_summary): Stream malloc_state.
338 (pure_const_read_summary): Add support for reading malloc_state.
339 (dump_malloc_lattice): New function.
340 (propagate_malloc): New function.
341 (warn_function_malloc): New function.
342 (ipa_pure_const::execute): Call propagate_malloc and
344 (pass_local_pure_const::execute): Add support for malloc attribute.
345 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
346 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
348 2017-10-27 Martin Liska <mliska@suse.cz>
350 PR gcov-profile/82457
351 * doc/invoke.texi: Document that one needs a non-strict ISO mode
352 for fork-like functions to be properly instrumented.
354 2017-10-27 Richard Biener <rguenther@suse.de>
357 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
358 info when we redirected EH.
360 2017-10-26 Michael Collison <michael.collison@arm.com>
362 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
364 (<optab>_trunchf<GPI:mode>2: New pattern.
365 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
366 * config/aarch64/iterators.md (wv): New mode attribute.
367 (vf, VF): New mode attributes.
368 (vgp, VGP): New mode attributes.
369 (s): Update attribute with SImode and DImode prefixes.
371 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
373 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
374 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
375 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
376 (nios2_option_overide): Initialize it. Don't allow R0-relative
378 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
379 (nios2_symbolic_constant_p): Likewise.
380 (nios2_legitimate_address_p): Likewise.
381 (nios2_r0rel_section_name_p): New.
382 (nios2_symbol_ref_in_r0rel_data_p): New.
383 (nios2_emit_move_sequence): Handle r0rel_constant_p.
384 (r0rel_constant_p): New.
385 (nios2_print_operand_address): Handle r0rel_constant_p.
386 (nios2_cdx_narrow_form_p): Likewise.
387 * config/nios2/nios2.opt (mr0rel-sec=): New option.
388 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
389 (Nios II Options): Document -mr0rel-sec.
391 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
393 * config/nios2/nios2.c: Include xregex.h.
394 (nios2_gprel_sec_regex): New.
395 (nios2_option_overide): Initialize it. Don't allow GP-relative
397 (nios2_small_section_name_p): Check for regex match.
398 * config/nios2/nios2.opt (mgprel-sec=): New option.
399 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
400 (Nios II Options): Document -mgprel-sec.
402 2017-10-26 Jim Wilson <wilson@tuliptree.org>
404 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
406 2017-10-26 Tom de Vries <tom@codesourcery.com>
408 PR tree-optimization/82707
409 * gimple.c (gimple_copy): Fix unsharing of
410 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
412 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
414 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
415 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
416 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
417 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
418 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
419 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
420 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
421 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
422 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
423 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
424 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
425 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
426 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
427 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
428 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
429 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
430 _mm512_mask_cmpunord_ps_mask): New intrinsics.
432 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
434 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
436 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
437 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
438 warning to rs6000.c. Remove the Undocumented flag, since it has
440 (-mabi=ibmlongdouble): Likewise.
441 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
442 already set, set the default format for long double.
443 (rs6000_debug_reg_global): Print whether long double is IBM or
445 (rs6000_option_override_internal): Rework setting long double
446 format. Only warn if the user is changing the long double default
447 and they did not use -Wno-psabi.
448 * doc/invoke.texi (PowerPC options): Update the documentation for
449 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
451 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
452 Alan Hayward <alan.hayward@arm.com>
453 David Sherwood <david.sherwood@arm.com>
455 * rtl.h (wider_subreg_mode): New function.
456 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
457 rather than an unsigned int *.
458 * ira-color.c (regno_max_ref_width): Replace with...
459 (regno_max_ref_mode): ...this new variable.
460 (coalesced_pseudo_reg_slot_compare): Update accordingly.
461 Use wider_subreg_mode.
462 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
463 rather than an unsigned int *.
464 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
465 (process_alt_operands): Likewise.
466 (invariant_p): Likewise.
467 * lra-spills.c (assign_mem_slot): Likewise.
468 (add_pseudo_to_slot): Likewise.
469 * lra.c (collect_non_operand_hard_regs): Likewise.
470 (add_regs_to_insn_regno_info): Likewise.
471 * reload1.c (regno_max_ref_width): Replace with...
472 (regno_max_ref_mode): ...this new variable.
473 (reload): Update accordingly. Update call to
474 ira_sort_regnos_for_alter_reg.
475 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
476 (init_eliminable_invariants): Update to use regno_max_ref_mode.
477 (scan_paradoxical_subregs): Likewise.
479 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
481 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
482 (aarch64_frame): Add emit_frame_chain boolean.
483 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
484 Move eh_return case to aarch64_layout_frame.
485 (aarch64_layout_frame): Initialize emit_frame_chain.
486 (aarch64_expand_prologue): Use emit_frame_chain.
488 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
490 * config/aarch64/aarch64.c (aarch64_layout_frame):
491 Ensure LR is always stored at the bottom of the callee-saves.
492 Remove rarely used frame layout which saves callee-saves at top of
493 frame, so the store of LR can be used as a valid probe in all cases.
495 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
497 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
498 Improve unaligned TImode/TFmode base/offset split.
500 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
501 Alan Hayward <alan.hayward@arm.com>
502 David Sherwood <david.sherwood@arm.com>
504 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
505 * combine.c (find_single_use_1): Likewise.
506 (expand_field_assignment): Likewise.
507 (move_deaths): Likewise.
508 * lra-constraints.c (simplify_operand_subreg): Likewise.
509 (curr_insn_transform): Likewise.
510 * lra.c (collect_non_operand_hard_regs): Likewise.
511 (add_regs_to_insn_regno_info): Likewise.
512 * rtlanal.c (reg_referenced_p): Likewise.
513 (covers_regno_no_parallel_p): Likewise.
515 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
517 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
518 Don't print any bits outside the precision of the value.
519 * wide-int.cc (test_printing): Add some new tests.
521 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
523 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
524 supports -xbrace_comment option.
525 * configure: Regenerate.
526 * config.in: Regenerate.
527 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
528 (ASM_CPU_SPEC): Use it.
530 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
532 * target.def (static_rtx_alignment): New hook.
533 * targhooks.h (default_static_rtx_alignment): Declare.
534 * targhooks.c (default_static_rtx_alignment): New function.
535 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
536 * doc/tm.texi: Regenerate.
537 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
538 instead of targetm.constant_alignment. Remove call to
540 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
541 (cris_preferred_mininum_alignment): New function, split out from...
542 (cris_constant_alignment): ...here.
543 (cris_static_rtx_alignment): New function.
544 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
546 (ix86_constant_alignment): ...here.
547 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
548 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
549 (mmix_static_rtx_alignment): New function.
550 * config/spu/spu.c (spu_static_rtx_alignment): New function.
551 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
553 2017-10-26 Tamar Christina <tamar.christina@arm.com>
556 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
557 Add flag_trapping_math and flag_fp_int_builtin_inexact.
559 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
561 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
562 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
563 mark as a sign-extending load.
564 (local_pic_load_u): Define.
566 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
569 * fold-const.c (operand_equal_for_comparison_p): Also return true
570 if ARG0 is a simple variant of ARG1 with narrower precision.
571 (fold_ternary_loc): Always pass unstripped operands to the predicate.
573 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
575 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
577 * i386.h (processor_costs): Add gather_static, gather_per_elt,
578 scatter_static, scatter_per_elt.
579 * x86-tune-costs.h: Add new cost entries.
581 2017-10-25 Richard Biener <rguenther@suse.de>
583 * tree-ssa-sccvn.h (vn_eliminate): Declare.
584 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
585 class pass_fre): Move to ...
586 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
587 class pass_fre): ... here and adjust for statistics.
589 2017-10-25 Jakub Jelinek <jakub@redhat.com>
592 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
593 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
594 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
595 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
598 2017-10-25 Richard Biener <rguenther@suse.de>
600 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
601 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
602 eliminate_push_avail, eliminate_insert): Move inside...
603 (class eliminate_dom_walker): ... this class in preparation
605 (fini_eliminate): Remove by merging with ...
606 (eliminate): ... this function. Adjust for class changes.
607 (pass_pre::execute): Remove fini_eliminate call.
608 (pass_fre::execute): Likewise.
610 2017-10-24 Jakub Jelinek <jakub@redhat.com>
613 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
614 (VPERMI2, VPERMI2I): New mode iterators.
615 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
616 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
618 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
619 mode iterator. Remove 3 old define_insn patterns.
620 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
621 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
622 VPERMI2 mode iterator, remove the other two expanders.
623 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
624 to use VPERMI2 mode iterator, add another alternative for vpermi2*
625 instructions, remove the other two patterns.
626 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
627 mode iterator, remove the other two patterns.
628 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
629 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
630 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
631 and adjust argument order accordingly.
632 (ix86_expand_vec_perm): Adjust caller.
633 (expand_vec_perm_1): Likewise.
634 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
635 (expand_vec_perm_vpermt2_vpshub2): ... this.
636 (ix86_expand_vec_perm_const_1): Adjust caller.
637 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
640 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
641 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
642 (vec_shl_<mode>): Remove unused expander.
643 (avx512bw_<shift_insn><mode>3): New define_insn.
644 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
645 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
647 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
650 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
653 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
655 PR rtl-optimization/82396
656 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
657 (autopref_multipass_init): Simplify initialization.
658 (autopref_rank_data): Simplify sort order.
659 * gcc/sched-int.h (autopref_multipass_data_): Remove
660 multi_mem_insn_p, min_offset and max_offset.
662 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
665 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
666 Check special value of flag_omit_frame_pointer.
667 (aarch64_can_eliminate): Likewise.
668 (aarch64_override_options_after_change_1): Simplify handling of
669 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
671 2017-10-24 Richard Biener <rguenther@suse.de>
673 PR tree-optimization/82697
674 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
675 zero for conditional load and unconditional store.
677 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
679 * doc/install.texi: Document bootstrap-cet.
681 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
684 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
685 ENDBR instruction at function entrance if function is only
688 2017-10-24 Jakub Jelinek <jakub@redhat.com>
691 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
692 patterns to better describe from which operation the CF is computed.
693 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
694 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
695 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
696 is 0, use _0 suffixed expanders instead of emitting a comparison
699 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
701 * config/i386/i386.md(*movsf_internal, *movdf_internal):
702 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
704 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
707 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
708 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
709 * loop-iv.c (iv_get_reaching_def): Likewise.
710 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
711 variable is promoted and the partition contains undefined values.
713 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
715 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
717 (nios2_address_cost): Define.
718 (nios2_legitimize_address): Recognize (exp + constant) directly.
719 (TARGET_ADDRESS_COST): Define.
721 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
723 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
724 (nios2_symbolic_memory_operand_p): Declare.
725 (nios2_split_large_constant): Declare.
726 (nios2_split_symbolic_memory_operand): Declare.
727 * config/nios2/nios2.c: Adjust includes.
728 (nios2_symbolic_constant_allowed): New.
729 (nios2_symbolic_constant_p): New.
730 (nios2_plus_symbolic_constant_p): New.
731 (nios2_valid_addr_expr_p): Recognize addresses involving
733 (nios2_legitimate_address_p): Likewise, also LO_SUM.
734 (nios2_symbolic_memory_operand_p): New.
735 (nios2_large_constant_p): New.
736 (nios2_split_large_constant): New.
737 (nios2_split_plus_large_constant): New.
738 (nios2_split_symbolic_memory_operand): New.
739 (nios2_legitimize_address): Code refactoring. Handle addresses
740 involving symbolic constants.
741 (nios2_emit_move_sequence): Likewise.
742 (nios2_print_operand): Improve error output.
743 (nios2_print_operand_address): Handle LO_SUM.
744 (nios2_cdx_narrow_form_p): Likewise.
745 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
746 operands involving symbolic constants.
747 (movhi_internal, movsi_internal): Likewise.
748 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
749 (extendhisi2, extendqi<mode>2): Likewise.
751 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
753 * tree-pass.h (PROP_rtl_split_insns): Define.
754 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
756 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
758 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
760 2017-10-23 Jakub Jelinek <jakub@redhat.com>
763 * target.def (const_not_ok_for_debug_p): Default to
764 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
765 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
766 * targhooks.c (default_const_not_ok_for_debug_p): New function.
767 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
768 which targetm.const_not_ok_for_debug_p returned true.
769 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
771 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
773 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
774 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
775 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
777 (ix86_const_not_ok_for_debug_p): New function.
778 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
779 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
781 2017-10-23 David Malcolm <dmalcolm@redhat.com>
784 * system.h: Conditionally include "unique-ptr.h" if
785 INCLUDE_UNIQUE_PTR is defined.
786 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
787 of defining INCLUDE_UNIQUE_PTR before including "system.h".
789 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
791 * config/rl78/rl78.md: New define_expand "subdi3".
793 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
796 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
797 DF_REF_INSN if DF_REF_INSN_INFO is false.
799 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
801 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
802 xmm_move instead of sse_move.
803 (sse_store_index): New function.
804 (ix86_register_move_cost): Be more sensible about mismatch stall;
805 model AVX moves correctly; make difference between sse->integer and
807 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
808 moves; make difference between SSE and AVX.
809 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
810 and zmm_move. Increase size of sse load and store tables;
811 add unaligned load and store tables; add ssemmx_to_integer.
812 * x86-tune-costs.h: Update all entries according to real
813 move latencies from Agner Fog's manual and chip documentation.
815 2017-10-23 Jakub Jelinek <jakub@redhat.com>
818 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
819 * config/i386/constraints.md (Wf): New constraint.
820 * config/i386/i386.md (UNSPEC_SBB): New unspec.
821 (cmp<dwi>_doubleword): Removed.
822 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
823 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
824 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
825 expand with cmp<dwi>_doubleword. For LTU and GEU use
826 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
828 * common.opt (gcolumn-info): Enable by default.
829 * doc/invoke.texi (gcolumn-info): Document new default.
831 2017-10-23 Richard Biener <rguenther@suse.de>
833 PR tree-optimization/82672
834 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
835 Fold the stmt if we propagated into it.
837 2017-10-23 Richard Biener <rguenther@suse.de>
839 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
840 (bitmap_remove_expr_from_set): ... this. All callers call this
841 for non-constant values.
842 (bitmap_set_subtract): Rename to...
843 (bitmap_set_subtract_expressions): ... this. Adjust and
845 (bitmap_set_contains_value): Remove superfluous check.
846 (bitmap_set_replace_value): Inline into single caller ...
847 (bitmap_value_replace_in_set): ... here and simplify.
848 (dependent_clean): Merge into ...
849 (clean): ... this using an overload. Adjust.
850 (prune_clobbered_mems): Adjust.
851 (compute_antic_aux): Likewise.
852 (compute_partial_antic_aux): Likewise.
854 2017-10-23 Richard Biener <rguenther@suse.de>
856 PR tree-optimization/82129
858 2017-08-01 Richard Biener <rguenther@suse.de>
860 PR tree-optimization/81181
861 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
862 (compute_antic): ... end of iteration here.
864 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
866 * target.def (starting_frame_offset): New hook.
867 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
868 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
869 * doc/tm.texi.in: Regenerate.
870 * hooks.h (hook_hwi_void_0): Declare.
871 * hooks.c (hook_hwi_void_0): New function.
872 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
873 STARTING_FRAME_OFFSET.
874 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
875 * reload1.c (reload): Likewise.
876 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
877 instead of STARTING_FRAME_OFFSET.
878 * function.c (try_fit_stack_local): Likewise.
879 (assign_stack_local_1): Likewise
880 (instantiate_virtual_regs): Likewise.
881 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
882 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
883 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
884 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
885 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
886 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
887 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
888 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
889 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
890 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
891 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
892 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
893 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
894 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
895 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
896 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
897 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
898 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
899 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
900 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
901 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
902 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
903 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
904 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
905 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
906 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
907 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
908 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
909 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
910 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
911 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
912 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
913 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
914 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
915 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
916 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
917 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
918 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
919 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
920 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
921 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
922 * config/avr/avr.c (avr_starting_frame_offset): Make static and
923 return a HOST_WIDE_INT.
924 (avr_builtin_setjmp_frame_value): Use it instead of
925 STARTING_FRAME_OFFSET.
926 (TARGET_STARTING_FRAME_OFFSET): Redefine.
927 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
928 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
930 (TARGET_STARTING_FRAME_OFFSET): Redefine.
931 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
932 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
933 (TARGET_CONSTANT_ALIGNMENT): Redefine.
934 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
935 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
936 (TARGET_STARTING_FRAME_OFFSET): Redefine.
937 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
938 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
939 (TARGET_STARTING_FRAME_OFFSET): Redefine.
940 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
941 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
943 (TARGET_STARTING_FRAME_OFFSET): Redefine.
944 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
945 * config/mips/mips.c (mips_compute_frame_info): Refer to
946 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
947 (mips_starting_frame_offset): New function.
948 (TARGET_STARTING_FRAME_OFFSET): Redefine.
949 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
950 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
951 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
952 and return a HOST_WIDE_INT.
953 (TARGET_STARTING_FRAME_OFFSET): Redefine.
954 (mmix_initial_elimination_offset): Refer to
955 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
956 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
957 * config/pa/pa.c (pa_starting_frame_offset): New function.
958 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
959 (pa_expand_prologue): Likewise.
960 (TARGET_STARTING_FRAME_OFFSET): Redefine.
961 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
962 !FRAME_GROWS_DOWNWARD handling to...
963 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
964 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
965 !FRAME_GROWS_DOWNWARD handling to...
966 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
967 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
968 !FRAME_GROWS_DOWNWARD handling to...
969 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
970 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
972 (rs6000_starting_frame_offset): New function.
973 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
974 !FRAME_GROWS_DOWNWARD handling to...
975 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
976 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
977 !FRAME_GROWS_DOWNWARD handling to...
978 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
979 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
980 !FRAME_GROWS_DOWNWARD handling to...
981 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
982 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
983 (rs6000_starting_frame_offset): New function.
984 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
985 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
986 * config/vax/vax.c (vax_starting_frame_offset): New function.
987 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
988 (TARGET_STARTING_FRAME_OFFSET): Redefine.
989 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
990 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
991 (TARGET_STARTING_FRAME_OFFSET): Redefine.
992 * system.h (STARTING_FRAME_OFFSET): Poison.
994 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
996 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
997 SCALAR_TYPE_MODE instead of TYPE_MODE.
999 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
1000 Alan Hayward <alan.hayward@arm.com>
1001 David Sherwood <david.sherwood@arm.com>
1003 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
1005 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
1006 Alan Hayward <alan.hayward@arm.com>
1007 David Sherwood <david.sherwood@arm.com>
1009 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
1011 2017-10-23 Richard Biener <rguenther@suse.de>
1013 PR tree-optimization/82129
1014 * tree-ssa-pre.c (bitmap_set_and): Remove.
1015 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
1016 canonicalizing expressions in the set to those with lowest
1017 ID rather than taking that from the first edge.
1019 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
1021 * combine.c (rtx_equal_for_field_assignment_p): Use
1022 byte_lowpart_offset.
1024 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1025 Alan Hayward <alan.hayward@arm.com>
1026 David Sherwood <david.sherwood@arm.com>
1028 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
1029 to a SUBREG_PROMOTED_VAR.
1031 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1032 Alan Hayward <alan.hayward@arm.com>
1033 David Sherwood <david.sherwood@arm.com>
1035 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
1036 (expand_debug_source_expr): Likewise.
1037 * combine.c (combine_simplify_rtx): Likewise.
1038 * cse.c (fold_rtx): Likewise.
1039 * optabs.c (expand_float): Likewise.
1040 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1041 (simplify_binary_operation_1): Likewise.
1043 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1044 Alan Hayward <alan.hayward@arm.com>
1045 David Sherwood <david.sherwood@arm.com>
1047 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
1048 (record_promoted_value): Likewise.
1049 * expr.c (expand_expr_real_2): Likewise.
1050 * ree.c (update_reg_equal_equiv_notes): Likewise.
1051 (combine_set_extension): Likewise.
1052 * rtlanal.c (low_bitmask_len): Likewise.
1053 * simplify-rtx.c (neg_const_int): Likewise.
1054 (simplify_binary_operation_1): Likewise.
1056 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1057 Alan Hayward <alan.hayward@arm.com>
1058 David Sherwood <david.sherwood@arm.com>
1060 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
1061 * regcprop.c (maybe_mode_change): Likewise.
1062 * reload1.c (alter_reg): Likewise.
1064 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1066 * inchash.h (inchash::hash::add_wide_int): New function.
1067 * lto-streamer-out.c (hash_tree): Use it.
1069 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1071 * inchash.h (inchash::hash::add_wide_int): Rename to...
1072 (inchash::hash::add_hwi): ...this.
1073 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
1074 (polymorphic_call_target_hasher::hash): Likewise.
1075 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
1076 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
1077 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
1078 * lto-streamer-out.c (hash_tree): Likewise.
1079 * optc-save-gen.awk: Likewise.
1080 * tree.c (add_expr): Likewise.
1082 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
1085 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
1086 for ordered inequality comparisons even with TARGET_IEEE_FP.
1088 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
1091 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
1092 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
1093 Expand with cmp<dwi>_doubleword.
1095 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1097 * extend.texi: Add x86 specific to 'nocf_check' attribute.
1098 List CET intrinsics.
1099 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
1100 specific to -fcf-protection option.
1102 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1104 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
1105 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
1106 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
1107 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
1108 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
1109 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
1110 (extra_objs): Add cet.o for Linux/x86 targets.
1111 (tmake_file): Add i386/t-cet for Linux/x86 targets.
1112 * config/i386/cet.c: New file.
1113 * config/i386/cetintrin.h: Likewise.
1114 * config/i386/t-cet: Likewise.
1115 * config/i386/cpuid.h (bit_SHSTK): New.
1116 (bit_IBT): Likewise.
1117 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
1118 pass IBT and SHSTK bits.
1119 * config/i386/i386-builtin-types.def
1120 (VOID_FTYPE_UNSIGNED_PVOID): New.
1121 (VOID_FTYPE_UINT64_PVOID): Likewise.
1122 * config/i386/i386-builtin.def: Add CET intrinsics.
1123 * config/i386/i386-c.c (ix86_target_macros_internal): Add
1124 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
1125 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
1126 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
1128 * config/i386/i386.c (rest_of_insert_endbranch): New.
1129 (pass_data_insert_endbranch): Likewise.
1130 (pass_insert_endbranch): Likewise.
1131 (make_pass_insert_endbranch): Likewise.
1132 (ix86_notrack_prefixed_insn_p): Likewise.
1133 (ix86_target_string): Add -mibt, -mshstk flags.
1134 (ix86_option_override_internal): Add flag_cf_protection
1136 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
1137 (ix86_print_operand): Add 'notrack' prefix output.
1138 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
1139 (ix86_expand_builtin): Expand CET intrinsics.
1140 (x86_output_mi_thunk): Add 'endbranch' instruction.
1141 * config/i386/i386.h (TARGET_IBT): New.
1142 (TARGET_IBT_P): Likewise.
1143 (TARGET_SHSTK): Likewise.
1144 (TARGET_SHSTK_P): Likewise.
1145 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
1146 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
1147 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
1148 (builtin_setjmp_setup): New pattern.
1149 (builtin_longjmp): Likewise.
1150 (rdssp<mode>): Likewise.
1151 (incssp<mode>): Likewise.
1152 (saveprevssp): Likewise.
1153 (rstorssp): Likewise.
1154 (wrss<mode>): Likewise.
1155 (wruss<mode>): Likewise.
1156 (setssbsy): Likewise.
1157 (clrssbsy): Likewise.
1158 (nop_endbr): Likewise.
1159 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
1161 * config/i386/immintrin.h: Include <cetintrin.h>.
1162 * config/i386/linux-common.h
1163 (file_end_indicate_exec_stack_and_cet): New prototype.
1164 (TARGET_ASM_FILE_END): New.
1166 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
1168 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
1169 latencies instead of having separate table; make difference between
1170 integer and float costs.
1171 * i386.h (processor_costs): Remove scalar_stmt_cost,
1172 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
1173 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
1175 * x86-tune-costs.h: Remove entries which has been removed in
1176 procesor_costs from all tables; make cond_taken_branch_cost
1177 and cond_not_taken_branch_cost COST_N_INSNS based.
1179 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
1181 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
1183 2017-10-20 Jakub Jelinek <jakub@redhat.com>
1185 * config/i386/i386.md (isa): Remove fma_avx512f.
1186 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
1187 <avx512>_fmadd_<mode>_mask3<round_name>,
1188 <avx512>_fmsub_<mode>_mask<round_name>,
1189 <avx512>_fmsub_<mode>_mask3<round_name>,
1190 <avx512>_fnmadd_<mode>_mask<round_name>,
1191 <avx512>_fnmadd_<mode>_mask3<round_name>,
1192 <avx512>_fnmsub_<mode>_mask<round_name>,
1193 <avx512>_fnmsub_<mode>_mask3<round_name>,
1194 <avx512>_fmaddsub_<mode>_mask<round_name>,
1195 <avx512>_fmaddsub_<mode>_mask3<round_name>,
1196 <avx512>_fmsubadd_<mode>_mask<round_name>,
1197 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
1198 (*vec_widen_umult_even_v16si<mask_name>,
1199 *vec_widen_smult_even_v16si<mask_name>): Likewise.
1200 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
1202 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1204 * extend.texi: Add 'nocf_check' documentation.
1205 * gimple.texi: Add second parameter to
1206 gimple_build_call_from_tree.
1207 * invoke.texi: Add -fcf-protection documentation.
1208 * rtl.texi: Add REG_CALL_NOTRACK documenation.
1210 2017-10-20 Richard Biener <rguenther@suse.de>
1212 PR tree-optimization/82473
1213 * tree-vect-loop.c (vectorizable_reduction): Properly get at
1214 the largest input type.
1216 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1218 * c-attribs.c (handle_nocf_check_attribute): New function.
1219 (c_common_attribute_table): Add 'nocf_check' handling.
1220 * gimple-parser.c: Add second argument NULL to
1221 gimple_build_call_from_tree.
1222 * attrib.c (comp_type_attributes): Check nocf_check attribute.
1223 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
1225 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
1226 * common.opt: Add fcf-protection flag.
1227 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
1228 * flag-types.h: Add enum cf_protection_level.
1229 * gimple.c (gimple_build_call_from_tree): Add second parameter.
1230 Add 'nocf_check' attribute propagation to gimple call.
1231 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
1232 (gimple_build_call_from_tree): Update prototype.
1233 (gimple_call_nocf_check_p): New function.
1234 (gimple_call_set_nocf_check): Likewise.
1235 * gimplify.c: Add second argument to gimple_build_call_from_tree.
1236 * ipa-icf.c: Add nocf_check attribute in statement hash.
1237 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
1238 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
1239 * toplev.c (process_options): Add flag_cf_protection handling.
1241 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
1243 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
1245 2017-10-20 Richard Biener <rguenther@suse.de>
1247 PR tree-optimization/82603
1248 * tree-if-conv.c (predicate_mem_writes): Make sure to only
1249 remove false predicated stores.
1251 2017-10-20 Richard Biener <rguenther@suse.de>
1253 * graphite-isl-ast-to-gimple.c
1254 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
1255 Remove return value and simplify, dump copied stmt after lhs
1257 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
1258 Reduce dump verbosity.
1259 (gsi_insert_earliest): Likewise.
1260 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
1261 * graphite.c (print_global_statistics): Adjust dumping.
1262 (print_graphite_scop_statistics): Likewise.
1263 (print_graphite_statistics): Do not dump loops here.
1264 (graphite_transform_loops): But here.
1266 2017-10-20 Nicolas Roche <roche@adacore.com>
1268 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
1269 * configure: Regenerate.
1271 2017-10-20 Jakub Jelinek <jakub@redhat.com>
1274 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
1275 functions when optimizing replace GIMPLE_RETURN stmts with
1276 calls to __builtin_unreachable ().
1279 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
1280 for -fsanitize=thread link of executables.
1281 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
1282 link of executables.
1285 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
1287 (<shift_insn><mode>3<mask_name>): Change the last of the 3
1288 define_insns for logical vector shifts to use VI248_AVX512BW
1289 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
1290 condition, useless isa and prefix attributes. Change the first
1291 2 of these define_insns to ...
1292 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
1293 define_insn for avx512vl.
1294 (<shift_insn><mode>3): ... and this, new define_insn without
1295 masking for non-avx512vl.
1298 * config/i386/sse.md (*andnot<mode>3,
1299 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
1300 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
1301 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
1302 not applied use empty suffix even for TARGET_AVX512VL.
1303 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
1304 is applied, supply evex,evex or evex,evex,evex instead of just
1307 2017-10-20 Julia Koval <julia.koval@intel.com>
1309 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
1310 (OPTION_MASK_ISA_GFNI_UNSET): New.
1311 (ix86_handle_option): Handle OPT_mgfni.
1312 * config/i386/cpuid.h (bit_GFNI): New.
1313 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
1314 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
1315 * config/i386/i386.c (ix86_target_string): Add -mgfni.
1316 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
1317 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
1318 * config/i386/i386.opt: Add mgfni.
1320 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
1322 * config/msp430/msp430.c (msp430_option_override): Disable
1323 -fdelete-null-pointer-checks.
1324 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
1326 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
1328 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
1329 of x87 and SSE instructions.
1331 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
1333 * asan.c (create_cond_insert_point): Do not update edge count.
1334 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
1335 (afdo_propagate_circuit): Likewise.
1336 (afdo_calculate_branch_prob): Likewise.
1337 (afdo_annotate_cfg): Likewise.
1338 * basic-block.h (struct edge_def): Remove count.
1339 (edge_def::count): New accessor.
1340 * bb-reorder.c (rotate_loop): Update.
1341 (find_traces_1_round): Update.
1342 (connect_traces): Update.
1343 (sanitize_hot_paths): Update.
1344 * cfg.c (unchecked_make_edge): Update.
1345 (make_single_succ_edge): Update.
1346 (check_bb_profile): Update.
1347 (dump_edge_info): Update.
1348 (update_bb_profile_for_threading): Update.
1349 (scale_bbs_frequencies_int): Update.
1350 (scale_bbs_frequencies_gcov_type): Update.
1351 (scale_bbs_frequencies_profile_count): Update.
1352 (scale_bbs_frequencies): Update.
1353 * cfganal.c (connect_infinite_loops_to_exit): Update.
1354 * cfgbuild.c (compute_outgoing_frequencies): Update.
1355 (find_many_sub_basic_blocks): Update.
1356 * cfgcleanup.c (try_forward_edges): Update.
1357 (try_crossjump_to_edge): Update
1358 * cfgexpand.c (expand_gimple_cond): Update
1359 (expand_gimple_tailcall): Update
1360 (construct_exit_block): Update
1361 * cfghooks.c (verify_flow_info): Update
1362 (redirect_edge_succ_nodup): Update
1363 (split_edge): Update
1364 (make_forwarder_block): Update
1365 (duplicate_block): Update
1366 (account_profile_record): Update
1367 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
1368 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
1369 * cfgloopmanip.c (scale_loop_profile): Update.
1371 (lv_adjust_loop_entry_edge): Update.
1372 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
1373 (force_nonfallthru_and_redirect): Update.
1374 (purge_dead_edges): Update.
1375 (rtl_flow_call_edges_add): Update.
1376 * cgraphunit.c (init_lowered_empty_function): Update.
1377 (cgraph_node::expand_thunk): Update.
1378 * gimple-pretty-print.c (dump_probability): Update.
1379 (dump_edge_probability): Update.
1380 * gimple-ssa-isolate-paths.c (isolate_path): Update.
1381 * haifa-sched.c (sched_create_recovery_edges): Update.
1382 * hsa-gen.c (convert_switch_statements): Update.
1383 * ifcvt.c (dead_or_predicable): Update.
1384 * ipa-inline-transform.c (inline_transform): Update.
1385 * ipa-split.c (split_function): Update.
1386 * ipa-utils.c (ipa_merge_profiles): Update.
1387 * loop-doloop.c (add_test): Update.
1388 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
1389 * lto-streamer-in.c (input_cfg): Update.
1390 (input_function): Update.
1391 * lto-streamer-out.c (output_cfg): Update.
1392 * modulo-sched.c (sms_schedule): Update.
1393 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
1394 * predict.c (maybe_hot_edge_p): Update.
1395 (unlikely_executed_edge_p): Update.
1396 (probably_never_executed_edge_p): Update.
1397 (dump_prediction): Update.
1398 (drop_profile): Update.
1399 (propagate_unlikely_bbs_forward): Update.
1400 (determine_unlikely_bbs): Update.
1401 (force_edge_cold): Update.
1402 * profile.c (compute_branch_probabilities): Update.
1403 * reg-stack.c (better_edge): Update.
1404 * shrink-wrap.c (handle_simple_exit): Update.
1405 * tracer.c (better_p): Update.
1406 * trans-mem.c (expand_transaction): Update.
1407 (split_bb_make_tm_edge): Update.
1408 * tree-call-cdce.c: Update.
1409 * tree-cfg.c (gimple_find_sub_bbs): Update.
1410 (gimple_split_edge): Update.
1411 (gimple_duplicate_sese_region): Update.
1412 (gimple_duplicate_sese_tail): Update.
1413 (gimple_flow_call_edges_add): Update.
1414 (insert_cond_bb): Update.
1415 (execute_fixup_cfg): Update.
1416 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
1417 * tree-complex.c (expand_complex_div_wide): Update.
1418 * tree-eh.c (lower_resx): Update.
1419 (unsplit_eh): Update.
1420 (cleanup_empty_eh_move_lp): Update.
1421 * tree-inline.c (copy_edges_for_bb): Update.
1422 (freqs_to_counts): Update.
1423 (copy_cfg_body): Update.
1424 * tree-ssa-dce.c (remove_dead_stmt): Update.
1425 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
1426 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
1427 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
1428 (unloop_loops): Update.
1429 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
1430 * tree-ssa-loop-split.c (connect_loops): Update.
1431 (split_loop): Update.
1432 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
1433 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
1434 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
1435 * tree-ssa-reassoc.c (branch_fixup): Update.
1436 * tree-ssa-tail-merge.c (replace_block_by): Update.
1437 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
1438 (compute_path_counts): Update.
1439 (update_profile): Update.
1440 (recompute_probabilities): Update.
1441 (update_joiner_offpath_counts): Update.
1442 (estimated_freqs_path): Update.
1443 (freqs_to_counts_path): Update.
1444 (clear_counts_path): Update.
1445 (ssa_fix_duplicate_block_edges): Update.
1446 (duplicate_thread_path): Update.
1447 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
1448 (case_bit_test_cmp): Update.
1449 (collect_switch_conv_info): Update.
1450 (gen_inbound_check): Update.
1451 (do_jump_if_equal): Update.
1452 (emit_cmp_and_jump_insns): Update.
1453 * tree-tailcall.c (decrease_profile): Update.
1454 (eliminate_tail_call): Update.
1455 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
1456 (vect_do_peeling): Update.
1457 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
1458 * ubsan.c (ubsan_expand_null_ifn): Update.
1459 (ubsan_expand_ptr_ifn): Update.
1460 * value-prof.c (gimple_divmod_fixed_value): Update.
1461 (gimple_mod_pow2): Update.
1462 (gimple_mod_subtract): Update.
1463 (gimple_ic): Update.
1464 (gimple_stringop_fixed_value): Update.
1466 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
1469 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
1471 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
1473 PR rtl-optimization/82395
1474 * ira-color.c (allocno_priority_compare_func): Fix comparison step
1475 based on non_spilled_static_chain_regno_p.
1477 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
1479 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
1480 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
1481 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
1483 2017-10-19 Martin Sebor <msebor@redhat.com>
1485 PR tree-optimization/82596
1486 * tree.c (array_at_struct_end_p): Handle STRING_CST.
1488 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
1490 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
1491 (get_mem_refs_of_builtin_call): Likewise.
1492 * builtins.c (expand_builtin_apply): Adjust call to
1493 allocate_dynamic_stack_space.
1494 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
1495 the third argument to allocate_dynamic_stack_space, otherwise -1.
1496 (expand_builtin): Deal with all alloca variants.
1497 (is_inexpensive_builtin): Likewise.
1498 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
1499 * calls.c (special_function_p): Deal with all alloca variants.
1500 (initialize_argument_information): Adjust call to
1501 allocate_dynamic_stack_space.
1502 (expand_call): Likewise.
1503 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
1504 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
1505 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
1506 use it for the stack usage computation.
1507 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
1508 * function.c (gimplify_parameters): Call build_alloca_call_expr.
1509 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
1510 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
1511 (in_loop_p): Remove first argument and useless check.
1512 (pass_walloca::execute): Remove useless test and adjust call to above.
1513 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
1514 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
1515 (gimplify_call_expr): Deal with all alloca variants.
1516 * hsa-gen.c (gen_hsa_alloca): Likewise.
1517 (gen_hsa_insns_for_call): Likewise.
1518 * ipa-pure-const.c (special_builtin_state): Likewise.
1519 * tree-chkp.c (chkp_build_returned_bound): Likewise.
1520 * tree-object-size.c (alloc_object_size): Likewise.
1521 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
1522 (call_may_clobber_ref_p_1): Likewise.
1523 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
1524 (ccp_fold_stmt): Likewise.
1525 (optimize_stack_restore): Likewise.
1526 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
1527 (mark_all_reaching_defs_necessary_1): Likewise.
1528 (propagate_necessity): Likewise.
1529 (eliminate_unnecessary_stmts): Likewise.
1530 * tree.c (build_common_builtin_nodes): Build
1531 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
1532 (build_alloca_call_expr): New function.
1533 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
1534 (CASE_BUILT_IN_ALLOCA): Likewise.
1535 (build_alloca_call_expr): Declare.
1536 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
1538 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
1541 * dwarf2out.c (new_die_raw): New static inline function.
1542 (new_die): Use it to create the DIE.
1543 (add_AT_external_die_ref): Likewise.
1544 (clone_die): Likewise.
1545 (clone_as_declaration): Likewise.
1546 (dwarf2out_vms_debug_main_pointer): Likewise.
1547 (base_type_die): Likewise. Remove early return for corner cases.
1548 Do not call add_pubtype on the DIE here.
1549 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
1550 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
1551 typedefs for base types with DW_AT_endianity. Make sure a DIE with
1552 native order exists for base types, attach the DIE manually and call
1553 add_pubtype on it. Do not equate a reverse order DIE to the type.
1555 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
1557 * config/arm/arm.c (align_ok_ldrd_strd): New function.
1558 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
1560 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
1562 2017-10-19 Jakub Jelinek <jakub@redhat.com>
1564 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
1565 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
1566 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
1567 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
1568 * opts.c (sanitizer_opts): Add builtin.
1569 * ubsan.c (instrument_builtin): New function.
1570 (pass_ubsan::execute): Call it.
1571 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
1572 * doc/invoke.texi: Document -fsanitize=builtin.
1574 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
1575 builtins, store max (log2 (align), 0) into uchar field instead of
1576 align into uptr field.
1577 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
1578 store uchar 0 field instead of uptr 0 field.
1579 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
1580 instead of passing one address of struct with 2 locations pass
1581 two addresses of structs with 1 location each.
1582 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
1583 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
1584 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
1585 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
1586 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
1587 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
1588 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
1589 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
1591 2017-10-19 Martin Liska <mliska@suse.cz>
1594 * file-find.c (remove_prefix): Remove.
1595 * file-find.h (remove_prefix): Likewise.
1596 * gcc-ar.c: Remove smartness of lookup.
1598 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
1600 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
1601 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
1602 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
1604 2017-10-19 Jakub Jelinek <jakub@redhat.com>
1607 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
1608 (setcc + and to xor + setcc): New peephole2.
1610 2017-10-19 Tom de Vries <tom@codesourcery.com>
1612 * doc/sourcebuild.texi (Test Directives, Variants of
1613 dg-require-support): Add dg-require-stack-size.
1615 2017-10-19 Martin Liska <mliska@suse.cz>
1618 * gimplify.c (gimplify_decl_expr): Do not instrument variables
1619 that have a large alignment.
1620 (gimplify_target_expr): Likewise.
1622 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
1624 PR rtl-optimization/82602
1625 * ira.c (rtx_moveable_p): Return false for volatile asm.
1627 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
1630 * config/i386/i386-modes.def (CCGZ): New CC mode.
1631 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
1632 * config/i386/predicates.md (ix86_comparison_operator):
1634 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
1635 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
1636 with double-word subtraction.
1637 (put_condition_code): Handle CCGZmode.
1639 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
1641 * wide-int.cc (debug (const wide_int &)): New.
1642 (debug (const wide_int *)): New.
1643 (debug (const widest_int &)): New.
1644 (debug (const widest_int *)): New.
1646 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
1649 * lra-constraints.c (curr_insn_transform): Use non-input operand
1650 instead of output one for matched reload.
1652 2017-10-18 Bin Cheng <bin.cheng@arm.com>
1654 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
1655 (tree-ssa-loop-ivopts.h): New header file.
1656 (struct builtin_info): New fields.
1657 (classify_builtin_1): Compute and record base and offset parts for
1658 memset builtin partition by calling strip_offset.
1659 (offset_cmp, fuse_memset_builtins): New functions.
1660 (finalize_partitions): Fuse adjacent memset partitions by calling
1662 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
1663 Expose the interface.
1664 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
1666 2017-10-18 Bin Cheng <bin.cheng@arm.com>
1668 PR tree-optimization/82574
1669 * tree-loop-distribution.c (find_single_drs): New parameter. Check
1670 that data reference must be executed exactly once per iteration
1671 against the outermost loop in nest.
1672 (classify_partition): Update call to above function.
1674 2017-10-18 Richard Biener <rguenther@suse.de>
1676 PR tree-optimization/82591
1677 * graphite.c (graphite_transform_loops): Move code gen message
1679 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
1680 Here. Handle scop_to_isl_ast failing.
1681 (scop_to_isl_ast): Limit the number of ISL operations.
1683 2017-10-18 Richard Biener <rguenther@suse.de>
1685 * graphite-isl-ast-to-gimple.c
1686 (translate_isl_ast_to_gimple::set_rename): Simplify.
1687 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
1688 (graphite_copy_stmts_from_block): ... here.
1689 (copy_bb_and_scalar_dependences): Simplify.
1690 (add_parameters_to_ivs_params): Canonicalize.
1691 (generate_entry_out_of_ssa_copies): Simplify.
1692 * graphite-sese-to-poly.c (extract_affine_name): Simplify
1693 by passing in ISL dimension.
1694 (parameter_index_in_region_1): Rename to ...
1695 (parameter_index_in_region): ... this.
1696 (extract_affine): Adjust assert, pass down parameter index.
1697 (add_param_constraints): Use range-info when available.
1698 (build_scop_context): Adjust.
1699 * sese.c (new_sese_info): Adjust.
1700 (free_sese_info): Likewise.
1701 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
1702 Remove unused typedefs.
1703 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
1705 2017-10-18 Martin Liska <mliska@suse.cz>
1707 * combine.c (simplify_compare_const): Add gcc_fallthrough.
1709 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
1711 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
1712 (s390_sched_init): Do not reset s390_sched_state if we entered the
1713 current basic block via a fallthru edge and all others are unlikely.
1715 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
1717 * config/s390/s390.c (NUM_SIDES): New variable.
1718 (LONGRUNNING_THRESHOLD): New variable.
1719 (LATENCY_FACTOR): New variable.
1720 (s390_sched_score): Decrease score for long-running instructions on
1722 (s390_sched_variable_issue): Perform bookkeeping for long-running
1725 2017-10-18 Richard Biener <rguenther@suse.de>
1727 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
1728 Simplify with removal of the parameter rename map.
1729 (set_rename): Likewise.
1730 (should_copy_to_new_region): Likewise.
1731 (graphite_copy_stmts_from_block): Likewise.
1732 (copy_bb_and_scalar_dependences): Remove initialization of
1733 unused copied_bb_map.
1735 (copy_internal_parameters): Likewise.
1736 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
1737 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
1738 Use INTEGRAL_TYPE_P.
1739 (parameter_index_in_region_1): Rename to ...
1740 (assign_parameter_index_in_region): ... this. Assert we have
1741 a parameter we handle.
1742 (scan_tree_for_params): Adjust.
1743 * sese.h (parameter_rename_map_t): Remove.
1744 (struct sese_info_t): Remove unused parameter_rename_map and
1745 copied_bb_map members.
1746 * sese.c (new_sese_info): Adjust.
1747 (free_sese_info): Likewise.
1749 2017-10-18 Martin Liska <mliska@suse.cz>
1752 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
1753 on an abnormal edge.
1755 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
1757 * doc/invoke.texi (ffunction-sections and fdata-sections):
1760 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
1762 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
1763 the use statement can throw internally.
1765 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
1767 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
1768 any RTX present on the RHS of a SET.
1769 * compare-elim.c (try_eliminate_compare): Restore comment.
1771 2017-10-17 Jakub Jelinek <jakub@redhat.com>
1773 * langhooks.h (struct lang_hooks): Document that tree_size langhook
1774 may be also called on tcc_type nodes.
1775 * langhooks.c (lhd_tree_size): Likewise.
1777 2017-10-17 David Malcolm <dmalcolm@redhat.com>
1779 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
1780 format_warning_at_substring.
1781 (maybe_warn): Convert source_range * param to a location_t. Pass
1782 UNKNOWN_LOCATION rather than NULL to fmtwarn.
1783 (format_directive): Remove code to extract source_ranges and
1784 source_range * in favor of just a location_t.
1785 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
1787 * substring-locations.c (format_warning_va): Convert
1788 source_range * param to a location_t.
1789 (format_warning_at_substring): Likewise.
1790 * substring-locations.h (format_warning_va): Likewise.
1791 (format_warning_at_substring): Likewise.
1793 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
1795 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
1797 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
1798 and scatter/gather ops.
1800 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
1801 vec_gather_load and vec_scatter_store.
1802 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
1803 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
1804 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
1805 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
1806 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
1807 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
1809 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
1811 * reg-stack.c (compare_for_stack_reg): Add bool argument.
1812 Detect FTST instruction and handle its register pops. Only pop
1813 second operand if can_pop_second_op is true.
1814 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
1815 set can_pop_second_op to false in the compare_for_stack_reg call.
1817 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
1818 output_fp_compare for stack register operands.
1819 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
1820 instructions here. Do not emit stack register pops here. Assert
1821 that FCOMPP pops next to top stack register. Rewrite function.
1823 2017-10-17 Nathan Sidwell <nathan@acm.org>
1826 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
1827 use DECL_ASSEMBLER_NAME_RAW.
1830 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
1833 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
1834 Wilco Dijkstra <wilco.dijkstra@arm.com>
1836 * builtins.c (expand_builtin_update_setjmp_buf): Add a
1837 converstion to Pmode from the buf_addr.
1839 2017-10-17 Richard Biener <rguenther@suse.de>
1841 * graphite-dependences.c (scop_get_reads_and_writes): Change
1842 output parameters to references.
1844 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
1846 PR 71026/tree-optimization
1847 * fold-const.c (distribute_real_division): Removed.
1848 (fold_binary_loc): Remove calls to distribute_real_divison.
1850 2017-10-17 Richard Biener <rguenther@suse.de>
1852 * graphite-scop-detection.c
1853 (scop_detection::stmt_has_simple_data_refs_p): Always use
1854 the full nest as region.
1855 (try_generate_gimple_bb): Likewise.
1856 * sese.c (scalar_evolution_in_region): Simplify now that
1857 SCEV can handle instantiation in regions.
1858 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
1859 in the non-loop part of a function if requested.
1861 2017-10-17 Richard Biener <rguenther@suse.de>
1863 PR tree-optimization/82563
1864 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
1866 (graphite_regenerate_ast_isl): Call it.
1867 * graphite-scop-detection.c (build_scops): Remove entry edge split.
1869 2017-10-17 Jakub Jelinek <jakub@redhat.com>
1871 PR tree-optimization/82549
1872 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
1873 Formatting fixes. Instead of calling make_bit_field_ref with negative
1876 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
1878 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
1879 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
1880 _mm_maskz_reduce_ss): New.
1881 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
1882 __builtin_ia32_reducess_mask): Ditto..
1883 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
1884 * config/i386/sse.md (reduces<mode>): Renamed to ...
1885 (reduces<mode><mask_scalar_name>): ... this.
1886 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
1888 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
1889 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
1891 2017-10-16 David Malcolm <dmalcolm@redhat.com>
1893 * Makefile.in (OBJS): Add unique-ptr-tests.o.
1894 * selftest-run-tests.c (selftest::run_tests): Call
1895 selftest::unique_ptr_tests_cc_tests.
1896 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
1897 * unique-ptr-tests.cc: New file.
1899 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
1902 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
1904 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
1905 (make_hard_regno_born, make_hard_regno_dead): Update
1906 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
1908 2017-10-16 Jeff Law <law@redhat.com>
1910 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
1912 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
1914 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
1916 2017-10-16 Olivier Hainque <hainque@adacore.com>
1918 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
1919 with_cpu if we were configured for an e500v2 target cpu name.
1921 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
1923 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
1924 * doc/invoke.texi: Document +nodsp as a valid extension for
1927 2017-10-16 Martin Liska <mliska@suse.cz>
1929 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
1930 (test_set_range): Likewise.
1931 (test_range_functions): Rename to ...
1932 (test_bit_in_range): ... this.
1933 (sbitmap_c_tests): Add new test.
1935 2017-10-16 Tamar Christina <tamar.christina@arm.com>
1937 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
1939 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
1940 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
1942 2017-10-16 Tamar Christina <tamar.christina@arm.com>
1944 * config/aarch64/aarch64-builtins.c
1945 (aarch64_types_quadopu_lane_qualifiers): New.
1946 (TYPES_QUADOPU_LANE): New.
1947 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
1948 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
1949 (aarch64_<sur>dot_laneq<vsi2qi>): New.
1950 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
1951 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
1952 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
1953 (Vdottype, DOTPROD): New.
1954 (sur): Add SDOT and UDOT.
1956 2017-10-16 Tamar Christina <tamar.christina@arm.com>
1958 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
1959 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
1960 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
1962 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
1963 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
1964 Enable TARGET_DOTPROD.
1965 (cortex-a75.cortex-a55): Likewise.
1966 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
1968 2017-10-16 Tamar Christina <tamar.christina@arm.com>
1970 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
1971 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
1973 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
1975 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
1976 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
1977 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
1978 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
1979 * config/arm/types.md (neon_dot, neon_dot_q): New.
1980 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
1982 2017-10-16 Tamar Christina <tamar.christina@arm.com>
1984 * config/arm/arm.h (TARGET_DOTPROD): New.
1985 * config/arm/arm.c (arm_arch_dotprod): New.
1986 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
1987 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
1988 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
1989 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
1990 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
1991 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
1992 * doc/invoke.texi (armv8.2-a): Document dotprod
1994 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
1996 * i386.c (ix86_vec_cost): New function.
1997 (ix86_rtx_costs): Handle vector operations better.
1998 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
1999 * x86-tune-costs.h: Add new costs to all tables.
2001 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
2003 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
2005 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
2006 divsd, sqrtss and sqrtsd
2007 * x86-tune-costs.h: Add new entries to all costs.
2008 (znver1_cost): Fix to match real instruction latencies.
2010 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2011 Michael Collison <michael.collison@arm.com>
2013 * compare-elim.c: Include emit-rtl.h.
2014 (can_merge_compare_into_arith): New function.
2015 (try_validate_parallel): Likewise.
2016 (try_merge_compare): Likewise.
2017 (try_eliminate_compare): Call the above when no previous clobber
2019 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
2022 2017-10-14 Jakub Jelinek <jakub@redhat.com>
2026 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
2027 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
2031 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
2032 to be any operand_equal_p operands. For & (B - 1) require
2033 B to be power of 2. Recognize
2034 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
2036 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
2039 * optabs.c (expand_memory_blockage): Fix call of
2040 targetm.have_memory_blockage.
2042 2017-10-14 Jakub Jelinek <jakub@redhat.com>
2045 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
2046 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
2047 objects to extra_objs instead of overwriting it.
2049 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
2051 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
2052 Use any_fp_register_operand as operand[3] predicate. Simplify
2053 equality test for operands[2] and operands[4] memory location.
2054 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
2055 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
2056 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
2057 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
2058 any_fp_register_operand as operand[1] predicate. Simplify
2059 equality test for operands[0] and operands[3] memory location.
2060 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
2061 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
2062 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
2064 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
2066 * target-insns.def: Add memory_blockage.
2067 * optabs.c (expand_memory_blockage): New function.
2068 (expand_asm_memory_barrier): Rename ...
2069 (expand_asm_memory_blockage): ... to this.
2070 (expand_mem_thread_fence): Call expand_memory_blockage
2071 instead of expand_asm_memory_barrier.
2072 (expand_mem_singnal_fence): Ditto.
2073 (expand_atomic_load): Ditto.
2074 (expand_atomic_store): Ditto.
2075 * doc/md.texi (Standard Pattern Names For Generation):
2076 Document memory_blockage instruction pattern.
2078 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
2080 * config/rl78/rl78.c (rl78_emit_libcall): New function.
2081 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
2082 * config/rl78/rl78.md: New define_expand "adddi3".
2084 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
2086 * cfghooks.c (verify_flow_info): Disable check that all probabilities
2089 2017-10-13 Jeff Law <law@redhat.com>
2091 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
2093 2017-10-13 Jakub Jelinek <jakub@redhat.com>
2096 * internal-fn.c (expand_mul_overflow): If both operands have
2097 the same highpart of -1 or 0 and the topmost bit of lowpart
2098 is different, overflow is if res <= 0 rather than res < 0.
2100 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
2102 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
2103 TARGET_P9_VECTOR code for unaligned_load case.
2105 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
2107 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
2109 2017-10-13 Nathan Sidwell <nathan@acm.org>
2111 * tree-core.h (tree_contains_struct): Make bool.
2112 * tree.c (tree_contains_struct): Likewise.
2113 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
2114 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
2115 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
2116 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
2117 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
2119 2017-10-13 Richard Biener <rguenther@suse.de>
2121 * graphite-isl-ast-to-gimple.c
2122 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
2123 parameters and dominance check.
2124 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
2125 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
2126 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
2127 Do not update SSA form here or do intermediate IL verification.
2128 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
2129 (graphite_initialize): Remove check on the number of loops in
2130 the function and inline into graphite_transform_loops.
2131 (graphite_finalize): Inline into graphite_transform_loops.
2132 (graphite_transform_loops): Perform SSA update and IL verification
2134 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
2136 2017-10-13 Richard Biener <rguenther@suse.de>
2138 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
2139 graphite_expression_type_precision): Avoid global constructor
2141 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
2142 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
2143 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
2144 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
2145 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
2146 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
2148 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
2151 * config/i386/i386.h (ix86_red_zone_size): New.
2152 * config/i386/i386.md (push peephole2s): Replace
2153 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
2155 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
2156 Alan Hayward <alan.hayward@arm.com>
2157 David Sherwood <david.sherwood@arm.com>
2159 * combine.c (can_change_dest_mode): Reject changes in
2160 REGMODE_NATURAL_SIZE.
2162 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
2163 Alan Hayward <alan.hayward@arm.com>
2164 David Sherwood <david.sherwood@arm.com>
2166 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
2167 (expand_debug_source_expr): Likewise.
2168 * combine.c (combine_simplify_rtx): Likewise.
2169 * cse.c (fold_rtx): Likewise.
2170 * fwprop.c (canonicalize_address): Likewise.
2171 * targhooks.c (default_shift_truncation_mask): Likewise.
2173 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
2174 Alan Hayward <alan.hayward@arm.com>
2175 David Sherwood <david.sherwood@arm.com>
2177 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
2178 (widened_mode): Likewise.
2179 (expand_unop): Likewise.
2180 * ree.c (transform_ifelse): Likewise.
2181 (merge_def_and_ext): Likewise.
2182 (combine_reaching_defs): Likewise.
2183 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2185 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
2186 Alan Hayward <alan.hayward@arm.com>
2187 David Sherwood <david.sherwood@arm.com>
2189 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
2190 * combine.c (gen_lowpart_for_combine): Likewise.
2191 * dwarf2out.c (rtl_for_decl_location): Likewise.
2192 * final.c (alter_subreg): Likewise.
2193 * rtlhooks.c (gen_lowpart_general): Likewise.
2194 (gen_lowpart_if_possible): Likewise.
2196 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
2197 Alan Hayward <alan.hayward@arm.com>
2198 David Sherwood <david.sherwood@arm.com>
2200 * calls.c (expand_call): Use subreg_lowpart_offset.
2201 * cse.c (cse_insn): Likewise.
2202 * regcprop.c (copy_value): Likewise.
2203 (copyprop_hardreg_forward_1): Likewise.
2205 2017-10-13 Jakub Jelinek <jakub@redhat.com>
2208 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
2209 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
2210 =Q constraints to +Q and into insn condition add check
2211 that operands[0] and operands[1] are equal.
2212 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
2213 =Q constraints to +Q and into insn condition add check
2214 that operands[0] is equal to either operands[1] or operands[2].
2217 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
2218 instead of handling MINUS_EXPR twice (once for each argument),
2219 canonicalize operand order and handle just once, use rtype where
2220 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
2223 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
2224 any values of __C while still being pattern recognizable as a simple
2227 2017-10-13 Richard Biener <rguenther@suse.de>
2229 PR tree-optimization/82451
2231 2017-10-02 Richard Biener <rguenther@suse.de>
2233 PR tree-optimization/82355
2234 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
2235 a mapping for the enclosing loop but avoid generating one for
2237 (copy_bb_and_scalar_dependences): Remove premature codegen
2238 error on PHIs in blocks duplicated into multiple places.
2239 * graphite-scop-detection.c
2240 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
2241 in the region use it as loop and nest to analyze the DR in.
2242 (try_generate_gimple_bb): Likewise.
2243 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
2244 (add_loop_constraints): For blocks in a loop not in the region
2245 create a dimension with a single iteration.
2246 * sese.h (gbb_loop_at_index): Remove assert.
2248 * cfgloop.c (loop_preheader_edge): For the loop tree root
2249 return the single successor of the entry block.
2250 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
2251 Reset the SCEV hashtable and niters.
2252 * graphite-scop-detection.c
2253 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
2254 assert that we only have POLYNOMIAL_CHREC that vary in loops
2255 contained in the region.
2256 (scop_detection::graphite_can_represent_expr): Adjust.
2257 (scop_detection::stmt_has_simple_data_refs_p): For loops
2258 not in the region set loop to NULL. The nest is now the
2259 entry edge to the region.
2260 (try_generate_gimple_bb): Likewise.
2261 * sese.c (scalar_evolution_in_region): Adjust for
2262 instantiate_scev change.
2263 * tree-data-ref.h (graphite_find_data_references_in_stmt):
2264 Make nest parameter the edge into the region.
2265 (create_data_ref): Likewise.
2266 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
2267 entry edge into a region and adjust instantiate_scev calls.
2268 (create_data_ref): Likewise.
2269 (graphite_find_data_references_in_stmt): Likewise.
2270 (find_data_references_in_stmt): Pass the loop preheader edge
2271 from the nest argument.
2272 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
2273 parameter the edge into the region.
2274 (instantiate_parameters): Use the loop preheader edge as entry.
2275 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
2277 (get_instantiated_value_entry): Make instantiate_below parameter
2278 the edge into the region.
2279 (instantiate_scev_name): Likewise. Adjust dominance checks,
2280 when we cannot use loop-based instantiation instantiate by
2281 walking use-def chains.
2282 (instantiate_scev_poly): Adjust.
2283 (instantiate_scev_binary): Likewise.
2284 (instantiate_scev_convert): Likewise.
2285 (instantiate_scev_not): Likewise.
2286 (instantiate_array_ref): Remove.
2287 (instantiate_scev_3): Likewise.
2288 (instantiate_scev_2): Likewise.
2289 (instantiate_scev_1): Likewise.
2290 (instantiate_scev_r): Do not blindly handle N-operand trees.
2291 Do not instantiate array-refs. Handle all constants and invariants.
2292 (instantiate_scev): Make instantiate_below parameter
2293 the edge into the region.
2294 (resolve_mixers): Use the loop preheader edge for the region
2295 parameter to instantiate_scev_r.
2296 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
2298 2017-10-13 Richard Biener <rguenther@suse.de>
2300 PR tree-optimization/82525
2301 * graphite-isl-ast-to-gimple.c
2302 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
2304 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
2305 Fail code generation when we cannot represent the isl integer.
2306 (binary_op_to_tree): Elide modulo operations that are no-ops
2307 in the type we code generate. Remove now superfluous code
2310 2017-10-13 Richard Biener <rguenther@suse.de>
2312 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
2313 (scop_detection::harmful_loop_in_region): Remove premature
2314 IV type restriction.
2315 (scop_detection::graphite_can_represent_scev): We can handle
2316 pointer IVs just fine.
2318 2017-10-13 Alan Modra <amodra@gmail.com>
2320 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
2321 "Clobbers and Scratch Registers". Add paragraph on
2322 alternative to clobbers for scratch registers and OpenBLAS
2325 2017-10-13 Alan Modra <amodra@gmail.com>
2327 * doc/extend.texi (Clobbers): Correct vax example. Delete old
2328 example of a memory input for a string of known length. Move
2329 commentary out of table. Add a number of new examples
2330 covering array memory inputs.
2332 2017-10-12 Martin Liska <mliska@suse.cz>
2334 PR tree-optimization/82493
2335 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
2336 (test_range_functions): New function.
2337 (sbitmap_c_tests): Likewise.
2338 * selftest-run-tests.c (selftest::run_tests): Run new tests.
2339 * selftest.h (sbitmap_c_tests): New function.
2341 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
2343 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
2345 * config/rs6000/amo.h: Fix spacing issue.
2347 2017-10-12 Jakub Jelinek <jakub@redhat.com>
2350 * config/i386/i386.md (*ashl<mode>3_mask_1,
2351 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
2352 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
2355 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
2357 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
2358 (profile_probability): Set max_probability
2359 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
2361 * profile-count.c (profile_probability::differs_from_p): Do not
2362 rely on max_probaiblity == 10000
2364 2017-10-12 Jeff Law <law@redhat.com>
2366 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
2369 2017-10-12 Martin Sebor <msebor@redhat.com>
2373 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
2374 (handle_alias_pairs): Call it.
2375 * common.opt (-Wattribute-alias): New option.
2376 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
2377 * doc/invoke.texi (-Wattribute-alias): Document.
2379 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
2382 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
2384 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
2386 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
2387 (make_hard_regno_born, make_hard_regno_dead): Update
2388 bb_killed_pseudos and bb_gen_pseudos.
2390 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
2392 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
2394 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
2396 * config/alpha/alpha.c (alpha_split_conditional_move):
2397 Use std::swap instead of manually swapping.
2398 (alpha_stdarg_optimize_hook): Ditto.
2399 (alpha_canonicalize_comparison): Ditto.
2401 2017-10-12 Bin Cheng <bin.cheng@arm.com>
2403 * tree-loop-distribution.c (struct builtin_info): New struct.
2404 (struct partition): Refactor fields into struct builtin_info.
2405 (partition_free): Free struct builtin_info.
2406 (build_size_arg_loc, build_addr_arg_loc): Delete.
2407 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
2408 information from struct builtin_info.
2409 (find_single_drs): New function refactored from classify_partition.
2410 Also moved builtin validity checks to this function.
2411 (compute_access_range, alloc_builtin): New functions.
2412 (classify_builtin_st, classify_builtin_ldst): New functions.
2413 (classify_partition): Refactor code into functions find_single_drs,
2414 classify_builtin_st and classify_builtin_ldst.
2415 (distribute_loop): Don't do runtime alias check when distributing
2417 (find_seed_stmts_for_distribution): New function.
2418 (pass_loop_distribution::execute): Refactor code finding seed
2419 stmts into above function. Support distribution for the innermost
2420 two-level loop nest. Adjust dump information.
2422 2017-10-12 Bin Cheng <bin.cheng@arm.com>
2424 * tree-loop-distribution.c: Adjust the general comment.
2425 (NUM_PARTITION_THRESHOLD): New macro.
2426 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
2427 (classify_partition): Skip builtin pattern of loop nest's inner loop.
2428 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
2429 in call to build_partition_graph.
2430 (finalize_partitions): New parameter. Make loop distribution more
2431 conservative by fusing more partitions.
2432 (distribute_loop): Don't do runtime alias check in case of loop nest
2434 (find_seed_stmts_for_distribution): New function.
2435 (prepare_perfect_loop_nest): New function.
2436 (pass_loop_distribution::execute): Refactor code finding seed stmts
2437 and loop nest into above functions. Support loop nest distribution.
2438 Adjust dump information accordingly.
2440 2017-10-12 Bin Cheng <bin.cheng@arm.com>
2442 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
2443 and set PTYPE_SEQUENTIAL for merged partition.
2445 2017-10-12 Richard Biener <rguenther@suse.de>
2447 PR tree-optimization/69728
2449 2017-09-19 Richard Biener <rguenther@suse.de>
2451 PR tree-optimization/69728
2452 * graphite-sese-to-poly.c (schedule_error): New global.
2453 (add_loop_schedule): Handle empty domain by failing the
2455 (build_original_schedule): Handle schedule_error.
2457 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
2458 domain by returning an unchanged schedule.
2460 2017-10-12 Jakub Jelinek <jakub@redhat.com>
2462 * genrecog.c (validate_pattern): For VEC_SELECT verify that
2463 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
2465 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
2467 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
2470 2017-10-12 Jakub Jelinek <jakub@redhat.com>
2473 * expr.c (store_field): Don't optimize away bitsize == 0 store
2474 from CALL_EXPR with addressable return type.
2476 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
2478 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
2479 * config/rs6000/rs6000.md (sel): Delete mode attribute.
2480 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
2481 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
2482 TARGET_ISEL instead of TARGET_ISEL<sel>.
2484 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
2486 * config/rs6000/rs6000.c
2487 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
2489 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
2491 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
2493 (reg_or_cint_operand): Fix comment.
2494 (reg_or_zero_operand): New predicate.
2495 * config/rs6000/rs6000-protos.h (output_isel): Delete.
2496 * config/rs6000/rs6000.c (output_isel): Delete.
2497 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
2498 instead of reg_or_cint_operand. Output instruction directly (not via
2500 (isel_unsigned_<mode>): Ditto.
2501 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
2502 gpc_reg_operand. Add an instruction alternative for this. Output
2503 instruction directly.
2504 (*isel_reversed_unsigned_<mode>): Ditto.
2506 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
2508 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
2509 (TARGET_CANONICALIZE_COMPARISON): Define.
2511 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
2514 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
2515 Check whether the dest is REG before adding REG_EQUIV note.
2517 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
2520 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
2522 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
2523 (make_hard_regno_born, make_hard_regno_dead): Update
2524 bb_killed_pseudos and bb_gen_pseudos.
2526 2017-10-11 Nathan Sidwell <nathan@acm.org>
2528 * incpath.h (enum incpath_kind): Name enum, prefix values.
2529 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
2530 * incpath.c (heads, tails): Use INC_MAX.
2531 (add_env_var_paths, add_standard_paths): Use incpath_kind.
2532 (merge_include_chains, split_quote_chain,
2533 register_include_chains): Update incpath_kind names.
2534 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
2535 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
2537 (add_framework_path, darwin_register_objc_includes): Likewise.
2538 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
2540 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
2542 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
2543 Do not use float_operator operator predicate.
2544 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
2545 * config/i386/predicates.md (float_operator): Remove predicate.
2547 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
2549 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
2550 (*jccxf_i387): Ditto.
2551 (*jcc<mode>_i387): Ditto.
2552 (*jccu<mode>_i387): Ditto.
2553 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
2554 (*jcc_*_i387 splitters): Remove.
2555 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
2556 * config/i386/i386.c (ix86_split_fp_branch): Remove.
2557 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
2560 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
2562 * profile-count.h (slow_safe_scale_64bit): New function.
2563 (safe_scale_64bit): New inline.
2564 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
2565 * profile-count.c: Include wide-int.h
2566 (slow_safe_scale_64bit): New.
2568 2017-10-11 Nathan Sidwell <nathan@acm.org>
2570 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
2571 HAS_DECL_ASSEMBLER_NAME_P.
2572 * gimple-expr.c (gimple_decl_printable_name: Check
2573 HAS_DECL_ASSEMBLER_NAME_P too.
2574 * ipa-utils.h (type_in_anonymous_namespace_p): Check
2575 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
2576 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
2577 * passes.c (rest_of_decl_compilation): Check
2578 HAS_DECL_ASSEMBLER_NAME_P too.
2579 * recog.c (verify_changes): Likewise.
2580 * tree-pretty-print.c (dump_decl_name): Likewise.
2581 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
2583 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
2584 (SET_DECL_ASSEMBLER_NAME): Use it.
2585 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
2586 (COPY_DECL_ASSEMBLER_NAME): Likewise.
2587 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
2589 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
2591 * config.gcc (i386, x86_64): Add extra objects.
2592 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
2593 (ix86_min_insn_size): Declare.
2594 (ix86_issue_rate): Declare.
2595 (ix86_adjust_cost): Declare.
2596 (ia32_multipass_dfa_lookahead): Declare.
2597 (ix86_macro_fusion_p): Declare.
2598 (ix86_macro_fusion_pair_p): Declare.
2599 (ix86_bd_has_dispatch): Declare.
2600 (ix86_bd_do_dispatch): Declare.
2601 (ix86_core2i7_init_hooks): Declare.
2602 (ix86_atom_sched_reorder): Declare.
2603 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
2604 (COSTS_N_BYTES): Move to x86-tune-costs.h.
2605 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
2606 (rip_relative_addr_p): Rename to ...
2607 (ix86_rip_relative_addr_p): ... this one; export.
2608 (memory_address_length): Update.
2609 (ix86_issue_rate): Move to x86-tune-sched.c.
2610 (ix86_flags_dependent): Move to x86-tune-sched.c.
2611 (ix86_agi_dependent): Move to x86-tune-sched.c.
2612 (exact_dependency_1): Move to x86-tune-sched.c.
2613 (exact_store_load_dependency): Move to x86-tune-sched.c.
2614 (ix86_adjust_cost): Move to x86-tune-sched.c.
2615 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
2616 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
2617 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
2618 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
2619 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
2620 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
2621 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
2622 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
2623 (min_insn_size): Rename to ...
2624 (ix86_min_insn_size): ... this one; export.
2625 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
2626 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
2627 (core2i7_first_cycle_multipass_backtrack): Move to
2628 x86-tune-sched-core.c.
2629 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
2630 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
2631 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
2632 (ix86_avoid_jump_mispredicts): Update.
2633 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
2634 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
2635 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
2636 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
2637 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
2638 (MAX_INSN): Move to ix86-tune-sched-bd.c.
2639 (MAX_IMM): Move to ix86-tune-sched-bd.c.
2640 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
2641 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
2642 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
2643 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
2644 (MAX_STORE): Move to ix86-tune-sched-bd.c.
2645 (BIG): Move to ix86-tune-sched-bd.c.
2646 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
2647 (enum insn_path): Move to ix86-tune-sched-bd.c.
2648 (get_mem_group): Move to ix86-tune-sched-bd.c.
2649 (is_cmp): Move to ix86-tune-sched-bd.c.
2650 (dispatch_violation): Move to ix86-tune-sched-bd.c.
2651 (is_branch): Move to ix86-tune-sched-bd.c.
2652 (is_prefetch): Move to ix86-tune-sched-bd.c.
2653 (init_window): Move to ix86-tune-sched-bd.c.
2654 (allocate_window): Move to ix86-tune-sched-bd.c.
2655 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
2656 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
2657 (process_end_window): Move to ix86-tune-sched-bd.c.
2658 (allocate_next_window): Move to ix86-tune-sched-bd.c.
2659 (find_constant): Move to ix86-tune-sched-bd.c.
2660 (get_num_immediates): Move to ix86-tune-sched-bd.c.
2661 (has_immediate): Move to ix86-tune-sched-bd.c.
2662 (get_insn_path): Move to ix86-tune-sched-bd.c.
2663 (get_insn_group): Move to ix86-tune-sched-bd.c.
2664 (count_num_restricted): Move to ix86-tune-sched-bd.c.
2665 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
2666 (add_insn_window): Move to ix86-tune-sched-bd.c.
2667 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
2668 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
2669 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
2670 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
2671 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
2672 (do_dispatch): Move to ix86-tune-sched-bd.c.
2673 (has_dispatch): Move to ix86-tune-sched-bd.c.
2674 * config/i386/t-i386: Add new object files.
2675 * config/i386/x86-tune-costs.h: New file.
2676 * config/i386/x86-tune-sched-atom.c: New file.
2677 * config/i386/x86-tune-sched-bd.c: New file.
2678 * config/i386/x86-tune-sched-core.c: New file.
2679 * config/i386/x86-tune-sched.c: New file.
2681 2017-10-11 Liu Hao <lh_mouse@126.com>
2683 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
2684 the generic version below instead.
2685 (should_colorize): Recognize Windows consoles as terminals
2687 * pretty-print.c [__MINGW32__] (write_all): New function.
2688 [__MINGW32__] (find_esc_head): Likewise.
2689 [__MINGW32__] (find_esc_terminator): Likewise.
2690 [__MINGW32__] (eat_esc_sequence): Likewise.
2691 [__MINGW32__] (mingw_ansi_fputs): New function that handles
2693 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
2696 2017-10-11 Richard Biener <rguenther@suse.de>
2698 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
2699 Properly call analyze_scalar_evolution with the loop of the stmt.
2701 2017-10-11 Richard Biener <rguenther@suse.de>
2703 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
2704 * tree-core.h (tree_base): Add chrec_var union member.
2705 * tree.h (CHREC_VAR): Remove.
2706 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
2707 * tree-chrec.h (build_polynomial_chrec): Adjust.
2708 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
2709 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
2711 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
2713 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
2714 * match.pd: ... here.
2715 ((T) X == (T) Y): Relax condition.
2717 2017-10-11 Bin Cheng <bin.cheng@arm.com>
2719 PR tree-optimization/82472
2720 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
2722 (break_alias_scc_partitions): Update postorder number.
2724 2017-10-11 Martin Liska <mliska@suse.cz>
2727 * opts.c (parse_no_sanitize_attribute): Do not use error_value
2729 * opts.h (parse_no_sanitize_attribute): Remove last argument.
2731 2017-10-11 Martin Liska <mliska@suse.cz>
2733 * print-rtl.c (print_insn): Move declaration of idbuf
2734 to same scope as name.
2736 2017-10-11 Martin Liska <mliska@suse.cz>
2741 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
2744 2017-10-11 Martin Liska <mliska@suse.cz>
2747 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
2750 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
2752 * config.gcc (default_gnu_indirect_function): Default to yes for
2753 arm*-*-linux* with glibc.
2755 2017-10-11 Richard Biener <rguenther@suse.de>
2757 * tree-scalar-evolution.c (get_scalar_evolution): Handle
2758 default-defs and types we do not want to analyze.
2759 (interpret_loop_phi): Replace unreachable code with an assert.
2760 (compute_scalar_evolution_in_loop): Remove and inline ...
2761 (analyze_scalar_evolution_1): ... here, replacing condition with
2762 what makes the intent clearer. Remove handling of cases
2763 get_scalar_evolution now handles.
2765 2017-10-10 Jim Wilson <wilson@tuliptree.org>
2767 PR rtl-optimization/81434
2768 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
2769 comment for main loop. In sched_group_found if, also add checks for
2770 pass and min_cost_group.
2772 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
2774 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
2775 (rs6000_insn_cost): New function.
2776 * config/rs6000/rs6000.md (cost): New attribute.
2778 2017-10-10 Jakub Jelinek <jakub@redhat.com>
2779 H.J. Lu <hongjiu.lu@intel.com>
2783 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
2784 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
2785 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
2786 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
2787 to OPTION_MASK_ISA_AVX512VL - builtins that have both
2788 OPTION_MASK_ISA_MMX and some other bit set require both
2789 mmx and the ISAs without the mmx bit.
2790 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
2791 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
2792 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
2793 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
2794 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
2795 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
2796 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
2797 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
2798 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
2799 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
2800 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
2801 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
2802 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
2803 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
2804 Add OPTION_MASK_ISA_MMX.
2806 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
2808 * config.gcc (armv7*-*-freebsd*): New target.
2809 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
2811 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
2813 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
2814 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
2817 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
2819 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
2820 a template parameter.
2821 (WIDE_INT_REF_FOR): Update accordingly.
2822 * tree.h (wi::int_traits <const_tree>): Delete.
2823 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
2824 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
2825 (wi::tree_to_wide_ref): New typedef.
2826 (wi::to_wide): New function.
2827 * calls.c (get_size_range): Use wi::to_wide when operating on
2829 * cgraph.c (cgraph_node::create_thunk): Likewise.
2830 * config/i386/i386.c (ix86_data_alignment): Likewise.
2831 (ix86_local_alignment): Likewise.
2832 * dbxout.c (stabstr_O): Likewise.
2833 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
2834 * expr.c (const_vector_from_tree): Likewise.
2835 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
2836 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
2837 (fold_negate_expr_1, int_const_binop_1, const_binop)
2838 (fold_convert_const_int_from_real, optimize_bit_field_compare)
2839 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
2840 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
2841 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
2842 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
2843 (fold_not_const, round_up_loc): Likewise.
2844 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
2845 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
2846 (alloca_call_type): Likewise.
2847 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
2848 * godump.c (go_output_typedef): Likewise.
2849 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
2850 * internal-fn.c (get_min_precision): Likewise.
2851 * ipa-cp.c (ipcp_store_vr_results): Likewise.
2852 * ipa-polymorphic-call.c
2853 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
2854 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
2855 (ipa_modify_call_arguments): Likewise.
2856 * match.pd: Likewise.
2857 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
2858 * print-tree.c (print_node_brief, print_node): Likewise.
2859 * stmt.c (expand_case): Likewise.
2860 * stor-layout.c (layout_type): Likewise.
2861 * tree-affine.c (tree_to_aff_combination): Likewise.
2862 * tree-cfg.c (group_case_labels_stmt): Likewise.
2863 * tree-data-ref.c (dr_analyze_indices): Likewise.
2864 (prune_runtime_alias_test_list): Likewise.
2865 * tree-dump.c (dequeue_and_dump): Likewise.
2866 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
2867 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
2868 * tree-pretty-print.c (dump_generic_node): Likewise.
2869 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
2870 (simple_iv_with_niters): Likewise.
2871 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
2872 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
2873 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
2874 * tree-ssa-loop-niter.c (split_to_var_and_offset)
2875 (refine_value_range_using_guard, number_of_iterations_ne_max)
2876 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
2877 (get_cst_init_from_scev, record_nonwrapping_iv)
2878 (scev_var_range_cant_overflow): Likewise.
2879 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
2880 * tree-ssa-pre.c (compute_avail): Likewise.
2881 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
2882 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
2883 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
2884 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
2885 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
2886 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
2887 (dump_case_nodes, try_switch_expansion): Likewise.
2888 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
2889 (vect_do_peeling): Likewise.
2890 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
2891 * tree-vect-stmts.c (vectorizable_load): Likewise.
2892 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
2893 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
2894 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
2895 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
2896 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
2897 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
2898 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
2899 (evrp_dom_walker::before_dom_children): Likewise.
2900 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
2901 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
2902 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
2903 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
2904 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
2905 (get_range_pos_neg): Likewise.
2906 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
2907 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
2908 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
2909 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
2910 * config/avr/avr.c (avr_fold_builtin): Likewise.
2911 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
2912 * config/msp430/msp430.c (msp430_attr): Likewise.
2913 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
2914 * config/powerpcspe/powerpcspe-c.c
2915 (altivec_resolve_overloaded_builtin): Likewise.
2916 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
2917 (rs6000_expand_ternop_builtin): Likewise.
2918 * config/rs6000/rs6000-c.c
2919 (altivec_resolve_overloaded_builtin): Likewise.
2920 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
2921 (rs6000_expand_ternop_builtin): Likewise.
2922 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
2924 2017-10-10 Bin Cheng <bin.cheng@arm.com>
2926 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
2927 when copying loop nest with only one inner loop.
2929 2017-10-10 Richard Biener <rguenther@suse.de>
2931 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
2932 blocks if SCEV is active.
2933 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
2935 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
2936 (scev_initialize): Assert we are not yet initialized.
2938 2017-10-10 Bin Cheng <bin.cheng@arm.com>
2940 * tree-loop-distribution.c (generate_loops_for_partition): Remove
2941 inner loop's exit stmt by making it always exit the loop, otherwise
2942 we would generate an infinite empty loop.
2944 2017-10-10 Bin Cheng <bin.cheng@arm.com>
2946 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
2947 renaming variables in new preheader if it's deleted.
2949 2017-10-10 Bin Cheng <bin.cheng@arm.com>
2951 * tree-loop-distribution.c (struct partition): Remove unused field
2952 loops of the structure.
2953 (partition_alloc, partition_free): Ditto.
2954 (build_rdg_partition_for_vertex): Ditto.
2956 2017-10-09 Jeff Law <law@redhat.com>
2958 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
2959 return type to match prototype and documentation.
2961 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
2963 * config/rs6000/rs6000.c (processor_costs): Move to ...
2964 * config/rs6000/rs6000.h: ... here.
2965 (rs6000_cost): Declare.
2967 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
2969 * except.c (setjmp_fn): New global variable.
2970 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
2971 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
2972 if DONT_USE_BUILTIN_SETJMP is defined.
2974 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
2976 * target.def (insn_cost): New hook.
2977 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
2978 * doc/tm.texi: Regenerate.
2979 * rtlanal.c (insn_cost): Use the new hook.
2981 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
2983 * combine.c (combine_validate_cost): Compute the new insn_cost,
2984 not just pattern_cost.
2985 (try_combine): Adjust comment.
2987 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
2989 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
2991 * combine.c (uid_insn_cost): Adjust comment.
2992 (combine_validate_cost): Adjust comment. Use pattern_cost instead
2994 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
2995 * dse.c (find_shift_sequence): Ditto.
2996 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
2997 (bb_valid_for_noce_process_p): Use pattern_cost.
2998 * rtl.h (insn_rtx_cost): Delete.
2999 (pattern_cost): New prototype.
3000 (insn_cost): New prototype.
3001 * rtlanal.c (insn_rtx_cost): Rename to...
3002 (pattern_cost): ... this.
3005 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
3007 * config/i386/i386.md (*jcc_2): Remove insn pattern.
3008 (*jcc<mode>_0_r_i387): Ditto.
3009 (*jccxf_r_i387): Ditto.
3010 (*jcc<mode>_r_i387): Ditto.
3011 (*jccu<mode>_r_i387): Ditto.
3012 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
3013 (*jcc): Rename from *jcc_1.
3015 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3017 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
3018 deferred rescans after the lvx/stvx recombination pre-pass.
3020 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
3022 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
3023 memory operation instruction support.
3024 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
3025 (rs6000-ibm-aix[789]*): Likewise.
3026 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
3027 Document new functions.
3029 2017-10-09 Richard Biener <rguenther@suse.de>
3031 PR tree-optimization/82397
3032 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
3033 equality only for semantically equal trees.
3035 2017-10-09 Richard Biener <rguenther@suse.de>
3037 PR tree-optimization/82449
3038 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
3039 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
3040 allow constant addresses.
3041 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
3044 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3046 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
3049 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3052 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
3055 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3058 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
3060 2017-10-09 Jakub Jelinek <jakub@redhat.com>
3063 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
3064 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
3066 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
3068 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
3069 (WI_BINARY_PREDICATE_RESULT): Likewise.
3070 (wi::binary_traits::operator_result): New type.
3071 (wi::binary_traits::predicate_result): Likewise.
3072 (generic_wide_int::operator~, unary generic_wide_int::operator-)
3073 (generic_wide_int::operator==, generic_wide_int::operator!=)
3074 (generic_wide_int::operator&, generic_wide_int::and_not)
3075 (generic_wide_int::operator|, generic_wide_int::or_not)
3076 (generic_wide_int::operator^, generic_wide_int::operator+
3077 (binary generic_wide_int::operator-, generic_wide_int::operator*):
3079 (operator~, unary operator-, operator==, operator!=, operator&)
3080 (operator|, operator^, operator+, binary operator-, operator*): New
3082 * expr.c (get_inner_reference): Use wi::bit_and_not.
3083 * fold-const.c (fold_binary_loc): Likewise.
3084 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
3085 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
3086 (bit_value_binop): Likewise.
3087 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
3088 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
3089 (extract_range_from_binary_expr_1): Likewise.
3090 (masked_increment): Likewise.
3091 (simplify_bit_ops_using_ranges): Likewise.
3093 2017-10-09 Martin Jambor <mjambor@suse.cz>
3096 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
3097 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
3098 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
3100 (hsa_fixup_mov_insn_type): New function.
3101 (hsa_op_with_type::get_in_type): Use it.
3102 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
3103 immediates in an assert.
3104 (hsa_op_with_type::extend_int_to_32bit): New method.
3105 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
3106 types. Convert to dest type if necessary.
3107 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
3108 (reg_for_gimple_ssa): Pass false as min32int to
3109 hsa_type_for_scalar_tree_type.
3110 (gen_hsa_addr): Fixup type when creating addresable temporary.
3111 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
3112 (gen_hsa_unary_operation): Extend operands and convert to dest type if
3113 necessary. Call hsa_fixup_mov_insn_type.
3114 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
3115 extend operands and convert to dest type if necessary.
3116 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
3117 to dest type if necessary.
3118 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
3119 if conversion nt necessary and size matches.
3120 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
3121 to dest type if necessary.
3122 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
3123 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
3125 (gen_hsa_clrsb): Likewise.
3126 (gen_hsa_ffs): Likewise.
3127 (gen_hsa_divmod): Extend operands and convert to dest type if
3129 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
3131 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
3133 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
3134 Remove empty default arguments. Use a brace block as output
3136 (conditional return): Ditto.
3138 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
3139 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
3141 (group_ending_nop): Ditto.
3142 (doloop_end): Ditto.
3143 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
3144 (splitters for those): Ditto.
3146 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
3148 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
3149 a conditional jump (and the compare for it) so that pc_rtx is the
3151 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
3152 for the deleted and renamed ctr<mode>_internal[234] patterns.
3153 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
3154 Delete second conditional return pattern.
3155 (ctr<mode>_internal2): Delete this second bdnz pattern.
3156 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
3157 (ctr<mode>_internal4): Delete this second bdz pattern.
3159 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
3161 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
3162 (always_initialized_rtx_for_ssa_name_p): New predicate.
3163 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
3164 (finish_out_of_ssa): Free new field of SA.
3165 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
3166 * tree-ssa-coalesce.c: Include tree-ssa.h.
3167 (get_parm_default_def_partitions): Remove extern keyword.
3168 (get_undefined_value_partitions): New function.
3169 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
3170 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
3172 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
3174 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
3176 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
3178 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
3180 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
3181 for targets that preffer 128bit.
3183 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
3185 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
3187 2017-10-08 Olivier Hainque <hainque@adacore.com>
3189 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
3190 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
3192 (thumb_set_return_address): Likewise.
3194 2017-10-08 Olivier Hainque <hainque@adacore.com>
3196 * common/config/arm/arm-common.c (arm_except_unwind_info):
3197 Handle DWARF2_UNWIND_INFO.
3199 2017-10-07 Michael Collison <michael.collison@arm.com>
3201 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
3204 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
3206 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
3207 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
3208 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
3209 defined, force the creation of a new block for a dispatch label.
3211 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
3213 * invoke.texi (Wsuggest-attribute=cold): Document.
3214 * common.opt (Wsuggest-attribute=cold): New
3215 * ipa-pure-const.c (warn_function_cold): New function.
3216 * predict.c (compute_function_frequency): Use it.
3217 * predict.h (warn_function_cold): Declare.
3219 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
3221 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
3224 2017-10-06 Martin Liska <mliska@suse.cz>
3226 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
3227 keyword for member functions.
3228 (struct sanopt_tree_couple): New struct.
3229 (struct sanopt_tree_couple_hash): New function.
3230 (struct sanopt_ctx): Add new hash_map.
3231 (has_dominating_ubsan_ptr_check): New function.
3232 (record_ubsan_ptr_check_stmt): Likewise.
3233 (maybe_optimize_ubsan_ptr_ifn): Likewise.
3234 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
3235 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
3237 2017-10-06 Sudakshina Das <sudi.das@arm.com>
3240 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
3241 aarch64_simd_valid_immediate on CONST_VECTORs.
3242 (aarch64_reg_or_bic_imm): Likewise.
3244 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
3246 PR rtl-optimization/82396
3247 * haifa-sched.c (ready_sort_real): Disable qsort checking.
3249 2017-10-06 Sebastian Pop <sebpop@gmail.com>
3251 * graphite-dependences.c (scop_get_reads): Move code to...
3252 (scop_get_must_writes): Move code to...
3253 (scop_get_may_writes): Move code to...
3254 (scop_get_reads_and_writes): ... here.
3255 (scop_get_dependences): Call scop_get_reads_and_writes.
3257 2017-10-06 Jakub Jelinek <jakub@redhat.com>
3259 PR tree-optimization/82434
3260 * fold-const.h (can_native_encode_type_p,
3261 can_native_encode_string_p): Remove.
3262 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
3263 don't encode anything, just return what would be otherwise returned.
3264 (native_encode_fixed, native_encode_complex, native_encode_vector):
3266 (native_encode_string): Likewise. Inline by hand
3267 can_native_encode_string_p.
3268 (can_native_encode_type_p): Remove.
3269 (can_native_encode_string_p): Remove.
3270 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
3271 STRING_CSTs using can_native_encode_string_p, test all
3272 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
3273 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
3274 argument from native_encode_expr.
3275 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
3276 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
3279 2017-10-06 Richard Biener <rguenther@suse.de>
3281 PR tree-optimization/82397
3282 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
3283 operand_equal_p but rely on data_ref_compare_tree for detecting
3285 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
3286 to match up with dr_group_sort_cmp.
3288 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3291 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
3293 * config/s390/s390-builtin-types.def: Regenerate.
3295 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3298 * config/s390/s390-builtin-types.def: Regenerate.
3299 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
3300 Change flag from B_VXE to B_VX.
3301 (s390_vec_min_dbl): Remove B_VXE flag.
3303 2017-10-06 Richard Biener <rguenther@suse.de>
3305 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
3306 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
3307 translate_isl_ast_to_gimple::is_valid_rename,
3308 translate_isl_ast_to_gimple::get_rename,
3309 translate_isl_ast_to_gimple::get_def_bb_for_const,
3310 translate_isl_ast_to_gimple::get_new_name,
3311 translate_isl_ast_to_gimple::collect_all_ssa_names,
3312 translate_isl_ast_to_gimple::copy_loop_phi_args,
3313 translate_isl_ast_to_gimple::collect_all_ssa_names,
3314 translate_isl_ast_to_gimple::copy_loop_phi_args,
3315 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
3316 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
3317 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
3318 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
3319 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
3320 translate_isl_ast_to_gimple::copy_cond_phi_args,
3321 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
3322 translate_isl_ast_to_gimple::edge_for_new_close_phis,
3323 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
3324 translate_isl_ast_to_gimple::rename_uses,
3325 translate_isl_ast_to_gimple::rename_all_uses): Remove.
3326 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
3327 (set_rename_for_each_def): Likewise.
3328 (graphite_copy_stmts_from_block): Handle debug stmt resetting
3329 here. Handle rewriting SCEV analyzable uses here.
3330 (copy_bb_and_scalar_dependences): Generate code for PHI
3332 (graphite_regenerate_ast_isl): Adjust.
3333 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
3334 (add_write, add_read): New functions.
3335 (build_cross_bb_scalars_def): Use it and simplify.
3336 (build_cross_bb_scalars_use): Likewise.
3337 (graphite_find_cross_bb_scalar_vars): Inline into...
3338 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
3339 simulating out-of-SSA. Compute liveout and add dependencies.
3340 (build_scops): Force an empty entry block.
3341 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
3343 (sese_build_liveouts): Declare.
3344 (sese_trivially_empty_bb_p): Likewise.
3345 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
3346 compute liveout and debug_liveout.
3347 (sese_bad_liveouts_use): Remove.
3348 (sese_reset_debug_liveouts_bb): Likewise.
3349 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
3350 (sese_build_liveouts): Build liveout and debug_liveout and store
3352 (new_sese_info): Adjust.
3353 (free_sese_info): Likewise.
3354 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
3355 do not build liveout here.
3356 (move_sese_in_condition): Adjust region entry.
3357 (scev_analyzable_p): Match up with chrec_apply requirements.
3358 (sese_trivially_empty_bb_p): New.
3359 * tree-into-ssa.c (get_reaching_def): Properly support generating
3360 default-defs for incremental rewrite of anonymous names.
3362 2017-10-06 Richard Biener <rguenther@suse.de>
3364 * graphite-sese-to-poly.c (extract_affine): For casts increasing
3365 precision do not perform modulo reduction.
3367 2017-10-06 Richard Biener <rguenther@suse.de>
3369 PR tree-optimization/82436
3370 * tree-vect-slp.c (vect_supported_load_permutation_p): More
3371 conservatively choose the vectorization factor when checking
3372 whether we can perform the required load permutation.
3373 (vect_transform_slp_perm_load): Assert when we may not fail.
3375 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
3377 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
3378 message for incompatible -msdata=* and -mcall-* options.
3380 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
3382 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
3383 rate for post-reload scheduling.
3385 2017-10-05 Tamar Christina <tamar.christina@arm.com>
3387 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
3389 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
3391 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
3392 to improve monte carlo in scimark.
3394 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
3396 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
3397 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
3398 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
3399 pentium4_cost, nocona_cost): Set reassociation width to 1.
3400 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
3401 width to 2 for fp operations and 1 otherwise.
3402 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
3404 (atom_cost): Set reassociation width to 2.
3405 (slm_cost, generic_cost): Set fp reassociation width
3406 to 2 and 1 otherwise.
3407 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
3408 (core_cost): Set fp reassociation width to 4 and vector to 2.
3409 (ix86_reassociation_width): Rewrite using cost table; special case
3410 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
3411 and TARGET_AVX128_OPTIMAL.
3412 * config/i386/i386.h (processor_costs): Add
3413 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
3414 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
3415 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
3416 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
3417 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
3418 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
3420 2017-10-05 Nathan Sidwell <nathan@acm.org>
3422 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
3424 2017-10-05 Tamar Christina <tamar.christina@arm.com>
3426 * config/arm/arm.c (arm_test_fpu_data): New.
3427 (arm_run_selftests): Call arm_test_fpu_data.
3429 2017-10-04 Nathan Sidwell <nathan@acm.org>
3431 * toplev.c (toplev::main): Remove excess parens on pretty_printer
3433 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
3435 2017-10-04 Sudakshina Das <sudi.das@arm.com>
3437 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
3438 check type for aarch64_simd_valid_immediate.
3439 (aarch64_output_simd_mov_immediate): Update prototype.
3440 (aarch64_simd_valid_immediate): Update prototype.
3441 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
3442 support for ORR-immediate.
3443 (and<mode>3): modified pattern to add support for BIC-immediate.
3444 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
3445 now checks for valid immediate for BIC and ORR based on new enum
3447 (aarch64_output_simd_mov_immediate): Function now used to output
3448 BIC/ORR imm as well based on new enum argument.
3449 * config/aarch64/constraints.md (Do): New vector immediate constraint.
3451 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
3452 (aarch64_reg_or_bic_imm): Likewise.
3454 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3456 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
3457 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
3459 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
3463 PR rtl-optimization/82396
3464 * haifa-sched.c (autopref_multipass_init): Simplify
3466 (autopref_rank_data): Simplify sort order.
3467 * sched-int.h (autopref_multipass_data_): Remove
3468 multi_mem_insn_p, min_offset and max_offset.
3470 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3472 * doc/sourcebuild.texi: Document vect_peeling_profitable.
3474 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3476 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
3479 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3481 * doc/sourcebuild.texi: Document vect_long_mult.
3483 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
3485 PR tree-optimization/82413
3486 * fold-const.c (build_range_check): Use widest_int when comparing
3487 the maximum ETYPE value with HIGH.
3489 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
3491 PR rtl-optimization/82396
3492 * haifa-sched.c (autopref_multipass_init): Simplify
3494 (autopref_rank_data): Simplify sort order.
3495 * sched-int.h (autopref_multipass_data_): Remove
3496 multi_mem_insn_p, min_offset and max_offset.
3498 2017-10-04 Jakub Jelinek <jakub@redhat.com>
3500 PR tree-optimization/82381
3501 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
3502 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
3505 PR tree-optimization/82374
3506 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
3507 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
3508 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
3509 current_function_decl to the new decl.
3511 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
3513 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
3514 helper macro for IEEE float128 hardware built-in functions.
3515 (SQRTF128_ODD): Add built-in functions with the round-to-odd
3517 (TRUNCF128_ODD): Likewise.
3518 (ADDF128_ODD): Likewise.
3519 (SUBF128_ODD): Likewise.
3520 (MULF128_ODD): Likewise.
3521 (DIVF128_ODD): Likewise.
3522 (FMAF128_ODD): Likewise.
3523 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
3524 UNSPEC_TRUNC_ROUND_TO_ODD.
3525 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
3526 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
3527 floating point round to odd instructions.
3528 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
3529 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
3530 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
3531 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
3532 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
3533 (trunc<mode>sf2_hw): Change the truncate with round to odd
3534 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
3535 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
3536 to odd hardware instructions.
3537 (sub<mode>3_odd): Likewise.
3538 (mul<mode>3_odd): Likewise.
3539 (div<mode>3_odd): Likewise.
3540 (sqrt<mode>2_odd): Likewise.
3541 (fma<mode>4_odd): Likewise.
3542 (fms<mode>4_odd): Likewise.
3543 (nfma<mode>4_odd): Likewise.
3544 (nfms<mode>4_odd): Likewise.
3545 (trunc<mode>df2_odd): Change the truncate with round to odd
3546 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
3548 * doc/extend.texi (PowerPC built-in functions): Update documentation
3549 for existing IEEE float128-bit built-in functions. Add built-in
3550 functions that generate the IEEE 128-bit floating point round to
3553 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
3555 PR rtl-optimization/77729
3556 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
3557 to (X&(C1&~C2))|C2 transformations.
3559 2017-10-03 Martin Jambor <mjambor@suse.cz>
3561 PR tree-optimization/82363
3562 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
3563 mismatch, mark lacc written regardless of racc.
3565 2017-10-03 Jakub Jelinek <jakub@redhat.com>
3567 PR tree-optimization/82381
3568 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
3569 stmt_to_insert nor wheather SSA_NAMEs are default defs.
3570 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
3571 fallthrough into reassoc_stmt_dominates_stmt_p.
3574 * combine.c (combine_instructions): Don't combine in unreachable
3577 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
3580 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
3581 function to not use the have_cpu variable. Do not set cpu_index,
3582 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
3584 (rs6000_valid_attribute_p): Remove duplicate initializations of
3585 old_optimize and func_optimize.
3586 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
3587 (rs6000_activate_target_options): Make global.
3588 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
3591 2017-10-02 Jakub Jelinek <jakub@redhat.com>
3593 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
3594 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
3595 Set *poffset to 0 and *psize and *pmax_size to -1 if
3596 *poffset + *psize overflows in HOST_WIDE_INT.
3598 PR tree-optimization/82387
3599 PR tree-optimization/82388
3600 PR tree-optimization/82389
3601 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
3602 instead of live_bytes non-NULL.
3604 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
3607 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
3610 2017-10-02 Richard Biener <rguenther@suse.de>
3612 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
3613 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
3614 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
3616 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
3618 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
3619 requested precision matches the type's.
3620 * calls.c (alloc_max_size): Calculate the new candidate size as
3621 a widest_int and use wi::to_widest when comparing it with the
3622 current candidate size.
3623 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
3624 zero rather than integer_zero_node.
3625 * match.pd: Check for a no-op conversion before using wi::add
3626 rather than after. Use tree_to_uhwi when summing small shift
3627 counts into an unsigned int.
3629 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
3630 Alan Hayward <alan.hayward@arm.com>
3631 David Sherwood <david.sherwood@arm.com>
3634 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
3635 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
3636 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
3637 POINTER_AND_FP_REGS.
3639 2017-10-02 Richard Biener <rguenther@suse.de>
3641 PR tree-optimization/82355
3642 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
3643 a mapping for the enclosing loop but avoid generating one for
3645 (copy_bb_and_scalar_dependences): Remove premature codegen
3646 error on PHIs in blocks duplicated into multiple places.
3647 * graphite-scop-detection.c
3648 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
3649 in the region use it as loop and nest to analyze the DR in.
3650 (try_generate_gimple_bb): Likewise.
3651 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
3652 (add_loop_constraints): For blocks in a loop not in the region
3653 create a dimension with a single iteration.
3654 * sese.h (gbb_loop_at_index): Remove assert.
3656 2017-10-01 Kevin Buettner <kevinb@redhat.com>
3658 * omp-expand.c (adjust_context_scope): New function.
3659 (expand_parallel_call): Call adjust_context_scope.
3661 2017-10-01 Jeff Law <law@redhat.com>
3663 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
3664 dom_opt_dom_walker class with direct access to private members.
3665 Add comments. Call test_for_singularity.
3666 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
3667 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
3668 m_dummy_cond anymore.
3669 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
3671 (pass_dominator:execute): Build the dummy_cond here and pass it
3672 to the dom_opt_dom_walker ctor.
3673 (test_for_singularity): New function.
3675 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
3676 Maya Rashish <coypu@sdf.org>
3678 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
3679 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
3680 (alpha*-*-netbsd*) Use nbsd_tm_file.
3681 (arm*-*-netbsdelf*) Likewise.
3682 (i[34567]86-*-netbsdelf*) Likewise.
3683 (x86_64-*-netbsd*) Likewise.
3684 (mips*-*-netbsd*) Likewise.
3685 (powerpc-*-netbsd*) Likewise.
3686 (sh*-*-netbsd*) Likewise.
3687 (sparc-*-netbsdelf*) Likewise.
3688 (sparc64-*-netbsd*) Likewise.
3689 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
3691 (vax-*-netbsdelf*) Likewise.
3692 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
3693 (UINT_FAST8_TYPE) Likewise.
3694 (INT_FAST16_TYPE) Check CHAR_FAST16.
3695 (UINT_FAST16_TYPE) Likewise.
3697 2017-09-30 Jakub Jelinek <jakub@redhat.com>
3700 * config/i386/i386.md
3701 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
3702 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
3703 *divmodsi4_zext_2): New define_insn_and_split.
3704 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
3705 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
3706 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
3707 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
3708 New define_insn_and_split.
3709 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
3710 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
3711 operands[1] having DImode when mode is SImode.
3713 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
3714 always SImode for DIV and MOD in REG_EQUAL notes.
3716 2017-09-29 Yury Gribov <tetra2005@gmail.com>
3719 * match.pd: Fix handling of NaNs in pattern.
3721 2017-09-29 Jeff Law <law@redhat.com>
3723 * sbitmap.c (bitmap_bit_in_range_p): New function.
3724 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
3725 * tree-ssa-dse.c (live_bytes_read): New function.
3726 (dse_classify_store): Ignore reads of dead bytes.
3728 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
3729 typos and whitespace errors.
3730 * config/i386/predicates.md (address_no_seg_operand): Likewise.
3731 * config/s390/s390.c (s390_emit_prologue): Likewise.
3733 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
3736 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
3737 with a symbol for LRA.
3739 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
3741 PR rtl-optimization/82338
3742 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
3744 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
3746 * genmodes.c (calc_wider_mode): Suppress qsort macro.
3747 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
3748 (qsort_chk): Declare.
3749 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
3750 (qsort_chk): New function.
3752 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3754 PR tree-optimization/82337
3755 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
3756 phi definition if the PHI result appears in an abnormal PHI.
3757 (find_basis_for_base_expr): Don't record a basis if the LHS of the
3758 basis appears in an abnormal PHI.
3760 2017-09-29 Richard Biener <rguenther@suse.de>
3762 * graphite-isl-ast-to-gimple.c
3763 (translate_isl_ast_to_gimple::set_codegen_error): New function.
3764 (binary_op_to_tree): Use it.
3765 (get_rename_from_scev): Likewise.
3766 (copy_loop_phi_nodes): Likewise.
3767 (copy_bb_and_scalar_dependences): Likewise.
3768 (translate_pending_phi_nodes): Likewise.
3770 2017-09-29 Jakub Jelinek <jakub@redhat.com>
3773 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
3774 for movabsq $(i32 << shift), r64.
3776 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
3778 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
3779 index when encoding %esp as %rsp to avoid 0x67 prefix.
3781 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
3783 * config/i386/i386.md (*movsf_internal, *movdf_internal):
3784 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
3786 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
3788 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
3789 Extensions with more than 16 double VFP registers.
3790 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
3791 to_clear_mask and all code related to it. Replace the remaining
3792 entry by a sbitmap and adapt code accordingly.
3794 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
3796 * brig-builtins.def: Change pure attributes to const.
3798 2017-09-28 Joseph Myers <joseph@codesourcery.com>
3800 * config.gcc (default_gnu_indirect_function): Default to yes for
3801 sparc*-*-linux* with glibc.
3803 2017-09-28 Joseph Myers <joseph@codesourcery.com>
3805 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
3806 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
3807 when creating .init_array and .fini_array sections with priority
3810 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
3813 * config/aarch64/aarch64.c
3814 (aarch64_builtin_support_vector_misalignment): Always return false
3815 when misalignment is unknown.
3817 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
3819 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
3820 this function to return false if the definition used by the swap
3821 instruction is artificial, or if the memory address from which the
3822 constant value is loaded is not represented by a base address held
3823 in a register or if the base address register is a frame or stack
3824 pointer. Additionally, return false if the base address of the
3825 loaded constant is a SYMBOL_REF but is not considered to be a
3827 (replace_swapped_load_constant): New function.
3828 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
3829 loaded constant vector with a load of a swapped constant vector.
3831 2017-09-27 Carl Love <cel@us.ibm.com>
3833 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
3834 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
3835 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
3838 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
3840 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
3841 first, always call autopref_rank_data otherwise.
3843 2017-09-27 Richard Biener <rguenther@suse.de>
3845 * graphite-scop-detection.c (find_scop_parameters): Move
3846 loop bound handling ...
3847 (gather_bbs::before_dom_children): ... here, avoiding the need
3848 to build scop_info->loop_nest.
3849 (record_loop_in_sese): Remove.
3850 * sese.h (sese_info_t::loop_nest): Remove.
3851 * sese.c (new_sese_info): Do not allocate loop_nest.
3852 (free_sese_info): Do not free loop_nest.
3854 2017-09-27 Jakub Jelinek <jakub@redhat.com>
3857 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
3858 lhs from calls if the lhs has addressable type.
3860 2017-09-27 Richard Biener <rguenther@suse.de>
3862 * graphite.h (scop::max_alias_set): New member.
3863 * graphite-scop-detection.c: Remove references to non-existing
3864 --param in comments.
3865 (build_alias_sets): Record the maximum alias set used for drs.
3866 (build_scops): Support zero as unlimited for
3867 --param graphite-max-arrays-per-scop.
3868 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
3870 (build_poly_sr_1): ... here. Compute alias set based on the
3871 maximum alias set used for drs rather than
3872 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
3874 2017-09-27 Richard Biener <rguenther@suse.de>
3876 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
3877 --param loop-block-tile-size=0 to disable tiling.
3879 2017-09-27 Richard Biener <rguenther@suse.de>
3881 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
3882 (graphite-max-nb-scop-params): Document special value zero.
3883 * domwalk.h (dom_walker::STOP): New symbolical constant.
3884 (dom_walker::dom_walker): Add optional parameter for bb to
3886 (dom_walker::~dom_walker): Declare.
3887 (dom_walker::before_dom_children): Document STOP return value.
3888 (dom_walker::m_user_bb_to_rpo): New member.
3889 (dom_walker::m_bb_to_rpo): Likewise.
3890 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
3891 mapping here if not provided by the user.
3892 (dom_walker::~dom_walker): Free bb to RPO mapping if not
3893 provided by the user.
3894 (dom_walker::STOP): Define.
3895 (dom_walker::walk): Do not compute bb to RPO mapping here.
3896 Support STOP return value from before_dom_children to stop
3898 * graphite-optimize-isl.c (optimize_isl): If the schedule
3899 is the same still generate code if -fgraphite-identity
3900 or -floop-parallelize-all are given.
3901 * graphite-scop-detection.c: Include cfganal.h.
3902 (gather_bbs::gather_bbs): Get and pass through bb to RPO
3904 (gather_bbs::before_dom_children): Return STOP for BBs
3906 (build_scops): Compute bb to RPO mapping and pass it to
3907 the domwalk. Treat --param graphite-max-nb-scop-params=0
3908 as not limiting the number of params.
3909 * graphite.c (graphite_initialize): Remove limit on the
3910 number of basic-blocks in a function.
3911 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
3912 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
3913 default value of 10.
3915 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
3917 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
3918 Adjust code to eliminate needing to do the shift right 32-bits
3919 operation after XSCVDPSPN.
3921 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3923 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
3924 ((X / Y) != 0 -> X >= Y): Likewise.
3926 2017-09-26 Carl Love <cel@us.ibm.com>
3928 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
3929 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
3930 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
3931 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
3932 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
3933 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
3934 definitions and overloading.
3935 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
3936 statement for P9V_BUILTIN_XST_LEN_R.
3937 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
3938 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
3939 define_expand and define_insn for the instructions and builtins.
3940 * doc/extend.texi: Update the built-in documentation file for the new
3942 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
3943 define_insn for the instructions
3945 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
3948 * gcc/config/netbsd-protos.h: New file.
3949 * gcc/config/netbsd.c: New file.
3950 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
3951 * gcc/config/t-netbsd: New file.
3952 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
3953 (tmake_file) Add t-netbsd.
3954 (extra_objs) Add netbsd.o.
3956 2017-09-26 Janus Weil <janus@gcc.gnu.org>
3960 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
3962 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
3964 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
3965 sign extension from a vector register to a GPR by doing a 32-bit
3966 direct move and then an EXTSW.
3967 (extendsi<mode>2 splitter): Likewise.
3968 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
3969 right or vector extract after doing XSCVDPSPN. Use
3970 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
3972 (movdi_from_sf_zero_ext): Likewise.
3973 (reload_gpr_from_vsxsf): Likewise.
3974 (p8_mfvsrd_4_disf): Delete, no longer used.
3975 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
3976 then needing to move the SFmode to a GPR to use the XSCVDPSP
3977 instruction instead of FRSP and XSCVDPSPN.
3978 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
3979 it is adjacent to the other XSCVSPDP insns.
3980 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
3981 SFmode to be in traditional Altivec registers.
3982 (vsx_xscvdpspn): Eliminate useless alternative constraint.
3983 (vsx_xscvspdpn): Likewise.
3984 (vsx_xscvspdpn_scalar): Likewise.
3986 2017-09-26 Martin Jambor <mjambor@suse.cz>
3988 * tree-sra.c (compare_access_positions): Put integral types first,
3989 stabilize sorting of integral types, remove conditions putting
3990 non-full-precision integers last.
3991 (sort_and_splice_var_accesses): Disable scalarization if a
3992 non-integert would be represented by a non-full-precision integer.
3994 2017-09-26 Joseph Myers <joseph@codesourcery.com>
3996 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
3997 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
3998 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
3999 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
4000 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
4001 conditionals inside the function instead of around it. Call
4002 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
4003 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
4005 2017-09-26 Richard Biener <rguenther@suse.de>
4007 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
4009 (scop_detection::build_scop_breadth): ... this. Removed.
4010 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
4011 (scop_detection::harmful_stmt_in_bb): Likewise.
4012 (scop_detection::graphite_can_represent_stmt): Likewise.
4013 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
4014 (scop_detection::can_represent_loop): Remove recursion, fold in ...
4015 (scop_detection::can_represent_loop_1): ... this. Removed.
4016 (scop_detection::harmful_loop_in_region): Simplify after inlining
4017 the above and remove more quadraticness.
4018 (build_scops): Adjust.
4019 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
4022 2017-09-26 Jakub Jelinek <jakub@redhat.com>
4025 * config/i386/i386.c (ix86_print_operand_address_as): Only test
4026 REGNO (base) == SP_REG if base is a REG.
4029 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
4030 if it is different SSA_NAME.
4031 (optimize_range_tests_cmp_bitwise): New function.
4032 (optimize_range_tests): Call it.
4034 2017-09-26 Richard Biener <rguenther@suse.de>
4036 PR tree-optimization/82321
4037 * graphite.c (canonicalize_loop_closed_ssa): Properly check
4038 for the def being inside the loop.
4040 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4042 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
4044 * config/s390/s390-builtins.def: Fix constraint on op4.
4046 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4048 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
4049 independent expanders.
4050 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
4051 ("vec_ordered", "vec_unordered"): New expanders.
4053 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4055 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
4058 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4060 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
4061 vec_unpacks_lo_v16qi.
4062 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
4064 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4066 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
4067 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
4068 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
4070 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4072 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
4074 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
4076 ("*vec_slb<mode>"): New insn pattern.
4077 ("vec_shr_<mode>"): New expander.
4078 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
4079 and force the shift count operand to V16QImode.
4080 ("vec_srb<mode>"): Set shift count mode to V16QI.
4082 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4084 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
4085 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
4086 ("vec_widen_smult_hi_<mode>"): New expander definitions.
4088 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
4091 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
4093 2017-09-26 Richard Biener <rguenther@suse.de>
4095 PR tree-optimization/82320
4096 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
4099 2017-09-25 Jeff Law <law@redhat.com>
4101 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
4102 prototype for new argument.
4103 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
4104 mostly extracted from rs6000_emit_allocate_stack.
4105 (rs6000_emit_probe_stack_range_stack_clash): New function.
4106 (rs6000_emit_allocate_stack): Call
4107 rs6000_emit_probe_stack_range_stack_clash as needed.
4108 (rs6000_emit_probe_stack_range): Add additional argument
4109 to call to gen_probe_stack_range{si,di}.
4110 (output_probe_stack_range): New.
4111 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
4112 (output_probe_stack_range_stack_clash): New.
4113 (rs6000_emit_prologue): Emit notes into dump file as requested.
4114 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
4115 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
4116 Add additional operand and pass it to output_probe_stack_range.
4118 2017-09-25 Bin Cheng <bin.cheng@arm.com>
4120 PR tree-optimization/82163
4121 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
4122 (checking_verify_loop_closed_ssa): New parameter.
4123 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
4124 (check_loop_closed_ssa_stmt): Delete.
4125 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
4126 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
4127 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
4130 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
4132 * brig-builtins.def: Treat HSAIL barrier builtins as
4133 setjmp/longjump style functions.
4135 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
4137 * target.def (constant_alignment): New hook.
4138 * defaults.h (CONSTANT_ALIGNMENT): Delete.
4139 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
4140 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
4141 * doc/tm.texi: Regenerate.
4142 * targhooks.h (default_constant_alignment): Declare.
4143 (constant_alignment_word_strings): Likewise.
4144 * targhooks.c (default_constant_alignment): New function.
4145 (constant_alignment_word_strings): Likewise.
4146 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
4147 instead of CONSTANT_ALIGNMENT.
4148 * varasm.c (align_variable, get_variable_align, build_constant_desc)
4149 (force_const_mem): Likewise.
4150 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
4151 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
4152 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
4153 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4154 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
4156 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
4157 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4158 constant_alignment_word_strings.
4159 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
4160 (CONSTANT_ALIGNMENT): Likewise.
4161 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4162 (arm_constant_alignment): New function.
4163 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
4164 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4165 constant_alignment_word_strings.
4166 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
4167 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4168 constant_alignment_word_strings.
4169 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
4170 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4171 (cris_constant_alignment): New function.
4172 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
4173 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4174 (epiphany_constant_alignment): New function.
4175 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
4176 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4177 constant_alignment_word_strings.
4178 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
4179 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4180 constant_alignment_word_strings.
4181 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
4182 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4183 constant_alignment_word_strings.
4184 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
4185 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
4186 * config/i386/i386.c (ix86_constant_alignment): Make static.
4187 Use the same interface as the target hook.
4188 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4189 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
4190 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4191 constant_alignment_word_strings.
4192 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
4193 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
4194 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4195 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
4196 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4197 constant_alignment_word_strings.
4198 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
4199 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4200 constant_alignment_word_strings.
4201 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
4202 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4203 constant_alignment_word_strings.
4204 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
4205 * config/microblaze/microblaze.c (microblaze_constant_alignment):
4207 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4208 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
4209 * config/mips/mips.c (mips_constant_alignment): New function.
4210 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4211 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
4212 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
4213 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4214 (mmix_constant_alignment): Make static. Use the same interface
4216 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
4217 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4218 constant_alignment_word_strings.
4219 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
4220 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4221 constant_alignment_word_strings.
4222 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
4223 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4224 constant_alignment_word_strings.
4225 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
4226 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4227 (rs6000_constant_alignment): New function.
4228 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
4229 * config/riscv/riscv.c (riscv_constant_alignment): New function.
4230 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4231 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
4232 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4233 (rs6000_constant_alignment): New function.
4234 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
4235 * config/s390/s390.c (s390_constant_alignment): New function.
4236 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4237 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
4238 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4239 constant_alignment_word_strings.
4240 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
4241 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4242 (sparc_constant_alignment): New function.
4243 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
4244 * config/spu/spu.c (spu_constant_alignment): New function.
4245 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4246 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
4247 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4248 constant_alignment_word_strings.
4249 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
4250 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4251 constant_alignment_word_strings.
4252 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
4253 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4254 constant_alignment_word_strings.
4255 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
4256 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4257 (visium_constant_alignment): New function.
4258 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
4259 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4260 (xtensa_constant_alignment): New function.
4261 * system.h (CONSTANT_ALIGNMENT): Poison.
4263 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
4265 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
4266 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
4267 (rs6000_builtin_valid_without_lhs): New helper function.
4268 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4269 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
4271 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
4273 * target.h (vec_perm_indices): Use unsigned short rather than
4275 (auto_vec_perm_indices): Likewise.
4276 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
4277 Use unsigned int rather than unsigned char.
4278 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
4280 2017-09-25 Richard Biener <rguenther@suse.de>
4282 * cfgloop.h (sort_sibling_loops): Declare.
4283 * cfgloop.c (sort_sibling_loops_cmp): New helper.
4284 (sort_sibling_loops): New function sorting the sibling loop list
4286 * graphite.c (graphite_transform_loops): Sort sibling loops.
4288 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
4290 * target.def (vec_perm_const_ok): Change sel parameter to
4292 * optabs-query.c (can_vec_perm_p): Update accordingly.
4293 * doc/tm.texi: Regenerate.
4294 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
4295 auto_vec_perm_indices and remove separate nelt field.
4296 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
4297 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
4298 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
4299 (aarch64_expand_vec_perm_const): Update accordingly.
4300 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
4301 to vec_perm_indices.
4302 * config/arm/arm.c (expand_vec_perm_d): Change perm to
4303 auto_vec_perm_indices and remove separate nelt field.
4304 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
4305 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
4306 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
4308 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
4309 to vec_perm_indices.
4310 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
4311 sel to vec_perm_indices.
4312 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
4313 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
4314 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
4316 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
4319 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
4322 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
4323 on the FUNCTION_DECL function context if it has a DIE that is a
4326 2017-09-25 Richard Biener <rguenther@suse.de>
4328 PR tree-optimization/82285
4329 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
4332 2017-09-25 Tom de Vries <tom@codesourcery.com>
4336 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
4339 2017-09-25 Richard Biener <rguenther@suse.de>
4341 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
4342 ISL errors other than isl_error_quota happen. Dump if the
4343 schedule is the same.
4344 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
4345 errors instead of aborting inside ISL.
4347 2017-09-25 Iain Sandoe <iain@codesourcery.com>
4350 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
4351 of libgcc_eh for m64.
4352 * config/i386/darwin64.h: Likewise.
4354 2017-09-25 Richard Biener <rguenther@suse.de>
4357 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
4358 attribute for incomplete types nor twice for complete ones.
4360 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
4363 * config/i386/i386.c (ix86_print_operand_address_as): Encode
4364 %esp as %rsp to avoid 0x67 prefix if there is no index or base
4367 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
4370 * config/i386/i386.opt (mprefer-avx256): Use
4371 ix86_target_flags variable.
4372 * config/i386/i386.c (ix86_target_string): Move
4373 -mprefer-avx256 to flag2_opts.
4375 2017-09-22 Jakub Jelinek <jakub@redhat.com>
4378 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
4379 and x != -1 | y != -1 into (x & y) != -1.
4381 2017-09-22 Steve Ellcey <sellcey@cavium.com>
4383 * config.gcc: Add new case statement to set
4384 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
4385 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
4386 s390x-*-linux* case statements. Added aarch64 to the list of
4387 supported architectures.
4389 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
4391 PR tree-optimization/82289
4392 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
4393 STMT_VINFO_RELEVANT_P.
4395 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
4396 Alan Hayward <alan.hayward@arm.com>
4397 David Sherwood <david.sherwood@arm.com>
4399 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
4400 for VR_RANGE only; don't allow VR_ANTI_RANGE.
4401 (extract_range_from_binary_expr_1): Don't call
4402 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
4404 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
4405 Alan Hayward <alan.hayward@arm.com>
4406 David Sherwood <david.sherwood@arm.com>
4408 * target.def (preferred_vector_alignment): New hook.
4409 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
4411 * doc/tm.texi: Regenerate.
4412 * targhooks.h (default_preferred_vector_alignment): Declare.
4413 * targhooks.c (default_preferred_vector_alignment): New function.
4414 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
4416 (DR_TARGET_ALIGNMENT): New macro.
4417 (aligned_access_p): Update commentary.
4418 (vect_known_alignment_in_bytes): New function.
4419 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
4421 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
4422 Calculate the misalignment based on the target alignment rather than
4424 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
4425 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
4426 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
4427 the target alignment, rather than masking the element misalignment
4428 with the number of elements in a vector. Also use the target
4429 alignment when calculating the maximum number of peels.
4430 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
4431 instead of TYPE_ALIGN_UNIT.
4432 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
4433 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
4434 (vect_create_addr_base_for_vector_ref): Update call accordingly.
4435 (vect_create_data_ref_ptr): Likewise.
4436 (vect_setup_realignment): Realign by ANDing with
4437 -DR_TARGET_MISALIGNMENT.
4438 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
4439 the number of peels based on DR_TARGET_ALIGNMENT.
4440 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
4441 with the guaranteed alignment boundary when deciding whether
4443 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
4444 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
4445 (ensure_base_align): Remove stmt_info parameter. Get the
4446 target base alignment from DR_TARGET_ALIGNMENT.
4447 (vectorizable_store): Update call accordingly. Interpret
4448 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
4450 (vectorizable_load): Likewise.
4452 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
4453 Alan Hayward <alan.hayward@arm.com>
4454 David Sherwood <david.sherwood@arm.com>
4456 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
4457 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
4458 (vect_enhance_data_refs_alignment): Likewise.
4460 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
4462 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
4463 error. Only quit immediately if parsing is complete.
4464 (BEGIN): Initialize fatal_err and parse_done.
4465 (begin fpu, end fpu): Check number of arguments.
4466 (begin arch, end arch): Likewise.
4467 (begin cpu, end cpu): Likewise.
4468 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
4469 (optalias): Likewise.
4471 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
4473 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
4474 * config/arm/arm-isa.h: Delete. Move definitions to ...
4475 * arm-cpus.in: ... here. Use new feature and fgroup values.
4476 * config/arm/arm.c (arm_option_override): Use lower case for feature
4478 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
4479 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
4480 * config/arm/parsecpu.awk (END): Add new command 'isa'.
4482 (print_isa_bits_for): New function.
4483 (gen_isa): New function.
4484 (gen_comm_data): Use print_isa_bits_for.
4485 (define feature): New keyword.
4486 (define fgroup): New keyword.
4487 * config/arm/t-arm (TM_H): Remove.
4488 (GTM_H): Add arm-isa.h.
4489 (arm-isa.h): Add rule to generate file.
4490 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
4491 case for feature bit names.
4493 2017-09-22 Richard Biener <rguenther@suse.de>
4495 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
4497 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
4498 print of no dependency loops ...
4499 * graphite.c (graphite_transform_loops): ... here.
4500 (canonicalize_loop_closed_ssa_form): Work from inner to outer
4502 (same_close_phi_node, remove_duplicate_close_phi,
4503 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
4504 (canonicalize_loop_closed_ssa): ... here and simplify.
4505 * graphite-optimize-isl.c: Include tree-vectorizer.h.
4506 (optimize_isl): Use dump_printf_loc to tell when we stopped
4507 optimizing because of an ISL timeout.
4509 2017-09-22 Richard Biener <rguenther@suse.de>
4511 PR tree-optimization/82291
4512 * tree-if-conv.c (predicate_mem_writes): Make sure to
4513 remove writes in blocks predicated with false.
4515 2017-09-22 Richard Biener <rguenther@suse.de>
4517 * sese.c: Include cfganal.h.
4518 (if_region_set_false_region): Remove.
4519 (create_if_region_on_edge): Likewise.
4520 (move_sese_in_condition): Re-implement without destroying
4523 2017-09-22 Richard Biener <rguenther@suse.de>
4525 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
4526 Verify both BBs contain loop PHI nodes before dispatching to
4528 (graphite_regenerate_ast_isl): Do not recompute dominators,
4529 do not verify three times. Restructure for clarity.
4530 * graphite-scop-detection.c (same_close_phi_node,
4531 remove_duplicate_close_phi, make_close_phi_nodes_unique,
4532 defined_in_loop_p, canonicalize_loop_closed_ssa,
4533 canonicalize_loop_closed_ssa_form): Simplify, remove excess
4534 checking and SSA rewrite, move to ...
4535 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
4536 (graphite_initialize): Do not pass in ctx, do not reset the
4537 SCEV cache, compute only dominators.
4538 (graphite_transform_loops): Allocate ISL ctx after
4539 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
4540 Maintain post-dominators only around build_scops.
4541 * sese.c (if_region_set_false_region): Make static. Free
4542 and recompute dominators.
4543 (move_sese_in_condition): Assert we don't get called with
4544 post-dominators computed.
4545 * sese.h (if_region_set_false_region): Remove.
4547 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
4549 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
4550 mode attribute for TARGET_AVX512VL.
4552 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
4554 * config/i386/i386.opt (mprefer-avx256): New option.
4555 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
4557 (ix86_preferred_simd_mode): Return 256-bit AVX modes
4558 for TARGET_PREFER_AVX256.
4559 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
4561 2017-09-21 Jeff Law <law@redhat.com>
4563 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
4564 Fix dump output if the only stack space is for pushed registers.
4566 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4568 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
4571 2017-09-21 Martin Sebor <msebor@redhat.com>
4574 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
4575 code (in C++) or code that triggers warnings.
4577 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
4579 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
4581 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
4583 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
4584 * sched-rgn.c: Ditto.
4585 * sel-sched-ir.c: Ditto.
4587 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
4589 * toplev.h (set_random_seed): Adjust return type.
4590 * toplev.c (init_local_tick): Move eager initialization of random_seed
4591 to get_random_seed. Adjust comment.
4592 (init_random_seed): Inline to get_random_seed, delete.
4593 (get_random_seed): Initialize random_seed lazily.
4594 (set_random_seed): Do not return previous value.
4595 (print_switch_value): Do not call get_random_seed.
4597 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
4599 * cgraph.c (delete_function_version): New, broken out from...
4600 (cgraph_node::delete_function_version): ...here. Rename to
4601 cgraph_node::delete_function_version_by_decl. Update all uses.
4602 (cgraph_node::remove): Call delete_function_version.
4604 2017-09-21 Jakub Jelinek <jakub@redhat.com>
4607 * tree-inline.c (expand_call_inline): Emit clobber stmts for
4608 VAR_DECLs to which addressable non-volatile parameters are mapped
4609 and for id->retvar after the return value assignment. Clear
4610 id->retval and id->retbnd after inlining.
4612 2017-09-21 Richard Biener <rguenther@suse.de>
4614 PR tree-optimization/82276
4615 PR tree-optimization/82244
4616 * tree-vrp.c (build_assert_expr_for): Set
4617 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
4619 (remove_range_assertions): Revert earlier change.
4621 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
4624 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
4626 2017-09-21 Richard Biener <rguenther@suse.de>
4628 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
4629 Restore valid IL after code generation errors.
4630 * graphite.c (graphite_transform_loops): Diagnose code
4631 generation issues as MSG_MISSED_OPTIMIZATION and continue
4632 with processing SCOPs.
4634 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4635 Alan Hayward <alan.hayward@arm.com>
4636 David Sherwood <david.sherwood@arm.com>
4638 * calls.c (compute_argument_addresses): Use simplify_gen_binary
4639 rather than choosing between plus_constant and gen_rtx_<CODE>.
4640 * expr.c (emit_push_insn): Likewise.
4641 (expand_expr_real_2): Likewise.
4643 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4644 Alan Hayward <alan.hayward@arm.com>
4645 David Sherwood <david.sherwood@arm.com>
4647 * loop-unroll.c (split_iv): Call copy_rtx on the step.
4649 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4650 Alan Hayward <alan.hayward@arm.com>
4651 David Sherwood <david.sherwood@arm.com>
4653 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
4654 calling tree_to_uhwi.
4656 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4657 Alan Hayward <alan.hayward@arm.com>
4658 David Sherwood <david.sherwood@arm.com>
4660 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
4661 INTEGER_CST rather than a negative test for ADDR_EXPR.
4663 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4664 Alan Hayward <alan.hayward@arm.com>
4665 David Sherwood <david.sherwood@arm.com>
4667 * tree-vrp.c (extract_range_from_binary_expr_1): Check
4668 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
4670 2017-09-21 Richard Biener <rguenther@suse.de>
4672 PR tree-optimization/71351
4673 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
4674 graphite_create_new_loop_guard): Remove, fold remaining parts
4676 (translate_isl_ast_node_for): ... here and simplify.
4678 2017-09-21 Jakub Jelinek <jakub@redhat.com>
4681 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
4682 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
4683 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
4684 alternative always use QI mode, for -Os imov (=R,R) alternative
4685 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
4688 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4689 Jeff Law <law@redhat.com>
4691 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
4692 (allocate_stack_space): New function, partially extracted from
4694 (s390_emit_prologue): Track offset to most recent stack probe.
4695 Code to allocate space moved into allocate_stack_space.
4696 Dump actions when no stack is allocated.
4697 (s390_prologue_plus_offset): New function.
4698 (s390_emit_stack_probe): Likewise.
4700 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
4702 * common.opt (Wa, Wl, Wp, g, gz=): Add
4704 (gno-column-info): Remove.
4705 (gcolumn-info): Drop RejectNegative.
4707 (gno-record-gcc-switches): Remove.
4708 (grecord-gcc-switches): Drop RejectNegative.
4709 (gno-split-dwarf): Remove.
4710 (gsplit-dwarf): Drop RejectNegative.
4711 (gno-strict-dwarf): Remove.
4712 (gstrict-dwarf): Drop RejectNegative.
4713 * config/darwin.opt (gfull, gused): Add RejectNegative.
4714 * dwarf2out.c (gen_producer_string): Drop
4715 gno-record-gcc-switches handler.
4716 * optc-gen.awk: Add g to prefixes with negative forms.
4717 * opts-common.c (remapping_prefix_p): New.
4718 (find_opt): Check it.
4719 (generate_canonical_option): Test g prefix.
4720 (option_map): Add -gno- mapping.
4721 (add_misspelling_candidates): Check remapping_prefix_p.
4723 2017-09-20 Jeff Law <law@redhat.com>
4725 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
4726 thinko in stack clash protection support.
4728 * explow.c (compute_stack_clash_protection_loop_data): Use
4729 CONST_INT_P instead of explicit test. Verify object is a
4730 CONST_INT_P before looking at INTVAL.
4731 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
4732 instead of explicit test.
4734 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
4737 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
4738 address instead of to r1 and r11.
4740 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
4742 * config.gcc: Support "knm".
4743 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
4744 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
4746 * config/i386/i386.c (m_KNM): Define.
4747 (processor_target_table): Add "knm".
4749 (ix86_option_override_internal): Add "knm".
4750 (ix86_issue_rate): Add PROCESSOR_KNM.
4751 (ix86_adjust_cost): Ditto.
4752 (ia32_multipass_dfa_lookahead): Ditto.
4753 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
4754 (fold_builtin_cpu): Add M_INTEL_KNM.
4755 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
4756 (processor_type): Add PROCESSOR_KNM.
4757 * config/i386/x86-tune.def: Add m_KNM.
4758 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
4760 2017-09-20 Richard Biener <rguenther@suse.de>
4762 PR tree-optimization/80213
4763 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
4764 are allowed in empty BBs as well.
4765 (canonicalize_loop_closed_ssa): Also look for other complex
4767 (scop_detection::get_sese): Include the loop-closed PHI block
4769 (scop_detection::merge_sese): Remove code adding extra blocks.
4770 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
4771 (build_scops): Assert the final returned scop is invalid.
4773 2017-09-20 Richard Biener <rguenther@suse.de>
4775 PR tree-optimization/82264
4776 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
4778 (vn_phi_lookup): Likewise.
4779 (vn_phi_insert): Likewise.
4781 2017-09-20 Jakub Jelinek <jakub@redhat.com>
4783 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
4784 that fits into uhwi or shwi, add DW_AT_const_value regardless
4785 of early_dwarf without going through RTL, using add_AT_unsigned
4788 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
4789 (DEBUG_LTO_ABBREV_SECTION): Likewise.
4790 (DEBUG_LTO_MACINFO_SECTION): Likewise.
4791 (DEBUG_MACRO_SECTION): Likewise.
4792 (DEBUG_LTO_MACRO_SECTION): Likewise.
4793 (DEBUG_STR_DWO_SECTION): Likewise.
4794 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
4795 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
4796 (DEBUG_LTO_DWO_LINE_SECTION): Define.
4797 (DEBUG_LTO_LINE_STR_SECTION): Define.
4798 (init_sections_and_labels): Initialize debug_line_str_section
4799 variable. Initialize debug_loc_section for -gdwarf-5 to
4800 DEBUG_LOCLISTS_SECTION. Formatting fixes.
4802 2017-09-20 Richard Biener <rguenther@suse.de>
4804 * graphite-sese-to-poly.c (extract_affine): Properly handle
4805 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
4807 2017-09-20 Richard Biener <rguenther@suse.de>
4809 PR tree-optimization/81373
4810 * graphite-scop-detection.c (build_cross_bb_scalars_def):
4811 Force SESE live-out defs to be handled even if they are
4814 2017-09-19 Jeff Law <law@redhat.com>
4816 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
4817 nothing for stack adjustments with REG_STACK_CHECK.
4818 * sched-deps.c (parse_add_or_inc): Reject insns with
4819 REG_STACK_CHECK from dependency breaking.
4820 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
4821 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
4822 * reg-notes.def (STACK_CHECK): New note.
4824 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
4825 (ix86_expand_prologue): Dump stack clash info as needed.
4826 Call ix86_adjust_stack_and_probe_stack_clash as needed.
4828 * function.c (dump_stack_clash_frame_info): New function.
4829 * function.h (dump_stack_clash_frame_info): Prototype.
4830 (enum stack_clash_probes): New enum.
4832 * config/alpha/alpha.c (alpha_expand_prologue): Also check
4833 flag_stack_clash_protection.
4834 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
4835 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
4836 (arm_frame_pointer_required): Likewise.
4837 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
4838 (ia64_expand_prologue): Likewise.
4839 * config/mips/mips.c (mips_expand_prologue): Likewise.
4840 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
4841 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
4842 (sparc_flat_expand_prologue): Likewise.
4843 * config/spu/spu.c (spu_expand_prologue): Likewise.
4845 * explow.c: Include "params.h".
4846 (anti_adjust_stack_and_probe_stack_clash): New function.
4847 (get_stack_check_protect): Likewise.
4848 (compute_stack_clash_protection_loop_data): Likewise.
4849 (emit_stack_clash_protection_loop_start): Likewise.
4850 (emit_stack_clash_protection_loop_end): Likewise.
4851 (allocate_dynamic_stack_space): Use get_stack_check_protect.
4852 Use anti_adjust_stack_and_probe_stack_clash.
4853 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
4854 (emit_stack_clash_protection_loop_start): Likewise.
4855 (emit_stack_clash_protection_loop_end): Likewise.
4856 * rtl.h (get_stack_check_protect): Prototype.
4857 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
4858 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
4859 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
4861 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
4863 * doc/tm.texi: Rebuilt.
4864 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
4865 get_stack_check_protect.
4866 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
4867 * config/arm/arm.c (arm_expand_prologue): Likewise.
4868 (arm_frame_pointer_required): Likewise.
4869 * config/i386/i386.c (ix86_expand_prologue): Likewise.
4870 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
4871 * config/mips/mips.c (mips_expand_prologue): Likewise.
4872 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
4873 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
4874 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
4875 (sparc_flat_expand_prologue): Likewise.
4877 * common.opt (-fstack-clash-protection): New option.
4878 * flag-types.h (enum stack_check_type): Note difference between
4879 -fstack-check= and -fstack-clash-protection.
4880 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
4881 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
4882 * toplev.c (process_options): Issue warnings/errors for cases
4883 not handled with -fstack-clash-protection.
4884 * doc/invoke.texi (-fstack-clash-protection): Document new option.
4885 (-fstack-check): Note additional problem with -fstack-check=generic.
4886 Note that -fstack-check is primarily for Ada and refer users
4887 to -fstack-clash-protection for stack-clash-protection.
4888 Document new params for stack clash protection.
4890 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
4892 * config/i386/i386.c (ix86_split_long_move): Do not handle
4893 address used for LEA in a special way.
4895 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
4897 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
4899 2017-09-19 Martin Sebor <msebor@redhat.com>
4902 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
4903 of incompatible types.
4905 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
4907 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
4908 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
4909 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4910 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
4912 2017-09-19 Richard Biener <rguenther@suse.de>
4914 PR tree-optimization/82244
4915 * tree-vrp.c (remove_range_assertions): Do not propagate
4916 a constant to abnormals but replace the assert with a copy.
4918 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
4920 PR rtl-optimization/57878
4921 PR rtl-optimization/68988
4922 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
4923 avoidance test involving non_reload_pseudos. Move frequency test
4924 below the general fragmentation avoidance test.
4926 2017-09-19 Richard Biener <rguenther@suse.de>
4928 PR tree-optimization/69728
4929 * graphite-sese-to-poly.c (schedule_error): New global.
4930 (add_loop_schedule): Handle empty domain by failing the
4932 (build_original_schedule): Handle schedule_error.
4934 2017-09-19 Richard Biener <rguenther@suse.de>
4936 * graphite-scop-detection.c (scop_detection::can_represent_loop):
4937 Do not iterate to sibling loops but only to siblings of inner
4940 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
4943 * config/m68k/m68k.md (moveq feeding equality comparison): Check
4944 that the registers are different.
4946 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
4948 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
4949 to processor_model and "amdfam17h" to arch_names_table.
4950 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
4952 2017-09-18 Jakub Jelinek <jakub@redhat.com>
4955 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
4957 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4958 Alan Hayward <alan.hayward@arm.com>
4959 David Sherwood <david.sherwood@arm.com>
4961 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
4963 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
4965 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
4966 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
4967 vect_schedule_slp_instance.
4968 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
4969 Update call to vect_slp_analyze_node_operations. Simplify return
4971 (vect_slp_analyze_bb_1): Update call accordingly.
4972 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
4973 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
4974 (vect_schedule_slp): Update call accordingly.
4976 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4977 Alan Hayward <alan.hayward@arm.com>
4978 David Sherwood <david.sherwood@arm.com>
4980 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
4981 with types that aren't in fact scalar.
4983 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4985 * tree-vect-slp.c (vect_record_max_nunits): New function,
4987 (vect_build_slp_tree_1): ...here.
4988 (vect_build_slp_tree_2): Call it for phis too.
4990 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4992 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
4993 to vect_get_vec_def_for_operand when getting the mask operand.
4995 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4996 Alan Hayward <alan.hayward@arm.com>
4997 David Sherwood <david.sherwood@arm.com>
4999 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
5002 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5003 Alan Hayward <alan.hayward@arm.com>
5004 David Sherwood <david.sherwood@arm.com>
5006 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
5007 calculation for vector booleans.
5009 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5010 Alan Hayward <alan.hayward@arm.com>
5011 David Sherwood <david.sherwood@arm.com>
5013 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
5015 (vect_transform_stmt): ...here.
5016 (vect_analyze_stmt): Use it instead of calling
5017 vectorizable_live_operation directly.
5019 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
5021 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
5022 non-SIMT targets in acc vector loops.
5024 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
5026 * configure.ac: Add arc and check if assembler supports gdwarf2.
5027 * configure: Regenerate.
5029 2017-09-18 Richard Biener <rguenther@suse.de>
5031 PR tree-optimization/82220
5032 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
5033 epilogue niters from the min_profitable_iters compute.
5035 2017-09-18 Jakub Jelinek <jakub@redhat.com>
5038 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
5039 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
5040 (ix86_init_pic_reg): Revert 2017-09-01 changes.
5042 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
5045 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
5046 switching to a new text section.
5048 2017-09-18 Richard Biener <rguenther@suse.de>
5050 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
5052 (build_alias_set): Reject aliases with no access function.
5054 2017-09-18 Richard Biener <rguenther@suse.de>
5056 PR tree-optimization/79622
5057 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
5059 (build_cross_bb_scalars_use): Likewise.
5061 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
5063 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
5065 2017-09-18 Alan Modra <amodra@gmail.com>
5068 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
5069 stack_pointer_rtx for count 0. Update comments. Break up
5070 large rtl expression.
5072 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
5074 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
5075 Increase to 20 bytes.
5076 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
5077 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
5078 or avx version of the stub.
5080 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
5083 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
5084 compute the minimum stack alignment. Also update preferred stack
5085 boundary for leaf functions.
5087 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
5089 PR tree-optimization/82228
5090 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
5093 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
5095 * common/config/nds32/nds32-common.c
5096 (nds32_option_optimization_table): Refine formatting.
5097 (nds32_option_optimization_table): Use -fsched-pressure and
5098 -fomit-frame-pointer for specific optimization level.
5100 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
5102 * config/nds32/nds32.c: Refine formatting and comments.
5103 * config/nds32/nds32.h: Likewise.
5104 * config/nds32/nds32.md: Likewise.
5105 * config/nds32/nds32-cost.c: Likewise.
5106 * config/nds32/nds32-isr.c: Likewise.
5107 * config/nds32/nds32-md-auxiliary.c: Likewise.
5108 * config/nds32/nds32-multiple.md: Likewise.
5109 * config/nds32/nds32-predicates.c: Likewise.
5111 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
5112 Jakub Jelinek <jakub@redhat.com>
5114 Add support for -std=c++2a.
5115 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
5117 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
5119 2017-09-15 Steve Ellcey <sellcey@cavium.com>
5122 * doc/extend.texi (Common Function Attributes): Add
5123 references to ARM, AArch64, and S/390 specific attributes.
5124 (Function Specific Option Pragmas): Add AArch64 and S/390
5125 to list of back ends that support the target pragma.
5127 2017-09-15 Nathan Sidwell <nathan@acm.org>
5129 * doc/standards.texi: Fix C++17 description. Update URLs for
5132 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
5134 * common.opt (Wcast-align=strict): New warning option.
5135 * doc/invoke.texi: Document -Wcast-align=strict.
5137 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
5139 * cgraph.h (cgraph_thunk_info): Add comments.
5140 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
5141 assert for VIRTUAL_* arguments stricter.
5143 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
5145 PR tree-optimization/71026
5146 * match.pd: Move RDIV patterns from fold-const.c
5147 * fold-const.c (distribute_real_division): Removed.
5148 (fold_binary_loc): Remove calls to distribute_real_divison.
5150 2017-09-15 Jakub Jelinek <jakub@redhat.com>
5152 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
5153 c++1z and gnu++1z as deprecated. Change other references to
5154 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
5155 Change -Wc++1z-compat to -Wc++17-compat.
5156 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
5157 * dwarf2out.c (highest_c_language): Handle C++17.
5158 (gen_compile_unit_die): Likewise.
5160 2017-09-15 Jakub Jelinek <jakub@redhat.com>
5162 PR rtl-optimization/82192
5163 * combine.c (make_extraction): Don't look through non-paradoxical
5164 SUBREGs or TRUNCATE if pos + len is or might be bigger than
5167 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
5168 Alan Hayward <alan.hayward@arm.com>
5169 David Sherwood <david.sherwood@arm.com>
5171 * target.def (function_arg_offset): New hook.
5172 * targhooks.h (default_function_arg_offset): Declare.
5173 * targhooks.c (default_function_arg_offset): New function.
5174 * function.c (locate_and_pad_parm): Use
5175 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
5176 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
5177 (TARGET_FUNCTION_ARG_OFFSET): ...this.
5178 * doc/tm.texi: Regenerate.
5179 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
5180 * config/spu/spu.c (spu_function_arg_offset): New function.
5181 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
5182 * system.h (FUNCTION_ARG_OFFSET): Poison.
5184 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
5185 Alan Hayard <alan.hayward@arm.com>
5186 David Sherwood <david.sherwood@arm.com>
5188 * target.def (truly_noop_truncation): New hook.
5189 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
5190 than TRULY_NOOP_TRUNCATION.
5191 * hooks.h (hook_bool_uint_uint_true): Declare.
5192 * hooks.c (hook_bool_uint_uint_true): New function.
5193 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
5194 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
5195 * doc/tm.texi: Regenerate.
5196 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
5197 rather than TRULY_NOOP_TRUNCATION in comments.
5198 (simplify_comparison): Likewise.
5199 (record_truncated_value): Likewise.
5200 * expmed.c (extract_bit_field_1): Likewise.
5201 (extract_split_bit_field): Likewise.
5202 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
5203 instead of TRULY_NOOP_TRUNCATION.
5204 * function.c (assign_parm_setup_block): Likewise.
5205 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
5206 * rtlhooks.c: Include target.h.
5207 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
5208 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
5209 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
5210 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
5211 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
5212 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
5213 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
5214 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
5215 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
5216 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
5217 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
5218 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
5219 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
5220 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
5221 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
5222 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
5223 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
5224 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
5225 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
5226 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
5227 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
5228 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
5229 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
5230 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
5231 * config/mips/mips.c (mips_truly_noop_truncation): New function.
5232 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
5233 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
5234 rather than TRULY_NOOP_TRUNCATION in comments.
5235 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
5236 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
5237 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
5238 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
5239 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
5240 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
5241 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
5242 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
5243 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
5244 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
5245 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
5246 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
5247 rather than TRULY_NOOP_TRUNCATION in comments.
5248 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
5249 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
5250 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
5251 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
5252 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
5253 TRULY_NOOP_TRUNCATION condition.
5254 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
5255 (TRULY_NOOP_TRUNCATION): Delete.
5256 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
5257 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
5258 * config/spu/spu.c (spu_truly_noop_truncation): New function.
5259 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
5260 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
5261 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
5262 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
5263 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
5264 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
5265 rather than TRULY_NOOP_TRUNCATION in comments.
5266 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
5267 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
5268 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
5269 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
5270 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
5271 * system.h (TRULY_NOOP_TRUNCATION): Poison.
5273 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
5276 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
5277 (*cmp_ior): Likewise.
5278 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
5279 (*ior_scc_scc_cmp): Likewise.
5280 (*and_scc_scc): Likewise.
5281 (*and_scc_scc_cmp): Likewise.
5283 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
5284 Alan Hayard <alan.hayward@arm.com>
5285 David Sherwood <david.sherwood@arm.com>
5287 * target.def (can_change_mode_class): New hook.
5288 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
5289 (hard_regno_nregs): Likewise.
5290 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
5291 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
5292 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
5293 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
5294 (LOAD_EXTEND_OP): Update accordingly.
5295 * doc/tm.texi: Regenerate.
5296 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
5297 CANNOT_CHANGE_MODE_CLASS.
5298 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
5299 (REG_CAN_CHANGE_MODE_P): ...this new macro.
5300 * combine.c (simplify_set): Update accordingly.
5301 * emit-rtl.c (validate_subreg): Likewise.
5302 * recog.c (general_operand): Likewise.
5303 * regcprop.c (mode_change_ok): Likewise.
5304 * reload1.c (choose_reload_regs): Likewise.
5305 (inherit_piecemeal_p): Likewise.
5306 * rtlanal.c (simplify_subreg_regno): Likewise.
5307 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
5308 instead of CANNOT_CHANGE_MODE_CLASS.
5309 (reload_cse_simplify_operands): Likewise.
5310 * reload.c (push_reload): Use targetm.can_change_mode_class
5311 instead of CANNOT_CHANGE_MODE_CLASS.
5312 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
5313 REG_CANNOT_CHANGE_MODE_P.
5314 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5315 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
5316 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5317 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5318 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5319 (arm_can_change_mode_class): New function.
5320 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
5321 than CANNOT_CHANGE_MODE_CLASS in comments.
5322 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5323 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
5324 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
5325 (ix86_can_change_mode_class): ...this new function, inverting the
5326 sense of the return value.
5327 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5328 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5329 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5330 (ia64_can_change_mode_class): New function.
5331 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5332 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
5333 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
5334 (m32c_can_change_mode_class): ...this new function, inverting the
5335 sense of the return value.
5336 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5337 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5338 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
5339 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
5340 (mips_can_change_mode_class): ...this new function, inverting the
5341 sense of the return value.
5342 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5343 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5344 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5345 (msp430_can_change_mode_class): New function.
5346 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5347 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
5348 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5349 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5350 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5351 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
5352 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5353 (pa_cannot_change_mode_class): Replace with...
5354 (pa_can_change_mode_class): ...this new function, inverting the
5355 sense of the return value.
5356 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
5357 than CANNOT_CHANGE_MODE_CLASS in comments.
5358 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5359 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
5360 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5361 (pdp11_cannot_change_mode_class): Replace with...
5362 (pdp11_can_change_mode_class): ...this new function, inverting the
5363 sense of the return value.
5364 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5365 * config/powerpcspe/powerpcspe-protos.h
5366 (rs6000_cannot_change_mode_class_ptr): Delete.
5367 * config/powerpcspe/powerpcspe.c
5368 (rs6000_cannot_change_mode_class_ptr): Delete.
5369 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5370 (rs6000_option_override_internal): Assign to
5371 targetm.can_change_mode_class instead of
5372 rs6000_cannot_change_mode_class_ptr.
5373 (rs6000_cannot_change_mode_class): Replace with...
5374 (rs6000_can_change_mode_class): ...this new function, inverting the
5375 sense of the return value.
5376 (rs6000_debug_cannot_change_mode_class): Replace with...
5377 (rs6000_debug_can_change_mode_class): ...this new function.
5378 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5379 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
5380 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5381 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5382 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
5384 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
5385 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5386 (rs6000_option_override_internal): Assign to
5387 targetm.can_change_mode_class instead of
5388 rs6000_cannot_change_mode_class_ptr.
5389 (rs6000_cannot_change_mode_class): Replace with...
5390 (rs6000_can_change_mode_class): ...this new function, inverting the
5391 sense of the return value.
5392 (rs6000_debug_cannot_change_mode_class): Replace with...
5393 (rs6000_debug_can_change_mode_class): ...this new function.
5394 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5395 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
5396 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
5397 (s390_can_change_mode_class): ...this new function, inverting the
5398 sense of the return value.
5399 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5400 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5401 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
5402 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5403 (sh_cannot_change_mode_class): Replace with...
5404 (sh_can_change_mode_class): ...this new function, inverting the
5405 sense of the return value.
5406 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5407 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5408 (sparc_can_change_mode_class): New function.
5409 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5410 * config/spu/spu.c (spu_can_change_mode_class): New function.
5411 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5412 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5413 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5414 (visium_can_change_mode_class): New function.
5415 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
5417 2017-09-15 Richard Biener <rguenther@suse.de>
5419 PR tree-optimization/82217
5420 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
5421 but not undefined case.
5423 2017-09-15 Jakub Jelinek <jakub@redhat.com>
5426 * postreload.c (reload_cse_simplify_operands): Skip
5427 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
5429 2017-09-15 Richard Biener <rguenther@suse.de>
5431 PR tree-optimization/68823
5432 * graphite-scop-detection.c (build_alias_set): If we have a
5433 possible dependence check whether we can handle them by just
5434 looking at the DRs DR_ACCESS_FNs.
5435 (build_scops): If build_alias_set fails, fail the SCOP.
5437 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
5439 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
5440 to support float128 built-in functions that require the ISA 3.0
5442 (BU_FLOAT128_3_HW): Likewise.
5443 (SQRTF128): Add support for the IEEE 128-bit square root and fma
5445 (FMAF128): Likewise.
5447 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
5448 support for built-in functions that need the ISA 3.0 IEEE 128-bit
5449 floating point instructions.
5450 (rs6000_invalid_builtin): Likewise.
5451 (rs6000_builtin_mask_names): Likewise.
5452 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
5453 (RS6000_BTM_FLOAT128_HW): Likewise.
5454 (RS6000_BTM_COMMON): Likewise.
5455 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
5457 * doc/extend.texi (RS/6000 built-in functions): Document the
5458 IEEE 128-bit floating point square root and fused multiply-add
5461 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
5463 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
5464 reg (r2) isn't in the set of registers defined in the prologue.
5466 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5467 Alan Hayward <alan.hayward@arm.com>
5468 David Sherwood <david.sherwood@arm.com>
5470 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
5471 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
5472 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
5473 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
5474 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
5476 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
5477 max_vectorization_factor.
5478 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
5480 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5481 Alan Hayward <alan.hayward@arm.com>
5482 David Sherwood <david.sherwood@arm.com>
5484 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
5485 (vect_worthwhile_without_simd_p): Declare.
5486 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
5487 (vectorizable_reduction): Use it.
5488 * tree-vect-stmts.c (vectorizable_shift): Likewise.
5489 (vectorizable_operation): Likewise.
5491 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5492 Alan Hayward <alan.hayward@arm.com>
5493 David Sherwood <david.sherwood@arm.com>
5495 * tree-vectorizer.h (vect_get_num_copies): New function.
5496 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
5497 * tree-vect-loop.c (vectorizable_reduction): Likewise.
5498 (vectorizable_induction): Likewise.
5499 (vectorizable_live_operation): Likewise.
5500 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
5501 (vectorizable_bswap): Likewise.
5502 (vectorizable_call): Likewise.
5503 (vectorizable_conversion): Likewise.
5504 (vectorizable_assignment): Likewise.
5505 (vectorizable_shift): Likewise.
5506 (vectorizable_operation): Likewise.
5507 (vectorizable_store): Likewise.
5508 (vectorizable_load): Likewise.
5509 (vectorizable_condition): Likewise.
5510 (vectorizable_comparison): Likewise.
5511 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
5513 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5514 Alan Hayward <alan.hayward@arm.com>
5515 David Sherwood <david.sherwood@arm.com>
5517 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
5518 of vect_init_vector.
5520 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5521 Alan Hayward <alan.hayward@arm.com>
5522 David Sherwood <david.sherwood@arm.com>
5524 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
5525 an inline wrapper that provides a location.
5526 (gimple_build_vector): Likewise.
5527 * gimple-fold.c (gimple_build_vector_from_val): New function.
5528 (gimple_build_vector): Likewise.
5529 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
5530 functions to build the initial value. Always return a gimple value.
5531 (get_initial_defs_for_reduction): Likewise. Only compute
5533 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
5534 vect_init_vector on the results from get_initial_def(s)_for_reduction.
5535 (vectorizable_induction): Use gimple_build_vector rather than
5538 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5539 Alan Hayward <alan.hayward@arm.com>
5540 David Sherwood <david.sherwood@arm.com>
5542 * target.h (vec_perm_indices): New typedef.
5543 (auto_vec_perm_indices): Likewise.
5544 * optabs-query.h: Include target.h
5545 (can_vec_perm_p): Take a vec_perm_indices *.
5546 * optabs-query.c (can_vec_perm_p): Likewise.
5547 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
5548 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
5549 * tree-vect-generic.c (lower_vec_perm): Likewise.
5550 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
5551 (vect_grouped_load_supported): Likewise.
5552 (vect_shift_permute_load_chain): Likewise.
5553 (vect_permute_store_chain): Use auto_vec_perm_indices.
5554 (vect_permute_load_chain): Likewise.
5555 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
5556 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
5557 Update uses of can_vec_perm_p.
5558 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
5559 mode with a number of elements. Take a vec_perm_indices *.
5560 (vect_create_epilog_for_reduction): Update accordingly.
5561 Use auto_vec_perm_indices.
5562 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
5563 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
5564 (vect_transform_slp_perm_load): Likewise.
5565 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
5566 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
5567 (vect_gen_perm_mask_checked): Likewise.
5568 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
5569 (vect_gen_perm_mask_checked): Likewise.
5570 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
5571 (vectorizable_store): Likewise.
5572 (vectorizable_load): Likewise.
5573 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
5574 (vectorizable_bswap): Likewise.
5576 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5577 Alan Hayward <alan.hayward@arm.com>
5578 David Sherwood <david.sherwood@arm.com>
5580 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
5581 * tree.c (build_vector): Likewise.
5582 (build_vector_from_ctor): Update accordingly.
5583 (build_vector_from_val): Likewise.
5584 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
5585 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
5586 * tree-vect-generic.c (add_rshift): Likewise.
5587 (expand_vector_divmod): Likewise.
5588 (optimize_vector_constructor): Likewise.
5589 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
5590 (vect_transform_slp_perm_load): Likewise.
5591 (vect_schedule_slp_instance): Likewise.
5592 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
5593 (vectorizable_call): Likewise.
5594 (vect_gen_perm_mask_any): Likewise. Add elements in order.
5595 * expmed.c (make_tree): Likewise.
5596 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
5597 a vector passed to build_vector.
5598 (fold_convert_const): Likewise.
5599 (exact_inverse): Likewise.
5600 (fold_ternary_loc): Likewise.
5601 (fold_relational_const): Likewise.
5602 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
5603 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
5604 (const_unop): Likewise. Store the reduction accumulator in a
5605 variable rather than an array.
5606 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
5607 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
5608 the new vector, rather than constructing it after the input arrays.
5609 (native_interpret_vector): Use auto_vec<tree> when building
5610 a vector passed to build_vector. Add elements in order.
5611 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
5612 auto_vec<tree> when building a vector passed to build_vector.
5613 (vect_create_epilog_for_reduction): Likewise.
5614 (vectorizable_induction): Likewise.
5615 (get_initial_def_for_reduction): Likewise. Fix indentation of
5617 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
5619 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
5620 passed to build_vector.
5622 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5623 Alan Hayward <alan.hayward@arm.com>
5624 David Sherwood <david.sherwood@arm.com>
5626 * tree-core.h (tree_base::u): Add an "nelts" field.
5627 (tree_vector): Use VECTOR_CST_NELTS as the length.
5628 * tree.c (tree_size): Likewise.
5629 (make_vector): Initialize VECTOR_CST_NELTS.
5630 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
5631 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
5632 TYPE_VECTOR_SUBPARTS.
5633 * expr.c (const_vector_mask_from_tree): Consistently use "units"
5634 as the number of units, setting it from VECTOR_CST_NELTS.
5635 (const_vector_from_tree): Likewise.
5636 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
5637 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
5638 (fold_negate_expr_1): Likewise.
5639 (fold_convert_const): Likewise.
5640 (const_binop): Likewise. Differentiate the number of output and
5642 (const_unop): Likewise.
5643 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
5644 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
5645 in cases that did the opposite.
5647 2017-09-14 Richard Biener <rguenther@suse.de>
5649 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
5652 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
5654 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
5656 2017-09-14 Jakub Jelinek <jakub@redhat.com>
5659 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
5660 if and where to split a bb, except for splitting before debug insn
5661 sequences followed by non-label real insn. Delete debug insns
5662 in between basic blocks.
5664 * combine.c (make_compound_operation_int): Formatting fixes.
5666 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
5667 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5668 * config/netbsd.h (LINK_EH_SPEC): Likewise.
5669 * config/sol2.h (LINK_EH_SPEC): Likewise.
5670 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5671 * config/s390/linux.h (LINK_SPEC): Likewise.
5672 * config/freebsd.h (LINK_EH_SPEC): Likewise.
5673 * config/openbsd.h (LINK_EH_SPEC): Likewise.
5674 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5675 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
5676 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
5677 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5678 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
5679 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
5681 2017-09-13 Jakub Jelinek <jakub@redhat.com>
5683 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
5685 (ENDFILE_LINUX_SPEC): Likewise.
5686 (LINK_EH_SPEC): Likewise.
5687 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
5688 (LINK_OS_LINUX_SPEC32): Likewise.
5689 (LINK_OS_LINUX_SPEC64): Likewise.
5690 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
5691 (LINK_OS_LINUX_SPEC): Likewise.
5693 2017-09-13 Martin Liska <mliska@suse.cz>
5696 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
5697 CASE_HIGH is NULL_TREE.
5699 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
5700 Alan Hayward <alan.hayward@arm.com>
5701 David Sherwood <david.sherwood@arm.com>
5703 * target.def (secondary_memory_needed): New hook.
5704 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
5705 instead of SECONDARY_MEMORY_NEEDED.
5706 (secondary_memory_needed_mode): Likewise.
5707 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
5708 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
5709 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
5710 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
5711 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
5712 * doc/tm.texi: Regenerate.
5713 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
5714 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
5715 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5716 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
5717 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
5718 * config/i386/i386.c (inline_secondary_memory_needed): Put the
5719 mode argument first and change the reg_class arguments to reg_class_t.
5720 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
5721 Make static. Update the call to inline_secondary_memory_needed.
5722 (ix86_register_move_cost): Update the call to
5723 inline_secondary_memory_needed.
5724 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5725 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
5727 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
5728 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
5730 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
5731 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
5732 * config/mips/mips.c (mips_secondary_memory_needed): Make static
5733 and match hook interface. Add comment from mips.h.
5734 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5735 * config/mmix/mmix.md (truncdfsf2): Refer to
5736 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
5738 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
5739 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
5740 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5741 (pa_secondary_memory_needed): New function.
5742 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
5743 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
5744 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5745 (pdp11_secondary_memory_needed): Make static and match hook interface.
5746 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
5747 * config/powerpcspe/powerpcspe-protos.h
5748 (rs6000_secondary_memory_needed_ptr): Delete.
5749 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
5751 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5752 (rs6000_option_override_internal): Assign to
5753 targetm.secondary_memory_needed rather than
5754 rs6000_secondary_memory_needed_ptr.
5755 (rs6000_secondary_memory_needed): Match hook interface.
5756 (rs6000_debug_secondary_memory_needed): Likewise.
5757 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
5758 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
5759 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
5760 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5761 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
5762 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
5764 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
5765 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5766 (rs6000_option_override_internal): Assign to
5767 targetm.secondary_memory_needed rather than
5768 rs6000_secondary_memory_needed_ptr.
5769 (rs6000_secondary_memory_needed): Match hook interface.
5770 (rs6000_debug_secondary_memory_needed): Likewise.
5771 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
5772 * config/s390/s390.c (s390_secondary_memory_needed): New function.
5773 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5774 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
5775 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5776 (sparc_secondary_memory_needed): New function.
5777 * lra-constraints.c (check_and_process_move): Refer to
5778 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
5780 (curr_insn_transform): Likewise.
5781 (process_alt_operands): Use targetm.secondary_memory_needed
5782 instead of TARGET_SECONDARY_MEMORY_NEEDED.
5783 (check_secondary_memory_needed_p): Likewise.
5784 (choose_split_class): Likewise.
5785 * reload.c: Unconditionally include code that was previously
5786 conditional on SECONDARY_MEMORY_NEEDED.
5787 (push_secondary_reload): Use targetm.secondary_memory_needed
5788 instead of TARGET_SECONDARY_MEMORY_NEEDED.
5789 (push_reload): Likewise.
5790 * reload1.c: Unconditionally include code that was previously
5791 conditional on SECONDARY_MEMORY_NEEDED.
5792 (choose_reload_regs): Use targetm.secondary_memory_needed
5793 instead of TARGET_SECONDARY_MEMORY_NEEDED.
5794 (gen_reload): Likewise.
5795 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
5797 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
5798 Alan Hayward <alan.hayward@arm.com>
5799 David Sherwood <david.sherwood@arm.com>
5801 * target.def (secondary_memory_needed_mode): New hook:
5802 * targhooks.c (default_secondary_memory_needed_mode): Declare.
5803 * targhooks.h (default_secondary_memory_needed_mode): New function.
5804 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
5805 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
5806 * doc/tm.texi: Regenerate.
5807 * lra-constraints.c (check_and_process_move): Use
5808 targetm.secondary_memory_needed_mode instead of
5809 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
5810 (curr_insn_transform): Likewise.
5811 * reload.c (get_secondary_mem): Likewise.
5812 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
5813 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
5815 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
5816 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
5817 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
5818 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
5819 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
5821 * config/powerpcspe/powerpcspe-protos.h
5822 (rs6000_secondary_memory_needed_mode): Delete.
5823 * config/powerpcspe/powerpcspe.c
5824 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
5825 (rs6000_secondary_memory_needed_mode): Make static.
5826 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
5827 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
5829 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
5831 (rs6000_secondary_memory_needed_mode): Make static.
5832 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
5833 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
5834 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
5835 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
5836 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
5838 (sparc_secondary_memory_needed_mode): New function.
5839 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
5841 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
5843 * config/aarch64/constraints.md (Umq): New constraint.
5844 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
5846 (mov<mode>): Update condition.
5848 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5850 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
5851 when bitposition is the same.
5853 2017-09-13 Richard Biener <rguenther@suse.de>
5855 * dwarf2out.c (output_die_symbol): Remove.
5856 (output_die): Do not output a DIEs symbol.
5858 2017-09-13 Richard Biener <rguenther@suse.de>
5861 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
5862 default-def to avoid breaking iterator update with the weird
5863 interaction with cgraph_update_edges_for_call_stmt_node.
5865 2017-09-13 Richard Biener <rguenther@suse.de>
5867 * tree-cfg.c (verify_gimple_assign_binary): Add verification
5868 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
5869 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
5870 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
5872 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
5874 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
5875 Disable pc relative literal load irrespective of
5876 TARGET_FIX_ERR_A53_84341 for default.
5878 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
5880 * config/sparc/sparc.c (output_return): Output the source location of
5881 the insn in the delay slot, if any.
5882 (output_sibcall): Likewise.
5884 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
5887 * common.opt (-static-pie): New alias.
5888 (shared): Negate static-pie.
5889 (-no-pie): Update help text.
5891 (static-pie): New option.
5892 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
5893 -static-pie support.
5894 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
5895 (LINK_EH_SPEC): Likewise.
5896 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5897 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
5898 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
5899 * gcc.c (LINK_COMMAND_SPEC): Likewise.
5900 (init_gcc_specs): Likewise.
5901 (init_spec): Likewise.
5902 (display_help): Update help message for -pie.
5903 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
5906 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
5908 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
5909 (movdi_aarch64): Likewise.
5910 (movti_aarch64): Likewise.
5912 2017-09-12 Simon Wright <simon@pushface.org>
5915 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
5916 calculation of the minor version, always output as 0.
5918 2017-09-12 Jakub Jelinek <jakub@redhat.com>
5921 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
5922 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
5923 on it early, rather than manual conversion late. For
5924 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
5925 instead of performing manual conversion.
5927 2017-09-12 Carl Love <cel@us.ibm.com>
5929 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
5930 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
5932 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
5933 VMULOSW): Add definitions.
5934 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5935 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
5936 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
5937 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
5938 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
5940 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
5942 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
5944 (movti_aarch64): Likewise.
5945 (movdf_aarch64): Likewise.
5946 (movtf_aarch64): Likewise.
5947 (load_pairdi): Likewise.
5948 (store_pairdi): Likewise.
5949 (load_pairdf): Likewise.
5950 (store_pairdf): Likewise.
5951 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
5952 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
5953 (ldr_got_small_<mode>): Likewise.
5954 (ldr_got_small_28k_<mode>): Likewise.
5955 (ldr_got_tiny): Likewise.
5956 * config/aarch64/iterators.md (ldst_sz): New.
5957 (ldpstp_sz): Likewise.
5958 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
5960 (thunderx_load): Split load_8 to load_16.
5961 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
5963 (thunderx2t99_storepair_basic): Split store_8 to store_16.
5964 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
5965 (xgene1_store_pair): Split store_8 to store_16.
5966 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
5967 (falkor_st_0_st_sd): Split store_8 to store_16.
5969 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
5971 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
5972 and store1/2/3/4 to store_4/8/12/16.
5973 * config/aarch64/aarch64.md: Update for rename.
5974 * config/arm/arm.md: Likewise.: Likewise.
5975 * config/arm/arm.c: Likewise.
5976 * config/arm/thumb1.md: Likewise.
5977 * config/arm/thumb2.md: Likewise.
5978 * config/arm/vfp.md: Likewise.
5979 * config/arm/arm-generic.md: Likewise.
5980 * config/arm/arm1020e.md: Likewise.
5981 * config/arm/arm1026ejs.md: Likewise.
5982 * config/arm/arm1136jfs.md: Likewise.
5983 * config/arm/arm926ejs.md: Likewise.
5984 * config/arm/cortex-a15.md: Likewise.
5985 * config/arm/cortex-a17.md: Likewise.
5986 * config/arm/cortex-a5.md: Likewise.
5987 * config/arm/cortex-a53.md: Likewise.
5988 * config/arm/cortex-a57.md: Likewise.
5989 * config/arm/cortex-a7.md: Likewise.
5990 * config/arm/cortex-a8.md: Likewise.
5991 * config/arm/cortex-a9.md: Likewise.
5992 * config/arm/cortex-m4.md: Likewise.
5993 * config/arm/cortex-m7.md: Likewise.
5994 * config/arm/cortex-r4.md: Likewise.
5995 * config/arm/exynos-m1.md: Likewise.
5996 * config/arm/fa526.md: Likewise.
5997 * config/arm/fa606te.md: Likewise.
5998 * config/arm/fa626te.md: Likewise.
5999 * config/arm/fa726te.md: Likewise.
6000 * config/arm/fmp626.md: Likewise.
6001 * config/arm/iwmmxt.md: Likewise.
6002 * config/arm/ldmstm.md: Likewise.
6003 * config/arm/marvell-pj4.md: Likewise.
6004 * config/arm/xgene1.md: Likewise.
6005 * config/aarch64/thunderx.md: Likewise.
6006 * config/aarch64/thunderx2t99.md: Likewise.
6007 * config/aarch64/falkor.md: Likewise.
6009 2017-09-12 Martin Liska <mliska@suse.cz>
6011 * attribs.c (private_lookup_attribute): New function.
6012 * attribs.h (private_lookup_attribute): Declared here.
6013 (lookup_attribute): Called from this place.
6015 2017-09-12 Richard Biener <rguenther@suse.de>
6017 PR tree-optimization/82157
6018 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
6019 stmts with side-effects.
6021 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6022 Alan Hayward <alan.hayward@arm.com>
6023 David Sherwood <david.sherwood@arm.com>
6025 * target.def (hard_regno_nregs): New hook.
6026 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
6027 * targhooks.h (default_hard_regno_nregs): Declare.
6028 * targhooks.c (default_hard_regno_nregs): New function.
6029 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
6030 (TARGET_HARD_REGNO_NREGS): ...this hook.
6031 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
6032 (CLASS_MAX_NREGS): Likewise.
6033 * doc/tm.texi: Regenerate.
6034 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
6035 instead of HARD_REGNO_NREGS.
6036 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
6037 HARD_REGNO_NREGS in the comment.
6038 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
6039 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
6040 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
6041 Return an unsigned int.
6042 (TARGET_HARD_REGNO_NREGS): Redefine.
6043 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
6044 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
6045 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
6046 (arc_hard_regno_nregs): New function.
6047 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
6048 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
6049 (arm_hard_regno_nregs): New function.
6050 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
6051 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
6052 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
6053 (TARGET_HARD_REGNO_NREGS): Redefine.
6054 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
6055 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
6056 (HARD_REGNO_NREGS): Delete.
6057 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
6058 (cr16_hard_regno_nregs): New function.
6059 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
6060 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
6061 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
6062 (cris_hard_regno_nregs): New function.
6063 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
6064 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
6065 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
6066 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
6067 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
6068 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
6069 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
6070 (frv_hard_regno_nregs): Make static. Take and return an
6072 (frv_class_max_nregs): Remove outdated copy of documentation.
6073 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
6074 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
6075 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
6076 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
6077 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
6078 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
6079 (TARGET_HARD_REGNO_NREGS): Redefine.
6080 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
6081 (CLASS_MAX_NREGS): Update comment.
6082 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
6083 (ia64_hard_regno_nregs): New function.
6084 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
6085 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
6086 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
6087 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
6088 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
6090 (m32c_hard_regno_nregs): Likewise. Make static.
6091 (TARGET_HARD_REGNO_NREGS): Redefine.
6092 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
6093 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
6094 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
6095 (m68k_hard_regno_nregs): New function.
6096 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
6097 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
6098 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
6099 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
6100 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
6101 Take and return an unsigned int.
6102 (TARGET_HARD_REGNO_NREGS): Redefine.
6103 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
6104 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
6105 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
6106 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
6107 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
6108 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
6109 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
6110 (msp430_hard_regno_nregs): Make static. Take and return an
6112 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
6113 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
6114 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
6115 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
6116 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
6117 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
6118 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
6119 (TARGET_HARD_REGNO_NREGS): Redefine.
6120 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
6121 (PA_HARD_REGNO_NREGS): ...this.
6122 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
6123 (PA_HARD_REGNO_NREGS): ...this.
6124 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
6125 (pa_hard_regno_nregs): New function.
6126 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
6127 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
6128 (pdp11_hard_regno_nregs): New function.
6129 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
6130 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
6131 (rs6000_hard_regno_nregs_hook): New function.
6132 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
6133 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
6134 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
6135 Take and return an unsigned int. Move earlier in file.
6136 (TARGET_HARD_REGNO_NREGS): Redefine.
6137 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
6138 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
6139 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
6140 (rl78_hard_regno_nregs): Make static. Take and return an
6142 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
6143 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
6144 (rs6000_hard_regno_nregs_hook): New function.
6145 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
6146 * config/rx/rx.c (rx_hard_regno_nregs): New function.
6147 (TARGET_HARD_REGNO_NREGS): Redefine.
6148 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
6149 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
6150 instead of HARD_REGNO_NREGS.
6151 (s390_hard_regno_nregs): New function.
6152 (s390_hard_regno_mode_ok): Add comment from s390.h.
6153 (TARGET_HARD_REGNO_NREGS): Redefine.
6154 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
6155 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
6156 (sh_hard_regno_nregs): New function.
6157 (sh_pass_in_reg_p): Use it.
6158 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
6159 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
6160 (sparc_hard_regno_nregs): New function.
6161 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
6162 * config/spu/spu.c (spu_hard_regno_nregs): New function.
6163 (spu_function_arg_advance): Use it, supplying a valid register number.
6164 (TARGET_HARD_REGNO_NREGS): Redefine.
6165 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
6166 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
6167 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
6168 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
6169 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
6170 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
6171 (CLASS_MAX_NREGS): Remove copy of old documentation.
6172 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
6173 (visium_hard_regno_nregs): New function.
6174 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
6175 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
6176 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
6177 xtensa_hard_regno_nregs): New function.
6178 * system.h (HARD_REGNO_NREGS): Poison.
6180 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6182 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
6183 hard_regno_nregs instead of HARD_REGNO_NREGS.
6184 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
6185 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
6186 (c6x_expand_epilogue): Likewise.
6187 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
6188 (frv_read_iacc_argument): Likewise.
6189 * config/sh/sh.c: Include regs.h.
6190 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
6191 (regs_used): Likewise.
6192 (output_stack_adjust): Likewise.
6193 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
6194 * expmed.c: Include regs.h.
6195 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
6196 * ree.c: Include regs.h.
6197 (combine_reaching_defs): Use hard_regno_nregs instead of
6199 (add_removable_extension): Likewise.
6201 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6203 * regs.h (hard_regno_nregs): Turn into a function.
6204 (end_hard_regno): Update accordingly.
6205 * caller-save.c (setup_save_areas): Likewise.
6206 (save_call_clobbered_regs): Likewise.
6207 (replace_reg_with_saved_mem): Likewise.
6208 (insert_restore): Likewise.
6209 (insert_save): Likewise.
6210 * combine.c (can_change_dest_mode): Likewise.
6211 (move_deaths): Likewise.
6212 (distribute_notes): Likewise.
6213 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
6214 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
6215 (rs6000_split_multireg_move): Likewise.
6216 (rs6000_register_move_cost): Likewise.
6217 (rs6000_memory_move_cost): Likewise.
6218 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
6219 (rs6000_split_multireg_move): Likewise.
6220 (rs6000_register_move_cost): Likewise.
6221 (rs6000_memory_move_cost): Likewise.
6222 * cselib.c (cselib_reset_table): Likewise.
6223 (cselib_lookup_1): Likewise.
6224 * emit-rtl.c (set_mode_and_regno): Likewise.
6225 * function.c (aggregate_value_p): Likewise.
6226 * ira-color.c (setup_profitable_hard_regs): Likewise.
6227 (check_hard_reg_p): Likewise.
6228 (calculate_saved_nregs): Likewise.
6229 (assign_hard_reg): Likewise.
6230 (improve_allocation): Likewise.
6231 (calculate_spill_cost): Likewise.
6232 * ira-emit.c (modify_move_list): Likewise.
6233 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
6234 (ira_hard_reg_in_set_p): Likewise.
6235 * ira.c (setup_reg_mode_hard_regset): Likewise.
6236 (clarify_prohibited_class_mode_regs): Likewise.
6237 (check_allocation): Likewise.
6238 * lra-assigns.c (find_hard_regno_for_1): Likewise.
6239 (lra_setup_reg_renumber): Likewise.
6240 (setup_try_hard_regno_pseudos): Likewise.
6241 (spill_for): Likewise.
6242 (assign_hard_regno): Likewise.
6243 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
6244 * lra-constraints.c (in_class_p): Likewise.
6245 (lra_constraint_offset): Likewise.
6246 (simplify_operand_subreg): Likewise.
6247 (lra_constraints): Likewise.
6248 (split_reg): Likewise.
6249 (split_if_necessary): Likewise.
6250 (invariant_p): Likewise.
6251 (inherit_in_ebb): Likewise.
6252 * lra-lives.c (process_bb_lives): Likewise.
6253 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
6254 (get_hard_regs): Likewise.
6255 (do_remat): Likewise.
6256 * lra-spills.c (assign_spill_hard_regs): Likewise.
6257 * mode-switching.c (create_pre_exit): Likewise.
6258 * postreload.c (reload_combine_recognize_pattern): Likewise.
6259 * recog.c (peep2_find_free_register): Likewise.
6260 * regcprop.c (kill_value_regno): Likewise.
6261 (set_value_regno): Likewise.
6262 (copy_value): Likewise.
6263 (maybe_mode_change): Likewise.
6264 (find_oldest_value_reg): Likewise.
6265 (copyprop_hardreg_forward_1): Likewise.
6266 * regrename.c (check_new_reg_p): Likewise.
6267 (regrename_do_replace): Likewise.
6268 * reload.c (push_reload): Likewise.
6269 (combine_reloads): Likewise.
6270 (find_dummy_reload): Likewise.
6271 (operands_match_p): Likewise.
6272 (find_reloads): Likewise.
6273 (find_equiv_reg): Likewise.
6274 (reload_adjust_reg_for_mode): Likewise.
6275 * reload1.c (count_pseudo): Likewise.
6276 (count_spilled_pseudo): Likewise.
6277 (find_reg): Likewise.
6278 (clear_reload_reg_in_use): Likewise.
6279 (free_for_value_p): Likewise.
6280 (allocate_reload_reg): Likewise.
6281 (choose_reload_regs): Likewise.
6282 (reload_adjust_reg_for_temp): Likewise.
6283 (emit_reload_insns): Likewise.
6284 (delete_output_reload): Likewise.
6285 * rtlanal.c (subreg_get_info): Likewise.
6286 * sched-deps.c (sched_analyze_reg): Likewise.
6287 * sel-sched.c (init_regs_for_mode): Likewise.
6288 (mark_unavailable_hard_regs): Likewise.
6289 (choose_best_reg_1): Likewise.
6290 (verify_target_availability): Likewise.
6291 * valtrack.c (dead_debug_insert_temp): Likewise.
6292 * var-tracking.c (track_loc_p): Likewise.
6293 (emit_note_insn_var_location): Likewise.
6294 * varasm.c (make_decl_rtl): Likewise.
6295 * reginfo.c (choose_hard_reg_mode): Likewise.
6296 (init_reg_modes_target): Refer directly to
6297 this_target_regs->x_hard_regno_nregs.
6299 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6301 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
6302 instead of hard_regno_nregs.
6304 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6306 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
6307 end_hard_regno instead of hard_regno_nregs.
6308 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
6309 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
6310 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
6311 * ira-color.c (improve_allocation): Likewise.
6312 * lra-assigns.c (find_hard_regno_for_1): Likewise.
6313 * lra-lives.c (mark_regno_live): Likewise.
6314 (mark_regno_dead): Likewise.
6315 * lra-remat.c (operand_to_remat): Likewise.
6316 * lra.c (collect_non_operand_hard_regs): Likewise.
6317 * postreload.c (reload_combine_note_store): Likewise.
6318 (move2add_valid_value_p): Likewise.
6319 * reload.c (regno_clobbered_p): Likewise.
6321 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6323 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
6325 * config/v850/v850.c (v850_reorg): Likewise.
6326 * reload.c (refers_to_regno_for_reload_p): Likewise.
6327 (find_equiv_reg): Likewise.
6328 * reload1.c (reload_reg_reaches_end_p): Likewise.
6330 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6332 * caller-save.c (add_used_regs): Use REG_NREGS instead of
6334 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
6335 * config/arm/arm.c (output_move_neon): Likewise.
6336 (arm_attr_length_move_neon): Likewise.
6337 (neon_split_vcombine): Likewise.
6338 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
6339 (c6x_mark_reg_written): Likewise.
6340 (c6x_dwarf_register_span): Likewise.
6341 * config/i386/i386.c (ix86_save_reg): Likewise.
6342 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
6343 (rws_access_reg): Likewise.
6344 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
6345 * mode-switching.c (create_pre_exit): Likewise.
6346 * ree.c (combine_reaching_defs): Likewise.
6347 (add_removable_extension): Likewise.
6348 * regcprop.c (find_oldest_value_reg): Likewise.
6349 (copyprop_hardreg_forward_1): Likewise.
6350 * reload.c (reload_inner_reg_of_subreg): Likewise.
6351 (push_reload): Likewise.
6352 (combine_reloads): Likewise.
6353 (find_dummy_reload): Likewise.
6354 (reload_adjust_reg_for_mode): Likewise.
6355 * reload1.c (find_reload_regs): Likewise.
6356 (forget_old_reloads_1): Likewise.
6357 (reload_reg_free_for_value_p): Likewise.
6358 (reload_adjust_reg_for_temp): Likewise.
6359 (emit_reload_insns): Likewise.
6360 (delete_output_reload): Likewise.
6361 * sel-sched.c (choose_best_reg_1): Likewise.
6362 (choose_best_pseudo_reg): Likewise.
6364 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6365 Alan Hayward <alan.hayward@arm.com>
6366 David Sherwood <david.sherwood@arm.com>
6368 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
6369 * target.def (slow_unaligned_access): New hook.
6370 * targhooks.h (default_slow_unaligned_access): Declare.
6371 * targhooks.c (default_slow_unaligned_access): New function.
6372 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
6373 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
6374 * doc/tm.texi: Regenerate.
6375 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
6376 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
6377 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
6379 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
6380 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
6382 (rs6000_slow_unaligned_access): New function.
6383 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
6384 (expand_block_compare): Likewise.
6385 (expand_strn_compare): Likewise.
6386 (rs6000_rtx_costs): Likewise.
6387 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
6388 (riscv_slow_unaligned_access): Likewise.
6389 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
6390 (riscv_slow_unaligned_access_p): ...this and make static.
6391 (riscv_option_override): Update accordingly.
6392 (riscv_slow_unaligned_access): New function.
6393 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
6394 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
6395 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
6396 (rs6000_slow_unaligned_access): New function.
6397 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
6398 (rs6000_rtx_costs): Likewise.
6399 * config/rs6000/rs6000-string.c (expand_block_compare)
6400 (expand_strn_compare): Use targetm.slow_unaligned_access instead
6401 of SLOW_UNALIGNED_ACCESS.
6402 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
6403 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
6404 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
6405 of SLOW_UNALIGNED_ACCESS.
6406 * expmed.c (simple_mem_bitfield_p): Likewise.
6407 * expr.c (alignment_for_piecewise_move): Likewise.
6408 (emit_group_load_1): Likewise.
6409 (emit_group_store): Likewise.
6410 (copy_blkmode_from_reg): Likewise.
6411 (emit_push_insn): Likewise.
6412 (expand_assignment): Likewise.
6413 (store_field): Likewise.
6414 (expand_expr_real_1): Likewise.
6415 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
6416 * lra-constraints.c (simplify_operand_subreg): Likewise.
6417 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
6418 * gimple-ssa-store-merging.c: Likewise in block comment at start
6420 * tree-ssa-strlen.c: Include target.h.
6421 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
6422 of SLOW_UNALIGNED_ACCESS.
6423 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
6425 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6427 PR rtl-optimization/82185
6428 * expmed.c (emit_store_flag_int): Only test tem if it has been
6431 2017-09-12 Richard Biener <rguenther@suse.de>
6434 * match.pd ((FTYPE) N CMP CST): Fix typo.
6436 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
6438 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
6440 (mips_near_type_p): Add 'short_call' attribute as a synonym
6442 * doc/extend.texi (short_call): Document new function attribute.
6444 2017-09-12 Jakub Jelinek <jakub@redhat.com>
6447 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
6448 assertion check that in the condition.
6449 (get_atomic_generic_size): Likewise. Before testing if parameter
6450 has pointer type, if it has array type, call for C++
6451 default_conversion to perform array-to-pointer conversion.
6453 2017-09-12 Richard Biener <rguenther@suse.de>
6455 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
6456 for operations we cannot scalarize.
6458 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
6460 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
6461 vectors heap vectors. Clean up comments.
6462 Make visited_bbs a reference.
6463 (profitable_jump_thread_path): Make GC
6464 vectors heap vectors. Clean up comments.
6466 (convert_and_register_jump_thread_path): Make GC vectors heap
6468 (check_subpath_and_update_thread_path): Same. Clean up comments.
6469 Make visited_bbs a reference.
6470 (handle_phi): Abstract common code to to
6471 register_jump_thread_path_if_profitable.
6472 Rename VAR_BB to DEF_BB.
6474 Make GC vectors heap vectors.
6475 Make visited_bbs a reference.
6476 (handle_assignment): Same.
6477 (register_jump_thread_path_if_profitable): New.
6478 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
6480 Make GC vectors heap vectors. Clean up comments.
6481 Make visited_bbs a reference.
6482 (find_jump_threads_backwards): Make visited_bbs live in the stack.
6483 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
6486 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
6489 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
6490 words of E_DImode object are reachable by xtensa_uimm8x4 access.
6492 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
6494 Revert r251800 and r251799.
6496 2017-09-11 Martin Jambor <mjambor@suse.cz>
6499 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
6500 arguments in advance.
6501 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
6502 use it to find predecessor edges.
6503 (naive_outof_ssa): Collect vector of predecessors.
6505 2017-09-08 Jason Merrill <jason@redhat.com>
6507 PR c++/70029 - ICE with ref-qualifier and -flto
6508 * langhooks.h (struct lang_hooks_for_types): Add
6509 copy_lang_qualifiers.
6510 * attribs.c (build_type_attribute_qual_variant): Use it.
6511 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
6513 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
6514 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
6516 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
6519 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
6520 (*mulsi3_sp64): New instruction.
6521 (mulsi3): New expander.
6523 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
6525 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
6527 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6529 * sancov.c: Include memmodel.h.
6531 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
6534 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
6537 2017-09-07 Carl Love <cel@us.ibm.com>
6539 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
6540 the sldi instruction.
6542 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
6544 * sancov.c: Include tm_p.h.
6546 2017-09-07 Jakub Jelinek <jakub@redhat.com>
6549 * output.h (switch_to_other_text_partition): New declaration.
6550 * varasm.c (switch_to_other_text_partition): New function.
6551 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
6552 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
6553 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
6554 to the other text partition before emitting LCL label and switch back
6555 after emitting the word after it.
6557 2017-09-07 Richard Biener <rguenther@suse.de>
6559 * passes.def (pass_split_crit_edges): Remove instance before PRE.
6560 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
6561 critical edges here, after loop init.
6562 (pass_data_pre): Remove PROP_no_crit_edges flags.
6563 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
6564 for valueization of call args to avoid leaking VN_TOP.
6565 (visit_use): Assert we do not visit default defs.
6566 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
6567 Use error_mark_node to more easily detect leaking VN_TOP.
6568 All default-defs are varying, not VN_TOP. Mark them visited.
6569 (run_scc_vn): Make code match comment.
6571 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
6573 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
6574 OPTION_MASK_FLOAT128_KEYWORD.
6575 (POWERPC_MASKS): Likewise.
6576 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
6577 support for the -mfloat128-type option, and make -mfloat128
6578 default on PowerPC Linux systems. Define or undefine
6579 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
6580 Define __float128 to be __ieee128 if IEEE 128-bit support is
6581 enabled, or undefine it.
6582 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
6583 Delete defining __FLOAT128_TYPE__.
6584 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
6585 -mfloat128-type option and make -mfloat128 default on PowerPC
6587 (TARGET_FLOAT128_TYPE): Likewise.
6588 (-mfloat128-type): Likewise.
6589 * config/rs6000/rs6000.c (rs6000_option_override_internal):
6590 Delete the -mfloat128-type option and make -mfloat128 default on
6591 PowerPC Linux systems. Always use __ieee128 to be the keyword for
6592 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
6593 128-bit floating point is enabled. Change tests from using
6594 -mfloat128-type to -mfloat128.
6595 (rs6000_mangle_type): Use the correct mangling for the __float128
6596 type even if normal long double is restricted to 64-bits.
6597 (floatn_mode): Enable the _Float128 type by default on VSX Linux
6599 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
6600 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
6601 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
6603 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
6604 documentation for -mfloat128.
6606 2017-09-06 Olivier Hainque <hainque@adacore.com>
6608 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
6610 2017-09-06 Wish Wu <wishwu007@gmail.com>
6611 Jakub Jelinek <jakub@redhat.com>
6613 * asan.c (initialize_sanitizer_builtins): Add
6614 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
6615 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
6616 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
6617 BT_FN_VOID_UINT64_PTR variables.
6618 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
6619 (BT_FN_VOID_UINT16_UINT16): Likewise.
6620 (BT_FN_VOID_UINT32_UINT32): Likewise.
6621 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
6622 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
6623 (BT_FN_VOID_UINT64_PTR): Likewise.
6624 * common.opt (flag_sanitize_coverage): New variable.
6625 (fsanitize-coverage=trace-pc): Remove.
6626 (fsanitize-coverage=): Add.
6627 * flag-types.h (enum sanitize_coverage_code): New enum.
6628 * fold-const.c (fold_range_test): Disable non-short-circuit
6629 optimization if flag_sanitize_coverage.
6630 (fold_truth_andor): Likewise.
6631 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
6632 * opts.c (COVERAGE_SANITIZER_OPT): Define.
6633 (coverage_sanitizer_opts): New array.
6634 (get_closest_sanitizer_option): Add OPTS argument, handle also
6635 OPT_fsanitize_coverage_.
6636 (parse_sanitizer_options): Adjusted to also handle
6637 OPT_fsanitize_coverage_.
6638 (common_handle_option): Add OPT_fsanitize_coverage_.
6639 * sancov.c (instrument_comparison, instrument_switch): New function.
6640 (sancov_pass): Add trace-cmp support.
6641 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
6642 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
6643 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
6644 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
6645 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
6646 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
6647 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
6648 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
6649 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
6650 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
6652 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
6654 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
6655 error. Only quit immediately if parsing is complete.
6656 (BEGIN): Initialize fatal_err and parse_done.
6657 (begin fpu, end fpu): Check number of arguments.
6658 (begin arch, end arch): Likewise.
6659 (begin cpu, end cpu): Likewise.
6660 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
6661 (optalias): Likewise.
6663 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
6665 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
6666 * config/arm/arm-isa.h: Delete. Move definitions to ...
6667 * arm-cpus.in: ... here. Use new feature and fgroup values.
6668 * config/arm/arm.c (arm_option_override): Use lower case for feature
6670 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
6671 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
6672 * config/arm/parsecpu.awk (END): Add new command 'isa'.
6674 (print_isa_bits_for): New function.
6675 (gen_isa): New function.
6676 (gen_comm_data): Use print_isa_bits_for.
6677 (define feature): New keyword.
6678 (define fgroup): New keyword.
6679 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
6680 (arm-isa.h): Add rule to generate file.
6681 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
6682 case for feature bit names.
6684 2017-09-06 Richard Biener <rguenther@suse.de>
6686 * tree-ssa-pre.c (NECESSARY): Remove.
6687 (create_expression_by_pieces): Do not touch pass-local flags.
6688 (insert_into_preds_of_block): Likewise.
6689 (do_pre_regular_insertion): Likewise.
6690 (eliminate_insert): Likewise.
6691 (eliminate_dom_walker::before_dom_children): Likewise.
6692 (fini_eliminate): Do not look at inserted_exprs.
6693 (mark_operand_necessary): Remove.
6694 (remove_dead_inserted_code): Replace with simple work-list
6695 algorithm based on inserted_exprs and SSA uses.
6696 (pass_pre::execute): Re-order fini_eliminate and
6697 remove_dead_inserted_code.
6699 2017-09-06 Olivier Hainque <hainque@adacore.com>
6701 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
6702 for VxWorks 7. Adjust surrounding comments.
6704 2017-09-06 Richard Biener <rguenther@suse.de>
6706 * gimple-ssa-strength-reduction.c
6707 (find_candidates_dom_walker::before_dom_children): Also allow
6710 2017-09-06 Richard Biener <rguenther@suse.de>
6712 PR tree-optimization/82108
6713 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
6714 for gap in the non-permutation SLP case.
6716 2017-09-06 Martin Jambor <mjambor@suse.cz>
6718 PR tree-optimization/82078
6719 * tree-sra.c (sort_and_splice_var_accesses): Move call to
6720 add_access_to_work_queue...
6721 (build_accesses_from_assign): ...here.
6722 (propagate_all_subaccesses): Make sure racc is the group
6723 representative, if there is one.
6725 2017-09-06 Jakub Jelinek <jakub@redhat.com>
6728 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
6731 2017-09-06 Richard Biener <rguenther@suse.de>
6733 * gimple-ssa-strength-reduction.c
6734 (find_candidates_dom_walker::before_doom_children): Use a
6735 type and not a mode check.
6737 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
6740 * config/arm/predicates.md (arm_general_adddi_operand): Create new
6742 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
6744 2017-09-05 Jeff Law <law@redhat.com>
6746 PR tree-optimization/64910
6747 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
6748 cases where we have 3 or more operands.
6750 2017-09-05 Jakub Jelinek <jakub@redhat.com>
6753 * omp-low.c (lower_omp_for): Recompute tree invariant if
6754 gimple_omp_for_initial/final is ADDR_EXPR.
6757 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
6758 into gimple val before gimplification fo the COND_EXPR.
6760 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
6762 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
6763 REGION_COPY argument.
6764 (thread_through_all_blocks): Remove unused argument to
6765 duplicate_thread_path.
6767 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6768 Alan Hayward <alan.hayward@arm.com>
6769 David Sherwood <david.sherwood@arm.com>
6771 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
6772 Take a scalar_mode rather than a machine_mode.
6773 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
6774 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
6775 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
6776 (aarch64_gen_adjusted_ldpstp): Likewise.
6777 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
6779 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6780 Alan Hayward <alan.hayward@arm.com>
6781 David Sherwood <david.sherwood@arm.com>
6783 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
6784 Take a scalar_int_mode instead of a machine_mode.
6785 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
6786 (aarch64_output_scalar_simd_mov_immediate): Likewise.
6787 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
6788 (aarch64_simd_attr_length_rglist): Delete.
6789 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
6790 a scalar_int_mode instead of a machine_mode.
6791 (aarch64_add_offset): Likewise.
6792 (aarch64_internal_mov_immediate): Likewise
6793 (aarch64_add_constant_internal): Likewise.
6794 (aarch64_add_constant): Likewise.
6795 (aarch64_movw_imm): Likewise.
6796 (aarch64_rtx_arith_op_extract_p): Likewise.
6797 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
6798 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
6799 Remove assert that the mode isn't a vector.
6800 (aarch64_output_scalar_simd_mov_immediate): Likewise.
6801 (aarch64_expand_mov_immediate): Update calls after above changes.
6802 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
6803 (aarch64_and_bitmask_imm): Check for scalar integer modes.
6804 (aarch64_move_imm): Likewise.
6805 (aarch64_can_const_movi_rtx_p): Likewise.
6806 (aarch64_strip_extend): Likewise.
6807 (aarch64_extr_rtx_p): Likewise.
6808 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
6809 a CONST_INT when the mode parameter is VOIDmode.
6810 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
6812 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6814 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
6815 * stor-layout.c (bitwise_mode_for_mode): Likewise.
6816 (bitwise_type_for_mode): Update accordingly.
6818 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6820 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
6821 * stor-layout.c (mode_for_size_tree): Likewise.
6822 (mode_for_array): Update accordingly.
6823 (layout_decl): Likewise.
6824 (compute_record_mode): Likewise. Only set the mode once.
6826 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6828 * target.def (get_mask_mode): Change return type to opt_mode.
6830 * doc/tm.texi: Regenerate.
6831 * targhooks.h (default_get_mask_mode): Return an opt_mode.
6832 * targhooks.c (default_get_mask_mode): Likewise.
6833 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
6834 * optabs-query.c (can_vec_mask_load_store_p): Update use of
6835 targetm.get_mask_mode.
6836 * tree.c (build_truth_vector_type): Likewise.
6838 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6840 * machmode.h (mode_for_vector): Return an opt_mode.
6841 * stor-layout.c (mode_for_vector): Likewise.
6842 (mode_for_int_vector): Update accordingly.
6843 (layout_type): Likewise.
6844 * config/i386/i386.c (emit_memmov): Likewise.
6845 (ix86_expand_set_or_movmem): Likewise.
6846 (ix86_expand_vector_init): Likewise.
6847 (ix86_get_mask_mode): Likewise.
6848 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
6850 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
6851 * expmed.c (extract_bit_field_1): Likewise.
6852 * expr.c (expand_expr_real_2): Likewise.
6853 * optabs-query.c (can_vec_perm_p): Likewise.
6854 (can_vec_mask_load_store_p): Likewise.
6855 * optabs.c (expand_vec_perm): Likewise.
6856 * targhooks.c (default_get_mask_mode): Likewise.
6857 * tree-vect-stmts.c (vectorizable_store): Likewise.
6858 (vectorizable_load): Likewise.
6859 (get_vectype_for_scalar_type_and_size): Likewise.
6861 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6863 * machmode.h (mode_for_int_vector): New function.
6864 * stor-layout.c (mode_for_int_vector): Likewise.
6865 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
6866 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
6867 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
6868 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
6869 (s390_expand_vcond): Likewise.
6871 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6873 * machmode.h (opt_machine_mode): New type.
6874 (opt_mode<T>): Allow construction from anything that can be
6876 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
6877 (mode_for_size): Return an opt_machine_mode.
6878 * stor-layout.c (mode_for_size): Likewise.
6879 (mode_for_size_tree): Update call accordingly.
6880 (bitwise_mode_for_mode): Likewise.
6881 (make_fract_type): Likewise.
6882 (make_accum_type): Likewise.
6883 * caller-save.c (replace_reg_with_saved_mem): Update call
6885 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
6886 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
6887 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
6888 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
6889 * expmed.c (extract_bit_field_1): Likewise.
6890 * reload.c (get_secondary_mem): Likewise.
6891 * varasm.c (assemble_integer): Likewise.
6892 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
6895 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6897 * machmode.h (decimal_float_mode_for_size): New function.
6898 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
6899 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
6900 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
6901 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
6902 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
6903 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
6905 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6907 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
6908 (get_builtin_sync_mode): Likewise.
6909 (expand_ifn_atomic_compare_exchange): Likewise.
6910 (expand_builtin_atomic_clear): Likewise.
6911 (expand_builtin_atomic_test_and_set): Likewise.
6912 (fold_builtin_atomic_always_lock_free): Likewise.
6913 * calls.c (compute_argument_addresses): Likewise.
6914 (emit_library_call_value_1): Likewise.
6915 (store_one_arg): Likewise.
6916 * combine.c (combine_instructions): Likewise.
6917 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
6918 * config/arm/arm.c (arm_function_value): Likewise.
6919 (aapcs_allocate_return_reg): Likewise.
6920 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
6921 * config/i386/i386.c (construct_container): Likewise.
6922 (ix86_gimplify_va_arg): Likewise.
6923 (ix86_expand_sse_cmp): Likewise.
6924 (emit_memmov): Likewise.
6925 (emit_memset): Likewise.
6926 (expand_small_movmem_or_setmem): Likewise.
6927 (ix86_expand_pextr): Likewise.
6928 (ix86_expand_pinsr): Likewise.
6929 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
6930 * config/microblaze/microblaze.c (microblaze_block_move_straight):
6932 * config/mips/mips.c (mips_function_value_1) Likewise.
6933 (mips_block_move_straight): Likewise.
6934 (mips_expand_ins_as_unaligned_store): Likewise.
6935 * config/powerpcspe/powerpcspe.c
6936 (rs6000_darwin64_record_arg_advance_flush): Likewise.
6937 (rs6000_darwin64_record_arg_flush): Likewise.
6938 * config/rs6000/rs6000.c
6939 (rs6000_darwin64_record_arg_advance_flush): Likewise.
6940 (rs6000_darwin64_record_arg_flush): Likewise.
6941 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
6942 (sparc_function_value_1): Likewise.
6943 * config/spu/spu.c (adjust_operand): Likewise.
6944 (spu_emit_branch_or_set): Likewise.
6945 (arith_immediate_p): Likewise.
6946 * emit-rtl.c (gen_lowpart_common): Likewise.
6947 * expr.c (expand_expr_real_1): Likewise.
6948 * function.c (assign_parm_setup_block): Likewise.
6949 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
6950 * reload1.c (alter_reg): Likewise.
6951 * stor-layout.c (mode_for_vector): Likewise.
6952 (layout_type): Likewise.
6954 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6956 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
6957 (spu_convert_move): Likewise.
6958 * lower-subreg.c (resolve_simple_move): Likewise.
6960 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6963 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
6964 define_insn to a define_expand.
6965 (altivec_vsum2sws_direct): New define_insn.
6966 (altivec_vsumsws): Convert from a define_insn to a define_expand.
6968 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
6970 * config/arm/arm.c (arm_option_params_internal): Improve setting of
6973 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
6978 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
6979 realign stack if stack alignment needed is less than incoming
6982 2017-09-05 Marek Polacek <polacek@redhat.com>
6985 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
6988 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
6990 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
6992 2017-09-05 Richard Biener <rguenther@suse.de>
6994 PR tree-optimization/82084
6995 * fold-const.c (can_native_encode_string_p): Handle wide characters.
6997 2017-09-05 Richard Biener <rguenther@suse.de>
6999 PR tree-optimization/82102
7000 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
7002 2017-09-05 Martin Liska <mliska@suse.cz>
7004 PR tree-optimization/82032
7005 * tree-cfg.c (generate_range_test): New function.
7006 * tree-cfg.h (generate_range_test): Declared here.
7007 * tree-cfgcleanup.c (convert_single_case_switch): New function.
7008 (cleanup_control_expr_graph): Use it.
7009 * tree-switch-conversion.c (try_switch_expansion): Remove
7011 (emit_case_nodes): Use generate_range_test.
7013 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
7016 * config/i386/i386.md (*<btsc><mode>_mask): Add
7017 TARGET_USE_BT to insn constraint.
7018 (*btr<mode>_mask): Ditto.
7020 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
7022 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
7023 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
7025 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
7028 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
7029 TARGET_NEON and TARGET_IWMMXT.
7030 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
7031 TARGET_NEON and TARGET_IWMMXT.
7032 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
7034 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
7036 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
7037 (ix86_rewrite_tls_address): Ditto.
7038 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
7039 (ix86_rewrite_tls_address_1): Ditto.
7040 (ix86_rewrite_tls_address): Ditto.
7041 * config/i386/predicates.md (tls_address_pattern): New predicate.
7042 * config/i386/i386.md (TLS address splitter): New splitter.
7044 2017-09-04 Richard Biener <rguenther@suse.de>
7046 PR tree-optimization/82084
7047 * fold-const.h (can_native_encode_string_p): Declare.
7048 * fold-const.c (can_native_encode_string_p): Factor out from ...
7049 (native_encode_string): ... here.
7050 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
7051 vectorizing stores from constants we later cannot handle.
7053 2017-09-04 Marek Polacek <polacek@redhat.com>
7056 * doc/invoke.texi: Update -Wtautological-compare documentation.
7058 2017-09-04 Jeff Law <law@redhat.com>
7060 PR tree-optimization/64910
7061 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
7062 swap the first and last operand if the last is a constant.
7064 2017-09-04 Marek Polacek <polacek@redhat.com>
7067 * convert.c (do_narrow): When sanitizing signed integer overflows,
7068 bail out for signed types.
7069 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
7071 2017-09-04 Richard Biener <rguenther@suse.de>
7073 PR tree-optimization/82060
7074 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
7075 Move devirtualization after stmt folding and before EH/AB/noreturn
7076 cleanup to get the stmt refs canonicalized. Use a bool instead
7077 of gimple_modified_p since that doesn't work for NOPs. Schedule
7078 NOPs generated by folding for removal.
7080 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7081 Alan Hayward <alan.hayward@arm.com>
7082 David Sherwood <david.sherwood@arm.com>
7084 * coretypes.h (pad_direction): New enum.
7085 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
7086 (FUNCTION_ARG_PADDING): Likewise.
7087 * target.def (function_arg_padding): New hook.
7088 * targhooks.h (default_function_arg_padding): Declare.
7089 * targhooks.c (default_function_arg_padding): New function.
7090 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
7091 (TARGET_FUNCTION_ARG_PADDING): ...this.
7092 * doc/tm.texi: Regenerate.
7093 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
7094 instead of direction.
7095 (compute_argument_addresses): Likewise.
7096 (load_register_parameters): Likewise.
7097 (emit_library_call_value_1): Likewise.
7098 (store_one_arg): Use targetm.calls.function_arg_padding instead
7099 of FUNCTION_ARG_PADDING.
7100 (must_pass_in_stack_var_size_or_pad): Likewise.
7101 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
7102 (emit_group_store): Likewise.
7103 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
7104 instead of FUNCTION_ARG_PADDING.
7105 (emit_push_insn): Likewise, and propagate enum change throughout
7107 * function.h (direction): Delete.
7108 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
7110 * function.c (assign_parm_find_stack_rtl): Likewise.
7111 (assign_parm_setup_block_p): Likewise.
7112 (assign_parm_setup_block): Likewise.
7113 (gimplify_parameters): Likewise.
7114 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
7115 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
7117 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
7118 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
7119 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
7120 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
7121 (aarch64_function_arg_padding): ...this new function.
7122 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
7123 (TARGET_FUNCTION_ARG_PADDING): Redefine.
7124 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
7125 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
7126 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
7127 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7128 (arm_pad_arg_upward): Replace with...
7129 (arm_function_arg_padding): ...this new function.
7130 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
7132 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
7133 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
7134 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7135 (ia64_hpux_function_arg_padding): Replace with...
7136 (ia64_function_arg_padding): ...this new function. Use pad_direction
7137 instead of direction. Check for TARGET_HPUX.
7138 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
7139 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7140 (iq2000_function_arg_padding): New function.
7141 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
7142 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
7143 (mips_function_arg_padding): ...this new function.
7144 (mips_pad_reg_upward): Update accordingly.
7145 (TARGET_FUNCTION_ARG_PADDING): Redefine.
7146 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
7147 targetm.calls.function_arg_padding.
7148 (FUNCTION_ARG_PADDING): Delete.
7149 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
7150 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
7151 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
7152 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
7153 (nios2_block_reg_padding): Return pad_direction instead of direction.
7154 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
7155 instead of direction.
7156 (nios2_function_arg_padding): Likewise. Make static.
7157 (TARGET_FUNCTION_ARG_PADDING): Redefine.
7158 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
7159 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
7160 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
7161 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7162 (pa_function_arg_padding): Make static. Return pad_direction instead
7164 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
7165 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
7166 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
7167 instead of direction. Use targetm.calls.function_arg_padding.
7168 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
7169 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
7170 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
7171 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
7172 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
7174 (function_arg_padding): Rename to...
7175 (rs6000_function_arg_padding): ...this. Make static. Return
7176 pad_direction instead of direction.
7177 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
7178 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
7179 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
7180 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
7181 instead of direction. Use targetm.calls.function_arg_padding.
7182 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
7183 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
7184 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
7185 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
7186 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7187 (function_arg_padding): Rename to...
7188 (rs6000_function_arg_padding): ...this. Make static. Return
7189 pad_direction instead of direction.
7190 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
7191 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
7192 * config/s390/s390.c (s390_function_arg_padding): New function.
7193 (TARGET_FUNCTION_ARG_PADDING): Redefine.
7194 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
7195 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
7196 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7197 (function_arg_padding): Rename to...
7198 (sparc_function_arg_padding): ...this. Make static. Return
7199 pad_direction instead of direction.
7200 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
7201 * config/spu/spu.c (spu_function_arg_padding): New function.
7202 (TARGET_FUNCTION_ARG_PADDING): Redefine.
7203 * system.h (FUNCTION_ARG_PADDING): Poison.
7205 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7206 Alan Hayward <alan.hayward@arm.com>
7207 David Sherwood <david.sherwood@arm.com>
7209 * target.def (modes_tieable_p): New hook.
7210 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
7211 (TARGET_MODES_TIEABLE_P): ...this.
7212 * doc/tm.texi.in: Regenerate.
7213 * hooks.h (hook_bool_mode_mode_true): Declare.
7214 * hooks.c (hook_bool_mode_mode_true): New function.
7215 * combine.c (subst): Use targetm.modes_tieable_p instead of
7217 * dse.c (find_shift_sequence): Likewise.
7218 * expmed.c (extract_low_bits): Likewise.
7219 * lower-subreg.c: Include target.h.
7220 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
7222 * rtlanal.c (rtx_cost): Likewise.
7223 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
7224 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
7225 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
7226 (TARGET_MODES_TIEABLE_P): Redefine.
7227 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
7228 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
7229 (TARGET_MODES_TIEABLE_P): Redefine.
7230 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
7231 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
7232 (arc_modes_tieable_p): New function.
7233 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
7234 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
7235 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
7236 (arm_modes_tieable_p): Make static.
7237 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
7238 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
7239 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
7240 (TARGET_MODES_TIEABLE_P): Redefine.
7241 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
7242 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
7243 (TARGET_MODES_TIEABLE_P): Redefine.
7244 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
7245 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
7246 (cr16_modes_tieable_p): New function.
7247 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
7248 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
7249 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
7250 (TRULY_NOOP_TRUNCATION): Update comment.
7251 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
7252 (TRULY_NOOP_TRUNCATION): Update comment.
7253 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
7254 (frv_modes_tieable_p): New function.
7255 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
7256 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
7257 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
7258 (TARGET_MODES_TIEABLE_P): Redefine.
7259 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
7260 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
7261 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
7262 (TARGET_MODES_TIEABLE_P): Redefine.
7263 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
7264 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
7265 (ia64_modes_tieable_p): New function.
7266 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
7267 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
7268 (iq2000_modes_tieable_p): New function.
7269 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
7270 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
7271 (lm32_modes_tieable_p): New function.
7272 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
7273 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
7274 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
7275 (TARGET_MODES_TIEABLE_P): Redefine.
7276 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
7277 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
7278 (m32r_modes_tieable_p): New function.
7279 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
7280 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
7281 (m68k_modes_tieable_p): New function.
7282 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
7283 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
7284 (mcore_modes_tieable_p): New function.
7285 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
7286 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
7288 (TARGET_MODES_TIEABLE_P): Redefine.
7289 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
7290 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
7291 * config/mips/mips.c (mips_modes_tieable_p): Make static.
7292 (TARGET_MODES_TIEABLE_P): Redefine.
7293 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
7294 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
7295 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
7296 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
7297 (mn10300_modes_tieable_p): ...this and make static.
7298 (TARGET_MODES_TIEABLE_P): Redefine.
7299 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
7300 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
7301 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
7302 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
7303 (msp430_modes_tieable_p): Make static.
7304 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
7305 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
7306 (TARGET_MODES_TIEABLE_P): Redefine.
7307 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
7308 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
7309 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
7310 (TARGET_MODES_TIEABLE_P): Redefine.
7311 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
7312 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
7313 * config/pa/pa.c (pa_modes_tieable_p): Make static.
7314 (TARGET_MODES_TIEABLE_P): Redefine.
7315 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
7316 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
7317 (pdp11_modes_tieable_p): New function.
7318 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
7319 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
7320 (rs6000_modes_tieable_p): New function.
7321 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
7322 * config/powerpcspe/powerpcspe.md: Update comment.
7323 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
7324 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
7325 (TARGET_MODES_TIEABLE_P): Redefine.
7326 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
7327 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
7328 (rl78_modes_tieable_p): New function.
7329 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
7330 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
7331 (rs6000_modes_tieable_p): New function.
7332 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
7333 * config/rs6000/rs6000.md: Update comment.
7334 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
7335 * config/rx/rx.c (rx_modes_tieable_p): New function.
7336 (TARGET_MODES_TIEABLE_P): Redefine.
7337 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
7338 * config/s390/s390.c (s390_modes_tieable_p): New function.
7339 (TARGET_MODES_TIEABLE_P): Redefine.
7340 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
7341 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
7342 (sh_modes_tieable_p): New function.
7343 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
7344 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
7345 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
7346 (sparc_modes_tieable_p): Make static.
7347 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
7348 * config/spu/spu.c (spu_modes_tieable_p): New function.
7349 (TARGET_MODES_TIEABLE_P): Redefine.
7350 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
7351 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
7352 (TARGET_MODES_TIEABLE_P): Redefine.
7353 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
7354 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
7355 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
7356 * config/v850/v850.c (v850_modes_tieable_p): New function.
7357 (TARGET_MODES_TIEABLE_P): Redefine.
7358 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
7359 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
7360 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
7361 (visium_modes_tieable_p): New function.
7362 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
7363 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
7364 (xtensa_modes_tieable_p): New function.
7365 * system.h (MODES_TIEABLE_P): Poison.
7367 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7368 Alan Hayward <alan.hayward@arm.com>
7369 David Sherwood <david.sherwood@arm.com>
7371 * target.def (hard_regno_mode_ok): New hook.
7372 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
7373 (TARGET_HARD_REGNO_MODE_OK): ...this.
7374 * doc/tm.texi.in: Regenerate.
7375 * hooks.h (hook_bool_uint_mode_true): Declare.
7376 * hooks.c (hook_bool_uint_mode_true): New function.
7377 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
7379 * genpreds.c (write_insn_preds_c): Add an include of target.h.
7380 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
7381 instead of HARD_REGNO_MODE_OK.
7382 * caller-save.c: Include target.h.
7383 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
7385 * combine.c (can_combine_p): Likewise.
7386 (combinable_i3pat): Likewise.
7387 (can_change_dest_mode): Likewise.
7388 * expr.c (init_expr_target): Likewise.
7389 (convert_move): Likewise.
7390 (convert_modes): Likewise.
7391 * ira.c (setup_prohibited_class_mode_regs): Likewise.
7392 (setup_prohibited_mode_move_regs): Likewise.
7393 * ira.h (target_ira): Likewise.
7394 * lra-assigns.c (find_hard_regno_for_1): Likewise.
7395 * lra-constraints.c (process_alt_operands): Likewise.
7396 (split_reg): Likewise.
7397 * recog.c (peep2_find_free_register): Likewise.
7398 * ree.c (combine_reaching_defs): Likewise.
7399 * regcprop.c (maybe_mode_change): Likewise.
7400 * reginfo.c (init_reg_sets_1): Likewise.
7401 (choose_hard_reg_mode): Likewise.
7402 (simplifiable_subregs): Likewise.
7403 * regrename.c (check_new_reg_p): Likewise.
7404 * reload.c (find_valid_class): Likewise.
7405 (find_valid_class_1): Likewise.
7406 (reload_inner_reg_of_subreg): Likewise.
7407 (push_reload): Likewise.
7408 (combine_reloads): Likewise.
7409 (find_dummy_reload): Likewise.
7410 (find_reloads): Likewise.
7411 * reload1.c (find_reg): Likewise.
7412 (set_reload_reg): Likewise.
7413 (allocate_reload_reg): Likewise.
7414 (choose_reload_regs): Likewise.
7415 (reload_adjust_reg_for_temp): Likewise.
7416 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
7417 (simplify_subreg_regno): Likewise.
7418 * sel-sched.c (init_regs_for_mode): Likewise.
7419 * varasm.c (make_decl_rtl): Likewise.
7420 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
7421 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
7423 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
7424 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
7425 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7426 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
7427 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
7428 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7429 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
7430 (arc_mode_class): Delete.
7431 (HARD_REGNO_MODE_OK): Delete.
7432 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7433 (arc_hard_regno_mode_ok): Rename old array to...
7434 (arc_hard_regno_mode_ok_modes): ...this.
7435 (arc_conditional_register_usage): Update accordingly.
7436 (arc_mode_class): Make static.
7437 (arc_hard_regno_mode_ok): New function.
7438 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
7439 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
7440 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7441 (arm_hard_regno_mode_ok): Make static.
7442 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
7444 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
7445 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
7446 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
7448 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7449 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
7450 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
7451 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
7452 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
7453 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7454 * config/bfin/predicates.md (valid_reg_operand): Use
7455 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
7456 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
7457 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
7458 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7459 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
7460 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
7461 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7462 (cr16_hard_regno_mode_ok): Make static and return a bool.
7463 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
7464 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7465 (cris_hard_regno_mode_ok): New function.
7466 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
7467 (epiphany_mode_class): Delete.
7468 (HARD_REGNO_MODE_OK): Delete.
7469 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
7470 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7471 (hard_regno_mode_ok): Rename to...
7472 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
7473 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
7474 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
7476 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
7477 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
7478 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7479 (frv_hard_regno_mode_ok): Make static and return a bool.
7480 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
7482 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
7483 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
7484 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
7485 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
7487 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7488 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
7489 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
7490 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
7492 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7493 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
7494 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7495 (ia64_hard_regno_mode_ok): New function.
7496 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
7497 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7498 (iq2000_hard_regno_mode_ok): New function.
7499 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
7500 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7501 (lm32_hard_regno_mode_ok): New function.
7502 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
7503 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
7504 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
7505 instead of HARD_REGNO_MODE_OK.
7506 (m32c_hard_regno_ok): Rename to...
7507 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
7508 (m32c_cannot_change_mode_class): Update accordingly.
7509 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7510 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
7511 (m32r_mode_class): Delete.
7512 (HARD_REGNO_MODE_OK): Delete.
7513 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7514 (m32r_hard_regno_mode_ok): Rename to...
7515 (m32r_hard_regno_modes): ...this.
7516 (m32r_mode_class): Make static.
7517 (m32r_hard_regno_mode_ok): New function.
7518 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
7519 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
7520 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7521 (m68k_hard_regno_mode_ok): Make static.
7522 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
7523 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7524 (mcore_hard_regno_mode_ok): New function.
7525 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
7526 (HARD_REGNO_MODE_OK): Delete.
7527 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
7529 (microblaze_hard_regno_mode_ok_p): ...this and make static.
7530 (microblaze_hard_regno_mode_ok): New function.
7531 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7532 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
7533 (mips_hard_regno_mode_ok): Delete.
7534 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
7535 (mips_hard_regno_mode_ok_p): ...this and make static.
7536 (mips_hard_regno_mode_ok_p): Rename to...
7537 (mips_hard_regno_mode_ok_uncached): ...this.
7538 (mips_hard_regno_mode_ok): New function.
7539 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
7540 of HARD_REGNO_MODE_OK.
7541 (mips_option_override): Update after above name changes.
7542 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7543 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
7544 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
7545 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
7546 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
7547 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7548 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
7549 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
7550 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
7551 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7552 (msp430_hard_regno_mode_ok): Make static and return a bool.
7553 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
7554 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
7555 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
7557 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7558 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
7559 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
7560 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
7561 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
7562 (PA_HARD_REGNO_MODE_OK): ...this
7563 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
7564 (PA_HARD_REGNO_MODE_OK): ...this.
7565 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7566 (pa_hard_regno_mode_ok): New function.
7567 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
7568 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7569 (pdp11_hard_regno_mode_ok): New function.
7570 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
7571 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
7573 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
7575 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7576 (rs6000_hard_regno_mode_ok): Rename to...
7577 (rs6000_hard_regno_mode_ok_uncached): ...this.
7578 (rs6000_init_hard_regno_mode_ok): Update accordingly.
7579 (rs6000_hard_regno_mode_ok): New function.
7580 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
7581 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
7582 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
7583 (riscv_hard_regno_mode_ok): ...this and make static.
7584 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7585 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
7586 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
7587 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7588 (rl78_hard_regno_mode_ok): Make static and return bool.
7589 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
7590 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
7592 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
7593 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7594 (rs6000_hard_regno_mode_ok): Rename to...
7595 (rs6000_hard_regno_mode_ok_uncached): ...this.
7596 (rs6000_init_hard_regno_mode_ok): Update accordingly.
7597 (rs6000_hard_regno_mode_ok): New function.
7598 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
7599 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
7600 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7601 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
7602 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
7603 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
7604 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7605 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
7606 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
7607 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7608 (sh_hard_regno_mode_ok): Make static.
7609 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
7610 instead of HARD_REGNO_MODE_OK.
7611 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
7612 (sparc_mode_class): Delete.
7613 (HARD_REGNO_MODE_OK): Delete.
7614 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7615 (hard_regno_mode_classes): Make static.
7616 (sparc_mode_class): Likewise.
7617 (sparc_hard_regno_mode_ok): New function.
7618 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
7619 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
7620 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
7622 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7623 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
7624 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
7625 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
7626 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
7627 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7628 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
7629 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
7630 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7631 (visium_hard_regno_mode_ok): New function.
7632 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
7633 instead of HARD_REGNO_MODE_OK.
7634 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
7635 (HARD_REGNO_MODE_OK): Delete.
7636 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
7637 (xtensa_hard_regno_mode_ok_p): ...this and make static.
7638 (xtensa_option_override): Update accordingly.
7639 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7640 (xtensa_hard_regno_mode_ok): New function.
7641 * system.h (HARD_REGNO_MODE_OK): Poison.
7643 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7644 Alan Hayward <alan.hayward@arm.com>
7645 David Sherwood <david.sherwood@arm.com>
7647 * target.def (hard_regno_call_part_clobbered): New hook.
7648 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
7649 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
7650 * doc/tm.texi: Regenerate.
7651 * hooks.h (hook_bool_uint_mode_false): Declare.
7652 * hooks.c (hook_bool_uint_mode_false): New function.
7653 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7654 * cselib.c (cselib_process_insn): Use
7655 targetm.hard_regno_call_part_clobbered instead of
7656 HARD_REGNO_CALL_PART_CLOBBERED.
7657 * ira-conflicts.c (ira_build_conflicts): Likewise.
7658 * ira-costs.c (ira_tune_allocno_costs): Likewise.
7659 * lra-constraints.c (need_for_call_save_p): Likewise.
7660 * lra-lives.c: Include target.h.
7661 (check_pseudos_live_through_calls): Use
7662 targetm.hard_regno_call_part_clobbered instead of
7663 HARD_REGNO_CALL_PART_CLOBBERED.
7664 * regcprop.c: Include target.h.
7665 (copyprop_hardreg_forward_1): Use
7666 targetm.hard_regno_call_part_clobbered instead of
7667 HARD_REGNO_CALL_PART_CLOBBERED.
7668 * reginfo.c (choose_hard_reg_mode): Likewise.
7669 * regrename.c (check_new_reg_p): Likewise.
7670 * reload.c (find_equiv_reg): Likewise.
7671 * reload1.c (emit_reload_insns): Likewise.
7672 * sched-deps.c (deps_analyze_insn): Likewise.
7673 * sel-sched.c (init_regs_for_mode): Likewise.
7674 (mark_unavailable_hard_regs): Likewise.
7675 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
7676 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7677 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
7679 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7680 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7681 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
7683 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
7685 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7686 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7687 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
7689 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7690 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7691 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
7693 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7694 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
7696 * config/powerpcspe/powerpcspe.c
7697 (rs6000_hard_regno_call_part_clobbered): New function.
7698 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7699 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7700 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
7702 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7703 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7704 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
7706 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7707 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7708 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
7710 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7711 Alan Hayward <alan.hayward@arm.com>
7712 David Sherwood <david.sherwood@arm.com>
7714 * rtl.h (subreg_memory_offset): Declare.
7715 * emit-rtl.c (subreg_memory_offset): New function.
7716 * expmed.c (store_bit_field_1): Use it.
7717 * expr.c (undefined_operand_subword_p): Likewise.
7718 * simplify-rtx.c (simplify_subreg): Likewise.
7720 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
7722 PR rtl-optimization/57448
7725 * optabs.c (expand_atomic_load): Place compiler memory barriers if
7726 using atomic_load pattern.
7727 (expand_atomic_store): Likewise.
7729 2017-09-04 Jakub Jelinek <jakub@redhat.com>
7732 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
7733 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
7734 handling. Use replace_call_with_value with NULL instead of
7735 gsi_replace, unlink_stmt_vdef and release_defs.
7737 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
7740 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
7742 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7745 * rtl.h (emit_library_call_value_1): Declare.
7746 (emit_library_call): Replace declaration with a series of overloads.
7747 Remove the parameter count argument.
7748 (emit_library_call_value): Likewise.
7749 * calls.c (emit_library_call_value_1): Make global. Replace varargs
7750 with an "rtx_mode_t *".
7751 (emit_library_call_value): Delete.
7752 (emit_library_call): Likewise.
7753 * asan.c (asan_emit_stack_protection): Update calls accordingly.
7754 (asan_emit_allocas_unpoison): Likewise.
7755 * builtins.c (expand_builtin_powi): Likewise.
7756 (expand_asan_emit_allocas_unpoison): Likewise.
7757 * cfgexpand.c (expand_main_function): Likewise.
7758 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
7759 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
7760 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
7761 * config/arm/arm.c (arm_trampoline_init): Likewise.
7762 (arm_call_tls_get_addr): Likewise.
7763 (arm_expand_divmod_libfunc): Likewise.
7764 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
7765 (smulsi3_highpart): Likewise.
7766 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
7767 (c6x_expand_compare): Likewise.
7768 (c6x_expand_movmem): Likewise.
7769 * config/frv/frv.c (frv_trampoline_init): Likewise.
7770 * config/i386/i386.c (ix86_trampoline_init): Likewise.
7771 (ix86_expand_divmod_libfunc): Likewise.
7772 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
7773 (ia64_expand_compare): Likewise.
7774 (ia64_profile_hook): Likewise.
7775 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
7776 (nonlocal_goto): Likewise.
7777 (restore_stack_nonlocal): Likewise.
7778 * config/m32r/m32r.c (block_move_call): Likewise.
7779 (m32r_trampoline_init): Likewise.
7780 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
7781 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
7782 (m68k_call_m68k_read_tp): Likewise.
7783 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
7784 (microblaze_expand_divide): Likewise.
7785 * config/mips/mips.h (mips_args): Likewise.
7786 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
7787 (MIPS_ICACHE_SYNC): Likewise.
7788 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
7789 (nios2_trampoline_init): Likewise.
7790 * config/pa/pa.c (hppa_tls_call): Likewise.
7791 (pa_trampoline_init): Likewise.
7792 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
7793 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
7794 (expand_strn_compare): Likewise.
7795 (rs6000_generate_compare): Likewise.
7796 (rs6000_expand_float128_convert): Likewise.
7797 (output_profile_hook): Likewise.
7798 (rs6000_trampoline_init): Likewise.
7799 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
7800 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
7801 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
7802 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
7803 (rs6000_generate_compare): Likewise.
7804 (rs6000_expand_float128_convert): Likewise.
7805 (output_profile_hook): Likewise.
7806 (rs6000_trampoline_init): Likewise.
7807 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
7808 * config/sh/sh.c (sh_trampoline_init): Likewise.
7809 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
7810 (sparc_emit_float_lib_cmp): Likewise.
7811 (sparc32_initialize_trampoline): Likewise.
7812 (sparc64_initialize_trampoline): Likewise.
7813 (sparc_profile_hook): Likewise.
7814 * config/spu/spu.c (ea_load_store): Likewise.
7815 * config/spu/spu.md (floatunssidf2): Likewise.
7816 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
7817 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
7818 * config/visium/visium.c (expand_block_move_4): Likewise.
7819 (expand_block_move_2): Likewise.
7820 (expand_block_move_1): Likewise.
7821 (expand_block_set_4): Likewise.
7822 (expand_block_set_2): Likewise.
7823 (expand_block_set_1): Likewise.
7824 (visium_trampoline_init): Likewise.
7825 (visium_profile_hook): Likewise.
7826 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
7827 (xtensa_setup_frame_addresses): Likewise.
7828 (xtensa_trampoline_init): Likewise.
7829 * except.c (sjlj_emit_function_enter): Likewise.
7830 (sjlj_emit_function_exit): Likewise.
7831 * explow.c (allocate_dynamic_stack_space): Likewise.
7832 (probe_stack_range): Likewise.
7833 * expr.c (convert_mode_scalar): Likewise.
7834 * optabs.c (expand_binop): Likewise.
7835 (expand_twoval_binop_libfunc): Likewise.
7836 (expand_unop): Likewise.
7837 (prepare_cmp_insn): Likewise.
7838 (prepare_float_lib_cmp): Likewise.
7839 (expand_float): Likewise.
7840 (expand_fix): Likewise.
7841 (expand_fixed_convert): Likewise.
7842 (maybe_emit_sync_lock_test_and_set): Likewise.
7843 (expand_atomic_compare_and_swap): Likewise.
7844 (expand_mem_thread_fence): Likewise.
7845 (expand_atomic_fetch_op): Likewise.
7847 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
7849 * doc/generic.texi (OpenACC): Adjust URL.
7850 * doc/invoke.texi (C Dialect Options): Ditto.
7852 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
7854 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
7855 predicate for operand 1. Add (m,<S>) constraint.
7856 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
7857 Prevent memory operand 1 with register operand 2.
7859 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
7861 PR rtl-optimization/82024
7862 * combine.c (try_combine): If the combination result is a PARALLEL,
7863 and we only need to retain the SET in there that would be placed
7864 at I2, check that we can place that at I3 instead, before doing so.
7866 2017-09-01 Jakub Jelinek <jakub@redhat.com>
7869 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
7871 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
7872 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
7875 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
7876 Jeff Law <law@redhat.com>
7878 * varasm.c (bss_initializer_p): Do not put constants into .bss
7879 (categorize_decl_for_section): Handle bss_initializer_p returning
7880 false when DECL_INITIAL is NULL.
7882 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7885 * config/s390/s390.c (s390_can_inline_p): New function.
7887 2017-09-01 Jeff Law <law@redhat.com>
7889 PR tree-optimization/82052
7890 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
7891 Always initialize the returned slot after a hash table miss
7892 when INSERT is true.
7894 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
7896 * config/s390/s390.md (mem_signal_fence): Remove.
7897 * doc/md.texi (mem_signal_fence): Remove.
7898 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
7900 * target-insns.def (mem_signal_fence): Remove.
7902 2017-09-01 Jakub Jelinek <jakub@redhat.com>
7905 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
7908 * asan.c (create_odr_indicator): Strip name encoding from assembler
7909 name before appending it after __odr_asan_.
7911 2017-09-01 Martin Liska <mliska@suse.cz>
7913 PR tree-optimization/82059
7914 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
7915 frequency only when an edge is redirected.
7917 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
7919 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
7920 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
7921 (arc_conditional_register_usage): Remove ARC600 lp_count
7923 (arc_file_start): Emit Tag_ARC_CPU_variation.
7924 (arc_can_use_doloop_p): New conditions to use ZOLs.
7925 (hwloop_fail): New function.
7926 (hwloop_optimize): Likewise.
7927 (hwloop_pattern_reg): Likewise.
7928 (arc_doloop_hooks): New struct, to be used with reorg_loops.
7929 (arc_reorg_loops): New function, calls reorg_loops.
7930 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
7931 (arc600_corereg_hazard): Remove ZOL checking, case handled by
7933 (arc_loop_hazard): Remove function, functionality moved into
7935 (arc_hazard): Remove arc_loop_hazard call.
7936 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
7937 into hwloop_optimize.
7938 (arc_label_align): Remove ZOL handling.
7939 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
7940 * config/arc/arc.md (doloop_begin): Remove pattern.
7941 (doloop_begin_i): Likewise.
7942 (doloop_end_i): Likewise.
7943 (doloop_fallback): Likewise.
7944 (doloop_fallback_m): Likewise.
7945 (doloop_end): Reimplement expand.
7946 (arc_lp): New pattern for LP instruction.
7947 (loop_end): New pattern.
7948 (loop_fail): Likewise.
7949 (decrement_and_branch_until_zero): Likewise.
7950 * config/arc/arc.opt (mlpc-width): New option.
7951 * doc/invoke.texi (mlpc-width): Document option.
7953 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
7955 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
7956 (arc_ccfsm_advance): Fix checking for delay slots.
7957 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
7959 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
7961 * config/arc/arc.md (movqi_insn): Add stores to save constant long
7963 (movhi_insn): Update store instruction constraint which are saving
7964 6-bit short immediates.
7965 (movsi_insn): Consider also short scaled load operations.
7966 (zero_extendhisi2_i): Use Usd constraint instead of T.
7967 (extendhisi2_i): Add q constraint.
7968 (arc_clzsi2): Add type and length attributes.
7969 (arc_ctzsi2): Likewise.
7970 * config/arc/constraints.md (Usc): Update constraint, the
7971 assembler can parse two relocations for a single instruction.
7973 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
7975 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
7976 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
7978 2017-08-31 Olivier Hainque <hainque@adacore.com>
7980 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
7981 match as powerpc-wrs-vxworks*.
7983 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
7985 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
7986 register constraint for by-element operand.
7987 (aarch64_mls_elt_merge<mode>): Likewise.
7989 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
7991 * config/arc/arc.c (arc_can_follow_jump): Check for short
7994 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
7996 * config.gcc: Use g.opt for arc.
7997 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
7998 functionality moved to ...
7999 (legitimate_scaled_address_p): New function, ...here.
8000 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
8001 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
8002 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
8004 (arc_override_options): Handle G option.
8005 (arc_output_pic_addr_const): Correct function definition.
8006 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
8007 (arc_decl_anon_ns_mem_p): Delete.
8008 (arc_in_small_data_p): Overhaul this function to take into
8009 consideration the value given via G option.
8010 (arc_rewrite_small_data_1): Renamed and corrected old
8011 arc_rewrite_small_data function.
8012 (arc_rewrite_small_data): New function.
8013 (small_data_pattern): Don't use pic_offset_table_rtx.
8014 * config/arc/arc.h (CC1_SPEC): Recognize G option.
8015 * config/arc/simdext.md (movmisalignv2hi): Use
8016 prepare_move_operands function.
8018 (movmisalign*): Likewise.
8019 * doc/invoke.texi (ARC options): Document -G option.
8021 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
8023 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
8025 * config/arc/arc.c (arc_print_operand): Output scalled address for
8026 sdata whenever is possible.
8027 (arc_in_small_data_p): Allow sdata for 64bit datum when double
8028 load/stores are available.
8029 (compact_sda_memory_operand): Check for the alignment required by
8030 code density instructions.
8031 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
8033 * config/arc/constraints.md (Usd): Update constraint.
8034 (Us0): New constraint.
8035 (Usc): Update constraint.
8037 2017-08-31 Richard Biener <rguenther@suse.de>
8040 * dwarf2out.c (dwarf2out_early_global_decl): Process each
8043 2017-08-31 Tamar Christina <tamar.christina@arm.com>
8045 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
8046 Resize type_signature.
8048 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
8049 Alan Hayward <alan.hayward@arm.com>
8050 David Sherwood <david.sherwood@arm.com>
8052 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
8053 subregs whose inner modes can be stored in GPRs.
8054 (aarch64_classify_index): Likewise.
8056 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
8057 Alan Hayward <alan.hayward@arm.com>
8058 David Sherwood <david.sherwood@arm.com>
8060 * config/aarch64/iterators.md (V_cmp_result): Rename to...
8061 (V_INT_EQUIV): ...this.
8062 (v_cmp_result): Rename to...
8063 (v_int_equiv): ...this.
8064 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
8065 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
8066 (copysign<mode>3): Likewise.
8067 (aarch64_simd_bsl<mode>_internal): Likewise.
8068 (aarch64_simd_bsl<mode>): Likewise.
8069 (vec_cmp<mode><mode>): Likewise.
8070 (vcond<mode><mode>): Likewise.
8071 (vcond<v_cmp_mixed><mode>): Likewise.
8072 (vcondu<mode><v_cmp_mixed>): Likewise.
8073 (aarch64_cm<optab><mode>): Likewise.
8074 (aarch64_cmtst<mode>): Likewise.
8075 (aarch64_fac<optab><mode>): Likewise.
8076 (vec_perm_const<mode>): Likewise.
8077 (vcond_mask_<mode><v_cmp_result>): Rename to...
8078 (vcond_mask_<mode><v_int_equiv>): ...this.
8079 (vec_cmp<mode><v_cmp_result>): Rename to...
8080 (vec_cmp<mode><v_int_equiv>): ...this.
8082 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
8083 Alan Hayward <alan.hayward@arm.com>
8084 David Sherwood <david.sherwood@arm.com>
8086 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
8088 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
8089 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
8090 (UNSPEC_LD4_DREG): New unspecs.
8091 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
8092 (aarch64_ld2<mode>_dreg_be): Replace with...
8093 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
8095 (aarch64_ld3<mode>_dreg_le)
8096 (aarch64_ld3<mode>_dreg_be): Replace with...
8097 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
8099 (aarch64_ld4<mode>_dreg_le)
8100 (aarch64_ld4<mode>_dreg_be): Replace with...
8101 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
8104 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8106 PR tree-optimization/81987
8107 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
8108 insert an initializer in a location not dominated by the stride
8111 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
8113 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
8114 on the entire header of the finally block in the fallthru case.
8116 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
8118 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
8120 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
8122 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
8123 rs6000_emit_move_from_cr and call renamed function.
8124 (rs6000_emit_prologue): Call renamed functions.
8125 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
8126 movesi_from_cr, remove volatile CRs.
8128 2017-08-30 Jon Beniston <jon@beniston.com>
8129 Richard Biener <rguenther@suse.de>
8131 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
8132 of VECTOR_MODE_P check.
8133 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
8134 element vector types.
8136 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8138 * df.h (df_read_modify_subreg_p): Remove in favor of...
8139 * rtl.h (read_modify_subreg_p): ...this new function. Take a
8140 const_rtx instead of an rtx.
8141 * cprop.c (local_cprop_find_used_regs): Update accordingly.
8142 * df-problems.c (df_word_lr_mark_ref): Likewise.
8143 * ira-lives.c (mark_pseudo_reg_live): Likewise.
8144 (mark_pseudo_reg_dead): Likewise.
8145 (mark_ref_dead): Likewise.
8146 * reginfo.c (init_subregs_of_mode): Likewise.
8147 * sched-deps.c (sched_analyze_1): Likewise.
8148 * df-scan.c (df_def_record_1): Likewise.
8149 (df_uses_record): Likewise.
8150 (df_read_modify_subreg_p): Remove in favor of...
8151 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
8152 const_rtx instead of an rtx.
8154 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8155 Alan Hayward <alan.hayward@arm.com>
8156 David Sherwood <david.sherwood@arm.com>
8158 * rtl.h (partial_subreg_p): New function.
8159 * caller-save.c (save_call_clobbered_regs): Use it.
8160 * calls.c (expand_call): Likewise.
8161 * combine.c (combinable_i3pat): Likewise.
8162 (simplify_set): Likewise.
8163 (make_extraction): Likewise.
8164 (make_compound_operation_int): Likewise.
8165 (gen_lowpart_or_truncate): Likewise.
8166 (force_to_mode): Likewise.
8167 (make_field_assignment): Likewise.
8168 (reg_truncated_to_mode): Likewise.
8169 (record_truncated_value): Likewise.
8170 (move_deaths): Likewise.
8171 * cse.c (record_jump_cond): Likewise.
8172 (cse_insn): Likewise.
8173 * cselib.c (cselib_lookup_1): Likewise.
8174 * expmed.c (extract_bit_field_using_extv): Likewise.
8175 * function.c (assign_parm_setup_reg): Likewise.
8176 * ifcvt.c (noce_convert_multiple_sets): Likewise.
8177 * ira-build.c (create_insn_allocnos): Likewise.
8178 * lra-coalesce.c (merge_pseudos): Likewise.
8179 * lra-constraints.c (match_reload): Likewise.
8180 (simplify_operand_subreg): Likewise.
8181 (curr_insn_transform): Likewise.
8182 * lra-lives.c (process_bb_lives): Likewise.
8183 * lra.c (new_insn_reg): Likewise.
8184 (lra_substitute_pseudo): Likewise.
8185 * regcprop.c (mode_change_ok): Likewise.
8186 (maybe_mode_change): Likewise.
8187 (copyprop_hardreg_forward_1): Likewise.
8188 * reload.c (push_reload): Likewise.
8189 (find_reloads): Likewise.
8190 (find_reloads_subreg_address): Likewise.
8191 * reload1.c (alter_reg): Likewise.
8192 (eliminate_regs_1): Likewise.
8193 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8195 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
8197 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
8198 back to if statements, including unpack.
8200 2017-08-30 Martin Liska <mliska@suse.cz>
8203 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
8205 (func_checker::compare_asm_inputs_outputs): ... this function.
8206 (func_checker::compare_gimple_asm): Use the function to compare
8207 also ASM constrains.
8208 * ipa-icf-gimple.h: Rename the function.
8210 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8211 Alan Hayward <alan.hayward@arm.com>
8212 David Sherwood <david.sherwood@arm.com>
8214 * coretypes.h (complex_mode): New type.
8215 * gdbhooks.py (build_pretty_printer): Handle it.
8216 * machmode.h (complex_mode): New class.
8217 (complex_mode::includes_p): New function.
8218 (is_complex_int_mode): Likewise.
8219 (is_complex_float_mode): Likewise.
8220 * genmodes.c (get_mode_class): Handle complex mode classes.
8221 * function.c (expand_function_end): Use is_complex_int_mode.
8223 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8224 Alan Hayward <alan.hayward@arm.com>
8225 David Sherwood <david.sherwood@arm.com>
8227 * coretypes.h (scalar_mode_pod): New typedef.
8228 * gdbhooks.py (build_pretty_printer): Handle it.
8229 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
8230 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
8231 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
8232 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
8235 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8236 Alan Hayward <alan.hayward@arm.com>
8237 David Sherwood <david.sherwood@arm.com>
8239 * machmode.h (mode_for_vector): Take a scalar_mode instead
8241 * stor-layout.c (mode_for_vector): Likewise.
8242 * explow.c (promote_mode): Use as_a <scalar_mode>.
8243 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
8245 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8246 Alan Hayward <alan.hayward@arm.com>
8247 David Sherwood <david.sherwood@arm.com>
8249 * target.def (preferred_simd_mode): Take a scalar_mode
8250 instead of a machine_mode.
8251 * targhooks.h (default_preferred_simd_mode): Likewise.
8252 * targhooks.c (default_preferred_simd_mode): Likewise.
8253 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
8254 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
8255 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
8256 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
8257 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
8258 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
8259 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
8260 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
8262 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
8263 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
8264 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
8265 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
8266 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
8267 * doc/tm.texi: Regenerate.
8268 * optabs-query.c (can_vec_mask_load_store_p): Return false for
8271 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8272 Alan Hayward <alan.hayward@arm.com>
8273 David Sherwood <david.sherwood@arm.com>
8275 * target.def (scalar_mode_supported_p): Take a scalar_mode
8276 instead of a machine_mode.
8277 * targhooks.h (default_scalar_mode_supported_p): Likewise.
8278 * targhooks.c (default_scalar_mode_supported_p): Likewise.
8279 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
8280 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
8281 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
8282 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
8283 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
8284 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
8285 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
8286 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
8287 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
8288 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
8289 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
8290 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
8292 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
8293 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
8294 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
8295 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
8296 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
8298 * doc/tm.texi: Regenerate.
8300 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8301 Alan Hayward <alan.hayward@arm.com>
8302 David Sherwood <david.sherwood@arm.com>
8304 * coretypes.h (opt_scalar_mode): New typedef.
8305 * gdbhooks.py (build_pretty_printers): Handle it.
8306 * machmode.h (mode_iterator::get_2xwider): Add overload for
8308 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
8310 * expr.c (convert_mode_scalar): Likewise.
8311 * omp-low.c (omp_clause_aligned_alignment): Likewise.
8312 * optabs.c (expand_float): Likewise.
8313 (expand_fix): Likewise.
8314 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
8316 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8317 Alan Hayward <alan.hayward@arm.com>
8318 David Sherwood <david.sherwood@arm.com>
8320 * optabs.c (expand_float): Explicitly check for scalars before
8321 using a branching expansion.
8322 (expand_fix): Likewise.
8324 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8325 Alan Hayward <alan.hayward@arm.com>
8326 David Sherwood <david.sherwood@arm.com>
8328 * expr.c (convert_mode): Split scalar handling out into...
8329 (convert_mode_scalar): ...this new function. Treat the modes
8332 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8333 Alan Hayward <alan.hayward@arm.com>
8334 David Sherwood <david.sherwood@arm.com>
8336 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
8338 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
8340 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8341 Alan Hayward <alan.hayward@arm.com>
8342 David Sherwood <david.sherwood@arm.com>
8344 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
8345 rather than a machine_mode.
8346 (fixed_from_string): Likewise.
8347 (fixed_convert): Likewise.
8348 (fixed_convert_from_int): Likewise.
8349 (fixed_convert_from_real): Likewise.
8350 (real_convert_from_fixed): Likewise.
8351 * fixed-value.c (fixed_from_double_int): Likewise.
8352 (fixed_from_string): Likewise.
8353 (fixed_convert): Likewise.
8354 (fixed_convert_from_int): Likewise.
8355 (fixed_convert_from_real): Likewise.
8356 (real_convert_from_fixed): Likewise.
8357 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
8359 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8360 Alan Hayward <alan.hayward@arm.com>
8361 David Sherwood <david.sherwood@arm.com>
8363 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
8364 of separate mode class checks. Do not allow vector modes here.
8365 (immed_wide_int_const): Use as_a <scalar_mode>.
8366 * explow.c (trunc_int_for_mode): Likewise.
8367 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
8368 (wi::shwi): Likewise.
8369 (wi::min_value): Likewise.
8370 (wi::max_value): Likewise.
8371 * dwarf2out.c (loc_descriptor): Likewise.
8372 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
8375 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8376 Alan Hayward <alan.hayward@arm.com>
8377 David Sherwood <david.sherwood@arm.com>
8379 * tree.h (SCALAR_TYPE_MODE): New macro.
8380 * expr.c (expand_expr_addr_expr_1): Use it.
8381 (expand_expr_real_2): Likewise.
8382 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
8383 (fold_convert_const_fixed_from_int): Likewise.
8384 (fold_convert_const_fixed_from_real): Likewise.
8385 (native_encode_fixed): Likewise
8386 (native_encode_complex): Likewise
8387 (native_encode_vector): Likewise.
8388 (native_interpret_fixed): Likewise.
8389 (native_interpret_real): Likewise.
8390 (native_interpret_complex): Likewise.
8391 (native_interpret_vector): Likewise.
8392 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
8393 (simd_clone_adjust_argument_types): Likewise.
8394 (simd_clone_init_simd_arrays): Likewise.
8395 (simd_clone_adjust): Likewise.
8396 * stor-layout.c (layout_type): Likewise.
8397 * tree.c (build_minus_one_cst): Likewise.
8398 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
8399 * tree-inline.c (estimate_move_cost): Likewise.
8400 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
8401 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
8402 (vectorizable_reduction): Likewise.
8403 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
8404 (vect_recog_mixed_size_cond_pattern): Likewise.
8405 (check_bool_pattern): Likewise.
8406 (adjust_bool_pattern): Likewise.
8407 (search_type_for_mask_1): Likewise.
8408 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
8409 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
8410 (vectorizable_load): Likewise.
8411 (vectorizable_store): Likewise.
8412 * ubsan.c (ubsan_encode_value): Likewise.
8413 * varasm.c (output_constant): Likewise.
8415 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8416 Alan Hayward <alan.hayward@arm.com>
8417 David Sherwood <david.sherwood@arm.com>
8419 * coretypes.h (scalar_mode): New class.
8420 * machmode.h (scalar_mode): Likewise.
8421 (scalar_mode::includes_p): New function.
8422 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
8423 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
8424 * genmodes.c (get_mode_class): Handle remaining scalar modes.
8425 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
8426 * expmed.c (store_bit_field_1): Likewise.
8427 (extract_bit_field_1): Likewise.
8428 * expr.c (write_complex_part): Likewise.
8429 (read_complex_part): Likewise.
8430 (emit_move_complex_push): Likewise.
8431 (expand_expr_real_2): Likewise.
8432 * function.c (assign_parm_setup_reg): Likewise.
8433 (assign_parms_unsplit_complex): Likewise.
8434 * optabs.c (expand_binop): Likewise.
8435 * rtlanal.c (subreg_get_info): Likewise.
8436 * simplify-rtx.c (simplify_immed_subreg): Likewise.
8437 * varasm.c (output_constant_pool_2): Likewise.
8439 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8440 Alan Hayward <alan.hayward@arm.com>
8441 David Sherwood <david.sherwood@arm.com>
8443 * expmed.c (extract_high_half): Use scalar_int_mode and remove
8445 (expmed_mult_highpart_optab): Likewise.
8446 (expmed_mult_highpart): Likewise.
8448 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8449 Alan Hayward <alan.hayward@arm.com>
8450 David Sherwood <david.sherwood@arm.com>
8452 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
8453 instead of a machine_mode.
8454 (builtin_memset_read_str): Likewise.
8455 * builtins.c (c_readstr): Likewise.
8456 (builtin_memcpy_read_str): Likewise.
8457 (builtin_strncpy_read_str): Likewise.
8458 (builtin_memset_read_str): Likewise.
8459 (builtin_memset_gen_str): Likewise.
8460 (expand_builtin_signbit): Use scalar_int_mode for local variables.
8461 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
8462 instead of a machine_mode.
8463 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
8465 (make_extraction): Likewise.
8466 (try_widen_shift_mode): Take and return scalar_int_modes instead
8468 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
8469 a scalar_int_mode instead of a machine_mode.
8470 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
8471 (avr_addr_space_pointer_mode): Likewise.
8472 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
8473 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
8474 (msp430_unwind_word_mode): Likewise.
8475 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
8476 (spu_addr_space_pointer_mode): Likewise.
8477 (spu_addr_space_address_mode): Likewise.
8478 (spu_libgcc_cmp_return_mode): Likewise.
8479 (spu_libgcc_shift_count_mode): Likewise.
8480 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
8481 (rl78_addr_space_pointer_mode): Likewise.
8482 (fl78_unwind_word_mode): Likewise.
8483 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
8485 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
8486 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
8487 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
8488 (mips_valid_pointer_mode): Likewise.
8489 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
8490 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
8491 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
8493 (ft32_addr_space_address_mode): Likewise.
8494 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
8495 scalar_int_mode instead of a machine_mode.
8496 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
8498 (m32c_addr_space_address_mode): Likewise.
8499 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
8500 (rs6000_eh_return_filter_mode): Likewise.
8501 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
8502 (rs6000_eh_return_filter_mode): Likewise.
8503 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
8504 (s390_libgcc_shift_count_mode): Likewise.
8505 (s390_unwind_word_mode): Likewise.
8506 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
8508 * target.def (mode_rep_extended): Likewise.
8509 (valid_pointer_mode): Likewise.
8510 (addr_space.valid_pointer_mode): Likewise.
8511 (eh_return_filter_mode): Return a scalar_int_mode rather than
8513 (libgcc_cmp_return_mode): Likewise.
8514 (libgcc_shift_count_mode): Likewise.
8515 (unwind_word_mode): Likewise.
8516 (addr_space.pointer_mode): Likewise.
8517 (addr_space.address_mode): Likewise.
8518 * doc/tm.texi: Regenerate.
8519 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
8521 (do_jump): Use scalar_int_mode for local variables.
8522 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
8523 rather than a machine_mode.
8524 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
8525 (scompare_loc_descriptor_wide): Likewise.
8526 (scompare_loc_descriptor_narrow): Likewise.
8527 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
8529 * except.c (sjlj_emit_dispatch_table): Likewise.
8530 (expand_builtin_eh_copy_values): Likewise.
8531 * explow.c (convert_memory_address_addr_space_1): Likewise.
8532 Take a scalar_int_mode rather than a machine_mode.
8533 (convert_memory_address_addr_space): Take a scalar_int_mode rather
8534 than a machine_mode.
8535 (memory_address_addr_space): Use scalar_int_mode for local variables.
8536 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
8537 rather than a machine_mode.
8538 * expmed.c (mask_rtx): Likewise.
8539 (init_expmed_one_conv): Likewise.
8540 (expand_mult_highpart_adjust): Likewise.
8541 (extract_high_half): Likewise.
8542 (expmed_mult_highpart_optab): Likewise.
8543 (expmed_mult_highpart): Likewise.
8544 (expand_smod_pow2): Likewise.
8545 (expand_sdiv_pow2): Likewise.
8546 (emit_store_flag_int): Likewise.
8547 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
8549 (extract_low_bits): Likewise.
8550 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
8552 * expr.c (pieces_addr::adjust): Likewise.
8553 (can_store_by_pieces): Likewise.
8554 (store_by_pieces): Likewise.
8555 (clear_by_pieces_1): Likewise.
8556 (expand_expr_addr_expr_1): Likewise.
8557 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
8558 (expand_expr_real_1): Likewise.
8559 (try_casesi): Likewise.
8560 * final.c (shorten_branches): Likewise.
8561 * fold-const.c (fold_convert_const_int_from_fixed): Change the
8562 type of "mode" to machine_mode.
8563 * internal-fn.c (expand_arith_overflow_result_store): Take a
8564 scalar_int_mode rather than a machine_mode.
8565 (expand_mul_overflow): Use scalar_int_mode for local variables.
8566 * loop-doloop.c (doloop_modify): Likewise.
8567 (doloop_optimize): Likewise.
8568 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
8569 than a machine_mode.
8570 (expand_doubleword_shift_condmove): Likewise.
8571 (expand_doubleword_shift): Likewise.
8572 (expand_doubleword_clz): Likewise.
8573 (expand_doubleword_popcount): Likewise.
8574 (expand_doubleword_parity): Likewise.
8575 (expand_absneg_bit): Use scalar_int_mode for local variables.
8576 (prepare_float_lib_cmp): Likewise.
8577 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
8578 rather than a machine_mode.
8579 (convert_memory_address_addr_space): Likewise.
8580 (get_mode_bounds): Likewise.
8581 (get_address_mode): Return a scalar_int_mode rather than a
8583 * rtlanal.c (get_address_mode): Likewise.
8584 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
8585 than a machine_mode.
8586 * targhooks.c (default_mode_rep_extended): Likewise.
8587 (default_valid_pointer_mode): Likewise.
8588 (default_addr_space_valid_pointer_mode): Likewise.
8589 (default_eh_return_filter_mode): Return a scalar_int_mode rather
8590 than a machine_mode.
8591 (default_libgcc_cmp_return_mode): Likewise.
8592 (default_libgcc_shift_count_mode): Likewise.
8593 (default_unwind_word_mode): Likewise.
8594 (default_addr_space_pointer_mode): Likewise.
8595 (default_addr_space_address_mode): Likewise.
8596 * targhooks.h (default_eh_return_filter_mode): Likewise.
8597 (default_libgcc_cmp_return_mode): Likewise.
8598 (default_libgcc_shift_count_mode): Likewise.
8599 (default_unwind_word_mode): Likewise.
8600 (default_addr_space_pointer_mode): Likewise.
8601 (default_addr_space_address_mode): Likewise.
8602 (default_mode_rep_extended): Take a scalar_int_mode rather than
8604 (default_valid_pointer_mode): Likewise.
8605 (default_addr_space_valid_pointer_mode): Likewise.
8606 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
8608 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
8609 rather than a machine_mode.
8610 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
8611 for local variables.
8612 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
8613 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
8614 than a machine_mode.
8616 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8617 Alan Hayward <alan.hayward@arm.com>
8618 David Sherwood <david.sherwood@arm.com>
8620 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
8621 the mode argument to scalar_int_mode.
8622 (do_jump_by_parts_zero_rtx): Likewise.
8623 (do_jump_by_parts_equality_rtx): Likewise.
8624 (do_jump_by_parts_greater): Take a mode argument.
8625 (do_jump_by_parts_equality): Likewise.
8626 (do_jump_1): Update calls accordingly.
8628 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8629 Alan Hayward <alan.hayward@arm.com>
8630 David Sherwood <david.sherwood@arm.com>
8632 * is-a.h (safe_dyn_cast): New function.
8633 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
8634 (jump_table_for_label): Likewise.
8635 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
8636 instead of an rtx_insn *.
8637 (shorten_branches): Use dyn_cast instead of LABEL_P and
8638 JUMP_TABLE_DATA_P. Use jump_table_for_label and
8639 rtx_jump_table_data::get_data_mode.
8640 (final_scan_insn): Likewise.
8642 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8643 Alan Hayward <alan.hayward@arm.com>
8644 David Sherwood <david.sherwood@arm.com>
8646 * combine.c (try_combine): Use is_a <scalar_int_mode> when
8647 trying to combine a full-register integer set with a subreg
8650 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8651 Alan Hayward <alan.hayward@arm.com>
8652 David Sherwood <david.sherwood@arm.com>
8654 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
8655 that is always either address_mode or pointer_mode.
8657 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8658 Alan Hayward <alan.hayward@arm.com>
8659 David Sherwood <david.sherwood@arm.com>
8661 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
8662 when the two are known to be equal.
8664 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8665 Alan Hayward <alan.hayward@arm.com>
8666 David Sherwood <david.sherwood@arm.com>
8668 * simplify-rtx.c (simplify_const_unary_operation): Use
8669 is_a <scalar_int_mode> instead of checking for a nonzero
8670 precision. Forcibly convert op_mode to a scalar_int_mode
8671 in that case. More clearly differentiate the operand and
8672 result modes and use the former when deciding what the value
8673 of a count-bits operation should be. Use is_int_mode instead
8674 of checking for a MODE_INT. Remove redundant check for whether
8675 this mode has a zero precision.
8677 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8678 Alan Hayward <alan.hayward@arm.com>
8679 David Sherwood <david.sherwood@arm.com>
8681 * optabs.c (widen_leading): Change the type of the mode argument
8682 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
8683 (widen_bswap): Likewise.
8684 (expand_parity): Likewise.
8685 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
8686 (expand_ffs): Likewise.
8687 (epand_unop): Check for scalar integer modes before calling the
8690 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8691 Alan Hayward <alan.hayward@arm.com>
8692 David Sherwood <david.sherwood@arm.com>
8694 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
8696 (expand_expr_real_1): Update call accordingly.
8698 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8699 Alan Hayward <alan.hayward@arm.com>
8700 David Sherwood <david.sherwood@arm.com>
8702 * expmed.c (store_bit_field_using_insv): Add op0_mode and
8703 value_mode arguments. Use scalar_int_mode internally.
8704 (store_bit_field_1): Rename the new integer mode from imode
8705 to op0_mode and use it instead of GET_MODE (op0). Update calls
8706 to store_split_bit_field, store_bit_field_using_insv and
8707 store_fixed_bit_field.
8708 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
8709 Use scalar_int_mode internally. Use a bit count rather than a mode
8710 when calculating the largest bit size for get_best_mode.
8711 Update calls to store_split_bit_field and store_fixed_bit_field_1.
8712 (store_fixed_bit_field_1): Add mode and value_mode arguments.
8713 Remove assertion that OP0 has a scalar integer mode.
8714 (store_split_bit_field): Add op0_mode and value_mode arguments.
8715 Update calls to extract_fixed_bit_field.
8716 (extract_bit_field_using_extv): Add an op0_mode argument.
8717 Use scalar_int_mode internally.
8718 (extract_bit_field_1): Rename the new integer mode from imode to
8719 op0_mode and use it instead of GET_MODE (op0). Update calls to
8720 extract_split_bit_field, extract_bit_field_using_extv and
8721 extract_fixed_bit_field.
8722 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
8723 to extract_split_bit_field and extract_fixed_bit_field_1.
8724 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
8725 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
8727 (extract_split_bit_field): Add an op0_mode argument. Update call
8728 to extract_fixed_bit_field.
8730 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8731 Alan Hayward <alan.hayward@arm.com>
8732 David Sherwood <david.sherwood@arm.com>
8734 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
8735 * explow.c (hard_function_value): Likewise.
8736 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
8737 convert_to_mode call outside the loop.
8738 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
8739 for the mode iterator. Require the mode specified by max_pieces
8741 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
8743 (copy_blkmode_to_reg): Likewise.
8744 (set_storage_via_setmem): Likewise.
8745 * optabs.c (prepare_cmp_insn): Likewise.
8746 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
8747 * stor-layout.c (finish_bitfield_representative): Likewise.
8749 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8750 Alan Hayward <alan.hayward@arm.com>
8751 David Sherwood <david.sherwood@arm.com>
8753 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
8754 * expr.c (convert_move): Use them.
8755 (convert_modes): Likewise.
8756 (store_expr_with_bounds): Likewise.
8758 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8759 Alan Hayward <alan.hayward@arm.com>
8760 David Sherwood <david.sherwood@arm.com>
8762 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
8763 parameter for the mode of "x". Remove the "known_x", "known_mode"
8764 and "known_ret" arguments. Change the type of the mode argument
8766 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
8767 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
8768 (reg_num_sign_bit_copies_for_combine): Likewise.
8769 * rtlanal.c (nonzero_bits1): Likewise.
8770 (num_sign_bit_copies1): Likewise.
8771 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
8772 (reg_num_sign_bit_copies_general): Likewise.
8773 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
8774 (reg_nonzero_bits_general): Likewise.
8776 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8777 Alan Hayward <alan.hayward@arm.com>
8778 David Sherwood <david.sherwood@arm.com>
8780 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
8781 than in subroutines. Return 1 for non-integer modes.
8782 (cached_num_sign_bit_copies): Change the type of the mode parameter
8784 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
8785 classes. Handle CONST_INT_P first and then check whether X also
8786 has a scalar integer mode. Check the same thing for inner registers
8787 of a SUBREG and for values that are being extended or truncated.
8789 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8790 Alan Hayward <alan.hayward@arm.com>
8791 David Sherwood <david.sherwood@arm.com>
8793 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
8794 in subroutines. Return the mode mask for non-integer modes.
8795 (cached_nonzero_bits): Change the type of the mode parameter
8797 (nonzero_bits1): Likewise. Remove early exit for other mode
8798 classes. Handle CONST_INT_P first and then check whether X
8799 also has a scalar integer mode.
8801 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8802 Alan Hayward <alan.hayward@arm.com>
8803 David Sherwood <david.sherwood@arm.com>
8805 * expr.c (widest_int_mode_for_size): Make the comment match the code.
8806 Return a scalar_int_mode and assert that the size is greater than
8808 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
8809 (op_by_pieces_d::op_by_pieces_d): Likewise.
8810 (op_by_pieces_d::run): Likewise.
8811 (can_store_by_pieces): Likewise.
8813 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8814 Alan Hayward <alan.hayward@arm.com>
8815 David Sherwood <david.sherwood@arm.com>
8817 * combine.c (extract_left_shift): Add a mode argument and update
8819 (make_compound_operation_int): Change the type of the mode parameter
8820 to scalar_int_mode and update the call to extract_left_shift.
8822 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8823 Alan Hayward <alan.hayward@arm.com>
8824 David Sherwood <david.sherwood@arm.com>
8826 * combine.c (simplify_and_const_int): Change the type of the mode
8827 parameter to scalar_int_mode.
8828 (simplify_and_const_int_1): Likewise. Update recursive call.
8830 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8831 Alan Hayward <alan.hayward@arm.com>
8832 David Sherwood <david.sherwood@arm.com>
8834 * combine.c (simplify_compare_const): Check that the mode is a
8835 scalar_int_mode (rather than VOIDmode) before testing its
8837 (simplify_comparison): Move COMPARISON_P handling out of the
8838 loop and restrict the latter part of the loop to scalar_int_modes.
8839 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
8840 and when considering SUBREG_REGs. Use is_int_mode instead of
8841 checking GET_MODE_CLASS against MODE_INT.
8843 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8844 Alan Hayward <alan.hayward@arm.com>
8845 David Sherwood <david.sherwood@arm.com>
8847 * combine.c (try_widen_shift_mode): Move check for equal modes to...
8848 (simplify_shift_const_1): ...here. Use scalar_int_mode for
8849 shift_unit_mode and for modes involved in scalar shifts.
8851 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8852 Alan Hayward <alan.hayward@arm.com>
8853 David Sherwood <david.sherwood@arm.com>
8855 * combine.c (force_int_to_mode): New function, split out from...
8856 (force_to_mode): ...here. Keep xmode up-to-date and use it
8857 instead of GET_MODE (x).
8859 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8860 Alan Hayward <alan.hayward@arm.com>
8861 David Sherwood <david.sherwood@arm.com>
8863 * optabs-query.h (extraction_insn::struct_mode): Change type to
8864 opt_scalar_int_mode and update comment.
8865 (extraction_insn::field_mode): Change type to scalar_int_mode.
8866 (extraction_insn::pos_mode): Likewise.
8867 * combine.c (make_extraction): Update accordingly.
8868 * optabs-query.c (get_traditional_extraction_insn): Likewise.
8869 (get_optab_extraction_insn): Likewise.
8870 * recog.c (simplify_while_replacing): Likewise.
8871 * expmed.c (narrow_bit_field_mem): Change the type of the mode
8872 parameter to opt_scalar_int_mode.
8874 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8875 Alan Hayward <alan.hayward@arm.com>
8876 David Sherwood <david.sherwood@arm.com>
8878 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
8879 to a scalar_int_mode instead of a machine_mode.
8880 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
8881 (get_best_mode): Return a boolean and use a pointer argument to store
8882 the selected mode. Replace the limit mode parameter with a bit limit.
8883 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
8884 for the values returned by bit_field_mode_iterator::next_mode.
8885 (store_bit_field): Update call to get_best_mode.
8886 (store_fixed_bit_field): Likewise.
8887 (extract_fixed_bit_field): Likewise.
8888 * expr.c (optimize_bitfield_assignment_op): Likewise.
8889 * fold-const.c (optimize_bit_field_compare): Likewise.
8890 (fold_truth_andor_1): Likewise.
8891 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
8892 Update for new type of m_mode.
8893 (get_best_mode): As above.
8895 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8896 Alan Hayward <alan.hayward@arm.com>
8897 David Sherwood <david.sherwood@arm.com>
8899 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
8900 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
8901 (store_bit_field): Check is_a <scalar_int_mode> before calling
8902 strict_volatile_bitfield_p.
8903 (extract_bit_field): Likewise.
8905 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8906 Alan Hayward <alan.hayward@arm.com>
8907 David Sherwood <david.sherwood@arm.com>
8909 * target.def (cstore_mode): Return a scalar_int_mode.
8910 * doc/tm.texi: Regenerate.
8911 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
8912 * targhooks.h (default_cstore_mode): Likewise.
8913 * targhooks.c (default_cstore_mode): Likewise, using a forced
8915 * expmed.c (emit_cstore): Expect the target of the cstore to be
8918 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8919 Alan Hayward <alan.hayward@arm.com>
8920 David Sherwood <david.sherwood@arm.com>
8922 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
8924 (niter_desc): Likewise mode.
8925 (iv_analyze): Add a mode parameter.
8927 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
8928 and change its type to scalar_int_mode.
8929 * loop-iv.c: Update commentary at head of file.
8930 (iv_constant): Pass the mode paraeter before the rtx it describes
8931 and change its type to scalar_int_mode. Remove VOIDmode handling.
8932 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
8933 (iv_extend): Likewise.
8934 (shorten_into_mode): Likewise.
8935 (iv_add): Use scalar_int_mode.
8936 (iv_mult): Likewise.
8937 (iv_shift): Likewise.
8938 (canonicalize_iv_subregs): Likewise.
8939 (get_biv_step_1): Pass the outer_mode parameter before the rtx
8940 it describes and change its mode to scalar_int_mode. Also change
8941 the type of the returned inner_mode to scalar_int_mode.
8942 (get_biv_step): Likewise, turning outer_mode from a pointer
8943 into a direct parameter. Update call to get_biv_step_1.
8944 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
8945 iv_constant and get_biv_step.
8946 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
8947 and change its type to scalar_int_mode. Don't initialise iv->mode
8948 to VOIDmode and remove later checks for its still being VOIDmode.
8949 Update calls to iv_analyze_op and iv_analyze_expr. Check
8950 is_a <scalar_int_mode> when changing the mode under consideration.
8951 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
8952 Update call to iv_analyze_expr.
8953 (iv_analyze_op): Add a mode parameter. Reject subregs whose
8954 inner register is not also a scalar_int_mode. Update call to
8956 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
8957 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
8958 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
8959 separate mode class checks. Update calls to iv_analyze. Remove
8960 fix-up of VOIDmodes after iv_analyze_biv.
8961 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
8962 don't have a scalar_int_mode. Update call to biv_p.
8964 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8965 Alan Hayward <alan.hayward@arm.com>
8966 David Sherwood <david.sherwood@arm.com>
8968 * cfgexpand.c (convert_debug_memory_address): Use
8969 as_a <scalar_int_mode>.
8970 * combine.c (expand_compound_operation): Likewise.
8971 (make_extraction): Likewise.
8972 (change_zero_ext): Likewise.
8973 (simplify_comparison): Likewise.
8974 * cse.c (cse_insn): Likewise.
8975 * dwarf2out.c (minmax_loc_descriptor): Likewise.
8976 (mem_loc_descriptor): Likewise.
8977 (loc_descriptor): Likewise.
8978 * expmed.c (init_expmed_one_mode): Likewise.
8979 (synth_mult): Likewise.
8980 (emit_store_flag_1): Likewise.
8981 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
8982 of a comparison with size.
8983 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
8984 (reduce_to_bit_field_precision): Likewise.
8985 * function.c (expand_function_end): Likewise.
8986 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
8987 * loop-doloop.c (doloop_modify): Likewise.
8988 * optabs.c (expand_binop): Likewise.
8989 (expand_unop): Likewise.
8990 (expand_copysign_absneg): Likewise.
8991 (prepare_cmp_insn): Likewise.
8992 (maybe_legitimize_operand): Likewise.
8993 * recog.c (const_scalar_int_operand): Likewise.
8994 * rtlanal.c (get_address_mode): Likewise.
8995 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8996 (simplify_cond_clz_ctz): Likewise.
8997 * tree-nested.c (get_nl_goto_field): Likewise.
8998 * tree.c (build_vector_type_for_mode): Likewise.
8999 * var-tracking.c (use_narrower_mode): Likewise.
9001 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9002 Alan Hayward <alan.hayward@arm.com>
9003 David Sherwood <david.sherwood@arm.com>
9005 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
9006 * builtins.c (expand_builtin_signbit): Use it.
9007 * cfgexpand.c (expand_debug_expr): Likewise.
9008 * dojump.c (do_jump): Likewise.
9009 (do_compare_and_jump): Likewise.
9010 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
9011 * expmed.c (make_tree): Likewise.
9012 * expr.c (expand_expr_real_2): Likewise.
9013 (expand_expr_real_1): Likewise.
9014 (try_casesi): Likewise.
9015 * fold-const-call.c (fold_const_call_ss): Likewise.
9016 * fold-const.c (unextend): Likewise.
9017 (extract_muldiv_1): Likewise.
9018 (fold_single_bit_test): Likewise.
9019 (native_encode_int): Likewise.
9020 (native_encode_string): Likewise.
9021 (native_interpret_int): Likewise.
9022 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
9023 * internal-fn.c (expand_addsub_overflow): Likewise.
9024 (expand_neg_overflow): Likewise.
9025 (expand_mul_overflow): Likewise.
9026 (expand_arith_overflow): Likewise.
9027 * match.pd: Likewise.
9028 * stor-layout.c (layout_type): Likewise.
9029 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
9030 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
9031 * tree-ssanames.c (get_range_info): Likewise.
9032 * tree-switch-conversion.c (array_value_type) Likewise.
9033 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
9034 (vect_recog_divmod_pattern): Likewise.
9035 (vect_recog_mixed_size_cond_pattern): Likewise.
9036 * tree-vrp.c (extract_range_basic): Likewise.
9037 (simplify_float_conversion_using_ranges): Likewise.
9038 * tree.c (int_fits_type_p): Likewise.
9039 * ubsan.c (instrument_bool_enum_load): Likewise.
9040 * varasm.c (mergeable_string_section): Likewise.
9041 (narrowing_initializer_constant_valid_p): Likewise.
9042 (output_constant): Likewise.
9044 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9045 Alan Hayward <alan.hayward@arm.com>
9046 David Sherwood <david.sherwood@arm.com>
9048 * machmode.h (NARROWEST_INT_MODE): New macro.
9049 * expr.c (alignment_for_piecewise_move): Use it instead of
9050 GET_CLASS_NARROWEST_MODE (MODE_INT).
9051 (push_block): Likewise.
9052 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
9054 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
9056 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9057 Alan Hayward <alan.hayward@arm.com>
9058 David Sherwood <david.sherwood@arm.com>
9060 * postreload.c (move2add_valid_value_p): Change the type of the
9061 mode parameter to scalar_int_mode.
9062 (move2add_use_add2_insn): Add a mode parameter and use it instead
9064 (move2add_use_add3_insn): Likewise.
9065 (reload_cse_move2add): Update accordingly.
9067 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9068 Alan Hayward <alan.hayward@arm.com>
9069 David Sherwood <david.sherwood@arm.com>
9071 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
9073 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
9074 * optabs.c (expand_unop): Likewise.
9076 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9077 Alan Hayward <alan.hayward@arm.com>
9078 David Sherwood <david.sherwood@arm.com>
9080 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
9081 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
9082 (popcount_loc_descriptor): Likewise.
9083 (bswap_loc_descriptor): Likewise.
9084 (rotate_loc_descriptor): Likewise.
9085 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
9086 calling the functions above.
9088 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9089 Alan Hayward <alan.hayward@arm.com>
9090 David Sherwood <david.sherwood@arm.com>
9092 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
9094 (try_combine): Likewise.
9095 (simplify_if_then_else): Likewise.
9096 * cse.c (cse_insn): Likewise.
9097 * dwarf2out.c (mem_loc_descriptor): Likewise.
9098 * emit-rtl.c (gen_lowpart_common): Likewise.
9099 * simplify-rtx.c (simplify_truncation): Likewise.
9100 (simplify_binary_operation_1): Likewise.
9101 (simplify_const_relational_operation): Likewise.
9102 (simplify_ternary_operation): Likewise.
9103 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
9105 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9106 Alan Hayward <alan.hayward@arm.com>
9107 David Sherwood <david.sherwood@arm.com>
9109 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
9110 * reload.c (push_reload): Likewise.
9111 (find_reloads): Likewise.
9113 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9114 Alan Hayward <alan.hayward@arm.com>
9115 David Sherwood <david.sherwood@arm.com>
9117 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
9118 (make_compound_operation_int): Likewise.
9119 (change_zero_ext): Likewise.
9120 * expr.c (convert_move): Likewise.
9121 (convert_modes): Likewise.
9122 * fwprop.c (forward_propagate_subreg): Likewise.
9123 * loop-iv.c (get_biv_step_1): Likewise.
9124 * optabs.c (widen_operand): Likewise.
9125 * postreload.c (move2add_valid_value_p): Likewise.
9126 * recog.c (simplify_while_replacing): Likewise.
9127 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9128 (simplify_binary_operation_1): Likewise. Remove redundant
9129 mode equality check.
9131 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9132 Alan Hayward <alan.hayward@arm.com>
9133 David Sherwood <david.sherwood@arm.com>
9135 * combine.c (combine_simplify_rtx): Add checks for
9136 is_a <scalar_int_mode>.
9137 (simplify_if_then_else): Likewise.
9138 (make_field_assignment): Likewise.
9139 (simplify_comparison): Likewise.
9140 * ifcvt.c (noce_try_bitop): Likewise.
9141 * loop-invariant.c (canonicalize_address_mult): Likewise.
9142 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9144 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9145 Alan Hayward <alan.hayward@arm.com>
9146 David Sherwood <david.sherwood@arm.com>
9148 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
9149 is_a <scalar_int_mode> instead of != BLKmode.
9151 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9152 Alan Hayward <alan.hayward@arm.com>
9153 David Sherwood <david.sherwood@arm.com>
9155 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
9156 instead of != VOIDmode.
9157 * combine.c (if_then_else_cond): Likewise.
9158 (change_zero_ext): Likewise.
9159 * dwarf2out.c (mem_loc_descriptor): Likewise.
9160 (loc_descriptor): Likewise.
9161 * rtlanal.c (canonicalize_condition): Likewise.
9162 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
9164 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9165 Alan Hayward <alan.hayward@arm.com>
9166 David Sherwood <david.sherwood@arm.com>
9168 * simplify-rtx.c (simplify_binary_operation_1): Use
9169 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
9171 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9172 Alan Hayward <alan.hayward@arm.com>
9173 David Sherwood <david.sherwood@arm.com>
9175 * wide-int.h (int_traits<unsigned char>) New class.
9176 (int_traits<unsigned short>) Likewise.
9177 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
9178 Use GET_MODE_UNIT_PRECISION and remove redundant test for
9180 * combine.c (set_nonzero_bits_and_sign_copies): Use
9181 is_a <scalar_int_mode>.
9182 (find_split_point): Likewise.
9183 (combine_simplify_rtx): Likewise.
9184 (simplify_logical): Likewise.
9185 (expand_compound_operation): Likewise.
9186 (expand_field_assignment): Likewise.
9187 (make_compound_operation): Likewise.
9188 (extended_count): Likewise.
9189 (change_zero_ext): Likewise.
9190 (simplify_comparison): Likewise.
9191 * dwarf2out.c (scompare_loc_descriptor): Likewise.
9192 (ucompare_loc_descriptor): Likewise.
9193 (minmax_loc_descriptor): Likewise.
9194 (mem_loc_descriptor): Likewise.
9195 (loc_descriptor): Likewise.
9196 * expmed.c (init_expmed_one_mode): Likewise.
9197 * lra-constraints.c (lra_constraint_offset): Likewise.
9198 * optabs.c (prepare_libcall_arg): Likewise.
9199 * postreload.c (move2add_note_store): Likewise.
9200 * reload.c (operands_match_p): Likewise.
9201 * rtl.h (load_extend_op): Likewise.
9202 * rtlhooks.c (gen_lowpart_general): Likewise.
9203 * simplify-rtx.c (simplify_truncation): Likewise.
9204 (simplify_unary_operation_1): Likewise.
9205 (simplify_binary_operation_1): Likewise.
9206 (simplify_const_binary_operation): Likewise.
9207 (simplify_const_relational_operation): Likewise.
9208 (simplify_subreg): Likewise.
9209 * stor-layout.c (bitwise_mode_for_mode): Likewise.
9210 * var-tracking.c (adjust_mems): Likewise.
9211 (prepare_call_arguments): Likewise.
9213 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9214 Alan Hayward <alan.hayward@arm.com>
9215 David Sherwood <david.sherwood@arm.com>
9217 * machmode.h (is_int_mode): New fuction.
9218 * combine.c (find_split_point): Use it.
9219 (combine_simplify_rtx): Likewise.
9220 (simplify_if_then_else): Likewise.
9221 (simplify_set): Likewise.
9222 (simplify_shift_const_1): Likewise.
9223 (simplify_comparison): Likewise.
9224 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
9225 * cse.c (notreg_cost): Likewise.
9226 (cse_insn): Likewise.
9227 * cselib.c (cselib_lookup_1): Likewise.
9228 * dojump.c (do_jump_1): Likewise.
9229 (do_compare_rtx_and_jump): Likewise.
9230 * dse.c (get_call_args): Likewise.
9231 * dwarf2out.c (rtl_for_decl_init): Likewise.
9232 (native_encode_initializer): Likewise.
9233 * expmed.c (emit_store_flag_1): Likewise.
9234 (emit_store_flag): Likewise.
9235 * expr.c (convert_modes): Likewise.
9236 (store_field): Likewise.
9237 (expand_expr_real_1): Likewise.
9238 * fold-const.c (fold_read_from_constant_string): Likewise.
9239 * gimple-ssa-sprintf.c (get_format_string): Likewise.
9240 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
9241 * optabs.c (expand_binop): Likewise.
9242 (expand_unop): Likewise.
9243 (expand_abs_nojump): Likewise.
9244 (expand_one_cmpl_abs_nojump): Likewise.
9245 * simplify-rtx.c (mode_signbit_p): Likewise.
9246 (val_signbit_p): Likewise.
9247 (val_signbit_known_set_p): Likewise.
9248 (val_signbit_known_clear_p): Likewise.
9249 (simplify_relational_operation_1): Likewise.
9250 * tree.c (vector_type_mode): Likewise.
9252 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9253 Alan Hayward <alan.hayward@arm.com>
9254 David Sherwood <david.sherwood@arm.com>
9256 * machmode.h (smallest_mode_for_size): Fix formatting.
9257 (smallest_int_mode_for_size): New function.
9258 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
9259 instead of smallest_mode_for_size.
9260 * combine.c (make_extraction): Likewise.
9261 * config/arc/arc.c (arc_expand_movmem): Likewise.
9262 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
9263 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
9264 * config/s390/s390.c (s390_expand_insv): Likewise.
9265 * config/sparc/sparc.c (assign_int_registers): Likewise.
9266 * config/spu/spu.c (spu_function_value): Likewise.
9267 (spu_function_arg): Likewise.
9268 * coverage.c (get_gcov_type): Likewise.
9269 (get_gcov_unsigned_t): Likewise.
9270 * dse.c (find_shift_sequence): Likewise.
9271 * expmed.c (store_bit_field_1): Likewise.
9272 * expr.c (convert_move): Likewise.
9273 (store_field): Likewise.
9274 * internal-fn.c (expand_arith_overflow): Likewise.
9275 * optabs-query.c (get_best_extraction_insn): Likewise.
9276 * optabs.c (expand_twoval_binop_libfunc): Likewise.
9277 * stor-layout.c (layout_type): Likewise.
9278 (initialize_sizetypes): Likewise.
9279 * targhooks.c (default_get_mask_mode): Likewise.
9280 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
9282 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9283 Alan Hayward <alan.hayward@arm.com>
9284 David Sherwood <david.sherwood@arm.com>
9286 * machmode.h (opt_mode::else_blk): New function.
9287 (int_mode_for_mode): Declare.
9288 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
9289 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
9291 * cfgexpand.c (expand_debug_expr): Likewise.
9292 * combine.c (gen_lowpart_or_truncate): Likewise.
9293 (gen_lowpart_for_combine): Likewise.
9294 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
9295 * config/avr/avr.c (avr_to_int_mode): Likewise.
9296 (avr_out_plus_1): Likewise.
9297 (avr_out_plus): Likewise.
9298 (avr_out_round): Likewise.
9299 * config/i386/i386.c (ix86_split_to_parts): Likewise.
9300 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
9301 (s390_expand_vcond): Likewise.
9302 * config/spu/spu.c (spu_split_immediate): Likewise.
9303 (spu_expand_mov): Likewise.
9304 * dse.c (get_stored_val): Likewise.
9305 * expmed.c (store_bit_field_1): Likewise.
9306 (convert_extracted_bit_field): Use int_mode_for_mode instead of
9308 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
9309 (extract_low_bits): Likewise.
9310 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
9311 handling rather than repeating the check.
9312 (emit_group_store): Likewise.
9313 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
9314 * optabs.c (expand_absneg_bit): Likewise.
9315 (expand_copysign_absneg): Likewise.
9316 (expand_copysign_bit): Likewise.
9317 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
9318 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
9319 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
9320 * var-tracking.c (prepare_call_arguments): Likewise.
9321 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
9322 int_mode_for_mode instead of mode_for_size.
9323 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
9325 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9326 Alan Hayward <alan.hayward@arm.com>
9327 David Sherwood <david.sherwood@arm.com>
9329 * machmode.h (int_mode_for_size): New function.
9330 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
9331 instead of mode_for_size.
9332 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
9334 * combine.c (expand_field_assignment): Use int_mode_for_size
9335 instead of mode_for_size.
9336 (make_extraction): Likewise.
9337 (simplify_shift_const_1): Likewise.
9338 (simplify_comparison): Likewise.
9339 * dojump.c (do_jump): Likewise.
9340 * dwarf2out.c (mem_loc_descriptor): Likewise.
9341 * emit-rtl.c (init_derived_machine_modes): Likewise.
9342 * expmed.c (flip_storage_order): Likewise.
9343 (convert_extracted_bit_field): Likewise.
9344 * expr.c (copy_blkmode_from_reg): Likewise.
9345 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
9346 * internal-fn.c (expand_mul_overflow): Likewise.
9347 * lower-subreg.c (simple_move): Likewise.
9348 * optabs-libfuncs.c (init_optabs): Likewise.
9349 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9350 * tree.c (vector_type_mode): Likewise.
9351 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
9352 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
9353 * tree-vect-generic.c (expand_vector_parallel): Likewise.
9354 * tree-vect-stmts.c (vectorizable_load): Likewise.
9355 (vectorizable_store): Likewise.
9357 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9358 Alan Hayward <alan.hayward@arm.com>
9359 David Sherwood <david.sherwood@arm.com>
9361 * coretypes.h (pod_mode): New type.
9362 (scalar_int_mode_pod): New typedef.
9363 * machmode.h (pod_mode): New class.
9364 (int_n_data_t::m): Change type to scalar_int_mode_pod.
9365 * genmodes.c (emit_mode_int_n): Update accordingly.
9366 * lower-subreg.h (target_lower_subreg): Change type to
9367 scalar_int_mode_pod.
9368 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
9369 scalar_int_mode_pod.
9371 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9372 Alan Hayward <alan.hayward@arm.com>
9373 David Sherwood <david.sherwood@arm.com>
9375 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
9376 machine_mode to scalar_int_mode.
9377 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
9378 (rs6000_option_override_internal): Remove cast to int.
9379 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
9380 machine_mode to scalar_int_mode.
9381 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
9382 (rs6000_option_override_internal): Remove cast to int.
9383 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
9384 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
9386 * config/s390/s390.c (s390_expand_builtin): Likewise.
9387 * coretypes.h (scalar_int_mode): New type.
9388 (opt_scalar_int_mode): New typedef.
9389 * machmode.h (scalar_int_mode): New class.
9390 (scalar_int_mode::includes_p): New function.
9391 (byte_mode): Change type to scalar_int_mode.
9392 (word_mode): Likewise.
9393 (ptr_mode): Likewise.
9394 * emit-rtl.c (byte_mode): Likewise.
9395 (word_mode): Likewise.
9396 (ptr_mode): Likewise.
9397 (init_derived_machine_modes): Update accordingly.
9398 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
9399 and MODE_PARTIAL_INT.
9400 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
9401 opt_scalar_int_mode.
9403 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9404 Alan Hayward <alan.hayward@arm.com>
9405 David Sherwood <david.sherwood@arm.com>
9407 * target.def (libgcc_floating_mode_supported_p): Take a
9409 * doc/tm.texi: Regenerate.
9410 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
9412 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
9413 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
9416 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9417 Alan Hayward <alan.hayward@arm.com>
9418 David Sherwood <david.sherwood@arm.com>
9420 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
9421 * doc/tm.texi: Regenerate.
9422 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
9423 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
9424 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
9425 * targhooks.h (default_floatn_mode): Likewise.
9426 * targhooks.c (default_floatn_mode): Likewise.
9427 * tree.c (build_common_tree_nodes): Update accordingly.
9429 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9430 Alan Hayward <alan.hayward@arm.com>
9431 David Sherwood <david.sherwood@arm.com>
9433 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
9434 (mode_iterator::iterate_p): Likewise.
9435 (mode_iterator::get_wider): Likewise.
9436 * expr.c (init_expr_target): Use opt_scalar_float_mode.
9438 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9439 Alan Hayward <alan.hayward@arm.com>
9440 David Sherwood <david.sherwood@arm.com>
9442 * coretypes.h (opt_scalar_float_mode): New typedef.
9443 * machmode.h (float_mode_for_size): New function.
9444 * emit-rtl.c (double_mode): Delete.
9445 (init_emit_once): Use float_mode_for_size.
9446 * stor-layout.c (layout_type): Likewise.
9447 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
9449 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9450 Alan Hayward <alan.hayward@arm.com>
9451 David Sherwood <david.sherwood@arm.com>
9453 * output.h (assemble_real): Take a scalar_float_mode.
9454 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
9455 * config/arm/arm.md (consttable_4): Likewise.
9456 (consttable_8): Likewise.
9457 (consttable_16): Likewise.
9458 * config/mips/mips.md (consttable_float): Likewise.
9459 * config/s390/s390.c (s390_output_pool_entry): Likewise.
9460 * varasm.c (assemble_real): Take a scalar_float_mode.
9461 (output_constant_pool_2): Update accordingly.
9462 (output_constant): Likewise.
9464 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9465 Alan Hayward <alan.hayward@arm.com>
9466 David Sherwood <david.sherwood@arm.com>
9468 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
9469 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
9470 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
9471 (native_encode_real): Likewise.
9472 (native_interpret_real): Likewise.
9473 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
9474 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
9476 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9477 Alan Hayward <alan.hayward@arm.com>
9478 David Sherwood <david.sherwood@arm.com>
9480 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
9481 <scalar_float_mode>. Simplify.
9482 (gen_extend_conv_libfunc): Likewise.
9484 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9485 Alan Hayward <alan.hayward@arm.com>
9486 David Sherwood <david.sherwood@arm.com>
9488 * coretypes.h (scalar_float_mode): New type.
9489 * machmode.h (mode_traits::from_int): Use machine_mode if
9490 USE_ENUM_MODES is defined.
9491 (is_a): New function.
9493 (dyn_cast): Likewise.
9494 (scalar_float_mode): New class.
9495 (scalar_float_mode::includes_p): New function.
9496 (is_float_mode): Likewise.
9497 * gdbhooks.py (MachineModePrinter): New class.
9498 (build_pretty_printer): Use it for scalar_float_mode.
9499 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
9500 (format_helper::format_helper): Turn into a template.
9501 * genmodes.c (get_mode_class): New function.
9502 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
9503 or machine_mode if none.
9504 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
9505 as_a <scalar_float_mode>.
9506 * dwarf2out.c (mem_loc_descriptor): Likewise.
9507 (insert_float): Likewise.
9508 (add_const_value_attribute): Likewise.
9509 * simplify-rtx.c (simplify_immed_subreg): Likewise.
9510 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
9511 (expand_unop): Update accordingly.
9512 (expand_abs_nojump): Likewise.
9513 (expand_copysign_absneg): Take a scalar_float_mode.
9514 (expand_copysign_bit): Likewise.
9515 (expand_copysign): Update accordingly.
9517 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9518 Alan Hayward <alan.hayward@arm.com>
9519 David Sherwood <david.sherwood@arm.com>
9521 * coretypes.h (opt_mode): New class.
9522 * machmode.h (opt_mode): Likewise.
9523 (opt_mode::else_void): New function.
9524 (opt_mode::require): Likewise.
9525 (opt_mode::exists): Likewise.
9526 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
9527 (GET_MODE_2XWIDER_MODE): Likewise.
9528 (mode_iterator::get_wider): Update accordingly.
9529 (mode_iterator::get_2xwider): Likewise.
9530 (mode_iterator::get_known_wider): Likewise, turning into a template.
9531 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
9532 forcing a wider mode to exist.
9533 * config/cr16/cr16.h (LONG_REG_P): Likewise.
9534 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
9535 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
9536 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
9537 * lower-subreg.c (init_lower_subreg): Likewise.
9538 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
9539 on the final iteration.
9540 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
9541 a wider mode exists before asking for a move pattern.
9542 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
9543 forcing a wider mode to exist.
9544 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
9545 returning false if no such mode exists.
9546 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
9547 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
9548 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
9549 Avoid checking for a MODE_INT if we already know the mode is not a
9551 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
9552 forcing a wider mode to exist.
9553 (expmed_mult_highpart_optab): Likewise.
9554 (expmed_mult_highpart): Likewise.
9555 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
9557 * lto-streamer-in.c (lto_input_mode_table): Likewise.
9558 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
9559 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
9560 * internal-fn.c (expand_mul_overflow): Update use of
9561 GET_MODE_2XWIDER_MODE.
9562 * omp-low.c (omp_clause_aligned_alignment): Likewise.
9563 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
9564 GET_MODE_WIDER_MODE.
9565 (convert_plusminus_to_widen): Likewise.
9566 * tree-switch-conversion.c (array_value_type): Likewise.
9567 * var-tracking.c (emit_note_insn_var_location): Likewise.
9568 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
9569 Return false inside rather than outside the loop if no wider mode
9571 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
9572 and GET_MODE_2XWIDER_MODE
9573 (can_compare_p): Use else_void.
9574 * gdbhooks.py (OptMachineModePrinter): New class.
9575 (build_pretty_printer): Use it for opt_mode.
9577 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9578 Alan Hayward <alan.hayward@arm.com>
9579 David Sherwood <david.sherwood@arm.com>
9581 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
9582 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
9584 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9585 Alan Hayward <alan.hayward@arm.com>
9586 David Sherwood <david.sherwood@arm.com>
9588 * machmode.h (mode_traits): New structure.
9589 (get_narrowest_mode): New function.
9590 (mode_iterator::start): Likewise.
9591 (mode_iterator::iterate_p): Likewise.
9592 (mode_iterator::get_wider): Likewise.
9593 (mode_iterator::get_known_wider): Likewise.
9594 (mode_iterator::get_2xwider): Likewise.
9595 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
9596 (FOR_EACH_MODE): Likewise.
9597 (FOR_EACH_MODE_FROM): Likewise.
9598 (FOR_EACH_MODE_UNTIL): Likewise.
9599 (FOR_EACH_WIDER_MODE): Likewise.
9600 (FOR_EACH_2XWIDER_MODE): Likewise.
9601 * builtins.c (expand_builtin_strlen): Use new mode iterators.
9602 * combine.c (simplify_comparison): Likewise
9603 * config/i386/i386.c (type_natural_mode): Likewise.
9604 * cse.c (cse_insn): Likewise.
9605 * dse.c (find_shift_sequence): Likewise.
9606 * emit-rtl.c (init_derived_machine_modes): Likewise.
9607 (init_emit_once): Likewise.
9608 * explow.c (hard_function_value): Likewise.
9609 * expmed.c (extract_fixed_bit_field_1): Likewise.
9610 (extract_bit_field_1): Likewise.
9611 (expand_divmod): Likewise.
9612 (emit_store_flag_1): Likewise.
9613 * expr.c (init_expr_target): Likewise.
9614 (convert_move): Likewise.
9615 (alignment_for_piecewise_move): Likewise.
9616 (widest_int_mode_for_size): Likewise.
9617 (emit_block_move_via_movmem): Likewise.
9618 (copy_blkmode_to_reg): Likewise.
9619 (set_storage_via_setmem): Likewise.
9620 (compress_float_constant): Likewise.
9621 * omp-low.c (omp_clause_aligned_alignment): Likewise.
9622 * optabs-query.c (get_best_extraction_insn): Likewise.
9623 * optabs.c (expand_binop): Likewise.
9624 (expand_twoval_unop): Likewise.
9625 (expand_twoval_binop): Likewise.
9626 (widen_leading): Likewise.
9627 (widen_bswap): Likewise.
9628 (expand_parity): Likewise.
9629 (expand_unop): Likewise.
9630 (prepare_cmp_insn): Likewise.
9631 (prepare_float_lib_cmp): Likewise.
9632 (expand_float): Likewise.
9633 (expand_fix): Likewise.
9634 (expand_sfix_optab): Likewise.
9635 * postreload.c (move2add_use_add2_insn): Likewise.
9636 * reg-stack.c (reg_to_stack): Likewise.
9637 * reginfo.c (choose_hard_reg_mode): Likewise.
9638 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
9639 * stor-layout.c (mode_for_size): Likewise.
9640 (smallest_mode_for_size): Likewise.
9641 (mode_for_vector): Likewise.
9642 (finish_bitfield_representative): Likewise.
9643 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
9644 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
9645 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
9646 * var-tracking.c (prepare_call_arguments): Likewise.
9648 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9649 Alan Hayward <alan.hayward@arm.com>
9650 David Sherwood <david.sherwood@arm.com>
9652 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
9653 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
9654 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
9655 * machmode.h (mode_size): Move earlier in file.
9656 (mode_precision): Likewise.
9657 (mode_inner): Likewise.
9658 (mode_nunits): Likewise.
9659 (mode_unit_size): Likewise.
9660 (unit_unit_precision): Likewise.
9661 (mode_wider): Likewise.
9662 (mode_2xwider): Likewise.
9663 (machine_mode): New class.
9664 (mode_to_bytes): New function.
9665 (mode_to_bits): Likewise.
9666 (mode_to_precision): Likewise.
9667 (mode_to_inner): Likewise.
9668 (mode_to_unit_size): Likewise.
9669 (mode_to_unit_precision): Likewise.
9670 (mode_to_nunits): Likewise.
9671 (GET_MODE_SIZE): Use mode_to_bytes.
9672 (GET_MODE_BITSIZE): Use mode_to_bits.
9673 (GET_MODE_PRECISION): Use mode_to_precision.
9674 (GET_MODE_INNER): Use mode_to_inner.
9675 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
9676 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
9677 (GET_MODE_NUNITS): Use mode_to_nunits.
9678 * system.h (ALWAYS_INLINE): New macro.
9679 * config/powerpcspe/powerpcspe-c.c
9680 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
9681 int for arg1_mode and arg2_mode.
9683 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9684 Alan Hayward <alan.hayward@arm.com>
9685 David Sherwood <david.sherwood@arm.com>
9687 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
9688 Prefix mode names with E_ in case statements.
9689 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9690 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
9691 (aarch64_split_simd_move): Likewise.
9692 (aarch64_gen_storewb_pair): Likewise.
9693 (aarch64_gen_loadwb_pair): Likewise.
9694 (aarch64_gen_store_pair): Likewise.
9695 (aarch64_gen_load_pair): Likewise.
9696 (aarch64_get_condition_code_1): Likewise.
9697 (aarch64_constant_pool_reload_icode): Likewise.
9698 (get_rsqrte_type): Likewise.
9699 (get_rsqrts_type): Likewise.
9700 (get_recpe_type): Likewise.
9701 (get_recps_type): Likewise.
9702 (aarch64_gimplify_va_arg_expr): Likewise.
9703 (aarch64_simd_container_mode): Likewise.
9704 (aarch64_emit_load_exclusive): Likewise.
9705 (aarch64_emit_store_exclusive): Likewise.
9706 (aarch64_expand_compare_and_swap): Likewise.
9707 (aarch64_gen_atomic_cas): Likewise.
9708 (aarch64_emit_bic): Likewise.
9709 (aarch64_emit_atomic_swap): Likewise.
9710 (aarch64_emit_atomic_load_op): Likewise.
9711 (aarch64_evpc_trn): Likewise.
9712 (aarch64_evpc_uzp): Likewise.
9713 (aarch64_evpc_zip): Likewise.
9714 (aarch64_evpc_ext): Likewise.
9715 (aarch64_evpc_rev): Likewise.
9716 (aarch64_evpc_dup): Likewise.
9717 (aarch64_gen_ccmp_first): Likewise.
9718 (aarch64_gen_ccmp_next): Likewise.
9719 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
9720 (alpha_emit_xfloating_libcall): Likewise.
9721 (emit_insxl): Likewise.
9722 (alpha_arg_type): Likewise.
9723 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
9724 (arc_preferred_simd_mode): Likewise.
9725 (arc_secondary_reload): Likewise.
9726 (get_arc_condition_code): Likewise.
9727 (arc_print_operand): Likewise.
9728 (arc_legitimate_constant_p): Likewise.
9729 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9730 * config/arc/arc.md (casesi_load): Likewise.
9731 (casesi_compact_jump): Likewise.
9732 * config/arc/predicates.md (proper_comparison_operator): Likewise.
9733 (cc_use_register): Likewise.
9734 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9735 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
9736 (arm_init_iwmmxt_builtins): Likewise.
9737 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
9738 (neon_expand_vector_init): Likewise.
9739 (arm_attr_length_move_neon): Likewise.
9740 (maybe_get_arm_condition_code): Likewise.
9741 (arm_emit_vector_const): Likewise.
9742 (arm_preferred_simd_mode): Likewise.
9743 (arm_output_iwmmxt_tinsr): Likewise.
9744 (thumb1_output_casesi): Likewise.
9745 (thumb2_output_casesi): Likewise.
9746 (arm_emit_load_exclusive): Likewise.
9747 (arm_emit_store_exclusive): Likewise.
9748 (arm_expand_compare_and_swap): Likewise.
9749 (arm_evpc_neon_vuzp): Likewise.
9750 (arm_evpc_neon_vzip): Likewise.
9751 (arm_evpc_neon_vrev): Likewise.
9752 (arm_evpc_neon_vtrn): Likewise.
9753 (arm_evpc_neon_vext): Likewise.
9754 (arm_validize_comparison): Likewise.
9755 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
9756 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
9757 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
9758 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
9759 (c6x_preferred_simd_mode): Likewise.
9760 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
9761 (epiphany_rtx_costs): Likewise.
9762 * config/epiphany/predicates.md (proper_comparison_operator):
9764 * config/frv/frv.c (condexec_memory_operand): Likewise.
9765 (frv_emit_move): Likewise.
9766 (output_move_single): Likewise.
9767 (output_condmove_single): Likewise.
9768 (frv_hard_regno_mode_ok): Likewise.
9769 (frv_matching_accg_mode): Likewise.
9770 * config/h8300/h8300.c (split_adds_subs): Likewise.
9771 (h8300_rtx_costs): Likewise.
9772 (h8300_print_operand): Likewise.
9773 (compute_mov_length): Likewise.
9774 (output_logical_op): Likewise.
9775 (compute_logical_op_length): Likewise.
9776 (compute_logical_op_cc): Likewise.
9777 (h8300_shift_needs_scratch_p): Likewise.
9778 (output_a_shift): Likewise.
9779 (compute_a_shift_length): Likewise.
9780 (compute_a_shift_cc): Likewise.
9781 (expand_a_rotate): Likewise.
9782 (output_a_rotate): Likewise.
9783 * config/i386/i386.c (classify_argument): Likewise.
9784 (function_arg_advance_32): Likewise.
9785 (function_arg_32): Likewise.
9786 (function_arg_64): Likewise.
9787 (function_value_64): Likewise.
9788 (ix86_gimplify_va_arg): Likewise.
9789 (ix86_legitimate_constant_p): Likewise.
9790 (put_condition_code): Likewise.
9791 (split_double_mode): Likewise.
9792 (ix86_avx256_split_vector_move_misalign): Likewise.
9793 (ix86_expand_vector_logical_operator): Likewise.
9794 (ix86_split_idivmod): Likewise.
9795 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
9796 (ix86_build_const_vector): Likewise.
9797 (ix86_build_signbit_mask): Likewise.
9798 (ix86_match_ccmode): Likewise.
9799 (ix86_cc_modes_compatible): Likewise.
9800 (ix86_expand_branch): Likewise.
9801 (ix86_expand_sse_cmp): Likewise.
9802 (ix86_expand_sse_movcc): Likewise.
9803 (ix86_expand_int_sse_cmp): Likewise.
9804 (ix86_expand_vec_perm_vpermi2): Likewise.
9805 (ix86_expand_vec_perm): Likewise.
9806 (ix86_expand_sse_unpack): Likewise.
9807 (ix86_expand_int_addcc): Likewise.
9808 (ix86_split_to_parts): Likewise.
9809 (ix86_vectorize_builtin_gather): Likewise.
9810 (ix86_vectorize_builtin_scatter): Likewise.
9811 (avx_vpermilp_parallel): Likewise.
9812 (inline_memory_move_cost): Likewise.
9813 (ix86_tieable_integer_mode_p): Likewise.
9814 (x86_maybe_negate_const_int): Likewise.
9815 (ix86_expand_vector_init_duplicate): Likewise.
9816 (ix86_expand_vector_init_one_nonzero): Likewise.
9817 (ix86_expand_vector_init_one_var): Likewise.
9818 (ix86_expand_vector_init_concat): Likewise.
9819 (ix86_expand_vector_init_interleave): Likewise.
9820 (ix86_expand_vector_init_general): Likewise.
9821 (ix86_expand_vector_set): Likewise.
9822 (ix86_expand_vector_extract): Likewise.
9823 (emit_reduc_half): Likewise.
9824 (ix86_emit_i387_round): Likewise.
9825 (ix86_mangle_type): Likewise.
9826 (ix86_expand_round_sse4): Likewise.
9827 (expand_vec_perm_blend): Likewise.
9828 (canonicalize_vector_int_perm): Likewise.
9829 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
9830 (expand_vec_perm_1): Likewise.
9831 (expand_vec_perm_interleave3): Likewise.
9832 (expand_vec_perm_even_odd_pack): Likewise.
9833 (expand_vec_perm_even_odd_1): Likewise.
9834 (expand_vec_perm_broadcast_1): Likewise.
9835 (ix86_vectorize_vec_perm_const_ok): Likewise.
9836 (ix86_expand_vecop_qihi): Likewise.
9837 (ix86_expand_mul_widen_hilo): Likewise.
9838 (ix86_expand_sse2_abs): Likewise.
9839 (ix86_expand_pextr): Likewise.
9840 (ix86_expand_pinsr): Likewise.
9841 (ix86_preferred_simd_mode): Likewise.
9842 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
9843 * config/i386/sse.md (*andnot<mode>3): Likewise.
9844 (<mask_codefor><code><mode>3<mask_name>): Likewise.
9845 (*<code><mode>3): Likewise.
9846 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
9847 (ia64_expand_atomic_op): Likewise.
9848 (ia64_arg_type): Likewise.
9849 (ia64_mode_to_int): Likewise.
9850 (ia64_scalar_mode_supported_p): Likewise.
9851 (ia64_vector_mode_supported_p): Likewise.
9852 (expand_vec_perm_broadcast): Likewise.
9853 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
9854 (iq2000_function_arg_advance): Likewise.
9855 (iq2000_function_arg): Likewise.
9856 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
9857 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
9858 (m68k_libcall_value): Likewise.
9859 (m68k_function_value): Likewise.
9860 (sched_attr_op_type): Likewise.
9861 * config/mcore/mcore.c (mcore_output_move): Likewise.
9862 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
9864 (microblaze_function_arg): Likewise.
9865 * config/mips/mips.c (mips16_build_call_stub): Likewise.
9866 (mips_print_operand): Likewise.
9867 (mips_mode_ok_for_mov_fmt_p): Likewise.
9868 (mips_vector_mode_supported_p): Likewise.
9869 (mips_preferred_simd_mode): Likewise.
9870 (mips_expand_vpc_loongson_even_odd): Likewise.
9871 (mips_expand_vec_unpack): Likewise.
9872 (mips_expand_vi_broadcast): Likewise.
9873 (mips_expand_vector_init): Likewise.
9874 (mips_expand_vec_reduc): Likewise.
9875 (mips_expand_msa_cmp): Likewise.
9876 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
9877 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
9878 (cc_flags_for_mode): Likewise.
9879 * config/msp430/msp430.c (msp430_print_operand): Likewise.
9880 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
9881 (nds32_output_casesi_pc_relative): Likewise.
9882 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9883 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
9884 (nvptx_gen_unpack): Likewise.
9885 (nvptx_gen_pack): Likewise.
9886 (nvptx_gen_shuffle): Likewise.
9887 (nvptx_gen_wcast): Likewise.
9888 (nvptx_preferred_simd_mode): Likewise.
9889 * config/pa/pa.c (pa_secondary_reload): Likewise.
9890 * config/pa/predicates.md (base14_operand): Likewise.
9891 * config/powerpcspe/powerpcspe-c.c
9892 (altivec_resolve_overloaded_builtin): Likewise.
9893 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
9895 (rs6000_preferred_simd_mode): Likewise.
9896 (output_vec_const_move): Likewise.
9897 (rs6000_expand_vector_extract): Likewise.
9898 (rs6000_split_vec_extract_var): Likewise.
9899 (reg_offset_addressing_ok_p): Likewise.
9900 (rs6000_legitimate_offset_address_p): Likewise.
9901 (rs6000_legitimize_address): Likewise.
9902 (rs6000_emit_set_const): Likewise.
9903 (rs6000_const_vec): Likewise.
9904 (rs6000_emit_move): Likewise.
9905 (spe_build_register_parallel): Likewise.
9906 (rs6000_darwin64_record_arg_recurse): Likewise.
9907 (swap_selector_for_mode): Likewise.
9908 (spe_init_builtins): Likewise.
9909 (paired_init_builtins): Likewise.
9910 (altivec_init_builtins): Likewise.
9911 (do_load_for_compare): Likewise.
9912 (rs6000_generate_compare): Likewise.
9913 (rs6000_expand_float128_convert): Likewise.
9914 (emit_load_locked): Likewise.
9915 (emit_store_conditional): Likewise.
9916 (rs6000_output_function_epilogue): Likewise.
9917 (rs6000_handle_altivec_attribute): Likewise.
9918 (rs6000_function_value): Likewise.
9919 (emit_fusion_gpr_load): Likewise.
9920 (emit_fusion_p9_load): Likewise.
9921 (emit_fusion_p9_store): Likewise.
9922 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
9923 (fusion_gpr_mem_load): Likewise.
9924 (fusion_addis_mem_combo_load): Likewise.
9925 (fusion_addis_mem_combo_store): Likewise.
9926 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
9927 (fusion_gpr_mem_load): Likewise.
9928 (fusion_addis_mem_combo_load): Likewise.
9929 (fusion_addis_mem_combo_store): Likewise.
9930 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9932 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
9933 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
9934 (rs6000_preferred_simd_mode): Likewise.
9935 (output_vec_const_move): Likewise.
9936 (rs6000_expand_vector_extract): Likewise.
9937 (rs6000_split_vec_extract_var): Likewise.
9938 (reg_offset_addressing_ok_p): Likewise.
9939 (rs6000_legitimate_offset_address_p): Likewise.
9940 (rs6000_legitimize_address): Likewise.
9941 (rs6000_emit_set_const): Likewise.
9942 (rs6000_const_vec): Likewise.
9943 (rs6000_emit_move): Likewise.
9944 (rs6000_darwin64_record_arg_recurse): Likewise.
9945 (swap_selector_for_mode): Likewise.
9946 (paired_init_builtins): Likewise.
9947 (altivec_init_builtins): Likewise.
9948 (rs6000_expand_float128_convert): Likewise.
9949 (emit_load_locked): Likewise.
9950 (emit_store_conditional): Likewise.
9951 (rs6000_output_function_epilogue): Likewise.
9952 (rs6000_handle_altivec_attribute): Likewise.
9953 (rs6000_function_value): Likewise.
9954 (emit_fusion_gpr_load): Likewise.
9955 (emit_fusion_p9_load): Likewise.
9956 (emit_fusion_p9_store): Likewise.
9957 * config/rx/rx.c (rx_gen_move_template): Likewise.
9958 (flags_from_mode): Likewise.
9959 * config/s390/predicates.md (s390_alc_comparison): Likewise.
9960 (s390_slb_comparison): Likewise.
9961 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
9962 (s390_vector_mode_supported_p): Likewise.
9963 (s390_cc_modes_compatible): Likewise.
9964 (s390_match_ccmode_set): Likewise.
9965 (s390_canonicalize_comparison): Likewise.
9966 (s390_emit_compare_and_swap): Likewise.
9967 (s390_branch_condition_mask): Likewise.
9968 (s390_rtx_costs): Likewise.
9969 (s390_secondary_reload): Likewise.
9970 (__SECONDARY_RELOAD_CASE): Likewise.
9971 (s390_expand_cs): Likewise.
9972 (s390_preferred_simd_mode): Likewise.
9973 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
9974 * config/sh/sh.c (sh_print_operand): Likewise.
9975 (dump_table): Likewise.
9976 (sh_secondary_reload): Likewise.
9977 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9978 * config/sh/sh.md (casesi_worker_1): Likewise.
9979 (casesi_worker_2): Likewise.
9980 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
9981 (fcc_comparison_operator): Likewise.
9982 * config/sparc/sparc.c (sparc_expand_move): Likewise.
9983 (emit_soft_tfmode_cvt): Likewise.
9984 (sparc_preferred_simd_mode): Likewise.
9985 (output_cbranch): Likewise.
9986 (sparc_print_operand): Likewise.
9987 (sparc_expand_vec_perm_bmask): Likewise.
9988 (vector_init_bshuffle): Likewise.
9989 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
9990 (spu_vector_mode_supported_p): Likewise.
9991 (spu_expand_insv): Likewise.
9992 (spu_emit_branch_or_set): Likewise.
9993 (spu_handle_vector_attribute): Likewise.
9994 (spu_builtin_splats): Likewise.
9995 (spu_builtin_extract): Likewise.
9996 (spu_builtin_promote): Likewise.
9997 (spu_expand_sign_extend): Likewise.
9998 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
9999 (tilegx_simd_int): Likewise.
10000 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
10001 (tilepro_simd_int): Likewise.
10002 * config/v850/v850.c (const_double_split): Likewise.
10003 (v850_print_operand): Likewise.
10004 (ep_memory_offset): Likewise.
10005 * config/vax/vax.c (vax_rtx_costs): Likewise.
10006 (vax_output_int_move): Likewise.
10007 (vax_output_int_add): Likewise.
10008 (vax_output_int_subtract): Likewise.
10009 * config/visium/predicates.md (visium_branch_operator): Likewise.
10010 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
10011 (visium_print_operand_address): Likewise.
10012 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10013 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
10014 (xtensa_expand_conditional_branch): Likewise.
10015 (xtensa_copy_incoming_a7): Likewise.
10016 (xtensa_output_literal): Likewise.
10017 * dfp.c (decimal_real_maxval): Likewise.
10018 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
10020 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10021 Alan Hayward <alan.hayward@arm.com>
10022 David Sherwood <david.sherwood@arm.com>
10024 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
10025 (mode_nunits_inline): Likewise.
10026 (mode_inner_inline): Likewise.
10027 (mode_unit_size_inline): Likewise.
10028 (mode_unit_precision_inline): Likewise.
10029 (emit_insn_modes_h): Likewise. Also emit a #define of the
10031 (emit_mode_wider): Add an E_ prefix to mode names.
10032 (emit_mode_complex): Likewise.
10033 (emit_mode_inner): Likewise.
10034 (emit_mode_adjustments): Likewise.
10035 (emit_mode_int_n): Likewise.
10036 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
10037 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
10038 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
10039 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
10040 (CRC32_BUILTIN, ENTRY): Likewise.
10041 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
10042 (aarch64_pop_regs): Likewise.
10043 (aarch64_process_components): Likewise.
10044 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
10045 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
10046 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
10047 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
10048 * config/arm/arm.c (arm_init_libfuncs): Likewise.
10049 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
10051 * config/i386/i386-builtin.def (pcmpestr): Likewise.
10052 (pcmpistr): Likewise.
10053 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
10054 * config/mmix/mmix.c (mmix_output_condition): Likewise.
10055 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
10057 * config/rl78/rl78.c (mduc_regs): Likewise.
10058 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
10059 (htm_expand_builtin): Likewise.
10060 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
10061 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
10062 * config/xtensa/xtensa.c (print_operand): Likewise.
10063 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
10064 (NUM_MODE_VECTOR_INT): Likewise.
10065 * genoutput.c (null_operand): Likewise.
10066 (output_operand_data): Likewise.
10067 * genrecog.c (print_parameter_value): Likewise.
10068 * lra.c (debug_operand_data): Likewise.
10070 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10072 * dwarf2out.c (scompare_loc_descriptor_wide)
10073 (scompare_loc_descriptor_narrow): New functions, split out from...
10074 (scompare_loc_descriptor): ...here.
10075 * expmed.c (emit_store_flag_int): New function, split out from...
10076 (emit_store_flag): ...here.
10078 2017-08-30 Richard Biener <rguenther@suse.de>
10080 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
10081 (dwarf2out_early_finish): Move setting of AT_pubnames from
10082 early debug output to early finish.
10084 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
10086 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
10087 and -mdata-region to the assembler.
10089 2017-08-30 Richard Biener <rguenther@suse.de>
10091 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
10093 (gen_subprogram_die): Add DW_AT_object_pointer only early.
10094 (dwarf2out_early_global_decl): Only generate a DIE for the
10095 abstract origin if it doesn't already exist or is a declaration DIE.
10096 (resolve_addr): Do not add the linkage name twice when
10097 generating a stub DIE for the DW_TAG_GNU_call_site target.
10099 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10101 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10102 Use machine_mode rather than int for arg1_mode.
10104 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
10107 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
10108 that the second argument of the built-in functions to unpack
10109 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
10110 switch statement instead a lot of if statements.
10111 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
10112 Allow 64-bit values to be in Altivec registers as well as
10113 traditional floating point registers.
10114 (pack<mode>, FMOVE128_VSX iterator): Likewise.
10116 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
10118 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
10119 MAX_REGS_PER_ADDRESS == 1.
10121 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
10123 * config/i386/i386.opt (flag_fentry): Do not init to -1.
10124 * config/i386/i386.c (ix86_option_override_internal): Simplify
10125 setting of opts->x_flag_entry.
10127 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10128 Jakub Jelinek <jakub@redhat.com>
10129 Richard Biener <rguenther@suse.de>
10131 PR tree-optimization/81503
10132 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
10133 folded constant fits in the target type; reorder tests for clarity.
10135 2017-08-29 Martin Liska <mliska@suse.cz>
10137 * passes.def: Include pass_lower_switch.
10138 * stmt.c (dump_case_nodes): Remove and move to
10139 tree-switch-conversion.
10140 (case_values_threshold): Likewise.
10141 (expand_switch_as_decision_tree_p): Likewise.
10142 (emit_case_decision_tree): Likewise.
10143 (expand_case): Likewise.
10144 (balance_case_nodes): Likewise.
10145 (node_has_low_bound): Likewise.
10146 (node_has_high_bound): Likewise.
10147 (node_is_bounded): Likewise.
10148 (emit_case_nodes): Likewise.
10149 (struct simple_case_node): New struct.
10150 (add_case_node): Remove.
10151 (emit_case_dispatch_table): Use vector instead of case_list.
10152 (reset_out_edges_aux): Remove.
10153 (compute_cases_per_edge): Likewise.
10154 (expand_case): Build list of simple_case_node.
10155 (expand_sjlj_dispatch_table): Use it.
10156 * tree-switch-conversion.c (struct case_node): Moved from
10157 stmt.c and adjusted.
10158 (emit_case_nodes): Likewise.
10159 (node_has_low_bound): Likewise.
10160 (node_has_high_bound): Likewise.
10161 (node_is_bounded): Likewise.
10162 (case_values_threshold): Likewise.
10163 (reset_out_edges_aux): Likewise.
10164 (compute_cases_per_edge): Likewise.
10165 (add_case_node): Likewise.
10166 (dump_case_nodes): Likewise.
10167 (balance_case_nodes): Likewise.
10168 (expand_switch_as_decision_tree_p): Likewise.
10169 (emit_jump): Likewise.
10170 (emit_case_decision_tree): Likewise.
10171 (try_switch_expansion): Likewise.
10172 (do_jump_if_equal): Likewise.
10173 (emit_cmp_and_jump_insns): Likewise.
10174 (fix_phi_operands_for_edge): New function.
10175 (record_phi_operand_mapping): Likewise.
10176 (class pass_lower_switch): New pass.
10177 (pass_lower_switch::execute): New function.
10178 (make_pass_lower_switch): Likewise.
10179 (conditional_probability):
10180 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
10181 * tree-pass.h: Add make_pass_lower_switch.
10183 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
10186 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
10189 2017-08-29 Richard Biener <rguenther@suse.de>
10191 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
10192 we do not add a DW_AT_inline attribute twice.
10193 (gen_subprogram_die): Remove code setting DW_AT_inline on
10194 DECL_ABSTRACT_P nodes.
10196 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
10198 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
10199 calls to internal functions.
10200 (gimplify_modify_expr): Likewise.
10201 * tree-call-cdce.c (use_internal_fn): Likewise.
10202 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
10203 (convert_to_divmod): Set the nothrow flag.
10204 * tree-if-conv.c (predicate_mem_writes): Likewise.
10205 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
10206 (vectorizable_call): Likewise.
10207 (vectorizable_store): Likewise.
10208 (vectorizable_load): Likewise.
10209 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
10210 (vect_recog_mask_conversion_pattern): Likewise.
10212 2017-08-29 Martin Liska <mliska@suse.cz>
10215 * gcc.c (driver_handle_option): Add new argument.
10216 * opts-common.c (handle_option): Pass
10217 target_option_override_hook.
10218 * opts-global.c (lang_handle_option): Add new option.
10219 (set_default_handlers): Add new argument.
10220 (decode_options): Likewise.
10221 * opts.c (target_handle_option): Likewise.
10222 (common_handle_option): Call target_option_override_hook.
10223 * opts.h (struct cl_option_handler_func): Add hook for
10224 target option override.
10225 (struct cl_option_handlers): Likewise.
10226 (set_default_handlers): Add new argument.
10227 (decode_options): Likewise.
10228 (common_handle_option): Likewise.
10229 (target_handle_option): Likewise.
10230 * toplev.c (toplev::main): Pass targetm.target_option.override
10233 2017-08-29 Richard Biener <rguenther@suse.de>
10234 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
10236 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
10237 life to the active subtree.
10239 2017-08-28 Jeff Law <law@redhat.com>
10241 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
10242 derive_equivalences.
10243 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
10244 Code moved into....
10245 (edge_info::derive_equivalences): New private member function
10247 * tree-ssa-dom.c (class edge_info): Changed from a struct
10248 to a class. Add ctor/dtor, methods and data members.
10249 (edge_info::edge_info): Renamed from allocate_edge_info.
10250 Initialize additional members.
10251 (edge_info::~edge_info): New.
10252 (free_dom_edge_info): Delete the edge info.
10253 (record_edge_info): Use new class & associated member functions.
10254 Tighten forms for testing for edge equivalences.
10255 (record_temporary_equivalences): Iterate over the simple
10256 equivalences rather than assuming there's only one per edge.
10257 (cprop_into_successor_phis): Iterate over the simple
10258 equivalences rather than assuming there's only one per edge.
10259 (optimize_stmt): Use operand_equal_p rather than pointer
10260 equality for mini-DSE code.
10262 2017-08-28 Nathan Sidwell <nathan@acm.org>
10264 * gcc.c (execute): Fold SIGPIPE handling into switch
10265 statement. Adjust internal error message.
10267 2017-08-28 Richard Biener <rguenther@suse.de>
10270 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
10271 Do nothing for removed DIEs.
10273 2017-08-28 Richard Biener <rguenther@suse.de>
10275 PR tree-optimization/81977
10276 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
10279 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
10282 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
10283 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
10284 using targetm.gen_mem_thread_fence.
10286 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
10289 * config/i386/i386.md (*<btsc><mode>): Change operand 2
10290 predicate to register_operand. Reorder operands.
10291 (*btr<mode>): Ditto.
10292 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
10293 (*btr<mode>_mask): Ditto.
10295 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
10297 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
10298 * config/rs6000/xmmintrin.h: New file.
10299 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
10301 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10304 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
10305 parameter and_insn and return it.
10306 (recombine_lvx_pattern): Insert a copy to ensure availability of
10307 the base register of the copied masking operation at the point of
10308 the instruction replacement.
10309 (recombine_stvx_pattern): Likewise.
10311 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
10313 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
10314 undocumented switches.
10315 (-mpower9-dform-vector): Likewise.
10316 (-mpower9-dform): Likewise.
10317 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
10318 comments to delete references to -mpower9-dform* switches.
10319 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
10320 Delete reference to -mpower9-dform* switches, test for
10321 -mpower9-vector instead.
10322 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
10323 (OTHER_P9_VECTOR_MASKS): Likewise.
10324 (POWERPC_MASKS): Likewise.
10325 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
10326 tests against -mpower9-dform* to -mpower9-vector. Delete code
10327 that checked for -mpower9-dform* consistancy with other options.
10328 Add test for -mpower9-misc to enable other power9 switches.
10329 (rs6000_init_hard_regno_mode_ok): Likewise.
10330 (rs6000_option_override_internal): Likewise.
10331 (rs6000_emit_prologue): Likewise.
10332 (rs6000_emit_epilogue): Likewise.
10333 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
10334 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
10335 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
10337 (emit_fusion_p9_store): Likewise.
10338 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
10339 resetting these macros if the assembler does not support ISA 3.0
10341 (TARGET_P9_DFORM_VECTOR): Likewise.
10342 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
10343 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
10345 2017-08-25 Alan Modra <amodra@gmail.com>
10347 PR rtl-optimization/81747
10348 * cse.c (cse_extended_basic_block): Don't attempt to record
10349 equivalences for degenerate conditional branches that branch
10350 to their fall-through.
10352 2017-08-24 Martin Sebor <msebor@redhat.com>
10354 PR middle-end/81908
10355 * gimple-fold.c (size_must_be_zero_p): New function.
10356 (gimple_fold_builtin_memory_op): Call it.
10358 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
10360 * config/rs6000/mm_malloc.h: New file.
10362 2017-08-24 Bin Cheng <bin.cheng@arm.com>
10364 PR tree-optimization/81913
10365 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
10366 analysis when either IVs in condition can wrap.
10368 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
10370 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
10371 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
10373 2017-08-24 Richard Biener <rguenther@suse.de>
10376 * targhooks.c (default_target_can_inline_p): Properly
10377 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
10378 is present and always compare.
10379 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
10380 infer -mfpmath=sse from TARGET_SSE_P.
10381 (ix86_can_inline_p): Properly use target_option_default_node when
10382 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
10384 2017-08-24 Richard Biener <rguenther@suse.de>
10387 * dwarf2out.c (output_die): Handle flag_generate_offload like
10389 (output_comp_unit): Likewise.
10390 (gen_array_type_die): Likewise.
10391 (dwarf2out_early_finish): Likewise.
10392 (note_variable_value_in_expr): Likewise.
10393 (dwarf2out_finish): Likewise. Adjust assert.
10394 * cgraphunit.c (symbol_table::compile): Move setting of
10395 flag_generate_offload earlier ...
10396 (symbol_table::finalize_compilation_unit): ... here, before
10397 early debug finalization.
10399 2017-08-24 Richard Biener <rguenther@suse.de>
10401 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
10402 and ipa-fnsummary.h.
10403 (ix86_can_inline_p): When ix86_fpmath flags do not match
10404 check whether the callee uses FP math at all.
10406 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
10408 PR middle-end/81931
10409 * tree-ssanames.c (get_nonzero_bits): Use element_precision
10410 instead of TYPE_PRECISION.
10412 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
10413 Alan Hayward <alan.hayward@arm.com>
10414 David Sherwood <david.sherwood@arm.com>
10416 * combine.c (make_extraction): Use subreg_offset_from_lsb.
10418 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
10420 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
10422 (ix86_frame::stack_realign_allocate): New field.
10423 (struct machine_frame_state): Modify comments.
10424 (machine_frame_state::sp_realigned_fp_end): New field.
10425 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
10426 layout calculation.
10427 (sp_valid_at): Add assertion to assure no attempt to access invalid
10428 offset of a realigned stack.
10429 (fp_valid_at): Likewise.
10430 (choose_baseaddr): Modify comments.
10431 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
10432 ix86_expand_prologue.
10433 (ix86_expand_prologue): Modify stack realignment and allocation.
10434 (ix86_expand_epilogue): Modify comments.
10435 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
10436 avx2_runtime, avx512f, and avx512f_runtime.
10438 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
10440 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
10441 (mstackrealign): Do not init to -1.
10442 * config/i386/i386.c (ix86_option_override_internal):
10443 Check opts_set, not opts when setting default value of
10444 opts->x_ix86_force_align_arg_pointer.
10446 2017-08-23 Richard Biener <rguenther@suse.de>
10448 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
10449 lang_hooks.decl_printable_name.
10450 * print-rtl-function.c (print_rtx_function): Likewise.
10451 * tree-pretty-print.c (dump_function_header): Likewise.
10453 2017-08-23 Richard Biener <rguenther@suse.de>
10456 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
10457 -g0 at compile-time.
10459 2017-08-23 Tamar Christina <tamar.christina@arm.com>
10461 PR middle-end/19706
10462 * doc/sourcebuild.texi (Other hardware attributes):
10465 2017-08-23 Tamar Christina <tamar.christina@arm.com>
10467 PR middle-end/19706
10468 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
10469 Fix single-use check.
10471 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
10473 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
10475 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
10477 * doc/install.texi: Modify to add more details on running selected
10480 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
10482 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
10483 is combined with -mabi=ms.
10484 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
10487 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10489 PR tree-optimization/81488
10490 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
10491 and cached_basis fields.
10492 (MAX_SPREAD): New constant.
10493 (alloc_cand_and_find_basis): Initialize new fields.
10494 (clear_visited): New function.
10495 (create_phi_basis_1): Rename from create_phi_basis, set visited
10496 and cached_basis fields.
10497 (create_phi_basis): New wrapper function.
10498 (phi_add_costs_1): Rename from phi_add_costs, add spread
10499 parameter, set visited field, short-circuit when limits reached.
10500 (phi_add_costs): New wrapper function.
10501 (record_phi_increments_1): Rename from record_phi_increments, set
10503 (record_phi_increments): New wrapper function.
10504 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
10505 (phi_incr_cost): New wrapper function.
10506 (all_phi_incrs_profitable_1): Rename from
10507 all_phi_incrs_profitable, set visited field.
10508 (all_phi_incrs_profitable): New wrapper function.
10510 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
10511 Alan Hayward <alan.hayward@arm.com>
10512 David Sherwood <david.sherwood@arm.com>
10514 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
10515 that takes the outer and inner modes.
10516 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
10517 comparison as the canonical test for a paradoxical subreg.
10518 * combine.c (simplify_set): Use paradoxical_subreg_p.
10519 (make_extraction): Likewise.
10520 (force_to_mode): Likewise.
10521 (rtx_equal_for_field_assignment_p): Likewise.
10522 (gen_lowpart_for_combine): Likewise.
10523 (simplify_comparison): Likewise.
10524 * cse.c (equiv_constant): Likewise.
10525 * expmed.c (store_bit_field_1): Likewise.
10526 * final.c (alter_subreg): Likewise.
10527 * fwprop.c (propagate_rtx): Likewise.
10528 (forward_propagate_subreg): Likewise.
10529 * ira-conflicts.c (ira_build_conflicts): Likewise.
10530 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
10531 * lra-constraints.c (curr_insn_transform): Likewise.
10532 (split_reg): Likewise.
10533 * lra-eliminations.c (move_plus_up): Likewise.
10534 (lra_eliminate_regs_1): Likewise.
10535 * recog.c (general_operand): Likewise.
10536 * ree.c (combine_reaching_defs): Likewise.
10537 * reload.c (push_reload): Likewise.
10538 (find_reloads): Likewise.
10539 * reload1.c (elimination_effects): Likewise.
10540 (compute_reload_subreg_offset): Likewise.
10541 (choose_reload_regs): Likewise.
10542 * rtlanal.c (subreg_lsb_1): Likewise.
10543 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10544 (simplify_subreg): Likewise.
10545 * var-tracking.c (track_loc_p): Likewise.
10546 * emit-rtl.c (byte_lowpart_offset): Likewise.
10547 (paradoxical_subreg_p): Delete out-of-line definition.
10549 2017-08-22 Jeff Law <law@redhat.com>
10551 PR tree-optimization/81741
10552 PR tree-optimization/71947
10553 * tree-ssa-dom.c: Include tree-inline.h.
10554 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
10555 equivalences if one is more expensive to compute than the other.
10556 * tree-ssa-scopedtables.h (class const_or_copies): Make
10557 record_const_or_copy_raw method private.
10558 (class avail_exprs_stack): New method simplify_binary_operation.
10559 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
10560 avail_exprs_stack::simplify_binary_operation as needed.
10561 (avail_exprs_stack::simplify_binary_operation): New function.
10563 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
10565 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
10566 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
10567 (DOT_SYMBOLS): Likewise.
10568 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
10569 (RELOCATABLE_NEEDS_FIXUP): Likewise.
10570 (RS6000_ABI_NAME): Likewise.
10571 (TARGET_CMODEL): Likewise.
10572 (TOC_SECTION_ASM_OP): Likewise.
10573 (SET_CMODEL): New macro.
10574 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
10576 2017-08-22 Richard Biener <rguenther@suse.de>
10578 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
10579 to allow for free-lang-data replacements similar to verify_type_variant.
10581 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
10582 Alan Hayward <alan.hayward@arm.com>
10583 David Sherwood <david.sherwood@arm.com>
10585 * config/aarch64/aarch64.md (casesi): Use DImode rather than
10586 VOIDmode for the LABEL_REF.
10588 2017-08-22 Richard Biener <rguenther@suse.de>
10590 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
10592 2017-08-22 Richard Biener <rguenther@suse.de>
10594 * common.opt (feliminate-dwarf2-dups): Ignore.
10595 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
10596 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
10597 same_die_p_wrap, compute_section_prefix,
10598 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
10599 (comdat_symbol_id, comdat_symbol_number): Likewise.
10600 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
10602 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
10603 (output_die): Mark unreachable path unreachable.
10604 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
10605 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
10606 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
10607 (dwarf2out_early_finish): Likewise.
10609 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
10611 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
10613 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
10616 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
10617 not VAR_P. Filter attribute warnings with OPT_Wattributes.
10618 (avr_attribute_table) <io, io_low, address>: Initialize
10619 .decl_required with true.
10621 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
10623 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
10624 undocumented debugging options.
10625 (-mvsx-scalar-double): Likewise.
10626 (-mallow-df-permute): Likewise.
10627 (-mvectorize-builtins): Likewise.
10628 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
10629 (rs6000_builtin_vectorized_function): Likewise.
10630 (rs6000_builtin_md_vectorized_function): Likewise.
10631 (rs6000_opt_vars): Likewise.
10633 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
10636 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
10637 (*btrq_imm): Rename from *btrq.
10638 (*btcq_imm): Rename from *btcq.
10639 (btsc): New code attribute.
10640 (*<btsc><mode>): New insn pattern.
10641 (*btr<mode>): Ditto.
10642 (*<btsc><mode>_mask): New insn_and_split pattern.
10643 (*btr<mode>_mask): Ditto.
10645 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10646 Alan Hayward <alan.hayward@arm.com>
10647 David Sherwood <david.sherwood@arm.com>
10649 * function.c (pad_below): Simplify padding calculation.
10651 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10652 Alan Hayward <alan.hayward@arm.com>
10653 David Sherwood <david.sherwood@arm.com>
10655 * target.def (function_prologue): Remove frame size argument.
10656 (function_epilogue): Likewise.
10657 * doc/tm.texi: Regenerate.
10658 * final.c (final_start_function): Update call to function_prologue.
10659 (final_end_function): Update call to function_epilogue.
10660 (default_function_pro_epilogue): Remove frame size argument.
10661 * output.h (default_function_pro_epilogue): Likewise.
10662 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
10663 (arm_output_function_prologue): Likewise.
10664 * config/frv/frv.c (frv_function_prologue): Likewise.
10665 (frv_function_epilogue): Likewise.
10666 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
10667 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
10668 (ia64_output_function_epilogue): Likewise.
10669 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
10670 (m32r_output_function_epilogue): Likewise.
10671 * config/microblaze/microblaze.c (microblaze_function_prologue)
10672 (microblaze_function_epilogue): Likewise.
10673 * config/mips/mips.c (mips_output_function_prologue): Likewise.
10674 (mips_output_function_epilogue): Likewise.
10675 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
10676 (mmix_target_asm_function_epilogue): Likewise.
10677 * config/msp430/msp430.c (msp430_start_function): Likewise.
10678 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
10679 (nds32_asm_function_epilogue): Likewise.
10680 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
10681 * config/pa/pa.c (pa_output_function_prologue): Likewise.
10682 (pa_output_function_epilogue): Likewise.
10683 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
10684 (rs6000_output_function_epilogue): Likewise.
10685 * config/rl78/rl78.c (rl78_start_function): Likewise.
10686 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
10687 (rs6000_output_function_epilogue): Likewise.
10688 * config/rx/rx.c (rx_output_function_prologue): Likewise.
10689 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
10690 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
10691 (sparc_asm_function_epilogue): Likewise.
10693 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10695 * tree.h (type_has_mode_precision_p): New function.
10696 * convert.c (convert_to_integer_1): Use it.
10697 * expr.c (expand_expr_real_2): Likewise.
10698 (expand_expr_real_1): Likewise.
10699 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
10700 * match.pd: Likewise.
10701 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
10702 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
10703 * tree-tailcall.c (process_assignment): Likewise.
10704 * tree-vect-loop.c (vectorizable_reduction): Likewise.
10705 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
10706 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
10707 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10708 (vectorizable_assignment): Likewise.
10709 (vectorizable_shift): Likewise.
10710 (vectorizable_operation): Likewise.
10711 * tree-vrp.c (register_edge_assert_for_2): Likewise.
10713 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
10715 * match.pd: Add pow (C, x) simplification.
10717 2017-08-21 Richard Biener <rguenther@suse.de>
10719 PR tree-optimization/81900
10720 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
10721 for blocks with abnormal predecessors.
10722 (compute_antic): Do not set visited flag prematurely.
10724 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
10727 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
10729 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10731 * stor-layout.h (vector_type_mode): Move to...
10732 * tree.h (vector_type_mode): ...here.
10733 * stor-layout.c (vector_type_mode): Move to...
10734 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
10736 2017-08-21 Richard Biener <rguenther@suse.de>
10738 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
10739 register_external_die hooks.
10740 (debug_false_tree_charstarstar_uhwistar): Declare.
10741 (debug_nothing_tree_charstar_uhwi): Likewise.
10742 * debug.c (do_nothing_debug_hooks): Adjust.
10743 (debug_false_tree_charstarstar_uhwistar): New do nothing.
10744 (debug_nothing_tree_charstar_uhwi): Likewise.
10745 * dbxout.c (dbx_debug_hooks): Adjust.
10746 (xcoff_debug_hooks): Likewise.
10747 * sdbout.c (sdb_debug_hooks): Likewise.
10748 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
10749 * dwarf2out.c (macinfo_label_base): New global.
10750 (dwarf2out_register_external_die): New function for the
10751 register_external_die hook.
10752 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
10753 (dwarf2_debug_hooks): Use them.
10754 (dwarf2_lineno_debug_hooks): Adjust.
10755 (struct die_struct): Add with_offset flag.
10756 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
10757 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
10758 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
10759 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
10760 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
10761 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
10762 defining section names for the early LTO debug variants.
10763 (reset_indirect_string): New helper.
10764 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
10765 (print_dw_val): Add support for offsetted symbol references.
10766 (get_ultimate_context): Split out from is_cxx.
10767 (is_cxx): Use get_ultimate_context.
10768 (is_fortran): Add decl overload.
10769 (compute_comp_unit_symbol): Split out worker from
10770 compute_section_prefix.
10771 (compute_section_prefix): Call compute_comp_unit_symbol and
10772 set comdat_type_p here.
10773 (output_die): Skip DIE symbol output for the LTO added one.
10774 Handle DIE symbol references with offset.
10775 (output_comp_unit): Guard section name mangling properly.
10776 For LTO debug sections emit a symbol at the section beginning
10777 which we use to refer to its DIEs.
10778 (add_abstract_origin_attribute): For DIEs registered via
10779 dwarf2out_register_external_die directly refer to the early
10780 DIE rather than indirectly through the shadow one we created.
10781 Remove obsolete call to dwarf2out_abstract_function for
10782 non-function/block origins.
10783 (gen_array_type_die): When generating early LTO debug do
10784 not emit DW_AT_string_length.
10785 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
10786 late when in LTO. As suggested place a gcc_unreachable for
10787 the DECL_ABSTRACT_P case.
10788 (gen_subprogram_die): Avoid another specification DIE
10789 for early built declarations/definitions for the late LTO case.
10790 (gen_variable_die): Add type references for late duplicated VLA dies
10792 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
10793 we have the abstract instance already.
10794 (process_scope_var): Adjust decl DIE contexts in LTO which
10795 first puts them in limbo.
10796 (gen_decl_die): Do not generate type DIEs late apart from
10797 types for VLAs or for decls we do not yet have a DIE. Do not
10798 call dwarf2out_abstract_function late.
10799 (dwarf2out_early_global_decl): Make sure to create DIEs
10800 for abstract instances of a decl first.
10801 (dwarf2out_late_global_decl): Adjust comment.
10802 (output_macinfo_op): With multiple macro sections use
10803 macinfo_label_base to distinguish labels.
10804 (output_macinfo): Likewise. Update macinfo_label_base.
10805 Pass in the line info label.
10806 (note_variable_value_in_expr): When generating LTO resolve
10807 all variable values here by generating DIEs as needed.
10808 (init_sections_and_labels): Add early LTO debug flag parameter
10809 and generate different sections and names if set. Add generation
10810 counter for the labels so we can have multiple of them.
10811 (reset_dies): Helper to allow DIEs to be output multiple times.
10812 (dwarf2out_finish): When outputting DIEs to the fat part of an
10813 LTO object first reset DIEs.
10814 (dwarf2out_early_finish): Output early DIEs when generating LTO.
10815 (modified_type_die): Check for decl_ultimate_origin being self
10817 (gen_type_die_with_usage): Likewise.
10818 (gen_typedef_die): Allow decl_ultimate_origin being self.
10819 (set_decl_abstract_flags): Remove.
10820 (set_block_abstract_flags): Likewise.
10821 (dwarf2out_abstract_function): Treat the early generated DIEs
10822 as the abstract copy and only add DW_AT_inline and
10823 DW_AT_artificial here and call set_decl_origin_self.
10824 If the DIE has an abstract origin don't do anything.
10825 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
10826 if we have none yet (Go fails to build one, PR78628).
10827 (variably_modified_type_p): Prevent endless recursion for Ada
10828 cyclic pointer types.
10829 * lto-streamer-in.c: Include debug.h.
10830 (dref_queue): New global.
10831 (lto_read_tree_1): Stream in DIE references.
10832 (lto_input_tree): Register DIE references.
10833 (input_function): Stream DECL_DEBUG_ARGS.
10834 * lto-streamer-out.c: Include debug.h.
10835 (lto_write_tree_1): Output DIE references.
10836 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
10837 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
10838 (output_function): Stream DECL_DEBUG_ARGS.
10839 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
10840 Stream DECL_ABSTRACT_ORIGIN.
10841 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
10842 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
10843 DECL_CONTEXT for file-scope decls.
10844 * lto-streamer.h (struct dref_entry): Declare.
10845 (dref_queue): Likewise.
10846 * cfgexpand.c (pass_expand::execute): Do not call the
10847 outlining_inline_function hook here.
10848 * lto-wrapper.c (debug_obj): New global.
10849 (tool_cleanup): Unlink it if required.
10850 (debug_objcopy): New function.
10851 (run_gcc): Handle early debug sections in the IL files by
10852 extracting them to separate files, partially linkin them and
10853 feeding the result back as result to the linker.
10854 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
10855 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
10856 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
10857 sections into a separate segment.
10858 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
10860 (darwin_asm_dwarf_section): Likewise.
10861 (darwin_asm_output_dwarf_offset): Likewise.
10862 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
10864 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10865 Alan Hayward <alan.hayward@arm.com>
10866 David Sherwood <david.sherwood@arm.com>
10868 * read-md.h (md_reader::record_potential_iterator_use): Replace
10869 pointer argument with an rtx and an index.
10870 * read-rtl.c (iterator_group::apply_iterator): Likewise.
10871 (apply_mode_iterator): Likewise.
10872 (apply_code_iterator): Likewise.
10873 (apply_int_iterator): Likewise.
10874 (apply_subst_iterator): Likewise.
10875 (record_iterator_use): Likewise.
10876 (record_attribute_use): Likewise.
10877 (md_reader::record_potential_iterator_use): Likewise. Update calls
10878 to record_iterator_use and apply_iterator.
10879 (iterator_use): Replace ptr with x and index.
10880 (attribute_use): Likewise.
10881 (apply_attribute_uses): Update calls to apply_iterator.
10882 (apply_iterators): Likewise. Update initialization of iterator_use.
10883 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
10884 and record_potential_iterator_use.
10885 (rtx_reader::read_rtx_operand): Likewise.
10887 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10888 Alan Hayward <alan.hayward@arm.com>
10889 David Sherwood <david.sherwood@arm.com>
10891 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
10894 2017-08-21 Richard Biener <rguenther@suse.de>
10896 PR middle-end/81884
10897 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
10898 at struct end conservatively when comparing common bases.
10900 2017-08-21 Richard Biener <rguenther@suse.de>
10902 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
10903 (mem_ref_in_stmt): Remove.
10904 (determine_max_movement): Use ref index to get at the reference.
10905 (invariantness_dom_walker::before_dom_children): Deal with
10906 lim data already initialized.
10907 (gather_mem_refs_stmt): Initialize lim data and record ref index.
10909 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
10911 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
10912 (TARGET_ISA_ROUND): Ditto.
10913 (TARGET_ROUND): Ditto.
10914 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
10915 * config/i386/i386.md: Ditto.
10916 * config/i386/sse.md: Ditto.
10917 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
10918 with OPTION_MASK_ISA_SSE4_1.
10920 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
10923 * doc/extend.texi (x86 Built-in Functions): Correct the name of
10924 __builtin_ia32_lzcnt_u16.
10926 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
10929 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
10930 (rs6000_set_current_function): Rewrite function to use it.
10932 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
10935 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
10936 and TYPE_WARN_IF_NOT_ALIGN.
10937 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
10938 (handle_warn_if_not_align): New.
10939 (place_union_field): Call handle_warn_if_not_align.
10940 (place_field): Call handle_warn_if_not_align.
10941 Copy TYPE_WARN_IF_NOT_ALIGN.
10942 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
10943 (layout_type): Likewise.
10944 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
10946 (tree_decl_common): Add warn_if_not_align.
10947 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
10948 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
10949 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
10950 (DECL_WARN_IF_NOT_ALIGN): Likewise.
10951 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
10952 * doc/extend.texi: Document warn_if_not_aligned attribute.
10953 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
10955 2017-08-17 Martin Liska <mliska@suse.cz>
10958 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
10959 (get_data_dependence): Use it as pointer type.
10960 (distribute_loop): Likewise.
10962 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10964 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
10965 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
10966 (p8_vmrgow_<mode>_direct): New define_insn.
10967 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
10968 handle endianness for vmrgew and vmrgow permute patterns.
10970 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
10972 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
10973 * config/rs6000/rs6000-cpus.def: Remove comment.
10974 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
10975 (POWERPC_MASKS): Likewise.
10976 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
10977 use of TARGET_VSX_TIMODE.
10978 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
10979 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
10980 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
10981 (rs6000_option_override_internal): Remove dead code.
10982 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
10983 (rs6000_legitimize_reload_address): Likewise.
10984 (rs6000_legitimate_address_p): Likewise.
10985 (rs6000_opt_masks): Delete "vsx-timode".
10986 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
10987 from function comment.
10988 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
10989 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
10990 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
10992 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
10993 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
10994 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
10995 (VSX_TI): Likewise.
10997 (define_peephole2): Likewise.
10999 2017-08-17 Martin Sebor <msebor@redhat.com>
11002 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
11003 past the end of an array.
11004 (test_pp_format): Add test cases.
11006 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
11008 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
11009 missing ECF_NOTHROW flags.
11011 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
11014 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
11015 operands residing in integer registers.
11016 (*vsx_le_perm_load_<mode>): Likewise.
11017 (*vsx_le_perm_store_<mode>): Likewise.
11018 (define_peephole2): Add peepholes to optimize the above.
11020 2017-08-17 Marek Polacek <polacek@redhat.com>
11022 PR middle-end/81814
11023 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
11024 to mimic what shorten_compare did. Change the return type to bool.
11025 (fold_cond_expr_with_comparison): Update call to
11026 operand_equal_for_comparison_p.
11027 (fold_ternary_loc): Likewise.
11029 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
11031 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
11033 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
11035 2017-08-17 Richard Biener <rguenther@suse.de>
11037 * tree-ssa-structalias.c (solve_graph): When propagating
11038 to successors update the graphs succ edges and avoid duplicate work.
11040 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
11043 * config/i386/i386.c (ix86_option_override_internal): Save target
11044 specific options after ix86_stack_protector_guard_reg was changed.
11046 2017-08-17 Richard Biener <rguenther@suse.de>
11048 PR tree-optimization/81827
11049 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
11050 (new_var_info): Initialize it conservatively.
11051 (get_call_vi): Mark register vars.
11052 (new_scalar_tmp_constraint_exp): Likewise.
11053 (handle_rhs_call): Likewise.
11054 (handle_const_call): Likewise.
11055 (create_function_info_for): Likewise.
11056 (solve_constraints): Sort varinfos to separate register from
11057 non-register vars to pack points-to solution bitmaps during
11060 2017-08-17 Marek Polacek <polacek@redhat.com>
11062 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
11064 2017-08-17 Richard Biener <rguenther@suse.de>
11066 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
11067 to true when overflow is undefined and we saturated the result.
11069 2017-08-17 Alan Modra <amodra@gmail.com>
11072 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
11073 Don't use store multiple if only one reg needs saving.
11074 (interesting_frame_related_regno): New function.
11075 (rs6000_frame_related): Don't emit frame info for regs that
11077 (rs6000_emit_epilogue): Likewise.
11079 2017-08-16 Nathan Sidwell <nathan@acm.org>
11081 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
11082 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
11083 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
11084 * tree.c (free_lang_data_in_type): Use else-if chain. Always
11085 clear TYPE_LANG_1. Remove obsolete member-function stripping.
11086 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
11087 (verify_type): Adjust for TYPE_BINFO move.
11088 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
11089 process TYPE_BINFO directly.
11090 (hash_tree): Likewise.
11091 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
11093 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
11096 2017-08-16 David Malcolm <dmalcolm@redhat.com>
11098 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
11100 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
11103 * config/i386/i386.md (*anddi_1_btr): Change predicates of
11104 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
11105 Add ix86_binary_operator_ok to insn constraint.
11106 (*iordi_1_bts): Ditto.
11107 (*xordi_1_btc): Ditto.
11108 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
11109 Update corresponding peephole2 pattern.
11113 2017-08-16 Bin Cheng <bin.cheng@arm.com>
11115 PR tree-optimization/81832
11116 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
11117 copy loop header which has IFN_LOOP_DIST_ALIAS call.
11119 2017-08-16 Marek Polacek <polacek@redhat.com>
11122 * fold-const.c (fold_indirect_ref_1): Restore original behavior
11123 regarding size_zero_node.
11125 2017-08-16 Martin Liska <mliska@suse.cz>
11128 * config.gcc: Respect previously set extra_objs in case
11131 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
11133 PR tree-optimization/81835
11134 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
11135 the phi SSA_NAME. Check that the condition in a COND_EXPR does
11136 not depend on the phi.
11138 2017-08-16 Alan Modra <amodra@gmail.com>
11140 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
11143 2017-08-16 Alan Modra <amodra@gmail.com>
11145 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
11146 (save_reg_p): ..into this. Update all callers.
11147 (first_reg_to_save): Simplify.
11149 2017-08-16 Alan Modra <amodra@gmail.com>
11151 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
11154 2017-08-15 Joseph Myers <joseph@codesourcery.com>
11158 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
11159 constant count if that count is less than 32.
11161 2017-08-15 Nathan Sidwell <nathan@acm.org>
11163 * gcc.c (execute): Emit friendlier message if inferior is killed
11164 by an external cause.
11166 2017-08-15 Richard Biener <rguenther@suse.de>
11168 PR tree-optimization/81790
11169 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
11170 CONSTRUCTORs from simplifying and VN.
11172 2017-08-14 Martin Sebor <msebor@redhat.com>
11174 * builtin-attrs.def: Add comments.
11176 2017-08-14 Martin Sebor <msebor@redhat.com>
11179 * doc/extend.texi (attribute nonstring): Document new attribute.
11181 2017-08-14 Martin Sebor <msebor@redhat.com>
11184 * tree-diagnostic.c (default_tree_printer): Handle %G.
11185 * gimple-pretty-print.h (percent_G_format): Declare new function.
11186 * gimple-pretty-print.c (percent_G_format): Define.
11187 * tree-pretty-print.c (percent_K_format): Add argument.
11189 2017-08-14 Martin Sebor <msebor@redhat.com>
11191 PR translation/79998
11192 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
11193 Remove a stray space.
11195 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
11198 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
11199 (*iordi_1_bts): Ditto.
11200 (*xordi_1_btc): Ditto.
11202 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11205 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
11206 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11208 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
11209 quoted strings, and make more translator-friendly.
11210 (darwin_rs6000_override_options): Likewise.
11211 (rs6000_option_override_internal): Likewise.
11212 (rs6000_return_in_memory): Fix overlong line.
11213 (init_cmulative_args): Use quoted strings, and make more
11214 translator-friendly.
11215 (rs6000_pass_by_reference): Fix overlong line.
11216 (def_builtin): Use quoted strings.
11217 (altivec_expand_predicate_builtin): Use quoted strings, and make
11218 more translator-friendly.
11219 (htm_expand_builtin): Use quoted strings.
11220 (cpu_expand_builtin): Use quoted strings, and make more
11221 translator-friendly.
11222 (altivec_expand_builtin): Likewise.
11223 (paired_expand_predicate_builtin): Likewise.
11224 (rs6000_invalid_builtin): Likewise.
11225 (builtin_function_type): Use quoted strings.
11226 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
11227 more translator-friendly.
11228 (rs6000_trampoline_init): Likewise.
11229 (rs6000_handle_altivec_attribute): Likewise.
11230 (rs6000_inner_target_options): Use quoted strings.
11231 (rs6000_disable_incompatible_switches): Likewise.
11232 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
11233 strings, and make more translator-friendly.
11234 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
11236 2017-08-14 Bin Cheng <bin.cheng@arm.com>
11238 PR tree-optimization/81799
11239 * tree-loop-distribution.c (version_loop_by_alias_check): Force
11240 cond_expr to simple gimple operand.
11242 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
11244 PR middle-end/46932
11245 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
11247 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
11251 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
11252 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
11253 TARGET_GASISR_PROLOGUES.
11254 * config/avr/avr.c (avr_option_override): Same.
11255 (avr_pass_pre_proep::execute): Same.
11257 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
11260 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
11261 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
11263 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
11265 * config/i386/i386.md (*load_tp_<mode>): Redefine as
11266 define_insn_and_split. Split to a memory load from 0 in
11267 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
11268 using PTR mode iterator.
11269 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
11270 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
11271 (*add_tp_<mode>): Redefine as define_insn_and_split.
11272 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
11273 address space. Merge with *add_tp_x32 using PTR mode iterator.
11274 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
11275 Split to an add with a memory load from 0 in
11276 DEFAULT_TLS_SEG_REG address space.
11278 2017-08-12 Andrew Pinski <apinski@cavium.com>
11280 * config/aarch64/aarch64-option-extensions.def (rdma):
11281 Fix feature string to what Linux prints out in /proc/cpuinfo.
11283 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
11286 * dwarf2out.c (modified_type_die): For C typedef types that have
11287 an ultimate origin, process the ultimate origin instead of the
11289 (gen_typedef_die): Assert that input DECLs have no ultimate
11291 (gen_type_die_with_usage): For typedef variants that have an
11292 ultimate origin, just call gen_decl_die on the original DECL.
11293 (process_scope_var): Avoid creating DIEs for local typedefs and
11294 concrete static variables.
11296 2017-08-12 Alan Modra <amodra@gmail.com>
11300 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
11301 match gnu-user.h startfile.
11302 (ENDFILE_LINUX_SPEC): Similarly.
11304 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
11307 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
11309 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
11311 2017-08-11 Tamar Christina <tamar.christina@arm.com>
11312 * config/aarch64/aarch64.md (mov<mode>): Change.
11313 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
11314 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
11315 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
11317 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
11319 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
11320 for storage order barriers.
11322 2017-08-11 Martin Liska <mliska@suse.cz>
11325 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
11326 variables of void type.
11328 2017-08-11 Martin Liska <mliska@suse.cz>
11330 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
11331 TARGET_SUPPORTS_ALIASES.
11332 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
11333 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
11334 (optimize_weakref): Likewise.
11335 * symtab.c (symtab_node::noninterposable_alias): Likewise.
11336 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
11337 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
11339 2017-08-11 Martin Liska <mliska@suse.cz>
11342 * config/i386/i386.c (make_resolver_func): Do complete
11343 refactoring of the function.
11345 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
11348 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
11349 * config/i386/i386.c (ix86_stack_protect_guard): Use
11350 ix86_stack_protect_guard_symbol_str to generate varible declaration.
11351 * doc/invoke.texi (x86 Options): Document
11352 -mstack-protector-guard-symbol= option.
11354 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
11356 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
11357 * config/i386/i386.c (ix86_split_stack_guard): New function.
11358 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
11359 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
11360 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
11361 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
11362 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
11363 (split_stack_space_check): Call ix86_split_stack_guard.
11365 2017-08-10 Martin Sebor <msebor@redhat.com>
11367 * print-tree.c (print_node): Print location using the established
11369 Replace spaces with colons.
11370 (debug_raw, debug): Ditto.
11372 2017-08-10 Martin Sebor <msebor@redhat.com>
11375 * pretty-print.c (pp_format): Correct the handling of %s precision.
11377 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
11380 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
11382 (ix86_finalize_stack_frame_flags): This. Also clear
11383 frame_pointer_needed if -fno-omit-frame-pointer is used without
11385 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
11386 with ix86_finalize_stack_frame_flags.
11387 (ix86_expand_epilogue): Likewise.
11388 (ix86_expand_split_stack_prologue): Likewise.
11389 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
11391 2017-08-10 Martin Liska <mliska@suse.cz>
11394 * c-attribs.c (handle_target_attribute):
11395 Report warning for an empty string argument of target attribute.
11397 2017-08-09 Jakub Jelinek <jakub@redhat.com>
11400 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
11402 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
11403 or DECL_NONLOCAL labels.
11404 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
11405 or DECL_NONLOCAL labels here.
11407 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
11409 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
11410 to indicate when early gimple folding has been disabled.
11411 (rs6000_gimple_fold_builtin): Add debug content.
11412 (rs6000_invalid_builtin): Fix whitespace.
11413 (rs6000_expand_builtin): Fix whitespace.
11414 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
11416 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
11419 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
11420 SAVE_MULTIPLE if not all the registers that saves, should be saved.
11422 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
11424 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
11425 (qdf24xx): Likewise.
11426 * config/aarch64/aarch64.md: Include falkor.md.
11427 * config/aarch64/falkor.md: New.
11429 2017-08-09 Marek Polacek <polacek@redhat.com>
11432 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
11433 * diagnostic.c (emit_diagnostic): Add a comment.
11434 (emit_diagnostic_valist): New function.
11436 2017-08-09 Marek Polacek <polacek@redhat.com>
11439 * input.c (make_location): New overload.
11440 * input.h (make_location): Declare.
11442 2017-08-08 Alan Modra <amodra@gmail.com>
11443 H.J. Lu <hongjiu.lu@intel.com>
11446 * gcc.c (NO_PIE_SPEC): Delete.
11447 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
11449 (LINK_PIE_SPEC): ..to here.
11450 (LINK_COMMAND_SPEC): Support -no-pie.
11451 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
11452 chain of crtbegin*.o selection, update for PIE_SPEC changes and
11454 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
11455 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
11456 (ENDFILE_CRTEND_SPEC): Similarly.
11458 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
11461 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
11462 (mstack-protector-guard-offset=): Ditto.
11463 * config/i386/i386.c (ix86_option_override): Handle
11464 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
11466 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
11467 ix86_stack_protect_guard_offset variables.
11468 (TARGET_STACK_PROTECT_GUARD): Always define.
11469 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
11470 and -mstack-protector-guard-offset= options.
11472 2017-08-08 Bin Cheng <bin.cheng@arm.com>
11474 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
11475 boundary case for the last candidate.
11477 2017-08-08 Bin Cheng <bin.cheng@arm.com>
11479 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
11480 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
11482 2017-08-08 Tamar Christina <tamar.christina@arm.com>
11484 PR middle-end/19706
11485 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
11486 * config/aarch64/aarch64-builtins.c
11487 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
11488 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
11489 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
11491 2017-08-08 Tamar Christina <tamar.christina@arm.com>
11492 Andrew Pinski <pinskia@gmail.com>
11494 PR middle-end/19706
11495 * internal-fn.def (XORSIGN): New.
11496 * optabs.def (xorsign_optab): New.
11497 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
11498 (convert_expand_mult_copysign): New.
11499 (pass_optimize_widening_mul::execute): Call
11500 convert_expand_mult_copysign.
11502 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11504 PR tree-optimization/81354
11505 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
11506 Insert on edges rather than explicitly creating landing pads.
11507 (analyze_candidates_and_replace): Commit edge inserts.
11509 2017-08-08 Richard Biener <rguenther@suse.de>
11511 PR middle-end/81719
11512 * tree-ssa-loop-niter.c: Include tree-dfa.h.
11513 (expand_simple_operations): Also look through ADDR_EXPRs with
11514 MEM_REF bases treating them as POINTER_PLUS_EXPR.
11516 2017-08-08 Richard Biener <rguenther@suse.de>
11518 PR tree-optimization/81723
11519 * tree-vect-slp.c (struct bst_traits): New hash traits.
11520 (bst_fail): New global.
11521 (vect_build_slp_tree_2): New worker, split out from ...
11522 (vect_build_slp_tree): ... this now wrapping it with using
11523 bst_fail set to cache SLP tree build fails. Properly handle
11525 (vect_analyze_slp_instance): Allocate and free bst_fail.
11527 2017-08-08 Martin Liska <mliska@suse.cz>
11530 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
11531 LABEL_DECLs that can be from a different function.
11533 2017-08-08 Bin Cheng <bin.cheng@arm.com>
11535 PR tree-optimization/81744
11536 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
11537 loop's number of iterations.
11539 2017-08-08 Martin Liska <mliska@suse.cz>
11541 * asan.c: Include header files.
11542 * attribs.c (build_decl_attribute_variant): New function moved
11544 (build_type_attribute_qual_variant): Likewise.
11545 (cmp_attrib_identifiers): Likewise.
11546 (simple_cst_list_equal): Likewise.
11547 (omp_declare_simd_clauses_equal): Likewise.
11548 (attribute_value_equal): Likewise.
11549 (comp_type_attributes): Likewise.
11550 (build_type_attribute_variant): Likewise.
11551 (lookup_ident_attribute): Likewise.
11552 (remove_attribute): Likewise.
11553 (merge_attributes): Likewise.
11554 (merge_type_attributes): Likewise.
11555 (merge_decl_attributes): Likewise.
11556 (merge_dllimport_decl_attributes): Likewise.
11557 (handle_dll_attribute): Likewise.
11558 (attribute_list_equal): Likewise.
11559 (attribute_list_contained): Likewise.
11560 * attribs.h (lookup_attribute): New function moved from tree.[ch].
11561 (lookup_attribute_by_prefix): Likewise.
11562 * bb-reorder.c: Include header files.
11563 * builtins.c: Likewise.
11564 * calls.c: Likewise.
11565 * cfgexpand.c: Likewise.
11566 * cgraph.c: Likewise.
11567 * cgraphunit.c: Likewise.
11568 * convert.c: Likewise.
11569 * dwarf2out.c: Likewise.
11570 * final.c: Likewise.
11571 * fold-const.c: Likewise.
11572 * function.c: Likewise.
11573 * gimple-expr.c: Likewise.
11574 * gimple-fold.c: Likewise.
11575 * gimple-pretty-print.c: Likewise.
11576 * gimple.c: Likewise.
11577 * gimplify.c: Likewise.
11578 * hsa-common.c: Likewise.
11579 * hsa-gen.c: Likewise.
11580 * internal-fn.c: Likewise.
11581 * ipa-chkp.c: Likewise.
11582 * ipa-cp.c: Likewise.
11583 * ipa-devirt.c: Likewise.
11584 * ipa-fnsummary.c: Likewise.
11585 * ipa-inline.c: Likewise.
11586 * ipa-visibility.c: Likewise.
11588 * lto-cgraph.c: Likewise.
11589 * omp-expand.c: Likewise.
11590 * omp-general.c: Likewise.
11591 * omp-low.c: Likewise.
11592 * omp-offload.c: Likewise.
11593 * omp-simd-clone.c: Likewise.
11594 * opts-global.c: Likewise.
11595 * passes.c: Likewise.
11596 * predict.c: Likewise.
11597 * sancov.c: Likewise.
11598 * sanopt.c: Likewise.
11599 * symtab.c: Likewise.
11600 * toplev.c: Likewise.
11601 * trans-mem.c: Likewise.
11602 * tree-chkp.c: Likewise.
11603 * tree-eh.c: Likewise.
11604 * tree-into-ssa.c: Likewise.
11605 * tree-object-size.c: Likewise.
11606 * tree-parloops.c: Likewise.
11607 * tree-profile.c: Likewise.
11608 * tree-ssa-ccp.c: Likewise.
11609 * tree-ssa-live.c: Likewise.
11610 * tree-ssa-loop.c: Likewise.
11611 * tree-ssa-sccvn.c: Likewise.
11612 * tree-ssa-structalias.c: Likewise.
11613 * tree-ssa.c: Likewise.
11614 * tree-streamer-in.c: Likewise.
11615 * tree-vectorizer.c: Likewise.
11616 * tree-vrp.c: Likewise.
11617 * tsan.c: Likewise.
11618 * ubsan.c: Likewise.
11619 * varasm.c: Likewise.
11620 * varpool.c: Likewise.
11621 * tree.c: Remove functions moved to attribs.[ch].
11622 * tree.h: Likewise.
11623 * config/aarch64/aarch64.c: Add attrs.h header file.
11624 * config/alpha/alpha.c: Likewise.
11625 * config/arc/arc.c: Likewise.
11626 * config/arm/arm.c: Likewise.
11627 * config/avr/avr.c: Likewise.
11628 * config/bfin/bfin.c: Likewise.
11629 * config/c6x/c6x.c: Likewise.
11630 * config/cr16/cr16.c: Likewise.
11631 * config/cris/cris.c: Likewise.
11632 * config/darwin.c: Likewise.
11633 * config/epiphany/epiphany.c: Likewise.
11634 * config/fr30/fr30.c: Likewise.
11635 * config/frv/frv.c: Likewise.
11636 * config/ft32/ft32.c: Likewise.
11637 * config/h8300/h8300.c: Likewise.
11638 * config/i386/winnt.c: Likewise.
11639 * config/ia64/ia64.c: Likewise.
11640 * config/iq2000/iq2000.c: Likewise.
11641 * config/lm32/lm32.c: Likewise.
11642 * config/m32c/m32c.c: Likewise.
11643 * config/m32r/m32r.c: Likewise.
11644 * config/m68k/m68k.c: Likewise.
11645 * config/mcore/mcore.c: Likewise.
11646 * config/microblaze/microblaze.c: Likewise.
11647 * config/mips/mips.c: Likewise.
11648 * config/mmix/mmix.c: Likewise.
11649 * config/mn10300/mn10300.c: Likewise.
11650 * config/moxie/moxie.c: Likewise.
11651 * config/msp430/msp430.c: Likewise.
11652 * config/nds32/nds32-isr.c: Likewise.
11653 * config/nds32/nds32.c: Likewise.
11654 * config/nios2/nios2.c: Likewise.
11655 * config/nvptx/nvptx.c: Likewise.
11656 * config/pa/pa.c: Likewise.
11657 * config/pdp11/pdp11.c: Likewise.
11658 * config/powerpcspe/powerpcspe.c: Likewise.
11659 * config/riscv/riscv.c: Likewise.
11660 * config/rl78/rl78.c: Likewise.
11661 * config/rx/rx.c: Likewise.
11662 * config/s390/s390.c: Likewise.
11663 * config/sh/sh.c: Likewise.
11664 * config/sol2.c: Likewise.
11665 * config/sparc/sparc.c: Likewise.
11666 * config/spu/spu.c: Likewise.
11667 * config/stormy16/stormy16.c: Likewise.
11668 * config/tilegx/tilegx.c: Likewise.
11669 * config/tilepro/tilepro.c: Likewise.
11670 * config/v850/v850.c: Likewise.
11671 * config/vax/vax.c: Likewise.
11672 * config/visium/visium.c: Likewise.
11673 * config/xtensa/xtensa.c: Likewise.
11675 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
11678 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
11679 constraints since the -mupper-regs-* switches have been
11681 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
11682 into a vector from a double word element that was extracted from
11683 another vector, and eliminate extra XXPERMDI instructions.
11684 (vsx_concat_<mode>_2): Likewise.
11685 (vsx_concat_<mode>_3): Likewise.
11686 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
11687 concat to allow optimizing inserts from previous extracts.
11689 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
11691 * config/i386/i386.c (ix86_stack_protect_guard): Generate
11692 memory reference to a SSP offset in TLS address space.
11693 (ix86_print_operand) <case '@'>: Remove.
11694 (ix86_print_operand_punct_valid_p): Remove '@' code.
11695 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
11696 UNSPEC_SP_TLS_TEST.
11697 (stack_tls_protect_set_<mode>): Remove.
11698 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
11699 (stack_tls_protect_test_<mode>): Remove.
11700 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
11702 2017-08-07 Olivier Hainque <hainque@adacore.com>
11705 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
11707 2017-08-07 Douglas Rupp <rupp@adacore.com>
11709 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
11710 variable was referenced as multidir in command.
11712 2017-08-07 Jakub Jelinek <jakub@redhat.com>
11715 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
11718 2017-08-07 Martin Liska <mliska@suse.cz>
11720 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
11721 * config/rl78/rl78.c: Add include of attribs.h.
11722 * config/sh/sh.c: Likewise.
11723 * config/v850/v850.c: Likewise.
11725 2017-08-07 Tom de Vries <tom@codesourcery.com>
11727 PR middle-end/78266
11728 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
11730 2017-08-07 Martin Liska <mliska@suse.cz>
11732 * config/mips/mips.c: Include attribs.h.
11734 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
11737 * doc/invoke.texi: Document change in behvaior for -Ofast for
11740 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
11742 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
11744 (aarch64_pop_regs): Likewise.
11745 (aarch64_gen_load_pair): Likewise.
11746 (aarch64_save_callee_saves): Likewise.
11747 (aarch64_restore_callee_saves): Likewise.
11749 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
11751 * config/i386/i386.c: Revert the last change.
11753 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
11756 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
11758 (ix86_finalize_stack_frame_flags): This. Also clear
11759 frame_pointer_needed if -fno-omit-frame-pointer is used without
11761 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
11762 with ix86_finalize_stack_frame_flags.
11763 (ix86_expand_epilogue): Likewise.
11764 (ix86_expand_split_stack_prologue): Likewise.
11766 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
11769 * config/i386/i386.c (get_builtin_code_for_version): Set priority
11770 to P_AES for Westmere.
11772 2017-08-07 Jonathan Yong <10walls@gmail.com>
11774 * config/i386/mingw.opt (fset-stack-executable): Removed.
11775 * config/i386/cygming.opt (fset-stack-executable): Moved
11777 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
11779 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
11781 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
11783 2017-08-07 Marek Polacek <polacek@redhat.com>
11785 PR middle-end/81737
11786 * fold-const.c (fold_indirect_ref_1): Check type_domain.
11788 2017-08-07 Martin Liska <mliska@suse.cz>
11790 * attribs.h (canonicalize_attr_name): New function.
11791 (cmp_attribs): Move from c-format.c and adjusted.
11792 (is_attribute_p): Moved from tree.h.
11793 * tree-inline.c: Add new includes.
11794 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
11795 (private_is_attribute_p): Remove.
11796 (private_lookup_attribute): Likewise.
11797 (private_lookup_attribute_by_prefix): Simplify.
11798 (remove_attribute): Use is_attribute_p.
11799 * tree.h: Remove removed declarations.
11801 2017-08-07 Jakub Jelinek <jakub@redhat.com>
11803 PR middle-end/81698
11804 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
11805 instead of computing it in the function. Formatting fix.
11806 (expand_case): Don't rely on default_edge being the first edge,
11807 clear it if removing it, pass default_edge to
11808 emit_case_dispatch_table.
11809 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
11812 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
11814 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
11815 insn in the function, emit NOP after the insn.
11817 2017-08-06 Tom de Vries <tom@codesourcery.com>
11819 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
11822 2017-08-06 Tom de Vries <tom@codesourcery.com>
11824 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
11827 2017-08-04 Yury Gribov <tetra2005@gmail.com>
11829 PR tree-optimization/57371
11830 * match.pd: New pattern.
11832 2017-08-04 Marek Polacek <polacek@redhat.com>
11834 PR middle-end/81695
11835 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
11836 perform the computation in offset_int.
11838 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
11840 PR tree-optimization/81136
11841 * tree-vectorizer.h: Include tree-hash-traits.h.
11842 (vec_base_alignments): New typedef.
11843 (vec_info): Add a base_alignments field.
11844 (vect_record_base_alignments): Declare.
11845 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
11847 (DR_IS_CONDITIONAL_IN_STMT): New macro.
11848 (create_data_ref): Add an is_conditional_in_stmt argument.
11849 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
11850 the is_conditional_in_stmt field.
11851 (data_ref_loc): Add an is_conditional_in_stmt field.
11852 (get_references_in_stmt): Set the is_conditional_in_stmt field.
11853 (find_data_references_in_stmt): Update call to create_data_ref.
11854 (graphite_find_data_references_in_stmt): Likewise.
11855 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
11856 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
11857 (vect_record_base_alignment): New function.
11858 (vect_record_base_alignments): Likewise.
11859 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
11860 for nested statements even if we fail to compute a misalignment.
11861 Use pooled base alignments for unconditional references.
11862 (vect_find_same_alignment_drs): Compare base addresses instead
11864 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
11865 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
11867 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
11869 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
11870 Add an explicit name for the enum. Use auto_vec for slp_instances
11871 and grouped_stores.
11872 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
11874 (_bb_vec_info): Add a constructor and destructor.
11875 (vinfo_for_stmt): Return NULL for uids of -1 as well.
11876 (destroy_loop_vec_info): Delete.
11877 (vect_destroy_datarefs): Likewise.
11878 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
11879 (vec_info::vec_info): New function.
11880 (vec_info::~vec_info): Likewise.
11881 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
11882 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
11883 destroy_loop_vec_info.
11884 * tree-vect-loop.c (new_loop_vec_info): Replace with...
11885 (_loop_vec_info::_loop_vec_info): ...this.
11886 (destroy_loop_vec_info): Replace with...
11887 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
11888 the stmt_vec_infos. Leave handling of vec_info information to its
11889 destructor. Remove explicit vector releases.
11890 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
11891 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
11892 * tree-vect-slp.c (new_bb_vec_info): Replace with...
11893 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
11894 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
11895 (destroy_bb_vec_info): Replace with...
11896 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
11897 information to its destructor.
11898 (vect_slp_analyze_bb_1): Use new and delete instead of
11899 new_bb_vec_info and destroy_bb_vec_info.
11900 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
11903 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
11905 * tree-data-ref.h (subscript): Add access_fn field.
11906 (data_dependence_relation): Add could_be_independent_p.
11907 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
11908 (same_access_functions): Move to tree-data-ref.c.
11909 * tree-data-ref.c (ref_contains_union_access_p): New function.
11910 (access_fn_component_p): Likewise.
11911 (access_fn_components_comparable_p): Likewise.
11912 (dr_analyze_indices): Add a reference to access_fn_component_p.
11913 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
11915 (constant_access_functions): Likewise.
11916 (add_other_self_distances): Likewise.
11917 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
11918 (initialize_data_dependence_relation): Use XCNEW and remove
11919 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
11920 of access functions that have the same type. Allow the
11921 subsequence to end with different bases in some circumstances.
11922 Record the chosen access functions in SUB_ACCESS_FN.
11923 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
11924 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
11925 (subscript_dependence_tester_1): Likewise dra and drb.
11926 (build_classic_dist_vector): Update calls accordingly.
11927 (subscript_dependence_tester): Likewise.
11928 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
11929 DDR_COULD_BE_INDEPENDENT_P.
11930 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
11931 comp_alias_ddrs instead of may_alias_ddrs.
11932 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
11934 (vect_analyze_data_ref_dependence): Use it if
11935 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
11936 distance vectors if that fails.
11937 (dependence_distance_ge_vf): New function.
11938 (vect_prune_runtime_alias_test_list): Use it. Don't clear
11939 LOOP_VINFO_MAY_ALIAS_DDRS.
11941 2017-08-04 Richard Biener <rguenther@suse.de>
11943 PR middle-end/81705
11944 * fold-const.c (fold_binary_loc): Properly restrict
11945 minus_var0 && minus_var1 case when associating undefined overflow
11948 2017-08-04 Tom de Vries <tom@codesourcery.com>
11950 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
11952 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11954 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11955 Don't start diagnostic messages with a capital letter.
11956 * config/rs6000/rs6000.c (rs6000_option_override_internal):
11958 (rs6000_invalid_builtin): Likewise.
11959 (rs6000_trampoline_init): Likewise.
11961 2017-08-03 Jakub Jelinek <jakub@redhat.com>
11964 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
11965 after setting changeable df flags.
11967 2017-08-03 Richard Biener <rguenther@suse.de>
11969 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
11970 up if the use is in USE - X.
11972 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
11974 * toplev.c (dumpfile.h): New include.
11975 (internal_error_reentered): New static function. Use it...
11976 (internal_error_function): ...here to handle reentered internal_error.
11978 2017-08-03 Richard Biener <rguenther@suse.de>
11980 PR middle-end/81148
11981 * fold-const.c (split_tree): Add minus_var and minus_con
11982 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
11983 here but always use minus_*.
11984 (associate_trees): Assert we never associate with MINUS_EXPR
11985 and NULL first operand. Do not recurse for PLUS_EXPR operands
11986 when associating as MINUS_EXPR either.
11987 (fold_binary_loc): Track minus_var and minus_con.
11989 2017-08-03 Tom de Vries <tom@codesourcery.com>
11992 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
11993 ACCEL_COMPILER, apply finish_options on
11994 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
11996 2017-08-03 Tom de Vries <tom@codesourcery.com>
11999 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
12000 function_entry_patch_area_size > 0.
12002 2017-08-03 Jakub Jelinek <jakub@redhat.com>
12005 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
12006 instead of 10??LU, perform unit multiplication in wide_int,
12007 don't change alloc_object_size_limit if the limit is larger
12010 PR tree-optimization/81655
12011 PR tree-optimization/81588
12012 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
12013 the case when ranges[i].low and high are 1 for unsigned type with
12016 PR middle-end/81052
12017 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
12018 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
12020 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12022 * tree-vrp.h: Add include guard.
12024 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
12027 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
12028 (ud2): New insn pattern.
12029 * config/i386/i386.c (ix86_expand_epilogue):
12030 For naked functions, generate ud2 instead of trap insn.
12032 2017-08-02 Marek Polacek <polacek@redhat.com>
12035 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
12037 2017-08-02 Tom de Vries <tom@codesourcery.com>
12038 Cesar Philippidis <cesar@codesourcery.com>
12040 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
12041 Add missing edge probabilities.
12043 2017-08-02 Tamar Christina <tamar.christina@arm.com>
12045 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
12046 Correct endianness.
12048 2017-08-02 Jakub Jelinek <jakub@redhat.com>
12050 PR middle-end/79499
12051 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
12052 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
12053 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
12054 prologue_seq sequences - if any.
12056 2017-08-02 Richard Biener <rguenther@suse.de>
12058 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
12059 via vectors if supported, integer extracts via punning if supported
12060 or otherwise vector extracts.
12062 2017-08-02 Richard Biener <rguenther@suse.de>
12064 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
12066 (bitmap_insert_into_set): ... this.
12068 2017-08-02 Richard Biener <rguenther@suse.de>
12070 PR tree-optimization/81633
12072 2015-08-17 Alan Hayward <alan.hayward@arm.com>
12074 PR tree-optimization/71752
12075 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
12077 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
12079 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
12080 (machine_function::call_ms2sysv_pad_out): Remove field.
12081 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
12082 (ix86_compute_frame_layout): Likewise.
12084 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
12087 * config/i386/i386.c (ix86_set_func_type): Disallow naked
12088 attribute with interrupt attribute.
12090 2017-08-01 Andrew Pinski <apinski@cavium.com>
12092 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
12093 BIT_INSERT_EXPR's operand 1
12094 to see if the types precision matches.
12096 2017-08-01 Martin Liska <mliska@suse.cz>
12098 PR middle-end/70140
12099 * builtins.c (expand_builtin_memcpy_args): Remove.
12100 (expand_builtin_memcpy): Call newly added function
12101 expand_builtin_memory_copy_args.
12102 (expand_builtin_memcpy_with_bounds): Likewise.
12103 (expand_builtin_mempcpy): Remove last argument.
12104 (expand_builtin_mempcpy_with_bounds): Likewise.
12105 (expand_builtin_memory_copy_args): New function created from
12106 expand_builtin_mempcpy_args with small modifications.
12107 (expand_builtin_mempcpy_args): Remove.
12108 (expand_builtin_stpcpy): Remove unused argument.
12109 (expand_builtin): Likewise.
12110 (expand_builtin_with_bounds): Likewise.
12112 2017-08-01 Martin Liska <mliska@suse.cz>
12115 Make mempcpy more optimal (PR middle-end/70140).
12117 2017-08-01 Jakub Jelinek <jakub@redhat.com>
12120 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
12121 __builtin_vec_cmpne verify both arguments are compatible vectors
12122 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
12123 verify arg1_type is a pointer or array type. For __builtin_vec_st,
12124 move computation of aligned to after checking the argument types.
12128 * config/rs6000/vsx.md (vextract_fp_from_shorth,
12129 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
12132 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
12133 Doug Rupp <rupp@adacore.com>
12134 Olivier Hainque <hainque@adacore.com>
12136 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
12137 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
12139 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
12140 for TARGET_OS_CPP_BUILTIN.
12141 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
12142 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
12144 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
12145 passing required abi options to the assembler for EABI configurations.
12146 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
12147 of .text.hot and .text.unlikely sections for kernel modules when
12148 using ARM style exceptions.
12149 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
12150 options. Add EXTRA_CC1_SPEC.
12151 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
12153 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
12155 (ARM_TARGET2_DWARF_FORMAT): Define.
12156 * config/arm/t-vxworks: Adjust multilib control to removal of the
12157 Diab command line options.
12159 2017-08-01 Martin Liska <mliska@suse.cz>
12161 PR gcov-profile/81561
12162 * gcov.c (unblock): Make unblocking safe as we need to preserve
12163 index correspondence of blocks and block_lists.
12165 2017-08-01 Richard Biener <rguenther@suse.de>
12167 PR tree-optimization/81181
12168 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
12169 (compute_antic): ... end of iteration here.
12171 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
12173 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
12174 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
12175 (ftree-slp-vectorize): Likewise.
12176 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
12177 can no longer be set independent of flag_tree_loop_vectorize.
12178 * omp-general.c (emp_max_vf): Likewise.
12179 * opts.c (enable_fdo_optimizations): Remove references to
12180 flag_tree_vectorize, these are now implicit.
12181 (common_handle_option): Remove handling for OPT_ftree_vectorize,
12182 and leave it for the options machinery.
12184 2017-08-01 Martin Liska <mliska@suse.cz>
12186 PR middle-end/70140
12187 * builtins.c (expand_builtin_memcpy_args): Remove.
12188 (expand_builtin_memcpy): Call newly added function
12189 expand_builtin_memory_copy_args.
12190 (expand_builtin_memcpy_with_bounds): Likewise.
12191 (expand_builtin_mempcpy): Remove last argument.
12192 (expand_builtin_mempcpy_with_bounds): Likewise.
12193 (expand_builtin_memory_copy_args): New function created from
12194 expand_builtin_mempcpy_args with small modifications.
12195 (expand_builtin_mempcpy_args): Remove.
12196 (expand_builtin_stpcpy): Remove unused argument.
12197 (expand_builtin): Likewise.
12198 (expand_builtin_with_bounds): Likewise.
12200 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
12203 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
12204 print "ds:" only for immediates in generic address space.
12206 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
12209 * config/i386/i386.c (ix86_funciton_naked): New prototype.
12210 (ix86_function_ok_for_sibcall): Return false for naked functions.
12212 2017-08-01 Richard Biener <rguenther@suse.de>
12214 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
12215 (compute_antic): Seed worklist with exit block predecessors.
12216 * cfganal.c (dfs_find_deadend): For a cycle return the source
12217 of the edge closing it.
12219 2017-08-01 Tamar Christina <tamar.christina@arm.com>
12221 * config/aarch64/aarch64.c
12222 (aarch64_can_const_movi_rtx_p): Move 0 check.
12224 2017-08-01 Bin Cheng <bin.cheng@arm.com>
12226 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
12227 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
12229 * match.pd: Ditto in address comparison pattern.
12231 2017-08-01 Bin Cheng <bin.cheng@arm.com>
12233 PR tree-optimization/81627
12234 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
12235 closed ssa form for store-store chain.
12237 2017-08-01 Bin Cheng <bin.cheng@arm.com>
12239 PR tree-optimization/81620
12240 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
12241 for store-store chain.
12243 2017-08-01 Jakub Jelinek <jakub@redhat.com>
12245 PR tree-optimization/81588
12246 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
12247 ranges[i].in_p, invert comparison code ccode. For >/>=,
12248 swap rhs1 and rhs2 and comparison code unconditionally,
12249 for </<= don't do that. Don't swap rhs1/rhs2 again if
12250 ranges[i].in_p, instead invert comparison code ccode if
12251 opcode or oe->rank is BIT_IOR_EXPR.
12254 * optabs.def (vec_extract_optab, vec_init_optab): Change from
12255 a direct optab to conversion optab.
12256 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
12257 with GET_MODE_INNER as last argument instead of optab_handler.
12258 * expmed.c (extract_bit_field_1): Likewise. Use vector from
12259 vector extraction if possible and optab is available.
12260 * expr.c (store_constructor): Use convert_optab_handler instead
12261 of optab_handler. Use vector initialization from smaller
12262 vectors if possible and optab is available.
12263 * tree-vect-stmts.c (vectorizable_load): Likewise.
12264 * doc/md.texi (vec_extract, vec_init): Document that the optabs
12265 now have two modes.
12266 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
12267 of vec_init from half-sized vectors with the same element mode.
12268 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
12269 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
12270 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
12271 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
12272 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
12273 after mode in gen_vec_extract* calls.
12274 (vec_extract<mode>): Renamed to ...
12275 (vec_extract<mode><ssescalarmodelower>): ... this.
12276 (vec_extract<mode><ssehalfvecmodelower>): New expander.
12277 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
12278 element mode after mode in gen_vec_init* calls.
12279 (VEC_INIT_HALF_MODE): New mode iterator.
12280 (vec_init<mode>): Renamed to ...
12281 (vec_init<mode><ssescalarmodelower>): ... this.
12282 (vec_init<mode><ssehalfvecmodelower>): New expander.
12283 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
12284 (vec_extractv2sfsf): ... this.
12285 (vec_initv2sf): Renamed to ...
12286 (vec_initv2sfsf): ... this.
12287 (vec_extractv2si): Renamed to ...
12288 (vec_extractv2sisi): ... this.
12289 (vec_initv2si): Renamed to ...
12290 (vec_initv2sisi): ... this.
12291 (vec_extractv4hi): Renamed to ...
12292 (vec_extractv4hihi): ... this.
12293 (vec_initv4hi): Renamed to ...
12294 (vec_initv4hihi): ... this.
12295 (vec_extractv8qi): Renamed to ...
12296 (vec_extractv8qiqi): ... this.
12297 (vec_initv8qi): Renamed to ...
12298 (vec_initv8qiqi): ... this.
12299 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
12300 (vec_init<mode>): Renamed to ...
12301 (vec_init<mode><VEC_base_l>): ... this.
12302 (vec_extract<mode>): Renamed to ...
12303 (vec_extract<mode><VEC_base_l>): ... this.
12304 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
12305 (vec_initv2sfsf): ... this.
12306 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
12307 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
12308 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
12309 element mode after mode in gen_vec_init* calls.
12310 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
12311 (vec_init<mode><Vel>): ... this.
12312 (vec_extract<mode>): Renamed to ...
12313 (vec_extract<mode><Vel>): ... this.
12314 * config/aarch64/iterators.md (Vel): New mode attribute.
12315 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
12316 Add element mode after mode in gen_vec_extract* calls.
12317 * config/s390/vector.md (non_vec_l): New mode attribute.
12318 (vec_extract<mode>): Renamed to ...
12319 (vec_extract<mode><non_vec_l>): ... this.
12320 (vec_init<mode>): Renamed to ...
12321 (vec_init<mode><non_vec_l>): ... this.
12322 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
12323 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
12325 * config/arm/iterators.md (V_elem_l): New mode attribute.
12326 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
12327 (vec_extract<mode><V_elem_l>): ... this.
12328 (vec_extractv2di): Renamed to ...
12329 (vec_extractv2didi): ... this.
12330 (vec_init<mode>): Renamed to ...
12331 (vec_init<mode><V_elem_l>): ... this.
12332 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
12333 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
12334 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
12335 Add element mode after gen_vec_extract* calls.
12336 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
12337 (vec_init<mode><unitmode>): ... this.
12338 (vec_extract<mode>): Renamed to ...
12339 (vec_extract<mode><unitmode>): ... this.
12340 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
12341 (vec_init<mode><unitmode>): ... this.
12342 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
12343 (vec_initv2sfsf): ... this.
12344 (vec_extractv2sf): Renamed to ...
12345 (vec_extractv2sfsf): ... this.
12346 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
12347 Add element mode after gen_vec_extract* calls.
12348 * config/mips/mips.md (unitmode): New mode iterator.
12349 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
12350 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
12351 * config/spu/spu.md (inner_l): New mode attribute.
12352 (vec_init<mode>): Renamed to ...
12353 (vec_init<mode><inner_l>): ... this.
12354 (vec_extract<mode>): Renamed to ...
12355 (vec_extract<mode><inner_l>): ... this.
12356 * config/sparc/sparc.md (veltmode): New mode iterator.
12357 (vec_init<VMALL:mode>): Renamed to ...
12358 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
12359 * config/ia64/vect.md (vec_initv2si): Renamed to ...
12360 (vec_initv2sisi): ... this.
12361 (vec_initv2sf): Renamed to ...
12362 (vec_initv2sfsf): ... this.
12363 (vec_extractv2sf): Renamed to ...
12364 (vec_extractv2sfsf): ... this.
12365 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
12366 (vec_init<mode>): Renamed to ...
12367 (vec_init<mode><VEC_base_l>): ... this.
12368 (vec_extract<mode>): Renamed to ...
12369 (vec_extract<mode><VEC_base_l>): ... this.
12370 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
12371 (vec_initv2sfsf): ... this.
12372 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
12373 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
12374 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
12375 gen_vec_init* calls.
12377 2017-08-01 Richard Biener <rguenther@suse.de>
12379 PR tree-optimization/81297
12380 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
12381 TREE_OVERFLOW from INTEGER_CSTs.
12383 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
12385 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
12387 2017-07-31 Carl Love <cel@us.ibm.com>
12389 * config/rs6000/rs6000-c: Add support for built-in functions
12390 vector signed char vec_xl_be (signed long long, signed char *);
12391 vector unsigned char vec_xl_be (signed long long, unsigned char *);
12392 vector signed int vec_xl_be (signed long long, signed int *);
12393 vector unsigned int vec_xl_be (signed long long, unsigned int *);
12394 vector signed long long vec_xl_be (signed long long, signed long long *);
12395 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
12396 vector signed short vec_xl_be (signed long long, signed short *);
12397 vector unsigned short vec_xl_be (signed long long, unsigned short *);
12398 vector double vec_xl_be (signed long long, double *);
12399 vector float vec_xl_be (signed long long, float *);
12400 * config/rs6000/altivec.h (vec_xl_be): Add #define.
12401 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
12402 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
12404 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
12405 (altivec_expand_builtin): Add switch statement to call
12406 altivec_expand_xl_be for each builtin.
12407 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
12408 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
12409 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
12410 __builtin_vsx_le_be_v16qi.
12411 * doc/extend.texi: Update the built-in documentation file for the
12412 new built-in functions.
12414 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
12417 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
12419 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
12421 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12423 * config.gcc: Add z14.
12424 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
12425 CPU model numbers for z13s and z14.
12426 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
12428 * config/s390/s390-opts.h (enum processor_type): Rename
12429 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
12430 * config/s390/s390.c (processor_table): Add field for CPU name to
12431 be passed to Binutils.
12432 (s390_asm_output_machine_for_arch): Use the new field in
12433 processor_table for Binutils.
12434 (s390_expand_builtin): Replace arch12 with z14.
12435 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
12436 (s390_get_sched_attrmask): Likewise.
12437 (s390_get_unit_mask): Likewise.
12438 * config/s390/s390.opt: Add z14 to processor_type enum.
12440 2017-07-31 Martin Jambor <mjambor@suse.cz>
12443 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
12444 regardless of optimization level.
12446 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
12447 Martin Liska <mliska@suse.cz>
12449 * predict.def: Remove old comment and adjust probability.
12450 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
12451 PREDICT statements.
12453 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
12456 * config/i386/i386.c (ix86_function_naked): New function.
12457 (ix86_can_use_return_insn_p): Return false for naked functions.
12458 (ix86_expand_prologue): Skip prologue for naked functions.
12459 (ix86_expand_epilogue): Skip epilogue for naked functions
12460 and emit trap instruction.
12461 (ix86_warn_func_return): New function.
12462 (ix86_attribute_table): Add "naked" attribute specification.
12463 (TARGET_WARN_FUNC_RETURN): Define.
12464 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
12466 2017-07-31 Martin Liska <mliska@suse.cz>
12468 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
12469 (dump_gimple_bb_header): Always dump BB info.
12470 (pp_cfg_jump): Do not append info about BB when dumping a jump.
12472 2017-07-31 Martin Liska <mliska@suse.cz>
12475 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
12476 also with current_function_decl non-null equality.
12478 2017-07-31 Jakub Jelinek <jakub@redhat.com>
12481 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
12482 change type to the element type, instead add eltype variable and
12483 use it where we are interested in the element type.
12485 PR tree-optimization/81603
12486 * ipa-polymorphic-call.c
12487 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
12488 offset arithmetic in offset_int, bail out if the resulting bit offset
12489 doesn't fit into shwi.
12491 2017-07-31 Martin Liska <mliska@suse.cz>
12493 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
12494 (gimplify_save_expr): Fix comment.
12496 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
12499 * config/i386/i386.c (ix86_function_arg): Update arguments for
12501 (ix86_compute_frame_layout): Set the initial stack offset to
12502 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
12503 INCOMING_FRAME_SP_OFFSET.
12504 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
12505 stack before exception handler returns.
12506 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
12507 the 'ERROR_CODE' for exception handler.
12509 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
12511 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
12512 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
12513 (ASM_OUTPUT_REG_POP): Ditto.
12514 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
12515 instead of asm_fprintf to output pure string.
12517 2017-07-29 Jakub Jelinek <jakub@redhat.com>
12519 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
12520 to imported_module_or_decl hook.
12521 (debug_nothing_tree_tree_tree_bool): Remove.
12522 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
12523 * debug.c (do_nothing_debug_hooks): Use
12524 debug_nothing_tree_tree_tree_bool_bool instead of
12525 debug_nothing_tree_tree_tree_bool.
12526 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
12527 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
12528 * sdbout.c (sdb_debug_hooks): Likewise.
12529 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
12530 (gen_namespace_die): Add DW_AT_export_symbols attribute if
12532 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
12533 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
12534 attribute, don't add anything.
12536 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12538 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
12539 (fold_build2_stat_loc): Likewise.
12540 (fold_build3_stat_loc): Likewise.
12541 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
12542 (fold_build1_loc): Remove macro.
12543 (fold_build2_loc): Likewise.
12544 (fold_build3_loc): Likewise.
12546 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12548 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
12549 (gimple_build_debug_bind_source_stat): Likewise.
12550 * gimple.h (gimple_build_debug_bind): Remove macro.
12551 (gimple_build_debug_bind_source): Likewise.
12553 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12555 * bitmap.c (bitmap_alloc): Adjust.
12556 (bitmap_gc_alloc): Likewise.
12557 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
12559 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12561 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
12562 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
12563 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
12564 (bitmap_gc_alloc_stat): Likewise.
12565 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
12567 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12569 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
12570 * rtl.h (shallow_copy_rtx): Remove macro.
12572 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12574 * emit-rtl.c (gen_raw_REG): Adjust.
12575 * gengenrtl.c (gendef): Likewise.
12576 * rtl.c (rtx_alloc_stat): Remove _stat from name.
12577 * rtl.h (rtx_alloc): Remove macro.
12579 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12581 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
12582 (build_tree_list_stat): Likewise.
12583 * tree.h (build_tree_list): Remove macro.
12584 (build_tree_list_vec): Likewise.
12586 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12588 * tree.c (make_vector_stat): Remove _stat from name.
12589 (build_vector_stat): Likewise.
12590 * tree.h (make_vector_stat): Remove macro.
12591 (build_vector_stat): Likewise.
12593 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12595 * tree.h (build_var_debug_value): Remove prototype.
12597 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12599 * tree.c (tree_cons_stat): Remove _stat from name.
12600 * tree.h (tree_cons): Remove macro.
12602 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12604 * tree.c (build_vl_exp_stat): Remove _stat from name.
12605 * tree.h (build_vl_exp): Remove macro.
12607 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12609 * tree.c (build_decl_stat): Remove _stat from name.
12610 * tree.h (build_decl): Remove macro.
12612 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12614 * gimple.c (gimple_build_with_ops_stat): Adjust.
12615 (gimple_alloc_stat): Remove _stat from name.
12616 * gimple.h (gimple_alloc): Remove macro.
12618 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12620 * tree.c (make_tree_vec_stat): Remove _stat from name.
12621 (grow_tree_vec_stat): Likewise.
12622 * tree.h (make_tree_vec_stat): Adjust prototype.
12623 (grow_tree_vec_stat): Likewise.
12624 (make_tree_vec): Remove macro.
12625 (grow_tree_vec): Likewise.
12627 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12629 * fold-const.c (fold_build1_stat_loc): Adjust.
12630 (fold_build2_stat_loc): Likewise.
12631 (fold_build3_stat_loc): Likewise.
12632 * tree.c (build0_stat): Remove _stat from name.
12633 (build1_stat): Likewise.
12634 (build2_stat): Likewise.
12635 (build3_stat): Likewise.
12636 (build4_stat): Likewise.
12637 (build5_stat): Likewise.
12638 * tree.h (build1_loc): Remove macro, and rename _stat function
12640 (build2_loc): Likewise.
12641 (build3_loc): Likewise.
12642 (build4_loc): Likewise.
12643 (build5_loc): Likewise.
12645 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12647 * tree.c (make_int_cst_stat): Remove _stat from name.
12648 * tree.h (make_int_cst_stat): Adjust prototype.
12649 (make_int_cst): Remove macro.
12651 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12653 * tree.c (make_tre_binfo_stat): Remove _stat from name.
12654 * tree.h (make_tree_binfo_stat): Adjust prototype.
12655 (make_tree_binfo): Remove.
12657 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12659 * tree.c (copy_node_stat): Rename to copy_node.
12660 (build_distinct_type_copy): Adjust.
12661 * tree.h (copy_node_stat): Adjust prototype.
12662 (copy_node): Remove macro.
12664 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12666 * tree.c (make_node_stat): rename to make_node.
12667 (build_tree_list_stat): Adjust.
12668 (build0_stat): Likewise.
12669 (build2_stat): Likewise.
12670 (build3_stat): Likewise.
12671 (build4_stat): Likewise.
12672 (build5_stat): Likewise.
12673 (build_decl_stat): Likewise.
12674 * tree.h (make_node_stat): Adjust prototype.
12675 (make_node): remove macro.
12677 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
12679 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
12680 (PPC_FEATURE2_SCV): Likewise.
12681 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
12683 2017-07-28 Tamar Christina <tamar.christina@arm.com>
12685 * config/aarch64/aarch64.c
12686 (aarch64_internal_mov_immediate): Add new special pattern.
12687 * config/aarch64/aarch64.md (*movdi_aarch64):
12688 Add reg/32bit const mov case.
12690 2017-07-28 Tamar Christina <tamar.christina@arm.com>
12691 Richard Sandiford <richard.sandiford@linaro.org>
12693 * config/aarch64/aarch64.md (mov<mode>): Generalize.
12694 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
12695 Add integer and movi cases.
12696 (movi-split-hf-df-sf split, fp16): New.
12697 (enabled): Added TARGET_FP_F16INST.
12698 * config/aarch64/iterators.md (GPF_HF): New.
12699 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
12701 2017-07-28 Tamar Christina <tamar.christina@arm.com>
12703 * config/aarch64/aarch64.c
12704 (aarch64_simd_container_mode): Add prototype.
12705 (aarch64_expand_mov_immediate): Add HI support.
12706 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
12707 (aarch64_can_const_movi_rtx_p): New.
12708 (aarch64_preferred_reload_class):
12709 Remove restrictions of using FP registers for certain SIMD operations.
12710 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
12711 (aarch64_valid_floating_const): Add integer move validation.
12712 (aarch64_simd_imm_scalar_p): Remove.
12713 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
12714 (aarch64_legitimate_constant_p): Expand list of supported cases.
12715 * config/aarch64/aarch64-protos.h
12716 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
12717 (aarch64_reinterpret_float_as_int): New.
12718 (aarch64_simd_imm_scalar_p): Remove.
12719 * config/aarch64/constraints.md (Uvi): New.
12720 (Dd): Split into Ds and new Dd.
12721 * config/aarch64/aarch64.md (*movsi_aarch64):
12723 (*movdi_aarch64): Add SIMD mov case.
12725 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12727 * tree-predcom.c: (struct chain): Handle store-store chain in which
12728 stores for elimination only store loop invariant values.
12729 (execute_pred_commoning_chain): Ditto.
12730 (prepare_initializers_chain_store_elim): Ditto.
12731 (prepare_finalizers): Ditto.
12732 (is_inv_store_elimination_chain): New function.
12733 (initialize_root_vars_store_elim_1): New function.
12735 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12737 * tree-predcom.c: Revise general description of the pass.
12738 (enum chain_type): New enum type for store elimination.
12739 (struct chain): New field supporting store elimination.
12740 (struct component): Ditto.
12741 (dump_chain): Dump store-stores chain.
12742 (release_chain): Release resources.
12743 (split_data_refs_to_components): Compute and create component
12744 contains only stores for elimination.
12745 (get_chain_last_ref_at): New function.
12746 (make_invariant_chain): Initialization.
12747 (make_rooted_chain): Specify chain type in parameter and record it.
12748 (add_looparound_copies): Skip for store-stores chain.
12749 (determine_roots_comp): Compute type of chain and pass it to
12751 (initialize_root_vars_store_elim_2): New function.
12752 (finalize_eliminated_stores): New function.
12753 (remove_stmt): Handle store for elimination.
12754 (execute_pred_commoning_chain): Execute predictive commoning on
12755 store-store chains.
12756 (determine_unroll_factor): Skip unroll for store-stores chain.
12757 (prepare_initializers_chain_store_elim): New function.
12758 (prepare_initializers_chain): Hanlde store-store chain.
12759 (prepare_finalizers_chain, prepare_finalizers): New function.
12760 (tree_predictive_commoning_loop): Return integer value indicating
12761 if loop is unrolled or lcssa form is corrupted.
12762 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
12764 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12766 * tree-predcom.c (initialize_root): Delete.
12767 (execute_pred_commoning_chain): Initialize root vars and replace
12768 reference of non-combined chain directly, rather than call above
12771 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12773 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
12774 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
12776 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12778 * tree-predcom.c (struct chain): New field init_seq.
12779 (release_chain): Release init_seq.
12780 (prepare_initializers_chain): Record intialization stmts in above
12782 (insert_init_seqs): New function.
12783 (tree_predictive_commoning_loop): Call insert_init_seqs.
12785 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12787 * tree-predcom.c (determine_roots_comp): Skip trivial components.
12789 2017-07-28 Richard Biener <rguenther@suse.de>
12791 * match.pd: Remove superfluous :c.
12792 * genmatch.c (simplify::id): Add member.
12793 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
12795 (current_id): New global.
12796 (dt_node::parent): Move from ...
12797 (dt_operand::parent): ... here. Add for_id member.
12798 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
12799 (decision_tree::find_node): Relax order requirement when
12800 merging DT_TRUE nodes to ones inbetween the current simplify
12801 and the one we try to merge with. Add diagnostic whenever
12802 we need to enforce pattern order by not merging.
12803 (decision_tree::insert): Set current_id.
12804 (decision_tree::print_node): Dump parent node and for_id.
12805 (parser::last_id): Add member.
12806 (parser::push_simplify): Assign unique id.
12807 (parser::parser): Initialize last_id.
12809 2017-07-28 Martin Liska <mliska@suse.cz>
12812 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
12813 gimple_build_debug_bind.
12815 2017-07-28 Richard Biener <rguenther@suse.de>
12817 PR tree-optimization/81502
12818 * match.pd: Add pattern combining BIT_INSERT_EXPR with
12820 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
12822 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
12823 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
12824 for BIT_FIELD_REF args.
12825 * fold-const.c (make_bit_field_ref): Likewise.
12826 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
12828 2017-07-28 Jakub Jelinek <jakub@redhat.com>
12831 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
12832 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
12833 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
12834 Or it into SANITIZER_UNDEFINED.
12835 * ubsan.c: Include gimple-fold.h and varasm.h.
12836 (ubsan_expand_ptr_ifn): New function.
12837 (instrument_pointer_overflow): New function.
12838 (maybe_instrument_pointer_overflow): New function.
12839 (instrument_object_size): Formatting fix.
12840 (pass_ubsan::execute): Call instrument_pointer_overflow
12841 and maybe_instrument_pointer_overflow.
12842 * internal-fn.c (expand_UBSAN_PTR): New function.
12843 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
12844 * sanitizer.def (__ubsan_handle_pointer_overflow,
12845 __ubsan_handle_pointer_overflow_abort): New builtins.
12846 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
12847 * internal-fn.def (UBSAN_PTR): New internal function.
12848 * opts.c (sanitizer_opts): Add pointer-overflow.
12849 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
12850 * fold-const.c (build_range_check): Compute pointer range check in
12851 integral type if pointer arithmetics would be needed. Formatting
12854 2017-07-28 Martin Liska <mliska@suse.cz>
12857 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
12858 parameters that are of a variable-length.
12860 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
12862 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
12864 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
12865 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
12866 (CRT_CALL_STATIC_FUNCTION): Likewise.
12867 (ASM_DEFAULT_SPEC): New define.
12868 (ASM_SPEC32): Likewise.
12869 (ASM_SPEC64): Likewise.
12870 (ASM_SPEC_COMMON): Likewise.
12871 (ASM_SPEC): Likewise.
12872 (INVALID_64BIT): Likewise.
12873 (LINK_OS_DEFAULT_SPEC): Likewise.
12874 (LINK_OS_SPEC32): Likewise.
12875 (LINK_OS_SPEC64): Likewise.
12876 (POWERPC_LINUX): Likewise.
12877 (PTRDIFF_TYPE): Likewise.
12878 (RESTORE_FP_PREFIX): Likewise.
12879 (RESTORE_FP_SUFFIX): Likewise.
12880 (SAVE_FP_PREFIX): Likewise.
12881 (SAVE_FP_SUFFIX): Likewise.
12882 (SIZE_TYPE): Likewise.
12883 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
12884 (TARGET_64BIT): Likewise.
12885 (TARGET_64BIT): Likewise.
12886 (TARGET_AIX): Likewise.
12887 (WCHAR_TYPE_SIZE): Likewise.
12888 (WCHAR_TYPE): Undefine.
12889 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
12890 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
12891 (CPP_OS_RTEMS_SPEC): Delete.
12892 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
12893 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
12895 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
12897 2017-07-28 Jakub Jelinek <jakub@redhat.com>
12899 PR tree-optimization/81578
12900 * tree-parloops.c (build_new_reduction): Bail out if
12901 reduction_code isn't one of the standard OpenMP reductions.
12902 Move the details printing after that decision.
12904 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
12906 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
12907 related to reload_in_progress.
12908 (splat_input_operand): Likewise.
12909 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
12911 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
12913 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
12914 (TARGET_INSTANTIATE_DECLS): Likewise.
12915 (legitimate_indexed_address_p): Delete reload_in_progress code.
12916 (rs6000_debug_legitimate_address_p): Likewise.
12917 (rs6000_eliminate_indexed_memrefs): Likewise.
12918 (rs6000_emit_le_vsx_store): Likewise.
12919 (rs6000_emit_move_si_sf_subreg): Likewise.
12920 (rs6000_emit_move): Likewise.
12921 (register_to_reg_type): Likewise.
12922 (rs6000_pre_atomic_barrier): Likewise.
12923 (rs6000_machopic_legitimize_pic_address): Likewise.
12924 (rs6000_allocate_stack_temp): Likewise.
12925 (rs6000_address_for_fpconvert): Likewise.
12926 (rs6000_address_for_altivec): Likewise.
12927 (rs6000_secondary_memory_needed_rtx): Delete function.
12928 (rs6000_check_sdmode): Likewise.
12929 (rs6000_alloc_sdmode_stack_slot): Likewise.
12930 (rs6000_instantiate_decls): Likewise.
12931 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
12932 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
12933 Delete reload_in_progress.
12934 (*vec_reload_and_plus_<mptrsize>): Likewise.
12935 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
12936 (vsx_div_v2di): Likewise.
12937 (vsx_udiv_v2di): Likewise.
12939 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
12941 * config/rs6000/rs6000.opt (mlra): Replace with stub.
12942 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
12943 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
12944 (rs6000_debug_reg_global): Delete print of LRA status.
12945 (rs6000_option_override_internal): Delete dead LRA related code.
12946 (rs6000_lra_p): Delete function.
12947 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
12949 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
12951 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
12952 * config/riscv/rtems.h: New file.
12954 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12955 Sudakshina Das <sudi.das@arm.com>
12957 * config/aarch64/aarch64.md
12958 (define_split for and<mode>3nr_compare): Move
12959 non aarch64_logical_operand to a register.
12960 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
12961 register immediate operand to a register.
12962 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
12964 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
12966 PR middle-end/81564
12967 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
12969 2017-07-27 Richard Biener <rguenther@suse.de>
12971 PR tree-optimization/81573
12972 PR tree-optimization/81494
12973 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
12974 multi defuse cycle case.
12976 2017-07-27 Richard Biener <rguenther@suse.de>
12978 PR tree-optimization/81571
12979 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
12982 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
12984 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
12985 earlier and only if MASK_FPU is set. Adjust formatting.
12987 2017-07-27 Martin Liska <mliska@suse.cz>
12989 * opt-functions.awk: Add validation of value of Init.
12990 * optc-gen.awk: Pass new argument.
12992 2017-07-27 Martin Liska <mliska@suse.cz>
12994 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
12995 Fix wrong condition.
12997 2017-07-27 Martin Liska <mliska@suse.cz>
12999 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
13000 BBs and edges seen by autoFDO.
13002 2017-07-27 Richard Biener <rguenther@suse.de>
13004 PR tree-optimization/81502
13005 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
13006 with incompatible but same sized type.
13007 (execute_update_addresses_taken): Likewise.
13009 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
13011 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
13012 flag_tree_loop_vectorize rather than flag_tree_vectorize.
13014 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13017 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
13018 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
13019 Change s_operand to memory_operand.
13021 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
13023 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
13024 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
13025 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
13026 Emit instructions rather than returning an expression. Handle TFmode
13027 and KFmode by casting to TImode.
13028 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
13029 (rs6000_emit_le_vsx_store): Likewise.
13030 * config/rs6000/vsx.md (VSX_TI): New iterator.
13031 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
13032 (*vsx_le_undo_permute_<mode>): Likewise.
13033 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
13034 emit the split sequence.
13035 (*vsx_le_perm_store_<mode>): Likewise.
13037 2017-07-27 Jakub Jelinek <jakub@redhat.com>
13039 PR tree-optimization/81555
13040 PR tree-optimization/81556
13041 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
13042 if true, force CHANGED for the recursive invocation.
13043 (reassociate_bb): Remember original length of ops array, pass
13044 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
13046 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
13047 attributes for noipa attribute. For naked attribute use
13048 lookup_attribute first before lookup_attribute_spec.
13049 * final.c (rest_of_handle_final): Disable IPA RA for functions with
13051 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
13052 for functions with noipa attribute.
13053 (cgraph_externally_visible_p): Return true for functions with noipa
13055 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
13056 for functions with noipa attribute.
13057 * doc/extend.texi: Document noipa function attribute.
13058 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
13059 also for functions with noipa attribute.
13060 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
13062 2017-07-26 Andrew Pinski <apinski@cavium.com>
13064 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
13065 vec_unalign_load_cost and vec_unalign_store_cost.
13067 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
13069 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
13070 -mvsx-small-integer option.
13071 (ISA_3_0_MASKS_IEEE): Likewise.
13072 (OTHER_VSX_VECTOR_MASKS): Likewise.
13073 (POWERPC_MASKS): Likewise.
13074 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
13075 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
13076 code, only testing for DImode being allowed in non-VSX floating
13078 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
13079 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
13081 (rs6000_option_override_internal): Delete -mvsx-small-integer.
13082 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
13083 TARGET_P8_VECTOR test.
13084 (rs6000_secondary_reload_simple_move): Likewise.
13085 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
13086 since TARGET_P9_VECTOR was already tested.
13087 (rs6000_opt_masks): Remove -mvsx-small-integer.
13088 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
13089 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
13091 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
13092 test for TARGET_VEXTRACTUB was used, and that uses
13094 (p9 extract splitter): Likewise.
13095 (vsx_extract_<mode>_di_p9): Likewise.
13096 (vsx_extract_<mode>_store_p9): Likewise.
13097 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
13098 for TARGET_P9_VECTOR was used. Delete code that is now dead with
13099 the elimination of TARGET_VSX_SMALL_INTEGER.
13100 (vsx_extract_<mode>_p8): Likewise.
13101 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
13102 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
13103 (vsx_set_<mode>_p9): Likewise.
13104 (vsx_set_v4sf_p9): Likewise.
13105 (vsx_set_v4sf_p9_zero): Likewise.
13106 (vsx_insert_extract_v4sf_p9): Likewise.
13107 (vsx_insert_extract_v4sf_p9_2): Likewise.
13108 * config/rs6000/rs6000.md (sign extend splitter): Change
13109 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
13110 (floatsi<mode>2_lfiwax_mem): Likewise.
13111 (floatunssi<mode>2_lfiwzx_mem): Likewise.
13112 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
13113 since a test for TARGET_P9_VECTOR was used.
13114 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
13115 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
13116 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
13117 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
13118 TARGET_P8_VECTOR test.
13119 (fix_trunc<mode>si2_stfiwx): Likewise.
13120 (fix_trunc<mode>si2_internal): Likewise.
13121 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
13122 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
13124 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
13125 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
13126 TARGET_P8_VECTOR test.
13127 (fixuns_trunc<mode>si2_stfiwx): Likewise.
13128 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
13129 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
13131 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
13132 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
13133 since a test for TARGET_P9_VECTOR was used.
13134 (splitter for loading small constants): Likewise.
13136 2017-07-26 Andrew Pinski <apinski@cavium.com>
13138 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
13141 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
13144 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
13145 (fp_valid_at): Likewise.
13147 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
13149 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
13150 (qdf24xx_addrcost_table): Likewise.
13151 (cortexa57_tunings): Update to use generic_branch_cost.
13152 (cortexa72_tunings): Likewise.
13153 (cortexa73_tunings): Likewise.
13154 (qdf24xx_tunings): Likewise.
13156 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
13158 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
13159 (thunderx2t99_branch_cost): Likewise.
13160 (cortexa35_tunings): Update to use generic_branch_cost.
13161 (cortexa53_tunings): Likewise.
13162 (cortexa57_tunings): Likewise.
13163 (cortexa72_tunings): Likewise.
13164 (cortexa73_tunings): Likewise.
13165 (thunderx2t99_tunings): Likewise.
13167 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
13169 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
13170 (sparc_option_override): Honour MASK_FSMULD.
13171 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
13172 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
13173 * config/sparc/sparc.opt (mfsmuld): New option.
13174 * doc/invoke.texi (mfsmuld): Document option.
13176 2017-07-26 Marek Polacek <polacek@redhat.com>
13178 PR middle-end/70992
13179 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
13181 2017-07-26 Richard Biener <rguenther@suse.de>
13183 * gimple-match-head.c (do_valueize): Return OP if valueize
13185 (get_def): New helper to get at the def stmt of a SSA name
13186 if valueize allows.
13187 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
13188 do_valueize to get at the def stmt.
13189 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
13191 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
13193 PR middle-end/46932
13194 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
13196 2017-07-26 Martin Liska <mliska@suse.cz>
13199 * function.c (expand_function_start): Make expansion of
13200 nonlocal_goto_save_area after parm_birth_insn.
13202 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
13204 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
13205 from all CPU target flags enable members.
13207 2017-07-26 Richard Biener <rguenther@suse.de>
13209 * genmatch.c (dt_simplify::gen): Make iterator vars const.
13210 (decision_tree::gen): Make 'type' const.
13211 (write_predicate): Likewise.
13213 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
13215 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
13216 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
13217 (rs6000_option_override_internal): Likewise.
13218 (rs6000_expand_vector_set): Likewise.
13219 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
13220 (TARGET_UPPER_REGS_SF): Likewise.
13221 (TARGET_UPPER_REGS_DI): Likewise.
13222 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
13223 (TARGET_DIRECT_MOVE_64BIT): Likewise.
13224 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
13225 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
13226 (Splitters for DI constants in Altivec registers): Likewise.
13227 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
13228 (vsx_set_v4sf_p9): Likewise.
13229 (vsx_set_v4sf_p9_zero): Likewise.
13230 (vsx_insert_extract_v4sf_p9): Likewise.
13231 (vsx_insert_extract_v4sf_p9_2): Likewise.
13233 2017-07-25 Carl Love <cel@us.ibm.com>
13235 * doc/extend.texi: Update the built-in documentation file for the
13236 existing built-in functions
13237 vector signed char vec_cnttz (vector signed char);
13238 vector unsigned char vec_cnttz (vector unsigned char);
13239 vector signed short vec_cnttz (vector signed short);
13240 vector unsigned short vec_cnttz (vector unsigned short);
13241 vector signed int vec_cnttz (vector signed int);
13242 vector unsigned int vec_cnttz (vector unsigned int);
13243 vector signed long long vec_cnttz (vector signed long long);
13244 vector unsigned long long vec_cnttz (vector unsigned long long);
13246 2017-07-25 Andrew Pinski <apinski@cavium.com>
13248 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
13249 accesses where the use is for the first operand of a BIT_INSERT.
13251 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
13254 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
13255 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
13257 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
13259 * config/i386/gstabs.h: Delete.
13260 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
13262 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
13264 * config/i386/i386.c (ix86_decompose_address): Do not check for
13265 register RTX when looking at index_reg or base_reg.
13266 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
13268 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
13270 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
13271 to update EH info here.
13273 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
13275 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
13277 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
13279 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
13281 2017-07-25 Torsten Duwe <duwe@suse.de>
13283 * common.opt: Introduce -fpatchable-function-entry
13284 command line option, and its variables function_entry_patch_area_size
13285 and function_entry_patch_area_start.
13286 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
13287 including a two-value parser.
13288 * target.def (print_patchable_function_entry): New target hook.
13289 * targhooks.h (default_print_patchable_function_entry): New function.
13290 * targhooks.c (default_print_patchable_function_entry): Likewise.
13291 * toplev.c (process_options): Switch off IPA-RA if
13292 patchable function entries are being generated.
13293 * varasm.c (assemble_start_function): Look at the
13294 patchable-function-entry command line switch and current
13295 function attributes and maybe generate NOP instructions by
13296 calling the print_patchable_function_entry hook.
13297 * doc/extend.texi: Document patchable_function_entry attribute.
13298 * doc/invoke.texi: Document -fpatchable_function_entry
13299 command line option.
13300 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
13302 * doc/tm.texi: Re-generate.
13304 2017-07-25 Jakub Jelinek <jakub@redhat.com>
13307 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
13308 TARGET_AVX512DQ rather than TARGET_AVX512BW.
13310 2017-07-25 Tamar Christina <tamar.christina@arm.com>
13312 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
13314 2017-07-25 Richard Biener <rguenther@suse.de>
13316 PR tree-optimization/81455
13317 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
13318 not walk in cycles when looking for guards.
13320 2017-07-25 Richard Biener <rguenther@suse.de>
13322 PR tree-optimization/81529
13323 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
13324 when optimizing backedge uses.
13326 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
13328 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
13330 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
13331 to dl_section_ref. On AIX, append an expression to subtract
13332 the size of the section length to dl_section_ref.
13334 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
13336 * configure.ac: If any of the config.* scripts fail, exit 1.
13337 * configure: Regenerate.
13339 2017-07-25 Richard Biener <rguenther@suse.de>
13341 PR middle-end/81546
13342 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
13343 of immediate uses, be more verbose on errors.
13345 2017-07-25 Richard Biener <rguenther@suse.de>
13347 PR tree-optimization/81510
13348 * tree-vect-loop.c (vect_is_simple_reduction): When the
13349 reduction stmt is not inside the loop bail out.
13351 2017-07-25 Richard Biener <rguenther@suse.de>
13353 PR tree-optimization/81303
13354 * tree-vect-loop-manip.c (vect_loop_versioning): Build
13355 profitability check against LOOP_VINFO_NITERSM1.
13357 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
13359 * domwalk.c (cmp_bb_postorder): Simplify.
13360 (sort_bbs_postorder): New function. Use it...
13361 (dom_walker::walk): ...here to optimize common cases.
13363 2017-07-25 Martin Liska <mliska@suse.cz>
13366 * ipa-visibility.c (function_and_variable_visibility): Make the
13367 redirection just on target that supports aliasing.
13368 Fix GNU coding style.
13370 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
13373 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
13375 * config/arm/rtems.h: Likewise.
13376 * config/bfin/rtems.h: Likewise.
13377 * config/i386/rtemself.h: Likewise.
13378 * config/lm32/rtems.h: Likewise.
13379 * config/m32c/rtems.h: Likewise.
13380 * config/m68k/rtemself.h: Likewise.
13381 * config/microblaze/rtems.h: Likewise.
13382 * config/mips/rtems.h: Likewise.
13383 * config/moxie/rtems.h: Likewise.
13384 * config/nios2/rtems.h: Likewise.
13385 * config/powerpcspe/rtems.h: Likewise.
13386 * config/rs6000/rtems.h: Likewise.
13387 * config/rtems.h: Likewise.
13388 * config/sh/rtems.h: Likewise.
13389 * config/sh/rtemself.h: Likewise.
13390 * config/sparc/rtemself.h: Likewise.
13392 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
13395 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
13396 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
13397 * tree-ssa-structalias.c (alias_get_name): Same.
13399 2017-07-25 Bin Cheng <bin.cheng@arm.com>
13402 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
13403 instructions if no du chain is found.
13405 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
13407 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
13409 2017-07-25 Richard Biener <rguenther@suse.de>
13411 PR middle-end/81505
13412 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
13415 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
13417 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
13418 upper-regs options.
13419 (ISA_2_7_MASKS_SERVER): Likewise.
13420 (ISA_3_0_MASKS_IEEE): Likewise.
13421 (OTHER_P8_VECTOR_MASKS): Likewise.
13422 (OTHER_VSX_VECTOR_MASKS): Likewise.
13423 (POWERPC_MASKS): Likewise.
13424 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
13425 duplicate list of options.
13426 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
13427 explicit -mupper-regs options.
13428 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
13429 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
13430 alias for -mupper-regs-df.
13431 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
13432 (rs6000_init_hard_regno_mode_ok): Likewise.
13433 (rs6000_option_override_internal): Likewise.
13434 (rs6000_opt_masks): Likewise.
13435 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
13436 options in terms of whether -mvsx or -mpower8-vector was used.
13437 (TARGET_UPPER_REGS_DI): Likewise.
13438 (TARGET_UPPER_REGS_SF): Likewise.
13439 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
13440 -mupper-regs-* options.
13442 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
13444 * passes.c (emergency_dump_function): Print some empty lines and a
13445 header before the RTL dump.
13447 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
13449 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
13451 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
13454 * config/aarch64/aarch64.c (aarch64_classify_symbol):
13455 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
13457 2017-07-24 Carl Love <cel@us.ibm.com>
13459 * config/rs6000/rs6000-c.c: Add support for built-in functions
13460 vector float vec_extract_fp32_from_shorth (vector unsigned short);
13461 vector float vec_extract_fp32_from_shortl (vector unsigned short);
13462 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
13463 vec_extract_fp_from_shortl): Add defines for the two builtins.
13464 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
13465 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
13467 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
13468 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
13469 * doc/extend.texi: Update the built-in documentation file for the
13470 new built-in function.
13472 2017-07-24 Jakub Jelinek <jakub@redhat.com>
13475 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
13477 * doc/generic.texi: Likewise.
13478 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
13479 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
13481 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
13483 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
13484 (aarch64_mls_elt_merge<mode>): Likewise.
13486 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
13488 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
13489 having __cxa_atexit.
13491 2017-07-23 Michael Collison <michael.collison@arm.com>
13493 * config/arm/arm.c (arm_option_override): Deprecate
13494 use of -mstructure-size-boundary.
13495 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
13496 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
13498 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13501 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
13502 Reduce cost estimate for direct moves.
13504 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
13507 * config/i386/i386.c (ix86_option_override_internal): Disable
13508 BMI, BMI2 and TBM instructions for -m16.
13510 2017-07-21 Carl Love <cel@us.ibm.com>
13512 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13513 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
13514 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
13515 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
13516 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
13517 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
13518 VMULOSW): New enum "unspec" values.
13519 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
13520 altivec_vmulosw): New patterns.
13521 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
13522 VMULOSW): Add definitions.
13524 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
13526 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
13527 (qdf24xx): Likewise.
13528 * config/aarch64/aarch64-options-extensions.def (rdma); New.
13529 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
13530 (AARCH64_FL_V8_1): Renumber.
13531 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
13532 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
13533 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
13534 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
13535 rdma to feature modifiers list.
13537 2017-07-21 Yury Gribov <tetra2005@gmail.com>
13539 PR middle-end/56727
13540 * ipa-visibility (function_and_variable_visibility): Convert
13541 recursive PLT call to direct call if appropriate.
13543 2017-07-21 Andrew Pinski <apinski@cavium.com>
13545 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
13546 operand 1 to see if the types precision matches.
13547 * fold-const.c (operand_equal_p): Likewise.
13549 2017-07-21 Richard Biener <rguenther@suse.de>
13551 PR tree-optimization/81303
13552 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
13553 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
13554 (vect_peeling_hash_get_lowest_cost): Adjust.
13555 (vect_enhance_data_refs_alignment): Likewise. Use
13556 vect_get_peeling_costs_all_drs to compute the penalty for no
13557 peeling to match up costs.
13559 2017-07-21 Richard Biener <rguenther@suse.de>
13561 PR tree-optimization/81500
13562 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
13563 we didn't identify a reduction path.
13565 2017-07-21 Tom de Vries <tom@codesourcery.com>
13566 Cesar Philippidis <cesar@codesourcery.com>
13568 PR gcov-profile/81442
13569 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
13572 2017-07-21 Tom de Vries <tom@codesourcery.com>
13575 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
13577 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
13578 nvptx_override_options_after_change.
13580 2017-07-21 Ulrich Drepper <drepper@redhat.com>
13582 * dwarf2out.c (output_file_names): Avoid double testing for
13583 dwarf_version >= 5.
13585 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
13587 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
13589 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
13591 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
13593 (try_crossjump_to_edge): Do not punt on partitioned functions.
13595 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
13597 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
13598 Put all BBs reachable only via paths crossing cold region to cold
13600 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
13602 2016-07-21 Richard Biener <rguenther@suse.de>
13604 PR tree-optimization/81303
13605 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
13606 into account prologue and epilogue iterations when raising
13607 min_profitable_iters to sth at least covering one vector iteration.
13609 2017-07-21 Tamar Christina <tamar.christina@arm.com>
13611 * config/arm/arm.c (arm_test_cpu_arch_dat):
13614 2017-07-20 Nathan Sidwell <nathan@acm.org>
13616 Remove TYPE_METHODS.
13617 * tree.h (TYPE_METHODS): Delete.
13618 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
13619 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
13620 (dbxout_type_methods): Scan TYPE_FIELDS.
13621 (dbxout_type): Don't check TYPE_METHODS here.
13622 * function.c (use_register_for_decl): Always ignore register for
13623 class types when not optimizing.
13624 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
13625 * tree.c (free_lang_data_in_type): Stitch out member functions and
13626 templates from TYPE_FIELDS.
13627 (build_distinct_type_copy, verify_type_variant,
13628 verify_type): Member fns are on TYPE_FIELDS.
13629 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
13630 * tree-pretty-print.c (dump_generic_node): Likewise.
13632 2017-07-20 Jakub Jelinek <jakub@redhat.com>
13635 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
13636 V2TImode and V4TImode.
13637 (ix86_expand_vector_extract): Likewise.
13638 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
13639 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
13640 (ssescalarmode): Handle V4TImode and V2TImode.
13641 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
13642 (*vec_extractv2ti, *vec_extractv4ti): New insns.
13643 (VEXTRACTI128_MODE): New mode iterator.
13644 (splitter for *vec_extractv?ti first element): New.
13645 (VEC_INIT_MODE): New mode iterator.
13646 (vec_init<mode>): Consolidate 3 expanders into one using
13647 VEC_INIT_MODE mode iterator.
13649 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
13651 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
13652 non_spilled_static_chain_regno_p.
13654 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
13656 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
13658 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
13660 * bb-reorder.c (connect_traces): Allow copying of blocks within
13663 2017-07-20 Richard Biener <rguenther@suse.de>
13665 * gimple.h (gimple_phi_result): Add gphi * overload.
13666 (gimple_phi_result_ptr): Likewise.
13667 (gimple_phi_arg): Likewise. Adjust index assert to only
13668 allow actual argument accesses rather than all slots available
13670 (gimple_phi_arg_def): Add gphi * overload.
13671 * tree-phinodes.c (make_phi_node): Initialize only actual
13673 (resize_phi_node): Clear memory not covered by old node,
13674 do not initialize excess argument slots.
13675 (reserve_phi_args_for_new_edge): Initialize new argument slot
13678 2017-07-20 Bin Cheng <bin.cheng@arm.com>
13680 PR tree-optimization/81388
13682 2016-07-21 Bin Cheng <bin.cheng@arm.com>
13684 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
13685 by removing computation of may_be_zero.
13687 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
13688 Tom de Vries <tom@codesourcery.com>
13690 PR middle-end/81030
13691 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
13692 when gimple level profile disagrees with what RTL expander did.
13694 2017-07-20 Richard Biener <rguenther@suse.de>
13696 PR tree-optimization/61171
13697 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
13698 (vect_analyze_stmt): Add slp instance parameter.
13699 (vectorizable_reduction): Likewise.
13700 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
13701 (vect_is_simple_reduction): Deal with chains not detected
13702 as SLP reduction chain, specifically not properly associated
13703 chains containing a mix of plus/minus.
13704 (get_reduction_op): Remove.
13705 (get_initial_defs_for_reduction): Simplify, pass in whether
13706 this is a reduction chain, pass in the SLP node for the PHIs.
13707 (vect_create_epilog_for_reduction): Get the SLP instance as
13709 (vectorizable_reduction): Get the SLP instance as arg.
13710 During analysis remember the SLP node with the PHIs in the
13711 instance. Simplify getting at the vectorized reduction PHIs.
13712 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
13713 through SLP instance.
13714 (vect_slp_analyze_operations): Likewise.
13715 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
13716 (vect_transform_stmt): Likewise.
13718 2017-07-20 Tom de Vries <tom@codesourcery.com>
13720 PR tree-optimization/81489
13721 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
13722 read of phi arg location to before loop that modifies phi.
13724 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13726 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
13729 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
13731 PR middle-end/81331
13732 * except.c (execute): Fix ordering issue.
13734 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
13736 PR rtl-optimization/81423
13737 * combine.c (make_compound_operation_int): Don't try to optimize
13738 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
13740 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
13742 PR rtl-optimization/81423
13743 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
13744 with a constant that is -1 in the truncated to mode.
13746 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
13748 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
13749 (determine_unlikely_bbs): ... here.
13750 * predict.h (propagate_unlikely_bbs_forward): Declare.
13751 * cfgexpand.c (pass_expand::execute): Use it.
13752 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
13754 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
13755 propagate_unlikely_bbs_forward.
13757 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
13759 PR middle-end/81331
13760 * except.c (maybe_add_nop_after_section_switch): New function.
13763 2017-07-19 Tom de Vries <tom@codesourcery.com>
13765 * gimple.h (gimple_phi_set_arg): Make assert more strict.
13767 2017-07-19 Tom de Vries <tom@codesourcery.com>
13769 * gimple.h (gimple_phi_arg): Make assert more strict.
13771 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
13773 * config.gcc (powerpc*-*-*): Add mmintrin.h.
13774 * config/rs6000/mmintrin.h: New file.
13775 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
13777 2017-07-19 Jakub Jelinek <jakub@redhat.com>
13779 PR tree-optimization/81346
13780 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
13782 2017-07-19 Tom de Vries <tom@codesourcery.com>
13784 * config/nvptx/nvptx.md (VECIM): Add V2DI.
13786 2017-07-19 Tom de Vries <tom@codesourcery.com>
13788 * config/nvptx/nvptx-modes.def: Add V2DImode.
13789 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
13790 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
13791 (nvptx_output_mov_insn): Handle lack of mov.b128.
13792 (nvptx_print_operand): Handle 'H' and 'L' codes.
13793 (nvptx_vector_mode_supported): Allow V2DImode.
13794 (nvptx_preferred_simd_mode): New function.
13795 (nvptx_data_alignment): New function.
13796 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
13797 nvptx_preferred_simd_mode.
13798 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
13800 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
13802 2017-07-19 Tom de Vries <tom@codesourcery.com>
13804 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
13805 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
13806 (nvptx_vector_mode_supported): New function. Allow V2SImode.
13807 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
13808 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
13809 (mov<VECIM>_insn): New define_insn.
13810 (define_expand "mov<VECIM>): New define_expand.
13812 2017-07-19 Tom de Vries <tom@codesourcery.com>
13814 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
13816 2017-07-19 Jakub Jelinek <jakub@redhat.com>
13818 PR tree-optimization/81346
13819 * fold-const.h (fold_div_compare, range_check_type): Declare.
13820 * fold-const.c (range_check_type): New function.
13821 (build_range_check): Use range_check_type.
13822 (fold_div_compare): No longer static, rewritten into
13823 a match.pd helper function.
13824 (fold_comparison): Don't call fold_div_compare here.
13825 * match.pd (X / C1 op C2): New optimization using fold_div_compare
13826 as helper function.
13828 2017-07-19 Nathan Sidwell <nathan@acm.org>
13830 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
13831 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
13832 * tree.c (find_decls_types_r, verify_type): Use
13833 TYPE_{MIN,MAX}_VALUE_RAW.
13834 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
13835 (hash_tree): Likewise.
13836 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
13838 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
13841 2017-07-18 Tom de Vries <tom@codesourcery.com>
13843 PR middle-end/81464
13844 * omp-expand.c (expand_omp_for_static_chunk): Handle
13845 equal-argument loop exit phi.
13847 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
13850 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
13851 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
13852 operand 2 predicate.
13853 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
13854 operand 2 predicate.
13855 (ror,rol -> rorx splitters): Use const_int_operand as
13856 operand 2 predicate.
13858 2017-06-18 Richard Biener <rguenther@suse.de>
13860 PR tree-optimization/81410
13861 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
13862 the gap in the ! slp_perm SLP case after each group.
13864 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
13866 PR middle-end/81463
13867 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
13870 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
13872 PR middle-end/81462
13873 * predict.c (set_even_probabilities): Cleanup; do not affect
13874 probabilities that are already known.
13875 (combine_predictions_for_bb): Call even when count is set.
13877 2017-07-18 Nathan Sidwell <nathan@acm.org>
13879 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
13882 2017-07-18 Bin Cheng <bin.cheng@arm.com>
13885 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
13886 optimization for loop niter analysis.
13888 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
13891 * config/avr/avr.c (avr_optimize_casesi): Don't use
13892 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
13894 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
13896 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
13897 body_cost_vec from _vect_peel_extended_info.
13898 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
13899 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
13902 2017-07-18 Bin Cheng <bin.cheng@arm.com>
13904 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
13906 2017-07-18 Richard Biener <rguenther@suse.de>
13908 PR tree-optimization/80620
13909 PR tree-optimization/81403
13910 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
13911 info when re-using a VN table entry.
13913 2017-07-18 Richard Biener <rguenther@suse.de>
13915 PR tree-optimization/81418
13916 * tree-vect-loop.c (vectorizable_reduction): Properly compute
13917 vectype_in. Verify that with lane-reducing reduction operations
13918 we have a single def-use cycle.
13920 2017-07-17 Carl Love <cel@us.ibm.com>
13922 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
13924 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13925 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
13926 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
13927 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
13928 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
13929 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
13930 VMULOSW): New enum "unspec" values.
13931 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
13932 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
13933 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
13934 altivec_vmulosw): New patterns.
13935 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
13936 VMULOSW): Add definitions.
13937 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
13939 * config/alpha/alpha.c: Include predict.h.
13941 2017-07-17 Yury Gribov <tetra2005@gmail.com>
13943 * tree-vrp.c (compare_assert_loc): Fix comparison function
13944 to return predictable results.
13946 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
13948 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
13950 (subdi3): Likewise.
13951 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
13952 * doc/invoke.texi (mexpand-adddi): Update text.
13954 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
13956 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
13957 that also clobbers the CC register. The old expand code is moved
13959 (*arc_clzsi2): ... here.
13960 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
13961 the CC register. The old expand code is moved to ...
13962 (arc_ctzsi2): ... here.
13964 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
13966 * config/arc/arc.opt (mindexed-loads): Use initial value
13967 TARGET_INDEXED_LOADS_DEFAULT.
13968 (mauto-modify-reg): Use initial value
13969 TARGET_AUTO_MODIFY_REG_DEFAULT.
13970 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
13971 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
13972 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
13973 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
13975 2017-07-17 Martin Liska <mliska@suse.cz>
13978 * opts.c (finish_options): Do not allow -fgnu-tm
13979 w/ -fsanitize={kernel-,}address. Say sorry.
13981 2017-07-17 Bin Cheng <bin.cheng@arm.com>
13984 * tree-loop-distribution.c (classify_partition): Only assert on
13985 numer of iterations.
13986 (merge_dep_scc_partitions): Delete prameter. Update function call.
13987 (distribute_loop): Remove code handling loop with unknown niters.
13988 (pass_loop_distribution::execute): Skip loop with unknown niters.
13990 2017-07-17 Bin Cheng <bin.cheng@arm.com>
13993 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
13994 function sort_partitions_by_post_order.
13996 2017-07-17 Bin Cheng <bin.cheng@arm.com>
13998 PR tree-optimization/81374
13999 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
14000 the max index of basic blocks, rather than number of basic blocks.
14002 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
14004 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
14006 (arc_legitimate_pic_operand_p): Likewise.
14007 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
14009 (arc_needs_pcl_p): Likewise.
14010 (arc_legitimate_pc_offset_p): Likewise.
14011 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
14012 function is also used in constrains.md.
14013 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
14014 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
14015 PLUS. Only return true/false in known cases, otherwise assert.
14016 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
14017 is already called in arc_legitimate_constant_p.
14018 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
14020 (LEGITIMATE_PIC_OPERAND_P): Use
14021 arc_raw_symbolic_reference_mentioned_p function.
14022 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
14027 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
14028 Andrew Burgess <andrew.burgess@embecosm.com>
14030 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
14031 (arc_return_address_register): New function.
14032 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
14033 (arc_handle_fndecl_attribute): Add naked attribute.
14034 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
14035 (TARGET_WARN_FUNC_RETURN): Likewise.
14036 (arc_allocate_stack_slots_for_args): New function.
14037 (arc_warn_func_return): Likewise.
14038 (machine_function): Change type fn_type.
14039 (arc_compute_function_type): Consider new naked function type,
14040 change function return type.
14041 (arc_must_save_register): Adapt to handle new
14042 arc_compute_function_type's return type.
14043 (arc_expand_prologue): Likewise.
14044 (arc_expand_epilogue): Likewise.
14045 (arc_return_address_regs): Delete.
14046 (arc_return_address_register): New function.
14047 (arc_epilogue_uses): Use above function.
14048 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
14049 (arc_function_type): Change encoding, add naked type.
14050 (ARC_INTERRUPT_P): Change to handle the new encoding.
14051 (ARC_FAST_INTERRUPT_P): Likewise.
14052 (ARC_NORMAL_P): Define.
14053 (ARC_NAKED_P): Likewise.
14054 (arc_compute_function_type): Delete prototype.
14055 * config/arc/arc.md (in_ret_delay_slot): Use
14056 arc_return_address_register function.
14057 (simple_return): Likewise.
14058 (p_return_i): Likewise.
14060 2017-07-17 Jakub Jelinek <jakub@redhat.com>
14062 PR tree-optimization/81428
14063 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
14064 can't be built for those types.
14066 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
14068 Remove stuff dead since r239246.
14070 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
14071 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
14072 (avr_inform_devices): Remove dead stuff.
14074 2017-07-17 Tamar Christina <tamar.christina@arm.com>
14076 * config/arm/arm_neon.h: Fix softp typo.
14078 2017-07-17 Jakub Jelinek <jakub@redhat.com>
14080 PR tree-optimization/81365
14081 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
14082 aggregate moves onto bb predecessor edges, make sure there are no
14083 loads that could alias the lhs in between the start of bb and the
14086 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
14089 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
14090 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
14091 [LSHIFTRT, outer_code = TRUNCATE]: Same.
14093 2017-07-17 Jakub Jelinek <jakub@redhat.com>
14095 PR tree-optimization/81396
14096 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
14097 (init_symbolic_number): Initialize it to 1.
14098 (perform_symbolic_merge): Add n_ops from both operands into the new
14100 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
14101 without base_addr as useless if they need more than one operation.
14102 (bswap_replace): Handle !bswap case for NULL base_addr.
14104 2017-07-17 Tom de Vries <tom@codesourcery.com>
14107 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
14110 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
14112 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
14113 conditional builtin define __FIX_LEON3FT_B2BST.
14115 2017-07-17 Daniel Cederman <cederman@gaisler.com>
14117 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
14118 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
14121 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
14123 PR rtl-optimization/81424
14124 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
14125 to remove potential trapping from operands if -fnon-call-exceptions.
14127 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14129 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
14130 profile_proability for scalling.
14131 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
14133 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14135 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
14137 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14139 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
14140 fixpoint arithmetics.
14142 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14144 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
14145 fixpoint arithmetics.
14147 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14149 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
14150 fixpoint arithmetics.
14152 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14154 * profile-count.h (profile_probability::from_reg_br_prob_note,
14155 profile_probability::to_reg_br_prob_note): New functions.
14156 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
14157 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
14158 * predict.c (probability_reliable_p): Update.
14159 (edge_probability_reliable_p): Update.
14160 (br_prob_note_reliable_p): Update.
14161 (invert_br_probabilities): Update.
14162 (add_reg_br_prob_note): New function.
14163 (combine_predictions_for_insn): Update.
14164 * asan.c (asan_clear_shadow): Update.
14165 * cfgbuild.c (compute_outgoing_frequencies): Update.
14166 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
14167 (update_br_prob_note): Update.
14168 (rtl_verify_edges): Update.
14169 (purge_dead_edges): Update.
14170 (fixup_reorder_chain): Update.
14171 * emit-rtl.c (try_split): Update.
14172 * ifcvt.c (cond_exec_process_insns): Update.
14173 (cond_exec_process_if_block): Update.
14174 (dead_or_predicable): Update.
14175 * internal-fn.c (expand_addsub_overflow): Update.
14176 (expand_neg_overflow): Update.
14177 (expand_mul_overflow): Update.
14178 * loop-doloop.c (doloop_modify): Update.
14179 * loop-unroll.c (compare_and_jump_seq): Update.
14180 * optabs.c (emit_cmp_and_jump_insn_1): Update.
14181 * predict.h: Update.
14182 * reorg.c (mostly_true_jump): Update.
14184 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
14185 * config/alpha/alpha.c (emit_unlikely_jump): Update.
14186 * config/arc/arc.c: (emit_unlikely_jump): Update.
14187 * config/arm/arm.c: (emit_unlikely_jump): Update.
14188 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
14189 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
14190 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
14191 (ix86_print_operand): Update.
14192 (ix86_split_fp_branch): Update.
14193 (predict_jump): Update.
14194 * config/ia64/ia64.c (ia64_print_operand): Update.
14195 * config/mmix/mmix.c (mmix_print_operand): Update.
14196 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
14197 (rs6000_expand_split_stack_prologue): Update.
14198 * config/rs6000/rs6000.c: Update.
14199 * config/s390/s390.c (s390_expand_vec_strlen): Update.
14200 (s390_expand_vec_movstr): Update.
14201 (s390_expand_cs_tdsi): Update.
14202 (s390_expand_split_stack_prologue): Update.
14203 * config/sh/sh.c (sh_print_operand): Update.
14204 (expand_cbranchsi4): Update.
14205 (expand_cbranchdi4): Update.
14206 * config/sparc/sparc.c (output_v9branch): Update.
14207 * config/spu/spu.c (get_branch_target): Update.
14208 (ea_load_store_inline): Update.
14209 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
14210 * config/tilepro/tilepro.c: Update.
14212 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
14214 * gimplify.c (mostly_copy_tree_r): Revert latest change.
14215 (gimplify_save_expr): Likewise.
14217 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
14219 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
14221 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
14223 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
14225 * timevar.def (TV_IPA_FNSUMMARY): Define.
14227 2017-07-16 Daniel Cederman <cederman@gaisler.com>
14229 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
14230 to back store errata sensitive sequence from being generated.
14231 (sqrtdf2_fix): Likewise.
14233 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
14235 * tree-ssa-threadupdate.c (compute_path_counts,
14236 update_joiner_offpath_counts): Use profile_probability.
14238 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
14241 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
14243 * config/arm/arm-c.c (arm_cpu_builtins): Define
14244 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
14246 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
14248 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14249 array entries to represent __ieee128 versions of the
14250 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
14251 scalar_extract_sig, and scalar_insert_exp built-in functions.
14252 (altivec_resolve_overloaded_builtin): Add special case handling
14253 for the __builtin_scalar_insert_exp function, as represented by
14254 the P9V_BUILTIN_VEC_VSIEDP constant.
14255 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
14256 exponent support for __ieee128 argument.
14257 (VSESQP): Add scalar extract signature support for __ieee128
14259 (VSTDCNQP): Add scalar test negative support for __ieee128
14261 (VSIEQP): Add scalar insert exponent support for __int128 argument
14262 with __ieee128 result.
14263 (VSIEQPF): Add scalar insert exponent support for __ieee128
14264 argument with __ieee128 result.
14265 (VSTDCQP): Add scalar test data class support for __ieee128
14267 (VSTDCNQP): Add overload support for scalar test negative with
14268 __ieee128 argument.
14269 (VSTDCQP): Add overload support for scalar test data class
14270 __ieee128 argument.
14271 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
14272 UNSPEC_VSX_SXSIGDP.
14273 (UNSPEC_VSX_SIEXPQP): New constant.
14274 (xsxexpqp): New insn for VSX scalar extract exponent quad
14276 (xsxsigqp): New insn for VSX scalar extract significand quad
14278 (xsiexpqpf): New insn for VSX scalar insert exponent quad
14279 precision with floating point argument.
14280 (xststdcqp): New expand for VSX scalar test data class quad
14282 (xststdcnegqp): New expand for VSX scalar test negative quad
14284 (xststdcqp): New insn to match expansions for VSX scalar test data
14285 class quad precision and VSX scalar test negative quad precision.
14286 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
14287 special case operand checking to enforce that second operand of
14288 VSX scalar test data class with quad precision argument is a 7-bit
14290 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
14291 prototypes and descriptions of __ieee128 versions of
14292 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
14293 scalar_test_data_class, and scalar_test_neg built-in functions.
14295 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14297 PR tree-optimization/81162
14298 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
14299 replace a negate with an add.
14301 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
14303 * doc/invoke.texi (arm/-mcpu): Document +crypto.
14305 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
14307 * config/arm/arm-c.c (arm_cpu_builtins): Define
14308 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
14310 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
14312 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
14313 (armv8-r): Set ARM Cortex-R52 as default CPU.
14314 * config/arm/arm-tables.opt: Regenerate.
14315 * config/arm/arm-tune.md: Regenerate.
14316 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
14318 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
14319 extension for -mcpu=cortex-r52.
14321 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
14323 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
14324 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
14325 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
14326 (fp-armv8): Define it as FP_ARMv8 only.
14327 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
14328 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
14330 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
14331 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
14332 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
14333 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
14334 than TARGET_FPU_ARMV8.
14335 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
14336 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
14337 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
14339 * config/arm/neon.md (neon_vrint): Likewise.
14340 (neon_vcvt): Likewise.
14341 (neon_<fmaxmin_op><mode>): Likewise.
14342 (<fmaxmin><mode>3): Likewise.
14343 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
14344 * config/arm/predicates.md (arm_cond_move_operator): Check against
14345 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
14347 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
14349 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
14350 to top of function.
14352 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14354 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
14355 loop in comment with memset.
14357 2017-07-14 Martin Liska <mliska@suse.cz>
14359 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
14360 * dwarf2out.c (is_java): Remove the function.
14361 (output_pubname): Remove usage of the function.
14362 (lower_bound_default): Remove usage of DW_LANG_Java.
14363 (gen_compile_unit_die): Likewise.
14364 * gcc.c: Remove compiler defaults for .java and .zip files.
14365 * gimple-expr.c (remove_suffix): Change as there's no longer
14366 extension than 4-letter one.
14367 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
14368 (gimplify_save_expr): Likewise.
14369 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
14370 as it's possible even for other languages than Java.
14371 * langhooks.h (struct lang_hooks): Remove Java from a comment.
14372 * lto-opts.c (lto_write_options): Remove reference to Java.
14373 * opts.c (strip_off_ending): Update file extension handling.
14374 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
14375 * tree-eh.c (lower_resx): Likewise.
14376 * tree.c (free_lang_data_in_type): Remove dead code.
14377 (find_decls_types_r): Likewise.
14378 (build_common_builtin_nodes): Remove Java from a comment.
14379 (verify_type): Remove dead code.
14380 * varasm.c (assemble_external): Remove Java from a comment.
14382 2017-07-14 Martin Liska <mliska@suse.cz>
14384 * opts.c (finish_options): Add quotes.
14385 (common_handle_option): Likewise.
14387 2017-07-14 Martin Liska <mliska@suse.cz>
14389 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
14390 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
14391 Remove N_SO_PASCAL.
14392 * dwarf2out.c (lower_bound_default): Do not handle
14394 (gen_compile_unit_die): Likewise.
14395 * gcc.c: Remove default extension binding for GNU Pascal.
14396 * stmt.c: Remove Pascal language from a comment.
14397 * xcoffout.c: Likewise.
14399 2017-07-13 David Malcolm <dmalcolm@redhat.com>
14402 * diagnostic-show-locus.c (fixit_cmp): New function.
14403 (layout::layout): Sort m_fixit_hints.
14404 (column_range::column_range): Assert that the values are valid.
14405 (struct char_span): New struct.
14406 (correction::overwrite): New method.
14407 (struct source_line): New struct.
14408 (line_corrections::add_hint): Add assertions. Reimplement memcpy
14409 calls in terms of classes source_line and char_span, and
14410 correction::overwrite.
14411 (selftest::test_overlapped_fixit_printing_2): New function.
14412 (selftest::diagnostic_show_locus_c_tests): Call it.
14414 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
14416 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
14417 early if there is no lhs.
14419 2017-07-13 Martin Liska <mliska@suse.cz>
14421 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
14422 (gen_reference_type_die): Likewise.
14423 * stor-layout.c: Remove Pascal-related comment.
14425 2017-07-13 Martin Liska <mliska@suse.cz>
14427 * opts.c (finish_options): Add quotes to error messages.
14428 (parse_sanitizer_options): Likewise.
14430 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
14432 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
14434 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
14436 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
14438 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
14440 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
14442 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
14444 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
14447 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
14448 provides the hardware capability bits, define the macro
14449 __BUILTIN_CPU_SUPPORTS__.
14450 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
14451 if GLIBC does not provide the hardware capability bits. Add a
14452 gcc_unreachable call if the built-in cpu function is neither
14453 __builtin_cpu_is nor __builtin_cpu_supports.
14454 (rs6000_get_function_versions_dispatcher): Change the warning
14455 that an old GLIBC is used which does not export the capability
14456 bits to be an error.
14457 * doc/extend.texi (target_clones attribute): Document the
14458 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
14459 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
14460 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
14461 the macros defined by GCC if the newer GLIBC is available.
14463 2017-07-12 Jeff Law <law@redhat.com>
14465 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
14466 remaining includes slightly.
14467 * config/riscv/riscv-builtins.c: Include profile-count.h.
14469 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
14472 * config/avr/avr.c (avr_set_current_function): In diagnostic
14473 messages: Quote keywords and (parts of) identifiers.
14474 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
14477 2017-07-12 Carl Love <cel@us.ibm.com>
14479 * config/rs6000/rs6000-c.c: Add support for built-in functions
14480 vector bool char vec_revb (vector bool char);
14481 vector bool short vec_revb (vector short char);
14482 vector bool int vec_revb (vector bool int);
14483 vector bool long long vec_revb (vector bool long long);
14484 * doc/extend.texi: Update the built-in documentation file for the
14485 new built-in functions.
14487 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14489 * config/s390/s390.md: Remove movcc splitter.
14491 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14493 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
14494 load/store on condition.
14496 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
14499 * config/avr/avr.c (avr_encode_section_info)
14500 [progmem && !TREE_READONLY]: Error if progmem object needs
14503 2017-07-11 Michael Collison <michael.collison@arm.com>
14505 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
14508 2017-07-11 Carl Love <cel@us.ibm.com>
14510 * config/rs6000/rs6000-c.c: Add support for builtins
14511 vector unsigned int vec_parity_lsbb (vector signed int);
14512 vector unsigned int vec_parity_lsbb (vector unsigned int);
14513 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
14514 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
14515 vector unsigned long long vec_parity_lsbb (vector signed long long);
14516 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
14517 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
14518 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
14519 * doc/extend.texi: Update the built-in documentation file for the
14520 new built-in functions.
14522 2017-07-11 David Malcolm <dmalcolm@redhat.com>
14524 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
14525 (layout::m_primary_loc): New field.
14526 (layout::layout): Initialize new field. Move location filtering
14527 logic from here to...
14528 (layout::maybe_add_location_range): ...this new method. Add
14529 support for filtering to just the lines already specified by other
14531 (layout::will_show_line_p): New method.
14532 (gcc_rich_location::add_location_if_nearby): New method.
14533 (selftest::test_add_location_if_nearby): New test function.
14534 (selftest::diagnostic_show_locus_c_tests): Call it.
14535 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
14538 2017-07-11 Tom de Vries <tom@codesourcery.com>
14540 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
14541 (bb_first_real_insn): New function.
14542 (nvptx_single): Add extra initialization of broadcasted condition
14545 2017-07-11 Nathan Sidwell <nathan@acm.org>
14547 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
14549 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
14551 * doc/extend.texi (AVR Function Attributes): Remove weblink to
14552 Binutils doc as TEXI will mess them up.
14553 * doc/invoke.texi (AVR Options): Same here.
14555 2017-07-11 Daniel Cederman <cederman@gaisler.com>
14557 * config/sparc/sparc.opt (mfix-ut700): New option.
14558 (mfix-gr712rc): Likewise.
14559 (sparc_fix_b2bst): New variable.
14560 * doc/invoke.texi (SPARC options): Document them.
14561 (ARM options): Fix warnings.
14562 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
14563 instructions to prevent sequences that can trigger the store-store
14564 errata for certain LEON3FT processors.
14565 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
14566 (sparc_option_override): Set sparc_fix_b2bst appropriately.
14567 * config/sparc/sparc.md (fix_b2bst): New attribute.
14568 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
14570 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
14573 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
14575 (*rsqrtsf2_sse): Ditto.
14577 (div<mode>3): Macroize insn from divdf3 and divsf3
14578 using MODEF mode iterator.
14580 2017-07-10 Martin Sebor <msebor@redhat.com>
14582 PR tree-optimization/80397
14583 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
14584 instead of testing for equality to INTEGER_TYPE.
14586 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
14588 * config.gcc: Remove uclibc from arc target spec.
14590 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
14592 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
14594 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
14597 * lto-wrapper.c (remove_option): New function.
14598 (merge_and_complain): Merge PIC/PIE options more realistically.
14600 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
14602 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
14606 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
14607 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
14608 * config.in: Regenerate.
14609 * configure: Regenerate.
14610 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
14611 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
14612 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
14613 (TARGET_GASISR_PROLOGUES): ...target mask.
14614 * common/config/avr/avr-common.c
14615 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
14616 Set -mgas-isr-prologues.
14617 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
14618 INSERT_PASS_BEFORE for it.
14619 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
14620 * config/avr/avr.c (avr_option_override)
14621 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
14622 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
14623 (avr_attribute_table) <no_gccisr>: Add new function attribute.
14624 (avr_set_current_function) <is_no_gccisr>: Init machine field.
14625 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
14627 (make_avr_pass_pre_proep): New function.
14628 (emit_push_sfr) <treg>: Add argument to function and use it
14629 instead of TMP_REG.
14630 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
14631 and set machine->gasisr.yes.
14632 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
14633 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
14634 __gcc_isr.n_pushed to .L__stack_usage.
14635 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
14636 (avr_asm_final_postscan_insn): ...this new static function.
14637 * config/avr/avr.h (machine_function)
14638 <is_no_gccisr, use_L__stack_usage>: New fields.
14639 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
14640 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
14641 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
14642 (gasisr, *gasisr): New expander and insn.
14643 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
14644 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
14645 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
14647 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
14649 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
14652 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
14654 Move jump-tables out of .text again.
14657 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
14658 (ASM_OUTPUT_ADDR_VEC): New function.
14659 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
14660 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
14661 INSN_ADDRESSes as asm comment.
14662 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
14663 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
14664 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
14665 * config/avr/avr.md (*tablejump): Adjust comment.
14666 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
14667 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
14669 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
14670 (avr_output_addr_vec): New proto.
14671 (avr_log_t) <insn_addresses>: New field.
14673 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
14676 * config/i386/i386.c (ix86_function_arg_advance): Set
14677 outgoing_args_on_stack to true if there are outgoing arguments
14679 (ix86_function_arg): Likewise.
14680 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
14681 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
14682 * config/i386/i386.h (machine_function): Add
14683 outgoing_args_on_stack.
14685 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
14687 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
14688 supporting pthreds.
14689 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
14691 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
14693 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
14694 (REAL_H): Remove $(MACHMODE_H).
14695 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
14697 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
14698 $(MACHMODE_H) and double-int.h.
14699 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
14701 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
14704 2017-07-07 Andrew Pinski <apinski@cavium.com>
14706 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
14707 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
14709 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
14711 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
14712 Add warning if GCC was not configured to link against a GLIBC that
14713 exports the hardware capability bits.
14714 (make_resolver_func): Make resolver function private and not a
14715 COMDAT function. Create the name with clone_function_name instead
14716 of make_unique_name.
14719 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
14720 correct operand in doing the split.
14722 2017-07-07 Carl Love <cel@us.ibm.com>
14724 * config/rs6000/rs6000-c: Add support for built-in function
14725 vector unsigned short vec_pack_to_short_fp32 (vector float,
14727 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
14728 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
14729 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
14730 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
14731 (convert_4f32_8i16): Add define_expand.
14732 * doc/extend.texi: Update the built-in documentation file for the
14733 new built-in function.
14735 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14737 * config/sparc/m8.md: New file.
14738 * config/sparc/sparc.md: Include m8.md.
14740 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14742 * config/sparc/sparc.opt: New option -mvis4b.
14743 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
14744 (sparc_option_override): Handle VIS4B.
14745 (enum sparc_builtins): Define
14746 SPARC_BUILTIN_DICTUNPACK{8,16,32},
14747 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
14748 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
14749 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
14750 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
14751 (check_constant_argument): New function.
14752 (sparc_vis_init_builtins): Define builtins
14753 __builtin_vis_dictunpack{8,16,32},
14754 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
14755 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
14756 __builtin_vis_fpcmpde{8,16,32}shl and
14757 __builtin_vis_fpcmpur{8,16,32}shl.
14758 (sparc_expand_builtin): Check that the constant operands to
14759 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
14760 constant and in range.
14761 * config/sparc/sparc-c.c (sparc_target_macros): Handle
14763 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
14764 (SPARC_IMM5_P): Likewise.
14765 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
14766 (enabled): Handle vis4b.
14767 (UNSPEC_DICTUNPACK): New unspec.
14768 (UNSPEC_FPCMPSHL): Likewise.
14769 (UNSPEC_FPUCMPSHL): Likewise.
14770 (UNSPEC_FPCMPDESHL): Likewise.
14771 (UNSPEC_FPCMPURSHL): Likewise.
14772 (cpu_feature): New CPU feature `vis4b'.
14773 (dictunpack{8,16,32}): New insns.
14774 (FPCSMODE): New mode iterator.
14775 (fpcscond): New code iterator.
14776 (fpcsucond): Likewise.
14777 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
14778 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
14779 (fpcmpde{8,16,32}{si,di}shl): Likewise.
14780 (fpcmpur{8,16,32}{si,di}shl): Likewise.
14781 * config/sparc/constraints.md: Define constraints `q' for unsigned
14782 2-bit integer constants and `t' for unsigned 5-bit integer
14784 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
14786 (imm5_operand_dictunpack16): Likewise.
14787 (imm5_operand_dictunpack32): Likewise.
14788 (imm2_operand): Likewise.
14789 * doc/invoke.texi (SPARC Options): Document -mvis4b.
14790 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
14791 ditunpack* and fpcmp*shl builtins.
14793 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14795 * config.gcc: Handle m8 in --with-{cpu,tune} options.
14796 * config.in: Add HAVE_AS_SPARC6 define.
14797 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
14799 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
14801 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
14802 (CPP_CPU_SPEC): Handle m8.
14803 (ASM_CPU_SPEC): Likewise.
14804 * config/sparc/sparc-opts.h (enum processor_type): Add
14806 * config/sparc/sparc.c (m8_costs): New struct.
14807 (sparc_option_override): Handle TARGET_CPU_m8.
14808 (sparc32_initialize_trampoline): Likewise.
14809 (sparc64_initialize_trampoline): Likewise.
14810 (sparc_issue_rate): Likewise.
14811 (sparc_register_move_cost): Likewise.
14812 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
14813 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
14814 (ASM_CPU64_DEFAULT_SPEC): Likewise.
14815 (CPP_CPU_SPEC): Handle M8.
14816 (ASM_CPU_SPEC): Likewise.
14817 (AS_M8_FLAG): Define.
14818 * config/sparc/sparc.md: Add m8 to the cpu attribute.
14819 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
14820 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
14822 * configure: Regenerate.
14823 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
14826 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14828 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
14830 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
14831 ("*movdi_insn_sp32"): Do not set v3pipe.
14832 ("*movsi_insn"): Likewise.
14833 ("*movdi_insn_sp64"): Likewise.
14834 ("*movsf_insn"): Likewise.
14835 ("*movdf_insn_sp32"): Likewise.
14836 ("*movdf_insn_sp64"): Likewise.
14837 ("*zero_extendsidi2_insn_sp64"): Likewise.
14838 ("*sign_extendsidi2_insn"): Likewise.
14839 ("*mov<VM32:mode>_insn"): Likewise.
14840 ("*mov<VM64:mode>_insn_sp64"): Likewise.
14841 ("*mov<VM64:mode>_insn_sp32"): Likewise.
14842 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
14843 ("<vlop:code><VL:mode>3"): Likewise.
14844 ("*not_<vlop:code><VL:mode>3"): Likewise.
14845 ("*nand<VL:mode>_vis"): Likewise.
14846 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
14847 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
14848 ("one_cmpl<VL:mode>2"): Likewise.
14849 ("faligndata<VM64:mode>_vis"): Likewise.
14850 ("alignaddrsi_vis"): Likewise.
14851 ("alignaddrdi_vis"): Likweise.
14852 ("alignaddrlsi_vis"): Likewise.
14853 ("alignaddrldi_vis"): Likewise.
14854 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
14855 ("bmaskdi_vis"): Likewise.
14856 ("bmasksi_vis"): Likewise.
14857 ("bshuffle<VM64:mode>_vis"): Likewise.
14858 ("cmask8<P:mode>_vis"): Likewise.
14859 ("cmask16<P:mode>_vis"): Likewise.
14860 ("cmask32<P:mode>_vis"): Likewise.
14861 ("pdistn<P:mode>_vis"): Likewise.
14862 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
14864 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14866 * config/sparc/sparc.md ("subtype"): New insn attribute.
14867 ("*wrgsr_sp64"): Set insn subtype.
14868 ("*rdgsr_sp64"): Likewise.
14869 ("alignaddrsi_vis"): Likewise.
14870 ("alignaddrdi_vis"): Likewise.
14871 ("alignaddrlsi_vis"): Likewise.
14872 ("alignaddrldi_vis"): Likewise.
14873 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
14874 ("fexpand_vis"): Likewise.
14875 ("fpmerge_vis"): Likewise.
14876 ("faligndata<VM64:mode>_vis"): Likewise.
14877 ("bshuffle<VM64:mode>_vis"): Likewise.
14878 ("cmask8<P:mode>_vis"): Likewise.
14879 ("cmask16<P:mode>_vis"): Likewise.
14880 ("cmask32<P:mode>_vis"): Likewise.
14881 ("fchksm16_vis"): Likewise.
14882 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
14883 ("fmean16_vis"): Likewise.
14884 ("fp<plusminus_insn>64_vis"): Likewise.
14885 ("<plusminus_insn>v8qi3"): Likewise.
14886 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
14887 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
14888 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
14889 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
14890 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
14891 ("*movqi_insn"): Likewise.
14892 ("*movhi_insn"): Likewise.
14893 ("*movsi_insn"): Likewise.
14894 ("movsi_pic_gotdata_op"): Likewise.
14895 ("*movdi_insn_sp32"): Likewise.
14896 ("*movdi_insn_sp64"): Likewise.
14897 ("movdi_pic_gotdata_op"): Likewise.
14898 ("*movsf_insn"): Likewise.
14899 ("*movdf_insn_sp32"): Likewise.
14900 ("*movdf_insn_sp64"): Likewise.
14901 ("*zero_extendhisi2_insn"): Likewise.
14902 ("*zero_extendqihi2_insn"): Likewise.
14903 ("*zero_extendqisi2_insn"): Likewise.
14904 ("*zero_extendqidi2_insn"): Likewise.
14905 ("*zero_extendhidi2_insn"): Likewise.
14906 ("*zero_extendsidi2_insn_sp64"): Likewise.
14907 ("ldfsr"): Likewise.
14908 ("prefetch_64"): Likewise.
14909 ("prefetch_32"): Likewise.
14910 ("tie_ld32"): Likewise.
14911 ("tie_ld64"): Likewise.
14912 ("*tldo_ldub_sp32"): Likewise.
14913 ("*tldo_ldub1_sp32"): Likewise.
14914 ("*tldo_ldub2_sp32"): Likewise.
14915 ("*tldo_ldub_sp64"): Likewise.
14916 ("*tldo_ldub1_sp64"): Likewise.
14917 ("*tldo_ldub2_sp64"): Likewise.
14918 ("*tldo_ldub3_sp64"): Likewise.
14919 ("*tldo_lduh_sp32"): Likewise.
14920 ("*tldo_lduh1_sp32"): Likewise.
14921 ("*tldo_lduh_sp64"): Likewise.
14922 ("*tldo_lduh1_sp64"): Likewise.
14923 ("*tldo_lduh2_sp64"): Likewise.
14924 ("*tldo_lduw_sp32"): Likewise.
14925 ("*tldo_lduw_sp64"): Likewise.
14926 ("*tldo_lduw1_sp64"): Likewise.
14927 ("*tldo_ldx_sp64"): Likewise.
14928 ("*mov<VM32:mode>_insn"): Likewise.
14929 ("*mov<VM64:mode>_insn_sp64"): Likewise.
14930 ("*mov<VM64:mode>_insn_sp32"): Likewise.
14932 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14934 * config/sparc/sparc.md ("type"): New insn type viscmp.
14935 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
14937 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
14938 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
14939 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
14940 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
14942 ("n7_vis_logical_11cycle"): Likewise.
14943 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
14944 * config/sparc/niagara2.md ("niag3_vis": Likewise.
14945 * config/sparc/niagara.md ("niag_vis"): Likewise.
14946 * config/sparc/ultra3.md ("us3_fga"): Likewise.
14947 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
14949 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14951 * config/sparc/sparc.md: New instruction type `bmask'.
14952 (bmaskdi_vis): Use the `bmask' type.
14953 (bmasksi_vis): Likewise.
14954 * config/sparc/ultra3.md (us3_array): Likewise.
14955 * config/sparc/niagara7.md (n7_array): Likewise.
14956 * config/sparc/niagara4.md (n4_array): Likewise.
14957 * config/sparc/niagara2.md (niag2_vis): Likewise.
14958 (niag3_vis): Likewise.
14959 * config/sparc/niagara.md (niag_vis): Likewise.
14961 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
14963 * ipa-comdats.c: Remove optimize check from gate.
14964 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
14965 for functions not optimized.
14966 (ipa_fn_summary_read): Skip optimize check.
14967 (ipa_fn_summary_write): Likewise.
14968 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
14970 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
14972 (can_inline_edge_p): Check flag_pcc_struct_return for match.
14973 (check_callers): Give up on caller which is not optimized.
14974 (inline_small_functions): Likewise.
14975 (ipa_inline): Do not give up when not optimizing.
14976 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
14977 away unoptimizes cdtors.
14978 (whole_program_function_and_variable_visibility): Do
14979 ipa_discover_readonly_nonaddressable_vars in LTO mode.
14980 * ipa.c (process_references): Do not check optimize.
14981 (symbol_table::remove_unreachable_nodes): Update optimize check.
14982 (set_writeonly_bit): Update optimize check.
14983 (pass_ipa_cdtor_merge::gate): Do not check optimize.
14984 (pass_ipa_single_use::gate): Remove.
14986 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
14988 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
14989 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
14990 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
14991 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
14992 permute_load, permute_store, adjust_extract, adjust_splat,
14993 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
14994 replace_swap_with_copy, dump_swap_insn_table,
14995 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
14996 recombine_lvx_pattern, recombine_stvx_pattern,
14997 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
14998 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
14999 to file rs6000-p8swap.c.
15000 * config/rs6000/rs6000-p8swap.c: New file.
15001 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
15002 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
15003 and rs6000*-*-* targets.
15005 2017-07-06 David Malcolm <dmalcolm@redhat.com>
15007 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
15009 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
15011 * lto-wrapper.c (merge_and_complain): Do not merge
15012 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
15013 fsigned_zeros, ftrapping_math, fwrapv.
15014 (append_compiler_options): Do not track these options.
15015 (append_linker_options): Likewie
15017 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
15019 * cgraphunit.c (cgraph_node::finalize_function): When
15020 !flag_toplevel_reorde set no_reorder flag.
15021 (varpool_node::finalize_decl): Likewise.
15022 (symbol_table::compile): Drop no toplevel reorder path.
15024 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
15026 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
15027 edges; zero probability is not better than uninitialized.
15029 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
15031 * asan.h (asan_sanitize_allocas_p): Declare.
15032 * asan.c (asan_sanitize_allocas_p): New function.
15033 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
15034 (handle_builtin_alloca): Likewise.
15035 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
15036 if !asan_sanitize_allocas_p.
15037 * params.def (asan-instrument-allocas): Add new option.
15038 * params.h (ASAN_PROTECT_ALLOCAS): Define.
15039 * opts.c (common_handle_option): Disable allocas sanitization for
15042 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
15044 * asan.c: Include gimple-fold.h.
15045 (get_last_alloca_addr): New function.
15046 (handle_builtin_stackrestore): Likewise.
15047 (handle_builtin_alloca): Likewise.
15048 (asan_emit_allocas_unpoison): Likewise.
15049 (get_mem_refs_of_builtin_call): Add new parameter, remove const
15050 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
15051 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
15052 (instrument_builtin_call): Pass gimple iterator to
15053 get_mem_refs_of_builtin_call.
15054 (last_alloca_addr): New global.
15055 * asan.h (asan_emit_allocas_unpoison): Declare.
15056 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
15057 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
15058 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
15059 if function calls alloca.
15060 * gimple-fold.c (replace_call_with_value): Remove static keyword.
15061 * gimple-fold.h (replace_call_with_value): Declare.
15062 * internal-fn.c: Include asan.h.
15063 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
15064 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
15066 2017-07-06 David Malcolm <dmalcolm@redhat.com>
15068 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
15069 (C_SELFTEST_FLAGS): New.
15070 (CPP_SELFTEST_FLAGS): New.
15071 (SELFTEST_DEPS): New, from deps of s-selftest.
15072 (C_SELFTEST_DEPS): New, from deps of s-selftest.
15073 (CPP_SELFTEST_DEPS): New.
15074 (selftest): Add dependency on s-selftest-c++.
15075 (s-selftest): Rename to...
15076 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
15077 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
15078 than SELFTEST_FLAGS.
15079 (selftest-gdb): Rename to...
15080 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
15082 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
15083 (selftest-valgrind): Rename to...
15084 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
15086 (selftest-valgrind): Reintroduce as an alias for
15087 selftest-c-valgrind.
15088 (s-selftest-c++): New.
15089 (selftest-c++-gdb): New.
15090 (selftest-c++-valgrind): New.
15092 2017-07-06 Olivier Hainque <hainque@adacore.com>
15094 * gcc.c (process_command): When deciding if undefined variables
15095 should be ignored when processing specs, accept "gcc -v" as well.
15097 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
15099 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
15100 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
15102 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
15104 * config/arm/arm-cpus.in (armv8-r): Add new entry.
15105 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
15106 * config/arm/arm-tables.opt: Regenerate.
15107 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
15109 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
15111 2017-07-06 Carl Love <cel@us.ibm.com>
15113 * ChangeLog: Clean up from mid air collision
15115 2017-07-06 Carl Love <cel@us.ibm.com>
15117 * config/rs6000/rs6000-c.c: Add support for built-in functions
15118 vector signed int vec_subc (vector signed int, vector signed int);
15119 vector signed __int128 vec_subc (vector signed __int128,
15120 vector signed __int128);
15121 vector unsigned __int128 vec_subc (vector unsigned __int128,
15122 vector unsigned __int128);
15123 vector signed int vec_sube (vector signed int, vector signed int,
15124 vector signed int);
15125 vector unsigned int vec_sube (vector unsigned int,
15126 vector unsigned int,
15127 vector unsigned int);
15128 vector signed __int128 vec_sube (vector signed __int128,
15129 vector signed __int128,
15130 vector signed__int128);
15131 vector unsigned __int128 vec_sube (vector unsigned __int128,
15132 vector unsigned __int128,
15133 vector unsigned __int128);
15134 vector signed int vec_subec (vector signed int, vector signed int,
15135 vector signed int);
15136 vector unsigned int vec_subec (vector unsigned int,
15137 vector unsigned int,
15138 vector unsigned int);
15139 vector signed __int128 vec_subec (vector signed __int128,
15140 vector signed __int128,
15141 vector signed__int128);
15142 vector unsigned __int128 vec_subec (vector unsigned __int128,
15143 vector unsigned __int128,
15144 vector unsigned __int128);
15145 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
15146 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
15147 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
15148 BU_ALTIVEC_OVERLOAD_X definitions.
15149 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
15150 * doc/extend.texi: Update the built-in documentation file for the new
15151 built-in functions.
15153 2017-07-06 David Malcolm <dmalcolm@redhat.com>
15156 * diagnostic-show-locus.c (layout::layout): Use start and finish
15157 spelling location for the start and finish of each range.
15158 * genmatch.c (linemap_client_expand_location_to_spelling_point):
15159 Add unused aspect param.
15160 * input.c (expand_location_1): Add "aspect" param, and use it
15161 to access the correct part of the location.
15162 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
15164 (expand_location_to_spelling_point): Likewise.
15165 (linemap_client_expand_location_to_spelling_point): Add "aspect"
15166 param, and pass it to expand_location_1.
15168 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
15170 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
15171 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
15172 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
15173 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
15174 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
15175 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
15176 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
15177 _mm_maskz_getmant_ss): New intrinsics.
15178 (__builtin_ia32_getexpss128_mask): Changed to ...
15179 __builtin_ia32_getexpss128_round ... this.
15180 (__builtin_ia32_getexpsd128_mask): Changed to ...
15181 __builtin_ia32_getexpsd128_round ... this.
15182 * config/i386/i386-builtin-types.def
15183 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
15184 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
15185 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
15186 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
15187 __builtin_ia32_getmantss_mask_round): New builtins.
15188 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
15189 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
15190 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
15191 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
15192 * config/i386/sse.md
15193 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
15194 avx512f_sgetexp<mode><mask_scalar_name>
15195 <round_saeonly_scalar_name> ... this.
15196 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
15197 %0, %1, %2<round_saeonly_op3>}): Changed to ...
15198 vgetexp<ssescalarmodesuffix>
15199 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
15200 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
15201 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
15202 avx512f_vgetmant<mode><mask_scalar_name>
15203 <round_saeonly_scalar_name> ... this.
15204 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
15205 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
15206 vgetmant<ssescalarmodesuffix>
15207 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
15208 %0<mask_scalar_operand4>, %1, %2
15209 <round_saeonly_scalar_mask_op4>, %3} ... this.
15210 * config/i386/subst.md (mask_scalar_operand4,
15211 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
15212 round_saeonly_scalar_nimm_predicate): New subst attributes.
15214 2017-07-06 Julia Koval <julia.koval@intel.com>
15216 * config/i386/i386.c (ix86_erase_embedded_rounding):
15217 Remove code for old rounding pattern.
15219 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
15221 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
15223 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
15225 * doc/sourcebuild.texi (Test Directives, Variants of
15226 dg-require-support): Add documentation for dg-require-stack-check.
15228 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
15230 * config/i386/subst.md (mask_scalar, round_scalar,
15231 round_saeonly_scalar): New meta-templates.
15232 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
15233 round_scalar_mask_operand3, round_scalar_mask_op3,
15234 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
15235 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
15236 round_saeonly_scalar_constraint,
15237 round_saeonly_scalar_prefix): New subst attribute.
15238 * config/i386/sse.md
15239 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
15240 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
15241 <round_scalar_name> ... this.
15242 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
15243 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
15244 <round_scalar_name> ... this.
15245 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
15246 <sse>_vm<code><mode>3<mask_scalar_name>
15247 <round_saeonly_scalar_name> ... this.
15248 (v<plusminus_mnemonic><ssescalarmodesuffix>
15249 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
15250 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
15251 v<plusminus_mnemonic><ssescalarmodesuffix>
15252 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
15253 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
15254 (v<multdiv_mnemonic><ssescalarmodesuffix>
15255 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
15256 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
15257 v<multdiv_mnemonic><ssescalarmodesuffix>
15258 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
15259 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
15260 (v<maxmin_float><ssescalarmodesuffix>
15261 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
15262 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
15263 v<maxmin_float><ssescalarmodesuffix>
15264 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
15265 %0<mask_scalar_operand3>, %1, %<iptr>2
15266 <round_saeonly_scalar_mask_op3>} ... this.
15268 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
15270 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
15271 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
15273 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
15274 Alan Hayward <alan.hayward@arm.com>
15275 David Sherwood <david.sherwood@arm.com>
15277 * combine.c (simplify_if_then_else): Remove "enum" before
15279 * compare-elim.c (can_eliminate_compare): Likewise.
15280 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
15282 (aarch64_lookup_simd_builtin_type): Likewise.
15283 (aarch64_simd_builtin_type): Likewise.
15284 (aarch64_init_simd_builtin_types): Likewise.
15285 (aarch64_simd_expand_args): Likewise.
15286 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
15288 (aarch64_reverse_mask): Likewise.
15289 (aarch64_simd_emit_reg_reg_move): Likewise.
15290 (aarch64_gen_adjusted_ldpstp): Likewise.
15291 (aarch64_ccmp_mode_to_code): Likewise.
15292 (aarch64_operands_ok_for_ldpstp): Likewise.
15293 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15294 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
15296 (aarch64_min_divisions_for_recip_mul): Likewise.
15297 (aarch64_reassociation_width): Likewise.
15298 (aarch64_get_condition_code_1): Likewise.
15299 (aarch64_simd_emit_reg_reg_move): Likewise.
15300 (aarch64_simd_attr_length_rglist): Likewise.
15301 (aarch64_reverse_mask): Likewise.
15302 (aarch64_operands_ok_for_ldpstp): Likewise.
15303 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15304 (aarch64_gen_adjusted_ldpstp): Likewise.
15305 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
15307 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
15308 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
15309 (arm_lookup_simd_builtin_type): Likewise.
15310 (arm_simd_builtin_type): Likewise.
15311 (arm_init_simd_builtin_types): Likewise.
15312 (arm_expand_builtin_args): Likewise.
15313 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
15314 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
15315 (ft32_setup_incoming_varargs): Likewise.
15316 (ft32_function_arg): Likewise.
15317 (ft32_function_arg_advance): Likewise.
15318 (ft32_pass_by_reference): Likewise.
15319 (ft32_arg_partial_bytes): Likewise.
15320 (ft32_valid_pointer_mode): Likewise.
15321 (ft32_addr_space_pointer_mode): Likewise.
15322 (ft32_addr_space_legitimate_address_p): Likewise.
15323 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
15325 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
15326 (ix86_emit_outlined_ms2sysv_restore): Likewise.
15327 (iamcu_alignment): Likewise.
15328 (canonicalize_vector_int_perm): Likewise.
15329 (ix86_noce_conversion_profitable_p): Likewise.
15330 (ix86_mpx_bound_mode): Likewise.
15331 (ix86_operands_ok_for_move_multiple): Likewise.
15332 * config/microblaze/microblaze-protos.h
15333 (microblaze_expand_conditional_branch_reg): Likewise.
15334 * config/microblaze/microblaze.c
15335 (microblaze_expand_conditional_branch_reg): Likewise.
15336 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
15338 (rs6000_reassociation_width): Likewise.
15339 (rs6000_invalid_binary_op): Likewise.
15340 (fusion_p9_p): Likewise.
15341 (emit_fusion_p9_load): Likewise.
15342 (emit_fusion_p9_store): Likewise.
15343 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
15345 (riscv_hard_regno_mode_ok_p): Likewise.
15346 (riscv_address_insns): Likewise.
15347 (riscv_split_symbol): Likewise.
15348 (riscv_legitimize_move): Likewise.
15349 (riscv_function_value): Likewise.
15350 (riscv_hard_regno_nregs): Likewise.
15351 (riscv_expand_builtin): Likewise.
15352 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
15353 (riscv_build_integer): Likewise.
15354 (riscv_split_integer): Likewise.
15355 (riscv_legitimate_constant_p): Likewise.
15356 (riscv_cannot_force_const_mem): Likewise.
15357 (riscv_regno_mode_ok_for_base_p): Likewise.
15358 (riscv_valid_base_register_p): Likewise.
15359 (riscv_valid_offset_p): Likewise.
15360 (riscv_valid_lo_sum_p): Likewise.
15361 (riscv_classify_address): Likewise.
15362 (riscv_legitimate_address_p): Likewise.
15363 (riscv_address_insns): Likewise.
15364 (riscv_load_store_insns): Likewise.
15365 (riscv_force_binary): Likewise.
15366 (riscv_split_symbol): Likewise.
15367 (riscv_force_address): Likewise.
15368 (riscv_legitimize_address): Likewise.
15369 (riscv_move_integer): Likewise.
15370 (riscv_legitimize_const_move): Likewise.
15371 (riscv_legitimize_move): Likewise.
15372 (riscv_address_cost): Likewise.
15373 (riscv_subword): Likewise.
15374 (riscv_output_move): Likewise.
15375 (riscv_canonicalize_int_order_test): Likewise.
15376 (riscv_emit_int_order_test): Likewise.
15377 (riscv_function_arg_boundary): Likewise.
15378 (riscv_pass_mode_in_fpr_p): Likewise.
15379 (riscv_pass_fpr_single): Likewise.
15380 (riscv_pass_fpr_pair): Likewise.
15381 (riscv_get_arg_info): Likewise.
15382 (riscv_function_arg): Likewise.
15383 (riscv_function_arg_advance): Likewise.
15384 (riscv_arg_partial_bytes): Likewise.
15385 (riscv_function_value): Likewise.
15386 (riscv_pass_by_reference): Likewise.
15387 (riscv_setup_incoming_varargs): Likewise.
15388 (riscv_print_operand): Likewise.
15389 (riscv_elf_select_rtx_section): Likewise.
15390 (riscv_save_restore_reg): Likewise.
15391 (riscv_for_each_saved_reg): Likewise.
15392 (riscv_register_move_cost): Likewise.
15393 (riscv_hard_regno_mode_ok_p): Likewise.
15394 (riscv_hard_regno_nregs): Likewise.
15395 (riscv_class_max_nregs): Likewise.
15396 (riscv_memory_move_cost): Likewise.
15397 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
15398 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
15399 (rl78_addr_space_address_mode): Likewise.
15400 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15402 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
15403 (rs6000_reassociation_width): Likewise.
15404 (rs6000_invalid_binary_op): Likewise.
15405 (fusion_p9_p): Likewise.
15406 (emit_fusion_p9_load): Likewise.
15407 (emit_fusion_p9_store): Likewise.
15408 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
15409 (ok_for_simple_move_operands): Likewise.
15410 (ok_for_simple_move_strict_operands): Likewise.
15411 (ok_for_simple_arith_logic_operands): Likewise.
15412 (visium_legitimize_reload_address): Likewise.
15413 (visium_select_cc_mode): Likewise.
15414 (output_cbranch): Likewise.
15415 (visium_split_double_move): Likewise.
15416 (visium_expand_copysign): Likewise.
15417 (visium_expand_int_cstore): Likewise.
15418 (visium_expand_fp_cstore): Likewise.
15419 * config/visium/visium.c (visium_pass_by_reference): Likewise.
15420 (visium_function_arg): Likewise.
15421 (visium_function_arg_advance): Likewise.
15422 (visium_libcall_value): Likewise.
15423 (visium_setup_incoming_varargs): Likewise.
15424 (visium_legitimate_constant_p): Likewise.
15425 (visium_legitimate_address_p): Likewise.
15426 (visium_legitimize_address): Likewise.
15427 (visium_secondary_reload): Likewise.
15428 (visium_register_move_cost): Likewise.
15429 (visium_memory_move_cost): Likewise.
15430 (prepare_move_operands): Likewise.
15431 (ok_for_simple_move_operands): Likewise.
15432 (ok_for_simple_move_strict_operands): Likewise.
15433 (ok_for_simple_arith_logic_operands): Likewise.
15434 (visium_function_value_1): Likewise.
15435 (rtx_ok_for_offset_p): Likewise.
15436 (visium_legitimize_reload_address): Likewise.
15437 (visium_split_double_move): Likewise.
15438 (visium_expand_copysign): Likewise.
15439 (visium_expand_int_cstore): Likewise.
15440 (visium_expand_fp_cstore): Likewise.
15441 (visium_split_cstore): Likewise.
15442 (visium_select_cc_mode): Likewise.
15443 (visium_split_cbranch): Likewise.
15444 (output_cbranch): Likewise.
15445 (visium_print_operand_address): Likewise.
15446 * expmed.c (flip_storage_order): Likewise.
15447 * expmed.h (emit_cstore): Likewise.
15448 (flip_storage_order): Likewise.
15449 * genrecog.c (validate_pattern): Likewise.
15450 * hsa-gen.c (gen_hsa_addr): Likewise.
15451 * internal-fn.c (expand_arith_overflow): Likewise.
15452 * ira-color.c (allocno_copy_cost_saving): Likewise.
15453 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15454 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
15455 (process_invariant_for_inheritance): Likewise.
15456 * lra-eliminations.c (move_plus_up): Likewise.
15457 * omp-low.c (lower_oacc_reductions): Likewise.
15458 * simplify-rtx.c (simplify_subreg): Likewise.
15459 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
15460 (TARGET_CHKP_BOUND_MODE): Likewise..
15461 * targhooks.c (default_chkp_bound_mode): Likewise.
15462 (default_setup_incoming_vararg_bounds): Likewise.
15463 * targhooks.h (default_chkp_bound_mode): Likewise.
15464 (default_setup_incoming_vararg_bounds): Likewise.
15465 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
15466 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
15467 (have_whole_vector_shift): Likewise.
15468 * tree-vect-stmts.c (vectorizable_load): Likewise.
15469 * doc/tm.texi: Regenerate.
15471 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
15473 Graceful degrade if Binutils PR21472 is not available.
15476 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
15477 .rodata in flash test fails.
15478 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
15479 * confgure: Regenerate.
15480 * config.in: Regenerate.
15481 * config/avr/avr.c (avr_asm_named_section)
15482 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
15483 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
15484 (avr_asm_init_sections): Same.
15486 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15488 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
15489 (fma<VH:mode>4_intrinsic): Likewise.
15490 (*fmsub<VCVTF:mode>4): Likewise.
15491 (*fmsub<VH:mode>4_intrinsic): Likewise.
15493 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
15496 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
15497 Don't depend on "optimize > 0".
15498 (out_movhi_r_mr, out_movqi_mr_r): Same.
15499 (out_movhi_mr_r, out_movqi_r_mr): Same.
15500 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
15501 io_address_operand on "optimize > 0".
15503 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15505 * tree-loop-distribution.c: Add general explanantion on the pass.
15506 (generate_loops_for_partition): Mark distributed loop.
15507 (pg_add_dependence_edges): New parameter. Handle alias data
15508 dependence specially and record it in the parameter if asked.
15509 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
15510 (init_partition_graph_vertices, add_partition_graph_edge): New.
15511 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
15512 (free_partition_graph_vdata, build_partition_graph): New.
15513 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
15514 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
15515 (data_ref_segment_size, latch_dominated_by_data_ref): New.
15516 (compute_alias_check_pairs, version_loop_by_alias_check): New.
15517 (version_for_distribution_p, finalize_partitions): New.
15518 (distribute_loop): Handle alias data dependence specially. Factor
15519 out loop fusion code as functions and call these functions.
15521 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15523 * tree-loop-distribution.c (classify_partition): New parameter and
15524 better handle reduction statement.
15525 (rdg_build_partitions): Revise comment.
15526 (distribute_loop): Compute statements in all partitions and pass it
15527 to classify_partition.
15529 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15531 * tree-loop-distribution.c (enum partition_type): New.
15532 (struct partition): New field type.
15533 (partition_merge_into): Add parameter. Update partition type.
15534 (data_dep_in_cycle_p, update_type_for_merge): New functions.
15535 (build_rdg_partition_for_vertex): Compute partition type.
15536 (rdg_build_partitions): Dump partition type.
15537 (distribute_loop): Update calls to partition_merge_into.
15539 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15541 * tree-loop-distribution.c (struct ddr_hasher): New.
15542 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
15544 (classify_partition): Call get_data_dependence.
15545 (pg_add_dependence_edges): Ditto.
15546 (distribute_loop): Release data dependence hash table.
15548 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15550 * tree-loop-distribution.c (ref_base_address): Delete.
15551 (similar_memory_accesses): Rename ...
15552 (share_memory_accesses): ... to this. Check if partitions access
15553 the same memory reference.
15554 (distribute_loop): Call share_memory_accesses.
15556 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15558 * tree-loop-distribution.c (struct partition): New field recording
15559 its data reference.
15560 (partition_alloc, partition_free): Init and release data refs.
15561 (partition_merge_into): Merge data refs.
15562 (build_rdg_partition_for_vertex): Collect data refs for partition.
15563 (pg_add_dependence_edges): Change parameters from vector to bitmap.
15565 (distribute_loop): Remve data refs from vertice data of partition
15568 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15570 * tree-loop-distribution.c (params.h): Include header file.
15571 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
15572 (datarefs_vec): New global var.
15573 (create_rdg_vertices): Use datarefs_vec directly.
15574 (free_rdg): Don't free data references.
15575 (build_rdg): Update use. Don't free data references.
15576 (distribute_loop): Compute global variable for data references.
15577 Bail out if there are too many data references.
15579 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15581 * tree-loop-distribution.c (loop_nest): New global var.
15582 (build_rdg): Use loop directly, rather than loop nest.
15583 (pg_add_dependence_edges): Remove loop nest parameter. Use global
15585 (distribute_loop): Compute global variable loop nest. Update use.
15587 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15589 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
15590 (partition_merge_into): New parameter. Dump reason for fusion.
15591 (distribute_loop): Update use of partition_merge_into.
15593 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15595 * tree-loop-distribution.c (bb_top_order_index): New.
15596 (bb_top_order_index_size, bb_top_order_cmp): New.
15597 (stmts_from_loop): Use topological order.
15598 (pass_loop_distribution::execute): Compute and release topological
15599 order for basic blocks.
15601 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15603 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
15606 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15608 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
15609 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
15610 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
15611 * internal-fn.def (LOOP_DIST_ALIAS): New.
15612 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
15613 (fold_loop_internal_call): ... this.
15614 (vect_loop_dist_alias_call): New function.
15615 (set_uid_loop_bbs): Call fold_loop_internal_call.
15616 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
15619 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
15622 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
15623 Require dead FLAGS_REG at the beginning of a peephole.
15625 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
15628 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
15629 arguments in the call to __builtin_ia32_sbb_u32.
15630 (_subborrow_u64): Swap _X and _Y arguments in the call to
15631 __builtin_ia32_sbb_u64.
15633 2017-07-04 Jakub Jelinek <jakub@redhat.com>
15636 * tree-vrp.c (compare_assert_loc): Turn into a function template
15637 with stable template parameter. Only test if a->e is NULL,
15638 !a->e == !b->e has been verified already. Use e == NULL or
15639 e != NULL instead of e or ! e tests. If stable is true, don't use
15640 iterative_hash_expr, on the other side allow a or b or both NULL
15641 and sort the NULLs last.
15642 (process_assert_insertions): Sort using compare_assert_loc<false>
15643 instead of compare_assert_loc, later sort using
15644 compare_assert_loc<true> before calling process_assert_insertions_for
15645 in a loop. Use break instead of continue once seen NULL pointer.
15647 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
15649 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
15650 Cortex-R7 and Cortex-R8 processors.
15652 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
15654 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
15655 uninitialized while src is not.
15657 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
15659 * common/config/arm/arm-common.c: Adjust include path for
15661 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
15662 (arm-cpu.h): Create in build directory. Adjust dependency rules.
15663 (arm-cpu-data.h): Likewise.
15664 (arm-cpu-cdata.h): Likewise.
15665 * config/arm/arm-cpu.h: Delete.
15666 * config/arm/arm-cpu-cdata.h: Delete.
15667 * config/arm/arm-cpu-data.h: Delete.
15669 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
15671 * config/arm/arm-cpus.in (cortex-a55): New.
15672 (cortex-a75): Likewise.
15673 (cortex-a75.cortex-a55): Likewise.
15674 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
15676 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
15677 * config/arm/arm-cpu-cdata.h: Regenerate.
15678 * config/arm/arm-cpu-data.h: Regenerate.
15679 * config/arm/arm-cpu.h: Regenerate.
15680 * config/arm/arm-tables.opt: Regenerate.
15681 * config/arm/arm-tune.md: Regenerate.
15683 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
15685 * haifa-sched.c (sched_create_recovery_edges): Update profile.
15687 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
15689 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
15692 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
15694 PR tree-optimization/81292
15695 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
15696 full_string_p, also call adjust_related_strinfos if the adjustment
15697 is simple, otherwise invalidate related strinfos.
15699 2017-07-04 Martin Liska <mliska@suse.cz>
15702 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
15703 newly created variable as DECL_IGNORED_P.
15705 2017-07-04 Martin Liska <mliska@suse.cz>
15708 * ipa-inline.c (inline_small_functions):
15709 Use xstrdup_for_dump.
15711 2017-07-04 Tom de Vries <tom@codesourcery.com>
15713 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
15715 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
15718 * config/darwin.c (darwin_function_switched_text_sections):
15721 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
15723 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
15725 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
15727 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
15729 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15731 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
15732 min_profitable_iters, and th as inclusive lower bounds.
15733 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
15734 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
15735 for min_profitable_iters and min_profitable_estimate.
15736 (vect_transform_loop): Treat th as an inclusive lower bound.
15737 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
15739 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
15742 * config/darwin.c (darwin_function_switched_text_sections):
15743 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
15744 in two pieces, and suppress the use of buf.
15746 2017-07-03 Nathan Sidwell <nathan@acm.org>
15748 * hash-table.h (hash_table_mod1): Fix indentation.
15750 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15752 PR middle-end/81290
15753 * predict.c (force_edge_cold): Be more careful about propagation
15755 * profile-count.h (profile_probability::guessed,
15756 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
15758 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
15760 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
15762 * doc/invoke.texi (rcpc architecture extension): Document it.
15764 2017-07-03 Richard Biener <rguenther@suse.de>
15766 PR tree-optimization/60510
15767 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
15768 the scalar reduction PHI and use it.
15769 (vectorizable_reduction): Properly guard the single_defuse_cycle
15770 path for non-SLP reduction chains where we cannot use it.
15771 Rework reduc_def/index and vector type deduction. Rework
15772 vector operand gathering during reduction op code-gen.
15773 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
15774 chains dissolve the chain and leave it to non-SLP reduction
15777 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15779 * tree-data-ref.h (dr_alignment): Declare.
15780 * tree-data-ref.c (dr_alignment): New function.
15781 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
15782 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
15784 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
15786 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15788 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
15789 and base_misalignment fields.
15790 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
15791 * tree-data-ref.c: Include builtins.h.
15792 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
15793 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
15794 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
15795 * tree-vect-data-refs.c: Include tree-cfg.h.
15796 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
15797 fields instead of calculating an alignment here.
15798 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
15799 innermost_loop_behavior fields.
15801 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15803 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
15805 (DR_STEP_ALIGNMENT): New macro.
15806 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
15807 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
15808 (create_data_ref): Print it.
15809 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
15810 to tell whether the step preserves vector (mis)alignment.
15811 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
15812 Move the check for an integer step and generalise to all INTEGER_CST.
15813 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
15814 Print the outer step alignment.
15816 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15818 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
15819 with offset_alignment.
15820 (DR_ALIGNED_TO): Delete.
15821 (DR_OFFSET_ALIGNMENT): New macro.
15822 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
15823 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
15824 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
15825 (create_data_ref): Likewise.
15826 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
15827 (vect_analyze_data_refs): Likewise.
15828 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
15829 creating dummy innermost behavior.
15831 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15833 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
15834 with a "innermost_loop_behavior *" and refeence tree.
15835 * tree-data-ref.c (dr_analyze_innermost): Likewise.
15836 (create_data_ref): Update call accordingly.
15837 * tree-predcom.c (find_looparound_phi): Likewise.
15839 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15841 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
15842 fields with dr_wrt_vec_loop.
15843 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
15844 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
15845 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
15846 (vect_dr_behavior): New function.
15847 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
15848 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
15849 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
15850 track whether the step preserves the misalignment.
15851 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
15852 Use vect_dr_behavior.
15853 (vect_setup_realignment): Update call accordingly.
15854 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
15855 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
15856 call to vect_create_addr_base_for_vector_ref.
15857 (vect_create_cond_for_align_checks): Likewise.
15858 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
15859 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
15860 (vect_recog_mask_conversion_pattern): Likewise.
15861 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
15862 (new_stmt_vec_info): Remove redundant zeroing.
15864 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
15866 * common/config/arm/arm-common.c (arm_be8_option): New function.
15867 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
15868 (ISA_ARMv6): Add isa_bit_be8.
15869 * config/arm/arm.h (arm_be8_option): Add prototype.
15870 (BE8_SPEC_FUNCTION): New define.
15871 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
15872 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
15873 (mlittle-endian): Similarly.
15874 (mbe8, mbe32): New options.
15875 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
15876 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
15878 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15880 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
15882 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15884 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
15885 (cleanup_tree_cfg_bb): Use it.
15886 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
15888 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
15890 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15893 * loop-doloop.c (add_test): Update profile.
15895 2017-07-03 Martin Liska <mliska@suse.cz>
15898 * sanopt.c (rewrite_usage_of_param): New function.
15899 (sanitize_rewrite_addressable_params): Likewise.
15900 (pass_sanopt::execute): Call rewrite_usage_of_param.
15902 2017-07-03 Richard Biener <rguenther@suse.de>
15904 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
15905 back to using VIEW_CONVERT_EXPR.
15907 2017-07-03 Martin Liska <mliska@suse.cz>
15910 * doc/extend.texi: Document when a resolver function is
15911 generated for target_clones.
15913 2017-07-03 Martin Liska <mliska@suse.cz>
15915 * asan.c (asan_emit_stack_protection): Unpoison just red zones
15916 and shadow memory of auto variables which are subject of
15917 use-after-scope sanitization.
15918 (asan_expand_mark_ifn): Add do set only when is_poison.
15920 2016-07-03 Richard Biener <rguenther@suse.de>
15922 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
15924 (vect_force_simple_reduction): Record reduction def -> phi mapping.
15925 (vectorizable_reduction): Perform reduction PHI creation when
15926 visiting a reduction PHI and adjust and simplify code generation
15927 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
15928 (vect_transform_loop): Visit reduction PHIs.
15929 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
15930 defs into the SLP tree.
15931 (vect_build_slp_tree): Reduction defs terminate the recursion.
15932 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
15934 (vect_get_vec_defs_for_stmt_copy): Export.
15935 (vect_get_vec_defs): Likewise.
15936 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
15938 (vect_get_vec_defs_for_stmt_copy): Declare.
15939 (vect_get_vec_defs): Likewise.
15941 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15943 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
15944 parameter with a "loop" parameter and use it instead of the
15945 loop containing DR_STMT. Don't check simple_iv when doing
15946 BB analysis. Describe the two analysis modes in the comment.
15948 2017-07-03 Tom de Vries <tom@codesourcery.com>
15950 PR tree-optimization/69468
15951 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
15952 (find_same_succ_bb): Handle ignore_edge_flags.
15954 2017-07-03 Tom de Vries <tom@codesourcery.com>
15956 PR tree-optimization/81192
15957 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
15959 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
15961 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
15963 2017-07-03 Tom de Vries <tom@codesourcery.com>
15965 PR tree-optimization/81192
15966 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
15967 BB_SAME_SUCC (bb) == NULL.
15969 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15971 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
15974 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15976 * dumpfile.c: Include profile-count.h
15977 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
15979 (insert_cond_bb): Update profile.
15980 * tree-cfg.h (insert_cond_bb): Update prototype.
15981 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
15982 * tree-dump.c: Do not include tree-cfg.
15984 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15986 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
15988 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15990 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
15993 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15995 * tree-complex.c (expand_complex_div_wide): update profile.
15997 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
15998 Alan Hayward <alan.hayward@arm.com>
15999 David Sherwood <david.sherwood@arm.com>
16001 * Makefile.in (MACHMODE_H): Remove insn-modes.h
16002 (CORETYPES_H): New define.
16003 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
16004 (insn-modes-inline.h, s-modes-inline-h): New rules.
16005 (generated_files): Add insn-modes-inline.h.
16006 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
16007 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
16008 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
16009 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
16010 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
16011 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
16012 (build/gencodes.o, build/genconditions.o): Likewise.
16013 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
16014 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
16015 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
16016 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
16017 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
16018 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
16019 * coretypes.h: Include everything up to real.h for generators.
16020 Include insn-modes.h first. Include wide-int-print.h after
16021 wide-int.h. Include insn-modes-inline.h and then machmode.h.
16022 * machmode.h: Don't include insn-modes.h here.
16023 * function-tests.c: Remove includes of signop.h, machmode.h,
16024 double-int.h and wide-int.h.
16026 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
16028 * optc-save-gen.awk: Likewise.
16029 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
16030 * godump.c: Remove include of wide-int-print.h.
16031 * pretty-print.h: Likewise.
16032 * wide-int-print.cc: Likewise.
16033 * wide-int.cc: Likewise.
16034 * hash-map-tests.c: Remove include of signop.h.
16035 * hash-set-tests.c: Likewise.
16036 * rtl-tests.c: Likewise.
16037 * mkconfig.sh: Remove include of machmode.h.
16038 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
16040 (emit_insn_modes_inline_h): ...this new function. Emit the code
16041 into an insn-modes-inline.h header file, adding appropriate
16042 include guards and end comments.
16043 (emit_insn_modes_c_header): Remove include of machmode.h.
16044 (emit_min_insn_modes_c_header): Include coretypes.h rather than
16046 (main): Handle -i flag and call emit_insn_modes_inline_h when
16049 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
16051 * tree-ssa-strlen.c (strinfo): Rename the length field to
16052 nonzero_chars. Add a full_string_p field.
16053 (compare_nonzero_chars, zero_length_string_p): New functions.
16054 (get_addr_stridx): Add an offset_out parameter.
16055 Use compare_nonzero_chars.
16056 (get_stridx): Update accordingly. Use compare_nonzero_chars.
16057 (new_strinfo): Update after above changes to strinfo.
16058 (set_endptr_and_length): Set full_string_p.
16059 (get_string_length): Update after above changes to strinfo.
16060 (unshare_strinfo): Update call to new_strinfo.
16061 (maybe_invalidate): Likewise.
16062 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
16063 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
16064 as a uhwi instead of an shwi. Update after above changes to
16065 strinfo and new_strinfo.
16066 (zero_length_string): Assert that chainsi contains full strings.
16067 Use zero_length_string_p. Update call to new_strinfo.
16068 (adjust_related_strinfos): Update after above changes to strinfo.
16069 Copy full_string_p from origsi.
16070 (adjust_last_stmt): Use zero_length_string_p.
16071 (handle_builtin_strlen): Update after above changes to strinfo and
16072 new_strinfo. Install the lhs as the string length if the previous
16073 entry didn't describe a full string.
16074 (handle_builtin_strchr): Update after above changes to strinfo
16076 (handle_builtin_strcpy): Likewise.
16077 (handle_builtin_strcat): Likewise.
16078 (handle_builtin_malloc): Likewise.
16079 (handle_pointer_plus): Likewise.
16080 (handle_builtin_memcpy): Likewise. Track nonzero characters
16081 that aren't necessarily followed by a nul terminator.
16082 (handle_char_store): Likewise.
16084 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
16086 PR tree-optimization/80769
16087 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
16088 for malloc and calloc. Document the new invariant that all related
16089 strinfos have delayed lengths or none do.
16090 (verify_related_strinfos): Move earlier in file.
16091 (set_endptr_and_length): New function, split out from...
16092 (get_string_length): ...here. Also set the lengths of related
16094 (zero_length_string): Assert that chainsi has known (rather than
16096 (adjust_related_strinfos): Likewise.
16098 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
16100 PR tree-optimization/81136
16101 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
16102 assert that two references with the same misalignment have the same
16103 compile-time misalignment if those compile-time misalignments
16106 2017-07-01 Andi Kleen <ak@linux.intel.com>
16108 * print-tree.c (print_node): Print all attributes.
16110 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16112 * cfg.c (scale_bbs_frequencies): New function.
16113 * cfg.h (scale_bbs_frequencies): Declare it.
16114 * cfgloopanal.c (single_likely_exit): Cleanup.
16115 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
16117 (scale_loop_profile): Likewise.
16118 (loop_version): Likewise.
16119 (create_empty_loop_on_edge): Update.
16120 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
16121 scale_loop_frequencies, scale_loop_profile, loopify,
16122 loop_version): Update prototypes.
16123 * modulo-sched.c (sms_schedule): Update.
16124 * predict.c (unlikely_executed_edge_p): Also check probability.
16125 (probably_never_executed_edge_p): Fix typo.
16126 * tree-if-conv.c (version_loop_for_if_conversion): Update.
16127 * tree-parloops.c (gen_parallel_loop): Update.
16128 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
16129 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
16130 * tree-ssa-loop-split.c (split_loop): Update.
16131 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
16132 * tree-vect-loop-manip.c (vect_do_peeling): Update.
16133 (vect_loop_versioning): Update.
16134 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
16136 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16138 * trans-mem.c (split_bb_make_tm_edge): Update profile.
16140 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16142 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
16143 to keep profile consistent.
16145 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16147 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
16148 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
16149 * profile-count.h (max_safe_multiplier): Make unsigned.
16150 (profile_count::guessed_zero): New.
16152 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16154 * bb-reorder.c (fix_up_crossing_landing_pad,
16155 fix_crossing_conditional_branches): Use make_single_succ_edge
16156 to keep profile consistent.
16158 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16160 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
16163 2017-07-01 Jakub Jelinek <jakub@redhat.com>
16166 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
16167 the right scopes, make sure cond_jump isn't preserved between multiple
16168 iterations. Search for fallthru edge whenever there are 3+ edges and
16169 use find_fallthru_edge for it.
16171 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
16173 Patch by Alexander Monakov <amonakov@ispras.ru>
16174 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
16175 probabilities consistently.
16177 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
16179 * pa.c (pa_expand_compare_and_swap_loop): Update call of
16180 emit_cmp_and_jump_insns.
16182 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
16185 * tree-inline.c (expand_call_inline): Combine profile statuses.
16187 2017-06-30 Andrew Pinski <apinski@cavium.com>
16189 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
16190 fold_stmt returned true.
16192 2017-06-30 Nathan Sidwell <nathan@acm.org>
16194 * ggc.h (empty_string): Delete.
16195 * cfgexpand.c (expand_asm_stmt): Use plain "".
16196 * optabs.c (expand_asm_memory_barrier): Likewise.
16197 * stringpool.c (empty_string): Delete.
16198 (digit_vector, digit_string): Delete.
16199 (ggc_alloc_string): Use plain "", don't optimize single digit
16200 strings. Use ggc_alloc_atomic.
16202 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
16204 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
16205 comparison set and one other set, use the cost of the non-comparison
16208 2017-06-30 Nathan Sidwell <nathan@acm.org>
16210 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
16213 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
16215 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
16216 loops. Remove now unneeded calls to gimple_switch_set_label() that
16217 just set removed labels to NULL_TREE.
16219 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
16221 * tree-ssanames.c (set_range_info_raw): Abstract from ...
16222 (set_range_info): ...here. Only call set_range_info_raw if domain
16224 (set_nonzero_bits): Call set_range_info_raw.
16225 * tree-ssanames.h (set_range_info_raw): New.
16227 2017-06-30 Jakub Jelinek <jakub@redhat.com>
16230 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
16231 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
16232 of nonimmediate_operand and <store_mask_constraint> instead of m
16233 for the input operand. For V8FI iterator, always split if input
16234 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
16235 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
16236 <store_mask_predicate> instead of register_operand and
16237 <store_mask_constraint> instead of v for the input operand. Make
16238 sure both operands aren't MEMs for if not <mask_applied>.
16240 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
16242 * lto-wrapper.c (copy_file) Close both file descriptors before
16245 2017-06-30 Martin Liska <mliska@suse.cz>
16248 * multiple_target.c (create_dispatcher_calls): Make ifunc
16249 also for function that don't have calls or are not referenced.
16251 2017-06-30 Richard Biener <rguenther@suse.de>
16253 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
16254 analyze the first scalar stmt. Move vector type computation
16255 for the BB case here from ...
16256 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
16257 live operation processing in the SLP case properly.
16259 2017-06-30 Richard Biener <rguenther@suse.de>
16261 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
16263 2017-06-30 Martin Liska <mliska@suse.cz>
16266 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
16267 before BUILT_IN_UNWIND_RESUME when ASAN is used.
16269 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
16271 * doc/invoke.texi (AArch64): Add missing options and remove redundant
16274 2017-06-30 Richard Biener <rguenther@suse.de>
16276 PR tree-optimization/81249
16277 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
16278 condition reduction result to original scalar type.
16280 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16282 * profile-count.h (enum profile_quality): Fix typos and whitespace
16285 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16287 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
16288 type for branch probabilities.
16290 2017-06-29 Julian Brown <julian@codesourcery.com>
16291 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16293 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
16294 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
16295 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
16296 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
16298 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16300 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
16301 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
16302 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
16303 CC usage from generic code to here.
16304 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
16305 CC usage into the target macros.
16307 2017-06-29 Maya Rashish <coypu@sdf.org>
16309 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
16312 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
16314 * arm/arm-builtins.c: Include profile-count.h
16315 * except.c (sjlj_emit_function_enter): Use
16316 profile_probability::unlikely.
16318 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16320 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
16321 and tocrel_offset be pointer args rather than implicitly using
16323 (legitimate_constant_pool_address_p, rs6000_emit_move,
16324 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
16325 tocrel_offset and use in toc_relative_expr_p call.
16326 (print_operand, print_operand_address): Use static tocrel_base_oac
16327 and tocrel_offset_oac.
16328 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
16331 2017-06-29 Maya Rashish <coypu@sdf.org>
16333 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
16335 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
16337 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
16338 objects, take into account only the alignment of 'op0' and 'mode1' if
16341 2017-06-29 Steve Ellcey <sellcey@cavium.com>
16343 * ccmp.c (ccmp_tree_comparison_p): New function.
16344 (ccmp_candidate_p): Update to use above function.
16345 (get_compare_parts): New function.
16346 (expand_ccmp_next): Update to use new functions.
16347 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
16349 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
16350 take mode as argument.
16351 * ccmp.h (expand_ccmp_expr): Add mode as argument.
16352 * expr.c (expand_expr_real_1): Pass mode as argument.
16354 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
16356 * combine.c (combine_instructions): Print insns to dump_file, together
16359 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
16361 * asan.c (asan_emit_stack_protection): Update.
16362 (create_cond_insert_point): Update.
16363 * auto-profile.c (afdo_propagate_circuit): Update.
16364 * basic-block.h (struct edge_def): Turn probability to
16365 profile_probability.
16366 (EDGE_FREQUENCY): Update.
16367 * bb-reorder.c (find_traces_1_round): Update.
16368 (better_edge_p): Update.
16369 (sanitize_hot_paths): Update.
16370 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
16371 (make_single_succ_edge): Update.
16372 (check_bb_profile): Update.
16373 (dump_edge_info): Update.
16374 (update_bb_profile_for_threading): Update.
16375 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
16376 probabilitycount to 0.
16377 * cfgbuild.c (compute_outgoing_frequencies): Update.
16378 * cfgcleanup.c (try_forward_edges): Update.
16379 (outgoing_edges_match): Update.
16380 (try_crossjump_to_edge): Update.
16381 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
16382 (expand_gimple_tailcall): Update.
16383 (construct_init_block): Use make_single_succ_edge.
16384 (construct_exit_block): Use make_single_succ_edge.
16385 * cfghooks.c (verify_flow_info): Update.
16386 (redirect_edge_succ_nodup): Update.
16387 (split_edge): Update.
16388 (account_profile_record): Update.
16389 * cfgloopanal.c (single_likely_exit): Update.
16390 * cfgloopmanip.c (scale_loop_profile): Update.
16391 (set_zero_probability): Remove.
16392 (duplicate_loop_to_header_edge): Update.
16393 * cfgloopmanip.h (loop_version): Update prototype.
16394 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
16395 (force_nonfallthru_and_redirect): Update.
16396 (update_br_prob_note): Update.
16397 (rtl_verify_edges): Update.
16398 (purge_dead_edges): Update.
16399 (rtl_lv_add_condition_to_bb): Update.
16400 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
16401 * cgraphunit.c (init_lowered_empty_function): Update.
16402 (cgraph_node::expand_thunk): Update.
16403 * cilk-common.c: Include profile-count.h
16404 * dojump.c (inv): Remove.
16405 (jumpifnot): Update.
16406 (jumpifnot_1): Update.
16407 (do_jump_1): Update.
16409 (do_jump_by_parts_greater_rtx): Update.
16410 (do_compare_rtx_and_jump): Update.
16411 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
16412 do_jump_1. do_compare_rtx_and_jump): Update prototype.
16413 * dwarf2cfi.c: Include profile-count.h
16414 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
16415 (sjlj_emit_dispatch_table): Likewise.
16416 * explow.c: Include profile-count.h
16417 * expmed.c (emit_store_flag_force): Update.
16418 (do_cmp_and_jump): Update.
16419 * expr.c (compare_by_pieces_d::generate): Update.
16420 (compare_by_pieces_d::finish_mode): Update.
16421 (emit_block_move_via_loop): Update.
16422 (store_expr_with_bounds): Update.
16423 (store_constructor): Update.
16424 (expand_expr_real_2): Update.
16425 (expand_expr_real_1): Update.
16426 * expr.h (try_casesi, try_tablejump): Update prototypes.
16427 * gimple-pretty-print.c (dump_probability): Update.
16428 (dump_profile): New.
16429 (dump_gimple_label): Update.
16430 (dump_gimple_bb_header): Update.
16431 * graph.c (draw_cfg_node_succ_edges): Update.
16432 * hsa-gen.c (convert_switch_statements): Update.
16433 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
16434 (find_if_case_1): Update.
16435 (find_if_case_2): Update.
16436 * internal-fn.c (expand_arith_overflow_result_store): Update.
16437 (expand_addsub_overflow): Update.
16438 (expand_neg_overflow): Update.
16439 (expand_mul_overflow): Update.
16440 (expand_vector_ubsan_overflow): Update.
16441 * ipa-cp.c (good_cloning_opportunity_p): Update.
16442 * ipa-split.c (split_function): Use make_single_succ_edge.
16443 * ipa-utils.c (ipa_merge_profiles): Update.
16444 * loop-doloop.c (add_test): Update.
16445 (doloop_modify): Update.
16446 * loop-unroll.c (compare_and_jump_seq): Update.
16447 (unroll_loop_runtime_iterations): Update.
16448 * lra-constraints.c (lra_inheritance): Update.
16449 * lto-streamer-in.c (input_cfg): Update.
16450 * lto-streamer-out.c (output_cfg): Update.
16451 * mcf.c (adjust_cfg_counts): Update.
16452 * modulo-sched.c (sms_schedule): Update.
16453 * omp-expand.c (expand_omp_for_init_counts): Update.
16454 (extract_omp_for_update_vars): Update.
16455 (expand_omp_ordered_sink): Update.
16456 (expand_omp_for_ordered_loops): Update.
16457 (expand_omp_for_generic): Update.
16458 (expand_omp_for_static_nochunk): Update.
16459 (expand_omp_for_static_chunk): Update.
16460 (expand_cilk_for): Update.
16461 (expand_omp_simd): Update.
16462 (expand_omp_taskloop_for_outer): Update.
16463 (expand_omp_taskloop_for_inner): Update.
16464 * omp-simd-clone.c (simd_clone_adjust): Update.
16465 * optabs.c (expand_doubleword_shift): Update.
16466 (expand_abs): Update.
16467 (emit_cmp_and_jump_insn_1): Update.
16468 (expand_compare_and_swap_loop): Update.
16469 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
16470 * predict.c (predictable_edge_p): Update.
16471 (edge_probability_reliable_p): Update.
16472 (set_even_probabilities): Update.
16473 (combine_predictions_for_insn): Update.
16474 (combine_predictions_for_bb): Update.
16475 (propagate_freq): Update.
16476 (estimate_bb_frequencies): Update.
16477 (force_edge_cold): Update.
16478 * profile-count.c (profile_count::dump): Add missing space into dump.
16479 (profile_count::debug): Add newline.
16480 (profile_count::differs_from_p): Explicitly convert to unsigned.
16481 (profile_count::stream_in): Update.
16482 (profile_probability::dump): New member function.
16483 (profile_probability::debug): New member function.
16484 (profile_probability::differs_from_p): New member function.
16485 (profile_probability::differs_lot_from_p): New member function.
16486 (profile_probability::stream_in): New member function.
16487 (profile_probability::stream_out): New member function.
16488 * profile-count.h (profile_count_quality): Rename to ...
16489 (profile_quality): ... this one.
16490 (profile_probability): New.
16491 (profile_count): Update.
16492 * profile.c (compute_branch_probabilities): Update.
16493 * recog.c (peep2_attempt): Update.
16494 * sched-ebb.c (schedule_ebbs): Update.
16495 * sched-rgn.c (find_single_block_region): Update.
16496 (compute_dom_prob_ps): Update.
16497 (schedule_region): Update.
16498 * sel-sched-ir.c (compute_succs_info): Update.
16499 * stmt.c (struct case_node): Update.
16500 (do_jump_if_equal): Update.
16501 (get_outgoing_edge_probs): Update.
16502 (conditional_probability): Update.
16503 (emit_case_dispatch_table): Update.
16504 (expand_case): Update.
16505 (expand_sjlj_dispatch_table): Update.
16506 (emit_case_nodes): Update.
16507 * targhooks.c: Update.
16508 * tracer.c (better_p): Update.
16509 (find_best_successor): Update.
16510 * trans-mem.c (expand_transaction): Update.
16511 * tree-call-cdce.c: Update.
16512 * tree-cfg.c (gimple_split_edge): Upate.
16513 (move_sese_region_to_fn): Upate.
16514 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
16515 * tree-eh.c (lower_resx): Upate.
16516 (cleanup_empty_eh_move_lp): Upate.
16517 * tree-if-conv.c (version_loop_for_if_conversion): Update.
16518 * tree-inline.c (copy_edges_for_bb): Update.
16519 (copy_cfg_body): Update.
16520 * tree-parloops.c (gen_parallel_loop): Update.
16521 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
16522 (gimple_gen_time_profiler): Update.
16523 * tree-ssa-dce.c (remove_dead_stmt): Update.
16524 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
16525 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
16526 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
16527 (unloop_loops): Update.
16528 (try_peel_loop): Update.
16529 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
16530 * tree-ssa-loop-split.c (connect_loops): Update.
16531 (split_loop): Update.
16532 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
16533 (hoist_guard): Update.
16534 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
16535 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
16536 (value_replacement): Update.
16537 * tree-ssa-reassoc.c (branch_fixup): Update.
16538 * tree-ssa-tail-merge.c (replace_block_by): Update.
16539 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
16540 (create_edge_and_update_destination_phis): Update.
16541 (compute_path_counts): Update.
16542 (recompute_probabilities): Update.
16543 (update_joiner_offpath_counts): Update.
16544 (freqs_to_counts_path): Update.
16545 (duplicate_thread_path): Update.
16546 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
16547 (struct switch_conv_info): Update.
16548 (gen_inbound_check): Update.
16549 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
16550 (vect_do_peeling): Update.
16551 (vect_loop_versioning): Update.
16552 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
16553 (optimize_mask_stores): Update.
16554 * ubsan.c (ubsan_expand_null_ifn): Update.
16555 * value-prof.c (gimple_divmod_fixed_value): Update.
16556 (gimple_divmod_fixed_value_transform): Update.
16557 (gimple_mod_pow2): Update.
16558 (gimple_mod_pow2_value_transform): Update.
16559 (gimple_mod_subtract): Update.
16560 (gimple_mod_subtract_transform): Update.
16561 (gimple_ic): Update.
16562 (gimple_stringop_fixed_value): Update.
16563 (gimple_stringops_transform): Update.
16564 * value-prof.h: Update.
16566 2017-06-29 Carl Love <cel@us.ibm.com>
16568 * config/rs6000/rs6000-c.c: Add support for built-in functions
16569 vector signed int vec_signed (vector float);
16570 vector signed long long vec_signed (vector double);
16571 vector signed int vec_signed2 (vector double, vector double);
16572 vector signed int vec_signede (vector double);
16573 vector signed int vec_signedo (vector double);
16574 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
16575 instruction generator.
16576 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
16577 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
16578 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
16580 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
16581 vunsignede_v2df): Add define_expands.
16582 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
16583 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
16584 VEC_UNSIGNEDO): Add definitions.
16585 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
16586 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
16587 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
16588 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
16589 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
16590 * config/rs6000/altivec.h (vec_signed, vec_signed2,
16591 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
16592 vec_unsignede, vec_unsignedo): Add builtin defines.
16593 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
16595 * doc/extend.texi: Update the built-in documentation file for the
16596 new built-in functions.
16598 2017-06-29 Richard Biener <rguenther@suse.de>
16600 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
16601 reduction chains to LOOP_VINFO_REDUCTIONS.
16602 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
16603 SLP reductions after processing reduction chains.
16605 2017-06-29 Nathan Sidwell <nathan@acm.org>
16607 * builtins.c (fold_builtin_FUNCTION): Use
16608 lang_hooks.decl_printable_name.
16610 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
16612 PR middle-end/81194
16613 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
16614 with only one label.
16615 * stmt.c (expand_case): Assert NCASES is greater than one.
16617 2017-06-29 Richard Biener <rguenther@suse.de>
16619 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
16621 (group_case_labels): Likewise.
16622 (find_taken_edge): Push sanity checking on val to workers...
16623 (find_taken_edge_cond_expr): ... here
16624 (find_taken_edge_switch_expr): ... and here, handle cases
16625 with just a default label.
16626 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
16627 (group_case_labels): Likewise.
16628 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
16629 group_case_labels does anything cleanup the CFG again.
16631 2017-06-29 Bin Cheng <bin.cheng@arm.com>
16633 PR tree-optimization/81196
16634 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
16635 exit condition comparing two IVs.
16637 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
16639 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
16640 profile to the dummy entry at the end of the list of architectures.
16641 * config/arm/arm-cpu-cdata.h: Regenerated.
16643 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16644 Michael Collison <michael.collison@arm.com>
16647 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
16649 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
16650 (*aarch64_reg_<mode>3_minus_mask): New pattern.
16651 (*aarch64_<optab>_reg_di3_mask2): New pattern.
16652 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
16653 of shift when the shift amount is masked with constant equal to
16654 the size of the mode.
16655 * config/aarch64/predicates.md (subreg_lowpart_operator): New
16658 2017-06-29 Martin Liska <mliska@suse.cz>
16660 * config/i386/i386.opt: Change range from [1,5] to [0,5].
16662 2017-06-29 Yury Gribov <tetra2005@gmail.com>
16665 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
16668 (edge_growth_cache_entry::edge_growth_cache_entry): New
16670 (reset_edge_growth_cache): Update to use constructor.
16672 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16674 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
16675 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
16676 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
16678 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
16680 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
16681 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
16683 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
16685 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
16686 (*-linux-uclibc*): Add t-uclibc tmake_file.
16687 * config/t-musl: New.
16688 * config/t-uclibc: New.
16690 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
16692 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
16694 (gen_comm_data): Emit architectural setting of arch_prof.
16695 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
16697 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
16698 (armv8-m.base, armv8-m.main): Likewise.
16699 * arm-protos.h (arm_build_target): Add profile field.
16700 (arch_option): Likewise.
16701 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
16703 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
16704 arm_active_target.profile.
16706 2017-06-28 Richard Biener <rguenther@suse.de>
16708 PR middle-end/81227
16709 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
16710 TYPE_OVERFLOW_WRAPS.
16711 * match.pd (negate_expr_p): Likewise.
16712 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
16713 fold_build2, not fold_binary.
16715 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
16717 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
16718 Convert memory address to Pmode.
16719 (aarch64_print_operand): Assert MEM operands are always Pmode.
16721 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
16724 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
16725 Remove redundant if.
16726 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
16727 * config/arm/aarch-common-protos.h
16728 (aarch_forward_to_shift_is_not_shifted_re): Remove.
16729 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
16731 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
16734 * multiple_target.c (create_dispatcher_calls): Set the default
16735 clone to be static, not public.
16737 2017-06-28 Richard Biener <rguenther@suse.de>
16739 * tree-vect-loop.c (vectorizable_reduction): Move special
16740 cond reduction IV var creation ...
16741 (vect_create_epilog_for_reduction): ... here. Remove induction_index
16742 parameter. Use STMT_VINFO_VECTYPE.
16743 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
16746 2017-06-28 Martin Liska <mliska@suse.cz>
16749 * ipa-visibility.c (non_local_p): Handle visibility.
16751 2017-06-28 Martin Liska <mliska@suse.cz>
16754 * common.opt: Add IntegerRange to various options.
16755 * opt-functions.awk (integer_range_info): New function.
16756 * optc-gen.awk: Add integer_range_info to cl_options struct.
16757 * opts-common.c (decode_cmdline_option): Handle
16758 CL_ERR_INT_RANGE_ARG.
16759 (cmdline_handle_error): Likewise.
16760 * opts.c (print_filtered_help): Show valid interval in
16761 when --help is provided.
16762 * opts.h (struct cl_option): Add range_min and range_max fields.
16763 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
16765 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
16767 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
16768 (x * C EQ/NE y * C): New transformation.
16770 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
16772 * genmultilib (combination_space): Accept '+' in option names.
16774 2017-06-28 Martin Liska <mliska@suse.cz>
16777 * asan.c (instrument_derefs): Bail out inner references
16778 that are hard register variables.
16780 2017-06-28 Jakub Jelinek <jakub@redhat.com>
16783 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
16784 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
16786 2017-06-28 Richard Biener <rguenther@suse.de>
16788 * tree-vectorizer.h (vect_get_vec_defs): Remove.
16789 (vect_get_slp_defs): Adjust.
16790 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
16792 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
16794 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
16795 get_initial_defs_for_reduction instead of vect_get_vec_defs.
16796 (vectorizable_reduction): Adjust.
16797 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
16799 (vect_get_slp_defs): Likewise.
16800 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
16801 (vectorizable_bswap): Adjust.
16802 (vectorizable_call): Likewise.
16803 (vectorizable_conversion): Likewise.
16804 (vectorizable_assignment): Likewise.
16805 (vectorizable_shift): Likewise.
16806 (vectorizable_operation): Likewise.
16807 (vectorizable_store): Likewise.
16808 (vectorizable_condition): Likewise.
16809 (vectorizable_comparison): Likewise.
16811 2017-06-28 Michael Collison <michael.collison@arm.com>
16814 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
16816 (arm_gen_movmemqi): Removed unused variable 'i'.
16817 Convert 'for' loop into 'while' loop.
16818 (arm_expand_prologue): Remove last unnecessary set of insn.
16819 (thumb_pop): Remove unused variable 'pushed_words'.
16820 (thumb_exit): Remove last unnecessary set of regs_to_pop.
16822 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16824 * config/s390/predicates.md: Use s390_rel_address_ok_p.
16825 * config/s390/s390-protos.h: Add prototype of
16826 s390_rel_address_ok_p.
16827 * config/s390/s390.c (s390_got_symbol): New function.
16828 (s390_rel_address_ok_p): New function.
16829 (legitimize_pic_address): Use s390_rel_address_ok_p.
16830 (s390_load_got): Use s390_got_symbol.
16831 (s390_option_override): Issue error if
16832 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
16833 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
16835 * config/s390/s390.opt: New option mpic-data-is-text-relative.
16837 2017-06-27 Andrew Pinski <apinski@cavium.com>
16839 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
16840 (X * copysign (1.0, X)): New pattern.
16841 (X * copysign (1.0, -X)): New pattern.
16842 (copysign (-1.0, CST)): New pattern.
16844 2017-06-27 Joseph Myers <joseph@codesourcery.com>
16846 * genmultilib (combination_space): Remove variable.
16847 Validate reuse rules against regular expression for any sequence
16848 of multilib options in any order.
16850 2017-06-27 Michael Collison <michael.collison@arm.com>
16852 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
16853 call aarch64_split_simd_combine.
16854 * (aarch64_combine_internal<mode>): Delete pattern.
16855 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
16856 Allow register and subreg operands.
16858 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16860 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
16861 specific need, just fallback on defaults.
16862 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
16864 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16865 Olivier Hainque <hainque@adacore.com>
16867 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
16869 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
16870 targets. Pick a default if no particular attempt applied.
16871 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
16874 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16876 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
16877 (x86_64-wrs-vxworks7): Likewise.
16879 2017-06-27 Marek Polacek <polacek@redhat.com>
16882 * ubsan.c (instrument_null): Check get_base_address's result for null.
16884 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
16886 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
16888 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
16890 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
16891 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
16892 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
16893 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
16894 New function types.
16895 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
16896 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
16897 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
16898 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
16899 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
16900 BUILT_IN_FEUPDATEENV): New builtins.
16901 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
16902 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
16903 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
16904 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
16906 (builtin_structptr_types): Adjust size.
16907 * tree.c (builtin_structptr_types): Add four entries.
16909 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16910 Olivier Hainque <hainque@adacore.com>
16912 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
16913 (TLS_SYM): New local macro, forcing reference to __tls__ on
16914 link command lines for VxWorks 7 RTPs, triggering initialization
16916 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
16917 OS features TLS support, true for RTPs on VxWorks 7.
16918 * config/vxworks.c (vxworks_override_options): Setup emutls
16921 2017-06-27 Jakub Jelinek <jakub@redhat.com>
16923 * predict.c (test_prediction_value_range): Use -1U instead of -1
16924 to avoid narrowing conversion warning.
16925 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
16926 to avoid narrowing conversion warning.
16927 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
16929 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
16931 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16933 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
16934 64bit configurations.
16935 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
16936 (SIZE_TYPE): Likewise.
16937 * config/vxworks.c (vxworks_emutls_var_fields): Use
16938 long_unsigned_type_node instead of unsigned_type_node as the offset
16939 field type, which is "pointer" mode in emutls.c.
16941 2017-06-27 Jakub Jelinek <jakub@redhat.com>
16944 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
16946 PR middle-end/81207
16947 * gimple-fold.c (replace_call_with_call_and_fold): Handle
16948 gimple_vuse copying separately from gimple_vdef copying.
16950 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16952 * value-prof.c (free_hist): Remove call to memset and the enclosing if
16955 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
16956 Olivier Hainque <hainque@adacore.com>
16958 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
16959 for all vxworks7 targets.
16960 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
16961 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
16962 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
16963 variations for VX6/VX7 and 32/64bits later on in ...
16964 (VXWORKS_LIB_SPEC): Leverage new macros.
16965 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
16966 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
16968 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
16970 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
16971 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
16973 2017-06-26 Carl Love <cel@us.ibm.com>
16975 * config/rs6000/rs6000-c.c: Add support for built-in functions
16976 vector bool char vec_reve (vector bool char);
16977 vector signed char vec_reve (vector signed char);
16978 vector unsigned char vec_reve (vector unsigned char);
16979 vector bool int vec_reve (vector bool int);
16980 vector signed int vec_reve (vector signed int);
16981 vector unsigned int vec_reve (vector unsigned int);
16982 vector bool long long vec_reve (vector bool long long);
16983 vector signed long long vec_reve (vector signed long long);
16984 vector unsigned long long vec_reve (vector unsigned long long);
16985 vector bool short vec_reve (vector bool short);
16986 vector signed short vec_reve (vector signed short);
16987 vector double vec_reve (vector double);
16988 vector float vec_reve (vector float);
16989 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
16990 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
16991 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
16992 (altivec_vreve): New pattern.
16993 * config/rs6000/altivec.h (vec_reve): New define.
16994 * doc/extend.texi (vec_rev): Update the built-in documentation file
16995 for the new built-in functions.
16997 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16999 PR tree-optimization/71815
17000 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
17002 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
17004 (slsr_process_phi): Likewise.
17005 (replace_uncond_cands_and_profitable_phis): Don't replace a
17006 multiply candidate with a stride of 1 (copy or cast).
17007 (phi_incr_cost): Call uses_consumed_by_stmt rather than
17009 (lowest_cost_path): Likewise.
17010 (total_savings): Likewise.
17012 2017-06-26 Richard Biener <rguenther@suse.de>
17015 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
17016 Use def_builtin_pure for all gather builtins.
17018 2017-06-26 Richard Biener <rguenther@suse.de>
17020 PR tree-optimization/81203
17021 * tree-tailcall.c (find_tail_calls): Do not move stmts into
17022 non-dominating BBs.
17024 2017-06-26 Marek Polacek <polacek@redhat.com>
17027 * doc/invoke.texi: Document -Wmultistatement-macros.
17029 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
17031 * doc/sourcebuild.texi (ARM-specific attributes): Document new
17032 arm_neon_ok_no_float_abi effective target.
17034 2017-06-26 Richard Biener <rguenther@suse.de>
17036 PR tree-optimization/80928
17037 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
17038 (copy_bbs): Set BB_DUPLICATED flag early.
17039 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
17041 (execute_on_shrinking_pred): Likewise.
17042 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
17043 BB_DUPLICATED blocks.
17044 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
17045 iterate over all PHIs considering removal of *gsi.
17047 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
17049 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
17052 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17054 * config/rs6000/rs6000-string.c: (expand_block_clear,
17055 do_load_for_compare, select_block_compare_mode,
17056 compute_current_alignment, expand_block_compare,
17057 expand_strncmp_align_check, expand_strn_compare,
17058 expand_block_move, rs6000_output_load_multiple)
17059 Move functions related to string/block move/compare
17060 to a separate file.
17061 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
17062 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
17063 for this function which is now used in two files.
17064 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
17065 * config.gcc: Add rs6000-string.o to extra_objs for
17066 targets powerpc*-*-* and rs6000*-*-*.
17068 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
17071 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
17072 32-bit, since indexed is not valid for DImode.
17073 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
17074 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
17075 (define_peephole2 for Altivec d-form load): Add 32-bit support.
17076 (define_peephole2 for Altivec d-form store): Likewise.
17079 * multiple_target.c (create_dispatcher_calls): Only create the
17080 dispatcher call if the function is the default clone of a
17081 versioned function.
17083 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
17085 PR middle-end/80902
17086 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
17087 a call, force the call to not be a tail call.
17089 2017-06-23 Jeff Law <law@redhat.com>
17091 * doc/contrib.texi: Add entry for Steven Pemberton's work on
17094 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
17096 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
17097 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
17098 handling for early expansion of vector shifts (sl,sr,sra,rl).
17099 (builtin_function_type): Add vector shift right instructions
17100 to the unsigned argument list.
17102 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
17104 rtl-optimizatoin/79286
17105 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
17106 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
17107 trap. PIC register plus a const unspec without offset can never trap.
17109 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
17111 * tree.h (builtin_structptr_type): New type.
17112 (builtin_structptr_types): Declare new array.
17113 * tree.c (builtin_structptr_types): New array.
17114 (free_lang_data, build_common_tree_nodes): Use it.
17116 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
17119 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
17122 2017-06-22 Matt Turner <mattst88@gmail.com>
17124 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
17125 Lake models to skylake case. Assume skylake for unknown
17126 models with clflushopt.
17128 2017-06-22 Jeff Law <law@redhat.com>
17130 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
17131 frame sizes that do not satisfy aarch64_uimm12_shift.
17133 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
17135 * profile-count.h (apply_probability,
17136 apply_scale, probability_in): Fix checks for zero.
17138 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17140 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
17141 * doc/cppdiropts.texi (-I @var{dir}): Document it.
17143 2016-06-22 Richard Biener <rguenther@suse.de>
17145 * tree-vect-loop.c (vect_model_reduction_cost): Handle
17146 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
17147 REDUC_MAX_EXPR support.
17148 (vectorizable_reduction): Likewise.
17149 (vect_create_epilog_for_reduction): Likewise.
17151 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
17153 * match.pd (A / (1 << B) -> A >> B): New.
17154 * generic-match-head.c: Include optabs-tree.h.
17155 * gimple-match-head.c: Likewise.
17156 * optabs-tree.h (target_supports_op_p): New.
17157 * optabs-tree.c (target_supports_op_p): New.
17159 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17161 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
17162 $gcc_cv_ld --help output.
17163 (gcc_cv_ld_demangle): Likewise.
17164 (gcc_cv_ld_eh_frame_hdr): Likewise.
17165 (gcc_cv_ld_pie): Likewise.
17166 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
17167 (gcc_cv_ld_buildid): Likewise.
17168 (gcc_cv_ld_sysroot): Likewise.
17169 (ld_bndplt_support): Likewise.
17170 (ld_pushpopstate_support): Likewise.
17171 * configure: Regenerate.
17172 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
17174 2017-06-21 Jakub Jelinek <jakub@redhat.com>
17177 * config/i386/sse.md (round<mode>2): Renumber match_dup and
17178 operands indexes to avoid gap between operands and match_dups.
17180 2017-06-21 Andrew Pinski <apinski@cavium.com>
17182 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
17183 Increment Arith_shift and Arith_shift_reg by 1.
17184 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
17186 * config/aarch64/aarch64.c (thunderx_tunings): Enable
17187 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
17188 (aarch64_strip_extend): Add new argument and test for it.
17189 (aarch64_cheap_mult_shift_p): New function.
17190 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
17191 add a cost if it is true.
17192 Update calls to aarch64_strip_extend.
17193 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
17195 2017-06-21 Andrew Pinski <apinski@cavium.com>
17197 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
17199 (thunderxt88): Likewise.
17200 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
17201 (thunderx_prefetch_tune): New variable.
17202 (thunderx2t99_prefetch_tune): Update for the correct values.
17203 (thunderxt88_tunings): New variable.
17204 (thunderx_tunings): Use thunderx_prefetch_tune instead of
17205 generic_prefetch_tune.
17206 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
17208 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17210 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
17211 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
17212 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
17213 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
17214 (aarch64_atomic_cas<mode>, GPI): Likewise.
17216 2017-06-21 Martin Liska <mliska@suse.cz>
17218 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
17219 statements on cold and hot labels.
17220 * predict.c (tree_estimate_probability_bb): Remove the
17221 prediction from this place.
17223 2017-06-21 Martin Liska <mliska@suse.cz>
17225 PR tree-optimization/79489
17226 * gimplify.c (maybe_add_early_return_predict_stmt): New
17228 (gimplify_return_expr): Call the function.
17229 * predict.c (tree_estimate_probability_bb): Remove handling
17231 * predict.def: Update comment about early return predictor.
17232 * gimple-predict.h (is_gimple_predict): New function.
17233 * predict.def: Change default value of early return to 66.
17234 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
17236 * passes.def: Put pass_strip_predict_hints to the beginning of
17239 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
17241 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
17242 FUNCTION_DECL declarations.
17243 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
17245 (dwaf2out_decl): Likewise.
17246 * godump.c (go_early_global_decl): Skip call to the real debug hook
17247 for FUNCTION_DECL declarations.
17248 * passes.c (rest_of_decl_compilation): Skip call to the
17249 early_global_decl debug hook for FUNCTION_DECL declarations, unless
17250 -fdump-go-spec is passed.
17252 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
17254 * config/i386/i386.c (struct builtin_isa): New field pure_p.
17255 Reorder for compactness.
17256 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
17257 (def_builtin_pure, def_builtin_pure2): New functions.
17258 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
17260 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
17262 * match.pd (nop_convert): New predicate.
17263 ((A +- CST1) +- CST2): Allow some NOP conversions.
17265 2017-06-21 Jakub Jelinek <jakub@redhat.com>
17268 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
17269 with ctors/dtors if GOVD_SHARED is set.
17271 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
17273 * config/aarch64/aarch64.md (movti_aarch64):
17274 Emit mov rather than orr.
17275 (movtf_aarch64): Likewise.
17276 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
17277 Emit mov rather than orr.
17279 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
17281 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
17282 Swap alternatives, make integer dup more expensive.
17284 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
17286 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
17287 Return true for non-tls symbols.
17289 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
17291 * config/aarch64/aarch64-cores.def (cortex-a55): New.
17292 (cortex-a75): Likewise.
17293 (cortex-a75.cortex-a55): Likewise.
17294 * config/aarch64/aarch64-tune.md: Regenerate.
17295 * doc/invoke.texi (-mtune): Document new values for -mtune.
17297 2017-06-21 Tom de Vries <tom@codesourcery.com>
17299 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
17300 stack_size feature.
17301 (Effective-Target Keywords, Other attributes): Suggest using
17302 dg-add-options stack_size feature to get stack limit in stack_size
17303 effective target documentation.
17305 2017-06-21 Julian Brown <julian@codesourcery.com>
17306 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17308 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
17309 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
17310 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
17312 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
17313 attribute type list for neon_multiply.
17314 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
17315 attribute type list for neon_multiply.
17316 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
17317 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
17318 attribute type list for neon_multiply.
17319 * config/arm/types.md (crypto_pmull): Add.
17320 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
17321 attribute type list.
17323 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
17325 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
17328 2017-06-20 Jakub Jelinek <jakub@redhat.com>
17330 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
17331 to make sure not to dereference a NULL cost_classes_ptr pointer.
17333 2017-06-20 Carl Love <cel@us.ibm.com>
17335 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17336 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
17337 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
17338 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
17339 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
17340 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
17341 VMULOSW): New enum "unspec" values.
17342 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
17343 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
17344 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
17345 altivec_vmulosw): New patterns.
17346 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
17347 VMULOSW): Add definitions.
17349 2017-06-20 Julia Koval <julia.koval@intel.com>
17351 * config/i386/i386.c: Fix rounding expand for new pattern.
17352 * config/i386/subst.md: Fix pattern (parallel -> unspec).
17354 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
17356 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
17357 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
17359 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
17361 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
17364 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
17366 * config/aarch64/aarch64-cores.def: Rearrange to sort by
17367 architecture, then by implementer ID.
17368 * config/aarch64/aarch64-tune.md: Regenerate.
17370 2017-06-20 Richard Biener <rguenther@suse.de>
17372 PR middle-end/81097
17373 * fold-const.c (split_tree): Fold to type before negating.
17375 2017-06-20 David Malcolm <dmalcolm@redhat.com>
17377 * diagnostic-show-locus.c
17378 (selftest::test_fixit_deletion_affecting_newline): New function.
17379 (selftest::diagnostic_show_locus_c_tests): Call it.
17381 2017-06-20 Andreas Schwab <schwab@suse.de>
17384 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
17387 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17389 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
17390 __ARM_FEATURE_COPROC according to support.
17392 2017-06-20 Jakub Jelinek <jakub@redhat.com>
17394 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
17395 Rewritten to avoid overflow for > 32-bit pointers.
17398 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
17399 by removing enum keyword.
17400 (ubsan_type_descriptor): Likewise. Formatting fix.
17403 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
17404 splitter): Require TARGET_SSE2 in the condition.
17406 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
17409 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
17410 for doing vector set of SFmode on ISA 3.0.
17411 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
17412 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
17414 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
17415 SFmode value into a V4SF variable that was extracted from another
17416 V4SF variable without converting the element to double precision
17417 and back to single precision vector format.
17418 (vsx_insert_extract_v4sf_p9_2): Likewise.
17420 2017-06-19 Jakub Jelinek <jakub@redhat.com>
17422 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
17423 in UWHI to avoid undefined overflow.
17426 * ubsan.h (enum ubsan_encode_value_phase): New.
17427 (ubsan_encode_value): Change second argument to
17428 enum ubsan_encode_value_phase with default value of
17429 UBSAN_ENCODE_VALUE_GENERIC.
17430 * ubsan.c (ubsan_encode_value): Change second argument to
17431 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
17432 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
17433 create_tmp_var_raw instead of create_tmp_var and use a
17435 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
17436 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
17437 ubsan_encode_value callers.
17440 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
17441 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
17442 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
17444 2017-06-19 Richard Biener <rguenther@suse.de>
17446 PR middle-end/81118
17447 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
17448 estimates if we changed anything.
17450 2017-06-19 Richard Biener <rguenther@suse.de>
17452 PR tree-optimization/80887
17453 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
17454 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
17455 simplified lookups, then reset mprts_hook.
17456 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
17458 (try_to_simplify): Likewise.
17460 2017-06-19 Martin Liska <mliska@suse.cz>
17463 * gimplify.c (gimplify_switch_expr):
17464 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
17466 2017-06-19 Martin Liska <mliska@suse.cz>
17468 * doc/install.texi: Document that PGO runs in 4 stages.
17470 2017-06-19 Martin Liska <mliska@suse.cz>
17473 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
17474 to dispatcher function name.
17475 * multiple_target.c (replace_function_decl): New function.
17476 (create_dispatcher_calls): Redirect both edges and references.
17478 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
17480 * profile-count.c (profile_count::dump): Dump quality.
17481 (profile_count::differs_from_p): Update for unsigned val.
17482 * profile-count.h (profile_count_quality): New enum.
17483 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
17485 2017-06-19 Richard Biener <rguenther@suse.de>
17487 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
17488 struct function as arg.
17489 (estimate_numbers_of_iterations): Export overload with loop arg.
17490 (free_numbers_of_iterations_estimates_loop): Use an overload of
17491 free_numbers_of_iterations_estimates instead.
17492 * tree-cfg.c (remove_bb): Adjust.
17493 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
17494 * tree-parloops.c (gen_parallel_loop): Likewise.
17495 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
17497 (tree_unroll_loops_completely): Likewise.
17498 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
17499 Use an overload instead and export.
17500 (estimated_loop_iterations): Adjust.
17501 (max_loop_iterations): Likewise.
17502 (likely_max_loop_iterations): Likewise.
17503 (estimate_numbers_of_iterations): Take struct function as arg
17505 (loop_exits_before_overflow): Adjust.
17506 (free_numbers_of_iterations_estimates_loop): Use an overload.
17507 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
17508 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
17510 2017-06-19 Richard Biener <rguenther@suse.de>
17513 * ipa-prop.c (find_constructor_constant_at_offset): Handle
17514 RANGE_EXPR conservatively.
17516 2017-06-16 Carl Love <cel@us.ibm.com>
17518 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17519 definitions for vec_float, vec_float2, vec_floato,
17520 vec_floate built-ins.
17521 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
17522 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
17524 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
17525 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
17526 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
17527 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
17528 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
17529 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
17530 vec_floato): Add builtin defines.
17531 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
17532 Update the built-in documentation file for the new built-in
17535 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17537 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
17538 (mthumb): Mark as the negative of -marm.
17540 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17542 * doc/invoke.texi (ARM Options, -mcpu): Document supported
17544 (ARM Options, -mtune): Document that this accepts the same
17545 extension options as -mcpu.
17546 (ARM Options, -mfpu): Document addition of -mfpu=auto.
17548 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17550 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
17551 permitted extensions.
17553 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17555 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
17556 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
17557 (armv8-m.main): Add option +nodsp.
17558 * config/arm/arm-cpu-cdata.h: Regenerated.
17560 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17562 * config/arm/t-fuchsia: New file.
17563 * config.gcc (arm*-*-fuchsia*): Use it.
17565 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17567 * config/arm/t-symbian: Rewrite for new option infrastructure.
17569 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17571 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
17572 (MULTILIB_REQUIRED): Likewise.
17574 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17576 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
17577 (MULTILIB_RESUE): Likewise.
17578 (MULTILIB_MATCHES): Likewise.
17579 (MULTLIB_REQUIRED): Likewise.
17581 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17583 * config/arm/t-rtems: Rewrite for new option framework.
17585 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17587 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
17588 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
17589 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
17590 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
17591 * config/arm/t-multilib: ... here.
17592 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
17593 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
17594 armv7-a and armv8*-a when A-profile libraries have not been built.
17595 * config/arm/t-rmprofile: Rewrite.
17597 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17599 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
17600 with a backslash. Remove the backslash after substituting unescaped
17602 * doc/fragments.texi (MULTILIB_REUSE): Document it.
17604 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17606 * config.gcc: (arm*-*-*): When building a-profile libraries, force
17607 the driver to pass through the default setting of -mfloat-abi.
17608 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
17610 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
17611 (all_feat_combs): New rule.
17612 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
17614 * config/arm/t-aprofile: Rewrite.
17616 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17618 * config/arm/arm.h (FPUTYPE_AUTO): Define.
17619 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
17620 fpu is not specified by the user/command-line.
17621 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
17622 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
17623 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
17624 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
17625 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
17626 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
17628 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17630 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
17631 * config/arm/t-arm-elf: Rewritten.
17633 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17635 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
17636 have some floating-point instructions.
17637 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
17638 (TARGET_MAYBE_HARD_FLOAT): New macro.
17639 * config/arm/arm-builtins.c (arm_init_builtins): Use
17640 TARGET_MAYBE_HARD_FLOAT.
17641 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
17643 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17645 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
17646 (configargs.h): Include it.
17647 (arm_print_hint_for_fpu_option): New function.
17648 (arm_parse_fpu_option): New function.
17649 (candidate_extension): New class.
17650 (arm_canon_for_multilib): New function.
17651 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
17652 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
17653 (ARCH_CANONICAL_SPECS): New macro.
17654 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
17656 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17658 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
17659 are set after handling multilib fragments. Set target_cpu_default2
17662 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17664 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
17666 (arm*-*-*): Set target_cpu_default2 to a quoted string.
17667 * config/arm/parsecpu.awk (check_cpu): Validate any extension
17669 (check_arch): Likewise.
17670 * config/arm/arm.c (arm_configure_build_target): Handle
17671 TARGET_CPU_DEFAULT being a string constant. Scan any feature
17672 options in the default.
17674 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17676 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
17677 when an option is an alias of another.
17678 * config/arm/parsecpu.awk (optalias): New parser token.
17679 (gen_comm_data): Mark non-alias options as such. Emit entries
17680 for extension aliases.
17681 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
17682 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
17683 (armv6kz, armv6zk, armv6t2): Likewise.
17684 (armv7): Make vfpv3-d16 an alias.
17685 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
17687 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
17688 Sort in canonical order.
17689 (armv8-a): Sort in canonical order.
17690 (armv8.1-a, armv8.2-a): Likewise.
17691 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
17693 (cortex-a9): Sort in canonical order.
17694 * config/arm/arm.c (selftests.h): Include it.
17695 (arm_test_cpu_arch_data): New function.
17696 (arm_run_self_tests): New function.
17697 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
17698 (targetm): Move declaration to the end of the file.
17699 * arm-cpu-cdata.h: Regenerated.
17701 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17703 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
17704 call to target_mode_check describing the type of option passed.
17705 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
17706 (arm_target_thumb_only): Use arm_parse_arch_option_name or
17707 arm_parse_cpu_option_name to match parameters against list of
17709 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
17710 arm_arch_core_flags data structure.
17711 * config/arm/arm-cpu_cdata.h: Regenerated.
17713 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17715 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
17717 (arm_print_hint_for_cpu_option): Likewise.
17718 (arm_print_hint_for_arch_option): Likewise.
17719 (arm_parse_cpu_option_name): Likewise.
17720 (arm_parse_arch_option_name): Likewise.
17721 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
17722 of entries in the all_fpus list.
17723 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
17724 (arm_parse_cpu_option_name): Declare.
17725 (arm_parse_arch_option_name): Declare.
17726 (arm_parse_option_features): Declare.
17727 (arm_intialize_isa): Declare.
17728 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
17730 (gen_comm_data): ... here. Make definitions non-static.
17731 * config/arm/arm-cpu-data.h: Regenerated.
17732 * config/arm/arm-cpu-cdata.h: Regenerated.
17734 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17736 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
17737 (cpu_arch_extension): New structure.
17738 (cpu_arch_option, arch_option, cpu_option): New structures.
17739 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
17740 architecture types.
17741 (gen_data): Generate new format data tables.
17742 * config/arm/arm.c (cpu_tune): New structure.
17743 (cpu_option, processors): Delete.
17744 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
17745 (arm_print_hint_for_cpu_option): ... this and ...
17746 (arm_print_hint_for_arch_option): ... this.
17747 (arm_parse_arch_cpu_name): Delete. Replace with ...
17748 (arm_parse_cpu_option_name): ... this and ...
17749 (arm_parse_arch_option_name): ... this.
17750 (arm_unrecognized_feature): Change type of target parameter to
17752 (arm_parse_arch_cpu_features): Delete. Replace with ...
17753 (arm_parse_option_features): ... this.
17754 (arm_configure_build_target): Rework to use new configuration data
17756 (arm_print_tune_info): Rework for new configuration data tables.
17757 * config/arm/arm-cpu-data.h: Regenerated.
17758 * config/arm/arm-cpu.h: Regenerated.
17760 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17762 * Makefile.in (OBJS): Move sbitmap.o from here ...
17763 (OBJS-libcommon): ... to here.
17765 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17767 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
17768 (ISA_ALL_CRYPTO): New macro.
17769 (ISA_ALL_SIMD): New macro
17770 (ISA_ALL_FP): New macro.
17771 * config/arm/arm.c (fpu_bitlist): Update initializer.
17772 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
17774 (arm9e): Add fpu. Add option for nofp
17775 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
17776 (arm926ej-s, arm1026ej-s): Likewise.
17777 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
17778 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
17779 neon-fp16, neon-vfpv4, nofp and nosimd.
17780 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
17781 (cortex-a8): Add fpu. Add option for nofp.
17782 (cortex-a9): Add fpu. Add options for nosimd and nofp.
17783 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
17784 (cortex-r4f): Add fpu.
17785 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
17786 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
17788 (cortex-r8): Likewise.
17789 (cortex-m4): Add fpu. Add option for nofp.
17790 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
17791 (cortex-a17.cortex-a7): Likewise.
17792 (cortex-a32): Add fpu. Add options for crypto and nofp.
17793 (cortex-a35, cortex-a53): Likewise.
17794 (cortex-a57): Add fpu. Add option for crypto.
17795 (cortex-a72, cortex-a73): Likewise.
17796 (exynos-m1): Likewise.
17797 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
17798 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
17799 (cortex-m33): Add fpu. Add option for nofp.
17800 * config/arm/arm-cpu-cdata.h: Regenerated
17801 * config/arm/arm-cpu-data.h: Regenerated.
17803 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17805 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
17806 (armv5te, armv5tej): Likewise.
17807 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
17808 (armv7): Add options fp and vfpv3-d16.
17809 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
17810 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
17812 (armv7ve): Likewise.
17813 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
17814 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
17815 (armv8-a): Add nocrypto option.
17816 (armv8.1-a, armv8.2-a): Likewise.
17817 (armv8-m.main): add options fp, fp.dp and nofp.
17819 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17821 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
17823 (armv8-a+crc): Delete.
17824 (armv8.1-a): Add options simd, crypto and nofp.
17825 (armv8.2-a): Add options fp16, simd, crypto and nofp.
17826 (armv8.2-a+fp16): Delete.
17827 (armv8-m.main): Add option dsp.
17828 (armv8-m.main+dsp): Delete.
17829 (cortex-a8): Add fpu. Add nofp option.
17830 (cortex-a9): Add fpu. Add nofp and nosimd options.
17831 * config/arm/parsecpu.awk (gen_data): Generate option tables and
17832 link to main cpu and architecture data structures.
17833 (gen_comm_data): Only put isa attributes from the main architecture
17835 (option): New statement for architecture and CPU entries.
17836 * arm.c (struct cpu_option): New structure.
17837 (struct processors): Add entry for options.
17838 (arm_unrecognized_feature): New function.
17839 (arm_parse_arch_cpu_name): Ignore any characters after the first
17841 (arm_parse_arch_cpu_feature): New function.
17842 (arm_configure_build_target): Separate out any CPU and architecture
17843 features and parse separately. Don't error out if -mfpu=auto is
17844 used with only an architecture string.
17845 (arm_print_asm_arch_directives): New function.
17846 (arm_file_start): Call it.
17847 * config/arm/arm-cpu-cdata.h: Regenerated.
17848 * config/arm/arm-cpu-data.h: Likewise.
17849 * config/arm/arm-tables.opt: Likewise.
17851 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17853 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
17854 assembler when it is not -mfpu=auto.
17856 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17858 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
17859 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
17860 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
17861 (ASM_CPU_SPEC): Rewrite.
17862 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
17863 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
17864 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
17865 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
17866 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
17867 copied string is NUL-terminated. Also strip any characters prefixed
17869 (arm_rewrite_selected_arch): New function.
17870 (arm_rewrite_march): New function.
17872 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17874 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
17875 (x_arm_cpu_string, x_arm_tune_string): Likewise.
17876 (march, mcpu, mtune): Convert to string-based options.
17877 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
17878 (arm_parse_arch_cpu_name): New function.
17879 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
17880 identify selected architecture or CPU.
17881 (arm_option_save): New function.
17882 (TARGET_OPTION_SAVE): Redefine.
17883 (arm_option_restore): Restore string options.
17884 (arm_option_print): Print string options.
17886 2017-06-16 Martin Sebor <msebor@redhat.com>
17888 PR tree-optimization/80933
17889 PR tree-optimization/80934
17890 * builtins.c (fold_builtin_3): Do not handle bcmp here.
17891 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
17892 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
17893 (gimple_fold_builtin): Call them.
17895 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17897 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
17898 as unlikely; update profile.
17900 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17902 * predict.c (force_edge_cold): Handle declaring edges impossible
17905 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17907 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
17909 (try_unroll_loop_completely): Fix reporting.
17911 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17913 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
17915 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
17918 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
17919 if given a non-constant argument for an intrinsic which requires a
17922 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17924 * profile.c (compare_freqs): New function.
17925 (branch_prob): Sort edge list.
17926 (find_spanning_tree): Assume that the list is priority sorted.
17928 2017-06-16 Richard Biener <rguenther@suse.de>
17930 PR tree-optimization/81090
17931 * passes.def (pass_record_bounds): Remove.
17932 * tree-pass.h (make_pass_record_bounds): Likewise.
17933 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
17934 make_pass_record_bounds): Likewise.
17935 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
17936 not free niter estimates at the beginning but at the end.
17937 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
17939 2017-06-16 Richard Biener <rguenther@suse.de>
17941 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
17942 initializer to workaround ICE in host GCC 4.8.
17944 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17946 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
17948 (clone_inlined_nodes): Update.
17950 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17952 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
17953 prefetch settings, and enable prefetching by default at -O3.
17955 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17957 * config/aarch64/aarch64.c (aarch64_override_options_internal):
17958 Set flag_prefetch_loop_arrays according to tuning data.
17960 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17962 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
17963 New tune structure.
17964 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
17965 [Unrelated to main purpose of the patch] Place the pointer field last
17966 to enable type checking errors when tune structure are wrongly merged.
17967 * config/aarch64/aarch64.c (generic_prefetch_tune,)
17968 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
17969 (thunderx2t99_prefetch_tune): New tune constants.
17970 (tune_params *_tunings): Update all tunings (no functional change).
17971 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
17972 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
17973 from tunings structures.
17975 2017-06-16 Jakub Jelinek <jakub@redhat.com>
17978 * ubsan.c (instrument_null): Add T argument, use it instead
17979 of computing it based on IS_LHS.
17980 (instrument_object_size): Likewise.
17981 (pass_ubsan::execute): Adjust instrument_null and
17982 instrument_object_size callers to pass gimple_get_lhs or
17983 gimple_assign_rhs1 result to it. Use instrument_null instead of
17984 calling get_base_address and instrument_mem_ref. Handle
17985 aggregate call arguments for object-size sanitization.
17987 2017-06-16 Yury Gribov <tetra2005@gmail.com>
17989 PR tree-optimization/81089
17990 * tree-vrp.c (is_masked_range_test): Validate operands of
17993 2017-06-15 Martin Sebor <msebor@redhat.com>
17996 * dumpfile.c (dump_register): Avoid calling memset to initialize
17997 a class with a default ctor.
17998 * gcc.c (struct compiler): Remove const qualification.
17999 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
18000 * hash-table.h: Ditto.
18001 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
18003 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
18004 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
18006 * params.h (struct param_info): Make struct members non-const.
18007 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
18008 with default initialization.
18009 * vec.h (vec_copy_construct, vec_default_construct): New helper
18011 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
18012 with vec_copy_construct.
18013 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
18014 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
18015 * doc/invoke.texi (-Wclass-memaccess): Document.
18017 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18019 * emit-rtl.h (is_leaf): Update comment about local
18020 register allocator.
18022 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
18025 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
18026 for a variable to have a section before checking if the section has a
18028 Set section to.persistent if persistent attribute is set.
18029 Warn if .persistent attribute is used on an automatic variable.
18031 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
18033 PR rtl-optimization/80474
18034 * reorg.c (update_block): Do not ignore instructions in a delay slot.
18036 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
18038 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
18041 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
18043 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
18044 (casesi): Emit bounds checking as RTL.
18045 (casesi_internal_mips16_<mode>): Remove bounds checking.
18047 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
18049 * config/xtensa/xtensa.c (xtensa_option_override): Append
18050 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
18051 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
18052 xtensa_doloop_hooks): Define unconditionally.
18053 (xtensa_reorg_loops): Only call reorg_loops in the presence of
18055 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
18056 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
18057 for it in xtensa_option_override.
18058 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
18059 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
18061 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
18063 * doc/cppopts.texi: Document '-' special value to -MF.
18065 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
18067 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
18068 (cortex_a53_fconst): Likewise.
18069 (cortex_a53_fpmul): Likewise.
18070 (cortex_a53_f_load_64): Likewise.
18071 (cortex_a53_f_load_many): Likewise.
18072 (cortex_a53_advsimd_alu): Likewise.
18073 (cortex_a53_advsimd_alu_q): Likewise.
18074 (cortex_a53_advsimd_mul): Likewise.
18075 (cortex_a53_advsimd_mul_q): Likewise.
18076 (fpmac bypass): Add new bypass for fpmac-fpmac case.
18077 Add missing fmul, r2f_cvt and fconst cases.
18079 2017-06-14 Richard Biener <rguenther@suse.de>
18081 PR middle-end/81088
18082 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
18084 (fold_binary_loc): When associating do not treat pre-existing
18085 TREE_OVERFLOW on literal constants as a reason to allow
18086 TREE_OVERFLOW on associated literal constants.
18088 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
18090 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
18091 (MASK_FEATURES): New macro.
18092 * config/sparc/sparc.c (sparc_option_override): Remove the special
18093 handling of -mfpu and generalize it to all MASK_FEATURES switches.
18095 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
18097 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
18098 a division of 0 if non-call exceptions are enabled.
18100 2017-06-14 Andrew Pinski <apinski@cavium.com>
18101 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18104 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
18105 Improve vector initialization code gen for only variable case.
18107 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
18109 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
18111 2017-06-14 Richard Biener <rguenther@suse.de>
18113 PR tree-optimization/81083
18114 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
18117 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18119 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
18120 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
18121 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
18122 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
18123 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
18124 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
18126 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18128 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
18129 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
18131 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18133 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
18135 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18137 * config/rs6000/t-rtems: Don't handle SPE.
18139 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18141 * config/rs6000/t-linux: Don't handle SPE.
18143 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18145 * config/rs6000/eabispe.h: Delete file.
18147 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18149 * config/rs6000/t-spe: Delete file.
18151 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18153 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
18154 (rs6000_legitimate_offset_address_p): Return false for anything in
18155 V2SImode or V2SFmode.
18157 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18159 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
18160 except V2SF and V2SI. Rearrange the vector modes, and add comments.
18161 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
18163 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
18164 (rs6000_legitimate_offset_address_p): Ditto.
18165 (rs6000_emit_move): Ditto.
18166 (rs6000_init_builtins): Remove V4HI_type_node.
18168 2017-06-13 Martin Liska <mliska@suse.cz>
18171 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
18172 (gate_asan): Likewise.
18173 * asan.h (asan_no_sanitize_address_p): Remove the function.
18174 (sanitize_flags_p): New function.
18175 * builtins.def: Fix coding style.
18176 * common.opt: Use renamed enum value.
18177 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
18178 * doc/extend.texi: Document no_sanitize attribute.
18179 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
18180 to SANITIZE_UNDEFINED_NONDEFAULT.
18181 * gcc.c (sanitize_spec_function): Use the renamed enum value.
18182 * gimple-fold.c (optimize_atomic_compare_exchange_p):
18183 Use sanitize_flags_p.
18184 * gimplify.c (gimplify_function_tree): Likewise.
18185 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
18186 * opts.c (parse_no_sanitize_attribute): New function.
18187 (common_handle_option): Use renamed enum value.
18188 * opts.h (parse_no_sanitize_attribute): Declare.
18189 * tree.c (sanitize_flags_p): New function.
18190 * tree.h: Declared here.
18191 * tsan.c: Use sanitize_flags_p.
18192 * ubsan.c (ubsan_expand_null_ifn): Likewise.
18193 (instrument_mem_ref): Likewise.
18194 (instrument_bool_enum_load): Likewise.
18195 (do_ubsan_in_current_function): Remove the function.
18196 (pass_ubsan::execute): Use sanitize_flags_p.
18197 * ubsan.h: Remove do_ubsan_in_current_function
18198 * tree-cfg.c (print_no_sanitize_attr_value): New function.
18199 (dump_function_to_file): Use it here.
18201 2017-06-13 Martin Jambor <mjambor@suse.cz>
18203 PR tree-optimization/80803
18204 PR tree-optimization/81063
18205 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
18206 (propagate_subaccesses_across_link): Enqueue subtree whenever
18207 necessary instead of relying on the caller.
18209 2017-06-13 Martin Jambor <mjambor@suse.cz>
18211 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
18212 that have a first_link.
18213 (sort_and_splice_var_accesses): Do not check first_link before
18215 (subtree_mark_written_and_enqueue): Likewise.
18216 (propagate_all_subaccesses): Likewise and do not stop at first
18217 parent with a first_link.
18219 2017-06-13 Martin Jambor <mjambor@suse.cz>
18221 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
18224 2017-06-13 Yury Gribov <tetra2005@gmail.com>
18226 * match.pd: New pattern.
18228 2017-06-13 Yury Gribov <tetra2005@gmail.com>
18230 * tree-vrp.c (is_masked_range_test): New function.
18231 (register_edge_assert_for): Determine ranges for
18234 2017-06-13 Yury Gribov <tetra2005@gmail.com>
18236 PR tree-optimization/67328
18237 * fold-const.c (maskable_range_p): New function.
18238 (build_range_check): Generate bittests if possible.
18240 2017-06-13 Martin Liska <mliska@suse.cz>
18242 * gimple-pretty-print.c (dump_probability): Add new argument.
18243 (dump_edge_probability): Dump both probability and count.
18244 (dump_gimple_label): Likewise.
18245 (dump_gimple_bb_header): Likewise.
18247 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
18250 * config/avr/avr-devices.c: Fix indentation.
18251 * config/avr/gen-avr-mmcu-specs.c: Dito.
18253 2017-06-13 Richard Biener <rguenther@suse.de>
18255 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
18256 instead get vector type from stmt_info.
18257 (vectorizable_reduction): Adjust. Remove dead code.
18259 2017-06-13 Richard Biener <rguenther@suse.de>
18261 PR middle-end/81065
18262 * fold-const.c (extract_muldiv_1): Remove bogus distribution
18263 case of C * (x * C2 + C3).
18264 (fold_addr_of_array_ref_difference): Properly fold index difference.
18266 2017-06-12 David S. Miller <davem@davemloft.net>
18269 * config/sparc/sparc.md (return expander): Emit frame blockage if
18270 function uses alloca.
18272 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
18274 * combine.c (make_field_assignment): Check len rather than the mode
18275 precision when calling force_to_mode.
18277 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
18279 Support multilibs and devices that see flash in RAM address range.
18282 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
18283 (avr_mcu_t) <flash_pm_offset>: New field.
18284 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
18285 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
18286 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
18287 (AVR_TINY_PM_OFFSET): Remove macro.
18288 * config/avr/avr.opt (-mshort-calls): New option.
18289 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
18290 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
18291 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
18292 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
18293 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
18294 instead of avr_arch->have_jmp_call.
18295 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
18296 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
18297 avr_arch->flash_pm_offset to define.
18298 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
18299 new field flash_pm_offset. Add entry for avrxmega3.
18300 (avr_texinfo): Add entry for avrxmega3.
18301 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
18302 attiny212, attiny214,
18303 attiny412, attiny414, attiny416, attiny417,
18304 attiny814, attiny816, attiny817,
18305 attiny1614, attiny1616, attiny1617,
18306 attiny3214, attiny3216, attiny3217.
18307 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
18308 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
18309 (avr_print_operand_address) [AVR_TINY]: Same.
18310 (avr_asm_init_sections) <readonly_data_section>: Only patch
18311 callback if avr_arch->flash_pm_offset = 0.
18312 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
18313 for rodata if avr_arch->flash_pm_offset != 0.
18314 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
18315 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
18316 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
18317 (m_options): Append opt_rcall.
18318 (m_dirnames): Append dir_rcall.
18319 * config/avr/t-multilib: Regenerate.
18321 * configure.ac [target=avr]: Check whether avrxmega3 default
18322 linker description file works as needed.
18323 * configure: Regenerate.
18324 * doc/avr-mmcu.texi: Regenerate.
18325 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
18326 <__AVR_ARCH__>: Document avrxmega3 and 103.
18327 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
18328 <__AVR_SHORT_CALLS__>: Document it.
18329 <__AVR_PM_BASE_ADDRESS__>: Document it.
18330 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
18331 (AVR Variable Attributes) <progmem>: Document this is
18332 not needed for avrxmega3.
18333 (AVR Named Address Spaces) <__flash>: Dito.
18335 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
18337 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
18339 2017-06-12 Doug Rupp <rupp@adacore.com>
18341 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
18342 Append vxworks-stdint.h to the tm_file list.
18343 * config/vxworks-stdint.h: New file.
18345 2017-06-12 Martin Liska <mliska@suse.cz>
18347 PR tree-optimization/81041
18348 * tree-profile.c (gimple_gen_ic_func_profiler):
18349 Create an extra BB in profile-generate
18350 (gimple_gen_time_profiler): Likewise.
18352 2017-06-12 Jakub Jelinek <jakub@redhat.com>
18354 PR tree-optimization/81003
18355 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
18356 (update_range_test): Use it instead of force_gimple_operand_gsi.
18358 2017-06-12 Richard Biener <rguenther@suse.de>
18360 PR tree-optimization/81053
18361 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
18362 with backedge value not defined in loop. Simplify def stmt
18365 2017-06-11 Tom de Vries <tom@codesourcery.com>
18368 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
18370 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
18371 nvptx_cannot_force_const_mem.
18373 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
18375 * opts.c (finish_options): Move test for flag_split_stack after
18376 it has been initialized.
18378 2017-06-11 Jason Merrill <jason@redhat.com>
18380 * tree.h (id_equal): New.
18381 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
18382 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
18383 instead of strcmp of IDENTIFIER_POINTER.
18385 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
18387 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
18388 (mark_all_inlined_calls_cdtor): Fix formating.
18389 (inline_transform): Rescale profile before inlining.
18391 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
18393 * cgraph.h (cgraph_edge::clone): Update prototype.
18394 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
18395 (cgraph_node::create_clone): Update.
18396 (cgraph_node::create_version_clone): Update.
18397 * tree-inline.c (copy_bb): Update.
18398 (expand_call_inline): Update.
18400 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
18402 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
18403 factored out from ...
18404 (rs6000_emit_prologue): ... here.
18406 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
18408 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
18409 factored out from ...
18410 (rs6000_emit_prologue): ... here.
18412 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
18414 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
18416 (handle_missing_profiles): Fix computation of tp_first_run.
18417 (counts_to_freqs): Do not touch freqs when count is 0.
18419 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
18421 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
18424 2017-06-10 Tom de Vries <tom@codesourcery.com>
18426 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
18427 attributes): Document signal effective target.
18429 2017-06-10 Tom de Vries <tom@codesourcery.com>
18431 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
18432 Document effective target stack_size.
18434 2017-06-09 David Malcolm <dmalcolm@redhat.com>
18436 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
18437 to the edit_context if they can be auto-applied.
18439 2017-06-9 Ian Lance Taylor <iant@golang.org>
18441 * opts.c (finish_options): If -fsplit-stack, disable implicit
18442 -forder-blocks-and-partition.
18443 * doc/invoke.texi (Optimize Options): Document that when using
18444 -fsplit-stack -forder-blocks-and-partition is not implicitly
18447 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
18449 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
18450 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
18451 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
18452 * builtins.def (abort, trap, unreachable): Declare cold.
18453 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
18454 * tree-core.h (ECF_COLD): New.
18455 * tree.c (set_call_expr_flags): Handle ECF_COLD.
18456 (build_common_builtin_nodes): Mark unreachable and abort as cold.
18458 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
18460 * predict.c (unlikely_executed_stmt_p): Cleanup.
18462 2017-06-09 Richard Biener <rguenther@suse.de>
18464 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
18465 model if the ref is always written to.
18467 2017-06-09 Tamar Christina <tamar.christina@arm.com>
18469 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
18471 2017-06-09 Tamar Christina <tamar.christina@arm.com>
18473 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
18476 2017-06-09 Tom de Vries <tom@codesourcery.com>
18479 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
18480 "target cannot support label values" when encountering LABEL_REF.
18482 2017-06-09 Martin Liska <mliska@suse.cz>
18484 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
18485 (gimple_gen_ic_func_profiler): Emit direct comparison
18486 of __gcov_indirect_call_callee with NULL.
18487 (gimple_gen_time_profiler): Change probability from
18488 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
18490 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
18492 * profile.c (edge_gcov_counts): Turn to pointer.
18493 (compute_branch_probabilities, compute_branch_probabilities): Update.
18494 (branch_prob): Do not clear edge_gcov_count.
18495 * profile.h (edge_gcov_counts): Turn to pointer.
18496 (edge_gcov_count): Update.
18498 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
18500 * gimple.h (gimple_check_failed): Mark cold.
18502 2017-06-09 Richard Biener <rguenther@suse.de>
18504 PR tree-optimization/66623
18505 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
18506 refactor check_reduction into two parts, properly computing
18507 whether we have to check reduction validity for outer loop
18510 2017-06-09 Richard Biener <rguenther@suse.de>
18512 PR tree-optimization/79483
18513 * graphite-scop-detection.c (order): New global.
18514 (get_order): Compute bb to order mapping that satisfies code
18515 generation constraints.
18516 (cmp_pbbs): New helper.
18517 (build_scops): Start domwalk at entry block, sort generated
18520 2017-06-09 Richard Biener <rguenther@suse.de>
18522 PR middle-end/81007
18523 * ipa-polymorphic-call.c
18524 (ipa_polymorphic_call_context::restrict_to_inner_class):
18525 Skip FIELD_DECLs with error_mark_node type.
18526 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
18529 2017-06-09 Martin Liska <mliska@suse.cz>
18531 * predict.c (struct branch_predictor): New struct.
18532 (test_prediction_value_range): New test.
18533 (predict_c_tests): New function.
18534 * selftest-run-tests.c (selftest::run_tests): Run the function.
18535 * selftest.h: Declare new tests.
18537 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
18540 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
18541 gen_add3_insn did not fail.
18542 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
18543 r0, construct that number in a temporary reg and add that reg to r0.
18544 If asked to put the result in r0 as well, fail.
18546 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
18548 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
18549 for early expansion of vec_eqv.
18551 2017-06-08 Jakub Jelinek <jakub@redhat.com>
18553 PR middle-end/81005
18554 * ubsan.c (instrument_null): Avoid pointless code temporary.
18555 (pass_ubsan::execute): Instrument aggregate arguments of calls.
18557 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
18561 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
18564 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
18567 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18569 * predict.c (unlikely_executed_edge_p): Move ahead.
18570 (probably_never_executed_edge_p): Use it.
18572 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
18574 PR middle-end/79988
18575 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
18576 gimple_call_builtin_p call.
18578 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18580 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
18581 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
18582 rtl_check_failed_type2, rtl_check_failed_code1,
18583 rtl_check_failed_code2, rtl_check_failed_code_mode,
18584 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
18585 rtvec_check_failed_bounds, rtl_check_failed_flag,
18586 _fatal_insn_not_found, _fatal_insn): Likewise.
18587 * tree.h (tree_contains_struct_check_failed,
18588 tree_check_failed, tree_not_check_failed,
18589 tree_class_check_failed, tree_range_check_failed,
18590 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
18591 tree_vec_elt_check_failed, phi_node_elt_check_failed,
18592 tree_operand_check_failed, omp_clause_check_failed,
18593 omp_clause_operand_check_failed, omp_clause_range_check_failed):
18596 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18598 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
18599 flag_branch_probabilities.
18600 * ipa-inline.c (edge_badness): Likewise.
18601 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
18602 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
18603 * predict.c (maybe_hot_frequency_p): Likewise.
18604 (probably_never_executed): Likewise.
18605 * sched-ebb.c (schedule_ebbs): Likewise.
18606 * sched-rgn.c (find_single_block_region): Likewise.
18607 * tracer.c (tail_duplicate): Likewise.
18609 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18611 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
18612 longer requires x_flag_profile_use.
18614 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18616 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
18617 instead of flag_reorder_blocks_and_partition.
18618 * dbxout.c (dbxout_function_end): Likewise.
18619 * dwarf2out.c (gen_subprogram_die): Likewise.
18620 * haifa-sched.c (sched_create_recovery_edges): Likewise.
18621 * hw-doloop.c (reorg_loops): Likewise.
18622 * varasm.c (assemble_start_function,
18623 assemble_end_function): Likewise.
18624 (decide_function_section): Do not check for
18625 flag_reorder_blocks_and_partition.
18627 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
18629 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
18631 (chkp_get_hard_register_fake_addr_expr): Ditto.
18632 (chkp_build_addr_expr): Add check for hard reg case.
18633 (chkp_parse_array_and_component_ref): Ditto.
18634 (chkp_find_bounds_1): Ditto.
18635 (chkp_process_stmt): Don't generate bounds store for
18638 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18640 * predict.c (maybe_hot_bb_p): Do not check profile status.
18641 (maybe_hot_edge_p): Likewise.
18642 (probably_never_executed): Check for zero counts even if profile
18644 (unlikely_executed_edge_p): New function.
18645 (unlikely_executed_stmt_p): New function.
18646 (unlikely_executed_bb_p): New function.
18647 (set_even_probabilities): Use unlikely predicates.
18648 (combine_predictions_for_bb): Likewise.
18649 (predict_paths_for_bb): Likewise.
18650 (predict_paths_leading_to_edge): Likewise.
18651 (determine_unlikely_bbs): New function.
18652 (estimate_bb_frequencies): Use it.
18653 (compute_function_frequency): Use zero counts even if profile is
18655 * profile-count.h: Fix typo.
18657 2017-08-08 Julia Koval <julia.koval@intel.com>
18659 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
18660 _mm512_mask_cvtsepi16_storeu_epi8,
18661 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
18662 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
18663 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
18664 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
18665 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
18666 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
18667 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
18668 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
18669 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
18670 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
18671 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
18672 __builtin_ia32_pmovuswb256mem_mask,
18673 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
18674 __builtin_ia32_pmovwb512mem_mask): New builtins.
18676 2017-08-08 Julia Koval <julia.koval@intel.com>
18678 PR target/73350,80862
18679 * config/i386/subst.md (round): Fix round pattern.
18680 * config/i386/i386.c (ix86_erase_embedded_rounding):
18681 Fix erasing rounding for the fixed pattern.
18683 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18685 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
18687 2017-06-08 Martin Liska <mliska@suse.cz>
18689 PR gcov-profile/80911
18690 * gcov.c (block_info::block_info): New constructor.
18692 2017-06-07 Carl Love <cel@us.ibm.com>
18694 * config/rs6000/rs6000-c: The return type of the following
18695 built-in functions was implemented as int not long long. Fix sign
18696 of return value for the unsigned version of vec_mulo and vec_mule.
18697 vector unsigned long long vec_bperm (vector unsigned long long,
18698 vector unsigned char)
18699 vector signed long long vec_mule (vector signed int,
18701 vector unsigned long long vec_mule (vector unsigned int,
18702 vector unsigned int)
18703 vector signed long long vec_mulo (vector signed int,
18705 vector unsigned long long vec_mulo (vector unsigned int,
18706 vector unsigned int)
18707 * doc/extend.texi: Fix the documentation for the built-in
18710 2017-06-07 Carl Love <cel@us.ibm.com>
18713 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
18716 2017-06-07 Carl Love <cel@us.ibm.com>
18718 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
18719 support, Generate doublehv for signed int/float for BE case only.
18721 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
18723 * doc/invoke.texi (mcx16): Rewrite.
18725 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18727 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
18728 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
18729 *mov<mode>_softfloat, and an anonymous splitter): Use
18730 nonimmediate_operand instead of rs6000_nonimmediate_operand.
18732 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18734 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
18736 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
18737 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
18738 (rs6000_debug_reg_global): Adjust.
18739 (rs6000_init_hard_regno_mode_ok): Adjust.
18740 (rs6000_dbx_register_number): Adjust.
18741 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
18742 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
18743 Remove SPE_ACC and SPEFSCR.
18744 (REG_ALLOC_ORDER): Ditto.
18745 (FRAME_POINTER_REGNUM): Change to 111.
18746 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
18747 (REG_CLASS_NAMES): Ditto.
18748 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
18749 (REGISTER_NAMES): Ditto.
18750 (ADDITIONAL_REG_NAMES): Ditto.
18751 (rs6000_reg_names): Ditto.
18752 * config/rs6000/rs6000.md: Renumber some register number
18755 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18757 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
18759 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
18760 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
18762 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
18763 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
18764 Delete the SPE high registers.
18765 (REG_ALLOC_ORDER): Ditto.
18766 (enum reg_class): Remove SPE_HIGH_REGS.
18767 (REG_CLASS_NAMES): Ditto.
18768 (REG_CLASS_CONTENTS): Delete the SPE high registers.
18769 (REGISTER_NAMES): Ditto.
18770 (rs6000_reg_names): Ditto.
18771 * doc/tm.texi.in: Remove SPE as example.
18772 * doc/tm.texi: Regenerate.
18774 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18776 * config/rs6000/8540.md (ppc8540_brinc): Delete.
18777 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
18778 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
18779 * config/rs6000/rs6000.md (type): Remove "brinc".
18781 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18783 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
18784 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
18785 * config/rs6000/linuxspe.h: Delete file.
18786 * config/rs6000/rs6000.md: Don't include spe.md.
18787 * config/rs6000/spe.h: Delete file.
18788 * config/rs6000/spe.md: Delete file.
18789 * config/rs6000/t-rs6000: Remove spe.md.
18791 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18793 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
18794 (reg_or_none500mem_operand): Delete.
18795 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
18796 instead of reg_or_none500mem_operand.
18798 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18800 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
18801 handling of SPE flags.
18802 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
18804 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18806 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
18808 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
18809 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
18810 paired_divv2sf3): Similar.
18811 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
18812 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
18813 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
18815 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
18816 Rename the paired_* instruction patterns.
18817 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
18819 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
18820 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
18821 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
18822 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
18823 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
18824 PAIRED_VECTOR_MODE.
18825 (struct machine_function): Delete field spe_insn_chain_scanned_p.
18826 (spe_func_has_64bit_regs_p): Delete.
18827 (spe_expand_predicate_builtin): Delete.
18828 (spe_expand_evsel_builtin): Delete.
18829 (TARGET_DWARF_REGISTER_SPAN): Do not define.
18830 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
18831 (invalid_e500_subreg): Delete.
18832 (rs6000_legitimize_address): Always force_reg op2 as well, for
18833 paired single memory accesses.
18834 (rs6000_member_type_forces_blk): Delete.
18835 (rs6000_spe_function_arg): Delete.
18836 (rs6000_expand_unop_builtin): Delete SPE handling.
18837 (rs6000_expand_binop_builtin): Ditto.
18838 (spe_expand_stv_builtin): Delete.
18839 (bdesc_2arg_spe): Delete.
18840 (spe_expand_builtin): Delete.
18841 (spe_expand_predicate_builtin): Delete.
18842 (spe_expand_evsel_builtin): Delete.
18843 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
18844 (spe_init_builtins): Delete.
18845 (spe_func_has_64bit_regs_p): Delete.
18846 (savres_routine_name): Delete "info" parameter. Adjust callers.
18847 (rs6000_emit_stack_reset): Ditto.
18848 (rs6000_dwarf_register_span): Delete.
18849 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
18850 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
18851 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
18853 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
18855 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
18856 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
18857 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
18858 mulv2sf3, divv2sf3): Delete expanders.
18860 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18862 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
18864 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18866 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
18867 * config/rs6000/rs6000.c: Ditto.
18869 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18871 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
18872 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
18873 comparison_operator.
18875 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18877 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
18878 * config/rs6000/rs6000.opt: Ditto.
18879 * config/rs6000/t-rtems: Ditto.
18881 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18883 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
18884 TARGET_E500_SINGLE by 0, simplify.
18885 * config/rs6000/rs6000.c: Ditto.
18886 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
18887 (spe_build_register_parallel): Delete.
18888 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
18889 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
18890 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
18891 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
18892 (E500_CONVERT): Delete.
18893 * config/rs6000/spe.md: Remove many patterns and all define_constants.
18895 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18897 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
18898 * config/rs6000/dfp.md: Ditto.
18899 (negdd2, *negdd2_fpr): Merge.
18900 (absdd2, *absdd2_fpr): Merge.
18901 (negtd2, *negtd2_fpr): Merge.
18902 (abstd2, *abstd2_fpr): Merge.
18903 * config/rs6000/e500.h: Delete file.
18904 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
18905 TARGET_FPRS by 1 and simplify.
18906 * config/rs6000/rs6000-c.c: Ditto.
18907 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
18908 TARGET_DF_SPE by 0.
18909 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
18911 * config/rs6000/rs6000.md: Ditto.
18912 (floatdidf2, *floatdidf2_fpr): Merge.
18913 (move_from_CR_gt_bit): Delete.
18914 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
18915 (E500_CR_IOR_COMPARE): Delete.
18916 (All patterns that require !TARGET_FPRS): Delete.
18917 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
18919 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18921 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
18923 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18925 * graphds.c (add_edge): Intitialize edge's attached data.
18926 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
18927 pointer parameter. Call pointed function on each edge during
18928 graph traversing. Skip traversing the edge when the function
18930 (graphds_dfs, graphds_scc): Ditto.
18931 (for_each_edge): New parameter. Pass the new parameter to callback
18933 * graphds.h (skip_edge_callback): New function pointer type.
18934 (graphds_dfs, graphds_scc): New function pointer parameter.
18935 (graphds_edge_callback, for_each_edge): New parameter.
18937 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18939 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
18940 out code checking if runtime alias check is possible to below ...
18941 Call the new function.
18942 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
18943 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
18945 2017-06-07 Marek Polacek <polacek@redhat.com>
18948 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
18949 TYPE_OVERFLOW_WRAPS check.
18951 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18953 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
18954 if versioning is required.
18955 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
18956 peeling with the check for versioning.
18958 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18960 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
18961 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
18962 Set true to new parameter if new ssa variable is defined.
18963 (vect_gen_vector_loop_niters): Refactor. Set range information
18964 for the new vector loop bound variable.
18965 (vect_do_peeling): Ditto.
18967 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18969 * tree-affine.c (ssa.h): Include header file.
18970 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
18971 has wrapping overflow behavior.
18973 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18975 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
18977 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18979 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
18980 (tree_to_aff_combination): ... here.
18982 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18984 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
18985 reg_pressure model function.
18986 (ivopts_global_cost_for_size): Delete.
18987 (determine_set_costs, iv_ca_recount_cost): Call new model function
18988 ivopts_estimate_reg_pressure.
18990 2017-06-07 Tamar Christina <tamar.christina@arm.com>
18992 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
18993 expensive than udiv. Remove floating point cases from mod.
18995 2017-06-07 Tamar Christina <tamar.christina@arm.com>
18997 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
18998 Increase idiv cost.
19000 2017-06-07 Tamar Christina <tamar.christina@arm.com>
19002 * config/aarch64/aarch64.md
19003 (copysignsf3): Fix mask generation.
19005 2017-06-07 Jakub Jelinek <jakub@redhat.com>
19007 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
19009 (class dump_manager): Add register_dumps method.
19010 * dumpfile.c: Include langhooks.h.
19011 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
19012 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
19013 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
19014 (dump_manager::dump_register): Start with 512 entries instead of 32.
19015 (dump_manager::register_dumps): New method.
19016 * toplev.c (general_init): Instead of invoking register_dumps
19017 langhook, invoke register_dumps method on the dump manager.
19018 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
19021 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
19023 * doc/md.texi: Clarify the restrictions on a define_insn condition.
19024 Say that # requires an associated define_split to exist, and that
19025 the define_split must be suitable for use after register allocation.
19027 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19029 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
19030 (compute_outgoing_frequencies): Also initialize zero counts.
19031 (find_many_sub_basic_blocks): Do not produce uninitialized profile
19032 around loops; preserve more of profile when nothing changes.
19034 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
19036 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
19038 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
19039 * config/arm/arm-cpu-cdata.h: Regenerate.
19040 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
19041 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
19042 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
19043 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
19044 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
19046 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
19048 * config/arm/t-rmprofile: Likewise.
19049 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
19051 2017-06-06 David S. Miller <davem@davemloft.net>
19054 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
19055 blockage if function uses alloca.
19057 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19059 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
19060 New "uid" fields to hold pretty-print IDs of group and ref.
19061 Memory references are now identified as <group_id>:<ref_id>
19062 instead of using [random] addresses.
19063 (dump_mem_details): Simplify, no functional change.
19064 (dump_mem_ref): Simplify and make output more concise.
19065 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
19066 (find_or_create_group): Initialize group uid.
19067 (record_ref): Initialize ref uid. Improve debug output.
19068 (prune_group_by_reuse, should_issue_prefetch_p,)
19069 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
19070 (mark_nontemporal_store, determine_loop_nest_reuse):
19071 Improve debug output.
19073 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19075 * dbgcnt.def (prefetch): New debug counter.
19076 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
19077 (schedule_prefetches): Stop issueing prefetches if debug counter
19080 2017-06-06 Tom de Vries <tom@codesourcery.com>
19082 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
19083 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
19085 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19087 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
19088 Use aarch64_reg_or_zero predicate for operand 4.
19089 (aarch64_compare_and_swap<mode> define_insn_and_split):
19090 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
19091 (aarch64_store_exclusive<mode>): Likewise for operand 2.
19093 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
19095 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
19096 (arm_compute_save_core_reg_mask): This.
19097 (thumb1_compute_save_reg_mask): Rename into ...
19098 (thumb1_compute_save_core_reg_mask): This.
19099 (arm_compute_save_reg0_reg12_mask): Adapt comment.
19100 (arm_compute_frame_layout): Likewise.
19102 2017-06-06 Richard Biener <rguenther@suse.de>
19104 PR tree-optimization/80974
19105 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
19106 keep or clear leaders SSA info.
19108 2017-06-06 Tom de Vries <tom@codesourcery.com>
19110 * config/nvptx/nvptx.c (split_mode_p): New function.
19111 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
19113 2017-06-06 Tom de Vries <tom@codesourcery.com>
19115 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
19117 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19120 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
19123 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19125 * shrink-wrap.c (handle_simple_exit): Update profile.
19126 (try_shrink_wrapping): Upate profile.
19128 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19130 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
19131 (tree_guess_outgoing_edge_probabilities): New.
19132 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
19133 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
19135 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19137 * ipa-split.c (split_function): Initialize return bb profile.
19139 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19141 * profile.c (compute_branch_probabilities): Also initialize
19142 EXIT_BLOCK profile.
19144 2017-06-06 Richard Biener <rguenther@suse.de>
19146 PR tree-optimization/80928
19147 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
19148 (vect_analyze_loop_operations): Properly guard analysis for
19150 (vect_transform_loop): Likewise.
19151 (vect_analyze_loop_2): Also reset SLP type on PHIs.
19152 (vect_model_induction_cost): Do not cost for pure SLP.
19153 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
19154 of induction in inner loop vectorization.
19155 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
19156 (vect_get_and_check_slp_defs): Handle vect_induction_def.
19157 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
19159 (vect_analyze_slp_cost_1): Cost induction.
19160 (vect_detect_hybrid_slp_stmts): Handle PHIs.
19161 (vect_get_slp_vect_defs): Likewise.
19162 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
19163 (vect_transform_stmt): Handle SLP reductions.
19164 * tree-vectorizer.h (vectorizable_induction): Adjust.
19166 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
19168 * config/rs6000/rs6000.c (make_resolver_func): Update
19169 init_lowered_empty_function call.
19171 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
19173 * doc/invoke.texi: Document the -fprofile-abs-path option.
19174 * common.opt (fprofile-abs-path): New option.
19175 * gcov-io.h (gcov_write_filename): Declare.
19176 * gcov-io.c (gcov_write_filename): New function.
19177 * coverage.c (coverage_begin_function): Use gcov_write_filename.
19178 * profile.c (output_location): Likewise.
19180 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
19182 * shring-wrap.c: Revert accidental commit.
19184 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
19186 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
19188 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
19190 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
19192 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
19193 profile in callgraph edge.
19194 * profile-count.h (apply_probability): If THIS is 0, then result is 0
19195 (apply_scale): Likewise.
19196 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
19197 Also scale profile when inlining function with zero profile.
19198 (initialize_cfun): Update exit block profile even when it is zero.
19199 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
19200 when profile is read.
19202 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
19204 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
19205 (CLONE_*): New constants to define the processors we can generate
19206 code for with the target_clone attribute.
19207 (rs6000_clone_map): New array to identify which clone processors
19208 the current program is running on.
19209 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
19210 target_clone attribute.
19211 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
19212 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
19213 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
19214 (cpu_expand_builtin): Add support for target_clone attribute.
19215 (rs6000_valid_attribute_p): Allow "default" attribute.
19216 (get_decl_name): New debug function to simplify printing the
19217 current function name in debugging statements.
19218 (rs6000_clone_priority): New functions to support the target_clone
19219 attribute, and be able to generate code to switch between ISA 2.05
19220 through ISA 3.0 (power6 through power9).
19221 (rs6000_compare_version_priority): Likewise.
19222 (rs6000_get_function_versions_dispatcher): Likewise.
19223 (make_resolver_func): Likewise.
19224 (add_condition_to_bb): Likewise.
19225 (dispatch_function_versions): Likewise.
19226 (rs6000_generate_version_dispatcher_body): Likewise.
19227 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
19228 (fusion_gpr_load_p): Fix a spacing issue.
19229 * doc/extend.texi (Common Function Attributes): Document that the
19230 PowerPC supports the target_clone attribute.
19232 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
19234 * config/arm/arm.h: explain F symbol found in description of ARM
19235 register allocation in its legend.
19237 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
19239 * config/mips/frame-header-opt.c: Include profile-count.h.
19240 * config/riscv/riscv.c: Include profile-count.h
19242 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
19244 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
19246 (sm_set_flag_if_changed): Add bbs field.
19247 (execute_sm_if_changed_flag_set): Pass BBS.
19248 (execute_sm): Update.
19250 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19252 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
19255 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19257 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
19258 (peephole2): New peephole2 to emit the above.
19259 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
19261 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19263 * config/aarch64/aarch64.c (define_peephole2 above
19264 *sub_<shift>_<mode>): New peephole.
19266 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
19268 * config/i386/i386.c (make_resolver_func): Update.
19269 * Makefile.in: Add profile-count.h and profile-count.o
19270 * auto-profile.c (afdo_indirect_call): Update to new API.
19271 (afdo_set_bb_count): Update.
19272 (afdo_propagate_edge): Update.
19273 (afdo_propagate_circuit): Update.
19274 (afdo_calculate_branch_prob): Update.
19275 (afdo_annotate_cfg): Update.
19276 * basic-block.h: Include profile-count.h
19277 (struct edge_def): Turn count to profile_count.
19278 (struct basic_block_def): Likewie.
19279 (REG_BR_PROB_BASE): Move to profile-count.h
19280 (RDIV): Move to profile-count.h
19281 * bb-reorder.c (max_entry_count): Turn to profile_count.
19282 (find_traces): Update.
19283 (rotate_loop):Update.
19284 (connect_traces):Update.
19285 (sanitize_hot_paths):Update.
19286 * bt-load.c (migrate_btr_defs): Update.
19287 * cfg.c (RDIV): Remove.
19288 (init_flow): Use alloc_block.
19289 (alloc_block): Uninitialize count.
19290 (unchecked_make_edge): Uninitialize count.
19291 (check_bb_profile): Update.
19292 (dump_edge_info): Update.
19293 (dump_bb_info): Update.
19294 (update_bb_profile_for_threading): Update.
19295 (scale_bbs_frequencies_int): Update.
19296 (scale_bbs_frequencies_gcov_type): Update.
19297 (scale_bbs_frequencies_profile_count): New.
19298 * cfg.h (update_bb_profile_for_threading): Update.
19299 (scale_bbs_frequencies_profile_count): Declare.
19300 * cfgbuild.c (compute_outgoing_frequencies): Update.
19301 (find_many_sub_basic_blocks): Update.
19302 * cfgcleanup.c (try_forward_edges): Update.
19303 (try_crossjump_to_edge): Update.
19304 * cfgexpand.c (expand_gimple_tailcall): Update.
19305 (construct_exit_block): Update.
19306 * cfghooks.c (verify_flow_info): Update.
19307 (dump_bb_for_graph): Update.
19308 (split_edge): Update.
19309 (make_forwarder_block): Update.
19310 (duplicate_block): Update.
19311 (account_profile_record): Update.
19312 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
19313 (get_estimated_loop_iterations): Update.
19314 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
19315 (single_likely_exit): Update.
19316 * cfgloopmanip.c (scale_loop_profile): Update.
19318 (set_zero_probability): Update.
19319 (lv_adjust_loop_entry_edge): Update.
19320 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
19321 (purge_dead_edges): Update.
19322 (rtl_account_profile_record): Update.
19323 * cgraph.c (cgraph_node::create): Uninitialize count.
19324 (symbol_table::create_edge): Uninitialize count.
19325 (cgraph_update_edges_for_call_stmt_node): Update.
19326 (cgraph_edge::dump_edge_flags): Update.
19327 (cgraph_node::dump): Update.
19328 (cgraph_edge::maybe_hot_p): Update.
19329 * cgraph.h: Include profile-count.h
19330 (create_clone), create_edge, create_indirect_edge): Update.
19331 (cgraph_node): Turn count to profile_count.
19332 (cgraph_edge0: Likewise.
19333 (make_speculative, clone): Update.
19334 (create_edge): Update.
19335 (init_lowered_empty_function): Update.
19336 * cgraphclones.c (cgraph_edge::clone): Update.
19337 (duplicate_thunk_for_node): Update.
19338 (cgraph_node::create_clone): Update.
19339 * cgraphunit.c (cgraph_node::analyze): Update.
19340 (cgraph_node::expand_thunk): Update.
19341 * final.c (dump_basic_block_info): Update.
19342 * gimple-streamer-in.c (input_bb): Update.
19343 * gimple-streamer-out.c (output_bb): Update.
19344 * graphite.c (print_global_statistics): Update.
19345 (print_graphite_scop_statistics): Update.
19346 * hsa-brig.c: Include basic-block.h.
19347 * hsa-dump.c: Include basic-block.h.
19348 * hsa-gen.c (T sum_slice): Update.
19349 (convert_switch_statements):Update.
19350 * hsa-regalloc.c: Include basic-block.h.
19351 * ipa-chkp.c (chkp_produce_thunks): Update.
19352 * ipa-cp.c (struct caller_statistics): Update.
19353 (init_caller_stats): Update.
19354 (gather_caller_stats): Update.
19355 (ipcp_cloning_candidate_p): Update.
19356 (good_cloning_opportunity_p): Update.
19357 (get_info_about_necessary_edges): Update.
19358 (dump_profile_updates): Update.
19359 (update_profiling_info): Update.
19360 (update_specialized_profile): Update.
19361 (perhaps_add_new_callers): Update.
19362 (decide_about_value): Update.
19363 (ipa_cp_c_finalize): Update.
19364 * ipa-devirt.c (struct odr_type_warn_count): Update.
19365 (struct decl_warn_count): Update.
19366 (struct final_warning_record): Update.
19367 (possible_polymorphic_call_targets): Update.
19368 (ipa_devirt): Update.
19369 * ipa-fnsummary.c (redirect_to_unreachable): Update.
19370 * ipa-icf.c (sem_function::merge): Update.
19371 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
19372 * ipa-inline.c (compute_uninlined_call_time): Update.
19373 (compute_inlined_call_time): Update.
19374 (want_inline_small_function_p): Update.
19375 (want_inline_self_recursive_call_p): Update.
19376 (edge_badness): Update.
19377 (lookup_recursive_calls): Update.
19378 (recursive_inlining): Update.
19379 (inline_small_functions): Update.
19380 (dump_overall_stats): Update.
19381 (dump_inline_stats): Update.
19382 * ipa-profile.c (ipa_profile_generate_summary): Update.
19383 (ipa_propagate_frequency): Update.
19384 (ipa_profile): Update.
19385 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
19386 * ipa-utils.c (ipa_merge_profiles): Update.
19387 * loop-doloop.c (doloop_modify): Update.
19388 * loop-unroll.c (report_unroll): Update.
19389 (unroll_loop_runtime_iterations): Update.
19390 * lto-cgraph.c (lto_output_edge): Update.
19391 (lto_output_node): Update.
19392 (input_node): Update.
19393 (input_edge): Update.
19394 (merge_profile_summaries): Update.
19395 * lto-streamer-in.c (input_cfg): Update.
19396 * lto-streamer-out.c (output_cfg): Update.
19397 * mcf.c (create_fixup_graph): Update.
19398 (adjust_cfg_counts): Update.
19399 (sum_edge_counts): Update.
19400 * modulo-sched.c (sms_schedule): Update.
19401 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
19402 * predict.c (maybe_hot_count_p): Update.
19403 (probably_never_executed): Update.
19404 (dump_prediction): Update.
19405 (combine_predictions_for_bb): Update.
19406 (propagate_freq): Update.
19407 (handle_missing_profiles): Update.
19408 (counts_to_freqs): Update.
19409 (rebuild_frequencies): Update.
19410 (force_edge_cold): Update.
19411 * predict.h: Include profile-count.h
19412 (maybe_hot_count_p, counts_to_freqs): UPdate.
19413 * print-rtl-function.c: Do not include cfg.h
19414 * print-rtl.c: Include basic-block.h
19415 * profile-count.c: New file.
19416 * profile-count.h: New file.
19417 * profile.c (is_edge_inconsistent): Update.
19418 (correct_negative_edge_counts): Update.
19419 (is_inconsistent): Update.
19420 (set_bb_counts): Update.
19421 (read_profile_edge_counts): Update.
19422 (compute_frequency_overlap): Update.
19423 (compute_branch_probabilities): Update; Initialize and deinitialize
19425 (branch_prob): Update.
19426 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
19427 (edge_gcov_count): New.
19428 (bb_gcov_count): New.
19429 * shrink-wrap.c (try_shrink_wrapping): Update.
19430 * tracer.c (better_p): Update.
19431 * trans-mem.c (expand_transaction): Update.
19432 (ipa_tm_insert_irr_call): Update.
19433 (ipa_tm_insert_gettmclone_call): Update.
19434 * tree-call-cdce.c: Update.
19435 * tree-cfg.c (gimple_duplicate_sese_region): Update.
19436 (gimple_duplicate_sese_tail): Update.
19437 (gimple_account_profile_record): Update.
19438 (execute_fixup_cfg): Update.
19439 * tree-inline.c (copy_bb): Update.
19440 (copy_edges_for_bb): Update.
19441 (initialize_cfun): Update.
19442 (freqs_to_counts): Update.
19443 (copy_cfg_body): Update.
19444 (expand_call_inline): Update.
19445 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
19446 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
19447 (try_unroll_loop_completely): Update.
19448 (try_peel_loop): Update.
19449 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19450 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
19451 * tree-ssa-loop-split.c (connect_loops): Update.
19452 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
19453 * tree-ssa-reassoc.c (branch_fixup): Update.
19454 * tree-ssa-tail-merge.c (replace_block_by): Update.
19455 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
19456 (compute_path_counts): Update.
19457 (update_profile): Update.
19458 (recompute_probabilities): Update.
19459 (update_joiner_offpath_counts): Update.
19460 (estimated_freqs_path): Update.
19461 (freqs_to_counts_path): Update.
19462 (clear_counts_path): Update.
19463 (ssa_fix_duplicate_block_edges): Update.
19464 (duplicate_thread_path): Update.
19465 * tree-switch-conversion.c (case_bit_test_cmp): Update.
19466 (struct switch_conv_info): Update.
19467 * tree-tailcall.c (decrease_profile): Update.
19468 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
19469 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19470 * value-prof.c (check_counter): Update.
19471 (gimple_divmod_fixed_value): Update.
19472 (gimple_mod_pow2): Update.
19473 (gimple_mod_subtract): Update.
19474 (gimple_ic_transform): Update.
19475 (gimple_stringop_fixed_value): Update.
19476 * value-prof.h (gimple_ic): Update.
19478 2017-06-02 Carl Love <cel@us.ibm.com>
19480 * config/rs6000/rs6000-c: Add support for built-in functions
19481 vector double vec_doublee (vector signed int);
19482 vector double vec_doublee (vector unsigned int);
19483 vector double vec_doublee (vector float);
19484 vector double vec_doubleh (vector signed int);
19485 vector double vec_doubleh (vector unsigned int);
19486 vector double vec_doubleh (vector float);
19487 vector double vec_doublel (vector signed int);
19488 vector double vec_doublel (vector unsigned int);
19489 vector double vec_doublel (vector float);
19490 vector double vec_doubleo (vector signed int);
19491 vector double vec_doubleo (vector unsigned int);
19492 vector double vec_doubleo (vector float);.
19493 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
19494 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
19496 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
19497 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
19498 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
19500 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
19501 vec_doublel, vec_doubleh.
19502 * doc/extend.texi: Update the built-in documentation file for the
19503 new built-in functions.
19505 2017-06-02 David Malcolm <dmalcolm@redhat.com>
19508 * ipa-inline-analysis.c (free_growth_caches): Set
19509 edge_removal_hook_holder to NULL after removing it.
19511 2017-06-02 Sudakshina Das <sudi.das@arm.com>
19513 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
19514 comparision with zero.
19516 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
19517 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
19518 for early expansion of vec_min and vec_max builtins.
19519 (builtin_function_type): Add min/max unsigned variants to those
19520 identified as having unsigned arguments.
19522 2017-06-02 Olivier Hainque <hainque@adacore.com>
19524 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
19526 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19528 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
19529 Use VALL_F16 iterator rather than VALL.
19531 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19533 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
19534 Emit CBNZ inside loop when doing a strong exchange and comparing
19535 against zero. Generate the CC flags after the loop.
19537 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
19539 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
19540 (dl_section_ref): New.
19541 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
19542 On AIX, append an expression to subtract the size of the
19543 section length to dl_section_ref.
19545 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
19547 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
19548 for early expansion of vector absolute builtins.
19550 2017-06-02 Richard Biener <rguenther@suse.de>
19552 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
19553 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
19555 2017-06-02 Richard Biener <rguenther@suse.de>
19557 PR tree-optimization/80948
19558 * tree-tailcall.c (find_tail_calls): Track stmts to move in
19559 stmt order as well.
19561 2017-06-02 Richard Biener <rguenther@suse.de>
19563 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
19565 * tree-vect-stmts.c (process_use): Do not mark backedge defs
19566 for inductions as relevant.
19568 2017-06-02 Richard Biener <rguenther@suse.de>
19570 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
19571 (vectorizable_induction): ... this. Remove dead code.
19573 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
19575 * builtins. (expand_builtin_alloca): Remove second parameter and
19576 infer its value from the first parameter instead.
19577 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
19579 2017-06-02 Jakub Jelinek <jakub@redhat.com>
19581 PR rtl-optimization/80903
19582 * loop-doloop.c (add_test): Unshare sequence.
19584 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
19586 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
19588 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
19590 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
19592 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
19593 xlogue_layout::get_instance, logue_layout::xlogue_layout,
19594 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
19595 (xlogue_layout::get_stub_rtx): Make static.
19596 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
19597 (xlogue_layout::compute_stub_managed_regs): Rename to...
19598 (xlogue_layout::count_stub_managed_regs): ...this.
19599 (xlogue_layout::is_stub_managed_reg): New function.
19600 (xlogue_layout::m_stub_names): Rename to...
19601 (xlogue_layout::s_stub_names): ...this, make static.
19602 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
19603 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
19604 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
19605 xlogue_layout::s_stub_names): Instantiate statics.
19606 (stub_managed_regs): Remove.
19607 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
19608 (disable_call_ms2sysv_xlogues): Rename to...
19609 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
19610 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
19612 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
19613 change after reload_completed.
19614 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
19616 (ix86_expand_prologue): Likewise.
19617 (ix86_expand_epilogue): Likewise.
19618 (ix86_expand_split_stack_prologue): Likewise.
19619 (ix86_compute_frame_layout): Remove frame parameter ...
19620 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
19621 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
19623 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
19624 (ix86_frame): Move from here ...
19625 * config/i386/i386.h (ix86_frame): ... to here.
19626 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
19627 complete ix86_frame data structure instead. Remove some_ld_name.
19629 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
19631 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
19632 symbols that hold a DECL_VALUE_EXPR.
19634 2017-06-01 Martin Jambor <mjambor@suse.cz>
19636 PR tree-optimization/80898
19637 * tree-sra.c (process_subtree_disqualification): Removed.
19638 (disqualify_candidate): Do not acll
19639 process_subtree_disqualification.
19640 (subtree_mark_written_and_enqueue): New function.
19641 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
19642 RHS has been disqualified and re-queue LHS if necessary. Apart
19643 from that, ignore disqualified RHS.
19645 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19647 * config/s390/s390.c (s390_emit_epilogue): Disable early return
19648 address fetch for z10 or later.
19650 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19652 * config/arc/arc.md (tst_movb): Add guard when splitting.
19654 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19656 * config/arc/arc.c (arc_can_eliminate): Test against
19657 arc_frame_pointer_needed.
19659 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19661 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
19662 to prevent store reordering.
19663 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
19664 (type): Add block type.
19665 (stack_tie): Define special instruction to be used in
19668 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19670 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
19671 constraint. It is not valid for the pattern.
19672 (noncommutative_binary_comparison): Likewise.
19674 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19676 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
19679 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19681 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
19682 be used by the reg-alloc.
19684 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19686 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
19687 reg-alloc when having mul64 or mul32x16 instructions.
19688 (mulsidi3): Likewise.
19689 (umulsidi3): Likewise.
19690 (mulsi32x16): New pattern.
19691 (mulsi64): Likewise.
19692 (mulsidi64): Likewise.
19693 (umulsidi64): Likewise.
19694 (MUL32x16_REG): Define.
19695 (mul64_600): Use MUL32x16_REG.
19696 (mac64_600): Likewise.
19697 (umul64_600): Likewise.
19698 (umac64_600): Likewise.
19700 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19702 * config/arc/arc.md (mulsi3_700): Make it commutative.
19704 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
19706 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
19708 (*sign_extendsidi2_insn): Likewise for movstosw.
19710 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
19712 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
19713 the type of the input discriminant value. Convert the
19714 discriminant value of signedness vary.
19716 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
19718 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
19719 Add to -Wall section.
19721 2017-06-01 Richard Biener <rguenther@suse.de>
19723 PR middle-end/66313
19724 * fold-const.c (fold_plusminus_mult_expr): If the factored
19725 factor may be zero use a wrapping type for the inner operation.
19726 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
19727 and handle moved defs.
19728 (process_assignment): Properly guard the unary op case. Return a
19729 tri-state indicating that moving the stmt before the call may allow
19730 to continue. Pass through to_move.
19731 (find_tail_calls): Handle moving unrelated defs before
19734 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
19737 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
19738 splitter result in the canonical way.
19740 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
19742 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
19743 also for 32bit target. Update insn attributes.
19744 (zero-extendsidi2 splitter): Allow all registers for operand 1.
19746 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
19748 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
19749 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
19750 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
19751 (_mm_maskz_min_ss): New intrinsics.
19753 2017-05-31 Martin Liska <mliska@suse.cz>
19755 * tree-vect-loop.c (vect_create_epilog_for_reduction):
19756 Change comment style to one we normally use.
19757 (vectorizable_reduction): Likewise.
19758 (vectorizable_induction): Likewise.
19759 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
19760 (vectorizable_call): Likewise.
19761 (vectorizable_simd_clone_call): Likewise.
19762 (vectorizable_conversion): Likewise.
19763 (vectorizable_assignment): Likewise.
19764 (vectorizable_shift): Likewise.
19765 (vectorizable_operation): Likewise.
19766 (vectorizable_store): Likewise.
19767 (vectorizable_load): Likewise.
19768 * tree-vectorizer.h: Likewise.
19770 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
19772 * passes.c (emergency_dump_function): New.
19773 * tree-pass.h (emergency_dump_function): Declare.
19774 * plugin.c (plugins_internal_error_function): Remove.
19775 * plugin.h (plugins_internal_error_function): Remove declaration.
19776 * toplev.c (internal_error_function): New static function. Use it...
19777 (general_init): ...here.
19779 2017-05-31 Graham Markall <graham.markall@embecosm.com>
19781 * config/arc/arc.c (arc_print_operand): Handle constant operands.
19782 (arc_rtx_costs): Add costs for new patterns.
19783 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
19784 * config/arc/predicates.md: Add _1_2_3_operand predicate.
19786 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
19788 * tree-ssa-strlen.c (get_next_strinfo): New function.
19789 (get_stridx_plus_constant): Use it.
19790 (zero_length_string): Likewise.
19791 (adjust_related_strinfos): Likewise.
19792 (adjust_last_stmt): Likewise.
19794 2017-05-31 Richard Biener <rguenther@suse.de>
19797 * config/i386/i386.c (ix86_expand_builtin): Remove assert
19798 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
19800 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
19802 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
19803 loop_vinfo argument and use of dependence distance vectors.
19804 Check instead whether the two references differ only in their
19805 initial value and assume that they have the same alignment if the
19806 difference is a multiple of the vector alignment.
19807 (vect_analyze_data_refs_alignment): Update call accordingly.
19809 2017-05-31 Martin Liska <mliska@suse.cz>
19812 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
19814 2017-05-31 Bin Cheng <bin.cheng@arm.com>
19816 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
19817 (create_intersect_range_checks): Move from ...
19818 * tree-data-ref.c (create_intersect_range_checks_index)
19819 (create_intersect_range_checks): ... to here.
19820 (create_runtime_alias_checks): New function factored from ...
19821 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
19822 here. Call above function.
19823 * tree-data-ref.h (create_runtime_alias_checks): New function.
19825 2017-05-31 Bin Cheng <bin.cheng@arm.com>
19827 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
19828 segment length for dr_b and compute it in wide_int.
19830 2017-05-31 Richard Biener <rguenther@suse.de>
19832 PR tree-optimization/80906
19833 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
19834 and pass through iv_map.
19835 (copy_bb_and_scalar_dependences): Adjust.
19836 (translate_pending_phi_nodes): Likewise.
19837 (copy_loop_close_phi_args): Handle code-generating IVs instead
19840 2017-05-30 David Malcolm <dmalcolm@redhat.com>
19842 * diagnostic-color.c (color_dict): Add "type-diff".
19843 (parse_gcc_colors): Update comment.
19844 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
19845 -fdiagnostics-show-template-tree and -fno-elide-type.
19846 (GCC_COLORS): Add type-diff to example.
19848 (-fdiagnostics-show-template-tree): New.
19849 (-fno-elide-type): New.
19850 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
19851 the pp_format_decoder callback. Call any m_format_postprocessor's
19853 (pretty_printer::pretty_printer): Initialize
19854 m_format_postprocessor.
19855 (pretty_printer::~pretty_printer): Delete any
19856 m_format_postprocessor.
19857 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
19858 (class format_postprocessor): New class.
19859 (struct pretty_printer::format_decoder): Document the new parameters.
19860 (struct pretty_printer::m_format_postprocessor): New field.
19861 * tree-diagnostic.c (default_tree_printer): Update for new
19862 bool and const char ** params.
19863 * tree-diagnostic.h (default_tree_printer): Likewise.
19865 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
19867 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
19868 (lwa_operand): Delete rs6000_gen_cell_microcode test.
19869 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
19870 rs6000_gen_cell_microcode code.
19871 (rs6000_final_prescan_insn): Delete.
19872 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
19873 "warn-cell-microcode" entries.
19874 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
19875 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
19876 throughout. Change cc_reg_not_micro_cr0_operand to
19877 cc_reg_not_cr0_operand throughout.
19878 (*extendhi<mode>2_noload): Delete.
19879 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
19880 (mwarn-cell-microcode): Delete.
19881 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
19882 -mgen-cell-microcode and -mwarn-cell-microcode.
19884 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
19887 * config/i386/constraints.md (Yd): New constraint.
19889 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
19890 and (?Yd, r) alternatives. Update insn attributes.
19891 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
19892 and (?*Yd, r) alternatives. Update insn attributes.
19893 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
19895 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
19897 * gimplify.c (gimplify_modify_expr): Don't create a
19898 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
19901 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
19903 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
19905 2017-05-30 Richard Biener <rguenther@suse.de>
19907 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
19908 and reduc_def fields.
19909 (STMT_VINFO_REDUC_TYPE): New define.
19910 (STMT_VINFO_REDUC_DEF): Likewise.
19911 (vect_force_simple_reduction): Adjust prototype.
19912 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
19913 (vect_is_simple_reduction): Remove check_reduction argument.
19914 (vect_force_simple_reduction): Adjust and set
19915 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
19916 (vectorizable_reduction): Do not re-do reduction analysis
19917 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
19918 * tree-parloops.c (gather_scalar_reductions): Adjust.
19920 2017-05-30 Richard Biener <rguenther@suse.de>
19922 PR middle-end/80901
19923 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
19926 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
19928 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
19929 Introduce unknown_misalignment parameter and remove vf.
19930 (vect_peeling_hash_get_lowest_cost):
19931 Pass unknown_misalignment parameter.
19932 (vect_enhance_data_refs_alignment):
19933 Fix unsupportable data ref treatment.
19935 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
19937 * tree-vect-data-refs.c (vect_get_data_access_cost):
19938 Workaround for SLP handling.
19939 (vect_enhance_data_refs_alignment):
19940 Compute costs for doing no peeling at all, compare to the best
19941 peeling costs so far and avoid peeling if cheaper.
19943 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
19945 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
19946 Return peeling info and set costs to zero for unlimited cost
19948 (vect_enhance_data_refs_alignment): Also inspect all datarefs
19949 with unknown misalignment. Compute and costs for unknown
19950 misalignment, compare them to the costs for known misalignment
19951 and choose the cheapest for peeling.
19953 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
19955 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
19956 (vect_get_peeling_costs_all_drs): Create function.
19957 (vect_peeling_hash_get_lowest_cost):
19958 Use vect_get_peeling_costs_all_drs.
19959 (vect_peeling_supportable): Create function.
19960 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
19962 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
19964 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
19965 DR_HAS_NEGATIVE_STEP.
19966 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
19967 (vect_enhance_data_refs_alignment): Use.
19968 (vect_duplicate_ssa_name_ptr_info): Use.
19969 * tree-vectorizer.h (dr_misalignment): Use.
19970 (known_alignment_for_access_p): Use.
19972 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19975 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
19977 (has_section_name): New function.
19979 2017-05-30 Martin Liska <mliska@suse.cz>
19982 * auto-profile.c (get_function_decl_from_block): Fix
19985 2017-05-30 Richard Biener <rguenther@suse.de>
19987 PR middle-end/80876
19988 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
19990 2017-05-30 Martin Liska <mliska@suse.cz>
19992 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
19993 * dumpfile.h (struct dump_file_info): Remove ctors.
19995 2017-05-30 Martin Liska <mliska@suse.cz>
19997 * predict.def: Fix GNU coding style.
19999 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
20001 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
20002 Mark 'to' argument with ATTRIBUTE_UNUSED.
20004 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
20006 * config/xtensa/xtensa.c (xtensa_emit_call): Use
20007 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
20008 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
20011 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
20013 * doc/install.texi (Options specification): Restore entry of
20014 --enable-sjlj-exceptions.
20016 2017-05-27 Michael Eager <eager@eagercon.com>
20019 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
20021 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
20023 * config/microblaze/microblaze.h
20024 (FIXED_REGISTERS): Update in macro.
20025 (CALL_USED_REGISTERS): Update in macro.
20027 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
20029 * doc/install.texi: Add links to macOS binary distributions.
20031 2017-05-27 Jakub Jelinek <jakub@redhat.com>
20035 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
20037 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
20039 2017-05-26 Martin Liska <mliska@suse.cz>
20041 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
20043 2017-05-26 Martin Liska <mliska@suse.cz>
20045 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
20046 always leading ';; '.
20047 (dump_bb_info): Likewise.
20048 (brief_dump_cfg): Likewise.
20049 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
20050 * dumpfile.c: Remove usage of TDF_VERBOSE.
20051 * dumpfile.h (enum dump_kind): Likewise.
20052 (dump_gimple_bb_header): Do not use TDF_COMMENT.
20053 * print-tree.c (debug_verbose): Remove.
20054 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
20055 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
20056 * tree-diagnostic.c (default_tree_printer): Replace
20057 TDF_DIAGNOSTIC with TDF_SLIM.
20059 2017-05-26 Bin Cheng <bin.cheng@arm.com>
20061 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
20062 in parameter loop, rather than loop_vinfo.
20063 (create_intersect_range_checks): Ditto.
20064 (vect_create_cond_for_alias_checks): Update call to above functions.
20066 2017-05-26 Bin Cheng <bin.cheng@arm.com>
20068 PR tree-optimization/80815
20069 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
20070 for merging runtime alias checks. Handle negative DR_STEPs.
20072 2017-05-26 Bin Cheng <bin.cheng@arm.com>
20074 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
20076 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
20077 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
20078 out code pruning runtime alias checks.
20079 * tree-data-ref.c (prune_runtime_alias_test_list): New function
20080 factored out from above.
20081 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
20083 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
20085 (prune_runtime_alias_test_list): New decalaration.
20087 2017-05-26 Bin Cheng <bin.cheng@arm.com>
20089 * tree-vect-data-refs.c (compare_tree): Rename and move ...
20090 * tree-data-ref.c (data_ref_compare_tree): ... to here.
20091 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
20092 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
20093 (operator==, comp_dr_with_seg_len_pair): Ditto.
20094 (vect_prune_runtime_alias_test_list): Ditto.
20096 2017-05-26 Martin Liska <mliska@suse.cz>
20099 * params.def: Bound partial-inlining-entry-probability param.
20101 2017-05-26 Marek Polacek <polacek@redhat.com>
20104 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
20107 2017-05-26 Richard Biener <rguenther@suse.de>
20109 PR tree-optimization/80842
20110 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
20113 2017-05-26 Richard Biener <rguenther@suse.de>
20115 PR tree-optimization/80844
20116 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
20118 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
20120 * doc/md.texi (Machine Constraints): Update x86 family
20121 machine constraints section to match 'config/i386/constraints.md'.
20123 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
20125 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
20127 2017-05-25 Nathan Sidwell <nathan@acm.org>
20129 * doc/invoke.texi (--enable-languages): Update documentation.
20131 2017-05-25 Martin Liska <mliska@suse.cz>
20133 * dumpfile.c: Add TDF_FOLDING.
20134 * dumpfile.h (enum dump_kind): Likewise.
20135 * genmatch.c (dt_simplify::gen_1): Use it.
20137 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
20139 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
20141 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
20143 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
20144 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
20146 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
20148 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
20149 * match.pd (X == C): Rewrite it here.
20150 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
20151 with_certain_nonzero_bits2): New predicates.
20152 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
20154 2017-05-24 Nathan Sidwell <nathan@acm.org>
20156 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
20159 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
20162 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
20164 * config/powerpcspe: New port. Files are copied from the rs6000
20165 port, with "rs6000" in filenames replaced by "powerpcspe".
20167 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
20169 PR rtl-optimization/80754
20170 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
20172 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
20174 * config/sparc/sparc.md (length): Return the correct value for -mflat
20175 sibcalls to match output_sibcall.
20177 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
20181 * config/rs6000/rs6000.c (struct machine_function): Add new field
20183 (rs6000_get_separate_components): Init that field, use it.
20184 (rs6000_components_for_bb): Use the field.
20186 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
20188 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
20190 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
20192 PR middle-end/80823
20193 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
20195 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20198 * config/s390/s390.c (s390_check_qrst_address): Check incoming
20199 address against address_operand predicate.
20200 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
20202 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
20204 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
20205 parameters passed indirectly.
20207 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
20209 * config/i386/i386.md (*movdi_internal): Remove SSE4
20210 alternative 18 (?r, *v). Update insn attributes.
20211 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
20212 Update insn attributes.
20213 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
20214 Update insn attributes.
20215 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
20216 alternative 1 (r, v). Remove isa attribute.
20217 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
20218 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
20219 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
20221 2017-05-23 Tom de Vries <tom@codesourcery.com>
20223 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
20226 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
20228 * cgraphunit.c (symbol_table::process_new_functions): Update.
20229 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
20230 (inline_generate_summary): Rename to ...
20231 (ipa_fn_summary_generate): ... this one.
20232 (inline_read_summary): Rename to ...
20233 (ipa_fn_summary_read): ... this one.
20234 (inline_write_summary): Rename to ...
20235 (ipa_fn_summary_write): ... this one.
20236 (inline_free_summary): Rename to ...
20237 (ipa_free_fn_summary): ... this one.
20238 (pass_data_local_fn_summary, pass_local_fn_summary,
20239 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
20240 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
20241 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
20242 make_pass_ipa_fn_summary): New.
20243 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
20244 inline_write_summary, inline_free_summary): Remove.
20245 (ipa_free_fn_summary) : New.
20246 * ipa-inline.c (ipa_inline): Update.
20247 (pass_ipa_inline): Do not generate summaries.
20248 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
20250 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
20251 and add pass_ipa_fn_summary.
20252 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
20254 (make_pass_inline_parameters): Remove.
20256 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
20258 * omp-low.c (struct omp_context): Remove "default_kind" member.
20261 * omp-offload.c (execute_oacc_device_lower): Remove the
20262 parallelism dimensions function attributes for unparallelized
20263 OpenACC kernels constructs.
20265 2017-05-23 Martin Liska <mliska@suse.cz>
20267 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
20269 (cgraph_edge::make_speculative): Likewise.
20270 (cgraph_edge::resolve_speculation): Likewise.
20271 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
20272 (cgraph_node::dump): Likewise.
20273 * cgraph.h: Likewise.
20274 * cgraphunit.c (analyze_functions): Likewise.
20275 (symbol_table::compile): Likewise.
20276 * ipa-cp.c (print_all_lattices): Likewise.
20277 (determine_versionability): Likewise.
20278 (initialize_node_lattices): Likewise.
20279 (ipcp_verify_propagated_values): Likewise.
20280 (estimate_local_effects): Likewise.
20281 (update_profiling_info): Likewise.
20282 (create_specialized_node): Likewise.
20283 (perhaps_add_new_callers): Likewise.
20284 (decide_about_value): Likewise.
20285 (decide_whether_version_node): Likewise.
20286 (identify_dead_nodes): Likewise.
20287 (ipcp_store_bits_results): Likewise.
20288 * ipa-devirt.c (dump_targets): Likewise.
20289 (ipa_devirt): Likewise.
20290 * ipa-icf.c (sem_item::dump): Likewise.
20291 (sem_function::equals): Likewise.
20292 (sem_variable::equals): Likewise.
20293 (sem_item_optimizer::read_section): Likewise.
20294 (sem_item_optimizer::execute): Likewise.
20295 (congruence_class::dump): Likewise.
20296 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
20297 (dump_inline_summary): Likewise.
20298 (estimate_node_size_and_time): Likewise.
20299 (inline_analyze_function): Likewise.
20300 * ipa-inline-transform.c (inline_call): Likewise.
20301 * ipa-inline.c (report_inline_failed_reason): Likewise.
20302 (want_early_inline_function_p): Likewise.
20303 (edge_badness): Likewise.
20304 (update_edge_key): Likewise.
20305 (inline_small_functions): Likewise.
20306 * ipa-profile.c (ipa_profile): Likewise.
20307 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
20308 (ipa_make_edge_direct_to_target): Likewise.
20309 (remove_described_reference): Likewise.
20310 (ipa_impossible_devirt_target): Likewise.
20311 (propagate_controlled_uses): Likewise.
20312 (ipa_print_node_params): Likewise.
20313 (ipcp_transform_function): Likewise.
20314 * ipa-pure-const.c (pure_const_read_summary): Likewise.
20315 (propagate_pure_const): Likewise.
20316 * ipa-reference.c (generate_summary): Likewise.
20317 (read_write_all_from_decl): Likewise.
20318 (propagate): Likewise.
20319 (ipa_reference_read_optimization_summary): Likewise.
20320 * ipa-utils.c (ipa_merge_profiles): Likewise.
20321 * ipa.c (walk_polymorphic_call_targets): Likewise.
20322 (symbol_table::remove_unreachable_nodes): Likewise.
20323 (ipa_single_use): Likewise.
20324 * passes.c (execute_todo): Likewise.
20325 * predict.c (drop_profile): Likewise.
20326 * symtab.c (symtab_node::get_dump_name): New function.
20327 (symtab_node::dump_name): Likewise.
20328 (symtab_node::dump_asm_name): Likewise.
20329 (symtab_node::dump_references): Likewise.
20330 (symtab_node::dump_referring): Likewise.
20331 (symtab_node::dump_base): Likewise.
20332 (symtab_node::debug_symtab): Likewise.
20333 * tree-sra.c (convert_callers_for_node): Likewise.
20334 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
20335 * value-prof.c (init_node_map): Likewise.
20337 2017-05-23 Martin Liska <mliska@suse.cz>
20339 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
20340 and symtab_node::debug_symtab to symbol_table::debug.
20341 * cgraphunit.c (analyze_functions): Use the renamed function.
20342 (symbol_table::compile): Likewise.
20343 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
20344 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
20345 * passes.c (execute_todo): Likewise.
20346 * symtab.c (symbol_table::dump): New function.
20347 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
20349 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
20351 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
20352 that nonconst implies exec.
20354 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
20356 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
20357 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
20358 (inline_edge_summary_vec): Turn into ...
20359 (ipa_call_summaries): ... this one.
20360 (redirect_to_unreachable, edge_set_predicate,
20361 evaluate_properties_for_edge, inline_summary_alloc,
20362 reset_ipa_call_summary, reset_inline_summary,
20363 inline_summary_t::duplicate): Update.
20364 (inline_edge_duplication_hook): Turn to ...
20365 (ipa_call_summary_t::duplicate): ... this one.
20366 (inline_edge_removal_hook): Turn to ...
20367 (ipa_call_summary_t::remove): ... this one.
20368 (dump_inline_edge_summary): Turn to ...
20369 (dump_ipa_call_summary): ... this one.
20370 (estimate_function_body_sizes): Update.
20371 (inline_update_callee_summaries): Update.
20372 (remap_edge_change_prob): Update.
20373 (remap_edge_summaries): Update.
20374 (inline_merge_summary): Update.
20375 (do_estimate_edge_time): Update.
20376 (inline_generate_summary): Update.
20377 (inline_read_section): Update.
20378 (inline_read_summary): Update.
20379 (inline_free_summary): Update.
20380 * ipa-inline.c (can_inline_edge_p): Update.
20381 (compute_inlined_call_time): Update.
20382 (want_inline_small_function_p): Update.
20383 (edge_badness): Update.
20384 (early_inliner): Update.
20385 * ipa-inline.h (inline_edge_summary): Turn to ...
20386 (ipa_call_summary): ... this one.
20387 (ipa_call_summary_t): New class.
20388 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
20389 (ipa_call_summaries): New.
20390 (inline_edge_summary): Remove.
20391 (estimate_edge_growth): Update.
20392 * ipa-profile.c (ipa_propagate_frequency_1): Update.
20393 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
20394 * ipa-split.c (execute_split_functions): Update.
20395 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
20397 2017-05-23 Tom de Vries <tom@codesourcery.com>
20399 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
20400 attributes): Document rdrand effective target.
20402 2017-05-23 Tom de Vries <tom@codesourcery.com>
20404 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
20405 attributes): Sort alphabetically.
20407 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
20409 * config/avr/genmultilib.awk: Use gsub instead of gensub.
20411 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
20414 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
20415 V2DF/V2DI splat into two separate patterns, one that handles
20416 registers, and the other that only handles memory. Drop support
20417 for splatting from a GPR on ISA 2.07 and then splitting the
20418 splat into direct move and splat.
20419 (vsx_splat_<mode>_reg): Likewise.
20420 (vsx_splat_<mode>_mem): Likewise.
20422 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
20424 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
20426 2017-05-22 Jakub Jelinek <jakub@redhat.com>
20428 PR middle-end/80809
20429 * omp-low.c (finish_taskreg_remap): New function.
20430 (finish_taskreg_scan): If unit size of ctx->record_type
20431 is non-constant, unshare the size expression and replace
20432 decls in it with possible outer var refs.
20434 PR middle-end/80809
20435 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
20436 GOVD_SHARED rather than GOVD_PRIVATE with it.
20437 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
20438 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
20440 PR middle-end/80853
20441 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
20442 as last argument to build_outer_var_ref for pointer bases of array
20443 section reductions.
20445 2017-05-19 Martin Sebor <msebor@redhat.com>
20447 * print-tree.c (print_node): Print DECL_READ_P flag.
20449 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
20451 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
20452 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
20453 * cgraph.c: Likewise.
20454 * cgraphunit.c: Likewise.
20455 * gengtype.c: Likewise.
20456 * ipa-cp.c: Likewise.
20457 * ipa-devirt.c: Likewise.
20458 * ipa-icf.c: Likewise.
20459 * ipa-predicate.c: Likewise.
20460 * ipa-profile.c: Likewise.
20461 * ipa-prop.c: Likewise.
20462 * ipa-split.c: Likewise.
20464 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
20465 edge_predicate_pool, dump_inline_hints,
20466 inline_summary::account_size_time, redirect_to_unreachable,
20467 edge_set_predicate, set_hint_predicate,
20468 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
20469 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
20470 inline_summary_t::remove, remap_hint_predicate_after_duplication,
20471 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
20472 ipa_call_summary_t::remove, initialize_growth_caches,
20473 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
20474 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
20475 mark_modified, unmodified_parm_1, unmodified_parm,
20476 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
20477 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
20478 compute_bb_predicates, will_be_nonconstant_expr_predicate,
20479 will_be_nonconstant_predicate, record_modified_bb_info,
20480 get_minimal_bb, record_modified, param_change_prob,
20481 phi_result_unknown_predicate, predicate_for_phi_result,
20482 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
20483 estimate_function_body_sizes, compute_inline_parameters,
20484 compute_inline_parameters_for_curren, pass_data_inline_parameters,
20485 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
20486 inline_update_callee_summaries, remap_edge_change_prob,
20487 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
20488 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
20489 inline_analyze_function, inline_summary_t::insert,
20490 inline_generate_summary, read_ipa_call_summary, inline_read_section,
20491 inline_read_summary, write_ipa_call_summary, inline_write_summary,
20492 inline_free_summary): Move to ipa-fnsummary.h
20493 (predicate_t): Remove.
20494 * ipa-fnsummary.c: New file.
20495 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
20496 (enum inline_hints_vals, inline_hints, agg_position_info,
20497 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
20498 inline_summaries, ipa_call_summary, ipa_call_summary_t,
20499 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
20500 dump_inline_summary, dump_inline_hints, inline_generate_summary,
20501 inline_read_summary, inline_write_summary, inline_free_summary,
20502 inline_analyze_function, initialize_inline_failed,
20503 inline_merge_summary, inline_update_overall_summary,
20504 compute_inline_parameters): Move to ipa-fnsummary.h
20505 * ipa-fnsummary.h: New file.
20506 * ipa-inline-transform.h: Include ipa-inline.h.
20507 * ipa-inline.c: LIkewise.
20509 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
20511 * ipa-inline.c (edge_badness): Use inlined_time instead of
20512 inline_summaries->get.
20514 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
20516 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
20518 2017-05-22 Nathan Sidwell <nathan@acm.org>
20520 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
20521 (fdump-lang): Document 'raw' option.
20522 * dumpfile.h (TDI_tu): Delete.
20523 * dumpfile.c (dump_files): Remove translation-unit.
20524 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
20526 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
20528 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
20529 command option from $(AWK) call.
20530 * config/avr/genmultilib.awk: Simplify and rewrite so that it
20531 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
20532 [FORMAT]: Remove handling of variable.
20533 * config/avr/t-multilib: Regenerate.
20535 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
20537 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
20539 (dump_inline_summary): Do not print self_time.
20540 (estimate_function_body_sizes): Do not set self_time.
20541 (compute_inline_parameters): Likewise.
20542 (inline_read_section, inline_write_summary): Do not stream self_time.
20543 * ipa-inline.h (inline_summary): Drop self_time.
20545 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
20547 * ipa-inline-analysis.c (account_size_time): Rename to ...
20548 (inline_summary::account_size_time): ... this one.
20549 (reset_ipa_call_summary): Turn to ...
20550 (ipa_call_summary::reset): ... this one.
20551 (reset_inline_summary): Turn to ...
20552 (inline_summary::reset): ... this one.
20553 (inline_summary_t::remove): Update.
20554 (inline_summary_t::duplicate): Update.
20555 (ipa_call_summary_t::remove): Update.
20556 (dump_inline_summary): Update.
20557 (estimate_function_body_sizes): Update.
20558 (compute_inline_parameters): Update.
20559 (estimate_node_size_and_time): Update.
20560 (inline_merge_summary): Update.
20561 (inline_update_overall_summary): Update.
20562 (inline_read_section): Update.
20563 (inline_write_summary): Update.
20564 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
20565 add account_size_time and reset member functions.
20566 (ipa_call_summary): Add reset function.
20567 * ipa-predicate.h (predicate::operator &): Constify.
20569 2017-05-22 Richard Biener <rguenther@suse.de>
20571 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
20573 2017-05-19 Jason Merrill <jason@redhat.com>
20575 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
20577 2017-05-19 Marek Polacek <polacek@redhat.com>
20580 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
20581 TYPE_OVERFLOW_WRAPS checks.
20583 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
20585 * tree-core.h (enum omp_clause_default_kind): Add
20586 "OMP_CLAUSE_DEFAULT_PRESENT".
20587 * tree-pretty-print.c (dump_omp_clause): Handle it.
20588 * gimplify.c (enum gimplify_omp_var_data): Add
20589 "GOVD_MAP_FORCE_PRESENT".
20590 (gimplify_adjust_omp_clauses_1): Map it to
20591 "GOMP_MAP_FORCE_PRESENT".
20592 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
20594 * gimplify.c (oacc_default_clause): Clarify.
20596 2017-05-19 Nathan Sidwell <nathan@acm.org>
20598 LANG_HOOK_REGISTER_DUMPS
20599 * toplev.c (general_init): Call register dump lang hook.
20600 * doc/invoke.texi: Document -fdump-lang option family.
20601 * dumpfile.c (dump_files): Remove class dump here.
20602 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
20603 * dumpfile.h (tree_dump_index): Remove TDI_class.
20604 * langhooks-def.h (lhd_register_dumps): Declare.
20605 (LANG_HOOKS_REGISTER_DUMPS): Define.
20606 (LANG_HOOKS_INITIALIZER): Add it.
20607 * langhooks.c (lhd_register_dumps): Define.
20608 * langhooks.h (struct lang_hooks): Add register_dumps.
20610 2017-05-19 Nathan Sidwell <nathan@acm.org>
20612 * context.h (context::set_passes): New.
20613 * context.c (context::context): Do not create pass manager.
20614 * toplev.c (general_init): Create pass manager here.
20616 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
20618 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
20619 use this splitter if two add or or instructions would also work for
20620 the constant we want to generate.
20622 2017-05-19 Richard Biener <rguenther@suse.de>
20625 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
20626 predicate evaluation.
20628 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
20630 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
20632 * ipa-inline.c (want_inline_small_function_p): Do not cast to
20635 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
20637 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
20638 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
20639 (inline_edge_summary_vec): Turn into ...
20640 (ipa_call_summaries): ... this one.
20641 (redirect_to_unreachable, edge_set_predicate,
20642 evaluate_properties_for_edge, inline_summary_alloc,
20643 reset_ipa_call_summary, reset_inline_summary,
20644 inline_summary_t::duplicate): Update.
20645 (inline_edge_duplication_hook): Turn to ...
20646 (ipa_call_summary_t::duplicate): ... this one.
20647 (inline_edge_removal_hook): Turn to ...
20648 (ipa_call_summary_t::remove): ... this one.
20649 (dump_inline_edge_summary): Turn to ...
20650 (dump_ipa_call_summary): ... this one.
20651 (estimate_function_body_sizes): Update.
20652 (inline_update_callee_summaries): Update.
20653 (remap_edge_change_prob): Update.
20654 (remap_edge_summaries): Update.
20655 (inline_merge_summary): Update.
20656 (do_estimate_edge_time): Update.
20657 (inline_generate_summary): Update.
20658 (inline_read_section): Update.
20659 (inline_read_summary): Update.
20660 (inline_free_summary): Update.
20661 * ipa-inline.c (can_inline_edge_p): Update.
20662 (compute_inlined_call_time): Update.
20663 (want_inline_small_function_p): Update.
20664 (edge_badness): Update.
20665 (early_inliner): Update.
20666 * ipa-inline.h (inline_edge_summary): Turn to ...
20667 (ipa_call_summary): ... this one.
20668 (ipa_call_summary_t): New class.
20669 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
20670 (ipa_call_summaries): New.
20671 (inline_edge_summary): Remove.
20672 (estimate_edge_growth): Update.
20673 * ipa-profile.c (ipa_propagate_frequency_1): Update.
20674 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
20675 * ipa-split.c (execute_split_functions): Update.
20676 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
20678 2017-05-19 Richard Biener <rguenther@suse.de>
20680 PR middle-end/80764
20681 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
20683 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
20685 * config/rs6000/rs6000.c (struct machine_function): Add field
20686 fpr_is_wrapped_separately.
20687 (rs6000_get_separate_components): Use 64 components. Handle the
20688 new FPR components.
20689 (rs6000_components_for_bb): Handle the FPR components.
20690 (rs6000_emit_prologue_components): Handle the FPR components.
20691 (rs6000_emit_epilogue_components): Handle the FPR components.
20692 (rs6000_set_handled_components): Handle the FPR components.
20693 (rs6000_emit_prologue): Don't output prologue code for those FPRs
20694 that are already separately shrink-wrapped.
20695 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
20696 that are already separately shrink-wrapped.
20698 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
20701 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
20704 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
20705 (define_peephole2 for Altivec d-form load): Add peepholes to catch
20706 cases where the register allocator uses a move and an offsettable
20707 memory operation to/from a FPR register on ISA 2.06/2.07.
20708 (define_peephole2 for Altivec d-form store): Likewise.
20710 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
20713 * config/i386/mmx.md (*mov<mode>_internal): Enable
20714 alternatives 11, 12, 13 and 14 also for 32bit targets.
20715 Remove alternatives 15, 16, 17 and 18.
20716 * config/i386/sse.md (vec_concatv2di): Change
20717 alternative (!x, *y) to (x, ?!*Yn).
20719 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
20721 * dumpfile.h (enum dump_kind): Remove stray comma.
20723 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
20725 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
20726 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
20727 predicate::num_conditions
20728 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
20729 (CHANGED): turn into predicate::changed.
20730 (agg_position_info): Move to ipa-predicate.h
20731 (add_condition, predicate::add_clause, predicate::operator &=,
20732 predicate::or_with, predicate::evaluate, predicate::probability,
20733 dump_condition, dump_clause, predicate::dump,
20734 predicate::remap_after_duplication, predicate::remap_after_inlining,
20735 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
20736 (evaluate_conditions_for_known_args): Update.
20737 (set_cond_stmt_execution_predicate): Update.
20738 * ipa-inline.h: Include ipa-predicate.h
20739 (condition, inline_param_summary, conditions, agg_position_info,
20740 predicate): Move to ipa-predicate.h
20741 * ipa-predicate.c: New file.
20742 * ipa-predicate.h: New file.
20744 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
20746 * final.c (leaf_function_p): Check we are not in a sequence.
20748 2017-05-18 Martin Liska <mliska@suse.cz>
20750 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
20751 * dumpfile.c (dump_register): Use new enum dump_kind.
20752 (get_dump_file_name): Likewise.
20753 (dump_enable_all): Likewise.
20754 (dump_switch_p_1): Likewise.
20755 (enable_rtl_dump_file): Remove usage of TDF_RTL.
20756 * dumpfile.h (enum dump_kind): New enum type.
20757 (struct dump_file_info): Create constructor and
20758 format fields and comments.
20759 * passes.c (pass_manager::register_one_dump_file):
20761 * statistics.c (statistics_early_init): Likewise.
20762 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
20763 TDF_TREE with TDF_SLIM.
20764 (gather_memory_references_ref): Likewise.
20766 2017-05-18 Martin Liska <mliska@suse.cz>
20768 * vec.h (struct vnull): Use it.
20770 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
20772 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
20773 (true_predicate, false_predicate, true_predicate_p,
20774 false_predicate_p): Remove.
20775 (single_cond_predicate, not_inlined_predicate): Turn to member function
20777 (add_condition): Update.
20778 (add_clause): Turn to...
20779 (predicate::add_clause): ... this one; update; allow passing NULL
20781 (and_predicates): Turn to ...
20782 (predicate::operator &=): ... this one.
20783 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
20784 (or_predicates): Turn to ...
20785 (predicate::or_with): ... this one.
20786 (evaluate_predicate): Turn to ...
20787 (predicate::evaluate): ... this one.
20788 (predicate_probability): Turn to ...
20789 (predicate::probability): ... this one.
20790 (dump_condition): Update.
20791 (dump_predicate): Turn to ...
20792 (predicate::dump): ... this one.
20793 (account_size_time): Update.
20794 (edge_set_predicate): Update.
20795 (set_hint_predicate): UPdate.
20796 (evaluate_conditions_for_known_args): Update.
20797 (evaluate_properties_for_edge): Update.
20798 (remap_predicate_after_duplication): Turn to...
20799 (predicate::remap_after_duplication): ... this one.
20800 (remap_hint_predicate_after_duplication): Update.
20801 (inline_summary_t::duplicate): UPdate.
20802 (dump_inline_edge_summary): Update.
20803 (dump_inline_summary): Update.
20804 (set_cond_stmt_execution_predicate): Update.
20805 (set_switch_stmt_execution_predicate): Update.
20806 (compute_bb_predicates): Update.
20807 (will_be_nonconstant_expr_predicate): Update.
20808 (will_be_nonconstant_predicate): Update.
20809 (phi_result_unknown_predicate): Update.
20810 (predicate_for_phi_result): Update.
20811 (array_index_predicate): Update.
20812 (estimate_function_body_sizes): Update.
20813 (estimate_node_size_and_time): Update.
20814 (estimate_ipcp_clone_size_and_time): Update.
20815 (remap_predicate): Rename to ...
20816 (predicate::remap_after_inlining): ... this one.
20817 (remap_hint_predicate): Update.
20818 (inline_merge_summary): Update.
20819 (inline_update_overall_summary): Update.
20820 (estimate_size_after_inlining): Update.
20821 (read_predicate): Rename to ...
20822 (predicate::stream_in): ... this one.
20823 (read_inline_edge_summary): Update.
20824 (write_predicate): Rename to ...
20825 (predicate::stream_out): ... this one.
20826 (write_inline_edge_summary): Update.
20827 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
20828 (clause_t): Turn to uint32_t
20829 (predicate): Turn to class; implement constructor and operators
20831 (size_time_entry): Update.
20832 (inline_summary): Update.
20833 (inline_edge_summary): Update.
20835 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
20837 * fold-const.c (fold_binary_loc): Move transformation...
20838 * match.pd (C - X CMP X): ... here.
20840 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
20842 * config/sparc/sparc.c (sparc_option_override): Set function
20843 alignment for -mcpu=niagara7 to 64 to match the I$ line.
20844 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
20846 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
20848 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
20850 2017-05-18 Marek Polacek <polacek@redhat.com>
20853 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
20854 (pass_ubsan::execute): Call gimple_assign_single_p instead of
20855 gimple_assign_load_p.
20857 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
20859 PR middle-end/80692
20860 * real.c (do_compare): Give decimal_do_compare preference over
20861 comparing just the signs.
20863 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
20865 * doc/md.texi (Canonicalization of Instructions): Describe the
20866 canonical form of instructions that inherently set a condition
20869 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
20871 PR middle-end/80775
20872 * tree-cfg.c: Move deletion of unreachable case statements to after
20873 the merging of consecutive case labels.
20875 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
20877 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
20878 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
20879 restoring of callee-saved registers.
20881 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
20883 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
20884 * config/visium/visium.c (single_set_and_flags): Likewise.
20885 * config/visium/visium.md (Substitutions): Likewise.
20887 2017-05-17 Martin Liska <mliska@suse.cz>
20889 * cfg.c: Introduce dump_flags_t type and
20890 use it instead of int type.
20892 * cfghooks.c: Likewise.
20893 * cfghooks.h (struct cfg_hooks): Likewise.
20894 * cfgrtl.c: Likewise.
20895 * cfgrtl.h: Likewise.
20896 * cgraph.c (cgraph_node::get_body): Likewise.
20897 * coretypes.h: Likewise.
20898 * domwalk.c: Likewise.
20899 * domwalk.h: Likewise.
20900 * dumpfile.c (struct dump_option_value_info): Likewise.
20901 (dump_enable_all): Likewise.
20902 (dump_switch_p_1): Likewise.
20903 (opt_info_switch_p): Likewise.
20904 * dumpfile.h (enum tree_dump_index): Likewise.
20905 (struct dump_file_info): Likewise.
20906 * genemit.c: Likewise.
20907 * generic-match-head.c: Likewise.
20908 * gengtype.c (open_base_files): Likewise.
20909 * gimple-pretty-print.c: Likewise.
20910 * gimple-pretty-print.h: Likewise.
20911 * graph.c (print_graph_cfg): Likewise.
20912 * graphite-scop-detection.c (dot_all_sese): Likewise.
20913 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
20914 * loop-unroll.c (report_unroll): Likewise.
20915 * passes.c (pass_manager::register_one_dump_file): Likewise.
20916 * print-tree.c: Likewise.
20917 * statistics.c: Likewise.
20918 * tree-cfg.c: Likewise.
20919 * tree-cfg.h: Likewise.
20920 * tree-dfa.c: Likewise.
20921 * tree-dfa.h: Likewise.
20922 * tree-dump.c (dump_function): Likewise.
20923 * tree-dump.h (struct dump_info): Likewise.
20924 * tree-pretty-print.c: Likewise.
20925 * tree-pretty-print.h: Likewise.
20926 * tree-ssa-live.c: Likewise.
20927 * tree-ssa-live.h: Likewise.
20928 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
20929 * tree-vect-loop.c: Likewise.
20930 * tree-vect-slp.c: Likewise.
20932 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
20933 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20935 PR tree-optimization/80457
20936 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
20937 of all arguments to a statement as scalar_to_vec operations.
20938 (vectorizable_call): Adjust call to vect_model_simple_cost for
20940 (vectorizable_conversion): Likewise.
20941 (vectorizable_assignment): Likewise.
20942 (vectorizable_shift): Likewise.
20943 (vectorizable_operation): Likewise.
20944 (vectorizable_comparison): Likewise.
20945 (vect_is_simple_cond): Record the def types for operands.
20946 (vectorizable_condition): Likewise, call vect_model_simple_cost.
20947 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
20948 for statement argument count.
20950 2017-05-16 Carl Love <cel@us.ibm.com>
20952 * config/rs6000/rs6000-c: Add support for built-in functions
20953 vector unsigned long long vec_bperm (vector unsigned long long,
20954 vector unsigned char)
20955 vector signed long long vec_mule (vector signed int,
20957 vector unsigned long long vec_mule (vector unsigned int,
20958 vector unsigned int)
20959 vector signed long long vec_mulo (vector signed int,
20961 vector unsigned long long vec_mulo (vector unsigned int,
20962 vector unsigned int)
20963 vector signed char vec_sldw (vector signed char,
20964 vector signed char,
20966 vector unsigned char vec_sldw (vector unsigned char,
20967 vector unsigned char,
20969 vector signed short vec_sldw (vector signed short,
20970 vector signed short,
20972 vector unsigned short vec_sldw (vector unsigned short,
20973 vector unsigned short,
20975 vector signed int vec_sldw (vector signed int,
20978 vector unsigned int vec_sldw (vector unsigned int,
20979 vector unsigned int,
20981 vector signed long long vec_sldw (vector signed long long,
20982 vector signed long long,
20984 vector unsigned long long vec_sldw (vector unsigned long long,
20985 vector unsigned long long,
20987 * config/rs6000/rs6000-c: Add support for built-in functions
20988 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
20989 * config/rs6000/altivec.h: Add defintion for vec_sldw.
20990 * doc/extend.texi: Update the built-in documentation for the
20991 new built-in functions.
20993 2017-05-16 Marek Polacek <polacek@redhat.com>
20997 * tree.c (save_expr): Don't fold the expression.
20999 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
21001 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
21002 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
21003 and (?*y,m). Update insn attributes.
21005 2017-05-16 Martin Liska <mliska@suse.cz>
21007 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
21008 flags argument of print_gimple_stmt, print_gimple_expr,
21009 print_generic_stmt and print_generic_expr.
21010 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
21011 * coretypes.h: Likewise.
21012 * except.c (dump_eh_tree): Likewise.
21013 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
21014 * gimple-pretty-print.h: Likewise.
21015 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
21016 (backprop::push_to_worklist): Likewise.
21017 (backprop::pop_from_worklist): Likewise.
21018 (backprop::process_use): Likewise.
21019 (backprop::intersect_uses): Likewise.
21020 (note_replacement): Likewise.
21021 * gimple-ssa-store-merging.c
21022 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
21023 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
21024 (pass_store_merging::execute): Likewise.
21025 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
21026 (ssa_base_cand_dump_callback): Likewise.
21027 (dump_incr_vec): Likewise.
21028 (replace_refs): Likewise.
21029 (replace_mult_candidate): Likewise.
21030 (create_add_on_incoming_edge): Likewise.
21031 (create_phi_basis): Likewise.
21032 (insert_initializers): Likewise.
21033 (all_phi_incrs_profitable): Likewise.
21034 (introduce_cast_before_cand): Likewise.
21035 (replace_one_candidate): Likewise.
21036 * gimplify.c (gimplify_expr): Likewise.
21037 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
21038 (set_rename): Likewise.
21039 (rename_uses): Likewise.
21040 (copy_loop_phi_nodes): Likewise.
21041 (add_close_phis_to_merge_points): Likewise.
21042 (copy_loop_close_phi_args): Likewise.
21043 (copy_cond_phi_args): Likewise.
21044 (graphite_copy_stmts_from_block): Likewise.
21045 (translate_pending_phi_nodes): Likewise.
21046 * graphite-poly.c (print_pdr): Likewise.
21047 (dump_gbb_cases): Likewise.
21048 (dump_gbb_conditions): Likewise.
21049 (print_scop_params): Likewise.
21050 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
21051 (build_cross_bb_scalars_use): Likewise.
21052 (gather_bbs::before_dom_children): Likewise.
21053 * hsa-dump.c (dump_hsa_immed): Likewise.
21054 * ipa-cp.c (print_ipcp_constant_value): Likewise.
21055 (get_replacement_map): Likewise.
21056 * ipa-inline-analysis.c (dump_condition): Likewise.
21057 (estimate_function_body_sizes): Likewise.
21058 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
21059 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
21060 * ipa-prop.c (ipa_dump_param): Likewise.
21061 (ipa_print_node_jump_functions_for_edge): Likewise.
21062 (ipa_modify_call_arguments): Likewise.
21063 (ipa_modify_expr): Likewise.
21064 (ipa_dump_param_adjustments): Likewise.
21065 (ipa_dump_agg_replacement_values): Likewise.
21066 (ipcp_modif_dom_walker::before_dom_children): Likewise.
21067 * ipa-pure-const.c (check_stmt): Likewise.
21068 (pass_nothrow::execute): Likewise.
21069 * ipa-split.c (execute_split_functions): Likewise.
21070 * omp-offload.c (dump_oacc_loop_part): Likewise.
21071 (dump_oacc_loop): Likewise.
21072 * trans-mem.c (tm_log_emit): Likewise.
21073 (tm_memopt_accumulate_memops): Likewise.
21074 (dump_tm_memopt_set): Likewise.
21075 (dump_tm_memopt_transform): Likewise.
21076 * tree-cfg.c (gimple_verify_flow_info): Likewise.
21077 (print_loop): Likewise.
21078 * tree-chkp-opt.c (chkp_print_addr): Likewise.
21079 (chkp_gather_checks_info): Likewise.
21080 (chkp_get_check_result): Likewise.
21081 (chkp_remove_check_if_pass): Likewise.
21082 (chkp_use_outer_bounds_if_possible): Likewise.
21083 (chkp_reduce_bounds_lifetime): Likewise.
21084 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
21085 (chkp_mark_completed_bounds): Likewise.
21086 (chkp_register_incomplete_bounds): Likewise.
21087 (chkp_mark_invalid_bounds): Likewise.
21088 (chkp_maybe_copy_and_register_bounds): Likewise.
21089 (chkp_build_returned_bound): Likewise.
21090 (chkp_get_bound_for_parm): Likewise.
21091 (chkp_build_bndldx): Likewise.
21092 (chkp_get_bounds_by_definition): Likewise.
21093 (chkp_generate_extern_var_bounds): Likewise.
21094 (chkp_get_bounds_for_decl_addr): Likewise.
21095 * tree-chrec.c (chrec_apply): Likewise.
21096 * tree-data-ref.c (dump_data_reference): Likewise.
21097 (dump_subscript): Likewise.
21098 (dump_data_dependence_relation): Likewise.
21099 (analyze_overlapping_iterations): Likewise.
21100 * tree-inline.c (expand_call_inline): Likewise.
21101 (tree_function_versioning): Likewise.
21102 * tree-into-ssa.c (dump_defs_stack): Likewise.
21103 (dump_currdefs): Likewise.
21104 (dump_names_replaced_by): Likewise.
21105 (dump_update_ssa): Likewise.
21106 (update_ssa): Likewise.
21107 * tree-object-size.c (pass_object_sizes::execute): Likewise.
21108 * tree-parloops.c (build_new_reduction): Likewise.
21109 (try_create_reduction_list): Likewise.
21110 (ref_conflicts_with_region): Likewise.
21111 (oacc_entry_exit_ok_1): Likewise.
21112 (oacc_entry_exit_single_gang): Likewise.
21113 * tree-pretty-print.h: Likewise.
21114 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
21115 (get_scalar_evolution): Likewise.
21116 (add_to_evolution): Likewise.
21117 (get_loop_exit_condition): Likewise.
21118 (analyze_evolution_in_loop): Likewise.
21119 (analyze_initial_condition): Likewise.
21120 (analyze_scalar_evolution): Likewise.
21121 (instantiate_scev): Likewise.
21122 (number_of_latch_executions): Likewise.
21123 (gather_chrec_stats): Likewise.
21124 (final_value_replacement_loop): Likewise.
21125 (scev_const_prop): Likewise.
21126 * tree-sra.c (dump_access): Likewise.
21127 (disqualify_candidate): Likewise.
21128 (create_access): Likewise.
21129 (reject): Likewise.
21130 (maybe_add_sra_candidate): Likewise.
21131 (create_access_replacement): Likewise.
21132 (analyze_access_subtree): Likewise.
21133 (analyze_all_variable_accesses): Likewise.
21134 (sra_modify_assign): Likewise.
21135 (initialize_constant_pool_replacements): Likewise.
21136 (find_param_candidates): Likewise.
21137 (decide_one_param_reduction): Likewise.
21138 (replace_removed_params_ssa_names): Likewise.
21139 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
21140 * tree-ssa-copy.c (dump_copy_of): Likewise.
21141 (copy_prop_visit_cond_stmt): Likewise.
21142 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
21143 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
21144 (record_equivalences_from_stmt): Likewise.
21145 * tree-ssa-dse.c (compute_trims): Likewise.
21146 (delete_dead_call): Likewise.
21147 (delete_dead_assignment): Likewise.
21148 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
21149 (forward_propagate_into_cond): Likewise.
21150 (pass_forwprop::execute): Likewise.
21151 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
21152 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
21154 (move_computations_worker): Likewise.
21155 (execute_sm): Likewise.
21156 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
21157 (remove_exits_and_undefined_stmts): Likewise.
21158 (remove_redundant_iv_tests): Likewise.
21159 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
21160 (adjust_iv_update_pos): Likewise.
21161 * tree-ssa-math-opts.c (bswap_replace): Likewise.
21162 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
21163 (value_replacement): Likewise.
21164 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
21165 * tree-ssa-pre.c (print_pre_expr): Likewise.
21166 (get_representative_for): Likewise.
21167 (create_expression_by_pieces): Likewise.
21168 (insert_into_preds_of_block): Likewise.
21169 (eliminate_insert): Likewise.
21170 (eliminate_dom_walker::before_dom_children): Likewise.
21171 (eliminate): Likewise.
21172 (remove_dead_inserted_code): Likewise.
21173 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
21174 * tree-ssa-reassoc.c (get_rank): Likewise.
21175 (eliminate_duplicate_pair): Likewise.
21176 (eliminate_plus_minus_pair): Likewise.
21177 (eliminate_not_pairs): Likewise.
21178 (undistribute_ops_list): Likewise.
21179 (eliminate_redundant_comparison): Likewise.
21180 (update_range_test): Likewise.
21181 (optimize_range_tests_var_bound): Likewise.
21182 (optimize_vec_cond_expr): Likewise.
21183 (rewrite_expr_tree): Likewise.
21184 (rewrite_expr_tree_parallel): Likewise.
21185 (linearize_expr): Likewise.
21186 (break_up_subtract): Likewise.
21187 (linearize_expr_tree): Likewise.
21188 (attempt_builtin_powi): Likewise.
21189 (attempt_builtin_copysign): Likewise.
21190 (transform_stmt_to_copy): Likewise.
21191 (transform_stmt_to_multiply): Likewise.
21192 (dump_ops_vector): Likewise.
21193 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
21194 (print_scc): Likewise.
21195 (set_ssa_val_to): Likewise.
21196 (visit_reference_op_store): Likewise.
21197 (visit_use): Likewise.
21198 (sccvn_dom_walker::before_dom_children): Likewise.
21199 (run_scc_vn): Likewise.
21200 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
21202 (expr_hash_elt::print): Likewise.
21203 (const_and_copies::pop_to_marker): Likewise.
21204 (const_and_copies::record_const_or_copy_raw): Likewise.
21205 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
21206 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
21207 (dump_predicates): Likewise.
21208 (find_uninit_use): Likewise.
21209 (warn_uninitialized_phi): Likewise.
21210 (pass_late_warn_uninitialized::execute): Likewise.
21211 * tree-ssa.c (verify_vssa): Likewise.
21212 (verify_ssa): Likewise.
21213 (maybe_optimize_var): Likewise.
21214 * tree-vrp.c (dump_value_range): Likewise.
21215 (dump_all_value_ranges): Likewise.
21216 (dump_asserts_for): Likewise.
21217 (register_edge_assert_for_2): Likewise.
21218 (vrp_visit_cond_stmt): Likewise.
21219 (vrp_visit_switch_stmt): Likewise.
21220 (vrp_visit_stmt): Likewise.
21221 (vrp_visit_phi_node): Likewise.
21222 (simplify_cond_using_ranges_1): Likewise.
21223 (fold_predicate_in): Likewise.
21224 (evrp_dom_walker::before_dom_children): Likewise.
21225 (evrp_dom_walker::push_value_range): Likewise.
21226 (evrp_dom_walker::pop_value_range): Likewise.
21227 (execute_early_vrp): Likewise.
21229 2017-05-16 Richard Biener <rguenther@suse.de>
21231 * dwarf2out.c (loc_list_from_tree_1): Do not create
21232 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
21234 2017-05-16 Richard Biener <rguenther@suse.de>
21236 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
21238 (note_variable_value_in_expr): If we resolved the decl ref
21239 do not push to the stack.
21241 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
21243 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
21244 operations in fast-math mode.
21245 (vaddq_f16): Likewise.
21246 (vmul_f16): Likewise.
21247 (vmulq_f16): Likewise.
21248 (vsub_f16): Likewise.
21249 (vsubq_f16): Likewise.
21250 * config/arm/neon.md (add<mode>3): New.
21252 (fma:<VH:mode>3): New. Also remove outdated comment.
21255 2017-05-16 Martin Liska <mliska@suse.cz>
21259 * ipa-devirt.c (warn_types_mismatch): Fix typo.
21260 (odr_types_equivalent_p): Likewise.
21262 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
21264 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
21266 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
21269 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
21270 non-interunit SSE move alternatives with '?'.
21271 (zero-extendsidi peephole2): New peephole to skip intermediate
21272 general register in SSE zero-extend sequence.
21274 2017-05-15 Jeff Law <law@redhat.com>
21276 * reorg.c (relax_delay_slots): Create a new variable to hold
21277 the temporary target rather than clobbering TARGET_LABEL.
21279 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
21280 missing argument to extract_bit_field call.
21281 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
21283 2017-05-15 Martin Liska <mliska@suse.cz>
21286 * gcc.c (process_command): Do not allow empty argument of -o option.
21288 2017-05-15 Renlin Li <renlin.li@arm.com>
21290 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
21291 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
21292 * config/aarch64/constraints.md (Usf): Add long call check.
21293 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
21294 (call_value): Likewise.
21295 (sibcall): Likewise.
21296 (sibcall_value): Likewise.
21298 (call_value_insn): New.
21299 (sibcall_insn): Update rtx pattern.
21300 (sibcall_value_insn): Likewise.
21301 (call_internal): Remove.
21302 (call_value_internal): Likewise.
21303 (sibcall_internal): Likewise.
21304 (sibcall_value_internal): Likewise.
21305 (call_reg): Likewise.
21306 (call_symbol): Likewise.
21307 (call_value_reg): Likewise.
21308 (call_value_symbol): Likewise.
21310 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
21313 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
21315 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
21317 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
21318 compatible with CCGOCmode and with CCZmode.
21320 2017-05-14 Martin Sebor <msebor@redhat.com>
21322 PR middle-end/77671
21323 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
21324 (gimple_fold_builtin_snprintf): Same.
21325 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
21326 (gimple_fold_builtin_snprintf): Same.
21327 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
21328 of character types.
21329 (is_call_safe): New function.
21330 (try_substitute_return_value): Call it.
21331 (try_simplify_call): New function.
21332 (pass_sprintf_length::handle_gimple_call): Call it.
21334 2017-05-14 Martin Sebor <msebor@redhat.com>
21336 PR middle-end/80669
21337 * builtins.c (expand_builtin_stpncpy): Simplify.
21339 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
21341 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
21342 * config/i386/i386.h
21343 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
21344 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
21345 (struct machine_function): Add new members call_ms2sysv,
21346 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
21347 (struct machine_frame_state): New fields sp_realigned and
21348 sp_realigned_offset.
21349 * config/i386/i386.c
21350 (enum xlogue_stub): New enum.
21351 (enum xlogue_stub_sets): New enum.
21352 (class xlogue_layout): New class.
21353 (struct ix86_frame): New fields stack_realign_allocate_offset,
21354 stack_realign_offset and outlined_save_offset. Modify comments to
21355 detail stack layout when using out-of-line stubs.
21356 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
21357 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
21358 -mcall-ms2sysv-xlogues.
21359 (stub_managed_regs): New static variable.
21360 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
21361 registers managed by out-of-line stub.
21362 (disable_call_ms2sysv_xlogues): New function.
21363 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
21364 m->call_ms2sysv when appropriate and compute frame layout for
21366 (sp_valid_at, fp_valid_at): New inline functions.
21367 (choose_basereg): New function.
21368 (choose_baseaddr): Add align parameter, use choose_basereg and modify
21370 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
21371 Use align parameter of choose_baseaddr to generated aligned SSE movs
21373 (pro_epilogue_adjust_stack): Modify to track
21374 machine_frame_state::sp_realigned.
21375 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
21376 (ix86_nsaved_sseregs): Likewise.
21377 (ix86_emit_save_regs): Likewise.
21378 (ix86_emit_save_regs_using_mov): Likewise.
21379 (ix86_emit_save_sse_regs_using_mov): Likewise.
21380 (get_scratch_register_on_entry): Likewise.
21381 (gen_frame_set): New function.
21382 (gen_frame_load): Likewise.
21383 (gen_frame_store): Likewise.
21384 (emit_outlined_ms2sysv_save): Likewise.
21385 (emit_outlined_ms2sysv_restore): Likewise.
21386 (ix86_expand_prologue): Modify stack re-alignment code and call
21387 emit_outlined_ms2sysv_save when appropriate.
21388 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
21389 parameter rtx_insn *insn, which allows the function to be used to only
21390 generate the notes.
21391 (ix86_expand_epilogue): Modify validity checks of frame and stack
21392 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
21393 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
21394 * config/i386/predicates.md
21395 (save_multiple): New predicate.
21396 (restore_multiple): Likewise.
21397 * config/i386/sse.md
21398 (save_multiple<mode>): New pattern.
21399 (save_multiple_realign<mode>): Likewise.
21400 (restore_multiple<mode>): Likewise.
21401 (restore_multiple_and_return<mode>): Likewise.
21402 (restore_multiple_leave_return<mode>): Likewise.
21403 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
21405 2017-05-14 Julia Koval <julia.koval@intel.com>
21407 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
21408 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
21409 (__builtin_ia32_xsetbv): New builtins.
21410 * config/i386/i386.c (ix86_expand_special_args_builtin):
21412 (ix86_expand_builtin): Special expand for new intrinsics.
21413 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
21414 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
21415 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
21417 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21419 * cfganal.c (inverted_post_order_compute): Change argument type
21421 * cfganal.h (inverted_post_order_compute): Adjust prototype.
21422 * df-core.c (rest_of_handle_df_initialize): Adjust.
21423 (rest_of_handle_df_finish): Likewise.
21424 (df_analyze_1): Likewise.
21425 (df_analyze): Likewise.
21426 (loop_inverted_post_order_compute): Change argument to be a vec *.
21427 (df_analyze_loop): Adjust.
21428 (df_get_n_blocks): Likewise.
21429 (df_get_postorder): Likewise.
21430 * df.h (struct df_d): Change field to be a vec.
21431 * lcm.c (compute_laterin): Adjust.
21432 (compute_available): Likewise.
21433 * lra-lives.c (lra_create_live_ranges_1): Likewise.
21434 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
21435 * tree-ssa-pre.c (compute_antic): Likewise.
21437 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21439 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
21440 (depth_first_search::depth_first_search): Change structure init
21441 function to this constructor.
21442 (depth_first_search::add_bb): Rename function to this member.
21443 (depth_first_search::execute): Likewise.
21444 (flow_dfs_compute_reverse_finish): Adjust.
21446 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21448 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
21449 (longest_simple_path): Likewise.
21450 * shrink-wrap.c (spread_components): Likewise.
21451 (disqualify_problematic_components): Likewise.
21452 (emit_common_heads_for_components): Likewise.
21453 (emit_common_tails_for_components): Likewise.
21454 (insert_prologue_epilogue_for_components): Likewise.
21456 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21458 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
21461 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21463 * df-core.c (df_set_blocks): Start using auto_bitmap.
21464 (df_compact_blocks): Likewise.
21465 * df-problems.c (df_rd_confluence_n): Likewise.
21466 * df-scan.c (df_insn_rescan_all): Likewise.
21467 (df_process_deferred_rescans): Likewise.
21468 (df_update_entry_block_defs): Likewise.
21469 (df_update_exit_block_uses): Likewise.
21470 (df_entry_block_bitmap_verify): Likewise.
21471 (df_exit_block_bitmap_verify): Likewise.
21472 (df_scan_verify): Likewise.
21473 * lra-constraints.c (lra_constraints): Likewise.
21474 (undo_optional_reloads): Likewise.
21475 (lra_undo_inheritance): Likewise.
21476 * lra-remat.c (calculate_gen_cands): Likewise.
21477 (do_remat): Likewise.
21478 * lra-spills.c (assign_spill_hard_regs): Likewise.
21479 (spill_pseudos): Likewise.
21480 * tree-ssa-pre.c (bitmap_set_and): Likewise.
21481 (bitmap_set_subtract_values): Likewise.
21483 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21485 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
21486 management with auto_bitmap.
21487 (fix_inter_tick): Likewise.
21488 (fix_recovery_deps): Likewise.
21489 * ira.c (add_store_equivs): Likewise.
21490 (find_moveable_pseudos): Likewise.
21491 (split_live_ranges_for_shrink_wrap): Likewise.
21492 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
21493 (rtx_reuse_manager::seen_def_p): Likewise.
21494 (rtx_reuse_manager::set_seen_def): Likewise.
21495 * print-rtl.h (class rtx_reuse_manager): Likewise.
21497 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21499 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
21501 (migrate_btr_def): Likewise.
21502 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
21503 * df-core.c (loop_post_order_compute): Likewise.
21504 (loop_inverted_post_order_compute): Likewise.
21505 * hsa-common.h: Likewise.
21506 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
21507 * init-regs.c (initialize_uninitialized_regs): Likewise.
21508 * ipa-inline.c (resolve_noninline_speculation): Likewise.
21509 (inline_small_functions): Likewise.
21510 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
21511 * ira.c (combine_and_move_insns): Likewise.
21512 (build_insn_chain): Likewise.
21513 * loop-invariant.c (find_invariants): Likewise.
21514 * lower-subreg.c (propagate_pseudo_copies): Likewise.
21515 * predict.c (tree_predict_by_opcode): Likewise.
21516 (predict_paths_leading_to): Likewise.
21517 (predict_paths_leading_to_edge): Likewise.
21518 (estimate_loops_at_level): Likewise.
21519 (estimate_loops): Likewise.
21520 * shrink-wrap.c (try_shrink_wrapping): Likewise.
21521 (spread_components): Likewise.
21522 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
21523 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
21524 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
21525 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
21526 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
21527 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
21528 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
21529 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
21530 (mark_threaded_blocks): Likewise.
21531 (thread_through_all_blocks): Likewise.
21532 * tree-ssa.c (verify_ssa): Likewise.
21533 (execute_update_addresses_taken): Likewise.
21534 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
21536 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21538 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
21540 (post_order_compute): Likewise.
21541 (inverted_post_order_compute): Likewise.
21542 (pre_and_rev_post_order_compute_fn): Likewise.
21544 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21546 * genrecog.c (int_set::int_set): Explicitly construct our
21547 auto_vec base class.
21548 * vec.h (auto_vec::auto_vec): New constructor.
21550 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21552 * bitmap.h (class auto_bitmap): New constructor taking
21553 bitmap_obstack * argument.
21555 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21557 * bitmap.h (class auto_bitmap): Change type of m_bits to
21558 bitmap_head, and adjust ctor / dtor and member operators.
21560 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
21562 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
21563 when returned register mode doesn't match original mode.
21565 2017-05-12 Jeff Law <law@redhat.com>
21566 Jakub Jelinek <jakub@redhat.com>
21568 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
21569 we look for cc setter after the compare-elim changes.
21570 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
21571 within the vector to match what compare-elim now expects.
21572 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
21573 (xorsi3_flags, one_cmplsi2_flags): Likewise.
21575 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
21576 after the compare-elim changes.
21577 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
21578 the vector to match what compare-elim now expects.
21579 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
21580 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
21581 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
21582 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
21583 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
21585 * config/visium/visium.c (single_set_and_flags): Fix where
21586 we look for cc setter after the compare-elim changes.
21587 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
21588 with the vector to match what compare-elim now expects.
21589 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
21590 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
21591 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
21592 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
21593 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
21594 (neg<mode>2_insn_set_overflow): Likewise.
21596 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
21598 PR middle-end/79794
21599 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
21600 maybe_expand_insn call, set ops[0].target. If still set after call,
21601 set alt_rtl. Add extra arg to recursive calls.
21602 (extract_bit_field): Add alt_rtl argument. Pass to
21604 * expmed.h (extract_bit_field): Fix prototype.
21605 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
21606 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
21607 to extract_bit_field_calls.
21608 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
21609 Pass alt_rtl to extract_bit_field calls.
21610 * calls.c (store_unaligned_arguments_into_psuedos)
21611 load_register_parameters): Pass extra NULL to extract_bit_field calls.
21612 * optabs.c (maybe_legitimize_operand): Clear op->target when call
21614 * optabs.h (struct expand_operand): Add target bitfield.
21616 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
21618 * compare-elim.c (try_eliminate_compare): Canonicalize
21619 operation with embedded compare to
21620 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
21621 (set (reg) (operation)].
21623 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
21625 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
21628 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
21629 cost of adding a carry flag for ADC instruction.
21630 [case MINUS]: Ignore the cost of subtracting a carry flag
21631 for SBB instruction.
21633 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
21635 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
21637 * config/rs6000/bmiintrin.h: New file.
21638 * config/rs6000/bmi2intrin.h: New file.
21639 * config/rs6000/x86intrin.h: New file.
21641 2017-05-12 Jeff Law <law@redhat.com>
21643 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
21646 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
21648 PR middle-end/80707
21649 * tree-cfg.c: Remove cfg edges of unreachable case statements.
21651 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
21653 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
21654 early expansion of vector divide builtins.
21655 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
21656 builtins identified as having unsigned arguments.
21658 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
21660 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
21661 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
21662 expansion of vector logical operations (and, andc, or, xor,
21665 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
21667 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
21668 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
21670 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
21672 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
21673 early GIMPLE expansion of vector multiplies.
21675 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21677 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
21678 TARGET_HAVE_MOVT conditional.
21679 (movt splitter): Likewise.
21681 2017-05-12 Richard Biener <rguenther@suse.de>
21683 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
21684 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
21685 Fold all stmts not inplace.
21687 2017-05-12 Richard Biener <rguenther@suse.de>
21689 PR tree-optimization/80713
21690 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
21691 inserted_exprs bit for not removed stmts.
21693 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
21695 PR middle-end/69921
21696 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
21697 parallelized" attribute for parallelized OpenACC kernels.
21698 * omp-offload.c (execute_oacc_device_lower): Use it.
21700 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
21701 Set "oacc kernels" attribute.
21702 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
21703 parameter. Adjust all users.
21704 (oacc_fn_attrib_kernels_p): Remove function.
21705 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
21706 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
21707 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
21708 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
21709 assert "oacc kernels" attribute is set.
21711 2017-05-11 Carl Love <cel@us.ibm.com>
21713 * config/rs6000/rs6000-c: Add support for built-in functions
21714 vector unsigned char vec_popcnt (vector signed char)
21715 vector unsigned char vec_popcnt (vector unsigned char)
21716 vector unsigned short vec_popcnt (vector signed short)
21717 vector unsigned short vec_popcnt (vector unsigned short)
21718 vector unsigned int vec_popcnt (vector signed int)
21719 vector unsigned int vec_popcnt (vector unsigned int)
21720 vector unsigned long long vec_popcnt (vector signed long long)
21721 vector unsigned long long vec_popcnt (vector unsigned long long)
21722 vector signed long long vec_slo (vector signed long long,
21723 vector signed char)
21724 vector signed long long vec_slo (vector signed long long,
21725 vector unsigned char)
21726 vector unsigned long long vec_slo (vector unsigned long long,
21727 vector signed char)
21728 vector unsigned long long vec_slo (vector unsigned long long,
21729 vector unsigned char)
21730 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
21731 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
21732 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
21733 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
21734 * doc/extend.texi: Update the built-in documentation file for the
21735 new built-in functions.
21737 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
21739 * attribs.h (sorted_attr_string): Move machine independent
21740 functions for target clone support from the i386 port to common
21741 code. Rename ix86_function_versions to common_function_versions.
21742 Rename make_name to make_unique_name.
21743 (common_function_versions): Likewise.
21744 (make_unique_name): Likewise.
21745 (make_dispatcher_decl): Likewise.
21746 (is_function_default_version): Likewise.
21747 * attribs.c (attr_strcmp): Likewise.
21748 (sorted_attr_string): Likewise.
21749 (common_function_versions): Likewise.
21750 (make_unique_name): Likewise.
21751 (make_dispatcher_decl): Likewise.
21752 (is_function_default_version): Likewise.
21753 * config/i386/i386.c (attr_strcmp): Likewise.
21754 (sorted_attr_string): Likewise.
21755 (ix86_function_versions): Likewise.
21756 (make_name): Likewise.
21757 (make_dispatcher_decl): Likewise.
21758 (is_function_default_version): Likewise.
21759 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
21761 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21764 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
21765 Account for direct move costs for vec_construct of integer
21768 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
21771 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
21772 (UNSPEC_STX_ATOMIC): Ditto.
21773 (loaddi_via_sse): New insn.
21774 (storedi_via_sse): Ditto.
21775 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
21776 Update corresponding peephole2 patterns.
21777 (atomic_storedi_fpu): Ditto.
21779 2017-05-11 Julia Koval <julia.koval@intel.com>
21781 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
21782 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
21784 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
21785 (__builtin_ia32_rsqrt14ss_mask): New builtins.
21786 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
21788 2017-05-11 Nathan Sidwell <nathan@acm.org>
21790 * graphite-poly.c: Include dumpfile.h.
21792 * dumpfle.h (dump_function): Declare here ...
21793 * tree-dump.h (dump_function): ... not here.
21794 * dumpfile.c: #include tree-cfg.h.
21795 (dump_function): Move here from ...
21796 * tree-dump.c (dump_function): ... here.
21797 * gimplify.c: #include splay-tree.h, not tree-dump.h.
21798 * graphite-poly.c: Don't include tree-dump.h.
21799 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
21800 * print-tree.c: Likewise.
21801 * stor-layout.c: Likewise.
21802 * tree-nested.c: Likewise.
21804 * dumpfile.c (dump_start): Use TDF_FLAGS.
21805 (dump_enable_all): Fix TDF_KIND check thinko.
21807 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
21809 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21810 array entries to represent two legal parameterizations of the
21811 overloaded __builtin_cmpb function, as represented by the
21812 P6_OV_BUILTIN_CMPB constant.
21813 (altivec_resolve_overloaded_builtin): Add special case handling
21814 for the __builtin_cmpb function, as represented by the
21815 P6_OV_BUILTIN_CMPB constant.
21816 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
21817 (BU_P6_64BIT_2): New macro.
21818 (BU_P6_OVERLOAD_2): New macro
21819 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
21820 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
21821 (CMPB): Add overload support to represent both 32-bit and 64-bit
21822 compare-bytes function.
21823 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
21824 support for TARGET_CMPB.
21825 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
21826 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
21827 documentation of the __builtin_cmpb overloaded built-in function.
21829 2017-05-11 Richard Biener <rguenther@suse.de>
21831 PR tree-optimization/80705
21832 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
21833 bases are not vectorizable.
21835 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21837 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
21838 when counting register pressure.
21840 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21842 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
21843 (iv_ca_more_deps): Renamed to ...
21844 (iv_ca_compare_deps): ... this.
21845 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
21847 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21849 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
21851 (determine_group_iv_costs): ... here.
21852 (find_inv_vars_cb): Record inv var if it's not recorded before.
21854 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21856 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
21857 (get_shiftadd_cost): Ditto.
21859 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21861 * tree-ssa-address.c: Include header file.
21862 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
21864 (add_to_parts): Refactor.
21865 (addr_to_parts): New parameter. Update use of move_hint_to_base.
21866 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
21869 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21871 PR tree-optimization/53090
21872 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
21874 (extract_cond_operands): Detect condition with IV on both sides
21875 and return COMP_IV_EXPR_2.
21876 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
21877 (rewrite_use_compare): Simplify by removing call to function
21878 extract_cond_operands.
21880 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21882 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
21883 (extract_cond_operands): Detect condition comparing against non-
21884 invariant bound and return appropriate enum value.
21885 (find_interesting_uses_cond): Update use of extract_cond_operands.
21886 Handle its return value accordingly.
21887 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
21889 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21891 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
21892 nonlinear iv_use computation in loop invariant sensitive way.
21894 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21896 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
21897 (find_iv_candidates): Call relate_compare_use_with_all_cands.
21899 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21901 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
21902 (dump_cand): Support iv_cand.inv_exprs.
21903 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
21905 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
21908 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21910 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
21912 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
21913 as local function. Include necessary header files.
21914 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
21916 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21918 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
21920 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21922 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
21923 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
21924 RSHIFT_EXPR and BIT_NOT_EXPR.
21926 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21928 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
21929 (adjust_setup_cost): New parameter supporting round up adjustment.
21930 (struct address_cost_data): Delete.
21931 (force_expr_to_var_cost): Don't bound cost with spill_cost.
21932 (split_address_cost, ptr_difference_cost): Delete.
21933 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
21934 (struct ainc_cost_data): New struct.
21935 (get_address_cost_ainc): New function.
21936 (get_address_cost, get_computation_cost): Reimplement.
21937 (determine_group_iv_cost_address): Record inv_expr for all uses of
21939 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
21940 (iv_ca_has_deps): Reimplemented to ...
21941 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
21943 (iv_ca_extend): Call iv_ca_more_deps.
21945 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21947 * tree-ssa-address.c (struct mem_address): Move to header file.
21948 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
21949 * tree-ssa-address.h (struct mem_address): Move from C file.
21950 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
21952 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21954 * tree-affine.h (aff_combination_type): New interface.
21955 (aff_combination_zero_p): Remove static.
21956 (aff_combination_const_p): New interface.
21957 (aff_combination_singleton_var_p): New interfaces.
21959 2017-05-11 Richard Biener <rguenther@suse.de>
21961 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
21962 Skip unreachable blocks and destinations.
21963 (eliminate): Move stmt removal and fixup ...
21964 (fini_eliminate): ... here. Skip inserted exprs.
21965 (pass_pre::execute): Move fini_pre after fini_eliminate.
21966 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
21967 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
21968 PRE to get rid of dead code that has invalid SSA form and
21969 split critical edges again.
21971 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21973 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
21975 2017-05-11 Richard Biener <rguenther@suse.de>
21977 * passes.c (execute_function_todo): Verify loops if they are
21978 said to be up-to-date.
21979 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
21980 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
21982 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
21985 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
21986 handle calling assemble_external ourself.
21989 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
21990 modes with zero size. Enhance comment.
21992 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21994 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
21995 built-ins for vec_xl and vec_xst with short and char pointer
21998 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
22000 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
22001 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
22002 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
22003 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
22004 (_mm_maskz_min_round_ss): New intrinsics.
22005 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
22006 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
22007 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
22008 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
22009 (__builtin_ia32_minss_mask_round): New builtins.
22010 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
22011 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
22012 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
22014 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
22015 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
22017 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
22020 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
22022 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
22023 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
22024 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
22025 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
22026 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
22027 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
22028 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
22029 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
22030 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
22031 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
22032 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
22033 (__builtin_ia32_mulss_mask_round): New builtins.
22034 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
22035 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
22036 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
22038 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
22039 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
22041 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
22044 2017-05-10 Julia Koval <julia.koval@intel.com>
22046 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
22047 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
22048 (_mm256_setr_m128i): New intrinsics.
22050 2017-05-10 Julia Koval <julia.koval@intel.com>
22052 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
22053 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
22054 (_mm_maskz_rcp14_ss): New intrinsics.
22055 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
22056 (__builtin_ia32_rcp14ss_mask): New builtins.
22057 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
22059 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
22061 PR tree-optimization/51513
22062 * tree-cfg.c (gimple_seq_unreachable_p): New function.
22063 (assert_unreachable_fallthru_edge_p): Use it.
22064 (group_case_labels_stmt): Likewise.
22065 * tree-cfg.h: Prototype it.
22066 * stmt.c: Include cfghooks.h and tree-cfg.h.
22067 (emit_case_dispatch_table) <gap_label>: New local variable.
22068 Use it to fill dispatch table gaps.
22069 Test for default_label before updating probabilities.
22070 (expand_case) <default_label>: Remove unneeded initialization.
22071 Test for unreachable default case statement and remove its edge.
22072 Set default_label accordingly.
22073 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
22075 2017-05-10 Carl Love <cel@us.ibm.com>
22077 * config/rs6000/rs6000-c: Add support for built-in functions
22078 vector signed char vec_neg (vector signed char)
22079 vector signed short int vec_neg (vector short int)
22080 vector signed int vec_neg (vector signed int)
22081 vector signed long long vec_neg (vector signed long long)
22082 vector float vec_neg (vector float)
22083 vector double vec_neg (vector double)
22084 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
22086 * config/rs6000/altivec.h: Add define for vec_neg
22087 * doc/extend.texi: Update the built-in documentation for the
22088 new built-in functions.
22090 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22092 PR tree-optimization/77644
22093 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
22095 2017-05-10 Nathan Sidwell <nathan@acm.org>
22097 * dumpfile.h (TDI_lang_all): New.
22098 (TDF_KIND): New. Renumber others
22099 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
22101 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
22103 (get_dump_file_name): Adjust suffix generation.
22104 (dump_enable_all): Use TDF_KIND.
22105 * doc/invoke.texi (-fdump-lang-all): Document.
22107 * dumpfile.h: Tabify.
22109 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
22112 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
22113 Move member access before delete.
22115 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
22117 * tree-inline.c (expand_call_inline): Split block at stmt
22120 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
22123 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
22124 are now unused after splitting mov{sf,sd}_hardfloat.
22125 (f32_lr2): Likewise.
22126 (f32_lm): Likewise.
22127 (f32_lm2): Likewise.
22128 (f32_li): Likewise.
22129 (f32_li2): Likewise.
22130 (f32_lv): Likewise.
22131 (f32_sr): Likewise.
22132 (f32_sr2): Likewise.
22133 (f32_sm): Likewise.
22134 (f32_sm2): Likewise.
22135 (f32_si): Likewise.
22136 (f32_si2): Likewise.
22137 (f32_sv): Likewise.
22138 (f32_dm): Likewise.
22139 (f32_vsx): Likewise.
22140 (f32_av): Likewise.
22141 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
22142 For movsf, order stores so the VSX stores occur before the GPR
22143 store which encourages the register allocator to use a traditional
22144 FPR instead of a GPR. For movsd, order the stores so that the GPR
22145 store comes before the VSX stores to allow the power6 to work.
22146 This is due to the power6 not having a 32-bit integer store
22147 instruction from a FPR.
22148 (movsf_hardfloat): Likewise.
22149 (movsd_hardfloat): Likewise.
22151 2017-05-09 Martin Sebor <msebor@redhat.com>
22153 PR translation/80280
22154 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
22157 PR translation/80280
22158 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
22159 data member added in r247778.
22160 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
22162 2017-05-09 Nathan Sidwell <nathan@acm.org>
22164 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
22166 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
22169 2017-05-09 Marek Polacek <polacek@redhat.com>
22171 * doc/invoke.texi: Fix typo.
22173 2017-05-09 Richard Biener <rguenther@suse.de>
22175 * tree-vrp.c (vrp_val_is_max): Adjust comment.
22176 (vrp_val_is_min): Likewise.
22177 (set_value_range_to_value): Likewise.
22178 (set_value_range_to_nonnegative): Likewise.
22179 (gimple_assign_nonzero_p): Likewise.
22180 (gimple_stmt_nonzero_p): Likewise.
22181 (vrp_int_const_binop): Likewise. Remove unreachable case.
22182 (adjust_range_with_scev): Adjust comments.
22183 (compare_range_with_value): Likewise.
22184 (extract_range_from_phi_node): Likewise.
22185 (test_for_singularity): Likewise.
22187 2017-05-09 Richard Biener <rguenther@suse.de>
22189 * tree-vrp.c (get_single_symbol): Add assert that we don't
22190 get overflowed constants as invariant part.
22191 (compare_values_warnv): Add comment before the TREE_NO_WARNING
22192 checks. Use wi::cmp instead of recursing for integer constants.
22193 (compare_values): Just ignore whether we assumed undefined
22194 overflow instead of failing the compare.
22195 (extract_range_for_var_from_comparison_expr): Add comment before the
22196 TREE_NO_WARNING sets.
22197 (test_for_singularity): Likewise.
22198 (extract_range_from_comparison): Do not disable optimization
22199 when we assumed undefined overflow.
22200 (extract_range_basic): Remove init of unused var.
22202 2017-05-09 Richard Biener <rguenther@suse.de>
22204 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
22205 (extract_range_from_multiplicative_op_1): Adjust.
22206 (extract_range_from_binary_expr_1): Use int_const_binop.
22208 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
22211 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
22212 rs6000_store_data_bypass_p in seven define_bypass directives and
22213 in several comments.
22214 * config/rs6000/rs6000-protos.h: Add prototype for
22215 rs6000_store_data_bypass_p function.
22216 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
22217 function implements slightly different (rs6000-specific) semantics
22218 than store_data_bypass_p, returning false rather than aborting
22219 with assertion error when arguments do not satisfy the
22220 requirements of store data bypass.
22221 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
22222 rs6000_store_data_bypass_p.
22224 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
22226 * config/xtensa/xtensa-protos.h
22227 (xtensa_initial_elimination_offset): New declaration.
22228 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
22229 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
22230 macro definition, add case for FRAME_POINTER_REGNUM when
22231 FRAME_GROWS_DOWNWARD.
22232 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
22233 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
22234 xtensa_initial_elimination_offset.
22236 2017-05-08 Nathan Sidwell <nathan@acm.org>
22238 * doc/invoke.texi: Alphabetize -fdump options.
22240 2017-05-08 Martin Sebor <msebor@redhat.com>
22242 PR translation/80280
22243 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
22245 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
22247 * target.def (compute_frame_layout): New optional target hook.
22248 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
22249 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
22250 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
22252 * reload1.c (verify_initial_elim_offsets): Likewise.
22253 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
22254 (use_simple_return_p): Call arm_compute_frame_layout if needed.
22255 (arm_get_frame_offsets): Split up into this ...
22256 (arm_compute_frame_layout): ... and this function.
22258 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
22260 * config/aarch64/constraints.md (Usa): New constraint.
22261 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
22263 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
22265 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
22266 with_multilib_list after it has been checked.
22268 2017-05-08 Richard Biener <rguenther@suse.de>
22270 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
22271 (bitmap_set_subtract_values): Likewise.
22273 2017-05-08 Richard Biener <rguenther@suse.de>
22275 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
22276 (gimple_assign_nonzero): ... this and remove strict_overflow_p
22278 (gimple_stmt_nonzero_warnv_p): Rename to ...
22279 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
22281 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
22282 (extract_range_basic): Adjust, do not disable propagation on
22283 strict overflow sensitive simplification.
22284 (vrp_visit_cond_stmt): Likewise.
22286 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
22288 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
22289 body size unconditionally.
22291 2017-05-07 Jeff Law <law@redhat.com>
22294 2017-05-06 Jeff Law <law@redhat.com>
22295 PR tree-optimization/78496
22296 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
22299 PR tree-optimization/78496
22300 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
22301 (simplify_stmt_using_ranges): Call it.
22302 (vrp_dom_walker::before_dom_children): Extract equivalences
22303 from an ASSERT_EXPR with an equality comparison against a
22306 2017-05-06 Jeff Law <law@redhat.com>
22308 PR tree-optimization/78496
22309 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
22312 PR tree-optimization/78496
22313 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
22314 (simplify_stmt_using_ranges): Call it.
22315 (vrp_dom_walker::before_dom_children): Extract equivalences
22316 from an ASSERT_EXPR with an equality comparison against a
22319 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
22321 * lra-constraints.c (lra_copy_reg_equiv): New function.
22322 (split_reg): Use it to copy equivalence information from the
22323 original register to the spill register.
22325 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
22327 PR rtl-optimization/75964
22328 * simplify-rtx.c (simplify_const_relational_operation): Remove
22329 invalid handling of comparisons of integer ABS.
22331 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
22333 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
22334 initialize to zero.
22335 (init_regs): Remove declaration.
22336 (function_arg_advance_32): Initialize error_p as boolean variable.
22338 2017-05-05 Nathan Sidwell <nathan@acm.org>
22340 * store-motion.c (remove_reachable_equiv_notes): Reformat long
22341 lines. Use for (;;).
22343 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22345 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
22346 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
22347 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
22348 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
22349 VF=2 that require versioning.
22351 2017-05-05 David Malcolm <dmalcolm@redhat.com>
22353 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
22356 2017-05-05 David Malcolm <dmalcolm@redhat.com>
22358 * diagnostic.h (diagnostic_override_option_index): Convert from
22359 macro to inline function.
22361 2017-05-05 David Malcolm <dmalcolm@redhat.com>
22363 * diagnostic.c (last_module_changed_p): New function.
22364 (set_last_module): New function.
22365 (diagnostic_report_current_module): Convert macro usage to
22366 the above functions.
22367 * diagnostic.h (diagnostic_context::last_module): Strengthen
22368 from const line_map * to const line_map_ordinary *.
22369 (diagnostic_last_module_changed): Delete macro.
22370 (diagnostic_set_last_module): Delete macro.
22372 2017-05-05 David Malcolm <dmalcolm@redhat.com>
22374 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
22375 with diagnostic_report_diagnostic.
22376 (diagnostic_n_impl_richloc): Likewise.
22377 * diagnostic.h (report_diagnostic): Delete macro.
22378 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
22379 with diagnostic_report_diagnostic.
22380 * substring-locations.c (format_warning_va): Likewise.
22382 2017-05-05 David Malcolm <dmalcolm@redhat.com>
22384 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
22385 save/restor of format_spec. Move option-printing code to...
22386 (print_option_information): ...this new function, and
22387 reimplement by simply printing to the pretty_printer,
22388 rather than appending to the format string.
22390 2017-05-05 David Malcolm <dmalcolm@redhat.com>
22392 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
22393 handling logic into...
22394 (update_effective_level_from_pragmas): ...this new function.
22396 2017-05-04 Andrew Waterman <andrew@sifive.com>
22398 * config/riscv/riscv.opt (mstrict-align): New option.
22399 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
22400 (SLOW_UNALIGNED_ACCESS): Define.
22401 (riscv_slow_unaligned_access): Declare.
22402 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
22404 (riscv_slow_unaligned_access): New variable.
22405 (rocket_tune_info): Set slow_unaligned_access to true.
22406 (optimize_size_tune_info): Set slow_unaligned_access to false.
22407 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
22408 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
22409 (riscv_option_override): Set riscv_slow_unaligned_access.
22410 * doc/invoke.texi: Add -mstrict-align to RISC-V.
22412 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
22414 * config/riscv/riscv.md: Unify indentation.
22416 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22421 * config/rs6000/rs6000.md (u code attribute): Add FIX and
22423 (extendsi<mode>2): Add support for doing sign extension via
22424 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
22425 don't have ISA 3.0 instructions.
22426 (extendsi<mode>2 splitter): Likewise.
22427 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
22428 generate the normal insns since SImode can now go in vector
22429 registers. Disallow the special UNSPECs needed for previous
22430 machines to hide SImode being used. Add new insns
22431 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
22432 (fix_trunc<mode>si2_stfiwx): Likewise.
22433 (fix_trunc<mode>si2_internal): Likewise.
22434 (fixuns_trunc<mode>si2): Likewise.
22435 (fixuns_trunc<mode>si2_stfiwx): Likewise.
22436 (fctiw<u>z_<mode>_smallint): Likewise.
22437 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
22438 of floating point to 32-bit integer from doing a direct move to
22439 the GPR registers to do a store.
22440 (fctiwz_<mode>): Break long line.
22442 2017-05-05 Bin Cheng <bin.cheng@arm.com>
22444 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
22445 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
22446 (addr_list, addr_offset_valid_p): New.
22447 (split_address_groups): Check offset validity with above function.
22448 (gt-tree-ssa-loop-ivopts.h): Include header file.
22450 2017-05-05 Nathan Sidwell <nathan@acm.org>
22452 * config.gcc (arm*-*-*): Add missing 'fi'.
22454 2017-05-05 Steve Ellcey <sellcey@cavium.com>
22456 * doc/invoke.texi (-fopt-info): Explicitly say order of options
22457 included in -fopt-info does not matter.
22458 * doc/optinfo.texi (-fopt-info): Fix description of default
22459 behavour. Explicitly say order of options included in -fopt-info
22462 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
22464 * config.gcc: Allow combinations of aprofile and rmprofile values for
22465 --with-multilib-list.
22466 * config/arm/t-multilib: New file.
22467 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
22468 variables. Remove setting of ISA and floating-point ABI in
22469 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
22470 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
22471 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
22472 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
22474 * config/arm/t-rmprofile: Likewise except for the matches changes.
22475 * doc/install.texi (--with-multilib-list): Document the combination of
22476 aprofile and rmprofile values and warn about pitfalls in doing that.
22478 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
22480 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
22481 (movdi_aarch64): Likewise.
22483 2017-05-05 Jakub Jelinek <jakub@redhat.com>
22485 PR tree-optimization/80632
22486 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
22488 (build_arrays): Initialize it for virtual phis.
22489 (fix_phi_nodes): Use it for virtual phis.
22491 PR tree-optimization/80558
22492 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
22493 [x, y] op z into [x op, y op z] for op & or | if conditions
22496 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
22497 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
22500 * config/arm/arm.md (use_literal_pool): Remove.
22501 (64-bit immediate split): No longer takes cost into consideration
22502 if arm_disable_literal_pool is enabled.
22503 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
22504 used when arm_disable_literal_pool is enabled.
22505 (arm_max_const_double_inline_cost): Remove use of
22506 arm_disable_literal_pool.
22507 (push_minipool_fix): Add assert.
22508 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
22509 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
22510 (no_literal_pool_sf_immediate): New.
22512 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22514 PR tree-optimization/80613
22515 * tree-ssa-dce.c (propagate_necessity): Remove cases for
22516 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
22518 2017-05-05 Richard Biener <rguenther@suse.de>
22520 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
22522 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
22524 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
22525 of this flag from insn conditions due to removal from r247495.
22527 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
22529 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
22531 (arm_early_store_addr_dep_ptr): Likewise.
22532 * config/arm/aarch-common-protos.h
22533 (arm_early_load_addr_dep_ptr): Add prototype.
22534 (arm_early_store_addr_dep_ptr): Likewise.
22535 * config/arm/cortex-a53.md: Add new bypasses.
22537 2017-05-05 Jakub Jelinek <jakub@redhat.com>
22539 * tree.c (next_type_uid): Change type to unsigned.
22540 (type_hash_canon): Decrement back next_type_uid if
22541 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
22542 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
22545 2017-05-04 Martin Sebor <msebor@redhat.com>
22547 * builtins.c: Fix a trivial typo in a comment.
22549 PR middle-end/79234
22550 * builtins.c (check_sizes): Adjust to handle reading past the end.
22551 Avoid printing excessive upper bound of ranges. Use %E to print
22552 tree nodes instead of converting them to %wu.
22553 (expand_builtin_memchr): New function.
22554 (compute_dest_size): Rename...
22555 (compute_objsize): ...to this.
22556 (expand_builtin_memcpy): Adjust.
22557 (expand_builtin_mempcpy): Adjust.
22558 (expand_builtin_strcat): Adjust.
22559 (expand_builtin_strcpy): Adjust.
22560 (check_strncat_sizes): Adjust.
22561 (expand_builtin_strncat): Adjust.
22562 (expand_builtin_strncpy): Adjust and simplify.
22563 (expand_builtin_memset): Adjust.
22564 (expand_builtin_bzero): Adjust.
22565 (expand_builtin_memcmp): Adjust.
22566 (expand_builtin): Handle memcmp.
22567 (maybe_emit_chk_warning): Check strncat just once.
22569 2017-05-04 Martin Sebor <msebor@redhat.com>
22571 PR preprocessor/79214
22572 PR middle-end/79222
22573 PR middle-end/79223
22574 * builtins.c (check_sizes): Add inlining context and issue
22575 warnings even when -Wno-system-headers is set.
22576 (check_strncat_sizes): Same.
22577 (expand_builtin_strncat): Same.
22578 (expand_builtin_memmove): New function.
22579 (expand_builtin_stpncpy): Same.
22580 (expand_builtin): Handle memmove and stpncpy.
22582 2017-05-04 Bin Cheng <bin.cheng@arm.com>
22584 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
22585 which is not used any more.
22587 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
22589 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
22591 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
22593 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
22594 (cortexa53_tunings): Likewise.
22595 (cortexa57_tunings): Likewise.
22596 (cortexa72_tunings): Likewise.
22597 (cortexa73_tunings): Likewise.
22599 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
22601 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
22602 Set loop alignment to 8.
22604 2017-05-04 Martin Sebor <msebor@redhat.com>
22606 PR translation/80280
22607 * builtins.c (expand_builtin_object_size): Add missing quoting to
22608 %D and like directives.
22609 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
22610 (hsa_type_for_tree_type): Same.
22611 (verify_function_arguments): Same.
22612 * symtab.c (symbol_table::change_decl_assembler_name): Same.
22613 * varasm.c (get_section): Same.
22616 2017-05-04 Martin Sebor <msebor@redhat.com>
22618 PR translation/80280
22619 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
22621 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
22623 * config/aarch64/aarch64.c (generic_addrcost_table):
22624 Change HI/TI mode setting.
22626 2017-05-04 Martin Jambor <mjambor@suse.cz>
22628 PR tree-optimization/80622
22629 * tree-sra.c (comes_initialized_p): New function.
22630 (build_accesses_from_assign): Only set write lazily when
22631 comes_initialized_p is false.
22632 (analyze_access_subtree): Use comes_initialized_p.
22633 (propagate_subaccesses_across_link): Assert !comes_initialized_p
22634 instead of testing for PARM_DECL.
22636 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22638 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
22639 constraint on operand 0 to allow more general addressing modes.
22640 Adjust output template.
22641 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
22643 * config/aarch64/aarch64-protos.h
22644 (aarch64_address_valid_for_prefetch_p): Declare prototype.
22645 * config/aarch64/constraints.md (Dp): New address constraint.
22646 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
22649 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
22651 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
22652 update use of estimate_ipcp_clone_size_and_time.
22653 (estimate_local_effects): Update use of
22654 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
22655 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
22656 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
22657 Return nonspecialized time.
22659 2017-05-04 Richard Biener <rguenther@suse.de>
22661 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
22662 for the last VUSE which def dominates the PHI. Directly call
22664 (get_continuation_for_phi_1): Remove.
22666 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
22668 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
22669 to explain the use of truncating division. Cap the number of
22670 iterations to the maximum given by nb_iterations_upper_bound,
22673 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
22675 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
22676 * configure: Regenerate.
22677 * config.in: Regenerate.
22678 * config/i386/driver-mingw32.c: new file.
22679 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
22680 * config.host: Link driver-mingw32.o on MinGW host.
22681 * doc/install.texi: Document new --enable-mingw-wildcard configure
22684 2017-05-04 Marek Polacek <polacek@redhat.com>
22686 PR tree-optimization/80612
22687 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
22689 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
22690 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
22692 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
22693 (movt splitter): Likewise.
22694 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
22695 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
22696 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
22697 block for Thumb-1 with MOVT.
22698 (thumb2_legitimate_address_p): Move code block ...
22699 (can_avoid_literal_pool_for_label_p): ... into this new function.
22700 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
22702 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
22703 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
22704 "M-profile targets with the MOVT instruction".
22706 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
22708 * config/arm/arm-builtins.c (arm_init_builtins): Rename
22709 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
22710 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
22712 2017-05-04 Martin Liska <mliska@suse.cz>
22714 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
22715 variable cond_code.
22717 2017-05-04 Richard Biener <rguenther@suse.de>
22719 * tree.c (array_at_struct_end_p): Handle arrays at struct
22720 end with flexarrays more conservatively. Refactor and treat
22721 arrays of arrays or aggregates more strict. Fix
22722 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
22723 * tree.c (array_at_struct_end_p): Adjust prototype.
22724 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
22725 * gimple-fold.c (get_range_strlen): Likewise.
22726 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
22728 2017-05-04 Richard Biener <rguenther@suse.de>
22730 PR tree-optimization/31130
22731 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
22733 (supports_overflow_infinity): Likewise.
22734 (is_negative_overflow_infinity): Likewise.
22735 (is_positive_overflow_infinity): Likewise.
22736 (is_overflow_infinity): Likewise.
22737 (stmt_overflow_infinity): Likewise.
22738 (overflow_infinity_range_p): Likewise.
22739 (usable_range_p): Remove as always returning true.
22740 (make_overflow_infinity): Remove.
22741 (negative_overflow_infinity): Likewise.
22742 (positive_overflow_infinity): Likewise.
22743 (avoid_overflow_infinity): Likewise.
22744 (set_value_range): Adjust accordingly.
22745 (set_value_range_to_nonnegative): Likewise, remove now unused
22746 overflow_infinity arg.
22747 (vrp_operand_equal_p): Adjust.
22748 (update_value_range): Likewise.
22749 (range_int_cst_singleton_p): Likewise.
22750 (operand_less_p): Likewise.
22751 (compare_values_warnv): Likewise.
22752 (extract_range_for_var_from_comparison_expr): Likewise.
22753 (vrp_int_const_binop): Likewise.
22754 (zero_nonzero_bits_from_vr): Likewise.
22755 (extract_range_from_multiplicative_op_1): Likewise.
22756 (extract_range_from_binary_expr_1): Likewise.
22757 (extract_range_from_unary_expr): Likewise.
22758 (extract_range_from_comparison): Likewise.
22759 (extract_range_basic): Likewise.
22760 (adjust_range_with_scev): Likewise.
22761 (compare_ranges): Likewise.
22762 (compare_range_with_value): Likewise.
22763 (dump_value_range): Likewise.
22764 (test_for_singularity): Likewise, remove strict_overflow_p parameter
22766 (simplify_cond_using_ranges): Adjust.
22768 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
22770 * brig-builtins.def: Added a builtin for class_f64.
22771 * builtin-types.def: Added a builtin type needed by class_f64.
22773 2017-05-03 Jason Merrill <jason@redhat.com>
22775 * timevar.def: Add TV_CONSTEXPR.
22777 2017-05-03 David Malcolm <dmalcolm@redhat.com>
22779 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
22781 2017-05-03 Martin Jambor <mjambor@suse.cz>
22783 * ipa-prop.c (ipa_update_after_lto_read): Removed.
22784 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
22785 * ipa-cp.c (ipcp_propagate_stage): Do not call
22786 ipa_update_after_lto_read.
22787 * ipa-inline.c (ipa_inline): Likewise.
22789 2017-05-03 Martin Jambor <mjambor@suse.cz>
22791 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
22792 tag. Added a default constructor and a destructor.
22793 (ipa_edge_args_sum_t): New class;
22794 (ipa_edge_args_sum): Declare.
22795 (ipa_edge_args_vector): Remove declaration.
22796 (IPA_EDGE_REF): Use ipa_edge_args_sum.
22797 (ipa_free_edge_args_substructures): Remove declaration.
22798 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
22799 (ipa_edge_args_info_available_for_edge_p): Likewise.
22800 * ipa-prop.c (ipa_edge_args_vector): Removed.
22801 (edge_removal_hook_holder): Likewise.
22802 (edge_duplication_hook_holder): Likewise.
22803 (ipa_edge_args_sum): New variable.
22804 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
22805 ipa_edge_args_vector.
22806 (ipa_free_edge_args_substructures): Likewise.
22807 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
22808 ipa_edge_args_vector.
22809 (ipa_edge_removal_hook): Turned into method
22810 ipa_edge_args_sum_t::remove.
22811 (ipa_edge_duplication_hook): Turned into method
22812 ipa_edge_args_sum_t::duplicate.
22813 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
22814 registering edge hooks.
22815 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
22816 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
22817 ipa_edge_args_sum instead of ipa_edge_args_vector.
22818 * ipa-profile.c (ipa_profile): Likewise.
22820 2017-05-03 Martin Jambor <mjambor@suse.cz>
22822 * symbol-summary.h (function_summary): New method exists.
22823 (function_summary::symtab_removal): Deallocate through release.
22824 (call_summary): New class.
22825 (gt_ggc_mx): New overload.
22826 (gt_pch_nx): Likewise.
22827 (gt_pch_nx): Likewise.
22829 2017-05-03 Jeff Law <law@redhat.com>
22831 PR tree-optimization/78496
22832 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
22833 from simplify_cond_using_ranges. Split off code to walk
22834 backwards through casts into ...
22835 (simplify_cond_using_ranges_2): New function.
22836 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
22837 (execute_vrp): After identifying jump threads, call
22838 simplify_cond_using_ranges_2.
22840 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
22843 * ipa-inline.h (inline_summary): Add ctor.
22844 (create_ggc): Do not use ggc_cleared_alloc.
22846 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
22847 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22849 * gcc.c (handle_braces): Support escaping in switch matching
22851 * doc/invoke.texi (Spec Files): Document it.
22852 Remove superfluous @code markup in items.
22854 2017-05-03 David Malcolm <dmalcolm@redhat.com>
22856 * diagnostic-show-locus.c (struct column_range): New struct.
22857 (get_affected_columns): New function.
22858 (get_printed_columns): New function.
22859 (struct correction): New struct.
22860 (correction::ensure_capacity): New function.
22861 (correction::ensure_terminated): New function.
22862 (struct line_corrections): New struct.
22863 (line_corrections::~line_corrections): New dtor.
22864 (line_corrections::add_hint): New function.
22865 (layout::print_trailing_fixits): Reimplement in terms of the new
22867 (selftest::test_overlapped_fixit_printing): New function.
22868 (selftest::diagnostic_show_locus_c_tests): Call it.
22870 2017-05-03 Nathan Sidwell <nathan@acm.org>
22872 Canonicalize canonical type hashing
22873 * tree.h (type_hash_canon_hash): Declare.
22874 * tree.c (type_hash_list, attribute_hash_list): Move into
22875 type_hash_canon_hash.
22876 (build_type_attribute_qual_variant): Break out hash code calc into
22877 type_hash_canon_hash.
22878 (type_hash_canon_hash): New. Generic type hash computation.
22879 (build_range_type_1, build_array_type_1, build_function_type,
22880 build_method_type_directly, build_offset_type, build_complex_type,
22881 make_vector_type): Call it.
22883 2017-05-03 Richard Biener <rguenther@suse.de>
22885 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
22886 When all DRs have unknown misaligned do not always peel
22887 when there is a store but apply the same costing model as if
22888 there were only loads.
22890 2017-05-03 Richard Biener <rguenther@suse.de>
22893 PR tree-optimization/80492
22894 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
22895 compare_base_decls returning dont-know properly.
22897 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
22899 * config/arm/iterators.md (CCSI): New mode iterator.
22900 (arch): New mode attribute.
22901 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
22902 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
22903 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
22904 code iterator for success result mode.
22905 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
22906 the corresponding new insn generators.
22908 2017-05-03 Bin Cheng <bin.cheng@arm.com>
22911 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22912 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
22914 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
22916 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
22917 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
22918 (DDR_A): Wrap DDR argument in brackets.
22919 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
22920 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
22921 (DDR_REVERSED_P): Likewise.
22923 2017-05-03 Jakub Jelinek <jakub@redhat.com>
22925 PR tree-optimization/79472
22926 * tree-switch-conversion.c (struct switch_conv_info): Add
22927 contiguous_range and default_case_nonstandard fields.
22928 (collect_switch_conv_info): Compute contiguous_range and
22929 default_case_nonstandard fields, don't clear final_bb if
22930 contiguous_range and only the default case doesn't have the required
22932 (check_all_empty_except_final): Set default_case_nonstandard instead
22933 of failing if contiguous_range and the default case doesn't have empty
22935 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
22936 and only the default case doesn't have the required constants. Skip
22938 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
22939 if default_case_nonstandard.
22940 (build_constructors): Build constant 1 just once. Assert that default
22941 values aren't inserted in between cases if contiguous_range. Skip
22943 (build_arrays): Skip virtual phis.
22944 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
22945 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
22946 Handle virtual phis.
22947 (gen_inbound_check): Handle default_case_nonstandard case.
22948 (process_switch): Adjust check_final_bb caller. Call
22949 gather_default_values with the first non-default case instead of
22950 default case if default_case_nonstandard.
22952 2017-05-02 Nathan Sidwell <nathan@acm.org>
22954 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
22955 check. Fix formatting.
22957 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
22959 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
22960 errors when comparing specialized and unspecialized times.
22962 2017-05-02 David Malcolm <dmalcolm@redhat.com>
22964 * diagnostic-show-locus.c
22965 (layout::should_print_annotation_line_p): Make private.
22966 (layout::print_annotation_line): Make private.
22967 (layout::annotation_line_showed_range_p): Make private.
22968 (layout::show_ruler): Make private.
22969 (layout::print_source_line): Make private. Pass in line and
22970 line_width, rather than calling location_get_source_line. Drop
22972 (layout::print_leading_fixits): New method.
22973 (layout::print_any_fixits): Rename to...
22974 (layout::print_trailing_fixits): ...this, and make private.
22975 Don't print newline fixits.
22976 (diagnostic_show_locus): Move logic for printing one row into...
22977 (layout::print_line): ...this new function. Move the
22978 location_get_source_line call and error-handling from
22979 print_source_line to here. Call print_leading_fixits, and rename
22980 print_any_fixits to print_trailing_fixits.
22981 (selftest::test_fixit_insert_containing_newline): Update now that
22982 newlines are partially supported.
22983 (selftest::test_fixit_insert_containing_newline_2): New test.
22984 (selftest::test_fixit_replace_containing_newline): Update comments.
22985 (selftest::diagnostic_show_locus_c_tests): Call the new test.
22986 * edit-context.c (class added_line): New class.
22987 (class edited_line): Describe newline handling in comment.
22988 (edited_line::actually_edited_p): New method.
22989 (edited_line::print_content): Delete redundant decl.
22990 (edited_line::m_predecessors): New field.
22991 (edited_file::print_content): Call edited_line::print_content.
22992 (edited_file::print_diff): Update to support newlines.
22993 (edited_file::print_diff_hunk): Likewise.
22994 (edited_file::print_run_of_changed_lines): New function.
22995 (edited_file::print_diff_line): Convert to...
22996 (print_diff_line): ...this.
22997 (edited_file::get_effective_line_count): New function.
22998 (edited_line::edited_line): Initialize new field m_predecessors.
22999 (edited_line::~edited_line): Clean up m_predecessors.
23000 (edited_line::apply_fixit): Handle newlines.
23001 (edited_line::get_effective_line_count): New function.
23002 (edited_line::print_content): New function.
23003 (edited_line::print_diff_lines): New function.
23004 (selftest::test_applying_fixits_insert_containing_newline): New
23006 (selftest::test_applying_fixits_replace_containing_newline): New
23008 (selftest::insert_line): New function.
23009 (selftest::test_applying_fixits_multiple_lines): Add example of
23011 (selftest::edit_context_c_tests): Call the new tests.
23013 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23015 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
23016 parameter cand. Update dump information.
23017 (get_computation_cost): Update uses.
23019 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23021 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
23022 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
23023 (get_computation_at, rewrite_use_address): Update use of
23024 get_computation_aff.
23026 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23028 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
23029 (get_computation): Delete.
23030 (get_computation_cost): Implement like get_computation_cost_at.
23031 Use get_computation_at.
23032 (get_computation_cost_at): Delete.
23033 (rewrite_use_nonlinear_expr): Use get_computation_at.
23034 (rewrite_use_compare, remove_unused_ivs): Ditto.
23036 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23038 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
23040 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23042 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
23043 (ivopts_global_cost_for_size): Rename parameter and update uses.
23044 (iv_ca_recount_cost): Update uses.
23045 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
23046 candidates seperately in n_invs and n_cands.
23047 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
23049 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23051 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
23052 (find_inv_vars_cb): New.
23053 (find_depends): Renamed to ...
23054 (find_inv_vars): ... this.
23055 (add_candidate_1, force_var_cost): Call find_inv_vars.
23056 (split_address_cost, determine_group_iv_cost_cond): Ditto.
23058 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23060 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
23061 inv_vars. Add inv_exprs.
23062 (struct iv_cand): Rename depends_on to inv_vars.
23063 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
23064 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
23065 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
23066 (dump_cand): Dump inv_vars.
23067 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
23068 (record_invariant, find_depends, add_candidate_1): Ditto.
23069 (set_group_iv_cost, force_var_cost): Ditto.
23070 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
23071 (get_computation_cost_at, get_computation_cost): Ditto.
23072 (determine_group_iv_cost_generic): Ditto.
23073 (determine_group_iv_cost_address): Ditto.
23074 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
23075 (determine_group_iv_costs): Ditto.
23076 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
23077 (iv_ca_set_remove_invariants): Renamed to ...
23078 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
23079 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
23080 (iv_ca_set_add_invariants): Renamed to ...
23081 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
23082 (iv_ca_set_cp): Use iv_ca_set_add_invs.
23083 (iv_ca_has_deps): Support inv_vars and inv_exprs.
23084 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
23085 (create_new_ivs): Remove useless dump.
23087 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23089 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
23091 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
23092 (iv_ca_set_no_cp, create_new_iv): Ditto.
23094 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23096 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
23098 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
23100 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
23101 function tree_check2.
23103 2017-05-02 Martin Liska <mliska@suse.cz>
23105 * doc/gcov.texi: Add missing preposition.
23106 * gcov.c (function_info::function_info): Properly fill up
23107 all member variables.
23109 2017-05-02 Tamar Christina <tamar.christina@arm.com>
23111 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
23113 2017-05-02 Tamar Christina <tamar.christina@arm.com>
23115 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
23117 2017-05-02 Martin Liska <mliska@suse.cz>
23120 * lto-streamer-in.c (lto_read_tree_1): Remove
23121 LTO_STREAMER_DEBUG.
23122 * lto-streamer.c (struct tree_hash_entry): Likewise.
23123 (struct tree_entry_hasher): Likewise.
23124 (tree_entry_hasher::hash): Likewise.
23125 (tree_entry_hasher::equal): Likewise.
23126 (lto_streamer_init): Likewise.
23127 (lto_orig_address_map): Likewise.
23128 (lto_orig_address_get): Likewise.
23129 (lto_orig_address_remove): Likewise.
23130 * lto-streamer.h: Likewise.
23131 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
23132 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
23134 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
23136 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
23137 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
23138 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
23139 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
23140 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
23141 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
23142 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
23143 (mm_maskz_sub_ss): New intrinsics.
23144 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23145 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
23146 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
23147 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
23148 (__builtin_ia32_subss_mask_round): New builtins.
23149 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23150 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
23151 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
23153 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
23154 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
23156 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
23159 2017-05-02 Martin Jambor <mjambor@suse.cz>
23161 PR tree-optimization/78687
23162 * tree-sra.c (access): New field parent.
23163 (process_subtree_disqualification): New function.
23164 (disqualify_candidate): Call it.
23165 (build_accesses_from_assign): Reset write flag if creating an
23167 (build_access_subtree): Fill in parent field and also prpagate
23168 down grp_write flag.
23169 (create_artificial_child_access): New parameter set_grp_write, set
23170 grp_write to its value.
23171 (propagate_subaccesses_across_link): Also propagate grp_write flag
23173 (propagate_all_subaccesses): Push the closest parent back to work
23174 queue if add_access_to_work_queue returned true.
23176 2017-05-02 Richard Biener <rguenther@suse.de>
23178 * common.opt (fstrict-overflow): Alias negative to fwrapv.
23179 * doc/invoke.texi (fstrict-overflow): Remove all traces of
23180 -fstrict-overflow documentation.
23181 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
23182 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
23183 flag_strict_overflow.
23184 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
23185 * lto-opts.c (lto_write_options): Do not stream it.
23186 * lto-wrapper.c (merge_and_complain): Do not handle it.
23187 * opts.c (default_options_table): Do not set -fstrict-overflow.
23188 (finish_options): Likewise do not clear it when sanitizing.
23189 * simplify-rtx.c (simplify_const_relational_operation): Do not
23190 test flag_strict_overflow.
23192 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
23194 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
23195 using enabled attribute.
23196 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
23197 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
23198 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
23199 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
23200 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
23201 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
23202 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
23203 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
23204 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
23205 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
23207 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
23209 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
23211 2017-05-02 Richard Biener <rguenther@suse.de>
23213 PR tree-optimization/80591
23215 2017-04-10 Richard Biener <rguenther@suse.de>
23217 * tree-ssa-structalias.c (find_func_aliases): Properly handle
23220 2017-05-02 Richard Biener <rguenther@suse.de>
23222 PR tree-optimization/80549
23223 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
23224 (cleanup_tree_cfg_noloop): Create forwarders to known loop
23225 headers if they do not have a preheader.
23227 2017-05-02 Martin Liska <mliska@suse.cz>
23230 * common.opt: Fix typo.
23231 * doc/invoke.texi: Likewise.
23233 2017-05-01 Jan Beulich <jbeulich@suse.com>
23235 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
23236 swapping, add (x,x,m,x,n) alternative.
23238 2017-05-01 Nathan Sidwell <nathan@acm.org>
23240 * calls.c (combine_pending_stack_adjustment_and_call): Remove
23241 unnecessary unadjusted_alignment check.
23243 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
23246 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
23248 * gimplify.c (gimplify_cilk_detach): New function.
23249 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
23250 * tree-core.h: Document EXPR_CILK_SPAWN.
23251 * tree.h (EXPR_CILK_SPAWN): Define.
23253 2017-05-01 David Malcolm <dmalcolm@redhat.com>
23255 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
23256 to use new fixit_hint representation, using the "replace" logic.
23257 (get_line_span_for_fixit_hint): Likewise.
23258 (layout::print_any_fixits): Likewise.
23259 (selftest::test_one_liner_many_fixits): Rename to...
23260 (selftest::test_one_liner_many_fixits_1): ...this, and update
23261 comment and expected output to reflect that the multiple fix-it
23262 hints are now consolidated into one insertion.
23263 (selftest::test_one_liner_many_fixits_2): New test.
23264 (selftest::test_diagnostic_show_locus_one_liner): Update for
23266 (selftest::test_fixit_consolidation): Update for fix-it API
23268 * diagnostic.c (print_parseable_fixits): Likewise.
23269 * edit-context.c (edited_line::m_line_events): Convert from
23270 auto_vec <line_event *> to auto_vec <line_event>.
23271 (class line_event): Convert from abstract base class to a concrete
23272 class, taking over the role of replace_event.
23273 (class insert_event): Delete.
23274 (class replace_event): Rename to class line_event. Convert to
23276 (edit_context::add_fixits): Reimplement.
23277 (edit_context::apply_insert): Delete.
23278 (edit_context::apply_replace): Rename to...
23279 (edit_context::apply_fixit): ...this. Convert to half-open range.
23280 (edited_file::apply_insert): Delete.
23281 (edited_file::apply_replace): Rename to...
23282 (edited_file::apply_fixit): ...this.
23283 (edited_line::~edited_line): Drop deletion of events.
23284 (edited_line::apply_insert): Delete.
23285 (edited_line::apply_replace): Rename to...
23286 (edited_line::apply_fixit): ...this. Convert to half-open range.
23287 Update for change to type of m_line_events.
23288 * edit-context.h (edit_context::apply_insert): Delete.
23289 (edit_context::apply_replace): Rename to...
23290 (edit_context::apply_fixit): ...this.
23292 2017-05-01 Martin Sebor <msebor@redhat.com>
23294 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
23297 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
23300 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
23301 __get_cpuid_max returns 0.
23302 (__get_cpuid_count): Ditto.
23304 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
23306 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
23307 replacement expression is another instance of one of its arguments.
23309 2017-05-01 Jakub Jelinek <jakub@redhat.com>
23312 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
23313 check for stack push/pop autoinc.
23314 * config/i386/i386.c (ix86_agi_dependent): Return false
23315 if the only reason why modified_in_p returned true is that
23316 addr is SP based and set_insn is a push or pop.
23318 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
23320 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
23323 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
23326 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
23327 (account_size_time): Use two predicates - exec_pred and
23329 (evaluate_conditions_for_known_args): Compute both clause and
23331 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
23332 (inline_summary_t::duplicate): Update.
23333 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
23335 (compute_inline_parameters): Likewise.
23336 (estimate_edge_size_and_time): Update caluclation of time.
23337 (estimate_node_size_and_time): Compute both time and nonspecialized
23339 (estimate_ipcp_clone_size_and_time): Update.
23340 (inline_merge_summary): Update.
23341 (do_estimate_edge_time): Update.
23342 (do_estimate_edge_size): Update.
23343 (do_estimate_edge_hints): Update.
23344 (inline_read_section, inline_write_summary): Stream both new predicates.
23345 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
23347 (compute_inlined_call_time): Cleanup.
23348 (big_speedup_p): Update.
23349 (edge_badness): Update.
23350 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
23351 (size_time_entry): Replace predicate by exec_predicate and
23352 nonconst_predicate.
23353 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
23354 (estimate_edge_time): Return also nonspec_time.
23355 (reset_edge_growth_cache): Update.
23357 2017-04-29 Jakub Jelinek <jakub@redhat.com>
23359 PR rtl-optimization/80491
23360 * ifcvt.c (noce_process_if_block): When looking for x setter
23361 with missing else_bb, don't check only the insn right before
23362 cond_earliest, but look for the last insn that x is modified in
23363 within the same bb.
23365 PR rtl-optimization/80491
23366 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
23368 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
23370 PR tree-optimization/80487
23371 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
23373 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23375 PR tree-optimization/79697
23376 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
23377 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
23378 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
23380 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
23381 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
23383 2017-04-28 Martin Sebor <msebor@redhat.com>
23385 PR tree-optimization/80523
23386 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
23387 (init_target_to_host_charmap, target_to_host, target_strtol10): New
23389 (maybe_warn, format_directive, parse_directive): Use new functions.
23390 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
23392 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
23394 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
23396 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
23398 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
23399 target_header_dir): Set correctly.
23400 * configure: Regenerated.
23401 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
23402 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
23403 instead of SYSTEM_HEADER_DIR.
23405 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
23407 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
23408 (estimate_local_effects): Likewise.
23409 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
23410 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
23411 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
23412 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
23413 do_estimate_edge_time, estimate_edge_time): Likewise.
23414 * ipa-inline-analysis.c (estimate_node_size_and_time,
23415 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
23416 (estimate_time_after_inlining): Remove.
23418 2017-04-28 Martin Liska <mliska@suse.cz>
23420 * doc/gcov.texi: Enhance documentation of gcov.
23422 2017-04-28 Martin Liska <mliska@suse.cz>
23424 * doc/gcov.texi: Sort options in alphabetic order.
23425 * doc/gcov-dump.texi: Likewise.
23426 * doc/gcov-tool.texi: Likewise.
23427 * gcov.c (print_usage): Likewise.
23428 * gcov-dump.c (print_usage): Likewise.
23429 * gcov-tool.c (print_merge_usage_message): Likewise.
23430 (print_rewrite_usage_message): Likewise.
23431 (print_overlap_usage_message): Likewise.
23433 2017-04-28 Martin Liska <mliska@suse.cz>
23435 PR gcov-profile/53915
23436 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
23438 2017-04-28 Martin Liska <mliska@suse.cz>
23440 PR gcov-profile/79891
23441 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
23442 is marked by compiler as living on a line.
23443 (get_cycles_count): Remove usage of the union.
23444 (output_intermediate_file): Likewise.
23445 (find_source): Fix GNU coding style.
23446 (accumulate_line_counts): Remove old non-all block mode.
23447 (output_lines): Remove usage of the union.
23448 * profile.c (output_location): Include all BBs, even if
23449 belonging to a same line (and file) as a previous BB.
23451 2017-04-28 Martin Liska <mliska@suse.cz>
23453 * gcov.c (process_args): Handle new argument 'w'.
23454 (read_graph_file): Assign ID to BBs.
23455 (output_branch_count): Display BB # if verbose flag is set.
23456 (output_lines): Likewise for arcs.
23457 (print_usage): Add '--verbose' option help.
23458 * doc/gcov.texi: Document --verbose (-w) option.
23460 2017-04-28 Martin Liska <mliska@suse.cz>
23462 * gcov.c (struct block_location_info): New struct.
23463 (process_file): Fill up the new structure.
23464 (read_graph_file): Replace usage of encoding by the newly added
23466 (add_line_counts): Likewise.
23467 (accumulate_line_counts): Remove usage of the union.
23468 (function_info::function_info): New function.
23469 (function_info::~function_info): Likewise.
23470 (process_file): Call delete instead of release_function.
23471 (release_function): Release the function.
23472 (release_structures): Call delete instead of release_function.
23473 (solve_flow_graph): Replace usage of num_blocks.
23474 (find_exception_blocks): Likewise.
23475 (output_lines): Fix GNU coding style.
23477 2017-04-28 Martin Liska <mliska@suse.cz>
23480 * coverage.c (coverage_remove_note_file): New function.
23481 * coverage.h: Declare the function.
23482 * toplev.c (finalize): Clean if an error has been seen.
23484 2017-04-28 Martin Liska <mliska@suse.cz>
23486 PR gcov-profile/80031
23487 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
23488 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
23489 * gcov.c (read_graph_file): Read just number of blocks.
23490 * profile.c (branch_prob): Do not stream 0 flags per a basic
23493 2017-04-28 Martin Liska <mliska@suse.cz>
23495 * gcov-dump.c (tag_*): Add new argument to declarations.
23496 (dump_gcov_file): Likewise.
23497 (tag_blocks): Add and use new argument depth.
23498 (tag_arcs): Likewise.
23499 (tag_lines): Likewise.
23500 (tag_counters): Likewise.
23501 (tag_summary): Likewise.
23502 (dump_working_sets): Use depth to do a proper indentation.
23504 2017-04-28 Jakub Jelinek <jakub@redhat.com>
23507 * cgraph.h (symtab_node::debug_symtab): No longer inline.
23508 * symtab.c (symtab_node::debug_symtab): Move definition here.
23510 2017-04-28 Richard Biener <rguenther@suse.de>
23512 * lto-streamer.h (LTO_major_version): Bump to 7.
23514 2017-04-28 Richard Biener <rguenther@suse.de>
23516 * tree-vrp.c (assert_info): New struct.
23517 (add_assert_info): New helper.
23518 (register_edge_assert_for_2): Refactor to add asserts to a vector
23520 (register_edge_assert_for_1): Likewise.
23521 (register_edge_assert_for): Likewise.
23522 (finish_register_edge_assert_for): New helper actually registering
23523 asserts where live on edge.
23524 (find_conditional_asserts): Adjust.
23525 (find_switch_asserts): Likewise.
23526 (evrp_dom_walker::try_find_new_range): Generalize.
23527 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
23529 2017-04-27 Marek Polacek <polacek@redhat.com>
23532 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
23533 arg10 and arg11 to itype.
23535 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
23537 * doc/extend.texi (Object Size Checking): Improve grammar.
23539 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
23542 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
23543 that the logic for permitting reciprocal estimates matches that
23546 2017-04-27 Jakub Jelinek <jakub@redhat.com>
23549 * tree.c (type_cache_hasher::equal): Only compare
23550 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
23551 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
23552 non-aggregate element types.
23553 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
23554 about the flag on ARRAY_TYPEs in the comment, formatting fix.
23556 2017-04-27 Richard Biener <rguenther@suse.de>
23558 PR middle-end/80533
23559 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
23560 stripping ARRAY_REFs from MEM_EXPR make sure we're not
23561 keeping a reference to a trailing array.
23563 2017-04-27 Richard Biener <rguenther@suse.de>
23565 PR middle-end/80539
23566 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
23567 being in loop-closed SSA form conservatively.
23568 (chrec_fold_multiply_poly_poly): Likewise.
23570 2017-04-27 Tamar Christina <tamar.christina@arm.com>
23572 PR middle-end/79665
23573 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
23574 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
23576 2017-04-27 Jakub Jelinek <jakub@redhat.com>
23579 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
23580 (aarch64_function_arg_alignment): Return unsigned int again, but still
23581 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
23582 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
23583 Don't emit -Wpsabi note.
23584 (aarch64_function_arg_boundary): Likewise.
23585 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
23588 2017-04-26 Nathan Sidwell <nathan@acm.org>
23590 * tree.h (crc32_unsigned_n): Declare.
23591 (crc32_unsigned, crc32_unsigned): Make inline.
23592 * tree.c (crc32_unsigned_bits): Replace with ...
23593 (crc32_unsigned_n): ... this.
23594 (crc32_unsigned, crc32_byte): Remove.
23595 (crc32_string): Remove unnecessary braces.
23597 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
23599 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
23600 * ipa-inline-analysis.c (MAX_TIME): Remove.
23601 (account_size_time): Use sreal for time.
23602 (dump_inline_summary): Update.
23603 (estimate_function_body_sizes): Update.
23604 (estimate_edge_size_and_time): Update.
23605 (estimate_calls_size_and_time): Update.
23606 (estimate_node_size_and_time): Update.
23607 (inline_merge_summary): Update.
23608 (inline_update_overall_summary): Update.
23609 (estimate_time_after_inlining): Update.
23610 (inline_read_section): Update.
23611 (inline_write_summary): Update.
23612 * ipa-inline.c (compute_uninlined_call_time): Update.
23613 (compute_inlined_call_time): Update.
23614 (recursive_inlining): Update.
23615 (inline_small_functions): Update.
23616 (dump_overall_stats): Update.
23617 * ipa-inline.h: Include sreal.h.
23618 (size_time_entry): Turn time to sreal.
23619 (inline_summary): Turn self_time nad time to sreal.
23621 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
23623 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
23625 (sreal::stream_out, sreal::stream_in): New.
23626 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
23628 2017-04-25 Jakub Jelinek <jakub@redhat.com>
23630 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
23633 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
23636 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
23638 (dimode_scalar_chain::compute_convert_gain): Ditto.
23639 (dimode_scalar_chain::make_vector_copies): Ditto.
23640 (dimode_scalar_chain::convert_reg): Ditto.
23641 (dimode_scalar_chain::convert_insn): Ditto.
23642 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
23643 (VI248_AVX512BW_1): New mode iterator.
23644 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
23645 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
23648 2017-04-25 Martin Sebor <msebor@redhat.com>
23650 PR tree-optimization/80497
23651 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
23652 constants are representable in HOST_WIDE_INT.
23653 (parse_directive): Ditto.
23655 2017-04-25 Martin Sebor <msebor@redhat.com>
23658 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
23659 (new_zero_array): Adjust signature.
23660 (dom_info::dom_init): Used unsigned rather that size_t.
23661 (dom_info::dom_info): Same.
23663 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23664 Jakub Jelinek <jakub@redhat.com>
23667 * config/arm/arm.c: Include gimple.h.
23668 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
23669 returns negative, increment ncrn only if it returned positive.
23670 (arm_needs_doubleword_align): Return int instead of bool,
23671 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
23672 members, but if there is any such non-FIELD_DECL
23673 > PARM_BOUNDARY aligned decl, return -1 instead of false.
23674 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
23675 returns negative, increment nregs only if it returned positive.
23676 (arm_setup_incoming_varargs): Likewise.
23677 (arm_function_arg_boundary): Emit -Wpsabi note if
23678 arm_needs_doubleword_align returns negative, return
23679 DOUBLEWORD_ALIGNMENT only if it returned positive.
23681 2017-04-25 Marek Polacek <polacek@redhat.com>
23684 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
23685 first argument to type.
23687 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
23690 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
23691 type checks to test for compatibility instead of equality.
23693 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23694 Jakub Jelinek <jakub@redhat.com>
23697 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
23699 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
23700 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
23701 the alignment computation, but return their maximum in warn_alignment.
23702 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
23703 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
23705 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
23706 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
23709 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23711 * config/arc/simdext.md (dmpyh): Fix typo.
23713 2017-04-25 Richard Biener <rguenther@suse.de>
23715 PR tree-optimization/80492
23716 * alias.c (compare_base_decls): Handle registers with asm
23717 specification conservatively.
23718 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
23719 compare_base_decls returning dont-know properly.
23721 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23723 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
23724 (legitimate_offset_address_p): New function.
23725 (arc_legitimate_address_p): Use above function.
23727 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23729 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
23731 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23733 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
23734 ACCH registers whenever they are available.
23736 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23738 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
23739 double regs fix when not used.
23741 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23743 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
23745 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
23746 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
23748 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23750 * config/arc/arc.c (arc_output_addsi): Check for h-register class
23751 when emitting short ADD instructions.
23753 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23755 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
23757 (cmpsi_cc_c_insn): Likewise.
23758 (cbranchsi4_scratch): Compute proper instruction length using
23759 compact_hreg_operand.
23760 * config/arc/predicates.md (compact_hreg_operand): New predicate.
23762 2017-04-25 Richard Biener <rguenther@suse.de>
23764 PR middle-end/80509
23765 * passes.c (pass_manager::pass_manager): Initialize
23766 m_name_to_pass_map.
23768 2017-04-25 Richard Biener <rguenther@suse.de>
23770 PR tree-optimization/79201
23771 * tree-ssa-sink.c (statement_sink_location): Handle calls.
23773 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23776 * config/s390/vector.md: Split MEM->GPR vector moves for
23777 non-s_operand addresses.
23779 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23782 * config/s390/predicates.md (reload_const_wide_int_operand): New
23784 * config/s390/s390.md ("movti"): Remove d/P alternative.
23785 ("movti_bigconst"): New pattern definition.
23787 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
23790 * s390-protos.h (s390_expand_cs_hqi): Removed.
23791 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
23792 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
23793 modes as well as CCZ1mode and CCZmode.
23794 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
23795 signature of s390_emit_compare_and_swap.
23796 (s390_expand_cs_hqi): Likewise, make static.
23797 (s390_expand_cs_tdsi): Generate an explicit compare before trying
23798 compare-and-swap, in some cases.
23799 (s390_expand_cs): Wrapper function.
23800 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
23802 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
23803 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
23804 patterns for small and large integers. Forbid symref memory operands.
23805 Move expander to s390.c. Require cc register.
23806 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
23807 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
23808 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
23809 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
23810 symref memory operands. Remove CC mode and call s390_match_ccmode
23812 ("atomic_exchange<mode>"): Allow and implement all integer modes.
23814 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
23816 * config/s390/s390.md (define_peephole2): New peephole to help
23817 combining the load-and-test pattern with volatile memory.
23819 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
23821 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
23822 with CCZmode for TARGET_Z196.
23824 2017-04-25 Jakub Jelinek <jakub@redhat.com>
23826 PR rtl-optimization/80501
23827 * combine.c (make_compound_operation_int): Set subreg_code to SET
23828 even for AND with mask of the sign bit of mode.
23830 PR rtl-optimization/80500
23831 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
23832 sum's initial value.
23834 2017-04-25 Julian Brown <julian@codesourcery.com>
23835 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23837 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
23839 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
23841 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
23843 2017-04-25 Julian Brown <julian@codesourcery.com>
23844 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23846 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
23847 (thunderx2t99_sha): New Reservation.
23849 2017-04-25 Julian Brown <julian@codesourcery.com>
23850 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23852 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
23853 type for 1-element load.
23855 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
23857 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
23859 2017-04-24 Martin Jambor <mjambor@suse.cz>
23861 PR tree-optimization/80293
23862 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
23863 char arrays not totally scalarizable if it is false.
23864 (analyze_all_variable_accesses): Pass correct value in the new
23865 parameter. Add a statistics counter.
23867 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
23869 PR middle-end/79931
23870 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
23872 2017-04-24 Richard Biener <rguenther@suse.de>
23874 PR tree-optimization/80494
23875 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
23876 out for complex types.
23878 2017-04-24 Richard Biener <rguenther@suse.de>
23880 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
23881 * tree-ssa-sccvn.c (print_scc): Print SCC size.
23882 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
23883 (DFS): Adjust and never fail.
23884 (sccvn_dom_walker::fail): Remove.
23885 (sccvn_dom_walker::before_dom_children): Adjust.
23886 (run_scc_vn): Likewise and never fail.
23887 * tree-ssa-pre.c (pass_pre::execute): Adjust.
23888 (pass_fre::execute): Likewise.
23890 2017-04-24 Richard Biener <rguenther@suse.de>
23892 PR tree-optimization/79725
23893 * tree-ssa-sink.c (statement_sink_location): Return whether
23894 failure reason was zero uses. Move that check later.
23895 (sink_code_in_bb): Deal with zero uses by removing the stmt
23898 2017-04-24 Richard Biener <rguenther@suse.de>
23901 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
23902 pointer-based references.
23904 2017-04-24 Richard Biener <rguenther@suse.de>
23907 * pass_manager.h (pass_manager::operator new): Remove.
23908 (pass_manager::operator delete): Likewise.
23909 * passes.c (pass_manager::operator new): Remove.
23910 (pass_manager::operator delete): Likewise.
23911 (pass_manager::pass_manager): Zero individual pass members.
23913 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
23916 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
23917 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
23918 Check "XEXP (src, 1)" operand here.
23919 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
23920 Check "XEXP (src, 1)" operand here.
23921 (dimode_scalar_chain::make_vector_copies): Detect count register
23922 of a shift instruction. Zero extend count register from QImode
23923 to DImode to satisfy vector shift pattern count operand predicate.
23924 Substitute vector shift count operand with a DImode copy.
23925 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
23928 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
23930 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
23931 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
23932 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
23933 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
23934 (UNSPEC_NOREX_MEM): Remove definition.
23936 2017-04-21 Richard Biener <rguenther@suse.de>
23938 PR tree-optimization/79547
23939 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
23940 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
23941 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
23942 without any constraints.
23944 2017-04-21 Richard Biener <rguenther@suse.de>
23946 PR tree-optimization/78847
23947 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
23949 2017-04-21 Richard Biener <rguenther@suse.de>
23951 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
23952 (build_distinct_type_copy): Likewise.
23953 (build_variant_type_copy): Likewise.
23954 * tree.c (build_qualified_type): Pass down mem-stat info.
23955 (build_distinct_type_copy): Likewise.
23956 (build_variant_type_copy): Likewise.
23958 2017-04-21 Richard Biener <rguenther@suse.de>
23960 PR tree-optimization/80237
23961 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
23963 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
23964 for a simplified result.
23966 2016-04-21 Richard Biener <rguenther@suse.de>
23968 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
23969 sth as strict as a simple_iv but a chrec without symbols and an
23970 operand defined in the loop we are peeling (and not some subloop).
23971 (propagate_constants_for_unrolling): Propagate all constants.
23973 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
23976 * config/i386/i386.c (print_reg): Remove assert for disalowed
23977 regno values, call output_operand_lossage instead.
23979 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
23982 * config/i386/constraints.md (Yc): New register constraint.
23983 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
23984 Use Yc constraint for alternative 2 of operand 0. Remove
23985 preferred_for_speed attribute.
23987 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
23989 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
23990 lastprivate clauses in SIMT case.
23992 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
23994 * doc/invoke.texi (-Wextra-semi): Document new warning option.
23996 2017-04-20 Richard Biener <rguenther@suse.de>
23998 PR tree-optimization/57796
23999 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
24000 as N scalar stores.
24001 (vect_model_load_cost): Cost gathers as N scalar loads.
24003 2017-04-20 Richard Biener <rguenther@suse.de>
24005 * ggc-page.c (ggc_allocated_p): Rename to ...
24006 (safe_lookup_page_table_entry): ... this and return the lookup
24008 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
24010 2017-04-20 Richard Biener <rguenther@suse.de>
24012 PR tree-optimization/80453
24013 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
24014 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
24015 from the conditions.
24016 (vn_phi_eq): Pass them down.
24017 (vn_phi_lookup): Record them.
24018 (vn_phi_insert): Likewise.
24020 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
24022 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
24023 uninitialized variable warning to avoid buffer overrun.
24025 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
24028 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
24029 is suppressed for '{ 0 }' in C.
24031 2017-04-20 Jakub Jelinek <jakub@redhat.com>
24033 * BASE-VER: Set to 8.0.0.
24035 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
24037 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
24038 priority .init_array and .fini_array section with SECTION_NOTYPE
24041 2017-04-20 Jakub Jelinek <jakub@redhat.com>
24043 PR middle-end/80423
24044 * tree.h (build_array_type): Add typeless_storage default argument.
24045 * tree.c (type_cache_hasher::equal): Also compare
24046 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
24047 (build_array_type): Add typeless_storage argument, set
24048 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
24050 (build_nonshared_array_type): Adjust build_array_type_1 caller.
24051 (build_array_type): Likewise. Add typeless_storage argument.
24053 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
24054 Jakub Jelinek <jakub@redhat.com>
24056 PR tree-optimization/80426
24057 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
24058 operation on symbolic operands, also compute the overflow for the
24059 invariant part when the operation degenerates into a negation.
24061 2017-04-19 Jakub Jelinek <jakub@redhat.com>
24064 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
24065 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
24068 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
24070 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
24073 * config/avr/avr.c (tree.h): Include it.
24074 (cgraph.h): Include it.
24075 (avr_encode_section_info): Don't warn for uninitialized progmem
24076 variable if it's just an alias.
24078 2017-04-19 Richard Biener <rguenther@suse.de>
24081 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
24082 when needed by AutoPGO.
24084 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
24087 * doc/lto.texi: Remove an extra 'that'.
24089 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
24091 PR rtl-optimization/80429
24092 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
24093 are only used in debug insns.
24095 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
24096 Vladimir Makarov <vmakarov@redhat.com>
24098 * config/sparc/predicates.md (input_operand): Add comment. Return
24099 true for any memory operand when LRA is in progress.
24100 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
24102 2017-04-18 Jeff Law <law@redhat.com>
24105 * mips.md ({return,simple_return}_internal): Do not overwrite
24108 2017-04-18 Jakub Jelinek <jakub@redhat.com>
24110 PR tree-optimization/80443
24111 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
24112 instead of adding 1, subtract -1 and similarly instead of subtracting
24115 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
24117 PR rtl-optimization/80357
24118 * haifa-sched.c (tmp_bitmap): New variable.
24119 (model_recompute): Handle duplicate use records.
24120 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
24121 (free_global_sched_pressure_data): Free it.
24123 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
24126 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
24127 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
24128 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
24129 instead of SYSTEM_HEADER_DIR.
24131 2017-04-18 Jeff Law <law@redhat.com>
24133 PR middle-end/80422
24134 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
24135 predecessors after walking up the insn chain.
24137 2017-04-18 Jakub Jelinek <jakub@redhat.com>
24140 * dwarf2out.c (modified_type_die): Try harder not to emit internal
24141 sizetype type into debug info.
24143 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
24146 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
24147 unneeded test for TARGET_UPPER_REGS_SF.
24148 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
24150 2017-04-18 Jakub Jelinek <jakub@redhat.com>
24153 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
24154 instead of gsi_after_labels.
24156 2017-04-18 Jeff Law <law@redhat.com>
24158 * regcprop.c (maybe_mode_change): Avoid creating copies of the
24162 2017-04-13 Jeff Law <law@redhat.com>
24163 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
24164 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
24166 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
24169 * config/avr/avr.c (intl.h): Include it.
24170 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
24172 2017-04-18 Martin Liska <mliska@suse.cz>
24174 PR gcov-profile/78783
24175 * gcov-tool.c (gcov_output_files): Validate that destination
24176 file is either removed by the tool or by a user.
24178 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
24179 Guy Benyei <guybe@mellanox.com>
24181 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
24182 block, and do not negate it, the stored id is already negative.
24184 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
24186 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
24188 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
24191 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
24192 masks of options that should be turned off if the VSX vector
24193 options are turned off.
24194 (OTHER_P8_VECTOR_MASKS): Likewise.
24195 (OTHER_VSX_VECTOR_MASKS): Likewise.
24196 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
24197 rs6000_disable_incompatible_switches to validate no type switches
24199 (rs6000_incompatible_switch): New function to disallow turning on
24200 other vector options if -mno-vsx, -mno-power8-vector, or
24201 -mno-power9-vector are specified.
24203 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
24205 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
24207 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
24209 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
24210 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
24211 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
24212 (ARG_POINTER_CFA_OFFSET): Likewise.
24214 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
24216 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
24217 conditions to take advantage of various optimizations.
24219 2017-04-13 Jeff Law <law@redhat.com>
24221 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
24222 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
24223 (zero_extendsidi2_dext): Likewise.
24225 2017-04-13 Jakub Jelinek <jakub@redhat.com>
24228 * fold-const.c (fold_ternary_loc): Revert
24229 use op0 instead of fold_convert_loc (loc, type, arg0) part of
24232 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
24234 PR rtl-optimization/80343
24235 * lra-remat.c (update_scratch_ops): Assign original hard reg to
24236 new scratch pseudo.
24238 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
24241 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
24242 to ubsan_encode_value.
24244 2017-04-13 Jeff Law <law@redhat.com>
24246 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
24247 appearing in DEBUG_INSNs.
24249 2017-04-13 Martin Liska <mliska@suse.cz>
24251 PR gcov-profile/80413
24252 * gcov-io.c (gcov_write_string): Copy to buffer just when
24253 allocated size is greater than zero.
24255 2017-04-13 Jakub Jelinek <jakub@redhat.com>
24258 * dwarf2out.c (decls_for_scope): Ignore declarations of
24259 current_function_decl in BLOCK_NONLOCALIZED_VARS.
24261 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
24264 * ipa-visibility.c (non_local_p): Fix typos.
24265 (localize_node): When localizing symbol in same comdat group,
24266 dissolve the group only when we know external symbols are going
24268 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
24270 2017-04-12 Jakub Jelinek <jakub@redhat.com>
24272 PR tree-optimization/79390
24273 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
24274 order does not result in usable sequence, retry with reversed operand
24280 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
24281 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
24282 op0 instead of fold_convert_loc (loc, type, arg0).
24284 2017-04-12 Jeff Law <law@redhat.com>
24286 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
24287 has a delay slot in the generated code.
24289 * config/cris/cris.md (cris_preferred_reload_class): Return
24290 GENNONACR_REGS rather than GENERAL_REGS.
24292 2017-04-12 Jakub Jelinek <jakub@redhat.com>
24295 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
24296 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
24297 signedness of the result type.
24299 2017-04-12 Richard Biener <rguenther@suse.de>
24300 Jeff Law <law@redhat.com>
24302 PR tree-optimization/80359
24303 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
24304 trim stores to TARGET_MEM_REFs.
24306 2017-04-12 Richard Biener <rguenther@suse.de>
24308 PR tree-optimization/79390
24309 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
24310 threading case even more.
24312 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
24315 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
24316 for quad_address_p for TImode, instead of just not indexed_address.
24318 2017-04-12 Richard Biener <rguenther@suse.de>
24319 Bernd Edlinger <bernd.edlinger@hotmail.de>
24321 PR middle-end/79671
24322 * alias.c (component_uses_parent_alias_set_from): Handle
24323 TYPE_TYPELESS_STORAGE.
24324 (get_alias_set): Likewise.
24325 * tree-core.h (tree_type_common): Add typeless_storage flag.
24326 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
24327 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
24328 for types containing members with TYPE_TYPELESS_STORAGE.
24329 (place_field): Likewise.
24330 (layout_type): Likewise for ARRAY_TYPE.
24331 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
24332 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
24333 TYPE_TYPELESS_STORAGE.
24334 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
24336 2017-04-12 Jakub Jelinek <jakub@redhat.com>
24339 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
24340 first argument to type.
24342 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24346 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
24347 CONST0_RTX (mode) rather than const0_rtx where appropriate.
24348 (rs6000_expand_binop_builtin): Likewise.
24349 (rs6000_expand_ternop_builtin): Likewise; also add missing
24350 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
24351 vshasigma built-ins.
24352 * doc/extend.texi: Document that vec_xxpermdi's third argument
24353 must be a constant.
24355 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
24357 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
24358 Use shift_const cost parameter when calculating gain of STV shifts.
24360 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
24362 PR rtl-optimization/70478
24363 * lra-constraints.c (process_alt_operands): Check memory for
24364 disfavoring memory insn operand.
24366 2017-04-11 Jakub Jelinek <jakub@redhat.com>
24368 PR middle-end/80100
24369 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
24370 left shift in unsigned HOST_WIDE_INT type.
24372 PR rtl-optimization/80385
24373 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
24374 (not (neg X)) into (plus X -1) for complex or non-integral modes.
24377 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
24378 if they have any depend clauses.
24380 2017-04-11 Martin Liska <mliska@suse.cz>
24383 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
24384 * ipa-split.c (split_function): Create a local comdat symbol
24385 if caller is in a comdat group.
24387 2017-04-11 Martin Liska <mliska@suse.cz>
24390 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
24393 2017-04-11 Martin Sebor <msebor@redhat.com>
24395 PR middle-end/80364
24396 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
24397 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
24399 (directive::set_width, directive::set_precision, format_character):
24401 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
24404 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
24407 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
24408 conflict, set target->arch_name instead of target->cpu_name.
24410 2017-04-11 Richard Biener <rguenther@suse.de>
24412 PR tree-optimization/80374
24413 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
24414 build_zero_cst, remove fold_convertible_p check again.
24416 2017-04-11 Martin Liska <mliska@suse.cz>
24419 * ubsan.c (instrument_object_size): Do not instrument register
24422 2017-04-11 Jakub Jelinek <jakub@redhat.com>
24425 * config/i386/i386-builtin-types.def
24426 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
24427 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
24428 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
24429 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
24430 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
24431 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
24432 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
24433 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
24434 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
24435 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
24436 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
24437 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
24438 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
24439 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
24440 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
24441 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
24442 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
24443 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
24444 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
24445 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
24446 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
24447 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
24448 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
24449 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
24450 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
24451 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
24452 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
24453 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
24454 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
24455 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
24456 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
24457 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
24458 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
24459 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
24460 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
24461 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
24462 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
24463 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
24464 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
24465 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
24466 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
24467 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
24468 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
24469 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
24470 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
24471 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
24473 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
24474 flag to second_arg_count, handle 4 argument function type _COUNT
24475 aliases, handle second_arg_count on second argument rather than last.
24477 2017-04-10 Jeff Law <law@redhat.com>
24479 PR tree-optimization/80374
24480 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
24481 record anything if we can not convert integer_zero_node to the
24484 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
24487 * config/rs6000/rs6000.c (rs6000_option_override_internal):
24488 Enhance special handling given to the TARGET_P9_MINMAX option in
24489 relation to certain other options.
24491 2017-04-10 Bin Cheng <bin.cheng@arm.com>
24493 PR tree-optimization/80153
24494 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
24495 remove POINTER_PLUS_EXPR's base part directly, rather than through
24498 2017-04-10 Richard Biener <rguenther@suse.de>
24499 Bin Cheng <bin.cheng@arm.com>
24501 PR tree-optimization/80153
24502 * tree-affine.c (aff_combination_to_tree): Get base pointer from
24503 the first element of pointer type aff_tree. Build result expr in
24505 (add_elt_to_tree): Convert to type unconditionally. Remove other
24506 fold_convert calls.
24507 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
24508 (rewrite_use_nonlinear_expr): Check invariant using iv information.
24510 2017-04-10 Richard Biener <rguenther@suse.de>
24512 * tree-ssa-structalias.c (find_func_aliases): Properly handle
24515 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
24517 PR rtl-optimization/70478
24518 * lra-constraints.c (curr_small_class_check): New.
24519 (update_and_check_small_class_inputs): New.
24520 (process_alt_operands): Update curr_small_class_check. Disfavor
24521 alternative insn memory operands. Check available regs for small
24524 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
24527 * config/mips/mips.opt (-mvirt): Update description.
24528 * doc/invoke.texi (-mvirt): Likewise.
24530 2017-04-10 Richard Biener <rguenther@suse.de>
24532 PR middle-end/80362
24533 * fold-const.c (fold_binary_loc): Look at unstripped ops when
24534 looking for NEGATE_EXPR in -A / -B to A / B folding.
24536 2017-04-10 Martin Liska <mliska@suse.cz>
24538 PR gcov-profile/80224
24539 * gcov.c (print_usage): Fix usage string.
24540 (get_gcov_intermediate_filename): Remove.
24541 (output_gcov_file): Use both for normal and intermediate format.
24542 (generate_results): Do not initialize special file for
24543 intermediate format.
24545 2017-04-10 Richard Biener <rguenther@suse.de>
24547 PR tree-optimization/80304
24548 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
24551 2017-04-10 Nathan Sidwell <nathan@acm.org>
24554 * config/rs6000/rs6000.c (rs6000_vector_type): New.
24555 (rs6000_init_builtins): Use it.
24557 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24559 * config/arm/arm.md (<mrc>): Add mode to SET source.
24560 (<mrrc>): Likewise.
24562 2017-04-10 Richard Biener <rguenther@suse.de>
24564 PR middle-end/80344
24565 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
24567 2017-04-10 Jakub Jelinek <jakub@redhat.com>
24570 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
24571 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
24572 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
24573 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
24574 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
24575 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
24576 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
24577 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
24578 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
24579 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
24580 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
24581 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
24582 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
24583 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
24584 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
24585 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
24586 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
24587 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
24588 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
24589 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
24590 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
24591 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
24592 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
24594 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
24596 PR rtl-optimization/70478
24597 * lra-constraints.c: Reverse the last patch.
24599 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
24601 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
24602 Add comment for WCHAR_T.
24604 2017-04-08 Martin Liska <mliska@suse.cz>
24607 2017-04-07 Martin Liska <mliska@suse.cz>
24610 * ipa-split.c (split_function): Add function part to a same comdat
24613 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24616 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
24618 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
24620 * rs6000/rs6000.c (vec_load_pendulum): Rename...
24621 (vec_pairing): ...to this.
24622 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
24623 (rs6000_sched_init): Adjust for name change.
24624 (struct rs6000_sched_context): Likewise.
24625 (rs6000_init_sched_context): Likewise.
24626 (rs6000_set_sched_context): Likewise.
24628 2017-04-07 Jakub Jelinek <jakub@redhat.com>
24634 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
24636 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
24637 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
24638 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
24640 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
24642 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
24644 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
24646 PR rtl-optimization/70703
24647 * ira-color.c (update_conflict_hard_regno_costs): Use
24648 int64_t instead of HOST_WIDE_INT.
24650 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
24652 PR rtl-optimization/70478
24653 * lra-constraints.c (process_alt_operands): Disfavor alternative
24654 insn memory operands.
24656 2017-04-07 Jeff Law <law@redhat.com>
24658 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
24659 CALL and NOTE_INSN_CALL_ARG_LOCATION.
24661 2017-04-07 Martin Liska <mliska@suse.cz>
24664 * config/aarch64/aarch64.c (aarch64_process_target_attr):
24665 Show error message instead of an ICE.
24667 2017-04-07 Martin Liska <mliska@suse.cz>
24670 * ipa-split.c (split_function): Add function part to a same comdat
24673 2017-04-07 Richard Biener <rguenther@suse.de>
24675 PR middle-end/80341
24676 * tree.c (get_unwidened): Also handle ! for_type case for
24678 * convert.c (do_narrow): Split out from ...
24679 (convert_to_integer_1): ... here. Do not pass final truncation
24680 type to get_unwidened for TRUNC_DIV_EXPR.
24682 2017-04-07 Richard Biener <rguenther@suse.de>
24684 * tree-affine.c (wide_int_ext_for_comb): Take type rather
24686 (aff_combination_const): Adjust.
24687 (aff_combination_scale): Likewise.
24688 (aff_combination_add_elt): Likewise.
24689 (aff_combination_add_cst): Likewise.
24690 (aff_combination_convert): Likewise.
24691 (add_elt_to_tree): Likewise. Remove unused argument.
24692 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
24694 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
24696 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
24698 * config/arm/arm.c (arm_default_short_enums): Use
24699 ARM_DEFAULT_SHORT_ENUMS.
24700 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
24702 2017-04-06 Jakub Jelinek <jakub@redhat.com>
24705 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
24706 members with redundant out-of-class redeclaration.
24708 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
24711 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
24712 * config/i386/i386.md (*zero_extendsidi2):
24713 Add (?*x,*x) and (?*v,*v) alternatives.
24715 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
24718 * config/i386/i386.c (ix86_expand_builtin)
24719 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
24720 mode from insn data. Convert operands to insn operand mode.
24721 Copy operands that don't satisfy insn predicate to a register.
24723 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
24725 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
24728 2017-04-06 Richard Biener <rguenther@suse.de>
24730 PR tree-optimization/80334
24731 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
24732 preserve alignment of accesses.
24734 2017-04-06 Richard Biener <rguenther@suse.de>
24736 PR tree-optimization/80262
24737 * tree-sra.c (build_ref_for_offset): Preserve address-space
24739 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
24740 Drop useless address-space information on MEM_REF offsets.
24742 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
24744 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
24746 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
24748 PR rtl-optimization/70703
24749 * ira-color.c (update_conflict_hard_regno_costs): Use
24750 HOST_WIDE_INT instead of long.
24752 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
24755 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
24756 not defined for x86_64 target. Add -mmmx target option when __SSE2__
24758 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
24759 for x86_64 target. Handle -m3dnowa option.
24761 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
24763 PR rtl-optimization/70703
24764 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
24765 (update_conflict_hard_regno_costs): Use long instead of unsigned
24766 arithmetic for cost calculation.
24768 2017-04-05 Jakub Jelinek <jakub@redhat.com>
24769 Bernd Edlinger <bernd.edlinger@hotmail.de>
24772 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
24775 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
24778 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
24779 ptr_mode with Pmode throughout.
24780 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
24781 into probe_stack_range and use DImode.
24783 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
24786 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
24787 call_eh_return is true.
24789 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24791 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
24792 Initialize last_match_fntype_index.
24794 2017-04-05 Jakub Jelinek <jakub@redhat.com>
24797 * tree-nvr.c: Include internal-fn.h.
24798 (pass_return_slot::execute): Ignore internal calls without
24801 2017-04-04 Jakub Jelinek <jakub@redhat.com>
24802 Richard Biener <rguenther@suse.de>
24805 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
24806 captures used multiple times, except for the last use.
24807 * generic-match-head.c: Include gimplify.h.
24809 2017-04-04 Jakub Jelinek <jakub@redhat.com>
24811 PR tree-optimization/79390
24812 * target.h (struct noce_if_info): Declare.
24813 * targhooks.h (default_noce_conversion_profitable_p): Declare.
24814 * target.def (noce_conversion_profitable_p): New target hook.
24815 * ifcvt.h (struct noce_if_info): New type, moved from ...
24816 * ifcvt.c (struct noce_if_info): ... here.
24817 (noce_conversion_profitable_p): Renamed to ...
24818 (default_noce_conversion_profitable_p): ... this. No longer
24820 (noce_try_store_flag_constants, noce_try_addcc,
24821 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
24822 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
24823 instead of noce_conversion_profitable_p.
24824 * config/i386/i386.c: Include ifcvt.h.
24825 (ix86_option_override_internal): Don't override
24826 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
24827 (ix86_noce_conversion_profitable_p): New function.
24828 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
24829 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
24830 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
24831 * doc/tm.texi: Regenerated.
24833 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24835 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
24838 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24841 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
24842 instructions for small multiply cores.
24844 2017-04-04 Jeff Law <law@redhat.com>
24846 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
24848 (mips_expand_vec_perm_const): Initialize elements in orig_perm
24849 that are not set by the loop over the elements.
24851 2017-04-04 Jakub Jelinek <jakub@redhat.com>
24854 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
24855 int mode, convert_modes it to mode as unsigned, otherwise use
24856 lowpart_subreg to mode rather than SImode.
24857 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
24858 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
24859 Use DImode instead of SImode for the shift count operand.
24860 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
24863 2017-04-04 Richard Biener <rguenther@suse.de>
24865 PR middle-end/80281
24866 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
24867 arithmetic done for the negate or the plus. Simplify.
24868 (A - (-B) -> A + B): Likewise.
24869 * fold-const.c (split_tree): Make sure to not negate pointers.
24871 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
24873 PR rtl-optimization/60818
24874 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
24875 a compare of comparisons with the thing compared if this results
24876 in a different machine mode.
24878 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
24880 * alias.c (base_alias_check): Fix typo in comment.
24881 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
24882 * cgraphunit.c (symbol_table::compile): Likewise.
24883 * collect2.c (maybe_run_lto_and_relink): Likewise.
24884 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
24885 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
24886 * config/avr/avr.c (avr_map_op_t): Likewise.
24887 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
24888 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
24889 * config/epiphany/epiphany.md (movcc): Likewise.
24890 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
24891 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
24893 * config/mips/mips.c (mips_save_restore_reg): Likewise.
24894 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
24895 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
24896 * config/sh/sh.c (sh_rtx_costs): Likewise.
24897 * fold-const.c (fold_truth_andor): Likewise.
24898 * genautomata.c (collapse_flag): Likewise.
24899 * gengtype.h (struct type::u::s): Likewise.
24900 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
24901 * input.c (FORMAT_AMOUNT): Likewise.
24902 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
24903 (known_aggs_to_agg_replacement_list): Likewise.
24904 * ipa-inline-analysis.c: Likewise.
24905 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
24906 * ipa-polymorphic-call.c
24907 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
24908 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
24909 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
24911 * modulo-sched.c (apply_reg_moves): Likewise.
24912 * omp-expand.c (build_omp_regions_1): Likewise.
24913 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
24914 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
24915 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
24916 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
24917 * value-prof.c: Likewise.
24918 * var-tracking.c (val_reset): Likewise.
24920 2017-04-03 Richard Biener <rguenther@suse.de>
24922 PR tree-optimization/80275
24923 * fold-const.c (split_address_to_core_and_offset): Handle
24926 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
24928 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
24929 descriptors is at least equal to that of functions.
24931 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
24933 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
24935 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
24938 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
24939 (mov<IMOD4:mode>): New expander.
24940 (*mov<IMOD4:mode>_internal): New insn and split pattern.
24942 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
24944 PR rtl-optimization/79405
24945 * fwprop.c (propagations_left): New variable.
24946 (forward_propagate_into): Decrement it.
24947 (fwprop_init): Initialize it.
24948 (fw_prop): If the variable has reached zero, stop propagating.
24949 (fwprop_addr): Ditto.
24951 2017-03-31 Jakub Jelinek <jakub@redhat.com>
24954 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
24955 a FUNCTION_DECL, pass it as decl instead of origin to
24958 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
24960 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
24963 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
24966 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
24967 TARGET_VSX_SMALL_INTEGER.
24969 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24971 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
24972 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
24974 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
24976 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
24977 extraction from odd-numbered MSA register.
24979 2017-03-31 Jakub Jelinek <jakub@redhat.com>
24981 PR middle-end/80173
24982 * expmed.c (store_bit_field_1): Don't attempt to create
24983 a word subreg out of hard registers wider than word if they
24984 have HARD_REGNO_NREGS of 1 for their mode.
24986 PR middle-end/80163
24987 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
24988 conversions to integer types wider than word and pointer.
24991 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
24992 (rtx_equal_for_cselib_p): Pass 0 to it.
24993 * cselib.c (cselib_hasher::equal): Likewise.
24994 (rtx_equal_for_cselib_1): Add depth argument. If depth
24995 is 128, don't look up VALUE locs and punt. Increment
24996 depth in recursive calls when walking VALUE locs.
24998 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
25000 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
25001 (make_gcov_file_name): Use the canonical path name for generating
25003 (read_line): Fix handling of files with ascii null bytes.
25005 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
25007 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
25008 to initialise a vector register instead
25009 of using a const_int.
25011 2017-03-30 Jakub Jelinek <jakub@redhat.com>
25013 PR translation/80189
25014 * gimplify.c (omp_default_clause): Use %qs instead of %s in
25015 diagnostic messages.
25017 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
25020 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
25021 (dfp_diex_<mode>): Update mode of operand 1.
25022 * doc/extend.texi (dxex, dxexq): Document change to return type.
25023 (diex, diexq): Document change to argument type.
25025 2017-03-30 Martin Jambor <mjambor@suse.cz>
25028 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
25029 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
25030 it reflects the signature changes performed at the callee side.
25031 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
25032 to cgraph_build_function_type_skip_args.
25033 (build_function_decl_skip_args): Adjust call to the above function.
25035 2017-03-30 Jakub Jelinek <jakub@redhat.com>
25038 * config/i386/sse.md
25039 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
25040 register as dest whenever it is a MEM not rtx_equal_p to the
25041 corresponding dup operand, and when forcing into reg move the
25042 reg into the memory afterwards.
25043 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
25044 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
25045 for the force_reg mode.
25046 (avx512vl_vextractf128<mode>): Use register as dest either
25047 always when a MEM, or when it is a MEM not rtx_equal_p to the
25048 corresponding dup operand, or even not when it is a CONST_VECTOR
25049 depending on the mode and lo vs. hi.
25050 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
25052 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
25053 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
25054 Likewise. Require that operands[2] is even.
25055 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
25056 Remove extraneous parens. Require that operands[2] is a multiple
25058 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
25059 operands[0] is a MEM if <mask_applied>, the predicates/constraints
25060 disallow memory then.
25062 2017-03-30 Richard Biener <rguenther@suse.de>
25064 PR tree-optimization/77498
25065 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
25066 to non-constants over backedges.
25068 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
25070 PR rtl-optimization/80233
25071 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
25072 as last_combined_insn. Do not test for BARRIER_P separately.
25074 2017-03-29 Andreas Schwab <schwab@suse.de>
25077 * calls.c (prepare_call_address): Convert funexp to Pmode before
25078 copying to temp reg.
25080 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25082 PR tree-optimization/80158
25083 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
25084 Handle possible future case of more than one alternate
25086 (replace_rhs_if_not_dup): Likewise.
25087 (replace_one_candidate): Likewise.
25089 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
25091 PR rtl-optimization/80193
25092 * ira.c (ira): Do not check allocation for LRA.
25094 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
25096 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
25097 (nvptx_output_simt_exit): Declare.
25098 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
25099 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
25100 (init_softstack_frame): Move initialization of crtl->is_leaf to...
25101 (nvptx_declare_function_name): ...here. Emit declaration of local
25102 memory space buffer for omp_simt_enter insn.
25103 (nvptx_output_unisimt_switch): New.
25104 (nvptx_output_softstack_switch): New.
25105 (nvptx_output_simt_enter): New.
25106 (nvptx_output_simt_exit): New.
25107 * config/nvptx/nvptx.h (struct machine_function): New fields
25108 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
25109 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
25110 (UNSPECV_SIMT_EXIT): Ditto.
25111 (omp_simt_enter_insn): New insn.
25112 (omp_simt_enter): New expansion.
25113 (omp_simt_exit): New insn.
25114 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
25116 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
25117 (expand_GOMP_SIMT_ENTER_ALLOC): New.
25118 (expand_GOMP_SIMT_EXIT): New.
25119 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
25120 (GOMP_SIMT_ENTER_ALLOC): Ditto.
25121 (GOMP_SIMT_EXIT): Ditto.
25122 * target-insns.def (omp_simt_enter): New insn.
25123 (omp_simt_exit): Ditto.
25124 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
25126 (lower_rec_simd_input_clauses): Implement SIMT privatization.
25127 (lower_rec_input_clauses): Likewise.
25128 (lower_lastprivate_clauses): Handle SIMT privatization.
25130 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
25131 (ompdevlow_adjust_simt_enter): New.
25132 (find_simtpriv_var_op): New.
25133 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
25134 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
25136 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
25137 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
25138 (copy_decl_for_dup_finish): Ditto.
25140 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
25142 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
25145 * config/i386/i386.c (ix86_option_override_internal): Always
25146 allow -mpreferred-stack-boundary=3 for 64-bit targets.
25148 2017-03-28 Bin Cheng <bin.cheng@arm.com>
25150 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
25152 2017-03-28 Bin Cheng <bin.cheng@arm.com>
25154 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
25155 mark new edge's irreducible flag accordign to it.
25156 (vect_do_peeling): Check loop preheader edge's irreducible flag
25157 and pass it to function slpeel_add_loop_guard.
25159 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
25161 PR tree-optimization/80218
25162 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
25163 Update block frequencies and counts.
25165 2017-03-28 Richard Biener <rguenther@suse.de>
25167 PR tree-optimization/78644
25168 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
25169 of a simplification result we may not use it at all.
25171 2017-03-28 Richard Biener <rguenther@suse.de>
25174 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
25175 without arguments, generate default definition of a SSA name.
25177 2017-03-28 Richard Biener <rguenther@suse.de>
25179 PR middle-end/80222
25180 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
25181 TYPE_REF_CAN_ALIAS_ALL references.
25182 * fold-const.c (fold_indirect_ref_1): Likewise.
25184 2017-03-28 Martin Liska <mliska@suse.cz>
25187 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
25188 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
25190 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
25191 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
25193 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
25194 (EXTRA_SPECS): Define.
25195 (SUBTARGET_EXTRA_SPECS): Likewise.
25196 (SUBTARGET_CPP_SPEC): Likewise.
25197 * config/arc/elf.h (EXTRA_SPECS): Renamed to
25198 SUBTARGET_EXTRA_SPECS.
25199 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
25201 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
25203 * config/arc/simdext.md (vst64_insn): Update pattern.
25204 (vld32wh_insn): Likewise.
25205 (vld32wl_insn): Likewise.
25206 (vld64_insn): Likewise.
25207 (vld32_insn): Likewise.
25209 2017-03-28 Marek Polacek <polacek@redhat.com>
25212 * fold-const.c (fold_comparison): Use protected_set_expr_location
25213 instead of SET_EXPR_LOCATION.
25215 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
25217 * tree.c (add_expr): Avoid name lookup warning.
25219 2017-03-27 Jeff Law <law@redhat.com>
25221 PR tree-optimization/80216
25222 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
25223 function name. Limit recursion depth.
25224 (record_temporary_equivalences): Corresponding changes.
25226 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
25228 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
25231 2017-03-27 Jakub Jelinek <jakub@redhat.com>
25234 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
25236 * cfgcleanup.c (reg_note_cfa_p): New array.
25237 (insns_have_identical_cfa_notes): New function.
25238 (old_insns_match_p): Don't cross-jump in between /f
25239 and non-/f instructions. If both i1 and i2 are frame related,
25240 verify all CFA notes, their order and content.
25242 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
25245 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
25246 HImode and SImode with zero extend to DImode to one insn.
25247 (bswap<mode>2_extenddi): Likewise.
25248 (bswapsi2_extenddi): Likewise.
25249 (bswaphi2_extendsi): Likewise.
25250 (bswaphi2): Combine bswap HImode and SImode into one insn.
25251 Separate memory insns from swapping register.
25252 (bswapsi2): Likewise.
25253 (bswap<mode>2): Likewise.
25254 (bswaphi2_internal): Delete, no longer used.
25255 (bswapsi2_internal): Likewise.
25256 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
25257 store, and gpr<-gpr swap insns.
25258 (bswap<mode>2_store): Likewise.
25259 (bswaphi2_reg): Register only splitter, combine with the splitter.
25260 (bswaphi2 splitter): Likewise.
25261 (bswapsi2_reg): Likewise.
25262 (bswapsi2 splitter): Likewise.
25263 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
25264 the insns into load, store, and register/register insns.
25265 (bswapdi2_ldbrx): Likewise.
25266 (bswapdi2_load): Likewise.
25267 (bswapdi2_store): Likewise.
25268 (bswapdi2_reg): Likewise.
25270 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
25272 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
25273 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
25275 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
25278 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
25280 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
25281 special handling for target option conflicts between dform
25282 options (-mpower9-dform, -mpower9-dform-vector,
25283 -mpower9-dform-scalar) and -mno-direct-move.
25285 2017-03-27 Richard Biener <rguenther@suse.de>
25287 PR tree-optimization/80181
25288 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
25290 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
25292 * config/arc/predicates.md (move_double_src_operand): Replace the
25293 call to move_double_src_operand with a call to address_operand.
25295 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
25297 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
25298 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
25299 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
25301 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
25303 * config/arc/predicates.md (long_immediate_loadstore_operand):
25304 Consider scaled addresses cases.
25306 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
25308 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
25309 restored when in interrupt.
25310 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
25311 doesn't have delay slot.
25313 2017-03-27 Richard Biener <rguenther@suse.de>
25316 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
25317 inlined thunk clones.
25319 2017-03-27 Jakub Jelinek <jakub@redhat.com>
25322 * asan.c (instrument_derefs): Copy over last operand from
25323 original COMPONENT_REF to the new COMPONENT_REF with
25324 DECL_BIT_FIELD_REPRESENTATIVE.
25325 * ubsan.c (instrument_object_size): Likewise.
25327 2017-03-27 Richard Biener <rguenther@suse.de>
25329 PR tree-optimization/80170
25330 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
25331 sure DR/SCEV didnt fold in constants we do not see when looking
25332 at the reference base alignment.
25334 2017-03-27 Richard Biener <rguenther@suse.de>
25336 PR middle-end/80171
25337 * gimple-fold.c (fold_ctor_reference): Properly guard against
25338 NULL return value from canonicalize_constructor_val.
25340 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
25343 * config/i386/i386.c (ix86_expand_builtin)
25344 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
25345 flags reg setting and flags reg using instructions.
25346 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
25347 clobbering instructions to zero extend op2.
25349 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
25351 * doc/install.texi (Configuration) <--with-aix-soname>:
25352 Update link to AIX ld.
25354 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
25356 PR rtl-optimization/80160
25357 PR rtl-optimization/80159
25358 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
25359 reg_alternate_class into account.
25361 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
25364 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
25365 to consider in curr_insn_transform.
25367 2017-03-24 Jakub Jelinek <jakub@redhat.com>
25369 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
25370 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
25371 and emit_mode_inner.
25373 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25375 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
25376 argument to the overloaded builtin variants. Use the new flag to
25377 deprecate certain builtin variants.
25378 * config/s390/s390-builtin-types.def: Add new builtin types.
25379 * config/s390/s390-builtins.h: Support new flags field for
25380 overloaded builtins.
25381 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
25382 (s390_macro_to_expand): Enable vector float data type.
25383 (s390_cpu_cpp_builtins_internal): Indicate support of the new
25384 builtins by incrementing the __VEC__ version number.
25385 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
25387 (s390_resolve_overloaded_builtin): Emit error messages depending
25388 on the builtin flags.
25389 * config/s390/s390.c (s390_expand_builtin): Support additional
25390 flags argument. Change error message to match the messages
25391 emitted in s390-c.c.
25392 * config/s390/s390.md: New UNSPEC_* constants.
25393 (op_type): Add new instruction types.
25394 * config/s390/vecintrin.h: Add new builtins and test data class
25396 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
25397 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
25398 (VEC_INEXACT, VEC_NOINEXACT): New constants.
25399 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
25400 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
25401 ("vec_mergel<mode>"): V_HW -> VEC_HW.
25403 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
25404 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
25405 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
25406 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
25408 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
25409 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
25410 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
25411 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
25413 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
25414 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
25415 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
25416 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
25417 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
25418 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
25419 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
25421 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
25422 ("vec_scatter_element<V_HW_4:mode>_DI")
25423 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
25424 ("vec_fpint<mode>", "vflls")
25425 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
25426 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
25427 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
25428 ("*vec_cmphe<mode>_cc"): ... these.
25430 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
25431 mode constant instead of magic value.
25433 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25435 * config/s390/s390.c (s390_expand_vec_compare): Support other
25436 vector floating point modes than just V2DF.
25437 (s390_expand_vcond): Likewise.
25438 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
25439 (s390_cannot_change_mode_class): Prevent mode changes between TF
25440 and V1TF in vector registers.
25441 * config/s390/s390.md (DF, SF): New mode attributes.
25442 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
25443 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
25444 SFmode support for VRs.
25445 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
25447 (VFT, VF_HW): New mode iterators.
25448 (vw, sdx): New mode attributes.
25449 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
25450 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
25451 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
25452 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
25453 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
25454 also the new vector floating point modes. Renaming to ...
25456 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
25457 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
25458 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
25459 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
25460 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
25461 ("vec_unordered<mode>"): ... these.
25463 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
25464 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
25465 ("*vec_extendv2df"): New insn definitions.
25467 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25469 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
25470 ("mulditi3_2", "*muldi3_sign"): New patterns.
25471 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
25472 rename the pattern definition.
25474 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25476 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
25478 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
25480 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25482 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
25483 instruction if possible.
25484 * config/s390/vector.md (vec_halfnumelts): New mode
25486 ("*vec_vllezlf<mode>"): New pattern.
25488 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25490 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
25491 ("popcountv4si2", "popcountv2di2"): Rename to ...
25492 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
25493 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
25495 ("popcount<mode>2_vxe"): New pattern.
25497 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25499 * common/config/s390/s390-common.c (processor_flags_table): Add
25501 * config.gcc: Add arch12.
25502 * config/s390/driver-native.c (s390_host_detect_local_cpu):
25503 Default to arch12 for unknown CPU model numbers.
25504 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
25505 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
25506 PROCESSOR_max sanity check.
25507 * config/s390/s390-opts.h (enum processor_type): Add
25509 * config/s390/s390.c (processor_table): Add arch12.
25510 (s390_expand_builtin): Add check for B_VXE flag.
25511 (s390_issue_rate): Add PROCESSOR_ARCH12.
25512 (s390_get_sched_attrmask): Likewise.
25513 (s390_get_unit_mask): Likewise.
25514 (s390_sched_score): Enable z13 scheduling for arch12.
25515 (s390_sched_reorder): Likewise.
25516 (s390_sched_variable_issue): Likewise.
25517 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
25519 (s390_tune_attr): Use z13 scheduling also for arch12.
25520 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
25521 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
25522 (TARGET_VXE_P): New macros.
25523 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
25524 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
25525 * config/s390/s390.opt: Add arch12 as processor_type.
25527 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25529 * config/s390/s390.md
25530 ("fixuns_truncdddi2", "fixuns_trunctddi2")
25531 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
25532 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
25534 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
25535 Rename expanders to ...
25537 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
25538 ("fixuns_truncdddi2_emu"): ... these.
25540 ("fixuns_trunc<mode>si2_emu"): New expander.
25542 ("*fixuns_truncdfdi2_z13"): Rename to ...
25543 ("*fixuns_truncdfdi2_vx"): ... this.
25545 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25547 * config/s390/2964.md: Remove the single element vector compare
25548 instructions which are no longer used.
25549 * config/s390/s390.c (s390_select_ccmode): Remove handling of
25551 (s390_canonicalize_comparison): Remove handling of DFmode
25553 (s390_expand_vec_compare_scalar): Remove function.
25554 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
25555 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
25557 ("*cmp<mode>_ccs"): Add wfcdb instruction.
25559 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25561 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
25563 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
25564 will anyway by matched by mov<mode>_64dfp.
25566 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25568 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
25569 vlef/vstef. Add missing operand to vleif.
25571 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25573 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
25574 pair for all vector types with 64 bit elements.
25575 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
25576 * config/s390/vector.md (V_HW_64): ... here.
25577 (V_128_NOSINGLE): New mode iterator.
25578 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
25579 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
25580 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
25581 ("*vec_load_pairv2di"): Change to ...
25582 ("*vec_load_pair<mode>"): ... this one.
25584 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25586 * config/s390/constraints.md: Add comments.
25587 (jKK): Reject element sizes > 8 bytes.
25588 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
25590 * config/s390/s390.md: Add the s_operand checks formerly in
25591 s390_split_ok_p to various splitters where they are still
25593 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
25594 for 128 bit vectors. Plus two splitters.
25596 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25598 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
25601 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25604 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
25605 error if the boundary argument is not constant.
25607 2017-03-24 Jakub Jelinek <jakub@redhat.com>
25609 PR rtl-optimization/80112
25610 * loop-doloop.c (doloop_condition_get): Don't check condition
25611 if cmp isn't SET with IF_THEN_ELSE src.
25613 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25615 PR tree-optimization/80158
25616 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
25617 replacing a candidate statement, also replace it for the
25618 candidate's alternate interpretation.
25619 (replace_rhs_if_not_dup): Likewise.
25620 (replace_one_candidate): Likewise.
25622 2017-03-24 Richard Biener <rguenther@suse.de>
25624 PR tree-optimization/80167
25625 * graphite-isl-ast-to-gimple.c
25626 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
25628 (translate_isl_ast_to_gimple::get_rename): Likewise.
25630 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
25632 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
25633 handling of certain combinations of target options, including the
25634 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
25635 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
25637 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25640 * config/arm/arm.md (*load_multiple): Add reload_completed to
25641 matching condition.
25643 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25644 Richard Biener <rguenth@suse.de>
25646 PR tree-optimization/79908
25647 PR tree-optimization/80136
25648 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
25649 been cast away, gimplify_and_add suffices.
25651 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
25653 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
25655 2017-03-23 Richard Biener <rguenther@suse.de>
25657 PR tree-optimization/80032
25658 * gimplify.c (gimple_push_cleanup): Forced unconditional
25659 cleanups still have to go to the conditional_cleanups
25662 2017-03-22 Jakub Jelinek <jakub@redhat.com>
25664 PR tree-optimization/80072
25665 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
25667 (next_operand_entry_id): Change type to unsigned int.
25668 (sort_by_operand_rank): Make sure to return the right return value
25669 even if unsigned fields are bigger than INT_MAX.
25670 (struct oecount): Change cnt and id type to unsigned int.
25671 (oecount_hasher::equal): Formatting fix.
25672 (oecount_cmp): Make sure to return the right return value
25673 even if unsigned fields are bigger than INT_MAX.
25674 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
25677 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
25678 TREE_READONLY on result if writing it more than once.
25681 * doc/invoke.texi (-fsanitize=thread): Document that with
25682 -fnon-call-exceptions atomics are not able to throw
25686 * tsan.c: Include tree-eh.h.
25687 (instrument_builtin_call): Call maybe_clean_eh_stmt or
25688 maybe_clean_or_replace_eh_stmt where needed.
25689 (instrument_memory_accesses): Add cfg_changed argument.
25690 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
25691 if it returned true.
25692 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
25694 PR rtl-optimization/63191
25695 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
25696 wrapper function, moved the whole old content into ...
25697 (ix86_delegitimize_address_1): ... this. New inline function.
25698 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
25699 true as last argument instead of ix86_delegitimize_address.
25701 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
25703 * config/aarch64/aarch64.c (generic_branch_cost): Copy
25704 cortexa57_branch_cost.
25706 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
25708 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
25710 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25713 * doc/md.texi (Constraints): Document wA constraint.
25714 * config/rs6000/constraints.md (wA): New.
25715 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
25716 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
25717 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
25718 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
25720 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
25723 * gimplify.c (is_oacc_declared): New function.
25724 (oacc_default_clause): Use it to set default flags for acc declared
25725 variables inside parallel regions.
25726 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
25727 declared variables.
25728 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
25729 declare attribute to any decl as necessary.
25731 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
25734 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
25735 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
25736 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
25737 (arm_arch_lpae): This.
25738 * config/arm/arm.c (arm_arch7ve): Rename into ...
25739 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
25740 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
25743 2017-03-22 Martin Liska <mliska@suse.cz>
25746 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
25747 error message instead of an ICE.
25749 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25751 * doc/extend.texi (6.11 Additional Floating Types): Revise.
25753 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
25755 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
25757 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
25760 2017-03-21 Martin Sebor <msebor@redhat.com>
25762 * doc/extend.texi: Use "cannot" instead of "can't."
25763 * doc/hostconfig.texi: Same.
25764 * doc/install.texi: Same.
25765 * doc/invoke.texi: Same.
25766 * doc/loop.texi: Same.
25767 * doc/md.texi: Same.
25768 * doc/objc.texi: Same.
25769 * doc/rtl.texi: Same.
25770 * doc/tm.texi: Same.
25771 * doc/tm.texi.in: Same.
25772 * doc/trouble.texi: Same.
25774 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
25777 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
25778 (collect_checksum_attributes): Set it.
25779 (die_checksum_ordered): Use it.
25781 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25783 PR tree-optimization/79908
25784 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
25785 change: For a VA_ARG whose LHS has been cast away, use
25786 force_gimple_operand to construct the side effects.
25788 2017-03-21 David Malcolm <dmalcolm@redhat.com>
25790 PR translation/80001
25791 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
25792 more amenable to translation.
25793 (oacc_loop_auto_partitions): Likewise.
25795 2017-03-21 Marek Polacek <polacek@redhat.com>
25796 Martin Sebor <msebor@redhat.com>
25798 PR tree-optimization/80109
25799 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
25800 on INTEGRAL_TYPE_P.
25802 2017-03-21 Jakub Jelinek <jakub@redhat.com>
25803 Segher Boessenkool <segher@kernel.crashing.org>
25806 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
25807 check reg_used_between_p between insn and one of succ or succ2
25808 depending on if succ is artificial insn not inserted into insn
25811 2017-03-21 Martin Liska <mliska@suse.cz>
25813 PR gcov-profile/80081
25814 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
25815 * doc/gcc.texi: Include gcov-dump stuff.
25816 * doc/gcov-dump.texi: New file.
25818 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
25820 PR rtl-optimization/79150
25821 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
25822 conditional jump, if the jump is the last insn of the loop.
25824 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25825 Richard Biener <rguenth@suse.de>
25827 PR tree-optimization/79908
25828 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
25829 been cast away, use force_gimple_operand to construct the side
25832 2017-03-21 Martin Liska <mliska@suse.cz>
25834 PR libfortran/79956
25835 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
25838 2017-03-21 Brad Spengler <spender@grsecurity.net>
25841 * plugin.c (htab_hash_plugin): New function.
25842 (add_new_plugin): Use it and adjust.
25843 (parse_plugin_arg_opt): Adjust.
25844 (init_one_plugin): Likewise.
25846 2017-03-21 Richard Biener <rguenther@suse.de>
25848 PR tree-optimization/80032
25849 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
25850 if set force the cleanup to happen unconditionally.
25851 (gimplify_target_expr): Push inserted clobbers with force_uncond
25852 to avoid them being removed by control-dependent DCE.
25854 2017-03-21 Richard Biener <rguenther@suse.de>
25856 PR tree-optimization/80122
25857 * tree-inline.c (copy_bb): Do not expans va-arg packs or
25858 va_arg_pack_len when the inlined call stmt requires pack
25860 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
25862 2017-03-21 Jakub Jelinek <jakub@redhat.com>
25865 * tsan.c (instrument_builtin_call): If the memory model argument
25866 is not a constant, assume it is valid.
25869 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
25872 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
25874 PR rtl-optimization/79910
25875 * combine.c (can_combine_p): Do not allow combining an I0 or I1
25876 if its dest is used by an insn before I2 (other than the combined
25877 insns themselves, which are properly handled already).
25879 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
25882 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
25884 * combine.c (record_used_regs): New static function.
25885 (try_combine): Handle situations where there is an additional
25886 instruction between I2 and I3 which needs to have a LOG_LINK
25890 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
25892 * combine.c (try_combine): Delete redundant i1 test. Call
25893 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
25895 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25898 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
25899 alternatives 13/14.
25901 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25903 PR tree-optimization/80054
25904 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
25905 the optimization if a PHI or any of its arguments is not dominated
25906 by the candidate's basis. Use gphi* rather than gimple* as
25908 (replace_profitable_candidates): Clean up a gimple* variable that
25909 should be a gphi* variable.
25911 2017-03-20 Martin Sebor <msebor@redhat.com>
25914 * doc/extend.texi (attribute constructor): Document present limitation.
25916 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
25919 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
25920 __POWER9_VECTOR__ #ifdef control, change template definition to
25921 use Power9-specific built-in function.
25922 (vec_any_eq): Likewise.
25923 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
25924 to control outcomes from this test.
25925 (vector_ae_<mode>p): For VEC_F modes, likewise.
25927 2017-03-20 Ian Lance Taylor <iant@google.com>
25929 * config/i386/i386.c (ix86_function_regparm): Save an extra
25930 register for -fsplit-stack with DECL_STATIC_CHAIN.
25932 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
25935 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
25936 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
25938 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
25940 * config/riscv/riscv.c (riscv_print_operand): Use "fence
25942 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
25945 2017-03-20 Marek Polacek <polacek@redhat.com>
25948 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
25950 2017-03-20 Richard Biener <rguenther@suse.de>
25952 PR tree-optimization/80113
25953 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
25954 allocate extra SSA name for PHI def.
25955 (add_close_phis_to_outer_loops): Likewise.
25956 (add_close_phis_to_merge_points): Likewise.
25957 (copy_loop_close_phi_args): Likewise.
25958 (copy_cond_phi_nodes): Likewise.
25960 2017-03-20 Martin Liska <mliska@suse.cz>
25962 PR middle-end/79753
25963 * tree-chkp.c (chkp_build_returned_bound): Do not build
25964 returned bounds for a LHS that's not a BOUNDED_P type.
25966 2017-03-20 Martin Liska <mliska@suse.cz>
25970 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
25971 COMPLEX_CST and VECTOR_CST.
25973 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25976 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
25977 target operand. A new splitter adds the clobber statement in case
25978 the target operand is dead anyway.
25980 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
25982 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
25983 to age-old versions of binutils and glibc.
25985 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
25987 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
25989 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
25991 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
25993 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
25995 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
25996 requirement for binutils 2.13.
25998 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
26000 * combine.c (try_combine): Delete redundant i1 test. Call
26001 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
26003 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
26005 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
26006 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
26008 <riscv64-*-elf>: Re-arrange section
26009 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
26010 <riscv32-*-linux>: Likewise.
26011 <riscv64-*-elf>: Likewise
26012 <riscv64-*-linux>: Likewise.
26014 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
26017 * aarch64.opt(verbose-cost-dump): Fix typo.
26019 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
26022 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
26023 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
26025 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
26027 * reload.c (find_reloads): When reloading a nonoffsettable address,
26028 use RELOAD_OTHER for it and its address reloads.
26030 PR rtl-optimization/79910
26031 * combine.c (record_used_regs): New static function.
26032 (try_combine): Handle situations where there is an additional
26033 instruction between I2 and I3 which needs to have a LOG_LINK
26036 2017-03-17 Jeff Law <law@redhat.com>
26038 PR tree-optimization/71437
26039 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
26040 conditional in the hash table first.
26041 (vrp_dom_walker::before_dom_children): Extract condition from
26042 ASSERT_EXPR. Record condition, its inverion and any implied
26043 conditions as well.
26045 2017-03-17 Marek Polacek <polacek@redhat.com>
26046 Markus Trippelsdorf <markus@trippelsdorf.de>
26048 PR tree-optimization/80079
26049 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
26052 2017-03-17 Richard Biener <rguenther@suse.de>
26054 PR middle-end/80075
26055 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
26056 Properly verify the LHS before the RHS possibly claims to be
26058 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
26061 2017-03-17 Martin Jambor <mjambor@suse.cz>
26063 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
26064 (List of -O2 options): Likewise.
26065 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
26068 2017-03-17 Tom de Vries <tom@codesourcery.com>
26070 * gcov-dump.c (print_usage): Print bug_report_url.
26072 2017-03-17 Richard Biener <rguenther@suse.de>
26074 PR middle-end/80050
26075 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
26076 (parser::peek): Likewise.
26078 2017-03-17 Richard Biener <rguenther@suse.de>
26080 PR tree-optimization/80048
26081 * sese.c (free_sese_info): Properly release rename_map and
26082 copied_bb_map elements.
26084 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
26086 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
26087 Add linked-list forward and backlinks. Insert on
26088 construction, remove on destruction.
26089 (class pass_store_merging): Add m_stores_head field.
26090 (pass_store_merging::terminate_and_process_all_chains):
26091 Iterate over m_stores_head list.
26092 (pass_store_merging::terminate_all_aliasing_chains):
26094 (pass_store_merging::execute): Check for debug stmts first.
26095 Push new chains onto the m_stores_head stack.
26097 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
26100 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
26101 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
26102 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
26104 2017-03-16 Jeff Law <law@redhat.com>
26106 PR tree-optimization/71437
26107 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
26108 member function. Implementation moved into after_dom_children
26109 member function and into the threader's thread_outgoing_edges
26111 (dom_opt_dom_walker::after_dom_children): Simplify by moving
26112 some code into new thread_outgoing_edges.
26113 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
26114 definition. Simplify marker handling (do it here). Assume we always
26115 have the available expression and the const/copies tables.
26116 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
26118 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
26119 * tree-vrp.c (equiv_stack): No longer file scoped.
26120 (vrp_dom_walker): New class.
26121 (vrp_dom_walker::before_dom_children): New member function.
26122 (vrp_dom_walker::after_dom_children): Likewise.
26123 (identify_jump_threads): Setup domwalker. Use it rather than
26124 walking edges in a random order by hand. Simplify setup/finalization.
26125 (finalize_jump_threads): Remove.
26126 (vrp_finalize): Do not call identify_jump_threads here.
26127 (execute_vrp): Do it here instead and call thread_through_all_blocks
26130 PR tree-optimization/71437
26131 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
26133 (simplify_stmt_for_jump_threading): Add basic_block argument. All
26135 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
26136 (dom_opt_dom_walker::thread_across_edge): Remove
26137 handle_dominating_asserts argument. All callers changed.
26138 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
26139 changes. Remove calls to lhs_of_dominating_assert. Other
26140 uses of handle_dominating_asserts turn into unconditional code
26141 (simplify_control_stmt_condition_1): Likewise.
26142 (simplify_control_stmt_condition): Likewise.
26143 (thread_through_normal_block, thread_across_edge): Likewise.
26144 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
26145 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
26146 object if it is not an SSA_NAME.
26147 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
26148 before calling into the VRP specific simplifiers.
26149 (identify_jump_threads): Remove handle_dominating_asserts
26152 2017-03-16 Jakub Jelinek <jakub@redhat.com>
26155 * tree-diagnostic.c (default_tree_printer): No longer static.
26156 * tree-diagnostic.h (default_tree_printer): New prototype.
26158 2017-03-16 Tamar Christina <tamar.christina@arm.com>
26160 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
26161 Change ins into fmov.
26163 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26165 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
26166 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
26167 Use h_con constraint for operand 1.
26168 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
26169 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
26171 2017-03-15 Jeff Law <law@redhat.com>
26173 PR tree-optimization/71437
26174 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
26175 (record_temporary_equivalences): Use it.
26177 PR tree-optimization/71437
26178 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
26179 tree-ssa-scopedtables.
26180 (lookup_avail_expr, build_and_record_new_cond): Likewise.
26181 (record_conditions, record_cond, vuse_eq): Likewise.
26182 (record_edge_info): Adjust to API tweak of record_conditions.
26183 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
26184 (record_temporary_equivalences, optimize_stmt): Likewise.
26185 (eliminate_redundant_computations): Likewise.
26186 (record_equivalences_from_stmt): Likewise.
26187 * tree-ssa-scopedtables.c: Include options.h and params.h.
26188 (vuse_eq): New function, moved from tree-ssa-dom.c
26189 (build_and_record_new_cond): Likewise.
26190 (record_conditions): Likewise. Accept vector of conditions rather
26191 than edge_equivalence structure for first argument.
26192 for the first argument.
26193 (avail_exprs_stack::lookup_avail_expr): New member function, moved
26194 from tree-ssa-dom.c.
26195 (avail_exprs_stack::record_cond): Likewise.
26196 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
26197 from tree-ssa-dom.c.
26198 (avail_exprs_stack): Add new member functions lookup_avail_expr
26200 (record_conditions): Declare.
26202 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
26205 * lra-constraints.c (process_alt_operands): Increase reject for
26206 reloading an input/output operand.
26208 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
26211 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
26212 insns to convert from signed/unsigned char/short to IEEE 128-bit
26214 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
26216 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
26219 * config/i386/i386.c (ix86_vector_duplicate_value): Create
26220 subreg of inner mode for values already in registers.
26222 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
26224 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
26225 iteration reg is used after the loop.
26227 2017-03-14 Martin Sebor <msebor@redhat.com>
26229 PR tree-optimization/79800
26230 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
26231 precision in negative-positive range.
26232 (format_floating): Call non-const overload with adjusted precision.
26234 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
26237 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
26240 2017-03-14 Martin Sebor <msebor@redhat.com>
26242 PR middle-end/80020
26243 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
26244 * builtins.def (aligned_alloc): Use it.
26247 * Makefile.in (GTFILES): Add calls.c.
26248 * calls.c: Include "gt-calls.h".
26250 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
26252 PR rtl-optimization/79728
26253 * regs.h (struct target_regs): New field
26254 x_contains_allocatable_regs_of_mode.
26255 (contains_allocatable_regs_of_mode): New macro.
26256 * reginfo.c (init_reg_sets_1): Initialize it, and change
26257 contains_reg_of_mode so it includes global regs as well.
26258 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
26259 rather than contains_regs_of_mode.
26261 2017-03-14 Martin Liska <mliska@suse.cz>
26263 * doc/invoke.texi: Document options that can't be combined with
26264 -fcheck-pointer-bounds.
26266 2017-03-14 Martin Liska <mliska@suse.cz>
26268 PR middle-end/79831
26269 * doc/invoke.texi (-Wchkp): Document the option.
26271 2017-03-14 Martin Liska <mliska@suse.cz>
26273 * Makefile.in: Install gcov-dump.
26275 2017-03-14 Martin Liska <mliska@suse.cz>
26277 * multiple_target.c (expand_target_clones): Bail out for
26278 an invalid attribute.
26280 2017-03-14 Richard Biener <rguenther@suse.de>
26282 * alias.c (struct alias_set_entry): Pack properly.
26283 * cfgloop.h (struct loop): Likewise.
26284 * cse.c (struct set): Likewise.
26285 * ipa-utils.c (struct searchc_env): Likewise.
26286 * loop-invariant.c (struct invariant): Likewise.
26287 * lra-remat.c (struct cand): Likewise.
26288 * recog.c (struct change_t): Likewise.
26289 * rtl.h (struct address_info): Likewise.
26290 * symbol-summary.h (function_summary): Likewise.
26291 * tree-loop-distribution.c (struct partition): Likewise.
26292 * tree-object-size.c (struct object_size_info): Likewise.
26293 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
26294 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
26295 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
26296 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
26297 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
26298 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
26299 (struct _stmt_vec_info): Likewise.
26301 2017-03-14 Martin Liska <mliska@suse.cz>
26304 * multiple_target.c (create_dispatcher_calls): Check that
26305 a target can create a function dispatcher.
26307 2017-03-14 Martin Liska <mliska@suse.cz>
26310 * multiple_target.c (expand_target_clones): Drop local.local
26311 flag for default implementation.
26313 2017-03-14 Richard Biener <rguenther@suse.de>
26315 PR tree-optimization/80030
26316 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
26318 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
26320 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
26321 gcc_fallthrough() instead of __attribute__((fallthrough));
26323 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
26325 * doc/gcc.texi: Remove "up" link to (DIR).
26326 * doc/gccint.texi: Ditto.
26328 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
26330 * doc/install.texi (Specific) <avr>: Remove reference to
26333 2017-03-13 Jeff Law <law@redhat.com>
26335 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
26336 attribute rather than comments.
26338 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
26339 match_scratch operand is highest.
26341 2017-03-13 Martin Liska <mliska@suse.cz>
26343 PR middle-end/78339
26344 * ipa-pure-const.c (warn_function_noreturn): If the declarations
26345 is a CHKP clone, use original declaration.
26347 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
26349 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
26350 (arc_conditional_register_usage): Use a different allocation order
26351 when optimizing for size.
26352 * common/config/arc/arc-common.c (arc_option_optimization_table):
26353 Section anchors default on when optimizing for size.
26355 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
26357 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
26359 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
26361 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
26362 * config/arc/arc.md (cpu_facility): Add cd variant.
26363 (*movqi_insn): Add code density variant.
26364 (*movhi_insn): Likewise.
26365 (*movqi_insn): Likewise.
26366 (*addsi3_mixed): Likewise.
26367 (subsi3_insn): Likewise.
26369 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
26371 * config/arc/arc.md (movsi_cond_exec): Update constraint.
26373 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
26375 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
26376 expressions with MINUS and UNARY ops.
26378 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26381 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
26383 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
26384 between vec_select and vector argument.
26385 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
26386 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
26387 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
26388 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
26389 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
26390 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
26392 2017-03-13 Richard Biener <rguenther@suse.de>
26395 * params.def (vect-max-peeling-for-alignment): Fix typo.
26397 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
26399 * doc/install.texi (Specific) <mips-*-*>: Remove description of
26400 issue that only occurred with binutils below 2.18.
26402 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
26404 * doc/install.texi (Specific) <cris-axis-elf>: No longer
26405 refer to binutils 2.11/2.12 minimum.
26407 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
26409 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
26410 ftp.kernel.org and simplify binutils requirement.
26412 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
26414 * doc/invoke.texi (Warning Options): Fix spelling of link-time
26416 (Optimize Options): Ditto. Also remove redundancy.
26418 2017-03-10 David Malcolm <dmalcolm@redhat.com>
26420 PR translation/79848
26421 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
26423 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
26424 to G_ to avoid double translation.
26426 2017-03-10 David Malcolm <dmalcolm@redhat.com>
26428 PR translation/79923
26429 * auto-profile.c (get_combined_location): Convert leading
26430 character of diagnostics to lower case and remove trailing period.
26431 (read_profile): Likewise for various diagnostics.
26432 * config/arm/arm.c (arm_option_override): Remove trailing period
26433 from various diagnostics.
26434 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
26435 (msp430_expand_delay_cycles): Likewise.
26437 2017-03-10 David Malcolm <dmalcolm@redhat.com>
26440 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
26441 full command-line argument, rather than just "str".
26442 (aarch64_validate_march): Likewise.
26443 (aarch64_validate_mtune): Likewise.
26445 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
26447 PR rtl-optimization/78911
26448 * lra-assigns.c (must_not_spill_p): New function.
26449 (spill_for): Use it.
26451 2017-03-10 Jakub Jelinek <jakub@redhat.com>
26453 PR tree-optimization/79981
26454 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
26455 ATOMIC_COMPARE_EXCHANGE ifn result.
26456 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
26457 IFN_ATOMIC_COMPARE_EXCHANGE.
26459 2017-03-10 David Malcolm <dmalcolm@redhat.com>
26462 * opts.c (parse_sanitizer_options): Add missing question mark to
26463 "did you mean" message.
26465 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26467 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
26469 (VMULEUH_UNS): Likewise.
26470 (VMULOUB_UNS): Likewise.
26471 (VMULOUH_UNS): Likewise.
26472 * config/rs6000/rs6000.c (builtin_function_type): Remove
26473 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
26475 2017-03-10 David Malcolm <dmalcolm@redhat.com>
26478 * read-rtl-function.c (function_reader::read_rtx_operand): Update
26479 x with result of extra_parsing_for_operand_code_0.
26480 (function_reader::extra_parsing_for_operand_code_0): Convert
26481 return type from void to rtx, returning x. When reading
26482 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
26483 larger size containing struct block_symbol.
26485 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
26487 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
26488 -mfloat128-hardware without -m64.
26490 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
26493 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
26494 entries to the case statement that marks unsigned arguments to
26495 overloaded functions.
26497 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
26499 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
26500 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
26502 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
26505 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
26506 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
26508 2017-03-10 Martin Liska <mliska@suse.cz>
26512 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
26513 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
26516 2017-03-10 Olivier Hainque <hainque@adacore.com>
26518 * tree-switch-conversion (array_value_type): Start by resetting
26519 candidate type to it's main variant.
26521 2017-03-10 Jakub Jelinek <jakub@redhat.com>
26523 PR rtl-optimization/79909
26524 * combine.c (try_combine): Use simplify_replace_rtx on individual
26525 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
26526 of the whole CALL_INSN_FUNCTION_USAGE.
26528 PR tree-optimization/79972
26529 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
26530 get_range_info on SSA_NAMEs. Formatting fixes.
26532 2017-03-10 Richard Biener <rguenther@suse.de>
26533 Jakub Jelinek <jakub@redhat.com>
26535 PR tree-optimization/77975
26536 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
26537 edge to be constant.
26538 (get_val_for): For constant x return it. Formatting fix.
26539 (loop_niter_by_eval): Avoid pointless looping if the next iteration
26540 would use the same bases as the current one.
26542 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26544 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
26545 instead of vec_select for V1TImode.
26546 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
26548 (VSX_LE_128): Add V1TI to this mode iterator.
26549 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
26550 (*vsx_le_perm_store_<mode>): Likewise.
26551 (pre-reload splitter for VSX stores): Likewise.
26552 (post-reload splitter for VSX stores): Likewise.
26553 (*vsx_xxpermdi2_le_<mode>): Likewise.
26554 (*vsx_lxvd2x2_le_<mode>): Likewise.
26555 (*vsx_stxvd2x2_le_<mode>): Likewise.
26557 2017-03-09 Michael Eager <eager@eagercon.com>
26559 Correct failures with --enable-checking=yes,rtl.
26561 * config/microblaze/microblaze.c (microblaze_expand_shift):
26562 Replace GET_CODE test with CONST_INT_P and INTVAL test with
26563 test for const0_rtx.
26564 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
26565 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
26567 2017-03-09 Richard Biener <rguenther@suse.de>
26569 PR tree-optimization/79977
26570 * graphite-scop-detection.c (scop_detection::merge_sese):
26571 Handle the case of extra exits to blocks dominating the entry.
26573 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
26575 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
26578 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
26580 PR rtl-optimization/79949
26581 * lra-constraints.c (process_alt_operands): Check memory when
26582 trying to predict a cycle. Print about the overall increase.
26584 2017-03-09 Richard Biener <rguenther@suse.de>
26586 PR middle-end/79971
26587 * gimple-expr.c (useless_type_conversion_p): Preserve
26588 TYPE_SATURATING for fixed-point types.
26590 2017-03-09 Richard Biener <rguenther@suse.de>
26593 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
26594 alignment of BLKmode params.
26596 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26599 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
26600 (VALL_NO_V2Q): Likewise.
26602 * config/aarch64/aarch64-simd.md
26603 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
26605 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
26606 VALL_NO_V2Q mode iterator.
26607 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
26609 2017-03-09 Martin Liska <mliska@suse.cz>
26611 PR tree-optimization/79631
26612 * tree-chkp-opt.c (chkp_is_constant_addr): Call
26613 tree_int_cst_sign_bit just for INTEGER constants.
26615 2017-03-09 Martin Liska <mliska@suse.cz>
26619 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
26622 2017-03-09 Marek Polacek <polacek@redhat.com>
26625 * tree.c (inchash::add_expr): Handle TREE_VEC.
26627 2017-03-09 Martin Liska <mliska@suse.cz>
26630 (chkp_narrow_size_and_offset): New function.
26631 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
26632 (void chkp_parse_bit_field_ref): New function.
26633 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
26634 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
26636 2017-03-09 Martin Liska <mliska@suse.cz>
26639 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
26640 (chkp_find_bounds_1): Remove gcc_unreachable.
26642 2017-03-09 Jakub Jelinek <jakub@redhat.com>
26645 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
26646 BUILT_IN_SYNC*, determine the access type from the size suffix and
26647 always build a MEM_REF with that type. Handle forgotten
26648 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
26651 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
26652 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
26653 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
26654 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
26655 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
26656 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
26657 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
26658 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
26659 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
26660 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
26661 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
26662 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
26663 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
26664 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
26665 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
26666 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
26667 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
26668 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
26669 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
26670 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
26671 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
26672 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
26673 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
26674 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
26675 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
26676 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
26677 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
26678 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
26679 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
26680 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
26681 definitions outside of __OPTIMIZE__ guarded section.
26684 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
26685 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
26686 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
26687 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
26690 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26692 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
26693 ("vfenez<mode>"): Add missing constraints.
26695 2017-03-08 Martin Sebor <msebor@redhat.com>
26698 * config/nds32/nds32.c (nds32_option_override):
26699 Fix misspelled diagnostic.
26701 2017-03-08 Jakub Jelinek <jakub@redhat.com>
26704 * gimplify.c (gimplify_omp_for): Replace index var in outer
26705 taskloop statement with an artificial variable and add
26706 OMP_CLAUSE_PRIVATE clause for it.
26708 2017-03-08 Richard Biener <rguenther@suse.de>
26710 PR tree-optimization/79955
26711 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
26712 for accesses that are completely outside of the variable.
26714 2017-03-08 Andrew Haley <aph@redhat.com>
26716 PR tree-optimization/79943
26717 * tree-ssa-loop-split.c (compute_new_first_bound): When
26718 calculating the new upper bound, (END-BEG) should be added, not
26721 2017-03-08 Jakub Jelinek <jakub@redhat.com>
26723 * config/avr/avr.md (setmemhi): Make sure match_dup
26724 operand number comes before match_scratch.
26726 2017-03-08 Richard Biener <rguenther@suse.de>
26728 PR tree-optimization/79920
26729 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
26730 with ncopies == 1 to ...
26731 (vect_transform_slp_perm_load): ... here. Properly compute
26732 all element loads by iterating VF times over the group. Do
26733 not handle ncopies (computed in a broken way) in
26734 vect_create_mask_and_perm.
26736 2017-03-08 Jakub Jelinek <jakub@redhat.com>
26739 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
26740 is a uniform vector, use uniform_vector_p return value instead of
26741 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
26743 2017-03-07 Marek Polacek <polacek@redhat.com>
26745 PR middle-end/79809
26746 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
26747 (alloca_call_type): Likewise.
26749 2017-03-07 Martin Liska <mliska@suse.cz>
26751 * gcov.c (process_args): Put comment to correct location.
26753 2017-03-07 Martin Liska <mliska@suse.cz>
26755 PR middle-end/68270
26756 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
26757 Use array_at_struct_end_p instead of DECL_CHAIN (field).
26758 (chkp_narrow_bounds_for_field): Likewise.
26759 (chkp_parse_array_and_component_ref): Pass one more argument to
26762 2017-03-07 Richard Biener <rguenther@suse.de>
26764 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
26767 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
26769 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
26770 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
26772 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26775 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
26776 to end of description.
26777 (PARAM_MAX_STORES_TO_MERGE): Likewise.
26779 2017-03-07 Jakub Jelinek <jakub@redhat.com>
26781 PR rtl-optimization/79901
26782 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
26784 (*avx512f_<code><mode>3<mask_name>): ... this.
26785 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
26786 iterator instead of VI8_AVX2_AVX512BW.
26788 PR rtl-optimization/79901
26789 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
26790 min/max expander, expand it using expand_vec_cond_expr.
26793 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
26796 2017-03-06 Jakub Jelinek <jakub@redhat.com>
26799 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
26800 to void * for PCH reasons.
26801 * dwarf2out.c (output_loc_operands, output_die): Cast
26802 v.val_vec.array to unsigned char *.
26804 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
26807 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
26810 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
26812 PR rtl-optimization/79571
26813 * lra-constraints.c (process_alt_operands): Calculate static
26814 reject and subtract it from overall when only addresses will be
26817 2017-03-06 Julia Koval <julia.koval@intel.com>
26820 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
26821 incoming stack boundary to 128 for 64-bit targets.
26823 2017-03-06 Richard Biener <rguenther@suse.de>
26825 PR tree-optimization/79894
26826 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
26827 to NULL after folding it.
26829 2017-03-06 Richard Biener <rguenther@suse.de>
26831 PR tree-optimization/79824
26832 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
26833 check disabling peeling for gaps.
26835 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
26837 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
26838 attributes): Document gettimeofday.
26840 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
26842 * config/s390/s390.c (s390_option_override_internal): Set
26843 PARAM_MIN_VECT_LOOP_BOUND
26845 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
26847 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
26848 * config/s390/s390.md: Likewise.
26850 2017-03-06 Jakub Jelinek <jakub@redhat.com>
26853 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
26854 (<avx2_avx512>_perm<mode>): Rename to ...
26855 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
26857 (<avx512>_perm<mode>_mask): Rename to ...
26858 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
26860 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
26861 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
26862 instead of VI8F_256_512.
26863 (avx512f_perm<mode>): New define_expand.
26864 (avx512f_perm<mode>_mask): Likewise.
26865 (avx512f_perm<mode>_1<mask_name>): New define_insn.
26866 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
26868 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
26870 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
26871 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
26873 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
26875 2017-03-06 Martin Liska <mliska@suse.cz>
26878 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
26879 when having a SSA NAME w/o VAR_DECL assigned to it.
26881 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
26883 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
26884 msa_dpsub_<su>_d): Fix MODE for vec_select.
26886 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
26888 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
26890 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
26892 2017-03-06 Richard Biener <rguenther@suse.de>
26894 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
26895 * plugin.c (register_plugin_info): Likewise.
26896 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
26898 2017-03-05 Jakub Jelinek <jakub@redhat.com>
26900 * config/i386/sse.md (sse_storehps, sse_storelps,
26901 avx_<castmode><avxsizesuffix>_<castmode>,
26902 avx512f_<castmode><avxsizesuffix>_<castmode>,
26903 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
26904 in condition that at least one operand is not a MEM.
26906 2017-03-03 Jakub Jelinek <jakub@redhat.com>
26908 PR middle-end/79805
26909 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
26910 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
26912 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
26913 gimple_call_nothrow_p flag based on whether original builtin can throw.
26914 If it can, emit following stmts on the fallthrough edge.
26915 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
26916 don't create new bb if inserting just debug stmts on the edge, try to
26917 insert them on the fallthru bb or just reset debug stmts.
26919 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
26922 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
26923 restore recog_data (including the operand rtxes inside it) around
26924 the call to get_insn_template.
26926 2017-03-03 Martin Sebor <msebor@redhat.com>
26928 PR tree-optimization/79699
26929 * context.c (context::~context): Free MPFR caches to avoid
26930 a memory leak on program exit.
26932 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26934 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
26935 Use wide_int::ulow () instead of .elt (0).
26937 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
26939 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
26940 (*pushxf): Limit oF constraint to 32bit targets and add oC
26941 constraint for 64bit targets.
26942 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
26943 (*pushdf): Change rmF constraint to rmC.
26945 2017-03-03 Martin Liska <mliska@suse.cz>
26947 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
26948 Remove unused variable.
26950 2017-03-03 Jakub Jelinek <jakub@redhat.com>
26953 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
26954 is a memory operand, increase num_memory.
26955 (ix86_expand_args_builtin): Likewise.
26957 2017-03-03 Jan Hubicka <jh@suse.cz>
26960 * ipa-devirt.c (maybe_record_node): Properly handle
26961 __cxa_pure_virtual visibility.
26963 2017-03-03 Martin Liska <mliska@suse.cz>
26965 PR tree-optimization/79803
26966 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
26968 (pass_loop_prefetch::execute): Disabled optimization if an
26969 assumption about L1 cache size is not met.
26971 2017-03-03 Martin Liska <mliska@suse.cz>
26973 PR rtl-optimization/79574
26974 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
26975 (hash_scan_set): Likewise.
26976 (dump_hash_table): Likewise.
26977 (hoist_code): Likewise.
26979 2017-03-03 Richard Biener <rguenther@suse.de>
26981 * fixed-value.c (fixed_from_string): Restore use of elt (1)
26982 in place of uhigh ().
26983 (fixed_convert_from_real): Likewise.
26985 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
26988 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
26990 2017-03-03 Richard Biener <rguenther@suse.de>
26992 PR middle-end/79818
26993 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
26994 TYPE_OVERFLOW_UNDEFINED check.
26996 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26998 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
27000 (vector_ae_<mode>_p): Likewise.
27001 (vector_nez_<mode>_p): Likewise.
27002 (vector_ne_v2di_p): Likewise.
27003 (vector_ae_v2di_p): Likewise.
27004 (vector_ne_<mode>_p): Likewise.
27005 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
27007 (vsx_tsqrt<mode>2_fe): Likewise.
27009 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
27012 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
27014 2017-03-02 Jakub Jelinek <jakub@redhat.com>
27016 PR rtl-optimization/79780
27017 * cprop.c (one_cprop_pass): When second and further conditional trap
27018 in a single basic block is turned into an unconditional trap, turn it
27019 into a deleted note to avoid RTL verification failures.
27021 2017-03-02 Richard Biener <rguenther@suse.de>
27023 * fold-const.c (const_binop): Use ulow () instead of elt (0).
27025 2017-03-02 Richard Biener <rguenther@suse.de>
27027 PR tree-optimization/79345
27029 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
27030 param and abort the walk, returning -1 if it is hit.
27031 (walk_aliased_vdefs): Take a limit param and pass it on.
27032 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
27033 defaulting to 0 and return a signed int.
27034 * tree-ssa-uninit.c (struct check_defs_data): New struct.
27035 (check_defs): New helper.
27036 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
27037 about uninitialized memory.
27038 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
27039 bogus uninitialized warning.
27040 (fixed_convert_from_real): Likewise.
27042 2017-03-02 Bin Cheng <bin.cheng@arm.com>
27044 PR tree-optimization/66768
27045 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
27046 iv_use if base object can't be determined.
27048 2017-03-02 Jakub Jelinek <jakub@redhat.com>
27050 PR tree-optimization/79345
27051 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
27052 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
27053 (get_pattern_stats): Initialize it.
27054 * genemit.c (gen_expand): Verify match_scratch numbers come after
27055 match_operand/match_dup numbers.
27056 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
27057 match_scratch numbers.
27058 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
27060 * config/s390/s390.md (trunctdsd2): Likewise.
27062 2017-03-02 Richard Biener <rguenther@suse.de>
27064 * wide-int.h (wide_int_storage::operator=): Implement in terms
27067 2017-03-02 Richard Biener <rguenther@suse.de>
27069 PR tree-optimization/79777
27070 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
27071 the to insert expression to sth existing.
27073 2017-03-01 Martin Sebor <msebor@redhat.com>
27075 PR middle-end/79692
27076 * gimple-ssa-sprintf.c
27077 (directive::known_width_and_precision): New function.
27078 (format_integer): Use it.
27079 (get_mpfr_format_length): Consider the full range of precision
27080 when computing %g output with the # flag. Set the likely byte
27081 count to 3 rather than 1 when precision is indeterminate.
27082 (format_floating): Correct the lower bound of precision.
27084 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27086 * doc/invoke.texi: Document default code model for 64-bit Linux.
27088 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27091 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
27092 udiv rather than div since input pattern is unsigned.
27094 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
27096 * config/i386/i386.c (print_reg): Warn for values of
27097 unsupported size in integer register.
27099 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
27102 * config/rs6000/predicates.md (current_file_function_operand): Do
27103 not allow self calls to be local if the function is replaceable.
27105 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
27108 * config/rs6000/altivec.h (vec_ctz and others): Change the
27109 preprocessor macro that controls conditional compilation from
27110 _ARCH_PWR9 to __POWER9_VECTOR__.
27111 (vec_all_ne): Change parameterization of __altivec_scalar_pred
27112 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
27113 control (instead of _ARCH_PWR9 control) so that template
27114 definition uses power9-specific function.
27115 (vec_any_eq): Likewise.
27116 (vec_all_ne): Change macro definition to use a power9-specific
27117 expansion under #ifdef __POWER9_VECTOR__ control (instead of
27118 _ARCH_PWR9 control).
27119 (vec_any_eq) Likewise.
27120 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
27121 expansion for CMPNEF to remove support for xvcmpnesp instruction.
27122 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
27123 support for xvcmpnedp instruction.
27124 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
27125 macro expansion so that Power9 implementation of vec_all_ne does
27126 not use the AltiVec predicate framework.
27127 (VCMPNEH_P): Likewise.
27128 (VCMPNEW_P): Likewise.
27129 (VCMPNED_P): Likewise.
27130 (VCMPNEFP_P): Likewise.
27131 (VCMPNEDP_P): Likewise.
27132 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
27133 implementation of vec_any_eq to not use AltiVec predicate
27135 (VCMPAEH_P): Likewise.
27136 (VCMPAEW_P): Likewise.
27137 (VCMPAED_P): Likewise.
27138 (VCMPAEFP_P): Likewise.
27139 (VCMPAEDP_P): Likewise.
27140 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
27141 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
27142 not use the AltiVec predicate framework.
27143 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
27144 of vec_any_eq to not use AltiVec predicate framework.
27145 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
27146 support for predefined __POWER9_VECTOR__ macro to indicate that
27147 Power9 instruction selection is enabled.
27148 (altivec_overloaded_builtins): Remove extraneous
27149 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
27150 function argument types RS6000_BTI_bool_V16QI and
27151 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
27152 entry for overloaded function argument types RS6000_BTI_bool_V4SI
27153 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
27154 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
27155 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
27156 Power9 for implementations of vec_cmpne. Change the signature for
27157 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
27158 (representing vec_all_ne) to remove the previously described first
27159 argument of type RS6000_BTI_INTSI, as this was an artifact of
27160 reliance on the AltiVec predicate framework, which is no longer
27161 used in the implementation of these functions. Add
27162 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
27163 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
27164 since, unlike the AltiVec predicate framework implementation, we
27165 do not share function descriptors between vec_alle and vec_anyeq.
27166 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
27167 set of modes that receive special treatment even when
27168 TARGET_P9_VECTOR is true. The special treatment emits code that
27169 does not depend on Power9 instructions.
27170 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
27171 define_expand to not rely on AltiVec predicate framework.
27172 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
27174 (vector_ne_v2di_p): Change this define_expand to not rely on
27175 AltiVec predicate framework.
27176 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
27178 (vector_ne_<mode>_p): Change this define_expand to not rely on
27179 AltiVec predicate framework.
27180 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
27182 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
27183 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
27184 define_insn pattern.
27185 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
27186 define_insn pattern because the xvcmpne<VSs>. instruction is not
27188 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
27189 instruction is not supported.
27191 2017-03-01 Jakub Jelinek <jakub@redhat.com>
27193 * config/nvptx/nvptx.c: Include intl.h.
27195 2017-03-01 Martin Jambor <mjambor@suse.cz>
27198 * ipa-prop.h (ipa_bits): Removed field known.
27199 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
27200 to pointers. Adjusted their comments to warn about their sharing.
27201 (ipcp_transformation_summary): Change bits to a vector of pointers.
27202 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
27203 (ipa_get_ipa_bits_for_value): Declare.
27204 * tree-vrp.h (value_range): Mark as GTY((for_user)).
27205 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
27206 (ipa_bits_hash_table): Likewise.
27207 (ipa_vr_ggc_hash_traits): Likewise.
27208 (ipa_vr_hash_table): Likewise.
27209 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
27210 being pointers and vr_known being removed.
27211 (ipa_set_jf_unknown): Likewise.
27212 (ipa_get_ipa_bits_for_value): New function.
27213 (ipa_set_jfunc_bits): Likewise.
27214 (ipa_get_value_range): New overloaded functions.
27215 (ipa_set_jfunc_vr): Likewise.
27216 (ipa_compute_jump_functions_for_edge): Use the above functions to
27217 construct bits and vr parts of jump functions.
27218 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
27219 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
27221 (ipcp_grow_transformations_if_necessary): Also allocate
27222 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
27224 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
27225 them. Fix too long lines.
27226 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
27227 vr_known being removed.
27228 (ipa_read_jump_function): Use new setter functions to construct bits
27229 and vr parts of jump functions or set them to NULL.
27230 (write_ipcp_transformation_info): Adjust for bits being pointers.
27231 (read_ipcp_transformation_info): Likewise.
27232 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
27234 Include gt-ipa-prop.h.
27235 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
27237 (ipcp_store_bits_results): Likewise.
27238 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
27239 Do not write to existing jump functions but use a temporary instead.
27241 2017-03-01 Jakub Jelinek <jakub@redhat.com>
27244 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
27245 attempt to use its first operand as BIT_FIELD_REF base.
27247 2017-03-01 Richard Biener <rguenther@suse.de>
27249 PR middle-end/79721
27250 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
27251 interpolating formula in wrapping arithmetic.
27252 (chrec_apply): Convert chrec_evaluate return value to wanted type.
27254 2017-03-01 Jakub Jelinek <jakub@redhat.com>
27256 PR tree-optimization/79734
27257 * tree-vect-generic.c (expand_vector_condition): Optimize
27258 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
27259 Handle VEC_COND_EXPR where comparison has different inner width from
27260 type's inner width.
27262 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
27264 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
27265 markup, and similar issues. Remove @opindex entries for things
27266 that aren't options. Add missing -mmpy-option entries.
27268 2017-02-28 Jakub Jelinek <jakub@redhat.com>
27270 PR tree-optimization/79737
27271 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
27272 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
27273 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
27274 instead of byte_size. Formatting fix.
27275 (shift_bytes_in_array_right): Formatting fix.
27277 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
27280 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
27281 condition on optimize for the leaf function test.
27283 2017-02-28 Martin Liska <mliska@suse.cz>
27286 * read-rtl-function.c (function_reader::handle_unknown_directive):
27287 Bail out when one uses -flto.
27289 2017-02-28 Martin Liska <mliska@suse.cz>
27291 * common.opt: Replace space with tabular for options of <number>
27293 * config/i386/i386.opt: Show <number> value for
27294 -mlarge-data-threshold.
27295 * opts.c (print_filtered_help): Do not display number in hexadecimal
27298 2017-02-28 Martin Liska <mliska@suse.cz>
27300 * common.opt: Fix --help=option -Q for options which are of
27303 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
27305 * config/i386/i386.c (print_reg): Error out for values
27306 of 8-bit size in invalid integer register.
27308 2017-02-28 Martin Sebor <msebor@redhat.com>
27310 PR tree-optimization/79691
27311 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
27313 2017-02-28 Jakub Jelinek <jakub@redhat.com>
27316 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
27317 gcc_unreachable with output_operand_lossage.
27319 2017-02-28 Richard Biener <rguenther@suse.de>
27321 PR tree-optimization/79740
27322 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
27324 (visit_nary_op): Insert the nary into the hashtable if we
27325 pattern-matched sth.
27326 * tree-ssa-pre.c (eliminate_insert): Robustify.
27328 2017-02-28 Richard Biener <rguenther@suse.de>
27330 PR middle-end/79731
27331 * fold-const.c (decode_field_reference): Reject out-of-bound
27334 2017-02-28 Jakub Jelinek <jakub@redhat.com>
27336 * config/i386/i386.c: Include intl.h.
27337 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
27338 instead of just cond ? "..." : "...".
27339 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
27340 * coverage.c (read_counts_file): Likewise.
27341 * omp-offload.c: Include intl.h.
27342 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
27343 of just cond ? "..." : "...".
27344 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
27345 of just cond ? "..." : "...".
27347 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
27350 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
27352 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
27353 'tune for' CPU name.
27354 * config/arm/arm-cpu-data.h: Regenerated.
27356 2017-02-28 Richard Biener <rguenther@suse.de>
27358 PR tree-optimization/79732
27359 * tree-inline.c (expand_call_inline): Do not shadow var.
27361 2017-02-28 Richard Biener <rguenther@suse.de>
27363 PR tree-optimization/79723
27364 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
27365 address-space properly.
27367 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
27369 * doc/optinfo.texi (Optimization groups): Fix option used for
27371 * doc/invoke.texi (-fopt-info): Document "omp".
27372 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
27373 (OPTGROUP_ALL): Add OPTGROUP_OMP.
27374 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
27375 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
27376 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
27378 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
27380 * dumpfile.c (optgroup_options): Instead of "openmp", associate
27381 OPTGROUP_OMP with "omp".
27383 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
27386 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
27387 for arithmetic shift of unsigned V2DI.
27389 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
27391 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
27392 arc/linux.h headers.
27393 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
27394 (LINK_SPEC): Likewise.
27395 (ARC_TLS_EXTRA_START_SPEC): Likewise.
27396 (EXTRA_SPECS): Likewise.
27397 (STARTFILE_SPEC): Likewise.
27398 (ENDFILE_SPEC): Likewise.
27399 (LIB_SPEC): Likewise.
27400 (TARGET_SDATA_DEFAULT): Likewise.
27401 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
27402 (MULTILIB_DEFAULTS): Likewise.
27403 (DWARF2_UNWIND_INFO): Likewise.
27404 * config/arc/big.h: New file.
27405 * config/arc/elf.h: Likewise.
27406 * config/arc/linux.h: Likewise.
27407 * config/arc/t-uClibc: Remove.
27409 2017-02-27 Bin Cheng <bin.cheng@arm.com>
27411 PR tree-optimization/77536
27412 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
27413 (tree_transform_and_unroll_loop): Use above function to compute the
27414 estimated niter of unrolled loop and use it when scaling profile.
27415 Also use count info rather than frequency if it's non-zero.
27416 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
27417 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
27418 (vect_transform_loop): Call above function.
27420 2017-02-27 Richard Biener <rguenther@suse.de>
27422 PR tree-optimization/45397
27423 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
27424 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
27425 (visit_nary_op): Add pattern matching for CSEing sign-changed
27426 or truncated operations with wider ones.
27428 2017-02-27 Richard Biener <rguenther@suse.de>
27430 PR tree-optimization/79690
27431 * tree-vect-stmts.c (vectorizable_store): Use vector type
27432 built from the DR with address-space.
27434 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
27436 * doc/invoke.texi (Optimize Options): Refine the description
27437 of asan-use-after-return.
27439 2017-02-25 Alan Modra <amodra@gmail.com>
27441 PR rtl-optimization/79584
27442 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
27443 base, not ad->base_term, the reg within base. Remove assertion
27444 that ad->base == ad->base_term. Replace gen_int_mode using
27445 bogus mode with const0_rtx.
27447 2017-02-25 Jakub Jelinek <jakub@redhat.com>
27449 PR middle-end/79396
27450 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
27451 FMA_EXPR like tcc_binary or tcc_unary.
27453 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
27456 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
27458 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
27459 (output_loc_operands): Handle DW_OP_call_ref and
27460 DW_OP_GNU_variable_value.
27461 (struct variable_value_struct): New type.
27462 (struct variable_value_hasher): Likewise.
27463 (variable_value_hash): New variable.
27464 (string_types): Remove.
27465 (copy_loc_descr): New function.
27466 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
27467 (prepend_loc_descr_to_each): New function.
27468 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
27469 instead of add_loc_descr_to_each if the first argument is single
27470 location list and the second has multiple.
27471 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
27472 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
27473 when looking for variable value which doesn't have other location info.
27474 (loc_list_from_tree): Formatting fix.
27475 (gen_array_type_die): Simplify DW_AT_string_length handling.
27476 (adjust_string_types): Remove.
27477 (gen_subprogram_die): Don't call adjust_string_types nor test/set
27478 string_types. Call resolve_variable_values.
27479 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
27480 (resolve_addr_in_expr): Likewise. Add A argument.
27481 (copy_deref_exprloc): Remove deref argument. Adjust for the
27482 original expression being DW_OP_GNU_variable_value with optionally
27483 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
27484 optionally after it.
27485 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
27486 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
27487 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
27488 (variable_value_hasher::hash, variable_value_hasher::equal): New
27490 (resolve_variable_value_in_expr, resolve_variable_value,
27491 resolve_variable_values, note_variable_value_in_expr,
27492 note_variable_value): New functions.
27493 (dwarf2out_early_finish): Call note_variable_value on all toplevel
27496 2017-02-24 Jakub Jelinek <jakub@redhat.com>
27499 * opts.h (handle_generated_option): Add GENERATED_P argument.
27500 * opts-common.c (handle_option): Adjust function comment.
27501 (handle_generated_option): Add GENERATED_P argument, pass it to
27503 (control_warning_option): Pass false to handle_generated_option
27505 * opts.c (maybe_default_option): Pass true to handle_generated_option
27507 * optc-gen.awk: Likewise.
27509 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27511 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
27512 a REG, look at the REG it is a SUBREG of.
27513 (splitter for cmpeqsi_t): Ditto.
27515 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27517 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
27518 the special USEs with the pattern of the insn, not the insn itself.
27520 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
27523 * doc/invoke.texi: Document -mload-store-pairs.
27525 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27526 Sandra Loosemore <sandra@codesourcery.com>
27528 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
27529 argument isn't a CONST_INT.
27530 (nios2_alternate_compare_const): Assert op is a CONST_INT.
27531 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
27532 (nios2_validate_compare): Bypass alternate compare logic if *op2
27533 is not a CONST_INT.
27534 (ldstwm_operation_p): Return false if first_base is not a REG or
27535 if first_offset is not a CONST_INT.
27537 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27539 * config/cris/cris.md: Use correct operand in a define_peephole2.
27541 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27543 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
27545 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27547 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
27548 this_insn if it is an INSN or JUMP_INSN.
27549 (force_offsettable): Look at base, not at addr.
27550 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
27551 on things that aren't necessarily CONST_INTs.
27553 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
27555 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
27556 -mfpmath=sse is the default also for x86-32 targets with SSE2
27557 instruction set when @option{-ffast-math} is enabled
27559 2017-02-24 Jeff Law <law@redhat.com>
27561 PR rtl-optimizatoin/79286
27562 * ira.c (update_equiv_regs): Drop may_trap_p exception to
27565 2017-02-24 Richard Biener <rguenther@suse.de>
27567 PR tree-optimization/79389
27568 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
27571 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
27573 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
27574 function comment to reflect reality.
27575 (loop_exits_before_overflow): Fix typo in function description.
27577 2017-02-24 Richard Biener <rguenther@suse.de>
27579 PR tree-optimization/79389
27580 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
27581 properly that a threading opportunity exists. Detect conditional
27582 copy/constant propagation opportunities.
27584 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
27586 * config/visium/visium.md (type): Add trap.
27587 (b): New mode attribute.
27588 (*btst): Rename into...
27589 (*btst<mode>): ...this and adjust.
27590 (*cbranchsi4_btst_insn): Rename into...
27591 (*cbranch<mode>4_btst_insn): ...this and adjust.
27592 (trap): New define_insn.
27594 2017-02-23 Jakub Jelinek <jakub@redhat.com>
27596 PR tree-optimization/79389
27597 * ifcvt.c (struct noce_if_info): Add rev_cond field.
27598 (noce_reversed_cond_code): New function.
27599 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
27600 reversed_comparison_code. Formatting fix.
27601 (noce_try_store_flag): Test rev_cond != NULL in addition to
27602 reversed_comparison_code.
27603 (noce_try_store_flag_constants): Likewise.
27604 (noce_try_store_flag_mask): Likewise.
27605 (noce_try_addcc): Use rev_cond if non-NULL instead of
27606 reversed_comparison_code.
27607 (noce_try_cmove_arith): Likewise. Formatting fixes.
27608 (noce_try_minmax, noce_try_abs): Clear rev_cond.
27609 (noce_find_if_block): Initialize rev_cond.
27610 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
27611 instead of false as last argument never attempt to reverse it
27614 2017-02-23 Bin Cheng <bin.cheng@arm.com>
27616 PR tree-optimization/79663
27617 * tree-predcom.c (combine_chains): Process refs in reverse order
27618 only for ZERO length chains, and add explaining comment.
27620 2017-02-23 Jeff Law <law@redhat.com>
27622 PR tree-optimization/79578
27623 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
27624 in call to operand_equal_p.
27626 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
27629 * config/i386/cpuid.h: Fix another undefined behavior.
27631 2017-02-23 Richard Biener <rguenther@suse.de>
27633 PR tree-optimization/79683
27634 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
27635 vector types for data-refs.
27637 2017-02-23 Martin Liska <mliska@suse.cz>
27639 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
27641 2017-02-23 Jakub Jelinek <jakub@redhat.com>
27643 PR middle-end/79665
27644 * internal-fn.c (get_range_pos_neg): Moved to ...
27645 * tree.c (get_range_pos_neg): ... here. No longer static.
27646 * tree.h (get_range_pos_neg): New prototype.
27647 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
27648 are known to be in between 0 and signed maximum inclusive, try to
27649 expand both unsigned and signed divmod and use the cheaper one from
27652 2017-02-22 Jeff Law <law@redhat.com>
27654 PR tree-optimization/79578
27655 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
27656 to compare base operands.
27658 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
27661 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
27662 gpc_reg_operand instead of fpr_reg_operand.
27664 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
27666 * config/mips/mips.c (mips_return_in_memory): Force FP
27667 vector types to be returned in memory for o32 ABI.
27669 2017-02-22 Jakub Jelinek <jakub@redhat.com>
27671 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
27672 instead of DW_TAG_member for static data member declarations and don't
27673 set no_linkage_name for static inline data members.
27674 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
27677 2017-02-22 Martin Liska <mliska@suse.cz>
27679 * doc/invoke.texi: Replace inequality signs with square brackets
27682 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27684 PR tree-optimization/68644
27685 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
27687 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
27690 * lra-constraints.c (simplify_operand_subreg): Handle
27691 WORD_REGISTER_OPERATIONS targets.
27693 2017-02-22 Jakub Jelinek <jakub@redhat.com>
27696 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
27697 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
27698 elimination by swapping fld*.
27700 2017-02-22 Richard Biener <rguenther@suse.de>
27702 PR tree-optimization/79673
27703 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
27704 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
27705 irrelevant address-space qualifiers and avoiding a
27706 ADDR_SPACE_CONVERT_EXPR from fold_convert.
27708 2017-02-22 Richard Biener <rguenther@suse.de>
27710 PR tree-optimization/79666
27711 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
27712 to not symbolically negate if that may introduce undefined
27715 2017-02-22 Martin Liska <mliska@suse.cz>
27718 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
27719 * data-streamer-out.c (streamer_write_gcov_count_stream):
27721 * value-prof.c (stream_out_histogram_value): Make assert more
27722 precise based on type of counter.
27724 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
27727 * config/i386/i386.md (standard_x87sse_constant_load splitter):
27728 Use nonimmediate_operand instead of memory_operand for operand 1.
27729 (float-extend standard_x87sse_constant_load splitter): Ditto.
27731 2017-02-21 Jeff Law <law@redhat.com>
27733 PR tree-optimization/79621
27734 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
27735 blocks with edges to themselves.
27737 2017-02-21 Jakub Jelinek <jakub@redhat.com>
27740 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
27741 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
27742 Use gimple_call_builtin_p.
27745 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
27746 on temporarily removed DEBUG_INSNs.
27748 PR tree-optimization/79649
27749 * tree-loop-distribution.c (classify_partition): Give up on
27750 non-generic address space loads/stores.
27752 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
27754 * doc/loop.texi (Loop manipulation): Remove nonexistent
27755 tree_ssa_loop_version from the documentation.
27756 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
27758 2017-02-21 Jakub Jelinek <jakub@redhat.com>
27761 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
27762 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
27763 * config/rs6000/rs6000.c: Include except.h.
27764 (rs6000_expand_split_stack_prologue): Call
27765 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
27767 2017-02-21 Martin Jambor <mjambor@suse.cz>
27770 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
27771 have been analyzed.
27773 2017-02-21 Martin Jambor <mjambor@suse.cz>
27775 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
27776 for backward compatibility only.
27777 * doc/invoke.texi (Option Summary): Remove all references to
27778 -fipa-cp-alignment.
27780 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
27784 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27786 * lra-constraints.c (curr_insn_transform): Handle
27787 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
27789 2017-02-21 Martin Liska <mliska@suse.cz>
27791 * config/i386/i386.opt: Replace -masm-dialect with -masm.
27793 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
27795 PR translation/79638
27796 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
27798 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
27801 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
27802 (arm_function_ok_for_sibcall): Return false for an indirect call by
27803 descriptor if all the argument registers are used.
27804 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
27805 alignment of the function.
27807 2017-02-21 Jakub Jelinek <jakub@redhat.com>
27809 PR tree-optimization/61441
27810 * simplify-rtx.c (simplify_const_unary_operation): For
27811 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
27812 the sNaN unmodified.
27814 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
27816 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
27817 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
27818 instead of SYSTEM_HEADER_DIR.
27820 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
27821 Martin Liška <mliska@suse.cz>
27823 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
27824 Fix typos and grammar, use active voice, and clarify.
27826 2017-02-20 Marek Polacek <polacek@redhat.com>
27828 PR middle-end/79537
27829 * gimplify.c (gimplify_expr): Handle unused *&&L;.
27832 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
27834 2017-02-20 Jakub Jelinek <jakub@redhat.com>
27837 * config/i386/i386.c (ix86_expand_builtin): Handle
27838 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
27839 ix86_builtins_isa[fcode].isa as a requirement of those
27840 flags and any other flag in the bitmask.
27841 (ix86_init_mmx_sse_builtins): Use 0 instead of
27842 ~OPTION_MASK_ISA_64BIT as mask.
27843 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
27844 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
27845 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
27846 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
27848 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27851 * lra-constraints.c (split_reg): Check requested split mode
27852 is supported by the register.
27854 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27856 * lra-constraints.c (simplify_operand_subreg): Remove early
27859 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27862 * lra-constraints.c (curr_insn_transform): Tighten condition
27863 for converting SUBREG reloads from OP_OUT to OP_INOUT.
27865 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27868 * lra-constraints.c (curr_insn_transform): Handle
27869 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
27871 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
27874 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
27876 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
27878 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
27881 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
27884 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27886 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
27889 2017-02-18 Jakub Jelinek <jakub@redhat.com>
27891 * final.c (last_columnnum, override_columnnum): New variables.
27892 (final_start_function): Set last_columnnum, pass it to begin_prologue
27893 hook and pass 0 to dwarf2out_begin_prologue.
27894 (final_scan_insn): Update override_columnnum. Pass last_columnnum
27895 to source_line debug hook.
27896 (notice_source_line): Compute last_columnnum and for debug_column_info
27897 return true on column changes.
27898 * debug.h (struct gcc_debug_hooks): Add column argument to
27899 source_line and begin_prologue hooks.
27900 (debug_nothing_int_charstar_int_bool): Remove prototype.
27901 (debug_nothing_int_int_charstar,
27902 debug_nothing_int_int_charstar_int_bool): New prototypes.
27903 (dwarf2out_begin_prologue): Add column argument.
27904 * debug.c (do_nothing_debug_hooks): Adjust source_line and
27905 begin_prologue hooks.
27906 (debug_nothing_int_charstar_int_bool): Remove.
27907 (debug_nothing_int_int_charstar,
27908 debug_nothing_int_int_charstar_int_bool): New functions.
27909 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
27910 through to dwarf2out_source_line.
27911 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
27912 (dwarf2out_source_line): Add column argument, emit it if requested.
27913 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
27915 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
27916 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
27917 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
27918 through to dwarf2out_begin_prologue.
27919 (vmsdbgout_source_line): Add column argument, pass it through to
27920 dwarf2out_source_line.
27921 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
27922 dbxout_source_line caller.
27923 (dbxout_source_line): Add column argument.
27925 * common.opt (gno-column-info, gcolumn-info): New options.
27926 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
27927 (check_die): Also test for multiple DW_AT_decl_column attributes.
27928 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
27929 DW_AT_decl_column if requested.
27930 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
27932 (gen_variable_die): Likewise.
27933 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
27934 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
27937 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
27938 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
27939 (ix86_handle_option): Handle OPT_m3dnowa.
27940 * doc/invoke.texi (-m3dnowa): Document.
27941 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
27942 -m3dnowa instead of -m3dnow -march=athlon.
27945 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
27946 instead of gcc_assert for K, r and R code checks. Formatting fixes.
27948 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27951 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
27952 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
27953 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
27954 generator for vsx_xxpermdi_<mode>_be.
27955 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
27956 force big-endian semantics.
27957 (vsx_xxpermdi_<mode>_be): New define_expand with same
27958 implementation as previous version of vsx_xxpermdi_<mode>.
27960 2017-02-17 Jakub Jelinek <jakub@redhat.com>
27962 PR tree-optimization/79327
27963 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
27964 variable, its initialization and use.
27966 2017-02-17 Julia Koval <julia.koval@intel.com>
27968 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
27969 (OPTION_MASK_ISA_PKU_UNSET): New.
27970 (ix86_handle_option): Handle -mrdpid.
27971 * config/i386/cpuid.h (bit_RDPID): New.
27972 * config/i386/driver-i386.c (host_detect_local_cpu):
27973 Detect RDPID feature.
27974 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
27975 * config/i386/i386-c.c (ix86_target_macros_internal):
27977 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
27978 (ix86_valid_target_attribute_inner_p): Add "rdpid".
27979 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
27980 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
27981 * config/i386/i386.md (define_insn "rdpid"): New.
27982 * config/i386/i386.opt Add -mrdpid.
27983 * config/i386/immintrin.h (_rdpid_u32): New.
27985 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
27987 PR rtl-optimization/79541
27988 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
27989 instead of transforming it into USE.
27991 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
27993 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
27994 If HONOR_SNANS (SFmode) force the input to a register.
27995 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
27996 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
27997 an frsp or similar insn.
27999 2017-02-17 Martin Liska <mliska@suse.cz>
28001 PR rtl-optimization/79577
28002 * params.def (selsched-max-sched-times): Increase minimum to 1.
28004 2017-02-17 Martin Liska <mliska@suse.cz>
28006 PR rtl-optimization/79574
28007 * gcse.c (want_to_gcse_p): Prevent integer overflow.
28009 2017-02-17 Martin Liska <mliska@suse.cz>
28011 PR tree-optimization/79529
28012 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
28013 ssa_defined_default_def_p to handle cases which are implicitly
28015 * tree-ssa.c (ssa_defined_default_def_p): New function.
28016 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
28017 which are implicitly defined.
28018 * tree-ssa.h (ssa_defined_default_def_p): Declare.
28020 2017-02-17 Richard Biener <rguenther@suse.de>
28022 PR middle-end/79576
28023 * params.def (max-ssa-name-query-depth): Limit to 10.
28025 2017-02-17 Richard Biener <rguenther@suse.de>
28027 PR tree-optimization/79552
28028 * tree-ssa-structalias.c (visit_loadstore): Properly verify
28031 2017-02-17 Richard Biener <rguenther@suse.de>
28034 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
28036 2017-02-17 Marek Polacek <polacek@redhat.com>
28038 PR middle-end/79536
28039 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
28040 (fold_negate_expr): New wrapper.
28042 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
28044 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
28045 Correct terminology and de-emphasize pre-standard behavior.
28047 2017-02-16 Alan Modra <amodra@gmail.com>
28049 PR rtl-optimization/79286
28050 * ira.c (def_dominates_uses): New function.
28051 (update_equiv_regs): Don't create an equivalence for insns that
28052 may trap where the register def does not dominate the use.
28054 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
28056 PR rtl-optimization/78127
28057 * lra.c (lra): Call lra_eliminate before finish the loop after
28060 2017-02-16 Richard Biener <rguenther@suse.de>
28062 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
28064 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
28065 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
28066 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
28067 (isl_val_int_from_wi): New function.
28068 (extract_affine_gmp): Rename to ...
28069 (extract_affine_wi): ... this, take a widest_int.
28070 (extract_affine_int): Just wrap extract_affine_wi.
28071 (add_param_constraints): Use isl_val_int_from_wi.
28072 (add_loop_constraints): Likewise, and extract_affine_wi.
28074 2017-02-15 Jeff Law <law@redhat.com>
28076 PR middle-end/79521
28077 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
28078 ira_init_register_move_cost_if_necessary.
28080 2017-02-15 Martin Sebor <msebor@redhat.com>
28082 PR middle-end/32003
28083 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
28084 removed in a prior commit.
28086 2017-02-15 Bin Cheng <bin.cheng@arm.com>
28088 PR tree-optimization/79347
28089 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
28090 counters during peeling.
28092 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
28094 * Makefile.in (site.exp): Remove "set ISLVER".
28096 2017-02-15 Jakub Jelinek <jakub@redhat.com>
28099 * real.c (real_from_integer): Call real_convert even for decimal.
28101 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
28104 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
28106 2017-02-14 Andrew Pinski <apinski@cavium.com>
28108 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
28109 cores and change the partno/implementer to be correct.
28110 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
28111 the 'B" as the implementer.
28112 * config/aarch64/aarch64-tune.md: Regenerate.
28114 2017-02-14 Carl Love <cel@us.ibm.com>
28116 * config/rs6000/rs6000.c: Add case statement entry to make the
28117 xvcvuxdsp built-in argument unsigned.
28118 * config/rs6000/vsx.md: Fix the source and return operand types so they
28119 match the instruction definitions from the ISA document. Fix typo
28120 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
28123 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
28126 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
28127 member early_clobber_alts.
28128 * lra-lives.c (reg_early_clobber_p): New.
28129 (process_bb_lives): Use it.
28130 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
28131 (debug_operand_data): Initialize early_clobber_alts.
28132 (setup_operand_alternative): Set up early_clobber_alts.
28133 (collect_non_operand_hard_regs): Ditto. Pass early clobber
28134 alternatives to new_insn_reg.
28135 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
28137 (lra_update_insn_regno_info): Pass the new arg.
28139 2017-02-14 Jakub Jelinek <jakub@redhat.com>
28141 PR middle-end/79505
28142 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
28143 (new_oacc_loop_raw): Don't clear already cleared fields.
28146 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
28147 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
28148 _mm512_prefetch_i64gather_ps): New inline functions and macros.
28150 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
28153 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
28155 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
28158 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
28159 the extra instruction to the right place to store 128-bit constant
28162 2017-02-14 Martin Sebor <msebor@redhat.com>
28164 PR middle-end/79448
28165 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
28166 warning for strings of unknown length.
28168 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
28170 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
28172 2017-02-14 Jeff Law <law@redhat.com>
28175 * ira-costs.c (scan_one_insn): Initialize register move costs
28176 for pseudos seen in USE/CLOBBER insns.
28178 PR tree-optimization/79095
28179 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
28180 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
28181 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
28182 if the operands are known to be not equal, then the resulting range
28184 (intersect_ranges): If the new range is ~[0,0] and the old range is
28185 wide, then prefer ~[0,0].
28186 * tree-vrp.c (overflow_comparison_p_1): New function.
28187 (overflow_comparison_p): New function.
28188 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
28189 if NAME is used in an overflow test.
28190 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
28191 overflow check that can be expressed as an equality test, then adjust
28192 ops to be that equality test.
28194 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28196 * config/s390/s390-builtin-types.def: Remove flags argument.
28197 * config/s390/s390.c (s390_init_builtins): Likewise.
28199 2017-02-14 Martin Liska <mliska@suse.cz>
28201 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
28202 vector. Fix trailing white spaces.
28204 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
28206 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
28209 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28211 PR rtl-optimization/68664
28212 * config/arm/arm.c (arm_sched_can_speculate_insn):
28213 New function. Declare prototype.
28214 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
28216 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28218 PR rtl-optimization/68664
28219 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
28221 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
28223 2017-02-14 Amit Pawar <amit.pawar@amd.com>
28225 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
28226 max skip bytes for function, loop and jump.
28228 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28230 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
28231 ABS_EXPR for gimple dump.
28233 2017-02-14 Jakub Jelinek <jakub@redhat.com>
28236 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
28238 PR tree-optimization/79408
28239 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
28240 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
28241 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
28242 also if rhs1 is INTEGER_CST.
28244 2017-02-14 Richard Biener <rguenther@suse.de>
28246 PR middle-end/79432
28247 * tree-into-ssa.c (insert_phi_nodes): When the function can
28248 have abnormal edges rewrite SSA names with broken use-def
28249 dominance out of SSA and register them for PHI insertion.
28251 2017-02-13 Martin Sebor <msebor@redhat.com>
28253 PR middle-end/79496
28254 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
28255 clearing info.nowrite flag when snprintf size argument is a range.
28257 2017-02-13 Jakub Jelinek <jakub@redhat.com>
28259 * cprop.c (cprop_jump): Add missing space in string literal.
28260 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
28261 (get_constraint_for_component_ref): Likewise.
28262 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
28263 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
28264 * lra-constraints.c (process_alt_operands): Likewise.
28265 * ipa-inline.c (inline_small_functions): Likewise.
28266 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
28267 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
28268 * trans-mem.c (diagnose_tm_1_op): Likewise.
28269 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
28270 (grid_parallel_clauses_gridifiable): Likewise.
28272 * config/nvptx/mkoffload.c (process): Add space in between
28275 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
28276 "MOD4_SSE_REGS" and "ALL_REGS".
28278 * spellcheck.c (test_data): Add , in between "foo" and "food".
28280 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28283 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
28284 boundary crossing check and subsequent code generation agree.
28286 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28288 * config/aarch64/aarch64.c (has_memory_op): Delete.
28289 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
28292 2017-02-13 Jakub Jelinek <jakub@redhat.com>
28294 PR rtl-optimization/79388
28295 PR rtl-optimization/79450
28296 * combine.c (distribute_notes): When removing TEM_INSN for which
28297 corresponding dest has last value recorded, invalidate that last
28300 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28302 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
28303 of explicit '@'. Add missing assembly comment marker on branch costs
28306 2017-02-13 Nathan Sidwell <nathan@acm.org>
28308 * gengtype-lex.l (<in_struct>): Add '/'.
28310 2017-02-13 Martin Liska <mliska@suse.cz>
28313 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
28315 2017-02-13 Richard Biener <rguenther@suse.de>
28317 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
28319 * configure: Re-generate.
28320 * config.in: Likewise.
28321 * graphite-dependences.c: Simplify as if
28322 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
28323 * graphite-isl-ast-to-gimple.c: Likewise.
28324 * graphite-optimize-isl.c: Likewise.
28325 * graphite-poly.c: Likewise.
28326 * graphite-sese-to-poly.c: Likewise.
28327 * graphite.h: Likewise.
28328 * toplev.c: Include isl/version.h and use isl_version () for
28329 printing the ISL version.
28330 * doc/install.texi: Update ISL requirement.
28332 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
28334 * doc/standards.texi (Standards): Update reference to
28337 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
28339 * doc/extend.texi (Named Address Spaces): sourceware.org now
28341 * doc/install.texi (Binaries): Ditto.
28344 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
28346 * doc/cpp.texi: Replace "stringify"/"stringification" with C
28347 standard terminology "stringize"/"stringizing" throughout.
28348 * doc/cppinternals.texi: Likewise.
28350 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
28352 * doc/extend.texi: Fix some spelling mistakes and typos.
28353 * doc/invoke.texi: Likewise.
28355 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
28358 * params.def (inline-min-speedup) Change from 10 to 8.
28360 2017-02-11 Jakub Jelinek <jakub@redhat.com>
28362 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
28365 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
28368 * ipa-inline-analysis.c (get_minimal_bb): New function.
28369 (record_modified): Use it.
28370 (remap_edge_change_prob): Handle also ancestor functions.
28372 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
28374 * doc/contrib.texi (Contributors): Remove broken link into
28375 the Mauve CVS repository.
28377 2017-02-11 Jakub Jelinek <jakub@redhat.com>
28379 PR middle-end/79454
28380 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
28381 result computation whenever lhs doesn't have vector mode, not
28382 just when it has BLKmode.
28384 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
28386 * doc/makefile.texi (profiledbootstrap): Refer to the
28387 installation instructions only in textual form.
28389 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28392 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
28394 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
28396 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
28397 (Specific): Update mingw-w64 reference.
28399 (Specific): Remove broken link to Renesas RX processor.
28401 2017-02-10 Richard Biener <rguenther@suse.de>
28403 * toplev.c (process_options): Do not mention obsolete graphite
28404 options when printing sorry message about missing graphite support.
28405 Mention -floop-nest-optimize.
28407 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
28409 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
28410 (vtst_p16): Likewise.
28411 (vtstq_p8): Likewise.
28412 (vtstq_p16): Likewise.
28414 (vtstq_p64): Likewise.
28415 * config/arm/arm_neon.h (vgetq_lane_p64): New.
28416 (vset_lane_p64): New.
28417 (vsetq_lane_p64): New.
28419 2017-02-10 Jakub Jelinek <jakub@redhat.com>
28421 PR tree-optimization/79411
28422 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
28423 stmt operands are SSA_NAMEs used in abnormal phis.
28424 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
28427 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
28430 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
28433 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
28435 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
28437 2017-02-09 Jakub Jelinek <jakub@redhat.com>
28439 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
28443 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
28444 not arbitrary TREE_CONSTANT.
28447 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
28448 "omp declare target link" attribute unless is_global_var.
28449 * omp-offload.c (find_link_var_op): Likewise.
28451 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
28452 Chung-Lin Tang <cltang@codesourcery.com>
28454 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
28456 (gimplify_adjust_omp_clauses): Don't delete TILE.
28457 (gimplify_omp_for): Deal with TILE.
28458 * internal-fn.c (expand_GOACC_TILE): New function.
28459 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
28461 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
28462 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
28464 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
28465 avoid DIV for outermost collapse var.
28466 (expand_oacc_for): Insert tile element loop as needed. Adjust.
28467 Remove out of date comments, fix whitespace.
28468 * omp-general.c (omp_extract_for_data): Deal with tiling.
28469 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
28470 adjust OLF_DIM_BASE value.
28471 (struct omp_for_data): Add tiling field.
28472 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
28473 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
28474 for auto loops. Remove default auto determining, moved to
28475 oacc_loop_fixed_partitions.
28476 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
28477 stmts, add e_mask field.
28478 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
28479 (oacc_thread_numbers): Use oacc_dim_call.
28480 (oacc_xform_tile): New.
28481 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
28482 (finish_oacc_loop): Adjust for ifns vector.
28483 (oacc_loop_discover_walk): Append loop abstraction sites to list,
28484 add case for GOACC_TILE fns.
28485 (oacc_loop_xform_loop): Delete.
28486 (oacc_loop_process): Iterate over call list directly, and add
28487 handling for GOACC_TILE fns.
28488 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
28490 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
28491 vector partitioning to outer loops. Assign 2 partitions to loops
28492 when available. Add TILE handling.
28493 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
28494 (execite_oacc_device_lower): Process GOACC_TILE fns,
28495 ignore unknown specs.
28496 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
28497 * tree.c (omp_clause_num_ops): Adjust TILE ops.
28498 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
28500 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
28502 * configure.ac (ACX_BUGURL): Update.
28503 * configure: Regenerate.
28505 2017-02-09 Richard Biener <rguenther@suse.de>
28507 PR tree-optimization/69823
28508 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
28509 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
28511 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
28513 * config/arc/arc-c.def: Add __NPS400__ definition.
28514 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
28515 (TARGET_NPS400): Define.
28517 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
28519 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
28521 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
28522 pointer, arch_info.
28523 (arc_cpu_types): Fill the arch_info field with a pointer into the
28524 arc_arch_types table.
28525 (arc_selected_cpu): Declare.
28526 * config/arc/arc.c (arc_selected_cpu): Make global.
28527 (arc_selected_arch): Delete.
28528 (arc_base_cpu): Delete.
28529 (arc_override_options): Remove references to deleted variables,
28530 update access to arch information.
28531 (ARC_OPT): Update access to arch information.
28532 (ARC_OPTX): Likewise.
28533 * config/arc/arc.h (arc_base_cpu): Remove declaration.
28534 (TARGET_ARC600): Update access to arch information.
28535 (TARGET_ARC601): Likewise.
28536 (TARGET_ARC700): Likewise.
28537 (TARGET_EM): Likewise.
28538 (TARGET_HS): Likewise.
28539 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
28542 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
28545 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
28546 condition/operands for integer GE/LE/GEU/LEU operations.
28548 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
28550 PR translation/79397
28551 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
28554 2017-02-08 Martin Jambor <mjambor@suse.cz>
28557 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
28558 whether allocation happened.
28559 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
28560 nothing was allocated.
28562 2017-02-08 Jakub Jelinek <jakub@redhat.com>
28564 PR tree-optimization/79408
28565 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
28566 constant, but SSA_NAME with a known integer range, use the minimum
28567 of that range instead of op1 to determine if modulo can be replaced
28568 with its first operand.
28570 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28572 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
28574 2017-02-08 Richard Biener <rguenther@suse.de>
28576 PR tree-optimization/71824
28577 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
28578 Check all loops contained in the merged region.
28580 2017-02-07 Andrew Pinski <apinski@cavium.com>
28582 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
28584 2017-02-07 Andrew Pinski <apinski@cavium.com>
28586 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
28587 (thunderxt88): Likewise.
28588 (thunderxt81): Disable LSE and change v8.1 to v8.
28589 (thunderxt83): Likewise.
28591 2017-02-07 Jakub Jelinek <jakub@redhat.com>
28592 Richard Biener <rguenther@suse.de>
28594 PR middle-end/79399
28595 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
28596 type from int to size_t.
28597 * ira-costs.c (struct_costs_size): Change type from int to size_t.
28599 2017-02-07 Jakub Jelinek <jakub@redhat.com>
28601 PR rtl-optimization/79386
28602 * cprop.c (bypass_conditional_jumps): Initialize
28603 bypass_last_basic_block already before splitting bbs after
28604 unconditional traps...
28605 (bypass_conditional_jumps): ... rather than here.
28608 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
28609 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
28610 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
28611 fix -masm=intel patterns.
28613 2017-02-07 Richard Biener <rguenther@suse.de>
28615 PR tree-optimization/79256
28616 PR middle-end/79278
28617 * builtins.c (get_object_alignment_2): Use min_align_of_type
28618 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
28619 and ADJUST_FIELD_ALIGN.
28621 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
28623 * doc/tm.texi: Regenerate.
28624 * stor-layout.c (layout_decl): Adjust.
28625 (update_alignment_for_field): Likewise.
28626 (place_field): Likewise.
28627 (min_align_of_type): Likewise.
28628 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
28629 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
28630 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
28631 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
28632 * config/frv/frv.c (frv_adjust_field_align): Likewise.
28633 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
28634 * config/i386/i386.c (x86_field_alignment): Likewise.
28635 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
28636 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
28637 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
28638 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
28639 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
28640 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
28644 2017-01-30 Richard Biener <rguenther@suse.de>
28646 PR tree-optimization/79256
28647 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
28648 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
28651 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
28653 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
28654 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
28655 builtins to SImode and emit a zero-extend, if necessary.
28657 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
28659 * docs/invoke.texi (RISC-V Options): Alphabetize.
28661 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
28663 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
28666 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
28668 * config/riscv/riscv.c: New file.
28669 * common/config/riscv/riscv-common.c: Likewise.
28670 * config.gcc: Likewise.
28671 * config/riscv/constraints.md: Likewise.
28672 * config/riscv/elf.h: Likewise.
28673 * config/riscv/generic.md: Likewise.
28674 * config/riscv/linux.h: Likewise.
28675 * config/riscv/multilib-generator: Likewise.
28676 * config/riscv/peephole.md: Likewise.
28677 * config/riscv/pic.md: Likewise.
28678 * config/riscv/predicates.md: Likewise.
28679 * config/riscv/riscv-builtins.c: Likewise.
28680 * config/riscv/riscv-c.c: Likewise.
28681 * config/riscv/riscv-ftypes.def: Likewise.
28682 * config/riscv/riscv-modes.def: Likewise.
28683 * config/riscv/riscv-opts.h: Likewise.
28684 * config/riscv/riscv-protos.h: Likewise.
28685 * config/riscv/riscv.h: Likewise.
28686 * config/riscv/riscv.md: Likewise.
28687 * config/riscv/riscv.opt: Likewise.
28688 * config/riscv/sync.md: Likewise.
28689 * config/riscv/t-elf-multilib: Likewise.
28690 * config/riscv/t-linux: Likewise.
28691 * config/riscv/t-linux-multilib: Likewise.
28692 * config/riscv/t-riscv: Likewise.
28693 * configure.ac: Likewise.
28694 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
28695 Waterman as RISC-V maintainers.
28696 * doc/install.texi: Add RISC-V entries.
28697 * doc/invoke.texi: Add RISC-V options section.
28698 * doc/md.texi: Add RISC-V constraints section.
28699 * configure: Regenerated.
28701 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
28704 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
28705 false values to be constant vectors with all 0 or all 1 bits set.
28706 (vcondu<mode><mode>): Likewise.
28707 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
28709 (fpmask_comparison_operator): Update comment.
28710 (vecint_comparison_operator): New predicate.
28711 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
28712 vector conditionals when the true and false values are constant
28713 vectors with all 0 bits or all 1 bits set.
28715 2017-02-06 Martin Sebor <msebor@redhat.com>
28717 PR tree-optimization/79376
28718 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
28720 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
28722 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
28723 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
28724 to simplify split condition.
28726 2017-02-06 Jakub Jelinek <jakub@redhat.com>
28728 * omp-expand.c (oxpand_omp_atomic_fetch_op,
28729 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
28732 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
28734 PR rtl-optimization/68664
28735 * target.def (can_speculate_insn): New hook.
28736 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
28737 * doc/tm.texi: Regenerate.
28738 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
28739 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
28740 (rs6000_sched_can_speculate_insn): New function.
28742 2017-02-06 Jakub Jelinek <jakub@redhat.com>
28744 PR tree-optimization/79284
28745 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
28746 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
28747 vectorizable_mask_load_store, vectorizable_operation,
28748 vect_is_simple_cond, get_same_sized_vectype): Use it instead
28749 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
28750 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
28751 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
28752 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
28753 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
28754 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
28755 is_gimple_assign (stmt). Replace another such test with
28756 is_gimple_assign (stmt).
28758 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
28761 * config/avr/avr.c (rtl-iter.h): Include it.
28762 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
28763 (avr_legitimate_combined_insn): ...and implementation.
28765 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
28767 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
28768 * config/s390/s390.c (s390_const_operand_ok)
28769 (s390_canonicalize_comparison, s390_extract_part)
28770 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
28771 (s390_contiguous_bitmask_p, s390_rtx_costs)
28772 (legitimize_pic_address): Likewise.
28773 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
28774 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
28775 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
28776 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
28777 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
28779 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
28781 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
28782 REGNO($0) == REGNO($1).
28784 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28786 * config/s390/linux.h(SIZE_TYPE): Add comment.
28788 2017-02-06 Julian Brown <julian@codesourcery.com>
28789 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28790 Virendra Pathak <virendra.pathak@broadcom.com>
28792 * config/aarch64/aarch64-cores.def: Change the scheduler
28794 * config/aarch64/aarch64.md: Include thunderx2t99.md.
28795 * config/aarch64/thunderx2t99.md: New file.
28797 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
28799 * doc/standards.texi (Go Language): Update link to language
28802 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
28804 * tree-eh.c (lower_resx): Sanitize profile.
28805 (cleanup_empty_eh_move_lp): Likewise.
28807 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
28810 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
28812 * cfgloopmanip.h (loop_version): Update prototype.
28813 * modulo-sched.c (sms_schedule): Update call of loop_version.
28814 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
28815 * tree-parloops.c (gen_parallel_loop): Likewise.
28816 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
28817 * tree-ssa-loop-split.c (split_loop): Likewise.
28818 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
28819 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
28821 2017-02-05 Martin Liska <mliska@suse.cz>
28824 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
28826 (print_operand_address): Initialize a struct to zero.
28828 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
28830 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
28831 garbage collector only in textual form.
28833 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
28835 * doc/extend.texi (x86 specific memory model extensions for
28836 transactional memory): Simplify a phrase.
28838 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
28841 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
28842 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
28843 (atomic_storedi_1): Likewise.
28845 2017-02-04 Jakub Jelinek <jakub@redhat.com>
28847 PR tree-optimization/79338
28848 * tree-parloops.c (gather_scalar_reductions): Don't call
28849 vect_analyze_loop_form for loop->inner before destroying loop's
28852 2017-02-03 Martin Sebor <msebor@redhat.com>
28854 PR tree-optimization/79327
28855 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
28856 when precision has resulted in leading zeros.
28857 (format_integer): Adjust the likely counter to assume an unknown
28858 argument that may be zero is non-zero.
28860 2017-02-03 Jason Merrill <jason@redhat.com>
28863 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
28864 avoid copying non-taken branch.
28866 2017-02-03 Jakub Jelinek <jakub@redhat.com>
28868 PR tree-optimization/79340
28869 * tree-vect-loop.c (vectorizable_reduction): Release
28870 vec_defs elements after safe_splicing them into other vectors.
28873 PR tree-optimization/79327
28874 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
28875 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
28877 (format_integer): Use wide_int_to_tree instead of build_int_cst
28878 + to_?hwi. If argmin is NULL, just set argmin and argmax to
28879 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
28880 of shortest and longest sequence.
28882 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
28884 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
28885 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
28887 2017-02-03 Walter Lee <walt@tilera.com>
28890 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
28891 after initial stackframe link reg save.
28892 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
28894 2017-02-03 Jakub Jelinek <jakub@redhat.com>
28897 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
28898 wu for stxssp alternative.
28900 2017-02-03 Martin Sebor <msebor@redhat.com>
28902 PR tree-optimization/79352
28903 * gimple-fold.c (get_range_strlen): Add argument.
28904 (get_range_strlen): Change return type to bool.
28905 (get_maxval_strlen): Pass in a dummy argument.
28906 * gimple-fold.h (get_range_strlen): Change return type to bool.
28907 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
28908 * tree.h (array_at_struct_end_p): Add argument.
28909 * tree.c (array_at_struct_end_p): Handle it.
28911 2017-02-03 Martin Liska <mliska@suse.cz>
28914 * multiple_target.c (create_dispatcher_calls): Redirect edge
28915 from a caller of a dispatcher.
28916 (expand_target_clones): Make the clones local.
28917 (ipa_target_clone): Do both target clones and resolvers.
28918 (ipa_dispatcher_calls): Remove the pass.
28919 (pass_dispatcher_calls::gate): Likewise.
28920 (make_pass_dispatcher_calls): Likewise.
28921 * passes.def (pass_target_clone): Put as very first IPA early
28924 2017-02-03 Martin Liska <mliska@suse.cz>
28926 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
28927 in case of a function with ifunc attribute.
28929 2017-02-03 Martin Liska <mliska@suse.cz>
28931 * cgraph.c (cgraph_node::dump): Dump function version info.
28932 * symtab.c (symtab_node::dump_base): Add missing new line.
28934 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
28936 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
28937 (ifcombine_ifandif): Use it.
28939 2017-02-03 Martin Liska <mliska@suse.cz>
28941 * doc/invoke.texi: Document default value for
28942 use-after-scope-direct-emission-threshold.
28944 2017-02-03 Martin Liska <mliska@suse.cz>
28946 PR tree-optimization/79339
28947 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
28948 (format_floating): Likewise.
28950 2017-02-03 Martin Liska <mliska@suse.cz>
28953 * ipa-prop.c (ipa_node_params_t::insert): Remove current
28955 (ipa_node_params_t::remove): Likewise.
28956 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
28957 initialization from removed ipa_node_params_t::insert.
28958 (ipa_node_params::~ipa_node_params): Move from removed
28959 ipa_node_params_t::release.
28960 * symbol-summary.h (symbol_summary::m_released): New member.
28961 Do not release a summary twice. Do not allow to call finalizer
28962 for types of a summary that live in GGC memory.
28964 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28966 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
28969 2017-02-02 Martin Sebor <msebor@redhat.com>
28971 PR middle-end/79275
28972 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
28973 (format_string): Tighten up the range of output for non-constant
28974 strings and correct the expected range for wide non-constant strings.
28976 2017-02-02 Martin Sebor <msebor@redhat.com>
28978 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
28980 PR middle-end/32003
28981 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
28983 (-fdump-tree-@var): Add to index and document how to come up
28984 with pass-specific option and dump file names.
28985 (-fdump-passes): Clarify where to look for output.
28987 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
28989 PR middle-end/77445
28990 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
28991 statistics of the analyzed path; allow threading for speed when
28992 any of BBs along the path are optimized for speed.
28994 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
28996 PR middle-end/78468
28997 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
28998 settings of the virtual registers.
29001 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
29003 * explow.c (get_dynamic_stack_size): Take known alignment of stack
29004 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
29007 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29009 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
29010 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
29012 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29014 * config/s390/s390.md: Add missing comments with the expanded
29016 * config/s390/vector.md: Likewise.
29017 * config/s390/vx-builtins.md: Likewise.
29019 2017-02-02 Jakub Jelinek <jakub@redhat.com>
29022 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
29023 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
29024 conditions on a single line.
29026 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29028 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
29029 __S390_VX__ to __VX__.
29031 2017-02-01 Andrew Pinski <apinski@cavium.com>
29033 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
29034 stmt_info to record_stmt_cost.
29035 (vect_get_known_peeling_cost): Pass stmt_info if known to
29037 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
29038 cpu_vector_cost field into
29039 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
29040 field into vec_int_stmt_cost and vec_fp_stmt_cost.
29041 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
29042 splitting of scalar_stmt_cost and vec_stmt_cost.
29043 (thunderx_vector_cost): Likewise.
29044 (cortexa57_vector_cost): LIkewise.
29045 (exynosm1_vector_cost): Likewise.
29046 (xgene1_vector_cost): Likewise.
29047 (thunderx2t99_vector_cost): Improve after the splitting of the two
29049 (aarch64_builtin_vectorization_cost): Update for the splitting of
29050 scalar_stmt_cost and vec_stmt_cost.
29052 2017-02-01 Torvald Riegel <triegel@redhat.com>
29053 Richard Henderson <rth@redhat.com>
29055 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
29056 conditional on existance of a fast atomic load.
29057 * optabs-query.c (can_atomic_load_p): New function.
29058 * optabs-query.h (can_atomic_load_p): Declare it.
29059 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
29060 no fast atomic load is available for the particular size of access.
29061 (expand_atomic_compare_and_swap): Likewise.
29062 (expand_atomic_load): Likewise.
29063 (expand_atomic_store): Likewise.
29064 (expand_atomic_fetch_op): Likewise.
29065 * testsuite/lib/target-supports.exp
29066 (check_effective_target_sync_int_128): Remove x86 because it provides
29067 no fast atomic load.
29068 (check_effective_target_sync_int_128_runtime): Likewise.
29070 2017-02-01 Richard Biener <rguenther@suse.de>
29072 * graphite.c: Include tree-vectorizer.h for find_loop_location.
29073 (graphite_transform_loops): Provide opt-info for optimized nests.
29074 * tree-parloop.c (parallelize_loops): Provide opt-info for
29075 parallelized loops.
29077 2017-02-01 Richard Biener <rguenther@suse.de>
29079 PR middle-end/79315
29080 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
29081 was not set before.
29083 2017-02-01 Richard Biener <rguenther@suse.de>
29085 PR tree-optimization/71824
29086 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
29087 Verify the loops are valid in the merged SESE region.
29088 (scop_detection::can_represent_loop_1): Check analyzing the
29089 evolution of the number of iterations in the region succeeds.
29091 2017-01-31 Ian Lance Taylor <iant@golang.org>
29093 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
29094 REG_ARGS_SIZE note to 32-bit push insns and call insn.
29096 2017-01-31 David Malcolm <dmalcolm@redhat.com>
29098 PR preprocessor/79210
29099 * input.c (get_substring_ranges_for_loc): Replace line_width
29100 assertion with error-handling.
29102 2017-01-31 Richard Biener <rguenther@suse.de>
29104 PR tree-optimization/77318
29105 * graphite-sese-to-poly.c (extract_affine): Fix assert.
29106 (create_pw_aff_from_tree): Take loop parameter.
29107 (add_condition_to_pbb): Pass loop of the condition to
29108 create_pw_aff_from_tree.
29110 2017-01-31 Jakub Jelinek <jakub@redhat.com>
29112 * config/s390/s390.c (s390_asan_shadow_offset): New function.
29113 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
29115 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
29119 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
29121 (convert_int_to_float128): Likewise.
29122 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
29123 (convert_int_to_float128): Likewise.
29124 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
29125 (UNSPEC_IEEE128_CONVERT): Likewise.
29126 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
29127 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
29128 Use local variables for IBM extended format.
29129 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
29130 (fix_trunc<mode>si2_fprs): Likewise.
29131 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
29132 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
29133 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
29134 to know that we can now have integers of all sizes in vector
29136 (fix<uns>_<mode>di2_hw): Likewise.
29137 (float<uns>_<mode>si2_hw): Likewise.
29138 (fix_<mode>si2_hw): Likewise.
29139 (fixuns_<mode>si2_hw): Likewise.
29140 (float<uns>_<mode>di2_hw): Likewise.
29141 (float_<mode>di2_hw): Likewise.
29142 (float_<mode>si2_hw): Likewise.
29143 (floatuns_<mode>di2_hw): Likewise.
29144 (floatuns_<mode>si2_hw): Likewise.
29145 (xscvqp<su>wz_<mode>): Delete, no longer used.
29146 (xscvqp<su>dz_<mode>): Likewise.
29147 (xscv<su>dqp_<mode>): Likewise.
29148 (ieee128_mfvsrd_64bit): Likewise.
29149 (ieee128_mfvsrd_32bit): Likewise.
29150 (ieee128_mfvsrwz): Likewise.
29151 (ieee128_mtvsrw): Likewise.
29152 (ieee128_mtvsrd_64bit): Likewise.
29153 (ieee128_mtvsrd_32bit): Likewise.
29155 2017-01-31 Martin Liska <mliska@suse.cz>
29158 * ipa-prop.c (ipa_free_all_node_params): Call release method
29159 instead of ~sumbol_summary to not to trigger double times
29162 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
29164 PR tree-optimization/71691
29165 * bitmap.h (class auto_bitmap): New.
29166 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
29167 is_maybe_undefined instead of ssa_undefined_value_p.
29169 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29171 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
29172 __S390_ARCH_LEVEL__ to __ARCH__.
29174 2017-01-31 Jakub Jelinek <jakub@redhat.com>
29176 PR tree-optimization/79267
29177 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
29178 if should_remove_lhs_p is true.
29180 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
29183 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
29184 (add_alignment_attribute): New.
29185 (base_type_die): Add alignment attribute.
29186 (subrange_type_die): Likewise.
29187 (modified_type_die): Likewise.
29188 (gen_array_type_die): Likewise.
29189 (gen_descr_array_type_die: Likewise.
29190 (gen_enumeration_type_die): Likewise.
29191 (gen_subprogram_die): Likewise.
29192 (gen_variable_die): Likewise.
29193 (gen_field_die): Likewise.
29194 (gen_ptr_to_mbr_type_die): Likewise.
29195 (gen_struct_or_union_type_die): Likewise.
29196 (gen_subroutine_type_die): Likewise.
29197 (gen_typedef_die): Likewise.
29198 (base_type_cmp): Compare alignment attribute.
29200 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29203 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
29204 (setb_unsigned) New pattern for setb with CCUNS.
29205 * config/rs6000/rs6000.c (expand_block_compare): Use a different
29206 subfc./subfe sequence to avoid overflow problems. Generate a
29207 shorter sequence with cmpld/setb for power9.
29208 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
29209 for generating subfc. instruction.
29210 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
29211 now uses this instruction.
29213 2017-01-30 Ian Lance Taylor <iant@google.com>
29216 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
29217 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
29219 2017-01-30 Martin Sebor <msebor@redhat.com>
29221 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
29222 Move constant to the right of a relational operator.
29223 (get_mpfr_format_length, format_character, format_string): Ditto.
29224 (should_warn_p, maybe_warn): Same.
29226 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
29228 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
29231 * asan.c (get_translation_unit_decl): Remove function.
29232 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
29234 2017-01-30 Martin Liska <mliska@suse.cz>
29236 PR gcov-profile/79259
29237 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
29238 -fprofile-generate.
29240 2017-01-30 Martin Liska <mliska@suse.cz>
29243 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
29244 Initialize variables with NULL value.
29246 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
29249 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
29251 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
29253 2017-01-30 Richard Biener <rguenther@suse.de>
29255 PR tree-optimization/79276
29256 * tree-vrp.c (process_assert_insertions): Properly adjust common
29257 when removing a duplicate.
29259 * gcc.dg/torture/pr79276.c: New testcase.
29261 2017-01-30 Richard Biener <rguenther@suse.de>
29263 PR tree-optimization/79256
29264 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
29265 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
29267 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
29269 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
29272 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
29273 ("*r<noxa>sbg_<mode>_sll_bitmask")
29274 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
29275 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
29276 Use contiguous_bitmask_nowrap_operand.
29278 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29281 * config/rs6000/altivec.h (vec_xl): Revise #define.
29282 (vec_xst): Likewise.
29284 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
29286 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
29288 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
29290 PR rtl-optimization/79194
29291 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
29292 traps before call to bypass_conditional_jumps.
29294 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
29296 PR tree-optimization/71374
29297 * lra-constraints.c (check_conflict_input_operands): New.
29298 (match_reload): Use it.
29300 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
29303 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
29304 account to calculate conflict_set.
29306 2017-01-27 Bin Cheng <bin.cheng@arm.com>
29308 PR rtl-optimization/78559
29309 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
29310 other_insn in combine.
29312 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
29314 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
29315 uint16_type_node for BT_UINT16.
29317 2017-01-27 David Malcolm <dmalcolm@redhat.com>
29319 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
29320 "RTL Tests" to menu.
29321 (GIMPLE Tests): New node.
29322 (RTL Tests): New node.
29324 2017-01-27 Richard Biener <rguenther@suse.de>
29326 PR tree-optimization/79245
29327 * tree-loop-distribution.c (distribute_loop): Apply cost
29328 modeling also to detected patterns.
29330 2017-01-27 Richard Biener <rguenther@suse.de>
29332 PR tree-optimization/71433
29333 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
29334 (compare_assert_loc): New function.
29335 (process_assert_insertions): Sort and optimize assert locations
29336 to remove duplicates and push down identical assertions on
29337 edges to their destination block.
29339 2017-01-27 Richard Biener <rguenther@suse.de>
29341 PR tree-optimization/79244
29342 * tree-vrp.c (remove_range_assertions): Forcefully propagate
29343 out SSA names even if abnormal.
29345 2017-01-27 Jakub Jelinek <jakub@redhat.com>
29347 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
29348 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
29349 instead of MPFR_RNDN.
29351 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
29354 * arm.c (arm_option_override): Don't call build_target_option_node
29355 until after doing all option overrides.
29356 (arm_valid_target_attribute_tree): Likewise.
29358 2017-01-27 Martin Liska <mliska@suse.cz>
29360 * doc/invoke.texi (-fprofile-arcs): Document profiling support
29361 for {cd}tors and C++ {cd}tors.
29363 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
29365 * config/s390/s390.md ("*setmem_long_and")
29366 ("*setmem_long_and_31z"): Use zero_extend instead of and.
29368 2017-01-26 Martin Sebor <msebor@redhat.com>
29370 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
29373 2017-01-26 Martin Sebor <msebor@redhat.com>
29375 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
29376 HAVE_DFmode before using XFmode or DFmode.
29377 (parse_directive): Avoid using the z length modifier to avoid
29378 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
29380 PR middle-end/78703
29381 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
29382 to accept adjustment as an array.
29383 (get_int_range): New function.
29384 (struct directive): Make width and prec arrays.
29385 (directive::set_width, directive::set_precision): Call get_int_range.
29386 (format_integer, format_floating): Handle width and precision ranges.
29387 (format_string, parse_directive): Same.
29389 2017-01-26 Jakub Jelinek <jakub@redhat.com>
29392 * dwarf2out.c (generate_skeleton_bottom_up): For children with
29393 comdat_type_p set, just clone them, but keep the children in the
29397 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
29398 which have direct callers with -fvar-tracking-assignments enabled
29400 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
29401 inside of type units.
29403 2017-01-26 Martin Sebor <msebor@redhat.com>
29405 PR middle-end/78703
29406 * gimple-ssa-sprintf.c (struct result_range): Add likely and
29408 (struct format_result): Replace number_chars, number_chars_min,
29409 and number_chars_max with a single member of struct result_range.
29411 (format_result::operator+=): Adjust.
29412 (struct fmtresult): Remove bounded. Handle likely and unlikely
29414 (fmtresult::adjust_for_width_or_precision): New function.
29415 (fmtresult:type_max_digits): New function.
29416 (bytes_remaining): Handle likely and unlikely counters.
29417 (min_bytes_remaining): Remove.
29418 (format_percent): Simplify.
29419 (format_integer, format_floating): Set likely and unlikely counters.
29420 (get_string_length, format_character, format_string): Same.
29421 (format_plain, should_warn_p): New function.
29422 (maybe_warn): Call should_warn_p. Update diagnostic messages
29423 and handle those for all directives, including plain strings.
29424 (format_directive): Handle likely and unlikely counters.
29425 Remove unnecessary quoting from diagnostics. Add an informational
29427 (add_bytes): Remove.
29428 (pass_sprintf_length::compute_format_length): Simplify.
29429 (try_substitute_return_value): Handle likely and unlikely counters.
29431 2017-01-26 Carl Love <cel@us.ibm.com>
29433 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
29434 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
29436 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
29439 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
29440 endianess for subregs into account.
29441 * lra-constraints.c (lra_constraints): Do risky transformations
29442 always on the first iteration.
29443 * lra-lives.c (check_pseudos_live_through_calls): Add arg
29444 last_call_used_reg_set.
29445 (process_bb_lives): Define and use last_call_used_reg_set.
29446 * lra.c (lra): Always continue after lra_constraints on the first
29449 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
29451 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
29453 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
29455 2017-01-26 Jakub Jelinek <jakub@redhat.com>
29457 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
29458 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
29459 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
29460 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
29461 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
29462 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
29463 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
29464 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
29465 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
29467 2017-01-26 Marek Polacek <polacek@redhat.com>
29470 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
29471 for the third operand.
29473 2017-01-26 Jakub Jelinek <jakub@redhat.com>
29475 PR middle-end/79236
29476 * omp-low.c (struct omp_context): Add simt_stmt field.
29477 (scan_omp_for): Return omp_context *.
29478 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
29479 context to the _simt_ SIMD stmt.
29480 (lower_omp_for): For combined SIMD with sibling _simt_
29481 SIMD, make sure to use the same decls in _looptemp_
29482 clauses as in the sibling.
29484 2017-01-26 David Sherwood <david.sherwood@arm.com>
29486 PR middle-end/79212
29487 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
29490 2017-01-26 Jakub Jelinek <jakub@redhat.com>
29493 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
29494 emit fld b; fld a; if possible.
29496 * brig-builtins.def: Update copyright years.
29497 * config/arm/arm_acle_builtins.def: Update copyright years.
29499 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
29502 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
29503 constraint instead of o for the stxsd instruction.
29505 2017-01-25 Carl Love <cel@us.ibm.com>
29507 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
29508 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
29510 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
29512 * doc/invoke.texi (C++ Dialect Options): Fix typo.
29514 2017-01-25 Richard Biener <rguenther@suse.de>
29516 PR tree-optimization/69264
29517 * target.def (vector_alignment_reachable): Improve documentation.
29518 * doc/tm.texi: Regenerate.
29519 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
29521 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
29522 earlier changes with respect to TYPE_USER_ALIGN.
29523 (vector_alignment_reachable_p): Likewise. Improve dumping.
29525 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29528 * config/arm/arm.md (xordi3): Force constant operand into a register
29531 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29533 * doc/invoke.texi (-fstore-merging): Correct default optimization
29534 levels at which it is enabled.
29535 (-O): Move -fstore-merging from list to...
29538 2017-01-25 Richard Biener <rguenther@suse.de>
29541 * omp-expand.c: Include debug.h.
29542 (expand_omp_taskreg): Make sure to generate early debug before
29543 outlining anything from a function.
29544 (expand_omp_target): Likewise.
29545 (grid_expand_target_grid_body): Likewise.
29547 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
29550 * asan.c (get_translation_unit_decl): New function.
29551 (asan_add_global): Extract modules file name from globals
29552 TRANSLATION_UNIT_DECL name.
29554 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
29557 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
29558 for long calls with APCS frame and VFP.
29560 2017-01-24 David Malcolm <dmalcolm@redhat.com>
29562 * cfg.c (original_copy_tables_initialized_p): New function.
29563 * cfg.h (original_copy_tables_initialized_p): New decl.
29564 * cfgrtl.c (relink_block_chain): Guard the call to
29565 free_original_copy_tables with a call to
29566 original_copy_tables_initialized_p.
29567 * cgraph.h (symtab_node::native_rtl_p): New decl.
29568 * cgraphunit.c (symtab_node::native_rtl_p): New function.
29569 (symtab_node::needed_p): Don't assert for early assembly output
29570 for __RTL functions.
29571 (cgraph_node::finalize_function): Set "force_output" for __RTL
29573 (cgraph_node::analyze): Bail out early for __RTL functions.
29574 (analyze_functions): Update assertion to support __RTL functions.
29575 (cgraph_node::expand): Bail out early for __RTL functions.
29576 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
29578 * function.h (struct function): Update comment for field
29580 * gimple-expr.c: Include "tree-pass.h".
29581 (gimple_has_body_p): Return false for __RTL functions.
29582 * Makefile.in (OBJS): Add run-rtl-passes.o.
29583 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
29585 (gcc::pass_manager::get_clean_slate): New accessor.
29586 * passes.c: Include "insn-addr.h".
29587 (should_skip_pass_p): Add logging. Update logic for running
29588 "expand" to be compatible with both __GIMPLE and __RTL. Guard
29589 property-provider override so it is only done for gimple passes.
29591 (skip_pass): New function.
29592 (execute_one_pass): Call skip_pass when skipping passes.
29593 * read-md.c (md_reader::read_char): Support filtering
29594 the input to a subset of line numbers.
29595 (md_reader::md_reader): Initialize fields
29596 m_first_line and m_last_line.
29597 (md_reader::read_file_fragment): New function.
29598 * read-md.h (md_reader::read_file_fragment): New decl.
29599 (md_reader::m_first_line): New field.
29600 (md_reader::m_last_line): New field.
29601 * read-rtl-function.c (function_reader::create_function): Only
29602 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
29603 curr_properties. Set DECL_INITIAL to a dummy block.
29604 (read_rtl_function_body_from_file_range): New function.
29605 * read-rtl-function.h (read_rtl_function_body_from_file_range):
29607 * run-rtl-passes.c: New file.
29608 * run-rtl-passes.h: New file.
29610 2017-01-24 Jeff Law <law@redhat.com>
29612 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
29615 2017-01-24 Bin Cheng <bin.cheng@arm.com>
29617 PR tree-optimization/79159
29618 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
29619 (record_nonwrapping_iv): Improve boundary using above function if no
29620 value range information.
29622 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
29623 Martin Jambor <mjambor@suse.cz>
29625 * brig-builtins.def: New file.
29626 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
29627 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
29628 (DEF_HSAIL_SAT_BUILTIN): Likewise.
29629 (DEF_HSAIL_INTR_BUILTIN): Likewise.
29630 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
29631 * builtin-types.def (BT_INT8): New.
29632 (BT_INT16): Likewise.
29633 (BT_UINT8): Likewise.
29634 (BT_UINT16): Likewise.
29635 (BT_FN_ULONG): Likewise.
29636 (BT_FN_UINT_INT): Likewise.
29637 (BT_FN_UINT_ULONG): Likewise.
29638 (BT_FN_UINT_LONG): Likewise.
29639 (BT_FN_UINT_PTR): Likewise.
29640 (BT_FN_ULONG_PTR): Likewise.
29641 (BT_FN_INT8_FLOAT): Likewise.
29642 (BT_FN_INT16_FLOAT): Likewise.
29643 (BT_FN_UINT32_FLOAT): Likewise.
29644 (BT_FN_UINT16_FLOAT): Likewise.
29645 (BT_FN_UINT8_FLOAT): Likewise.
29646 (BT_FN_UINT64_FLOAT): Likewise.
29647 (BT_FN_UINT16_UINT32): Likewise.
29648 (BT_FN_UINT32_UINT16): Likewise.
29649 (BT_FN_UINT16_UINT16_UINT16): Likewise.
29650 (BT_FN_INT_PTR_INT): Likewise.
29651 (BT_FN_UINT_PTR_UINT): Likewise.
29652 (BT_FN_LONG_PTR_LONG): Likewise.
29653 (BT_FN_ULONG_PTR_ULONG): Likewise.
29654 (BT_FN_VOID_UINT64_UINT64): Likewise.
29655 (BT_FN_UINT8_UINT8_UINT8): Likewise.
29656 (BT_FN_INT8_INT8_INT8): Likewise.
29657 (BT_FN_INT16_INT16_INT16): Likewise.
29658 (BT_FN_INT_INT_INT): Likewise.
29659 (BT_FN_UINT_FLOAT_UINT): Likewise.
29660 (BT_FN_FLOAT_UINT_UINT): Likewise.
29661 (BT_FN_ULONG_UINT_UINT): Likewise.
29662 (BT_FN_ULONG_UINT_PTR): Likewise.
29663 (BT_FN_ULONG_ULONG_ULONG): Likewise.
29664 (BT_FN_UINT_UINT_UINT): Likewise.
29665 (BT_FN_VOID_UINT_PTR): Likewise.
29666 (BT_FN_UINT_UINT_PTR: Likewise.
29667 (BT_FN_UINT32_UINT64_PTR): Likewise.
29668 (BT_FN_INT_INT_UINT_UINT): Likewise.
29669 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
29670 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
29671 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
29672 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
29673 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
29674 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
29675 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
29676 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
29677 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
29678 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
29679 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
29680 * doc/frontends.texi: List BRIG FE.
29681 * doc/install.texi (Testing): Add BRIG tesring requirements.
29682 * doc/invoke.texi (Overall Options): Mention BRIG.
29683 * doc/standards.texi (Standards): Doucment BRIG HSA version.
29685 2017-01-24 Richard Biener <rguenther@suse.de>
29687 PR translation/79208
29688 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
29690 2017-01-24 Martin Jambor <mjambor@suse.cz>
29693 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
29694 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
29695 and known_contexts.
29697 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
29699 PR middle-end/79123
29700 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
29701 casts from signed to unsigned really don't have a range.
29703 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
29705 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
29706 GMP_RNDx for compatiblity.
29708 2017-01-24 Martin Liska <mliska@suse.cz>
29711 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
29712 that would prevent us to call alloca with -1 as argument.
29714 2017-01-24 Jakub Jelinek <jakub@redhat.com>
29716 * dwarf2out.c (output_compilation_unit_header, output_file_names):
29717 Avoid -Wformat-security warning.
29719 2017-01-23 Andrew Pinski <apinski@cavium.com>
29721 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
29724 2017-01-23 Martin Sebor <msebor@redhat.com>
29726 PR middle-end/78703
29727 * gimple-ssa-sprintf.c (warn_level): New global.
29728 (format_integer): Use it here and throughout the rest of the file.
29729 Use the same switch to compute sign as base.
29730 (maybe_warn): New function.
29731 (format_directive): Factor out warnings into maybe_warn.
29732 Add debugging output. Use warn_level.
29733 (add_bytes): Use warn_level.
29734 (pass_sprintf_length::compute_format_length): Add debugging output.
29735 (try_substitute_return_value): Same.
29736 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
29738 PR middle-end/78703
29739 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
29740 (struct fmtresult, format_integer, format_floating): Adjust.
29741 (fmtresult::fmtresult): Set max correctly in two argument ctor.
29742 (get_string_length, format_string,format_directive): Same.
29743 (pass_sprintf_length::compute_format_length): Same.
29744 (try_substitute_return_value): Simplify slightly.
29746 PR middle-end/78703
29747 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
29748 (fmtresult::operator+=): Outlined.
29749 (struct fmtresult): Add ctors.
29750 (struct conversion_spec): Rename...
29751 (struct directive): ...to this. Add and remove data members.
29752 (directive::set_width, directive::set_precision): New functions.
29753 (format_percent): Use fmtresult ctor.
29754 (get_width_and_precision): Remove.
29755 (format_integer): Make naming changes. Avoid computing width and
29757 (format_floating): Same. Adjust indentation.
29758 (format_character, format_none): New functions.
29759 (format_string): Moved character handling to format_character.
29760 (format_directive): Remove arguments, change return type.
29761 (parse_directive): New function.
29762 (pass_sprintf_length::compute_format_length): Move directive
29763 parsing to parse_directive.
29765 2017-01-23 Jakub Jelinek <jakub@redhat.com>
29767 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
29768 (assign_assembler_name_if_needed): ... this.
29769 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
29770 (assign_assembler_name_if_needed): ... this.
29771 (free_lang_data_in_cgraph): Adjust callers.
29772 * cgraphunit.c (cgraph_node::analyze): Likewise.
29773 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
29776 2017-01-23 Richard Biener <rguenther@suse.de>
29778 PR tree-optimization/79088
29779 PR tree-optimization/79188
29780 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
29781 resetting loop bounds after last path deletion. Reset loop
29782 bounds of the target loop, make code match the comments.
29783 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
29784 Make sure loops need no fixups.
29786 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
29788 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
29789 exponent support with double type for first argument.
29790 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
29791 type returned by __builtin_vec_extract_sig,
29792 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
29793 functions from "vector int" to "vector unsigned int" or from
29794 "vector long long int" to "vector unsigned long long int".
29795 Changed type returned by __builtin_vec_extract_exp,
29796 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
29797 functions from "vector int" to "vector unsigned int" or from
29798 "vector long long int" to "vector unsigned long long int".
29799 Changed return type of __builtin_vec_test_data_class,
29800 __builtin_vec_test_data_class_sp, and
29801 __builtin_vec_test_data_class_dp from "vector int" to
29802 "vector bool int" or from "vector long long int" to "vector bool
29803 long long int" and changed second argument type from "unsigned
29804 int" to "int". Added new overloaded function forms "vector float
29805 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
29806 "vector float __builtin_vec_insert_exp_sp (vector float, vector
29807 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
29808 double, vector unsigned long long int)" and "vector double
29809 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
29810 long int)". Changed return type of
29811 __builtin_scalar_test_data_class and
29812 __builtin_scalar_test_data_class_sp and
29813 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
29814 int" and changed second argument from "unsigned int" to "int".
29815 Changed type returned by __builtin_scalar_test_neg,
29816 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
29817 from "int" to "bool int". Added new overloaded function form
29818 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
29819 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
29820 exponent double-precision with floating point first argument.
29821 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
29822 documentation of scalar_test_data_class, scalar_test_neg,
29823 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
29824 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
29825 vec_test_data_class built-in functions to reflect refinements in
29826 their type signatures.
29828 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
29830 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
29832 (aarch64_elf_asm_destructor): Likewise.
29834 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
29836 PR rtl-optimization/78634
29837 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
29838 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
29839 * ifcvt.c (noce_try_cmove): Add missing cost check.
29841 PR rtl-optimization/71724
29842 * combine.c (if_then_else_cond): Look for situations where it is
29843 beneficial to undo the work of one of the recursive calls.
29845 2017-01-23 Bin Cheng <bin.cheng@arm.com>
29847 PR tree-optimization/70754
29848 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
29849 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
29850 combined stmt before it if not NULL.
29851 (combine_chains): Process refs reversely and compute dominance point
29854 2017-01-23 Martin Liska <mliska@suse.cz>
29856 PR tree-optimization/79196
29857 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
29858 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
29860 (strlen_optimize_stmt): Call the renamed function.
29862 2017-01-23 Michael Matz <matz@suse.de>
29864 PR tree-optimization/78384
29865 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
29867 2017-01-23 Richard Biener <rguenther@suse.de>
29869 PR tree-optimization/79186
29870 * tree-vrp.c (register_new_assert_for): Make sure we've seen
29871 both incoming edges before moving an assert.
29873 2017-01-23 Martin Jambor <mjambor@suse.cz>
29875 * ipa-prop.c (load_from_param_1): Removed.
29876 (load_from_unmodified_param): Bits from load_from_param_1 put back
29878 (load_from_param): Removed.
29879 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
29880 with stmt. Reverted back to use of load_from_unmodified_param.
29882 2017-01-23 Martin Jambor <mjambor@suse.cz>
29885 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
29886 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
29887 field a pointer to garbage collected vector, mark lattices and
29888 ipcp_orig_node with GTY((skip)).
29889 (ipa_get_param_count): Adjust to descriptors being a pointer.
29890 (ipa_get_param): Likewise.
29891 (ipa_get_type): Likewise.
29892 (ipa_get_param_move_cost): Likewise.
29893 (ipa_set_param_used): Likewise.
29894 (ipa_get_controlled_uses): Likewise.
29895 (ipa_set_controlled_uses): Likewise.
29896 (ipa_is_param_used): Likewise.
29897 (ipa_node_params_t): Move into garbage collector. New methods insert
29899 (ipa_node_params_sum): Annotate wth GTY(()).
29900 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
29902 (ipa_load_from_parm_agg): Adjust declaration.
29903 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
29904 * ipa-profile.c (ipa_profile): Likewise.
29905 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
29906 (ipa_populate_param_decls): Make descriptors parameter garbage
29908 (ipa_dump_param): Adjust to descriptors being a pointer.
29909 (ipa_alloc_node_params): Likewise.
29910 (ipa_initialize_node_params): Likewise.
29911 (load_from_param_1): Make descriptors parameter garbage collected.
29912 (load_from_unmodified_param): Likewise.
29913 (load_from_param): Likewise.
29914 (ipa_load_from_parm_agg): Likewise.
29915 (ipa_node_params::~ipa_node_params): Removed.
29916 (ipa_free_all_node_params): Remove call to delete operator.
29917 (ipa_node_params_t::insert): New.
29918 (ipa_node_params_t::remove): Likewise.
29919 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
29920 copy known_csts and known_contexts vectors.
29921 (ipa_read_node_info): Adjust to descriptors being a pointer.
29922 (ipcp_modif_dom_walker): Make m_descriptors field garbage
29924 (ipcp_transform_function): Make descriptors variable garbage
29927 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
29929 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
29930 * config/i386/avx512dqintrin.h: Ditto.
29931 * config/i386/avx512fintrin.h: Ditto.
29932 * config/i386/i386.c: Handle new builtins.
29933 * config/i386/i386-builtin.def: Add new builtins.
29934 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
29935 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
29937 2017-01-23 Jakub Jelinek <jakub@redhat.com>
29938 Martin Liska <mliska@suse.cz>
29940 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
29941 * asan.c (asan_expand_poison_ifn): Support stores and use
29942 appropriate ASAN report function.
29943 * internal-fn.c (expand_ASAN_POISON_USE): New function.
29944 * internal-fn.def (ASAN_POISON_USE): Declare.
29945 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
29946 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
29947 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
29948 ASAN_POISON calls w/o LHS.
29949 * tree-ssa.c (execute_update_addresses_taken): Create clobber
29950 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
29951 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
29952 * gimplify.c (asan_poison_variables): Add attribute
29953 use_after_scope_memory to variables that really needs to live
29955 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
29956 having the attribute.
29958 2017-01-23 Martin Liska <mliska@suse.cz>
29960 * asan.c (create_asan_shadow_var): New function.
29961 (asan_expand_poison_ifn): Likewise.
29962 * asan.h (asan_expand_poison_ifn): New declaration.
29963 * internal-fn.c (expand_ASAN_POISON): Likewise.
29964 * internal-fn.def (ASAN_POISON): New builtin.
29965 * sanopt.c (pass_sanopt::execute): Expand
29966 asan_expand_poison_ifn.
29967 * tree-inline.c (copy_decl_for_dup_finish): Make function
29969 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
29970 * tree-ssa.c (is_asan_mark_p): New function.
29971 (execute_update_addresses_taken): Rewrite local variables
29972 (identified just by use-after-scope as addressable) into SSA.
29974 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
29976 * doc/install.texi (Specific): opensource.apple.com uses https
29977 now. Remove trailing slash.
29979 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
29981 * README.Portability: Remove note on an Irix compatibility issue.
29983 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
29985 * gcov.c (INCLUDE_ALGORITHM): Define.
29986 (INCLUDE_VECTOR): Define.
29987 No longer include <vector> and <algorithm> directly.
29989 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
29991 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
29993 * doc/invoke.texi (Code Gen Options): Ditto.
29995 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
29998 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
30000 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
30002 rtl-optimization/79125
30003 * cprop.c (local_cprop_pass): Handle cases where we make an
30004 unconditional trap.
30006 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
30010 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
30011 read from, for big endian.
30013 2017-01-20 Jiong Wang <jiong.wang@arm.com>
30015 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
30016 register pauth builtins for LP64 only.
30018 2017-01-20 Marek Polacek <polacek@redhat.com>
30021 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
30024 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
30026 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
30028 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
30029 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
30030 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
30032 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30035 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
30036 in big-endian mode when they are not a single duplicated value.
30038 2017-01-20 Richard Biener <rguenther@suse.de>
30040 * BASE-VER: Bump to 7.0.1.
30042 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
30044 * omp-low.c (omplow_simd_context): New struct. Use it...
30045 (lower_rec_simd_input_clauses): ...here and...
30046 (lower_rec_input_clauses): ...here to hold common data. Adjust all
30047 references to idx, lane, max_vf, is_simt.
30049 2017-01-20 Graham Markall <graham.markall@embecosm.com>
30051 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
30054 2017-01-20 Martin Jambor <mjambor@suse.cz>
30056 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
30057 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
30059 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
30060 (GTFILES): Rename hsa.c to hsa-common.c.
30061 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
30062 * hsa-dump.c: Likewise.
30063 * hsa-gen.c: Likewise.
30064 * hsa-regalloc.c: Likewise.
30065 * ipa-hsa.c: Likewise.
30066 * omp-expand.c: Likewise.
30067 * omp-low.c: Likewise.
30068 * toplev.c: Likewise.
30070 2017-01-20 Marek Polacek <polacek@redhat.com>
30073 * doc/invoke.texi: Document -Wduplicated-branches.
30074 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
30075 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
30076 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
30077 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
30078 return 0 only when not OEP_LEXICOGRAPHIC.
30079 (fold_build_cleanup_point_expr): Use the expression
30080 location when building CLEANUP_POINT_EXPR.
30081 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
30082 * tree.c (add_expr): Handle error_mark_node.
30084 2017-01-20 Martin Liska <mliska@suse.cz>
30087 * tree-profile.c (init_ic_make_global_vars): Do not call
30089 (gimple_init_gcov_profiler): Likewise.
30091 2017-01-20 Martin Liska <mliska@suse.cz>
30094 * cgraph.h (maybe_create_reference): Remove argument and
30096 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
30098 * ipa-cp.c (create_specialized_node): Likewise.
30099 * symtab.c (symtab_node::maybe_create_reference): Handle
30100 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
30102 2017-01-20 Martin Liska <mliska@suse.cz>
30104 * read-rtl-function.c (function_reader::create_function): Use
30105 build_decl instread of build_decl_stat.
30107 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
30109 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
30110 * config/i386/avx512dqintrin.h: Ditto.
30111 * config/i386/avx512fintrin.h: Ditto.
30112 * config/i386/i386-builtin-types.def: Add new types.
30113 * config/i386/i386.c: Handle new types.
30114 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
30115 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
30116 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
30117 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
30118 (__builtin_ia32_kshiftridi): New.
30119 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
30121 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
30125 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
30126 define to rs6000_init_stack_protect_guard.
30127 (rs6000_init_stack_protect_guard): New function.
30129 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
30130 Yunqiang Su <yunqiang.su@imgtec.com>
30132 * config.gcc (supported_defaults): Add madd4.
30133 (with_madd4): Add validation.
30134 (all_defaults): Add madd4.
30135 * config/mips/mips.opt (mmadd4): New option.
30136 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
30138 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
30140 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
30141 (ISA_HAS_FUSED_MADD4): Likewise.
30142 * doc/invoke.texi (-mmadd4): Document the new option.
30143 * doc/install.texi (--with-madd4): Document the new option.
30145 2017-01-19 Jiong Wang <jiong.wang@arm.com>
30147 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
30148 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
30149 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
30150 (aarch64_init_pauth_hint_builtins): New.
30151 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
30152 (aarch64_expand_builtin): Expand new builtins.
30154 2017-01-19 Jiong Wang <jiong.wang@arm.com>
30156 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
30157 * combine-stack-adj.c (no_unhandled_cfa): Handle
30158 REG_CFA_TOGGLE_RA_MANGLE.
30159 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
30160 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
30161 info for return address signing.
30162 (aarch64_expand_epilogue): Likewise.
30164 2017-01-19 Jiong Wang <jiong.wang@arm.com>
30166 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
30167 * config/aarch64/aarch64-protos.h
30168 (aarch64_return_address_signing_enabled): New declaration.
30169 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
30171 (aarch64_expand_prologue): Sign return address before it's pushed onto
30173 (aarch64_expand_epilogue): Authenticate return address fetched from
30175 (aarch64_override_options): Sanity check for ILP32 and ISA level.
30176 (aarch64_attributes): New function attributes for "sign-return-address".
30177 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
30178 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
30179 ("*do_return"): Generate combined instructions according to key index.
30180 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
30181 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
30183 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
30184 * config/aarch64/aarch64.opt (msign-return-address=): New.
30185 * doc/extend.texi (AArch64 Function Attributes): Documents
30186 "sign-return-address=".
30187 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
30189 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
30191 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
30192 overall option summary.
30194 2017-01-19 Jiong Wang <jiong.wang@arm.com>
30196 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
30197 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
30198 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
30199 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
30201 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
30203 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
30204 -mpower9-minmax by default for -mcpu=power9.
30205 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
30206 128-bit floating point.
30208 2017-01-20 Alan Modra <amodra@gmail.com>
30210 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
30211 optimizing for size.
30213 2017-01-20 Alan Modra <amodra@gmail.com>
30216 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
30217 for strcmp and strncmp from corresponding builtin decl.
30219 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
30221 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
30222 instead of i386/rtems-64.h.
30223 * config/i386/rtems-64.h: Remove.
30225 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
30229 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
30231 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
30233 2017-01-19 Tamar Christina <tamar.christina@arm.com>
30235 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
30236 Change int to HOST_WIDE_INT.
30237 * config/aarch64/aarch64-protos.h
30238 (aarch64_simd_gen_const_vector_dup): Likewise.
30239 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
30241 2017-01-19 David Malcolm <dmalcolm@redhat.com>
30243 * langhooks-def.h (lhd_type_for_size): New decl.
30244 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
30245 * langhooks.c (lhd_type_for_size): New function, taken from
30248 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
30250 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
30251 define_bypass for CR latency.
30252 (power9-cracked-alu): Update bypass latency and remove power9-branch.
30253 (power9-alu2): Add define_bypass for CR latency.
30255 (power9-mul): Update insn latency.
30256 (power9-mul-compare): Update insn latency, bypass latency and remove
30259 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30261 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
30263 * config/aarch64/aarch64.md
30264 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
30265 aarch64_nopcrelative_literal_loads.
30266 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
30268 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
30270 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
30271 TARGET_LOONGSON_3A.
30272 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
30274 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
30277 * config.gcc (supported_defaults): Add lxc1-sxc1.
30278 (with_lxc1_sxc1): Add validation.
30279 (all_defaults): Add lxc1-sxc1.
30280 * config/mips/mips.opt (mlxc1-sxc1): New option.
30281 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
30283 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
30284 __mips_no_lxc1_sxc1.
30285 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
30286 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
30287 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
30289 2017-01-19 Richard Biener <rguenther@suse.de>
30291 PR tree-optimization/72488
30292 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
30293 sure to restore SSA info.
30294 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
30296 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
30298 PR rtl-optimization/79121
30299 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
30300 of the inner type when shifting an extended value.
30302 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
30305 * symtab.c (symtab_node::equal_address_to): Fix comparing of
30306 interposable aliases.
30308 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
30311 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
30312 Use the evmergelohi instruction.
30313 (mov_si<mode>_e500_subreg4_2_le): Likewise.
30314 (mov_sitf_e500_subreg8_2_be): Likewise.
30315 (mov_sitf_e500_subreg12_2_le): Likewise.
30316 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
30317 (mov_si<mode>_e500_subreg4_2_be): Likewise.
30318 (mov_sitf_e500_subreg8_2_le): Likewise.
30319 (mov_sitf_e500_subreg12_2_be): Likewise.
30321 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30323 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
30324 attribute from vecsimple to vecperm.
30325 (altivec_vbpermq2): Likewise.
30327 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30330 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
30332 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30333 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
30334 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
30335 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
30336 case where N arg is SIZE_MAX.
30337 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
30338 (cmpstrsi): Add pattern.
30340 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
30342 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
30343 __builtin_vec_revb builtins.
30344 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
30345 built-in functions to support generation of the ISA 3.0 XXBR<x>
30346 vector byte reverse instructions.
30347 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
30348 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
30349 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
30350 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
30351 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
30352 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
30353 (P9V_BUILTIN_VEC_REVB): Likewise.
30354 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
30355 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
30356 (p9_xxbrq_v16qi): Likewise.
30357 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
30358 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
30359 (p9_xxbrh_v8hi): Likewise.
30360 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
30361 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
30362 vec_revb built-in functions.
30364 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
30366 PR rtl-optimization/78952
30367 * config/i386/i386.md (any_extract): New code iterator.
30368 (*insvqi_2): Use any_extract for source operand.
30369 (*insvqi_3): Use any_shiftrt for source operand.
30371 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
30373 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
30375 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
30377 2017-01-18 Matthias Klose <doko@ubuntu.com>
30379 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
30381 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30383 * config/rs6000/altivec.h (vec_bperm): Change #define.
30384 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
30385 (altivec_vbpermq2): New define_insn.
30386 (altivec_vbpermd): Likewise.
30387 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
30388 function interface.
30389 (VBPERMD): Likewise.
30390 (VBPERM): New polymorphic function interface.
30391 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
30392 Add entries for P9V_BUILTIN_VEC_VBPERM.
30393 * doc/extend.texi: Add interfaces for vec_bperm.
30395 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30397 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
30398 first letter of error messages.
30399 (s390_resolve_overloaded_builtin): Likewise.
30400 * config/s390/s390.c (s390_expand_builtin): Likewise.
30401 (s390_invalid_arg_for_unprototyped_fn): Likewise.
30402 (s390_valid_target_attribute_inner_p): Likewise.
30403 * config/s390/s390.md ("tabort"): Likewise.
30405 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
30407 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
30408 (ISA_AVOID_DIV_HILO): New macro.
30409 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
30410 (ISA_HAS_DDIV): Likewise.
30412 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
30414 * doc/invoke.texi (fabi-version): Correct number of occurrences.
30416 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
30418 * doc/invoke.texi (fabi-version): Spelling fix.
30420 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
30423 * doc/invoke.texi (fabi-version): Mention mangling fix for
30426 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
30429 * doc/invoke.texi (fabi-version): Document discriminator mangling.
30431 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
30434 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
30435 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
30437 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
30439 (stack_protect_test): Ditto.
30440 * config/rs6000/rs6000.opt (mstack-protector-guard=,
30441 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
30443 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
30444 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
30445 -mstack-protector-guard-offset=.
30446 (RS/6000 and PowerPC Options): Ditto.
30448 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
30450 * config/i386/i386.h (MASK_CLASS_P): New define.
30451 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
30452 there are no registers from different register sets also when
30453 mask registers are used. Update function comment.
30454 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
30455 to (*k/*r) and (*k/*km) alternatives.
30457 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
30459 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
30460 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
30461 (EH_RETURN_HANDLER_RTX): New define.
30462 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
30463 Force frame pointer in EH return functions.
30464 (aarch64_expand_epilogue): Add barrier for eh_return.
30465 (aarch64_final_eh_return_addr): Remove.
30466 (aarch64_eh_return_handler_rtx): New function.
30467 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
30469 (aarch64_eh_return_handler_rtx): New prototype.
30471 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30473 * config/rs6000/altivec.h (vec_rlmi): New #define.
30474 (vec_vrlnm): Likewise.
30475 (vec_rlnm): Likewise.
30476 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
30477 (UNSPEC_VRLNM): Likewise.
30478 (VIlong): New mode iterator.
30479 (altivec_vrl<VI_char>mi): New define_insn.
30480 (altivec_vrl<VI_char>nm): Likewise.
30481 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
30483 (VRLDNM): Likewise.
30484 (RLNM): New polymorphic function entry.
30485 (VRLWMI): New monomorphic function entry.
30486 (VRLDMI): Likewise.
30487 (RLMI): New polymorphic function entry.
30488 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
30489 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
30490 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
30493 2017-01-17 Jakub Jelinek <jakub@redhat.com>
30496 * dwarf2out.c (field_byte_offset): Restore the
30497 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
30498 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
30499 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
30502 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
30505 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
30507 2017-01-17 Jakub Jelinek <jakub@redhat.com>
30510 * dwarf2out.c (add_data_member_location_attribute): For constant
30511 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
30512 instead of DW_AT_data_member_location, DW_AT_bit_offset and
30513 DW_AT_byte_size attributes.
30515 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
30517 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
30518 after forcing to constant memory when the code model is medium.
30520 2017-01-17 Julia Koval <julia.koval@intel.com>
30523 * config/i386/avx512fintrin.h
30524 (_mm512_i32gather_ps): Change __addr type to void const*.
30525 (_mm512_mask_i32gather_ps): Ditto.
30526 (_mm512_i32gather_pd): Ditto.
30527 (_mm512_mask_i32gather_pd): Ditto.
30528 (_mm512_i64gather_ps): Ditto.
30529 (_mm512_mask_i64gather_ps): Ditto.
30530 (_mm512_i64gather_pd): Ditto.
30531 (_mm512_mask_i64gather_pd): Ditto.
30532 (_mm512_i32gather_epi32): Ditto.
30533 (_mm512_mask_i32gather_epi32): Ditto.
30534 (_mm512_i32gather_epi64): Ditto.
30535 (_mm512_mask_i32gather_epi64): Ditto.
30536 (_mm512_i64gather_epi32): Ditto.
30537 (_mm512_mask_i64gather_epi32): Ditto.
30538 (_mm512_i64gather_epi64): Ditto.
30539 (_mm512_mask_i64gather_epi64): Ditto.
30540 (_mm512_i32scatter_ps): Change __addr type to void*.
30541 (_mm512_mask_i32scatter_ps): Ditto.
30542 (_mm512_i32scatter_pd): Ditto.
30543 (_mm512_mask_i32scatter_pd): Ditto.
30544 (_mm512_i64scatter_ps): Ditto.
30545 (_mm512_mask_i64scatter_ps): Ditto.
30546 (_mm512_i64scatter_pd): Ditto.
30547 (_mm512_mask_i64scatter_pd): Ditto.
30548 (_mm512_i32scatter_epi32): Ditto.
30549 (_mm512_mask_i32scatter_epi32): Ditto.
30550 (_mm512_i32scatter_epi64): Ditto.
30551 (_mm512_mask_i32scatter_epi64): Ditto.
30552 (_mm512_i64scatter_epi32): Ditto.
30553 (_mm512_mask_i64scatter_epi32): Ditto.
30554 (_mm512_i64scatter_epi64): Ditto.
30555 (_mm512_mask_i64scatter_epi64): Ditto.
30556 * config/i386/avx512pfintrin.h
30557 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
30558 (_mm512_mask_prefetch_i32gather_ps): Ditto.
30559 (_mm512_mask_prefetch_i64gather_pd): Ditto.
30560 (_mm512_mask_prefetch_i64gather_ps): Ditto.
30561 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
30562 (_mm512_prefetch_i32scatter_ps): Ditto.
30563 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
30564 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
30565 (_mm512_prefetch_i64scatter_pd): Ditto.
30566 (_mm512_prefetch_i64scatter_ps): Ditto.
30567 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
30568 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
30569 * config/i386/avx512vlintrin.h
30570 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
30571 (_mm_mmask_i32gather_ps): Ditto.
30572 (_mm256_mmask_i32gather_pd): Ditto.
30573 (_mm_mmask_i32gather_pd): Ditto.
30574 (_mm256_mmask_i64gather_ps): Ditto.
30575 (_mm_mmask_i64gather_ps): Ditto.
30576 (_mm256_mmask_i64gather_pd): Ditto.
30577 (_mm_mmask_i64gather_pd): Ditto.
30578 (_mm256_mmask_i32gather_epi32): Ditto.
30579 (_mm_mmask_i32gather_epi32): Ditto.
30580 (_mm256_mmask_i32gather_epi64): Ditto.
30581 (_mm_mmask_i32gather_epi64): Ditto.
30582 (_mm256_mmask_i64gather_epi32): Ditto.
30583 (_mm_mmask_i64gather_epi32): Ditto.
30584 (_mm256_mmask_i64gather_epi64): Ditto.
30585 (_mm_mmask_i64gather_epi64): Ditto.
30586 (_mm256_i32scatter_ps): Change __addr type to void*.
30587 (_mm256_mask_i32scatter_ps): Ditto.
30588 (_mm_i32scatter_ps): Ditto.
30589 (_mm_mask_i32scatter_ps): Ditto.
30590 (_mm256_i32scatter_pd): Ditto.
30591 (_mm256_mask_i32scatter_pd): Ditto.
30592 (_mm_i32scatter_pd): Ditto.
30593 (_mm_mask_i32scatter_pd): Ditto.
30594 (_mm256_i64scatter_ps): Ditto.
30595 (_mm256_mask_i64scatter_ps): Ditto.
30596 (_mm_i64scatter_ps): Ditto.
30597 (_mm_mask_i64scatter_ps): Ditto.
30598 (_mm256_i64scatter_pd): Ditto.
30599 (_mm256_mask_i64scatter_pd): Ditto.
30600 (_mm_i64scatter_pd): Ditto.
30601 (_mm_mask_i64scatter_pd): Ditto.
30602 (_mm256_i32scatter_epi32): Ditto.
30603 (_mm256_mask_i32scatter_epi32): Ditto.
30604 (_mm_i32scatter_epi32): Ditto.
30605 (_mm_mask_i32scatter_epi32): Ditto.
30606 (_mm256_i32scatter_epi64): Ditto.
30607 (_mm256_mask_i32scatter_epi64): Ditto.
30608 (_mm_i32scatter_epi64): Ditto.
30609 (_mm_mask_i32scatter_epi64): Ditto.
30610 (_mm256_i64scatter_epi32): Ditto.
30611 (_mm256_mask_i64scatter_epi32): Ditto.
30612 (_mm_i64scatter_epi32): Ditto.
30613 (_mm_mask_i64scatter_epi32): Ditto.
30614 (_mm256_i64scatter_epi64): Ditto.
30615 (_mm256_mask_i64scatter_epi64): Ditto.
30616 (_mm_i64scatter_epi64): Ditto.
30617 (_mm_mask_i64scatter_epi64): Ditto.
30618 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
30619 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
30620 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
30621 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
30622 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
30623 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
30624 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
30625 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
30626 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
30627 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
30628 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
30629 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
30630 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
30631 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
30632 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
30633 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
30634 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
30635 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
30636 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
30637 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
30638 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
30639 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
30640 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
30641 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
30642 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
30643 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
30644 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
30645 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
30646 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
30647 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
30648 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
30649 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
30650 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
30651 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
30652 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
30653 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
30654 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
30655 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
30656 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
30657 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
30658 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
30659 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
30660 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
30661 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
30662 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
30663 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
30664 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
30665 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
30666 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
30667 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
30668 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
30669 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
30670 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
30671 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
30672 definitions accordingly.
30674 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
30675 Kuan-Lin Chen <kuanlinchentw@gmail.com>
30678 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
30681 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
30684 * ira-conflicts.c (ira_build_conflicts): Update total conflict
30685 hard regs for inner regno.
30687 2017-01-17 Martin Liska <mliska@suse.cz>
30690 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
30691 assumption and add comment.
30693 2017-01-17 Nathan Sidwell <nathan@acm.org>
30695 * ipa-visibility.c (localize_node): New function, broken out of ...
30696 (function_and_variable_visibility): ... here. Call it.
30698 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
30700 PR middle-end/77445
30701 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
30702 correctly set frequency of oudgoing edge.
30703 (duplicate_thread_path): Fix profile updating.
30705 2017-01-17 Jakub Jelinek <jakub@redhat.com>
30708 * configure.ac: Add GCC_BASE_VER.
30709 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
30710 version from BASE-VER file.
30711 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
30712 (gcc.o): Depend on $(BASEVER).
30713 * common.opt (dumpfullversion): New option.
30714 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
30715 * doc/invoke.texi: Document -dumpfullversion.
30716 * doc/install.texi: Document --with-gcc-major-version-only.
30717 * configure: Regenerated.
30719 2017-01-17 Richard Biener <rguenther@suse.de>
30721 PR tree-optimization/71433
30722 * tree-vrp.c (register_new_assert_for): Merge same asserts
30723 on all incoming edges.
30724 (process_assert_insertions_for): Handle insertions at the
30727 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
30729 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
30730 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
30732 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
30735 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
30738 2017-01-17 Alan Modra <amodra@gmail.com>
30741 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
30742 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
30743 symbolic stack limit when pic.
30745 2017-01-16 Martin Sebor <msebor@redhat.com>
30747 PR tree-optimization/78608
30748 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
30750 2017-01-16 Jeff Law <law@redhat.com>
30753 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
30754 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
30755 for several include directories that may be relative to sysroot.
30756 * config/i386/x-mingw32 (gplus_includedir): Define.
30757 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
30758 (native_system_includedir): Likewise.
30759 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
30760 override if TARGET_SYSTEM_ROOT is defined.
30761 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
30763 PR tree-optimization/79090
30764 PR tree-optimization/33562
30765 PR tree-optimization/61912
30766 PR tree-optimization/77485
30767 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
30768 and computed trims into the dump file.
30770 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
30772 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
30774 2017-01-16 Jakub Jelinek <jakub@redhat.com>
30777 * gimplify.c (gimplify_init_constructor): If want_value and
30778 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
30782 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
30783 sequence. Formatting fixes.
30784 (doloop_optimize): Formatting fixes.
30787 * gcc.c (debug_level_greater_than_spec_func): New function.
30788 (static_spec_functions): Add debug-level-gt spec function.
30789 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
30791 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
30792 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
30793 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
30794 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
30795 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
30796 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
30798 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
30800 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
30801 QImode fixups to general and mask registers only.
30803 2017-01-16 Carl Love <cel@us.ibm.com>
30805 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
30806 for built-in functions
30807 vector signed char vec_nabs (vector signed char)
30808 vector signed short vec_nabs (vector signed short)
30809 vector signed int vec_nabs (vector signed int)
30810 vector signed long long vec_nabs (vector signed long long)
30811 vector float vec_nabs (vector float)
30812 vector double vec_nabs (vector double)
30813 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
30815 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
30816 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
30817 * doc/extend.texi: Update the documentation file for the new built-in
30820 2017-01-16 Martin Sebor <msebor@redhat.com>
30822 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
30825 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30827 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
30828 UNSPEC_VSX__XXSPLTD to require special splat handling.
30830 2017-01-16 David Malcolm <dmalcolm@redhat.com>
30833 * system.h: Poison strndup.
30835 2017-01-16 Alan Modra <amodra@gmail.com>
30838 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
30841 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
30843 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
30845 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
30847 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
30848 call recog here. Assert that INSN_CODE (insn) is non-negative.
30850 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
30853 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
30855 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
30856 in the currently scheduled RTL fragment.
30858 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
30860 PR rtl-optimization/78751
30861 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
30864 2017-01-14 Jeff Law <law@redhat.com>
30866 PR tree-optimization/79090
30867 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
30868 variable length stores.
30869 (compute_trims): Delete dead assignment to *trim_tail.
30870 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
30873 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
30875 PR rtl-optimization/78626
30876 PR rtl-optimization/78727
30877 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
30878 of a block, and split such blocks after everything else is finished.
30880 2017-01-14 Alan Modra <amodra@gmail.com>
30883 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
30884 target legitimate_combined_insn.
30885 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
30886 (rs6000_legitimate_combined_insn): New function.
30887 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
30889 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
30890 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
30891 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
30893 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
30895 * doc/frontends.texi (G++ and GCC): Remove references to Java.
30897 2017-01-13 Jeff Law <law@redhat.com>
30899 PR tree-optimization/33562
30900 PR tree-optimization/61912
30901 PR tree-optimization/77485
30902 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
30904 (delete_dead_assignment): Likewise.
30905 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
30906 statement to delete_dead_call and delete_dead_assignment.
30908 2017-01-13 David Malcolm <dmalcolm@redhat.com>
30911 * substring-locations.c (format_warning_va): Strengthen case 1 so
30912 that both endpoints of the substring must be within the format
30913 range for just the substring to be printed.
30915 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
30917 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
30918 * config/i386/i386.c (ix86_target_string): Add missing options
30919 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
30920 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
30921 flags_other and ix86_target_other to flags2_other. Display unknown
30923 (ix86_valid_target_attribute_inner_p): Add missing options and
30924 reorder options by implied ISAs, as in ix86_target_string.
30926 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
30928 * hash-table.h (hash_table::too_empty_p): New function.
30929 (hash_table::expand): Use it.
30930 (hash_table::traverse): Likewise.
30931 (hash_table::empty_slot): Use sizeof (value_type) instead of
30932 sizeof (PTR) to convert bytes to elements. Shrink the table
30933 if the current size is excessive for the current number of
30936 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
30938 * ira-costs.c (record_reg_classes): Break from the inner loop
30939 early once alt_fail is known to be true. Update outer loop
30940 handling accordingly.
30942 2017-01-13 Jeff Law <law@redhat.com>
30944 * tree-ssa-dse.c (decrement_count): New function.
30945 (increment_start_addr, maybe_trim_memstar_call): Likewise.
30946 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
30947 when we know the partially dead statement is a mem* function.
30949 PR tree-optimization/61912
30950 PR tree-optimization/77485
30951 * tree-ssa-dse.c: Include expr.h.
30952 (maybe_trim_constructor_store): New function.
30953 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
30955 PR tree-optimization/33562
30956 PR tree-optimization/61912
30957 PR tree-optimization/77485
30958 * doc/invoke.texi: Document new dse-max-object-size param.
30959 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
30960 * tree-ssa-dse.c: Include params.h.
30961 (dse_store_status): New enum.
30962 (initialize_ao_ref_for_dse): New, partially extracted from
30964 (valid_ao_ref_for_dse, normalize_ref): New.
30965 (setup_live_bytes_from_ref, compute_trims): Likewise.
30966 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
30967 (maybe_trim_partially_dead_store): Likewise.
30968 (maybe_trim_complex_store): Likewise.
30969 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
30970 Track what bytes live from the original store. Return tri-state
30971 for dead, partially dead or live.
30972 (dse_dom_walker): Add constructor, destructor and new private members.
30973 (delete_dead_call, delete_dead_assignment): New extracted from
30975 (dse_optimize_stmt): Make a member of dse_dom_walker.
30976 Use initialize_ao_ref_for_dse.
30978 PR tree-optimization/33562
30979 PR tree-optimization/61912
30980 PR tree-optimization/77485
30981 * sbitmap.h (bitmap_count_bits): Prototype.
30982 (bitmap_clear_range, bitmap_set_range): Likewise.
30983 * sbitmap.c (bitmap_clear_range): New function.
30984 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
30986 2017-01-13 Martin Liska <mliska@suse.cz>
30989 * function.c (set_cfun): Add new argument force.
30990 * function.h (set_cfun): Likewise.
30991 * ipa-inline-transform.c (inline_call): Use the function when
30992 strict alising from is dropped for function we inline to.
30994 2017-01-13 Richard Biener <rguenther@suse.de>
30996 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
30997 for dumping GIMPLE INTEGER_CSTs.
30999 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
31001 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
31002 to 201112L since C++17.
31004 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
31007 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
31008 if -fsanitize=kernel-address is present.
31010 2017-01-13 Richard Biener <rguenther@suse.de>
31012 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
31013 as _Literal ( type ) number in case usual suffixes do not
31014 preserve all information.
31016 2017-01-13 Richard Biener <rguenther@suse.de>
31018 PR tree-optimization/77283
31019 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
31020 and ssa-iterators.h.
31021 (is_feasible_trace): Implement a cost model based on joiner
31024 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
31027 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
31028 char or short to __float128/_Float128 directly.
31030 2017-01-12 Martin Sebor <msebor@redhat.com>
31032 to -Wformat-overflow.
31033 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
31034 (min_bytes_remaining): Same.
31035 (get_string_length): Same.
31036 (format_string): Same.
31037 (format_directive): Same.
31039 (pass_sprintf_length::handle_gimple_call): Same.
31041 2017-01-12 Jakub Jelinek <jakub@redhat.com>
31043 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
31044 info.nowrite calls with no lhs that can't throw. Return bool
31045 whether gsi_remove has been called or not.
31046 (pass_sprintf_length::handle_gimple_call): Return bool whether
31047 try_substitute_return_value called gsi_remove. Formatting fix.
31048 (pass_sprintf_length::execute): Don't use gsi_remove if
31049 handle_gimple_call returned true.
31052 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
31053 be removed due to side-effects, don't remove following barrier nor
31054 turn the successor edge into fallthru edge.
31056 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31059 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
31060 element-reversing loads and stores as not swappable.
31062 2017-01-12 Nathan Sidwell <nathan@acm.org>
31063 Nicolai Stange <nicstange@gmail.com>
31065 * combine.c (try_combine): Don't ignore result of overlap checking
31066 loop. Combine overlap & asm check into single loop.
31068 2017-01-12 Richard Biener <rguenther@suse.de>
31070 * tree-pretty-print.c (dump_generic_node): Provide -gimple
31071 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
31073 2017-01-12 Richard Biener <rguenther@suse.de>
31075 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
31076 and TS_TARGET_OPTION directly derive from TS_BASE.
31077 * tree-core.h (tree_optimization_option): Derive from tree_base.
31078 (tree_target_option): Likewise.
31080 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
31082 * config/i386/i386.c (memory_address_length): Increase len
31083 only when rip_relative_addr_p returns false.
31085 2017-01-11 Julia Koval <julia.koval@intel.com>
31087 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
31088 (OPTION_MASK_ISA_SGX_SET): New.
31089 (ix86_handle_option): Handle OPT_msgx.
31090 * config.gcc: Added sgxintrin.h.
31091 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
31092 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
31093 * config/i386/i386.c (ix86_target_string): Add -msgx.
31095 (ix86_option_override_internal): Handle new options.
31096 (ix86_valid_target_attribute_inner_p): Add sgx.
31097 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
31098 * config/i386/i386.opt: Add msgx.
31099 * config/i386/sgxintrin.h: New file.
31100 * config/i386/x86intrin.h: Add sgxintrin.h.
31102 2017-01-11 Jakub Jelinek <jakub@redhat.com>
31105 * fold-const.c (maybe_nonzero_address): Return 1 for function
31107 (tree_single_nonzero_warnv_p): Don't handle function local objects
31111 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
31114 2017-01-11 David Malcolm <dmalcolm@redhat.com>
31117 * opts.c: Include "spellcheck.h"
31118 (struct string_fragment): New struct.
31119 (struct edit_distance_traits<const string_fragment &>): New
31121 (get_closest_sanitizer_option): New function.
31122 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
31124 2017-01-11 Jakub Jelinek <jakub@redhat.com>
31126 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
31128 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
31129 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
31130 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
31131 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
31132 for initial die_offset if dwarf_split_debug_info.
31133 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
31134 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
31136 (output_skeleton_debug_sections): Formatting fix. Use
31137 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
31138 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
31140 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
31142 * config/arm/cortex-a53.md: Add bypasses for
31143 cortex_a53_r2f_cvt.
31144 (cortex_a53_r2f): Only use for transfers.
31145 (cortex_a53_f2r): Likewise.
31146 (cortex_a53_r2f_cvt): Add reservation for conversions.
31147 (cortex_a53_f2r_cvt): Likewise.
31149 2017-01-11 Tamar Christina <tamar.christina@arm.com>
31151 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
31152 to all inlined functions, change static to extern.
31154 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
31157 * config/arm/arm.c (legitimize_pic_address): Handle reference to
31159 (arm_assemble_integer): Likewise.
31161 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
31163 * config.gcc: Use new awk script to check CPU, FPU and architecture
31164 parameters for --with-... options.
31165 * config/arm/parsecpu.awk: New file
31166 * config/arm/arm-cpus.in: New file.
31167 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
31169 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
31171 * config/arm/t-arm: Update dependency rules.
31172 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
31173 of processing .def files.
31174 * config/arm/genopt.sh: Deleted.
31175 * config/arm/gentune.sh: Deleted.
31176 * config/arm/arm-cores.def: Deleted.
31177 * config/arm/arm-arches.def: Deleted.
31178 * config/arm/arm-fpus.def: Deleted.
31179 * config/arm/arm-tune.md: Regenerated.
31180 * config/arm/arm-tables.opt: Regenerated.
31181 * config/arm/arm-cpu.h: New generated file.
31182 * config/arm/arm-cpu-data.h: New generated file.
31183 * config/arm/arm-cpu-cdata.h: New generated file.
31185 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
31188 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
31190 (input_varpool_node): Unpack dynamically_initialized bit.
31192 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
31194 PR rtl-optimization/79032
31195 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
31196 the alignment of the adjusted memory reference against that of MODE,
31197 instead of the alignment of the original memory reference.
31199 2017-01-11 Martin Jambor <mjambor@suse.cz>
31201 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
31203 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
31204 decorated functions.
31206 2017-01-11 Richard Biener <rguenther@suse.de>
31208 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
31209 set range/nonnull info for PHI results. Do not set it on
31210 stmts marked for removal.
31212 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
31214 * expr.c (store_field): In the bitfield case, fetch the return value
31215 from the registers before applying a single big-endian adjustment.
31216 Always do a final load for a BLKmode value not larger than a word.
31218 2017-01-10 David Malcolm <dmalcolm@redhat.com>
31221 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
31222 that we correctly handle column numbers greater than
31223 LINE_MAP_MAX_COLUMN_NUMBER.
31225 2017-01-10 Martin Sebor <msebor@redhat.com>
31227 PR middle-end/78245
31228 * gimple-ssa-sprintf.c (get_destination_size): Call
31229 {init,fini}object_sizes.
31230 * tree-object-size.c (addr_object_size): Adjust.
31231 (pass_through_call): Adjust.
31232 (pass_object_sizes::execute): Adjust.
31233 * tree-object-size.h (fini_object_sizes): Declare.
31235 2017-01-10 Martin Sebor <msebor@redhat.com>
31237 PR tree-optimization/78775
31238 * builtins.c (get_size_range): Move...
31239 * calls.c: ...to here.
31240 (alloc_max_size): Accept zero argument.
31241 (operand_signed_p): Remove.
31242 (maybe_warn_alloc_args_overflow): Call get_size_range.
31243 * calls.h (get_size_range): Declare.
31245 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
31247 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
31248 from TI's devices.csv file as of September 2016.
31249 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
31251 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
31253 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
31254 * doc/invoke.texi: Likewise.
31255 * doc/md.texi: Likewise.
31256 * doc/objc.texi: Likewise.
31258 2017-01-10 Joshua Conner <joshconner@google.com>
31260 * config/arm/fuchsia-elf.h: New file.
31261 * config/fuchsia.h: New file.
31262 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
31263 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
31265 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
31267 2016-01-10 Richard Biener <rguenther@suse.de>
31269 PR tree-optimization/79034
31270 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
31271 Propagate out degenerate PHIs in the joiner.
31273 2017-01-10 Martin Liska <mliska@suse.cz>
31275 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
31276 (sort_congruence_classes_by_decl_uid): Likewise.
31277 (sort_congruence_class_groups_by_decl_uid): Likewise.
31278 (sem_item_optimizer::merge_classes): Sort class, groups in these
31279 classes and members in the groups by DECL_UID of declarations.
31280 This would make merge operations stable.
31282 2017-01-10 Martin Liska <mliska@suse.cz>
31284 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
31285 usage of m_classes_vec.
31286 (sem_item_optimizer::~sem_item_optimizer): Likewise.
31287 (sem_item_optimizer::get_group_by_hash): Likewise.
31288 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
31289 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
31290 (sem_item_optimizer::verify_classes): Likewise.
31291 (sem_item_optimizer::process_cong_reduction): Likewise.
31292 (sem_item_optimizer::dump_cong_classes): Likewise.
31293 (sem_item_optimizer::merge_classes): Likewise.
31294 * ipa-icf.h (congruence_class_hash): Rename from
31295 congruence_class_group_hash. Remove declaration of m_classes_vec.
31297 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
31299 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
31300 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
31301 * config.gcc: Add avx512vpopcntdqintrin.h.
31302 * config/i386/avx512vpopcntdqintrin.h: New.
31303 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
31304 * config/i386/i386-builtin-types.def: Add new types.
31305 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
31306 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
31307 __builtin_ia32_vpopcountq_v8di_mask): New.
31308 * config/i386/i386-c.c (ix86_target_macros_internal): Define
31309 __AVX512VPOPCNTDQ__.
31310 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
31311 (PTA_AVX512VPOPCNTDQ): Define.
31312 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
31313 TARGET_AVX512VPOPCNTDQ_P): Define.
31314 * config/i386/i386.opt: Add mavx512vpopcntdq.
31315 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
31316 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
31318 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
31320 PR middle-end/77484
31321 * predict.def (PRED_CALL): Set to 67.
31323 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
31325 * expr.c (store_field): In the bitfield case, if the value comes from
31326 a function call and is of an aggregate type returned in registers, do
31327 not modify the field mode; extract the value in all cases if the mode
31328 is BLKmode and the size is not larger than a word.
31330 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
31333 * config/i386/cpuid.h: Fix undefined behavior.
31335 2017-01-04 Jeff Law <law@redhat.com>
31337 PR tree-optimization/79007
31338 PR tree-optimization/67955
31339 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
31340 conservative for pt.null when flag_non_call_exceptions is on.
31342 2017-01-09 Jakub Jelinek <jakub@redhat.com>
31344 PR translation/79019
31345 PR translation/79020
31346 * params.def (PARAM_INLINE_MIN_SPEEDUP,
31347 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
31348 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
31350 * config/avr/avr.opt (maccumulate-args): Likewise.
31351 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
31352 * common.opt (freport-bug): Likewise.
31353 * cif-code.def (CIF_FINAL_ERROR): Likewise.
31354 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
31355 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
31356 translatable string.
31357 * config/i386/i386.c (function_value_32): Likewise.
31358 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
31359 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
31361 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
31362 * common/config/msp430/msp430-common.c (msp430_handle_option):
31364 * symtab.c (symtab_node::verify_base): Likewise.
31365 * opts.c (set_debug_level): Likewise.
31366 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
31367 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
31368 missing whitespace to translatable strings.
31369 * config/avr/avr.md (bswapsi2): Fix typo in comment.
31370 * config/sh/superh.h: Likewise.
31371 * config/i386/xopintrin.h: Likewise.
31372 * config/i386/znver1.md: Likewise.
31373 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
31374 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
31375 * double-int.h (struct double_int): Likewise.
31376 * double-int.c (div_and_round_double): Likewise.
31377 * wide-int.cc: Likewise.
31378 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
31379 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
31380 * cfgcleanup.c (crossjumps_occured): Renamed to ...
31381 (crossjumps_occurred): ... this.
31382 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
31385 PR tree-optimization/78899
31386 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
31387 returning bool return struct loop *, NULL for failure and the new
31389 (versionable_outer_loop_p): Don't version outer loop if it has
31390 dont_vectorized bit set.
31391 (tree_if_conversion): When versioning outer loop, ensure
31392 tree_if_conversion is performed also on the inner loop of the
31393 non-vectorizable outer loop copy.
31394 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
31395 LOOP_VECTORIZED in inner loop of the scalar outer loop and
31396 prevent vectorization of it.
31397 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
31398 the outer loop vectorization of the non-scalar version is attempted
31399 before vectorization of the inner loop in scalar version. If
31400 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
31401 vectorization of its inner loop.
31402 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
31403 has 2 inner loops, rename also on edges from bb whose single pred
31404 is outer_loop->header. Fix typo in function comment.
31406 2017-01-09 Martin Sebor <msebor@redhat.com>
31409 * asan.c (asan_emit_stack_protection): Increase local buffer size
31410 to avoid snprintf truncation warning.
31412 2017-01-09 Andrew Pinski <apinski@cavium.com>
31414 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
31415 to reference thunderx2t99 for the tuning structure
31416 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
31418 (thunderx2t99_extra_costs): This.
31419 * config/aarch64/aarch64-tune.md: Regenerate.
31420 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
31421 (vulcan_addrcost_table): This.
31422 (vulcan_regmove_cost): Rename to ...
31423 (thunderx2t99_regmove_cost): This.
31424 (vulcan_vector_cost): Rename to ...
31425 (thunderx2t99_vector_cost): this.
31426 (vulcan_branch_cost): Rename to ...
31427 (thunderx2t99_branch_cost): This.
31428 (vulcan_tunings): Rename to ...
31429 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
31430 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
31432 2017-01-09 Martin Jambor <mjambor@suse.cz>
31436 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
31437 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
31438 (propagate_vr_accross_jump_function): Use the above function for all
31439 value range computations for pass-through jump functions and type
31440 converasion from explicit value range values.
31441 (ipcp_propagate_stage): Do not attempt to deduce types of formal
31442 parameters from TYPE_ARG_TYPES.
31443 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
31444 (ipa_write_node_info): Stream type of the actual argument.
31445 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
31447 2017-01-09 Martin Liska <mliska@suse.cz>
31450 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
31451 (lookup_compiler): Do not show error message with have_E.
31453 2017-01-09 Jakub Jelinek <jakub@redhat.com>
31455 PR tree-optimization/78938
31456 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
31457 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
31458 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
31459 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
31462 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31464 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
31467 2017-01-09 Richard Biener <rguenther@suse.de>
31469 PR tree-optimization/78997
31470 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
31471 name condition properly.
31473 2017-01-09 Richard Biener <rguenther@suse.de>
31476 * dwarf2out.c (is_cxx): New overload with context.
31477 (is_naming_typedef_decl): Use it.
31479 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
31481 * invoke.texi (Option Summary): Correct spacing in option lists
31482 and add line breaks to fix over-long lines.
31484 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
31486 PR middle-end/17660
31488 * extend.texi (Common Variable Attributes): Add xref to GCC
31489 Internals manual to explain mode attribute keywords.
31491 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
31494 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
31495 and Preprocessor Options.
31496 (Options for Linking): Document -pthread here....
31497 (RS/6000 and PowerPC Options): ...not here.
31498 (Solaris 2 Options): ...or here.
31499 * doc/cppopts.texi: Document -pthread.
31501 2017-01-08 Martin Sebor <msebor@redhat.com>
31503 PR middle-end/77708
31504 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
31505 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
31506 New member functions.
31507 (format_directive): Used them.
31509 (pass_sprintf_length::handle_gimple_call): Same.
31510 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
31511 to avoid truncation for any argument.
31512 (extract_affine_mul): Same.
31513 * tree.c (get_file_function_name): Same.
31515 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
31517 PR middle-end/77484
31518 * predict.def (PRED_INDIR_CALL): Set to 86.
31520 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
31522 PR preprocessor/54124
31523 * doc/cppopts.texi: Reformat -d subtable to list the full name
31524 of the options. Add cross-reference to the docs for the general
31525 compiler -d options.
31526 * doc/invoke.texi (Developer Options): Add cross-reference to the
31527 preprocessor-specific -d option documentation.
31529 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
31531 PR preprocessor/13498
31532 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
31533 redudant material, and reflect new command-line options.
31534 (System Headers): Likewise.
31536 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
31538 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
31539 -isystem, and -idirafter. Copy-edit.
31540 * doc/cppopts.texi: Copy-edit. Remove contradiction about
31541 default for -ftrack-macro-expansion. Delete obsolete and
31542 badly-formatted implementation details about -fdebug-cpp output.
31543 * doc/cppwarnopts.texi: Copy-edit.
31545 2017-01-07 David Malcolm <dmalcolm@redhat.com>
31548 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
31549 that the transition from a max line width >= 1<<10 to narrower
31550 lines works correctly.
31552 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
31554 * doc/options.texi (PerFunction): New.
31555 * opt-functions.awk (switch_flags): Map both Optimization and
31556 PerFunction to CL_OPTIMIZATION.
31557 * opth-gen.awk: Test for PerFunction flag along with
31559 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
31560 it only when the latter is present. Skip those that don't in
31561 the hash function generator.
31562 * common.opt (fvar-tracking): Mark as PerFunction instead of
31564 (fvar-tracking-assignments): Likewise.
31565 (fvar-tracking-assignments-toggle): Likewise.
31566 (fvar-tracking-uninit): Likewise.
31568 2017-01-07 Jakub Jelinek <jakub@redhat.com>
31570 PR translation/79018
31571 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
31574 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
31577 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
31578 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
31579 TARGET_LONG_JUMP_TABLE_OFFSETS.
31580 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
31581 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
31582 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
31583 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
31584 * config/m68k/m68k.md (tablejump expander): Likewise.
31585 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
31586 TARGET_LONG_JUMP_TABLE_OFFSETS.
31587 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
31588 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
31590 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
31591 David Holsgrove <david.holsgrove@xilinx.com>
31593 * common/config/microblaze/microblaze-common.c
31594 (TARGET_EXCEPT_UNWIND_INFO): Remove.
31595 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
31597 * config/microblaze/microblaze.c (microblaze_must_save_register)
31598 (microblaze_expand_epilogue, microblaze_return_addr): Handle
31600 (microblaze_eh_return): New function.
31601 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
31602 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
31603 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
31604 * config/microblaze/microblaze.md (eh_return): New pattern.
31606 2017-01-06 Jakub Jelinek <jakub@redhat.com>
31608 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
31609 GCC_DIAGNOSTIC_STRINGIFY): Define.
31611 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
31613 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31615 * config/arm/arm.md (<mcrr>): New.
31617 * config/arm/arm.c (arm_arch5te): New.
31618 (arm_option_override): Set arm_arch5te.
31619 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
31621 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
31622 (arm_mcrr_qualifiers): ... this. New.
31623 (MRRC_QUALIFIERS): Define to...
31624 (arm_mrrc_qualifiers): ... this. New.
31625 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
31627 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
31628 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
31629 (MRRCI, mrrc, MRRC): New.
31630 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
31631 VUNSPEC_MRRC2): New.
31633 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31635 * config/arm/arm.md (<mcr>): New.
31637 * config/arm/arm.c (arm_coproc_builtin_available): Add
31638 support for mcr, mrc, mcr2 and mrc2.
31639 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
31640 (arm_mcr_qualifiers): ... this. New.
31641 (MRC_QUALIFIERS): Define to ...
31642 (arm_mrc_qualifiers): ... this. New.
31643 (MCR_QUALIFIERS): Define to ...
31644 (arm_mcr_qualifiers): ... this. New.
31645 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
31647 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
31648 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
31649 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
31650 VUNSPEC_MRC2): New.
31652 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31654 * config/arm/arm.md (*ldc): New.
31658 * config/arm/arm.c (arm_coproc_builtin_available): Add
31659 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
31660 (arm_coproc_ldc_stc_legitimate_address): New.
31661 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
31662 'qualifier_const_pointer'.
31663 (LDC_QUALIFIERS): Define to...
31664 (arm_ldc_qualifiers): ... this. New.
31665 (STC_QUALIFIERS): Define to...
31666 (arm_stc_qualifiers): ... this. New.
31667 * config/arm/arm-protos.h
31668 (arm_coproc_ldc_stc_legitimate_address): New.
31669 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
31670 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
31671 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
31672 stc2, stcl, stc2l): New.
31673 * config/arm/constraints.md (Uz): New.
31674 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
31675 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
31676 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
31677 VUNSPEC_STC2L): New.
31679 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31681 * config/arm/arm.md (<cdp>): New.
31682 * config/arm/arm.c (neon_const_bounds): Rename this ...
31683 (arm_const_bounds): ... this.
31684 (arm_coproc_builtin_available): New.
31685 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
31686 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
31687 (CDP_QUALIFIERS): Define to...
31688 (arm_cdp_qualifiers): ... this. New.
31690 (arm_expand_builtin_args): Add case for 6 arguments.
31691 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
31692 (arm_const_bounds): ... this.
31693 (arm_coproc_builtin_available): New.
31694 * config/arm/arm_acle.h (__arm_cdp): New.
31696 * config/arm/arm_acle_builtins.def (cdp): New.
31698 * config/arm/iterators.md (CDPI,CDP,cdp): New.
31699 * config/arm/neon.md: Rename all 'neon_const_bounds' to
31700 'arm_const_bounds'.
31701 * config/arm/types.md (coproc): New.
31702 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
31703 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
31704 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
31705 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
31707 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31709 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
31710 (UBINOP_QUALIFIERS): New.
31712 (acle_builtin_data): New. Change comment.
31713 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
31714 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
31715 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
31716 arm_acle_builtins.def.
31717 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
31718 (arm_init_acle_builtins): New.
31719 (CRC32_BUILTIN): Remove.
31720 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
31721 crc32cb, crc32ch and crc32cw.
31722 (arm_init_crc32_builtins): Remove.
31723 (arm_init_builtins): Use arm_init_acle_builtins rather
31724 than arm_init_crc32_builtins.
31725 (arm_expand_acle_builtin): New.
31726 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
31727 * config/arm/arm_acle_builtins.def: New.
31729 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31731 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
31732 (arm_builtin_datum): ... this.
31733 (arm_init_neon_builtin): Rename to ...
31734 (arm_init_builtin): ... this. Add a new parameters PREFIX
31735 and USE_SIG_IN_NAME.
31736 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
31737 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
31738 'arm_builtin_datum'.
31739 (arm_init_vfp_builtins): Likewise.
31740 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
31741 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
31742 (arm_expand_neon_args): Rename to ...
31743 (arm_expand_builtin_args): ... this. Rename builtin_arg
31744 enum values and differentiate between ARG_BUILTIN_MEMORY
31745 and ARG_BUILTIN_NEON_MEMORY.
31746 (arm_expand_neon_builtin_1): Rename to ...
31747 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
31748 values, arm_expand_builtin_args and add bool parameter NEON.
31749 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
31750 (arm_expand_vfp_builtin): Likewise.
31751 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
31753 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
31755 PR middle-end/77484
31756 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
31757 * predict.c (tree_estimate_probability_bb): Reverse direction of
31758 polymorphic call predictor.
31760 2017-01-06 David Malcolm <dmalcolm@redhat.com>
31762 * passes.c (execute_one_pass): Split out pass-skipping logic into...
31763 (determine_pass_name_match): ...this new function and...
31764 (should_skip_pass_p): ...this new function.
31766 2017-01-06 Nathan Sidwell <nathan@acm.org>
31768 * ipa-visibility.c (function_and_variable_visibility): Reformat
31769 comments and long lines. Remove extrneous if.
31770 * symtab.c (symtab_node::make_decl_local): Fix code format.
31771 (symtab_node::set_section_for_node): Fix comment typo.
31773 2017-01-06 Martin Liska <mliska@suse.cz>
31776 * lra-constraints.c: Rename invariant to lra_invariant.
31777 * predict.c (set_even_probabilities): Initialize e to NULL.
31779 2017-01-05 Martin Sebor <msebor@redhat.com>
31781 PR tree-optimization/78910
31782 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
31783 (format_integer): Correct off-by-one error in the handling
31784 of precision with negative numbers in signed conversions..
31786 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
31788 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
31790 2017-01-05 Jakub Jelinek <jakub@redhat.com>
31792 PR tree-optimization/71016
31793 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
31794 factor_out_conditional_conversion. Formatting fix.
31795 (factor_out_conditional_conversion): Add cond_stmt argument.
31796 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
31797 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
31800 2017-01-05 David Malcolm <dmalcolm@redhat.com>
31802 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
31803 read-rtl-function.o, and selftest-rtl.o.
31804 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
31805 (selftest::aarch64_test_loading_full_dump): New function.
31806 (selftest::aarch64_run_selftests): New function.
31807 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
31808 selftest::aarch64_run_selftests.
31809 * config/i386/i386.c
31810 (selftest::ix86_test_loading_dump_fragment_1): New function.
31811 (selftest::ix86_test_loading_call_insn): New function.
31812 (selftest::ix86_test_loading_full_dump): New function.
31813 (selftest::ix86_test_loading_unspec): New function.
31814 (selftest::ix86_run_selftests): Call the new functions.
31815 * emit-rtl.c (maybe_set_max_label_num): New function.
31816 * emit-rtl.h (maybe_set_max_label_num): New decl.
31817 * function.c (instantiate_decls): Guard call to
31818 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
31819 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
31821 * gensupport.c (gen_reader::gen_reader): Pass "false"
31822 for new "compact" param of rtx_reader.
31823 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
31824 rather than an empty string for NULL strings.
31825 * read-md.c: Potentially include config.h rather than bconfig.h.
31826 Wrap include of errors.h with #ifdef GENERATOR_FILE.
31827 (have_error): New global, copied from errors.c.
31828 (md_reader::read_name): Rename to...
31829 (md_reader::read_name_1): ...this, adding "out_loc" param,
31830 and converting "missing name or number" to returning false, rather
31832 (md_reader::read_name): Reimplement in terms of read_name_1.
31833 (md_reader::read_name_or_nil): New function.
31834 (md_reader::read_string): Handle "(nil)" by returning NULL.
31835 (md_reader::md_reader): Add new param "compact".
31836 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
31837 (md_reader::read_file): New method.
31838 * read-md.h (md_reader::md_reader): Add new param "compact".
31839 (md_reader::read_file): New method.
31840 (md_reader::is_compact): New accessor.
31841 (md_reader::read_name): Convert return type from void to file_location.
31842 (md_reader::read_name_or_nil): New decl.
31843 (md_reader::read_name_1): New decl.
31844 (md_reader::m_compact): New field.
31845 (noop_reader::noop_reader): Pass "false" for new "compact" param
31847 (rtx_reader::rtx_reader): Add new "compact" param.
31848 (rtx_reader::read_rtx_operand): Make virtual and convert return
31849 type from void to rtx.
31850 (rtx_reader::read_until): New decl.
31851 (rtx_reader::handle_any_trailing_information): New virtual function.
31852 (rtx_reader::postprocess): New virtual function.
31853 (rtx_reader::finalize_string): New virtual function.
31854 (rtx_reader::m_in_call_function_usage): New field.
31855 (rtx_reader::m_reuse_rtx_by_id): New field.
31856 * read-rtl-function.c: New file.
31857 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
31858 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
31859 (selftest::verify_three_block_rtl_cfg): New decl.
31860 * read-rtl-function.h: New file.
31861 * read-rtl.c: Potentially include config.h rather than bconfig.h.
31862 For host, include function.h, memmodel.h, and emit-rtl.h.
31863 (one_time_initialization): New function.
31864 (struct compact_insn_name): New struct.
31865 (compact_insn_names): New array.
31866 (find_code): Handle insn codes in compact dumps.
31867 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
31868 (bind_subst_iter_and_attr): Likewise.
31869 (add_condition_to_string): Likewise.
31870 (add_condition_to_rtx): Likewise.
31871 (apply_attribute_uses): Likewise.
31872 (add_current_iterators): Likewise.
31873 (apply_iterators): Likewise.
31874 (initialize_iterators): Guard usage of apply_subst_iterator with
31875 #ifdef GENERATOR_FILE.
31876 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
31877 (md_reader::read_mapping): Likewise.
31878 (add_define_attr_for_define_subst): Likewise.
31879 (add_define_subst_attr): Likewise.
31880 (read_subst_mapping): Likewise.
31881 (check_code_iterator): Likewise.
31882 (rtx_reader::read_rtx): Likewise. Move one-time initialization
31884 (one_time_initialization): New function.
31885 (rtx_reader::read_until): New method.
31886 (read_flags): New function.
31887 (parse_reg_note_name): New function.
31888 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
31889 Handle reuse_rtx ids.
31890 Wrap iterator lookup within #ifdef GENERATOR_FILE.
31891 Add parsing support for RTL dumps, mirroring the special-cases in
31892 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
31893 values, and calling handle_any_trailing_information.
31894 (rtx_reader::read_rtx_operand): Convert return type from void
31895 to rtx, returning return_rtx. Handle case 'e'. Call
31896 finalize_string on XSTR and XTMPL fields.
31897 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
31898 "(nil)" values were omitted. Call the postprocess vfunc on the
31900 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
31901 class ctor. Initialize m_in_call_function_usage. Call
31902 one_time_initialization.
31903 * rtl-tests.c (selftest::test_uncond_jump): Call
31904 set_new_first_and_last_insn.
31905 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
31906 * selftest-rtl.c: New file.
31907 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
31908 (selftest::get_insn_by_uid): New decl.
31909 * selftest-run-tests.c (selftest::run_tests): Call
31910 read_rtl_function_c_tests.
31911 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
31912 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
31915 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
31917 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
31918 operands in a special way. Assert that pos+len <= mode precision.
31920 2017-01-05 Jakub Jelinek <jakub@redhat.com>
31922 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
31923 3 argument Alias with unlimited for the negative form.
31924 (fno-vect-cost-model): Removed.
31926 2017-01-05 Martin Liska <mliska@suse.cz>
31928 * hsa-gen.c (gen_hsa_divmod): New function.
31929 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
31931 2017-01-05 Martin Liska <mliska@suse.cz>
31934 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
31937 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31939 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
31940 small constant length operands.
31942 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31944 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
31945 between loop iterations.
31947 2017-01-05 Martin Liska <mliska@suse.cz>
31950 * gimplify.c (gimplify_decl_expr): Compare to
31951 asan_poisoned_variables instread of checking flags.
31952 (gimplify_target_expr): Likewise.
31953 (gimplify_expr): Likewise.
31954 (gimplify_function_tree): Conditionally initialize
31955 asan_poisoned_variables.
31957 2017-01-04 Jeff Law <law@redhat.com>
31959 PR tree-optimizatin/78812
31960 * rtl.h (contains_mem_rtx_p): Prototype.
31961 * ifcvt.c (containts_mem_rtx_p): Move from here to...
31962 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
31963 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
31964 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
31965 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
31967 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
31969 * input.c (assert_char_at_range): Default-initialize actual_range.
31971 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
31973 * df-scan.c (df_ref_create_structure): Make regno unsigned,
31974 to match the caller.
31976 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
31978 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
31979 insns after final jump in test to emit dummy move.
31981 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
31983 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
31984 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
31986 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
31988 * multiple_target.c (create_dispatcher_calls): Init e_next.
31989 * tree-ssa-loop-split.c (split_loop): Init border.
31990 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
31993 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
31998 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
31999 (altivec_register_operand): Do not return true if the operand
32000 contains a SUBREG mixing SImode and SFmode.
32001 (vsx_register_operand): Likewise.
32002 (vsx_reg_sfsubreg_ok): New predicate.
32003 (vfloat_operand): Do not return true if the operand contains a
32004 SUBREG mixing SImode and SFmode.
32005 (vint_operand): Likewise.
32006 (vlogical_operand): Likewise.
32007 (gpc_reg_operand): Likewise.
32008 (int_reg_operand): Likewise.
32009 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
32010 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
32011 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
32013 (rs6000_emit_move_si_sf_subreg): New helper function.
32014 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
32015 fixup SUBREGs involving SImode and SFmode.
32016 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
32017 numbers for the new peephole2 optimization.
32018 (peephole2 for SFmode unions): New peephole2 to optimize cases in
32019 the GLIBC math library that do AND/IOR/XOR operations on single
32020 precision floating point.
32021 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
32022 target macros to say whether we need to avoid SUBREGs mixing
32024 (TARGET_ALLOW_SF_SUBREG): Likewise.
32025 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
32026 (UNSPEC_SI_FROM_SF): Likewise.
32027 (iorxor): Change spacing.
32028 (and_ior_xor): New iterator for AND, IOR, and XOR.
32029 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
32030 (movdi_from_sf_zero_ext): Likewise.
32031 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
32032 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
32033 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
32034 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
32035 (fms<mode>4): Likewise.
32036 (fnma<mode>4): Likewise.
32037 (fnms<mode>4): Likewise.
32038 (nfma<mode>4): Likewise.
32039 (nfms<mode>4): Likewise.
32041 2017-01-04 Marek Polacek <polacek@redhat.com>
32044 * doc/invoke.texi: Document -Wpointer-compare.
32046 2017-01-04 Jakub Jelinek <jakub@redhat.com>
32048 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
32051 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
32052 descriptions for -gdwarf-5 and emit them as uleb128 instead of
32055 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
32058 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
32059 documentation of the powerpc_popcntb_ok attribute.
32060 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
32061 code to issue warning messages if a requested CPU configuration is
32062 not supported by the binary (assembler and loader) toolchain.
32063 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
32064 made to define a built-in function that has been disabled.
32065 (paired_init_builtins): Add assertion to prevent ICE if attempt is
32066 made to define a built-in function that has been disabled.
32067 (altivec_init_builtins): Add comment explaining why definition
32068 of the DST built-in functions is not preceded by an assertion
32069 check. Add assertions to prevent ICE if attempts are made to
32070 define an altivec predicate or an abs* built-in function that has
32072 (htm_init_builtins): Add comment explaining why definition of the
32073 htm built-in functions is not preceded by an assertion check.
32075 2017-01-04 Jeff Law <law@redhat.com>
32077 PR tree-optimizatin/67955
32078 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
32079 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
32080 the points-to solution does not include pt_null. Use DECL_PT_UID
32083 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
32085 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
32086 Use gen_int_mode instead of gen_lopwart for const_int operands.
32088 2017-01-04 Jakub Jelinek <jakub@redhat.com>
32090 PR tree-optimization/71563
32091 * match.pd: Simplify X << Y into X if Y is known to be 0 or
32092 out of range value - has low bits known to be zero.
32094 2017-01-04 Alan Modra <amodra@gmail.com>
32096 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
32097 * configure: Regenerate.
32098 * config.in: Regenerate.
32100 2017-01-04 Jakub Jelinek <jakub@redhat.com>
32103 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
32104 a substring of the message, but strcmp with the whole message. Ifdef
32105 ENABLE_NLS, translate the message first using dgettext.
32107 2017-01-03 Jeff Law <law@redhat.com>
32109 PR tree-optimizatin/78856
32110 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
32111 (mark_threaded_blocks): Remove code to truncate thread paths that
32112 cross multiple loop headers. Instead invalidate the cached loop
32113 iteration information and handle case of a thread path walking
32114 into an irreducible region.
32116 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
32119 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
32120 assertions. Add support for doing the signbit if the IEEE 128-bit
32121 floating point value is in a GPR.
32122 * config/rs6000/rs6000.md (Fsignbit): Delete.
32123 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
32124 Update the length attribute if the value is in a GPR.
32125 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
32126 the sign or zero extension instruction, since the value is always 0/1.
32127 (signbit<mode>2_dm2): Delete using <Fsignbit>.
32130 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
32131 extracting SImode to a GPR register so that we can generate a
32132 store, limit the vector to be in a traditional Altivec register
32133 for the vextuwrx instruction.
32135 2017-01-03 Ian Lance Taylor <iant@google.com>
32137 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
32139 2017-01-03 Martin Sebor <msebor@redhat.com>
32141 PR tree-optimization/78696
32142 * gimple-ssa-sprintf.c (format_floating): Correct handling of
32143 precision. Use MPFR for %f for greater fidelity. Correct handling
32145 (pass_sprintf_length::compute_format_length): Set width and precision
32146 specified by asrerisk to void_node for vararg functions.
32147 (try_substitute_return_value): Adjust dump output.
32149 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
32151 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
32153 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
32155 * doc/invoke.texi (SPARC options): Document -mlra as the default.
32156 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
32157 -mlra/-mno-lra was passed to the compiler.
32159 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
32161 PR rtl-optimization/65618
32162 * emit-rtl.c (try_split): Move initialization of "before" and
32163 "after" to just before the call to emit_insn_after_setloc.
32165 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
32167 * doc/md.texi (Standard Names): Remove reference to Java frontend.
32169 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
32171 * dwarf2out.c (gen_enumeration_type_die): When
32172 -gno-strict-dwarf, add a DW_AT_encoding attribute.
32174 2017-01-03 Jakub Jelinek <jakub@redhat.com>
32176 PR tree-optimization/78965
32177 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
32178 Change first argument from const call_info & to call_info &. For %n
32179 set info.nowrite to false.
32181 PR middle-end/78901
32182 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
32183 possibly throwing calls.
32185 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
32186 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
32187 and fns handling, rather than in a separate case SSA_NAME.
32189 2017-01-02 Jeff Law <law@redhat.com>
32191 * config/darwin-driver.c (darwin_driver_init): Const-correctness
32192 fixes for first_period and second_period variables.
32194 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
32197 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
32198 (*insvqi_1): New insn pattern.
32199 (*insvqi_1_mem_rex64): Ditto.
32200 (*insvqi_2): Ditto.
32201 (*insvqi_3): Rename from *insvqi.
32203 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
32205 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
32207 * doc/cfg.texi (Edges): Remove reference to Java.
32208 (Maintaining the CFG): Ditto.
32210 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
32212 PR middle-end/77674
32213 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
32214 transparent aliases.
32216 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
32218 PR middle-end/77484
32219 * predict.def (PRED_CALL): Update hitrate.
32220 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
32221 * predict.c (tree_estimate_probability_bb): Split CALL predictor
32222 into direct/indirect/polymorphic variants.
32224 2017-01-01 Jakub Jelinek <jakub@redhat.com>
32226 Update copyright years.
32228 * gcc.c (process_command): Update copyright notice dates.
32229 * gcov-dump.c (print_version): Ditto.
32230 * gcov.c (print_version): Ditto.
32231 * gcov-tool.c (print_version): Ditto.
32232 * gengtype.c (create_file): Ditto.
32233 * doc/cpp.texi: Bump @copying's copyright year.
32234 * doc/cppinternals.texi: Ditto.
32235 * doc/gcc.texi: Ditto.
32236 * doc/gccint.texi: Ditto.
32237 * doc/gcov.texi: Ditto.
32238 * doc/install.texi: Ditto.
32239 * doc/invoke.texi: Ditto.
32241 Copyright (C) 2017 Free Software Foundation, Inc.
32243 Copying and distribution of this file, with or without modification,
32244 are permitted in any medium without royalty provided the copyright
32245 notice and this notice are preserved.