1 2017-10-27 Richard Biener <rguenther@suse.de>
4 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
5 info when we redirected EH.
7 2017-10-26 Michael Collison <michael.collison@arm.com>
9 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
11 (<optab>_trunchf<GPI:mode>2: New pattern.
12 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
13 * config/aarch64/iterators.md (wv): New mode attribute.
14 (vf, VF): New mode attributes.
15 (vgp, VGP): New mode attributes.
16 (s): Update attribute with SImode and DImode prefixes.
18 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
20 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
21 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
22 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
23 (nios2_option_overide): Initialize it. Don't allow R0-relative
25 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
26 (nios2_symbolic_constant_p): Likewise.
27 (nios2_legitimate_address_p): Likewise.
28 (nios2_r0rel_section_name_p): New.
29 (nios2_symbol_ref_in_r0rel_data_p): New.
30 (nios2_emit_move_sequence): Handle r0rel_constant_p.
31 (r0rel_constant_p): New.
32 (nios2_print_operand_address): Handle r0rel_constant_p.
33 (nios2_cdx_narrow_form_p): Likewise.
34 * config/nios2/nios2.opt (mr0rel-sec=): New option.
35 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
36 (Nios II Options): Document -mr0rel-sec.
38 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
40 * config/nios2/nios2.c: Include xregex.h.
41 (nios2_gprel_sec_regex): New.
42 (nios2_option_overide): Initialize it. Don't allow GP-relative
44 (nios2_small_section_name_p): Check for regex match.
45 * config/nios2/nios2.opt (mgprel-sec=): New option.
46 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
47 (Nios II Options): Document -mgprel-sec.
49 2017-10-26 Jim Wilson <wilson@tuliptree.org>
51 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
53 2017-10-26 Tom de Vries <tom@codesourcery.com>
55 PR tree-optimization/82707
56 * gimple.c (gimple_copy): Fix unsharing of
57 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
59 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
61 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
62 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
63 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
64 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
65 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
66 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
67 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
68 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
69 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
70 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
71 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
72 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
73 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
74 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
75 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
76 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
77 _mm512_mask_cmpunord_ps_mask): New intrinsics.
79 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
81 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
83 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
84 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
85 warning to rs6000.c. Remove the Undocumented flag, since it has
87 (-mabi=ibmlongdouble): Likewise.
88 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
89 already set, set the default format for long double.
90 (rs6000_debug_reg_global): Print whether long double is IBM or
92 (rs6000_option_override_internal): Rework setting long double
93 format. Only warn if the user is changing the long double default
94 and they did not use -Wno-psabi.
95 * doc/invoke.texi (PowerPC options): Update the documentation for
96 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
98 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
99 Alan Hayward <alan.hayward@arm.com>
100 David Sherwood <david.sherwood@arm.com>
102 * rtl.h (wider_subreg_mode): New function.
103 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
104 rather than an unsigned int *.
105 * ira-color.c (regno_max_ref_width): Replace with...
106 (regno_max_ref_mode): ...this new variable.
107 (coalesced_pseudo_reg_slot_compare): Update accordingly.
108 Use wider_subreg_mode.
109 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
110 rather than an unsigned int *.
111 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
112 (process_alt_operands): Likewise.
113 (invariant_p): Likewise.
114 * lra-spills.c (assign_mem_slot): Likewise.
115 (add_pseudo_to_slot): Likewise.
116 * lra.c (collect_non_operand_hard_regs): Likewise.
117 (add_regs_to_insn_regno_info): Likewise.
118 * reload1.c (regno_max_ref_width): Replace with...
119 (regno_max_ref_mode): ...this new variable.
120 (reload): Update accordingly. Update call to
121 ira_sort_regnos_for_alter_reg.
122 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
123 (init_eliminable_invariants): Update to use regno_max_ref_mode.
124 (scan_paradoxical_subregs): Likewise.
126 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
128 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
129 (aarch64_frame): Add emit_frame_chain boolean.
130 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
131 Move eh_return case to aarch64_layout_frame.
132 (aarch64_layout_frame): Initialize emit_frame_chain.
133 (aarch64_expand_prologue): Use emit_frame_chain.
135 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
137 * config/aarch64/aarch64.c (aarch64_layout_frame):
138 Ensure LR is always stored at the bottom of the callee-saves.
139 Remove rarely used frame layout which saves callee-saves at top of
140 frame, so the store of LR can be used as a valid probe in all cases.
142 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
144 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
145 Improve unaligned TImode/TFmode base/offset split.
147 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
148 Alan Hayward <alan.hayward@arm.com>
149 David Sherwood <david.sherwood@arm.com>
151 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
152 * combine.c (find_single_use_1): Likewise.
153 (expand_field_assignment): Likewise.
154 (move_deaths): Likewise.
155 * lra-constraints.c (simplify_operand_subreg): Likewise.
156 (curr_insn_transform): Likewise.
157 * lra.c (collect_non_operand_hard_regs): Likewise.
158 (add_regs_to_insn_regno_info): Likewise.
159 * rtlanal.c (reg_referenced_p): Likewise.
160 (covers_regno_no_parallel_p): Likewise.
162 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
164 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
165 Don't print any bits outside the precision of the value.
166 * wide-int.cc (test_printing): Add some new tests.
168 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
170 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
171 supports -xbrace_comment option.
172 * configure: Regenerate.
173 * config.in: Regenerate.
174 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
175 (ASM_CPU_SPEC): Use it.
177 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
179 * target.def (static_rtx_alignment): New hook.
180 * targhooks.h (default_static_rtx_alignment): Declare.
181 * targhooks.c (default_static_rtx_alignment): New function.
182 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
183 * doc/tm.texi: Regenerate.
184 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
185 instead of targetm.constant_alignment. Remove call to
187 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
188 (cris_preferred_mininum_alignment): New function, split out from...
189 (cris_constant_alignment): ...here.
190 (cris_static_rtx_alignment): New function.
191 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
193 (ix86_constant_alignment): ...here.
194 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
195 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
196 (mmix_static_rtx_alignment): New function.
197 * config/spu/spu.c (spu_static_rtx_alignment): New function.
198 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
200 2017-10-26 Tamar Christina <tamar.christina@arm.com>
203 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): Add flag_trapping_math
204 and flag_fp_int_builtin_inexact.
206 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
208 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
209 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
210 mark as a sign-extending load.
211 (local_pic_load_u): Define.
213 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
216 * fold-const.c (operand_equal_for_comparison_p): Also return true
217 if ARG0 is a simple variant of ARG1 with narrower precision.
218 (fold_ternary_loc): Always pass unstripped operands to the predicate.
220 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
222 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
224 * i386.h (processor_costs): Add gather_static, gather_per_elt,
225 scatter_static, scatter_per_elt.
226 * x86-tune-costs.h: Add new cost entries.
228 2017-10-25 Richard Biener <rguenther@suse.de>
230 * tree-ssa-sccvn.h (vn_eliminate): Declare.
231 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
232 class pass_fre): Move to ...
233 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
234 class pass_fre): ... here and adjust for statistics.
236 2017-10-25 Jakub Jelinek <jakub@redhat.com>
239 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
240 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
241 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
242 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
245 2017-10-25 Richard Biener <rguenther@suse.de>
247 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
248 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
249 eliminate_push_avail, eliminate_insert): Move inside...
250 (class eliminate_dom_walker): ... this class in preparation
252 (fini_eliminate): Remove by merging with ...
253 (eliminate): ... this function. Adjust for class changes.
254 (pass_pre::execute): Remove fini_eliminate call.
255 (pass_fre::execute): Likewise.
257 2017-10-24 Jakub Jelinek <jakub@redhat.com>
260 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
261 (VPERMI2, VPERMI2I): New mode iterators.
262 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
263 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
265 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
266 mode iterator. Remove 3 old define_insn patterns.
267 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
268 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
269 VPERMI2 mode iterator, remove the other two expanders.
270 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
271 to use VPERMI2 mode iterator, add another alternative for vpermi2*
272 instructions, remove the other two patterns.
273 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
274 mode iterator, remove the other two patterns.
275 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
276 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
277 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
278 and adjust argument order accordingly.
279 (ix86_expand_vec_perm): Adjust caller.
280 (expand_vec_perm_1): Likewise.
281 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
282 (expand_vec_perm_vpermt2_vpshub2): ... this.
283 (ix86_expand_vec_perm_const_1): Adjust caller.
284 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
287 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
288 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
289 (vec_shl_<mode>): Remove unused expander.
290 (avx512bw_<shift_insn><mode>3): New define_insn.
291 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
292 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
294 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
297 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
300 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
302 PR rtl-optimization/82396
303 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
304 (autopref_multipass_init): Simplify initialization.
305 (autopref_rank_data): Simplify sort order.
306 * gcc/sched-int.h (autopref_multipass_data_): Remove
307 multi_mem_insn_p, min_offset and max_offset.
309 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
312 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
313 Check special value of flag_omit_frame_pointer.
314 (aarch64_can_eliminate): Likewise.
315 (aarch64_override_options_after_change_1): Simplify handling of
316 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
318 2017-10-24 Richard Biener <rguenther@suse.de>
320 PR tree-optimization/82697
321 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
322 zero for conditional load and unconditional store.
324 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
326 * doc/install.texi: Document bootstrap-cet.
328 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
331 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
332 ENDBR instruction at function entrance if function is only
335 2017-10-24 Jakub Jelinek <jakub@redhat.com>
338 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
339 patterns to better describe from which operation the CF is computed.
340 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
341 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
342 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
343 is 0, use _0 suffixed expanders instead of emitting a comparison
346 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
348 * config/i386/i386.md(*movsf_internal, *movdf_internal):
349 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
351 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
354 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
355 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
356 * loop-iv.c (iv_get_reaching_def): Likewise.
357 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
358 variable is promoted and the partition contains undefined values.
360 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
362 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
364 (nios2_address_cost): Define.
365 (nios2_legitimize_address): Recognize (exp + constant) directly.
366 (TARGET_ADDRESS_COST): Define.
368 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
370 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
371 (nios2_symbolic_memory_operand_p): Declare.
372 (nios2_split_large_constant): Declare.
373 (nios2_split_symbolic_memory_operand): Declare.
374 * config/nios2/nios2.c: Adjust includes.
375 (nios2_symbolic_constant_allowed): New.
376 (nios2_symbolic_constant_p): New.
377 (nios2_plus_symbolic_constant_p): New.
378 (nios2_valid_addr_expr_p): Recognize addresses involving
380 (nios2_legitimate_address_p): Likewise, also LO_SUM.
381 (nios2_symbolic_memory_operand_p): New.
382 (nios2_large_constant_p): New.
383 (nios2_split_large_constant): New.
384 (nios2_split_plus_large_constant): New.
385 (nios2_split_symbolic_memory_operand): New.
386 (nios2_legitimize_address): Code refactoring. Handle addresses
387 involving symbolic constants.
388 (nios2_emit_move_sequence): Likewise.
389 (nios2_print_operand): Improve error output.
390 (nios2_print_operand_address): Handle LO_SUM.
391 (nios2_cdx_narrow_form_p): Likewise.
392 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
393 operands involving symbolic constants.
394 (movhi_internal, movsi_internal): Likewise.
395 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
396 (extendhisi2, extendqi<mode>2): Likewise.
398 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
400 * tree-pass.h (PROP_rtl_split_insns): Define.
401 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
403 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
405 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
407 2017-10-23 Jakub Jelinek <jakub@redhat.com>
410 * target.def (const_not_ok_for_debug_p): Default to
411 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
412 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
413 * targhooks.c (default_const_not_ok_for_debug_p): New function.
414 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
415 which targetm.const_not_ok_for_debug_p returned true.
416 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
418 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
420 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
421 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
422 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
424 (ix86_const_not_ok_for_debug_p): New function.
425 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
426 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
428 2017-10-23 David Malcolm <dmalcolm@redhat.com>
431 * system.h: Conditionally include "unique-ptr.h" if
432 INCLUDE_UNIQUE_PTR is defined.
433 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
434 of defining INCLUDE_UNIQUE_PTR before including "system.h".
436 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
438 * config/rl78/rl78.md: New define_expand "subdi3".
440 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
443 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
444 DF_REF_INSN if DF_REF_INSN_INFO is false.
446 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
448 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
449 xmm_move instead of sse_move.
450 (sse_store_index): New function.
451 (ix86_register_move_cost): Be more sensible about mismatch stall;
452 model AVX moves correctly; make difference between sse->integer and
454 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
455 moves; make difference between SSE and AVX.
456 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
457 and zmm_move. Increase size of sse load and store tables;
458 add unaligned load and store tables; add ssemmx_to_integer.
459 * x86-tune-costs.h: Update all entries according to real
460 move latencies from Agner Fog's manual and chip documentation.
462 2017-10-23 Jakub Jelinek <jakub@redhat.com>
465 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
466 * config/i386/constraints.md (Wf): New constraint.
467 * config/i386/i386.md (UNSPEC_SBB): New unspec.
468 (cmp<dwi>_doubleword): Removed.
469 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
470 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
471 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
472 expand with cmp<dwi>_doubleword. For LTU and GEU use
473 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
475 * common.opt (gcolumn-info): Enable by default.
476 * doc/invoke.texi (gcolumn-info): Document new default.
478 2017-10-23 Richard Biener <rguenther@suse.de>
480 PR tree-optimization/82672
481 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
482 Fold the stmt if we propagated into it.
484 2017-10-23 Richard Biener <rguenther@suse.de>
486 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
487 (bitmap_remove_expr_from_set): ... this. All callers call this
488 for non-constant values.
489 (bitmap_set_subtract): Rename to...
490 (bitmap_set_subtract_expressions): ... this. Adjust and
492 (bitmap_set_contains_value): Remove superfluous check.
493 (bitmap_set_replace_value): Inline into single caller ...
494 (bitmap_value_replace_in_set): ... here and simplify.
495 (dependent_clean): Merge into ...
496 (clean): ... this using an overload. Adjust.
497 (prune_clobbered_mems): Adjust.
498 (compute_antic_aux): Likewise.
499 (compute_partial_antic_aux): Likewise.
501 2017-10-23 Richard Biener <rguenther@suse.de>
503 PR tree-optimization/82129
505 2017-08-01 Richard Biener <rguenther@suse.de>
507 PR tree-optimization/81181
508 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
509 (compute_antic): ... end of iteration here.
511 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
513 * target.def (starting_frame_offset): New hook.
514 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
515 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
516 * doc/tm.texi.in: Regenerate.
517 * hooks.h (hook_hwi_void_0): Declare.
518 * hooks.c (hook_hwi_void_0): New function.
519 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
520 STARTING_FRAME_OFFSET.
521 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
522 * reload1.c (reload): Likewise.
523 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
524 instead of STARTING_FRAME_OFFSET.
525 * function.c (try_fit_stack_local): Likewise.
526 (assign_stack_local_1): Likewise
527 (instantiate_virtual_regs): Likewise.
528 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
529 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
530 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
531 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
532 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
533 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
534 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
535 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
536 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
537 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
538 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
539 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
540 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
541 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
542 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
543 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
544 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
545 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
546 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
547 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
548 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
549 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
550 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
551 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
552 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
553 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
554 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
555 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
556 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
557 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
558 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
559 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
560 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
561 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
562 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
563 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
564 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
565 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
566 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
567 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
568 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
569 * config/avr/avr.c (avr_starting_frame_offset): Make static and
570 return a HOST_WIDE_INT.
571 (avr_builtin_setjmp_frame_value): Use it instead of
572 STARTING_FRAME_OFFSET.
573 (TARGET_STARTING_FRAME_OFFSET): Redefine.
574 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
575 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
577 (TARGET_STARTING_FRAME_OFFSET): Redefine.
578 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
579 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
580 (TARGET_CONSTANT_ALIGNMENT): Redefine.
581 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
582 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
583 (TARGET_STARTING_FRAME_OFFSET): Redefine.
584 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
585 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
586 (TARGET_STARTING_FRAME_OFFSET): Redefine.
587 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
588 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
590 (TARGET_STARTING_FRAME_OFFSET): Redefine.
591 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
592 * config/mips/mips.c (mips_compute_frame_info): Refer to
593 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
594 (mips_starting_frame_offset): New function.
595 (TARGET_STARTING_FRAME_OFFSET): Redefine.
596 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
597 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
598 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
599 and return a HOST_WIDE_INT.
600 (TARGET_STARTING_FRAME_OFFSET): Redefine.
601 (mmix_initial_elimination_offset): Refer to
602 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
603 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
604 * config/pa/pa.c (pa_starting_frame_offset): New function.
605 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
606 (pa_expand_prologue): Likewise.
607 (TARGET_STARTING_FRAME_OFFSET): Redefine.
608 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
609 !FRAME_GROWS_DOWNWARD handling to...
610 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
611 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
612 !FRAME_GROWS_DOWNWARD handling to...
613 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
614 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
615 !FRAME_GROWS_DOWNWARD handling to...
616 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
617 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
619 (rs6000_starting_frame_offset): New function.
620 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
621 !FRAME_GROWS_DOWNWARD handling to...
622 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
623 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
624 !FRAME_GROWS_DOWNWARD handling to...
625 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
626 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
627 !FRAME_GROWS_DOWNWARD handling to...
628 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
629 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
630 (rs6000_starting_frame_offset): New function.
631 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
632 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
633 * config/vax/vax.c (vax_starting_frame_offset): New function.
634 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
635 (TARGET_STARTING_FRAME_OFFSET): Redefine.
636 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
637 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
638 (TARGET_STARTING_FRAME_OFFSET): Redefine.
639 * system.h (STARTING_FRAME_OFFSET): Poison.
641 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
643 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
644 SCALAR_TYPE_MODE instead of TYPE_MODE.
646 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
647 Alan Hayward <alan.hayward@arm.com>
648 David Sherwood <david.sherwood@arm.com>
650 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
652 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
653 Alan Hayward <alan.hayward@arm.com>
654 David Sherwood <david.sherwood@arm.com>
656 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
658 2017-10-23 Richard Biener <rguenther@suse.de>
660 PR tree-optimization/82129
661 * tree-ssa-pre.c (bitmap_set_and): Remove.
662 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
663 canonicalizing expressions in the set to those with lowest
664 ID rather than taking that from the first edge.
666 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
668 * combine.c (rtx_equal_for_field_assignment_p): Use
671 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
672 Alan Hayward <alan.hayward@arm.com>
673 David Sherwood <david.sherwood@arm.com>
675 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
676 to a SUBREG_PROMOTED_VAR.
678 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
679 Alan Hayward <alan.hayward@arm.com>
680 David Sherwood <david.sherwood@arm.com>
682 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
683 (expand_debug_source_expr): Likewise.
684 * combine.c (combine_simplify_rtx): Likewise.
685 * cse.c (fold_rtx): Likewise.
686 * optabs.c (expand_float): Likewise.
687 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
688 (simplify_binary_operation_1): Likewise.
690 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
691 Alan Hayward <alan.hayward@arm.com>
692 David Sherwood <david.sherwood@arm.com>
694 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
695 (record_promoted_value): Likewise.
696 * expr.c (expand_expr_real_2): Likewise.
697 * ree.c (update_reg_equal_equiv_notes): Likewise.
698 (combine_set_extension): Likewise.
699 * rtlanal.c (low_bitmask_len): Likewise.
700 * simplify-rtx.c (neg_const_int): Likewise.
701 (simplify_binary_operation_1): Likewise.
703 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
704 Alan Hayward <alan.hayward@arm.com>
705 David Sherwood <david.sherwood@arm.com>
707 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
708 * regcprop.c (maybe_mode_change): Likewise.
709 * reload1.c (alter_reg): Likewise.
711 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
713 * inchash.h (inchash::hash::add_wide_int): New function.
714 * lto-streamer-out.c (hash_tree): Use it.
716 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
718 * inchash.h (inchash::hash::add_wide_int): Rename to...
719 (inchash::hash::add_hwi): ...this.
720 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
721 (polymorphic_call_target_hasher::hash): Likewise.
722 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
723 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
724 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
725 * lto-streamer-out.c (hash_tree): Likewise.
726 * optc-save-gen.awk: Likewise.
727 * tree.c (add_expr): Likewise.
729 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
732 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
733 for ordered inequality comparisons even with TARGET_IEEE_FP.
735 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
738 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
739 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
740 Expand with cmp<dwi>_doubleword.
742 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
744 * extend.texi: Add x86 specific to 'nocf_check' attribute.
746 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
747 specific to -fcf-protection option.
749 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
751 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
752 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
753 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
754 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
755 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
756 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
757 (extra_objs): Add cet.o for Linux/x86 targets.
758 (tmake_file): Add i386/t-cet for Linux/x86 targets.
759 * config/i386/cet.c: New file.
760 * config/i386/cetintrin.h: Likewise.
761 * config/i386/t-cet: Likewise.
762 * config/i386/cpuid.h (bit_SHSTK): New.
764 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
765 pass IBT and SHSTK bits.
766 * config/i386/i386-builtin-types.def
767 (VOID_FTYPE_UNSIGNED_PVOID): New.
768 (VOID_FTYPE_UINT64_PVOID): Likewise.
769 * config/i386/i386-builtin.def: Add CET intrinsics.
770 * config/i386/i386-c.c (ix86_target_macros_internal): Add
771 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
772 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
773 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
775 * config/i386/i386.c (rest_of_insert_endbranch): New.
776 (pass_data_insert_endbranch): Likewise.
777 (pass_insert_endbranch): Likewise.
778 (make_pass_insert_endbranch): Likewise.
779 (ix86_notrack_prefixed_insn_p): Likewise.
780 (ix86_target_string): Add -mibt, -mshstk flags.
781 (ix86_option_override_internal): Add flag_cf_protection
783 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
784 (ix86_print_operand): Add 'notrack' prefix output.
785 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
786 (ix86_expand_builtin): Expand CET intrinsics.
787 (x86_output_mi_thunk): Add 'endbranch' instruction.
788 * config/i386/i386.h (TARGET_IBT): New.
789 (TARGET_IBT_P): Likewise.
790 (TARGET_SHSTK): Likewise.
791 (TARGET_SHSTK_P): Likewise.
792 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
793 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
794 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
795 (builtin_setjmp_setup): New pattern.
796 (builtin_longjmp): Likewise.
797 (rdssp<mode>): Likewise.
798 (incssp<mode>): Likewise.
799 (saveprevssp): Likewise.
800 (rstorssp): Likewise.
801 (wrss<mode>): Likewise.
802 (wruss<mode>): Likewise.
803 (setssbsy): Likewise.
804 (clrssbsy): Likewise.
805 (nop_endbr): Likewise.
806 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
808 * config/i386/immintrin.h: Include <cetintrin.h>.
809 * config/i386/linux-common.h
810 (file_end_indicate_exec_stack_and_cet): New prototype.
811 (TARGET_ASM_FILE_END): New.
813 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
815 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
816 latencies instead of having separate table; make difference between
817 integer and float costs.
818 * i386.h (processor_costs): Remove scalar_stmt_cost,
819 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
820 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
822 * x86-tune-costs.h: Remove entries which has been removed in
823 procesor_costs from all tables; make cond_taken_branch_cost
824 and cond_not_taken_branch_cost COST_N_INSNS based.
826 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
828 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
830 2017-10-20 Jakub Jelinek <jakub@redhat.com>
832 * config/i386/i386.md (isa): Remove fma_avx512f.
833 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
834 <avx512>_fmadd_<mode>_mask3<round_name>,
835 <avx512>_fmsub_<mode>_mask<round_name>,
836 <avx512>_fmsub_<mode>_mask3<round_name>,
837 <avx512>_fnmadd_<mode>_mask<round_name>,
838 <avx512>_fnmadd_<mode>_mask3<round_name>,
839 <avx512>_fnmsub_<mode>_mask<round_name>,
840 <avx512>_fnmsub_<mode>_mask3<round_name>,
841 <avx512>_fmaddsub_<mode>_mask<round_name>,
842 <avx512>_fmaddsub_<mode>_mask3<round_name>,
843 <avx512>_fmsubadd_<mode>_mask<round_name>,
844 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
845 (*vec_widen_umult_even_v16si<mask_name>,
846 *vec_widen_smult_even_v16si<mask_name>): Likewise.
847 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
849 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
851 * extend.texi: Add 'nocf_check' documentation.
852 * gimple.texi: Add second parameter to
853 gimple_build_call_from_tree.
854 * invoke.texi: Add -fcf-protection documentation.
855 * rtl.texi: Add REG_CALL_NOTRACK documenation.
857 2017-10-20 Richard Biener <rguenther@suse.de>
859 PR tree-optimization/82473
860 * tree-vect-loop.c (vectorizable_reduction): Properly get at
861 the largest input type.
863 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
865 * c-attribs.c (handle_nocf_check_attribute): New function.
866 (c_common_attribute_table): Add 'nocf_check' handling.
867 * gimple-parser.c: Add second argument NULL to
868 gimple_build_call_from_tree.
869 * attrib.c (comp_type_attributes): Check nocf_check attribute.
870 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
872 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
873 * common.opt: Add fcf-protection flag.
874 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
875 * flag-types.h: Add enum cf_protection_level.
876 * gimple.c (gimple_build_call_from_tree): Add second parameter.
877 Add 'nocf_check' attribute propagation to gimple call.
878 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
879 (gimple_build_call_from_tree): Update prototype.
880 (gimple_call_nocf_check_p): New function.
881 (gimple_call_set_nocf_check): Likewise.
882 * gimplify.c: Add second argument to gimple_build_call_from_tree.
883 * ipa-icf.c: Add nocf_check attribute in statement hash.
884 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
885 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
886 * toplev.c (process_options): Add flag_cf_protection handling.
888 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
890 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
892 2017-10-20 Richard Biener <rguenther@suse.de>
894 PR tree-optimization/82603
895 * tree-if-conv.c (predicate_mem_writes): Make sure to only
896 remove false predicated stores.
898 2017-10-20 Richard Biener <rguenther@suse.de>
900 * graphite-isl-ast-to-gimple.c
901 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
902 Remove return value and simplify, dump copied stmt after lhs
904 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
905 Reduce dump verbosity.
906 (gsi_insert_earliest): Likewise.
907 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
908 * graphite.c (print_global_statistics): Adjust dumping.
909 (print_graphite_scop_statistics): Likewise.
910 (print_graphite_statistics): Do not dump loops here.
911 (graphite_transform_loops): But here.
913 2017-10-20 Nicolas Roche <roche@adacore.com>
915 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
916 * configure: Regenerate.
918 2017-10-20 Jakub Jelinek <jakub@redhat.com>
921 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
922 functions when optimizing replace GIMPLE_RETURN stmts with
923 calls to __builtin_unreachable ().
926 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
927 for -fsanitize=thread link of executables.
928 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
932 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
934 (<shift_insn><mode>3<mask_name>): Change the last of the 3
935 define_insns for logical vector shifts to use VI248_AVX512BW
936 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
937 condition, useless isa and prefix attributes. Change the first
938 2 of these define_insns to ...
939 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
940 define_insn for avx512vl.
941 (<shift_insn><mode>3): ... and this, new define_insn without
942 masking for non-avx512vl.
945 * config/i386/sse.md (*andnot<mode>3,
946 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
947 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
948 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
949 not applied use empty suffix even for TARGET_AVX512VL.
950 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
951 is applied, supply evex,evex or evex,evex,evex instead of just
954 2017-10-20 Julia Koval <julia.koval@intel.com>
956 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
957 (OPTION_MASK_ISA_GFNI_UNSET): New.
958 (ix86_handle_option): Handle OPT_mgfni.
959 * config/i386/cpuid.h (bit_GFNI): New.
960 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
961 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
962 * config/i386/i386.c (ix86_target_string): Add -mgfni.
963 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
964 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
965 * config/i386/i386.opt: Add mgfni.
967 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
969 * config/msp430/msp430.c (msp430_option_override): Disable
970 -fdelete-null-pointer-checks.
971 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
973 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
975 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
976 of x87 and SSE instructions.
978 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
980 * asan.c (create_cond_insert_point): Do not update edge count.
981 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
982 (afdo_propagate_circuit): Likewise.
983 (afdo_calculate_branch_prob): Likewise.
984 (afdo_annotate_cfg): Likewise.
985 * basic-block.h (struct edge_def): Remove count.
986 (edge_def::count): New accessor.
987 * bb-reorder.c (rotate_loop): Update.
988 (find_traces_1_round): Update.
989 (connect_traces): Update.
990 (sanitize_hot_paths): Update.
991 * cfg.c (unchecked_make_edge): Update.
992 (make_single_succ_edge): Update.
993 (check_bb_profile): Update.
994 (dump_edge_info): Update.
995 (update_bb_profile_for_threading): Update.
996 (scale_bbs_frequencies_int): Update.
997 (scale_bbs_frequencies_gcov_type): Update.
998 (scale_bbs_frequencies_profile_count): Update.
999 (scale_bbs_frequencies): Update.
1000 * cfganal.c (connect_infinite_loops_to_exit): Update.
1001 * cfgbuild.c (compute_outgoing_frequencies): Update.
1002 (find_many_sub_basic_blocks): Update.
1003 * cfgcleanup.c (try_forward_edges): Update.
1004 (try_crossjump_to_edge): Update
1005 * cfgexpand.c (expand_gimple_cond): Update
1006 (expand_gimple_tailcall): Update
1007 (construct_exit_block): Update
1008 * cfghooks.c (verify_flow_info): Update
1009 (redirect_edge_succ_nodup): Update
1010 (split_edge): Update
1011 (make_forwarder_block): Update
1012 (duplicate_block): Update
1013 (account_profile_record): Update
1014 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
1015 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
1016 * cfgloopmanip.c (scale_loop_profile): Update.
1018 (lv_adjust_loop_entry_edge): Update.
1019 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
1020 (force_nonfallthru_and_redirect): Update.
1021 (purge_dead_edges): Update.
1022 (rtl_flow_call_edges_add): Update.
1023 * cgraphunit.c (init_lowered_empty_function): Update.
1024 (cgraph_node::expand_thunk): Update.
1025 * gimple-pretty-print.c (dump_probability): Update.
1026 (dump_edge_probability): Update.
1027 * gimple-ssa-isolate-paths.c (isolate_path): Update.
1028 * haifa-sched.c (sched_create_recovery_edges): Update.
1029 * hsa-gen.c (convert_switch_statements): Update.
1030 * ifcvt.c (dead_or_predicable): Update.
1031 * ipa-inline-transform.c (inline_transform): Update.
1032 * ipa-split.c (split_function): Update.
1033 * ipa-utils.c (ipa_merge_profiles): Update.
1034 * loop-doloop.c (add_test): Update.
1035 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
1036 * lto-streamer-in.c (input_cfg): Update.
1037 (input_function): Update.
1038 * lto-streamer-out.c (output_cfg): Update.
1039 * modulo-sched.c (sms_schedule): Update.
1040 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
1041 * predict.c (maybe_hot_edge_p): Update.
1042 (unlikely_executed_edge_p): Update.
1043 (probably_never_executed_edge_p): Update.
1044 (dump_prediction): Update.
1045 (drop_profile): Update.
1046 (propagate_unlikely_bbs_forward): Update.
1047 (determine_unlikely_bbs): Update.
1048 (force_edge_cold): Update.
1049 * profile.c (compute_branch_probabilities): Update.
1050 * reg-stack.c (better_edge): Update.
1051 * shrink-wrap.c (handle_simple_exit): Update.
1052 * tracer.c (better_p): Update.
1053 * trans-mem.c (expand_transaction): Update.
1054 (split_bb_make_tm_edge): Update.
1055 * tree-call-cdce.c: Update.
1056 * tree-cfg.c (gimple_find_sub_bbs): Update.
1057 (gimple_split_edge): Update.
1058 (gimple_duplicate_sese_region): Update.
1059 (gimple_duplicate_sese_tail): Update.
1060 (gimple_flow_call_edges_add): Update.
1061 (insert_cond_bb): Update.
1062 (execute_fixup_cfg): Update.
1063 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
1064 * tree-complex.c (expand_complex_div_wide): Update.
1065 * tree-eh.c (lower_resx): Update.
1066 (unsplit_eh): Update.
1067 (cleanup_empty_eh_move_lp): Update.
1068 * tree-inline.c (copy_edges_for_bb): Update.
1069 (freqs_to_counts): Update.
1070 (copy_cfg_body): Update.
1071 * tree-ssa-dce.c (remove_dead_stmt): Update.
1072 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
1073 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
1074 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
1075 (unloop_loops): Update.
1076 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
1077 * tree-ssa-loop-split.c (connect_loops): Update.
1078 (split_loop): Update.
1079 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
1080 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
1081 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
1082 * tree-ssa-reassoc.c (branch_fixup): Update.
1083 * tree-ssa-tail-merge.c (replace_block_by): Update.
1084 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
1085 (compute_path_counts): Update.
1086 (update_profile): Update.
1087 (recompute_probabilities): Update.
1088 (update_joiner_offpath_counts): Update.
1089 (estimated_freqs_path): Update.
1090 (freqs_to_counts_path): Update.
1091 (clear_counts_path): Update.
1092 (ssa_fix_duplicate_block_edges): Update.
1093 (duplicate_thread_path): Update.
1094 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
1095 (case_bit_test_cmp): Update.
1096 (collect_switch_conv_info): Update.
1097 (gen_inbound_check): Update.
1098 (do_jump_if_equal): Update.
1099 (emit_cmp_and_jump_insns): Update.
1100 * tree-tailcall.c (decrease_profile): Update.
1101 (eliminate_tail_call): Update.
1102 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
1103 (vect_do_peeling): Update.
1104 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
1105 * ubsan.c (ubsan_expand_null_ifn): Update.
1106 (ubsan_expand_ptr_ifn): Update.
1107 * value-prof.c (gimple_divmod_fixed_value): Update.
1108 (gimple_mod_pow2): Update.
1109 (gimple_mod_subtract): Update.
1110 (gimple_ic): Update.
1111 (gimple_stringop_fixed_value): Update.
1113 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
1116 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
1118 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
1120 PR rtl-optimization/82395
1121 * ira-color.c (allocno_priority_compare_func): Fix comparison step
1122 based on non_spilled_static_chain_regno_p.
1124 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
1126 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
1127 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
1128 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
1130 2017-10-19 Martin Sebor <msebor@redhat.com>
1132 PR tree-optimization/82596
1133 * tree.c (array_at_struct_end_p): Handle STRING_CST.
1135 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
1137 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
1138 (get_mem_refs_of_builtin_call): Likewise.
1139 * builtins.c (expand_builtin_apply): Adjust call to
1140 allocate_dynamic_stack_space.
1141 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
1142 the third argument to allocate_dynamic_stack_space, otherwise -1.
1143 (expand_builtin): Deal with all alloca variants.
1144 (is_inexpensive_builtin): Likewise.
1145 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
1146 * calls.c (special_function_p): Deal with all alloca variants.
1147 (initialize_argument_information): Adjust call to
1148 allocate_dynamic_stack_space.
1149 (expand_call): Likewise.
1150 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
1151 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
1152 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
1153 use it for the stack usage computation.
1154 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
1155 * function.c (gimplify_parameters): Call build_alloca_call_expr.
1156 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
1157 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
1158 (in_loop_p): Remove first argument and useless check.
1159 (pass_walloca::execute): Remove useless test and adjust call to above.
1160 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
1161 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
1162 (gimplify_call_expr): Deal with all alloca variants.
1163 * hsa-gen.c (gen_hsa_alloca): Likewise.
1164 (gen_hsa_insns_for_call): Likewise.
1165 * ipa-pure-const.c (special_builtin_state): Likewise.
1166 * tree-chkp.c (chkp_build_returned_bound): Likewise.
1167 * tree-object-size.c (alloc_object_size): Likewise.
1168 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
1169 (call_may_clobber_ref_p_1): Likewise.
1170 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
1171 (ccp_fold_stmt): Likewise.
1172 (optimize_stack_restore): Likewise.
1173 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
1174 (mark_all_reaching_defs_necessary_1): Likewise.
1175 (propagate_necessity): Likewise.
1176 (eliminate_unnecessary_stmts): Likewise.
1177 * tree.c (build_common_builtin_nodes): Build
1178 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
1179 (build_alloca_call_expr): New function.
1180 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
1181 (CASE_BUILT_IN_ALLOCA): Likewise.
1182 (build_alloca_call_expr): Declare.
1183 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
1185 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
1188 * dwarf2out.c (new_die_raw): New static inline function.
1189 (new_die): Use it to create the DIE.
1190 (add_AT_external_die_ref): Likewise.
1191 (clone_die): Likewise.
1192 (clone_as_declaration): Likewise.
1193 (dwarf2out_vms_debug_main_pointer): Likewise.
1194 (base_type_die): Likewise. Remove early return for corner cases.
1195 Do not call add_pubtype on the DIE here.
1196 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
1197 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
1198 typedefs for base types with DW_AT_endianity. Make sure a DIE with
1199 native order exists for base types, attach the DIE manually and call
1200 add_pubtype on it. Do not equate a reverse order DIE to the type.
1202 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
1204 * config/arm/arm.c (align_ok_ldrd_strd): New function.
1205 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
1207 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
1209 2017-10-19 Jakub Jelinek <jakub@redhat.com>
1211 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
1212 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
1213 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
1214 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
1215 * opts.c (sanitizer_opts): Add builtin.
1216 * ubsan.c (instrument_builtin): New function.
1217 (pass_ubsan::execute): Call it.
1218 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
1219 * doc/invoke.texi: Document -fsanitize=builtin.
1221 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
1222 builtins, store max (log2 (align), 0) into uchar field instead of
1223 align into uptr field.
1224 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
1225 store uchar 0 field instead of uptr 0 field.
1226 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
1227 instead of passing one address of struct with 2 locations pass
1228 two addresses of structs with 1 location each.
1229 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
1230 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
1231 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
1232 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
1233 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
1234 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
1235 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
1236 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
1238 2017-10-19 Martin Liska <mliska@suse.cz>
1241 * file-find.c (remove_prefix): Remove.
1242 * file-find.h (remove_prefix): Likewise.
1243 * gcc-ar.c: Remove smartness of lookup.
1245 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
1247 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
1248 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
1249 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
1251 2017-10-19 Jakub Jelinek <jakub@redhat.com>
1254 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
1255 (setcc + and to xor + setcc): New peephole2.
1257 2017-10-19 Tom de Vries <tom@codesourcery.com>
1259 * doc/sourcebuild.texi (Test Directives, Variants of
1260 dg-require-support): Add dg-require-stack-size.
1262 2017-10-19 Martin Liska <mliska@suse.cz>
1265 * gimplify.c (gimplify_decl_expr): Do not instrument variables
1266 that have a large alignment.
1267 (gimplify_target_expr): Likewise.
1269 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
1271 PR rtl-optimization/82602
1272 * ira.c (rtx_moveable_p): Return false for volatile asm.
1274 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
1277 * config/i386/i386-modes.def (CCGZ): New CC mode.
1278 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
1279 * config/i386/predicates.md (ix86_comparison_operator):
1281 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
1282 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
1283 with double-word subtraction.
1284 (put_condition_code): Handle CCGZmode.
1286 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
1288 * wide-int.cc (debug (const wide_int &)): New.
1289 (debug (const wide_int *)): New.
1290 (debug (const widest_int &)): New.
1291 (debug (const widest_int *)): New.
1293 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
1296 * lra-constraints.c (curr_insn_transform): Use non-input operand
1297 instead of output one for matched reload.
1299 2017-10-18 Bin Cheng <bin.cheng@arm.com>
1301 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
1302 (tree-ssa-loop-ivopts.h): New header file.
1303 (struct builtin_info): New fields.
1304 (classify_builtin_1): Compute and record base and offset parts for
1305 memset builtin partition by calling strip_offset.
1306 (offset_cmp, fuse_memset_builtins): New functions.
1307 (finalize_partitions): Fuse adjacent memset partitions by calling
1309 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
1310 Expose the interface.
1311 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
1313 2017-10-18 Bin Cheng <bin.cheng@arm.com>
1315 PR tree-optimization/82574
1316 * tree-loop-distribution.c (find_single_drs): New parameter. Check
1317 that data reference must be executed exactly once per iteration
1318 against the outermost loop in nest.
1319 (classify_partition): Update call to above function.
1321 2017-10-18 Richard Biener <rguenther@suse.de>
1323 PR tree-optimization/82591
1324 * graphite.c (graphite_transform_loops): Move code gen message
1326 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
1327 Here. Handle scop_to_isl_ast failing.
1328 (scop_to_isl_ast): Limit the number of ISL operations.
1330 2017-10-18 Richard Biener <rguenther@suse.de>
1332 * graphite-isl-ast-to-gimple.c
1333 (translate_isl_ast_to_gimple::set_rename): Simplify.
1334 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
1335 (graphite_copy_stmts_from_block): ... here.
1336 (copy_bb_and_scalar_dependences): Simplify.
1337 (add_parameters_to_ivs_params): Canonicalize.
1338 (generate_entry_out_of_ssa_copies): Simplify.
1339 * graphite-sese-to-poly.c (extract_affine_name): Simplify
1340 by passing in ISL dimension.
1341 (parameter_index_in_region_1): Rename to ...
1342 (parameter_index_in_region): ... this.
1343 (extract_affine): Adjust assert, pass down parameter index.
1344 (add_param_constraints): Use range-info when available.
1345 (build_scop_context): Adjust.
1346 * sese.c (new_sese_info): Adjust.
1347 (free_sese_info): Likewise.
1348 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
1349 Remove unused typedefs.
1350 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
1352 2017-10-18 Martin Liska <mliska@suse.cz>
1354 * combine.c (simplify_compare_const): Add gcc_fallthrough.
1356 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
1358 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
1359 (s390_sched_init): Do not reset s390_sched_state if we entered the
1360 current basic block via a fallthru edge and all others are unlikely.
1362 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
1364 * config/s390/s390.c (NUM_SIDES): New variable.
1365 (LONGRUNNING_THRESHOLD): New variable.
1366 (LATENCY_FACTOR): New variable.
1367 (s390_sched_score): Decrease score for long-running instructions on
1369 (s390_sched_variable_issue): Perform bookkeeping for long-running
1372 2017-10-18 Richard Biener <rguenther@suse.de>
1374 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
1375 Simplify with removal of the parameter rename map.
1376 (set_rename): Likewise.
1377 (should_copy_to_new_region): Likewise.
1378 (graphite_copy_stmts_from_block): Likewise.
1379 (copy_bb_and_scalar_dependences): Remove initialization of
1380 unused copied_bb_map.
1382 (copy_internal_parameters): Likewise.
1383 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
1384 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
1385 Use INTEGRAL_TYPE_P.
1386 (parameter_index_in_region_1): Rename to ...
1387 (assign_parameter_index_in_region): ... this. Assert we have
1388 a parameter we handle.
1389 (scan_tree_for_params): Adjust.
1390 * sese.h (parameter_rename_map_t): Remove.
1391 (struct sese_info_t): Remove unused parameter_rename_map and
1392 copied_bb_map members.
1393 * sese.c (new_sese_info): Adjust.
1394 (free_sese_info): Likewise.
1396 2017-10-18 Martin Liska <mliska@suse.cz>
1399 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
1400 on an abnormal edge.
1402 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
1404 * doc/invoke.texi (ffunction-sections and fdata-sections):
1407 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
1409 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
1410 the use statement can throw internally.
1412 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
1414 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
1415 any RTX present on the RHS of a SET.
1416 * compare-elim.c (try_eliminate_compare): Restore comment.
1418 2017-10-17 Jakub Jelinek <jakub@redhat.com>
1420 * langhooks.h (struct lang_hooks): Document that tree_size langhook
1421 may be also called on tcc_type nodes.
1422 * langhooks.c (lhd_tree_size): Likewise.
1424 2017-10-17 David Malcolm <dmalcolm@redhat.com>
1426 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
1427 format_warning_at_substring.
1428 (maybe_warn): Convert source_range * param to a location_t. Pass
1429 UNKNOWN_LOCATION rather than NULL to fmtwarn.
1430 (format_directive): Remove code to extract source_ranges and
1431 source_range * in favor of just a location_t.
1432 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
1434 * substring-locations.c (format_warning_va): Convert
1435 source_range * param to a location_t.
1436 (format_warning_at_substring): Likewise.
1437 * substring-locations.h (format_warning_va): Likewise.
1438 (format_warning_at_substring): Likewise.
1440 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
1442 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
1444 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
1445 and scatter/gather ops.
1447 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
1448 vec_gather_load and vec_scatter_store.
1449 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
1450 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
1451 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
1452 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
1453 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
1454 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
1456 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
1458 * reg-stack.c (compare_for_stack_reg): Add bool argument.
1459 Detect FTST instruction and handle its register pops. Only pop
1460 second operand if can_pop_second_op is true.
1461 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
1462 set can_pop_second_op to false in the compare_for_stack_reg call.
1464 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
1465 output_fp_compare for stack register operands.
1466 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
1467 instructions here. Do not emit stack register pops here. Assert
1468 that FCOMPP pops next to top stack register. Rewrite function.
1470 2017-10-17 Nathan Sidwell <nathan@acm.org>
1473 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
1474 use DECL_ASSEMBLER_NAME_RAW.
1477 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
1480 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
1481 Wilco Dijkstra <wilco.dijkstra@arm.com>
1483 * builtins.c (expand_builtin_update_setjmp_buf): Add a
1484 converstion to Pmode from the buf_addr.
1486 2017-10-17 Richard Biener <rguenther@suse.de>
1488 * graphite-dependences.c (scop_get_reads_and_writes): Change
1489 output parameters to references.
1491 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
1493 PR 71026/tree-optimization
1494 * fold-const.c (distribute_real_division): Removed.
1495 (fold_binary_loc): Remove calls to distribute_real_divison.
1497 2017-10-17 Richard Biener <rguenther@suse.de>
1499 * graphite-scop-detection.c
1500 (scop_detection::stmt_has_simple_data_refs_p): Always use
1501 the full nest as region.
1502 (try_generate_gimple_bb): Likewise.
1503 * sese.c (scalar_evolution_in_region): Simplify now that
1504 SCEV can handle instantiation in regions.
1505 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
1506 in the non-loop part of a function if requested.
1508 2017-10-17 Richard Biener <rguenther@suse.de>
1510 PR tree-optimization/82563
1511 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
1513 (graphite_regenerate_ast_isl): Call it.
1514 * graphite-scop-detection.c (build_scops): Remove entry edge split.
1516 2017-10-17 Jakub Jelinek <jakub@redhat.com>
1518 PR tree-optimization/82549
1519 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
1520 Formatting fixes. Instead of calling make_bit_field_ref with negative
1523 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
1525 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
1526 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
1527 _mm_maskz_reduce_ss): New.
1528 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
1529 __builtin_ia32_reducess_mask): Ditto..
1530 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
1531 * config/i386/sse.md (reduces<mode>): Renamed to ...
1532 (reduces<mode><mask_scalar_name>): ... this.
1533 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}): Changed
1535 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
1536 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
1538 2017-10-16 David Malcolm <dmalcolm@redhat.com>
1540 * Makefile.in (OBJS): Add unique-ptr-tests.o.
1541 * selftest-run-tests.c (selftest::run_tests): Call
1542 selftest::unique_ptr_tests_cc_tests.
1543 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
1544 * unique-ptr-tests.cc: New file.
1546 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
1549 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
1551 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
1552 (make_hard_regno_born, make_hard_regno_dead): Update
1553 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
1555 2017-10-16 Jeff Law <law@redhat.com>
1557 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
1559 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
1561 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
1563 2017-10-16 Olivier Hainque <hainque@adacore.com>
1565 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
1566 with_cpu if we were configured for an e500v2 target cpu name.
1568 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
1570 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
1571 * doc/invoke.texi: Document +nodsp as a valid extension for
1574 2017-10-16 Martin Liska <mliska@suse.cz>
1576 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
1577 (test_set_range): Likewise.
1578 (test_range_functions): Rename to ...
1579 (test_bit_in_range): ... this.
1580 (sbitmap_c_tests): Add new test.
1582 2017-10-16 Tamar Christina <tamar.christina@arm.com>
1584 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32): New.
1585 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
1586 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
1588 2017-10-16 Tamar Christina <tamar.christina@arm.com>
1590 * config/aarch64/aarch64-builtins.c
1591 (aarch64_types_quadopu_lane_qualifiers): New.
1592 (TYPES_QUADOPU_LANE): New.
1593 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
1594 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
1595 (aarch64_<sur>dot_laneq<vsi2qi>): New.
1596 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
1597 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
1598 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
1599 (Vdottype, DOTPROD): New.
1600 (sur): Add SDOT and UDOT.
1602 2017-10-16 Tamar Christina <tamar.christina@arm.com>
1604 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
1605 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
1606 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Add TARGET_DOTPROD.
1607 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
1608 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75): Enable TARGET_DOTPROD.
1609 (cortex-a75.cortex-a55): Likewise.
1610 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
1612 2017-10-16 Tamar Christina <tamar.christina@arm.com>
1614 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
1615 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS): New.
1616 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane): new.
1617 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
1618 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
1619 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
1620 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
1621 * config/arm/types.md (neon_dot, neon_dot_q): New.
1622 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
1624 2017-10-16 Tamar Christina <tamar.christina@arm.com>
1626 * config/arm/arm.h (TARGET_DOTPROD): New.
1627 * config/arm/arm.c (arm_arch_dotprod): New.
1628 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
1629 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
1630 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
1631 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
1632 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
1633 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
1634 * doc/invoke.texi (armv8.2-a): Document dotprod
1636 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
1638 * i386.c (ix86_vec_cost): New function.
1639 (ix86_rtx_costs): Handle vector operations better.
1640 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
1641 * x86-tune-costs.h: Add new costs to all tables.
1643 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
1645 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
1647 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
1648 divsd, sqrtss and sqrtsd
1649 * x86-tune-costs.h: Add new entries to all costs.
1650 (znver1_cost): Fix to match real instruction latencies.
1652 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1653 Michael Collison <michael.collison@arm.com>
1655 * compare-elim.c: Include emit-rtl.h.
1656 (can_merge_compare_into_arith): New function.
1657 (try_validate_parallel): Likewise.
1658 (try_merge_compare): Likewise.
1659 (try_eliminate_compare): Call the above when no previous clobber
1661 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
1664 2017-10-14 Jakub Jelinek <jakub@redhat.com>
1668 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
1669 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
1673 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
1674 to be any operand_equal_p operands. For & (B - 1) require
1675 B to be power of 2. Recognize
1676 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
1678 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
1681 * optabs.c (expand_memory_blockage): Fix call of
1682 targetm.have_memory_blockage.
1684 2017-10-14 Jakub Jelinek <jakub@redhat.com>
1687 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
1688 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
1689 objects to extra_objs instead of overwriting it.
1691 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
1693 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
1694 Use any_fp_register_operand as operand[3] predicate. Simplify
1695 equality test for operands[2] and operands[4] memory location.
1696 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
1697 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
1698 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
1699 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
1700 any_fp_register_operand as operand[1] predicate. Simplify
1701 equality test for operands[0] and operands[3] memory location.
1702 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
1703 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
1704 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
1706 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
1708 * target-insns.def: Add memory_blockage.
1709 * optabs.c (expand_memory_blockage): New function.
1710 (expand_asm_memory_barrier): Rename ...
1711 (expand_asm_memory_blockage): ... to this.
1712 (expand_mem_thread_fence): Call expand_memory_blockage
1713 instead of expand_asm_memory_barrier.
1714 (expand_mem_singnal_fence): Ditto.
1715 (expand_atomic_load): Ditto.
1716 (expand_atomic_store): Ditto.
1717 * doc/md.texi (Standard Pattern Names For Generation):
1718 Document memory_blockage instruction pattern.
1720 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
1722 * config/rl78/rl78.c (rl78_emit_libcall): New function.
1723 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
1724 * config/rl78/rl78.md: New define_expand "adddi3".
1726 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
1728 * cfghooks.c (verify_flow_info): Disable check that all probabilities
1731 2017-10-13 Jeff Law <law@redhat.com>
1733 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
1735 2017-10-13 Jakub Jelinek <jakub@redhat.com>
1738 * internal-fn.c (expand_mul_overflow): If both operands have
1739 the same highpart of -1 or 0 and the topmost bit of lowpart
1740 is different, overflow is if res <= 0 rather than res < 0.
1742 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
1744 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
1745 TARGET_P9_VECTOR code for unaligned_load case.
1747 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
1749 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
1751 2017-10-13 Nathan Sidwell <nathan@acm.org>
1753 * tree-core.h (tree_contains_struct): Make bool.
1754 * tree.c (tree_contains_struct): Likewise.
1755 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
1756 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
1757 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
1758 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
1759 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
1761 2017-10-13 Richard Biener <rguenther@suse.de>
1763 * graphite-isl-ast-to-gimple.c
1764 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
1765 parameters and dominance check.
1766 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
1767 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
1768 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
1769 Do not update SSA form here or do intermediate IL verification.
1770 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
1771 (graphite_initialize): Remove check on the number of loops in
1772 the function and inline into graphite_transform_loops.
1773 (graphite_finalize): Inline into graphite_transform_loops.
1774 (graphite_transform_loops): Perform SSA update and IL verification
1776 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
1778 2017-10-13 Richard Biener <rguenther@suse.de>
1780 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
1781 graphite_expression_type_precision): Avoid global constructor
1783 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
1784 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
1785 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
1786 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
1787 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
1788 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
1790 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
1793 * config/i386/i386.h (ix86_red_zone_size): New.
1794 * config/i386/i386.md (push peephole2s): Replace
1795 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
1797 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
1798 Alan Hayward <alan.hayward@arm.com>
1799 David Sherwood <david.sherwood@arm.com>
1801 * combine.c (can_change_dest_mode): Reject changes in
1802 REGMODE_NATURAL_SIZE.
1804 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
1805 Alan Hayward <alan.hayward@arm.com>
1806 David Sherwood <david.sherwood@arm.com>
1808 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
1809 (expand_debug_source_expr): Likewise.
1810 * combine.c (combine_simplify_rtx): Likewise.
1811 * cse.c (fold_rtx): Likewise.
1812 * fwprop.c (canonicalize_address): Likewise.
1813 * targhooks.c (default_shift_truncation_mask): Likewise.
1815 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
1816 Alan Hayward <alan.hayward@arm.com>
1817 David Sherwood <david.sherwood@arm.com>
1819 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
1820 (widened_mode): Likewise.
1821 (expand_unop): Likewise.
1822 * ree.c (transform_ifelse): Likewise.
1823 (merge_def_and_ext): Likewise.
1824 (combine_reaching_defs): Likewise.
1825 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1827 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
1828 Alan Hayward <alan.hayward@arm.com>
1829 David Sherwood <david.sherwood@arm.com>
1831 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
1832 * combine.c (gen_lowpart_for_combine): Likewise.
1833 * dwarf2out.c (rtl_for_decl_location): Likewise.
1834 * final.c (alter_subreg): Likewise.
1835 * rtlhooks.c (gen_lowpart_general): Likewise.
1836 (gen_lowpart_if_possible): Likewise.
1838 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
1839 Alan Hayward <alan.hayward@arm.com>
1840 David Sherwood <david.sherwood@arm.com>
1842 * calls.c (expand_call): Use subreg_lowpart_offset.
1843 * cse.c (cse_insn): Likewise.
1844 * regcprop.c (copy_value): Likewise.
1845 (copyprop_hardreg_forward_1): Likewise.
1847 2017-10-13 Jakub Jelinek <jakub@redhat.com>
1850 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
1851 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
1852 =Q constraints to +Q and into insn condition add check
1853 that operands[0] and operands[1] are equal.
1854 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
1855 =Q constraints to +Q and into insn condition add check
1856 that operands[0] is equal to either operands[1] or operands[2].
1859 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
1860 instead of handling MINUS_EXPR twice (once for each argument),
1861 canonicalize operand order and handle just once, use rtype where
1862 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
1865 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
1866 any values of __C while still being pattern recognizable as a simple
1869 2017-10-13 Richard Biener <rguenther@suse.de>
1871 PR tree-optimization/82451
1873 2017-10-02 Richard Biener <rguenther@suse.de>
1875 PR tree-optimization/82355
1876 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
1877 a mapping for the enclosing loop but avoid generating one for
1879 (copy_bb_and_scalar_dependences): Remove premature codegen
1880 error on PHIs in blocks duplicated into multiple places.
1881 * graphite-scop-detection.c
1882 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
1883 in the region use it as loop and nest to analyze the DR in.
1884 (try_generate_gimple_bb): Likewise.
1885 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
1886 (add_loop_constraints): For blocks in a loop not in the region
1887 create a dimension with a single iteration.
1888 * sese.h (gbb_loop_at_index): Remove assert.
1890 * cfgloop.c (loop_preheader_edge): For the loop tree root
1891 return the single successor of the entry block.
1892 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
1893 Reset the SCEV hashtable and niters.
1894 * graphite-scop-detection.c
1895 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
1896 assert that we only have POLYNOMIAL_CHREC that vary in loops
1897 contained in the region.
1898 (scop_detection::graphite_can_represent_expr): Adjust.
1899 (scop_detection::stmt_has_simple_data_refs_p): For loops
1900 not in the region set loop to NULL. The nest is now the
1901 entry edge to the region.
1902 (try_generate_gimple_bb): Likewise.
1903 * sese.c (scalar_evolution_in_region): Adjust for
1904 instantiate_scev change.
1905 * tree-data-ref.h (graphite_find_data_references_in_stmt):
1906 Make nest parameter the edge into the region.
1907 (create_data_ref): Likewise.
1908 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
1909 entry edge into a region and adjust instantiate_scev calls.
1910 (create_data_ref): Likewise.
1911 (graphite_find_data_references_in_stmt): Likewise.
1912 (find_data_references_in_stmt): Pass the loop preheader edge
1913 from the nest argument.
1914 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
1915 parameter the edge into the region.
1916 (instantiate_parameters): Use the loop preheader edge as entry.
1917 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
1919 (get_instantiated_value_entry): Make instantiate_below parameter
1920 the edge into the region.
1921 (instantiate_scev_name): Likewise. Adjust dominance checks,
1922 when we cannot use loop-based instantiation instantiate by
1923 walking use-def chains.
1924 (instantiate_scev_poly): Adjust.
1925 (instantiate_scev_binary): Likewise.
1926 (instantiate_scev_convert): Likewise.
1927 (instantiate_scev_not): Likewise.
1928 (instantiate_array_ref): Remove.
1929 (instantiate_scev_3): Likewise.
1930 (instantiate_scev_2): Likewise.
1931 (instantiate_scev_1): Likewise.
1932 (instantiate_scev_r): Do not blindly handle N-operand trees.
1933 Do not instantiate array-refs. Handle all constants and invariants.
1934 (instantiate_scev): Make instantiate_below parameter
1935 the edge into the region.
1936 (resolve_mixers): Use the loop preheader edge for the region
1937 parameter to instantiate_scev_r.
1938 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
1940 2017-10-13 Richard Biener <rguenther@suse.de>
1942 PR tree-optimization/82525
1943 * graphite-isl-ast-to-gimple.c
1944 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
1946 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
1947 Fail code generation when we cannot represent the isl integer.
1948 (binary_op_to_tree): Elide modulo operations that are no-ops
1949 in the type we code generate. Remove now superfluous code
1952 2017-10-13 Richard Biener <rguenther@suse.de>
1954 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
1955 (scop_detection::harmful_loop_in_region): Remove premature
1956 IV type restriction.
1957 (scop_detection::graphite_can_represent_scev): We can handle
1958 pointer IVs just fine.
1960 2017-10-13 Alan Modra <amodra@gmail.com>
1962 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
1963 "Clobbers and Scratch Registers". Add paragraph on
1964 alternative to clobbers for scratch registers and OpenBLAS
1967 2017-10-13 Alan Modra <amodra@gmail.com>
1969 * doc/extend.texi (Clobbers): Correct vax example. Delete old
1970 example of a memory input for a string of known length. Move
1971 commentary out of table. Add a number of new examples
1972 covering array memory inputs.
1974 2017-10-12 Martin Liska <mliska@suse.cz>
1976 PR tree-optimization/82493
1977 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
1978 (test_range_functions): New function.
1979 (sbitmap_c_tests): Likewise.
1980 * selftest-run-tests.c (selftest::run_tests): Run new tests.
1981 * selftest.h (sbitmap_c_tests): New function.
1983 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
1985 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
1987 * config/rs6000/amo.h: Fix spacing issue.
1989 2017-10-12 Jakub Jelinek <jakub@redhat.com>
1992 * config/i386/i386.md (*ashl<mode>3_mask_1,
1993 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
1994 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
1997 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
1999 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
2000 (profile_probability): Set max_probability
2001 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
2003 * profile-count.c (profile_probability::differs_from_p): Do not
2004 rely on max_probaiblity == 10000
2006 2017-10-12 Jeff Law <law@redhat.com>
2008 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
2011 2017-10-12 Martin Sebor <msebor@redhat.com>
2015 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
2016 (handle_alias_pairs): Call it.
2017 * common.opt (-Wattribute-alias): New option.
2018 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
2019 * doc/invoke.texi (-Wattribute-alias): Document.
2021 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
2024 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
2026 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
2028 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
2029 (make_hard_regno_born, make_hard_regno_dead): Update
2030 bb_killed_pseudos and bb_gen_pseudos.
2032 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
2034 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
2036 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
2038 * config/alpha/alpha.c (alpha_split_conditional_move):
2039 Use std::swap instead of manually swapping.
2040 (alpha_stdarg_optimize_hook): Ditto.
2041 (alpha_canonicalize_comparison): Ditto.
2043 2017-10-12 Bin Cheng <bin.cheng@arm.com>
2045 * tree-loop-distribution.c (struct builtin_info): New struct.
2046 (struct partition): Refactor fields into struct builtin_info.
2047 (partition_free): Free struct builtin_info.
2048 (build_size_arg_loc, build_addr_arg_loc): Delete.
2049 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
2050 information from struct builtin_info.
2051 (find_single_drs): New function refactored from classify_partition.
2052 Also moved builtin validity checks to this function.
2053 (compute_access_range, alloc_builtin): New functions.
2054 (classify_builtin_st, classify_builtin_ldst): New functions.
2055 (classify_partition): Refactor code into functions find_single_drs,
2056 classify_builtin_st and classify_builtin_ldst.
2057 (distribute_loop): Don't do runtime alias check when distributing
2059 (find_seed_stmts_for_distribution): New function.
2060 (pass_loop_distribution::execute): Refactor code finding seed
2061 stmts into above function. Support distribution for the innermost
2062 two-level loop nest. Adjust dump information.
2064 2017-10-12 Bin Cheng <bin.cheng@arm.com>
2066 * tree-loop-distribution.c: Adjust the general comment.
2067 (NUM_PARTITION_THRESHOLD): New macro.
2068 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
2069 (classify_partition): Skip builtin pattern of loop nest's inner loop.
2070 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
2071 in call to build_partition_graph.
2072 (finalize_partitions): New parameter. Make loop distribution more
2073 conservative by fusing more partitions.
2074 (distribute_loop): Don't do runtime alias check in case of loop nest
2076 (find_seed_stmts_for_distribution): New function.
2077 (prepare_perfect_loop_nest): New function.
2078 (pass_loop_distribution::execute): Refactor code finding seed stmts
2079 and loop nest into above functions. Support loop nest distribution.
2080 Adjust dump information accordingly.
2082 2017-10-12 Bin Cheng <bin.cheng@arm.com>
2084 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
2085 and set PTYPE_SEQUENTIAL for merged partition.
2087 2017-10-12 Richard Biener <rguenther@suse.de>
2089 PR tree-optimization/69728
2091 2017-09-19 Richard Biener <rguenther@suse.de>
2093 PR tree-optimization/69728
2094 * graphite-sese-to-poly.c (schedule_error): New global.
2095 (add_loop_schedule): Handle empty domain by failing the
2097 (build_original_schedule): Handle schedule_error.
2099 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
2100 domain by returning an unchanged schedule.
2102 2017-10-12 Jakub Jelinek <jakub@redhat.com>
2104 * genrecog.c (validate_pattern): For VEC_SELECT verify that
2105 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
2107 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
2109 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
2112 2017-10-12 Jakub Jelinek <jakub@redhat.com>
2115 * expr.c (store_field): Don't optimize away bitsize == 0 store
2116 from CALL_EXPR with addressable return type.
2118 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
2120 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
2121 * config/rs6000/rs6000.md (sel): Delete mode attribute.
2122 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
2123 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
2124 TARGET_ISEL instead of TARGET_ISEL<sel>.
2126 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
2128 * config/rs6000/rs6000.c
2129 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
2131 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
2133 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
2135 (reg_or_cint_operand): Fix comment.
2136 (reg_or_zero_operand): New predicate.
2137 * config/rs6000/rs6000-protos.h (output_isel): Delete.
2138 * config/rs6000/rs6000.c (output_isel): Delete.
2139 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
2140 instead of reg_or_cint_operand. Output instruction directly (not via
2142 (isel_unsigned_<mode>): Ditto.
2143 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
2144 gpc_reg_operand. Add an instruction alternative for this. Output
2145 instruction directly.
2146 (*isel_reversed_unsigned_<mode>): Ditto.
2148 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
2150 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
2151 (TARGET_CANONICALIZE_COMPARISON): Define.
2153 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
2156 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
2157 Check whether the dest is REG before adding REG_EQUIV note.
2159 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
2162 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
2164 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
2165 (make_hard_regno_born, make_hard_regno_dead): Update
2166 bb_killed_pseudos and bb_gen_pseudos.
2168 2017-10-11 Nathan Sidwell <nathan@acm.org>
2170 * incpath.h (enum incpath_kind): Name enum, prefix values.
2171 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
2172 * incpath.c (heads, tails): Use INC_MAX.
2173 (add_env_var_paths, add_standard_paths): Use incpath_kind.
2174 (merge_include_chains, split_quote_chain,
2175 register_include_chains): Update incpath_kind names.
2176 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
2177 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
2179 (add_framework_path, darwin_register_objc_includes): Likewise.
2180 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
2182 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
2184 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
2185 Do not use float_operator operator predicate.
2186 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
2187 * config/i386/predicates.md (float_operator): Remove predicate.
2189 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
2191 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
2192 (*jccxf_i387): Ditto.
2193 (*jcc<mode>_i387): Ditto.
2194 (*jccu<mode>_i387): Ditto.
2195 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
2196 (*jcc_*_i387 splitters): Remove.
2197 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
2198 * config/i386/i386.c (ix86_split_fp_branch): Remove.
2199 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
2202 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
2204 * profile-count.h (slow_safe_scale_64bit): New function.
2205 (safe_scale_64bit): New inline.
2206 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
2207 * profile-count.c: Include wide-int.h
2208 (slow_safe_scale_64bit): New.
2210 2017-10-11 Nathan Sidwell <nathan@acm.org>
2212 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
2213 HAS_DECL_ASSEMBLER_NAME_P.
2214 * gimple-expr.c (gimple_decl_printable_name: Check
2215 HAS_DECL_ASSEMBLER_NAME_P too.
2216 * ipa-utils.h (type_in_anonymous_namespace_p): Check
2217 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
2218 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
2219 * passes.c (rest_of_decl_compilation): Check
2220 HAS_DECL_ASSEMBLER_NAME_P too.
2221 * recog.c (verify_changes): Likewise.
2222 * tree-pretty-print.c (dump_decl_name): Likewise.
2223 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
2225 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
2226 (SET_DECL_ASSEMBLER_NAME): Use it.
2227 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
2228 (COPY_DECL_ASSEMBLER_NAME): Likewise.
2229 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
2231 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
2233 * config.gcc (i386, x86_64): Add extra objects.
2234 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
2235 (ix86_min_insn_size): Declare.
2236 (ix86_issue_rate): Declare.
2237 (ix86_adjust_cost): Declare.
2238 (ia32_multipass_dfa_lookahead): Declare.
2239 (ix86_macro_fusion_p): Declare.
2240 (ix86_macro_fusion_pair_p): Declare.
2241 (ix86_bd_has_dispatch): Declare.
2242 (ix86_bd_do_dispatch): Declare.
2243 (ix86_core2i7_init_hooks): Declare.
2244 (ix86_atom_sched_reorder): Declare.
2245 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
2246 (COSTS_N_BYTES): Move to x86-tune-costs.h.
2247 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
2248 (rip_relative_addr_p): Rename to ...
2249 (ix86_rip_relative_addr_p): ... this one; export.
2250 (memory_address_length): Update.
2251 (ix86_issue_rate): Move to x86-tune-sched.c.
2252 (ix86_flags_dependent): Move to x86-tune-sched.c.
2253 (ix86_agi_dependent): Move to x86-tune-sched.c.
2254 (exact_dependency_1): Move to x86-tune-sched.c.
2255 (exact_store_load_dependency): Move to x86-tune-sched.c.
2256 (ix86_adjust_cost): Move to x86-tune-sched.c.
2257 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
2258 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
2259 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
2260 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
2261 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
2262 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
2263 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
2264 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
2265 (min_insn_size): Rename to ...
2266 (ix86_min_insn_size): ... this one; export.
2267 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
2268 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
2269 (core2i7_first_cycle_multipass_backtrack): Move to
2270 x86-tune-sched-core.c.
2271 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
2272 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
2273 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
2274 (ix86_avoid_jump_mispredicts): Update.
2275 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
2276 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
2277 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
2278 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
2279 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
2280 (MAX_INSN): Move to ix86-tune-sched-bd.c.
2281 (MAX_IMM): Move to ix86-tune-sched-bd.c.
2282 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
2283 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
2284 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
2285 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
2286 (MAX_STORE): Move to ix86-tune-sched-bd.c.
2287 (BIG): Move to ix86-tune-sched-bd.c.
2288 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
2289 (enum insn_path): Move to ix86-tune-sched-bd.c.
2290 (get_mem_group): Move to ix86-tune-sched-bd.c.
2291 (is_cmp): Move to ix86-tune-sched-bd.c.
2292 (dispatch_violation): Move to ix86-tune-sched-bd.c.
2293 (is_branch): Move to ix86-tune-sched-bd.c.
2294 (is_prefetch): Move to ix86-tune-sched-bd.c.
2295 (init_window): Move to ix86-tune-sched-bd.c.
2296 (allocate_window): Move to ix86-tune-sched-bd.c.
2297 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
2298 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
2299 (process_end_window): Move to ix86-tune-sched-bd.c.
2300 (allocate_next_window): Move to ix86-tune-sched-bd.c.
2301 (find_constant): Move to ix86-tune-sched-bd.c.
2302 (get_num_immediates): Move to ix86-tune-sched-bd.c.
2303 (has_immediate): Move to ix86-tune-sched-bd.c.
2304 (get_insn_path): Move to ix86-tune-sched-bd.c.
2305 (get_insn_group): Move to ix86-tune-sched-bd.c.
2306 (count_num_restricted): Move to ix86-tune-sched-bd.c.
2307 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
2308 (add_insn_window): Move to ix86-tune-sched-bd.c.
2309 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
2310 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
2311 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
2312 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
2313 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
2314 (do_dispatch): Move to ix86-tune-sched-bd.c.
2315 (has_dispatch): Move to ix86-tune-sched-bd.c.
2316 * config/i386/t-i386: Add new object files.
2317 * config/i386/x86-tune-costs.h: New file.
2318 * config/i386/x86-tune-sched-atom.c: New file.
2319 * config/i386/x86-tune-sched-bd.c: New file.
2320 * config/i386/x86-tune-sched-core.c: New file.
2321 * config/i386/x86-tune-sched.c: New file.
2323 2017-10-11 Liu Hao <lh_mouse@126.com>
2325 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
2326 the generic version below instead.
2327 (should_colorize): Recognize Windows consoles as terminals
2329 * pretty-print.c [__MINGW32__] (write_all): New function.
2330 [__MINGW32__] (find_esc_head): Likewise.
2331 [__MINGW32__] (find_esc_terminator): Likewise.
2332 [__MINGW32__] (eat_esc_sequence): Likewise.
2333 [__MINGW32__] (mingw_ansi_fputs): New function that handles
2335 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
2338 2017-10-11 Richard Biener <rguenther@suse.de>
2340 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
2341 Properly call analyze_scalar_evolution with the loop of the stmt.
2343 2017-10-11 Richard Biener <rguenther@suse.de>
2345 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
2346 * tree-core.h (tree_base): Add chrec_var union member.
2347 * tree.h (CHREC_VAR): Remove.
2348 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
2349 * tree-chrec.h (build_polynomial_chrec): Adjust.
2350 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
2351 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
2353 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
2355 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
2356 * match.pd: ... here.
2357 ((T) X == (T) Y): Relax condition.
2359 2017-10-11 Bin Cheng <bin.cheng@arm.com>
2361 PR tree-optimization/82472
2362 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
2364 (break_alias_scc_partitions): Update postorder number.
2366 2017-10-11 Martin Liska <mliska@suse.cz>
2369 * opts.c (parse_no_sanitize_attribute): Do not use error_value
2371 * opts.h (parse_no_sanitize_attribute): Remove last argument.
2373 2017-10-11 Martin Liska <mliska@suse.cz>
2375 * print-rtl.c (print_insn): Move declaration of idbuf
2376 to same scope as name.
2378 2017-10-11 Martin Liska <mliska@suse.cz>
2383 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
2386 2017-10-11 Martin Liska <mliska@suse.cz>
2389 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
2392 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
2394 * config.gcc (default_gnu_indirect_function): Default to yes for
2395 arm*-*-linux* with glibc.
2397 2017-10-11 Richard Biener <rguenther@suse.de>
2399 * tree-scalar-evolution.c (get_scalar_evolution): Handle
2400 default-defs and types we do not want to analyze.
2401 (interpret_loop_phi): Replace unreachable code with an assert.
2402 (compute_scalar_evolution_in_loop): Remove and inline ...
2403 (analyze_scalar_evolution_1): ... here, replacing condition with
2404 what makes the intent clearer. Remove handling of cases
2405 get_scalar_evolution now handles.
2407 2017-10-10 Jim Wilson <wilson@tuliptree.org>
2409 PR rtl-optimization/81434
2410 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
2411 comment for main loop. In sched_group_found if, also add checks for
2412 pass and min_cost_group.
2414 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
2416 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
2417 (rs6000_insn_cost): New function.
2418 * config/rs6000/rs6000.md (cost): New attribute.
2420 2017-10-10 Jakub Jelinek <jakub@redhat.com>
2421 H.J. Lu <hongjiu.lu@intel.com>
2425 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
2426 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
2427 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
2428 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
2429 to OPTION_MASK_ISA_AVX512VL - builtins that have both
2430 OPTION_MASK_ISA_MMX and some other bit set require both
2431 mmx and the ISAs without the mmx bit.
2432 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
2433 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
2434 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
2435 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
2436 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
2437 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
2438 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
2439 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
2440 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
2441 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
2442 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
2443 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
2444 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
2445 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
2446 Add OPTION_MASK_ISA_MMX.
2448 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
2450 * config.gcc (armv7*-*-freebsd*): New target.
2451 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
2453 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
2455 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
2456 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
2459 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
2461 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
2462 a template parameter.
2463 (WIDE_INT_REF_FOR): Update accordingly.
2464 * tree.h (wi::int_traits <const_tree>): Delete.
2465 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
2466 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
2467 (wi::tree_to_wide_ref): New typedef.
2468 (wi::to_wide): New function.
2469 * calls.c (get_size_range): Use wi::to_wide when operating on
2471 * cgraph.c (cgraph_node::create_thunk): Likewise.
2472 * config/i386/i386.c (ix86_data_alignment): Likewise.
2473 (ix86_local_alignment): Likewise.
2474 * dbxout.c (stabstr_O): Likewise.
2475 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
2476 * expr.c (const_vector_from_tree): Likewise.
2477 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
2478 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
2479 (fold_negate_expr_1, int_const_binop_1, const_binop)
2480 (fold_convert_const_int_from_real, optimize_bit_field_compare)
2481 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
2482 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
2483 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
2484 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
2485 (fold_not_const, round_up_loc): Likewise.
2486 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
2487 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
2488 (alloca_call_type): Likewise.
2489 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
2490 * godump.c (go_output_typedef): Likewise.
2491 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
2492 * internal-fn.c (get_min_precision): Likewise.
2493 * ipa-cp.c (ipcp_store_vr_results): Likewise.
2494 * ipa-polymorphic-call.c
2495 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
2496 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
2497 (ipa_modify_call_arguments): Likewise.
2498 * match.pd: Likewise.
2499 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
2500 * print-tree.c (print_node_brief, print_node): Likewise.
2501 * stmt.c (expand_case): Likewise.
2502 * stor-layout.c (layout_type): Likewise.
2503 * tree-affine.c (tree_to_aff_combination): Likewise.
2504 * tree-cfg.c (group_case_labels_stmt): Likewise.
2505 * tree-data-ref.c (dr_analyze_indices): Likewise.
2506 (prune_runtime_alias_test_list): Likewise.
2507 * tree-dump.c (dequeue_and_dump): Likewise.
2508 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
2509 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
2510 * tree-pretty-print.c (dump_generic_node): Likewise.
2511 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
2512 (simple_iv_with_niters): Likewise.
2513 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
2514 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
2515 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
2516 * tree-ssa-loop-niter.c (split_to_var_and_offset)
2517 (refine_value_range_using_guard, number_of_iterations_ne_max)
2518 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
2519 (get_cst_init_from_scev, record_nonwrapping_iv)
2520 (scev_var_range_cant_overflow): Likewise.
2521 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
2522 * tree-ssa-pre.c (compute_avail): Likewise.
2523 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
2524 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
2525 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
2526 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
2527 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
2528 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
2529 (dump_case_nodes, try_switch_expansion): Likewise.
2530 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
2531 (vect_do_peeling): Likewise.
2532 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
2533 * tree-vect-stmts.c (vectorizable_load): Likewise.
2534 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
2535 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
2536 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
2537 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
2538 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
2539 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
2540 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
2541 (evrp_dom_walker::before_dom_children): Likewise.
2542 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
2543 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
2544 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
2545 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
2546 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
2547 (get_range_pos_neg): Likewise.
2548 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
2549 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
2550 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
2551 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
2552 * config/avr/avr.c (avr_fold_builtin): Likewise.
2553 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
2554 * config/msp430/msp430.c (msp430_attr): Likewise.
2555 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
2556 * config/powerpcspe/powerpcspe-c.c
2557 (altivec_resolve_overloaded_builtin): Likewise.
2558 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
2559 (rs6000_expand_ternop_builtin): Likewise.
2560 * config/rs6000/rs6000-c.c
2561 (altivec_resolve_overloaded_builtin): Likewise.
2562 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
2563 (rs6000_expand_ternop_builtin): Likewise.
2564 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
2566 2017-10-10 Bin Cheng <bin.cheng@arm.com>
2568 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
2569 when copying loop nest with only one inner loop.
2571 2017-10-10 Richard Biener <rguenther@suse.de>
2573 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
2574 blocks if SCEV is active.
2575 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
2577 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
2578 (scev_initialize): Assert we are not yet initialized.
2580 2017-10-10 Bin Cheng <bin.cheng@arm.com>
2582 * tree-loop-distribution.c (generate_loops_for_partition): Remove
2583 inner loop's exit stmt by making it always exit the loop, otherwise
2584 we would generate an infinite empty loop.
2586 2017-10-10 Bin Cheng <bin.cheng@arm.com>
2588 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
2589 renaming variables in new preheader if it's deleted.
2591 2017-10-10 Bin Cheng <bin.cheng@arm.com>
2593 * tree-loop-distribution.c (struct partition): Remove unused field
2594 loops of the structure.
2595 (partition_alloc, partition_free): Ditto.
2596 (build_rdg_partition_for_vertex): Ditto.
2598 2017-10-09 Jeff Law <law@redhat.com>
2600 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
2601 return type to match prototype and documentation.
2603 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
2605 * config/rs6000/rs6000.c (processor_costs): Move to ...
2606 * config/rs6000/rs6000.h: ... here.
2607 (rs6000_cost): Declare.
2609 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
2611 * except.c (setjmp_fn): New global variable.
2612 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
2613 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
2614 if DONT_USE_BUILTIN_SETJMP is defined.
2616 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
2618 * target.def (insn_cost): New hook.
2619 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
2620 * doc/tm.texi: Regenerate.
2621 * rtlanal.c (insn_cost): Use the new hook.
2623 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
2625 * combine.c (combine_validate_cost): Compute the new insn_cost,
2626 not just pattern_cost.
2627 (try_combine): Adjust comment.
2629 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
2631 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
2633 * combine.c (uid_insn_cost): Adjust comment.
2634 (combine_validate_cost): Adjust comment. Use pattern_cost instead
2636 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
2637 * dse.c (find_shift_sequence): Ditto.
2638 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
2639 (bb_valid_for_noce_process_p): Use pattern_cost.
2640 * rtl.h (insn_rtx_cost): Delete.
2641 (pattern_cost): New prototype.
2642 (insn_cost): New prototype.
2643 * rtlanal.c (insn_rtx_cost): Rename to...
2644 (pattern_cost): ... this.
2647 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
2649 * config/i386/i386.md (*jcc_2): Remove insn pattern.
2650 (*jcc<mode>_0_r_i387): Ditto.
2651 (*jccxf_r_i387): Ditto.
2652 (*jcc<mode>_r_i387): Ditto.
2653 (*jccu<mode>_r_i387): Ditto.
2654 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
2655 (*jcc): Rename from *jcc_1.
2657 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2659 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
2660 deferred rescans after the lvx/stvx recombination pre-pass.
2662 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
2664 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
2665 memory operation instruction support.
2666 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
2667 (rs6000-ibm-aix[789]*): Likewise.
2668 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
2669 Document new functions.
2671 2017-10-09 Richard Biener <rguenther@suse.de>
2673 PR tree-optimization/82397
2674 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
2675 equality only for semantically equal trees.
2677 2017-10-09 Richard Biener <rguenther@suse.de>
2679 PR tree-optimization/82449
2680 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
2681 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
2682 allow constant addresses.
2683 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
2686 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2688 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
2691 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2694 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
2697 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2700 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
2702 2017-10-09 Jakub Jelinek <jakub@redhat.com>
2705 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
2706 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
2708 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
2710 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
2711 (WI_BINARY_PREDICATE_RESULT): Likewise.
2712 (wi::binary_traits::operator_result): New type.
2713 (wi::binary_traits::predicate_result): Likewise.
2714 (generic_wide_int::operator~, unary generic_wide_int::operator-)
2715 (generic_wide_int::operator==, generic_wide_int::operator!=)
2716 (generic_wide_int::operator&, generic_wide_int::and_not)
2717 (generic_wide_int::operator|, generic_wide_int::or_not)
2718 (generic_wide_int::operator^, generic_wide_int::operator+
2719 (binary generic_wide_int::operator-, generic_wide_int::operator*):
2721 (operator~, unary operator-, operator==, operator!=, operator&)
2722 (operator|, operator^, operator+, binary operator-, operator*): New
2724 * expr.c (get_inner_reference): Use wi::bit_and_not.
2725 * fold-const.c (fold_binary_loc): Likewise.
2726 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
2727 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
2728 (bit_value_binop): Likewise.
2729 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
2730 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
2731 (extract_range_from_binary_expr_1): Likewise.
2732 (masked_increment): Likewise.
2733 (simplify_bit_ops_using_ranges): Likewise.
2735 2017-10-09 Martin Jambor <mjambor@suse.cz>
2738 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
2739 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
2740 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
2742 (hsa_fixup_mov_insn_type): New function.
2743 (hsa_op_with_type::get_in_type): Use it.
2744 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
2745 immediates in an assert.
2746 (hsa_op_with_type::extend_int_to_32bit): New method.
2747 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
2748 types. Convert to dest type if necessary.
2749 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
2750 (reg_for_gimple_ssa): Pass false as min32int to
2751 hsa_type_for_scalar_tree_type.
2752 (gen_hsa_addr): Fixup type when creating addresable temporary.
2753 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
2754 (gen_hsa_unary_operation): Extend operands and convert to dest type if
2755 necessary. Call hsa_fixup_mov_insn_type.
2756 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
2757 extend operands and convert to dest type if necessary.
2758 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
2759 to dest type if necessary.
2760 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
2761 if conversion nt necessary and size matches.
2762 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
2763 to dest type if necessary.
2764 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
2765 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
2767 (gen_hsa_clrsb): Likewise.
2768 (gen_hsa_ffs): Likewise.
2769 (gen_hsa_divmod): Extend operands and convert to dest type if
2771 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
2773 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
2775 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
2776 Remove empty default arguments. Use a brace block as output
2778 (conditional return): Ditto.
2780 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
2781 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
2783 (group_ending_nop): Ditto.
2784 (doloop_end): Ditto.
2785 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
2786 (splitters for those): Ditto.
2788 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
2790 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
2791 a conditional jump (and the compare for it) so that pc_rtx is the
2793 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
2794 for the deleted and renamed ctr<mode>_internal[234] patterns.
2795 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
2796 Delete second conditional return pattern.
2797 (ctr<mode>_internal2): Delete this second bdnz pattern.
2798 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
2799 (ctr<mode>_internal4): Delete this second bdz pattern.
2801 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
2803 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
2804 (always_initialized_rtx_for_ssa_name_p): New predicate.
2805 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
2806 (finish_out_of_ssa): Free new field of SA.
2807 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
2808 * tree-ssa-coalesce.c: Include tree-ssa.h.
2809 (get_parm_default_def_partitions): Remove extern keyword.
2810 (get_undefined_value_partitions): New function.
2811 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
2812 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
2814 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
2816 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
2818 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
2820 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
2822 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
2823 for targets that preffer 128bit.
2825 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
2827 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
2829 2017-10-08 Olivier Hainque <hainque@adacore.com>
2831 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
2832 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
2834 (thumb_set_return_address): Likewise.
2836 2017-10-08 Olivier Hainque <hainque@adacore.com>
2838 * common/config/arm/arm-common.c (arm_except_unwind_info):
2839 Handle DWARF2_UNWIND_INFO.
2841 2017-10-07 Michael Collison <michael.collison@arm.com>
2843 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
2846 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
2848 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
2849 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
2850 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
2851 defined, force the creation of a new block for a dispatch label.
2853 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
2855 * invoke.texi (Wsuggest-attribute=cold): Document.
2856 * common.opt (Wsuggest-attribute=cold): New
2857 * ipa-pure-const.c (warn_function_cold): New function.
2858 * predict.c (compute_function_frequency): Use it.
2859 * predict.h (warn_function_cold): Declare.
2861 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
2863 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
2866 2017-10-06 Martin Liska <mliska@suse.cz>
2868 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
2869 keyword for member functions.
2870 (struct sanopt_tree_couple): New struct.
2871 (struct sanopt_tree_couple_hash): New function.
2872 (struct sanopt_ctx): Add new hash_map.
2873 (has_dominating_ubsan_ptr_check): New function.
2874 (record_ubsan_ptr_check_stmt): Likewise.
2875 (maybe_optimize_ubsan_ptr_ifn): Likewise.
2876 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
2877 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
2879 2017-10-06 Sudakshina Das <sudi.das@arm.com>
2882 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
2883 aarch64_simd_valid_immediate on CONST_VECTORs.
2884 (aarch64_reg_or_bic_imm): Likewise.
2886 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
2888 PR rtl-optimization/82396
2889 * haifa-sched.c (ready_sort_real): Disable qsort checking.
2891 2017-10-06 Sebastian Pop <sebpop@gmail.com>
2893 * graphite-dependences.c (scop_get_reads): Move code to...
2894 (scop_get_must_writes): Move code to...
2895 (scop_get_may_writes): Move code to...
2896 (scop_get_reads_and_writes): ... here.
2897 (scop_get_dependences): Call scop_get_reads_and_writes.
2899 2017-10-06 Jakub Jelinek <jakub@redhat.com>
2901 PR tree-optimization/82434
2902 * fold-const.h (can_native_encode_type_p,
2903 can_native_encode_string_p): Remove.
2904 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
2905 don't encode anything, just return what would be otherwise returned.
2906 (native_encode_fixed, native_encode_complex, native_encode_vector):
2908 (native_encode_string): Likewise. Inline by hand
2909 can_native_encode_string_p.
2910 (can_native_encode_type_p): Remove.
2911 (can_native_encode_string_p): Remove.
2912 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
2913 STRING_CSTs using can_native_encode_string_p, test all
2914 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
2915 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
2916 argument from native_encode_expr.
2917 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
2918 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
2921 2017-10-06 Richard Biener <rguenther@suse.de>
2923 PR tree-optimization/82397
2924 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
2925 operand_equal_p but rely on data_ref_compare_tree for detecting
2927 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
2928 to match up with dr_group_sort_cmp.
2930 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2933 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
2935 * config/s390/s390-builtin-types.def: Regenerate.
2937 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2940 * config/s390/s390-builtin-types.def: Regenerate.
2941 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
2942 Change flag from B_VXE to B_VX.
2943 (s390_vec_min_dbl): Remove B_VXE flag.
2945 2017-10-06 Richard Biener <rguenther@suse.de>
2947 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
2948 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
2949 translate_isl_ast_to_gimple::is_valid_rename,
2950 translate_isl_ast_to_gimple::get_rename,
2951 translate_isl_ast_to_gimple::get_def_bb_for_const,
2952 translate_isl_ast_to_gimple::get_new_name,
2953 translate_isl_ast_to_gimple::collect_all_ssa_names,
2954 translate_isl_ast_to_gimple::copy_loop_phi_args,
2955 translate_isl_ast_to_gimple::collect_all_ssa_names,
2956 translate_isl_ast_to_gimple::copy_loop_phi_args,
2957 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
2958 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
2959 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
2960 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
2961 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
2962 translate_isl_ast_to_gimple::copy_cond_phi_args,
2963 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
2964 translate_isl_ast_to_gimple::edge_for_new_close_phis,
2965 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
2966 translate_isl_ast_to_gimple::rename_uses,
2967 translate_isl_ast_to_gimple::rename_all_uses): Remove.
2968 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
2969 (set_rename_for_each_def): Likewise.
2970 (graphite_copy_stmts_from_block): Handle debug stmt resetting
2971 here. Handle rewriting SCEV analyzable uses here.
2972 (copy_bb_and_scalar_dependences): Generate code for PHI
2974 (graphite_regenerate_ast_isl): Adjust.
2975 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
2976 (add_write, add_read): New functions.
2977 (build_cross_bb_scalars_def): Use it and simplify.
2978 (build_cross_bb_scalars_use): Likewise.
2979 (graphite_find_cross_bb_scalar_vars): Inline into...
2980 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
2981 simulating out-of-SSA. Compute liveout and add dependencies.
2982 (build_scops): Force an empty entry block.
2983 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
2985 (sese_build_liveouts): Declare.
2986 (sese_trivially_empty_bb_p): Likewise.
2987 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
2988 compute liveout and debug_liveout.
2989 (sese_bad_liveouts_use): Remove.
2990 (sese_reset_debug_liveouts_bb): Likewise.
2991 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
2992 (sese_build_liveouts): Build liveout and debug_liveout and store
2994 (new_sese_info): Adjust.
2995 (free_sese_info): Likewise.
2996 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
2997 do not build liveout here.
2998 (move_sese_in_condition): Adjust region entry.
2999 (scev_analyzable_p): Match up with chrec_apply requirements.
3000 (sese_trivially_empty_bb_p): New.
3001 * tree-into-ssa.c (get_reaching_def): Properly support generating
3002 default-defs for incremental rewrite of anonymous names.
3004 2017-10-06 Richard Biener <rguenther@suse.de>
3006 * graphite-sese-to-poly.c (extract_affine): For casts increasing
3007 precision do not perform modulo reduction.
3009 2017-10-06 Richard Biener <rguenther@suse.de>
3011 PR tree-optimization/82436
3012 * tree-vect-slp.c (vect_supported_load_permutation_p): More
3013 conservatively choose the vectorization factor when checking
3014 whether we can perform the required load permutation.
3015 (vect_transform_slp_perm_load): Assert when we may not fail.
3017 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
3019 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
3020 message for incompatible -msdata=* and -mcall-* options.
3022 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
3024 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
3025 rate for post-reload scheduling.
3027 2017-10-05 Tamar Christina <tamar.christina@arm.com>
3029 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
3031 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
3033 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
3034 to improve monte carlo in scimark.
3036 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
3038 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
3039 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
3040 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
3041 pentium4_cost, nocona_cost): Set reassociation width to 1.
3042 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
3043 width to 2 for fp operations and 1 otherwise.
3044 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
3046 (atom_cost): Set reassociation width to 2.
3047 (slm_cost, generic_cost): Set fp reassociation width
3048 to 2 and 1 otherwise.
3049 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
3050 (core_cost): Set fp reassociation width to 4 and vector to 2.
3051 (ix86_reassociation_width): Rewrite using cost table; special case
3052 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
3053 and TARGET_AVX128_OPTIMAL.
3054 * config/i386/i386.h (processor_costs): Add
3055 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
3056 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
3057 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
3058 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
3059 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
3060 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
3062 2017-10-05 Nathan Sidwell <nathan@acm.org>
3064 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
3066 2017-10-05 Tamar Christina <tamar.christina@arm.com>
3068 * config/arm/arm.c (arm_test_fpu_data): New.
3069 (arm_run_selftests): Call arm_test_fpu_data.
3071 2017-10-04 Nathan Sidwell <nathan@acm.org>
3073 * toplev.c (toplev::main): Remove excess parens on pretty_printer
3075 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
3077 2017-10-04 Sudakshina Das <sudi.das@arm.com>
3079 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
3080 check type for aarch64_simd_valid_immediate.
3081 (aarch64_output_simd_mov_immediate): Update prototype.
3082 (aarch64_simd_valid_immediate): Update prototype.
3083 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
3084 support for ORR-immediate.
3085 (and<mode>3): modified pattern to add support for BIC-immediate.
3086 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
3087 now checks for valid immediate for BIC and ORR based on new enum
3089 (aarch64_output_simd_mov_immediate): Function now used to output
3090 BIC/ORR imm as well based on new enum argument.
3091 * config/aarch64/constraints.md (Do): New vector immediate constraint.
3093 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
3094 (aarch64_reg_or_bic_imm): Likewise.
3096 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3098 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
3099 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
3101 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
3105 PR rtl-optimization/82396
3106 * haifa-sched.c (autopref_multipass_init): Simplify
3108 (autopref_rank_data): Simplify sort order.
3109 * sched-int.h (autopref_multipass_data_): Remove
3110 multi_mem_insn_p, min_offset and max_offset.
3112 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3114 * doc/sourcebuild.texi: Document vect_peeling_profitable.
3116 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3118 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
3121 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3123 * doc/sourcebuild.texi: Document vect_long_mult.
3125 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
3127 PR tree-optimization/82413
3128 * fold-const.c (build_range_check): Use widest_int when comparing
3129 the maximum ETYPE value with HIGH.
3131 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
3133 PR rtl-optimization/82396
3134 * haifa-sched.c (autopref_multipass_init): Simplify
3136 (autopref_rank_data): Simplify sort order.
3137 * sched-int.h (autopref_multipass_data_): Remove
3138 multi_mem_insn_p, min_offset and max_offset.
3140 2017-10-04 Jakub Jelinek <jakub@redhat.com>
3142 PR tree-optimization/82381
3143 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
3144 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
3147 PR tree-optimization/82374
3148 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
3149 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
3150 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
3151 current_function_decl to the new decl.
3153 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
3155 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
3156 helper macro for IEEE float128 hardware built-in functions.
3157 (SQRTF128_ODD): Add built-in functions with the round-to-odd
3159 (TRUNCF128_ODD): Likewise.
3160 (ADDF128_ODD): Likewise.
3161 (SUBF128_ODD): Likewise.
3162 (MULF128_ODD): Likewise.
3163 (DIVF128_ODD): Likewise.
3164 (FMAF128_ODD): Likewise.
3165 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
3166 UNSPEC_TRUNC_ROUND_TO_ODD.
3167 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
3168 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
3169 floating point round to odd instructions.
3170 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
3171 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
3172 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
3173 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
3174 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
3175 (trunc<mode>sf2_hw): Change the truncate with round to odd
3176 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
3177 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
3178 to odd hardware instructions.
3179 (sub<mode>3_odd): Likewise.
3180 (mul<mode>3_odd): Likewise.
3181 (div<mode>3_odd): Likewise.
3182 (sqrt<mode>2_odd): Likewise.
3183 (fma<mode>4_odd): Likewise.
3184 (fms<mode>4_odd): Likewise.
3185 (nfma<mode>4_odd): Likewise.
3186 (nfms<mode>4_odd): Likewise.
3187 (trunc<mode>df2_odd): Change the truncate with round to odd
3188 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
3190 * doc/extend.texi (PowerPC built-in functions): Update documentation
3191 for existing IEEE float128-bit built-in functions. Add built-in
3192 functions that generate the IEEE 128-bit floating point round to
3195 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
3197 PR rtl-optimization/77729
3198 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
3199 to (X&(C1&~C2))|C2 transformations.
3201 2017-10-03 Martin Jambor <mjambor@suse.cz>
3203 PR tree-optimization/82363
3204 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
3205 mismatch, mark lacc written regardless of racc.
3207 2017-10-03 Jakub Jelinek <jakub@redhat.com>
3209 PR tree-optimization/82381
3210 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
3211 stmt_to_insert nor wheather SSA_NAMEs are default defs.
3212 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
3213 fallthrough into reassoc_stmt_dominates_stmt_p.
3216 * combine.c (combine_instructions): Don't combine in unreachable
3219 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
3222 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
3223 function to not use the have_cpu variable. Do not set cpu_index,
3224 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
3226 (rs6000_valid_attribute_p): Remove duplicate initializations of
3227 old_optimize and func_optimize.
3228 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
3229 (rs6000_activate_target_options): Make global.
3230 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
3233 2017-10-02 Jakub Jelinek <jakub@redhat.com>
3235 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
3236 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
3237 Set *poffset to 0 and *psize and *pmax_size to -1 if
3238 *poffset + *psize overflows in HOST_WIDE_INT.
3240 PR tree-optimization/82387
3241 PR tree-optimization/82388
3242 PR tree-optimization/82389
3243 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
3244 instead of live_bytes non-NULL.
3246 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
3249 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
3252 2017-10-02 Richard Biener <rguenther@suse.de>
3254 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
3255 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
3256 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
3258 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
3260 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
3261 requested precision matches the type's.
3262 * calls.c (alloc_max_size): Calculate the new candidate size as
3263 a widest_int and use wi::to_widest when comparing it with the
3264 current candidate size.
3265 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
3266 zero rather than integer_zero_node.
3267 * match.pd: Check for a no-op conversion before using wi::add
3268 rather than after. Use tree_to_uhwi when summing small shift
3269 counts into an unsigned int.
3271 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
3272 Alan Hayward <alan.hayward@arm.com>
3273 David Sherwood <david.sherwood@arm.com>
3276 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
3277 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
3278 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
3279 POINTER_AND_FP_REGS.
3281 2017-10-02 Richard Biener <rguenther@suse.de>
3283 PR tree-optimization/82355
3284 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
3285 a mapping for the enclosing loop but avoid generating one for
3287 (copy_bb_and_scalar_dependences): Remove premature codegen
3288 error on PHIs in blocks duplicated into multiple places.
3289 * graphite-scop-detection.c
3290 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
3291 in the region use it as loop and nest to analyze the DR in.
3292 (try_generate_gimple_bb): Likewise.
3293 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
3294 (add_loop_constraints): For blocks in a loop not in the region
3295 create a dimension with a single iteration.
3296 * sese.h (gbb_loop_at_index): Remove assert.
3298 2017-10-01 Kevin Buettner <kevinb@redhat.com>
3300 * omp-expand.c (adjust_context_scope): New function.
3301 (expand_parallel_call): Call adjust_context_scope.
3303 2017-10-01 Jeff Law <law@redhat.com>
3305 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
3306 dom_opt_dom_walker class with direct access to private members.
3307 Add comments. Call test_for_singularity.
3308 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
3309 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
3310 m_dummy_cond anymore.
3311 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
3313 (pass_dominator:execute): Build the dummy_cond here and pass it
3314 to the dom_opt_dom_walker ctor.
3315 (test_for_singularity): New function.
3317 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
3318 Maya Rashish <coypu@sdf.org>
3320 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
3321 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
3322 (alpha*-*-netbsd*) Use nbsd_tm_file.
3323 (arm*-*-netbsdelf*) Likewise.
3324 (i[34567]86-*-netbsdelf*) Likewise.
3325 (x86_64-*-netbsd*) Likewise.
3326 (mips*-*-netbsd*) Likewise.
3327 (powerpc-*-netbsd*) Likewise.
3328 (sh*-*-netbsd*) Likewise.
3329 (sparc-*-netbsdelf*) Likewise.
3330 (sparc64-*-netbsd*) Likewise.
3331 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
3333 (vax-*-netbsdelf*) Likewise.
3334 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
3335 (UINT_FAST8_TYPE) Likewise.
3336 (INT_FAST16_TYPE) Check CHAR_FAST16.
3337 (UINT_FAST16_TYPE) Likewise.
3339 2017-09-30 Jakub Jelinek <jakub@redhat.com>
3342 * config/i386/i386.md
3343 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
3344 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
3345 *divmodsi4_zext_2): New define_insn_and_split.
3346 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
3347 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
3348 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
3349 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
3350 New define_insn_and_split.
3351 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
3352 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
3353 operands[1] having DImode when mode is SImode.
3355 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
3356 always SImode for DIV and MOD in REG_EQUAL notes.
3358 2017-09-29 Yury Gribov <tetra2005@gmail.com>
3361 * match.pd: Fix handling of NaNs in pattern.
3363 2017-09-29 Jeff Law <law@redhat.com>
3365 * sbitmap.c (bitmap_bit_in_range_p): New function.
3366 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
3367 * tree-ssa-dse.c (live_bytes_read): New function.
3368 (dse_classify_store): Ignore reads of dead bytes.
3370 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
3371 typos and whitespace errors.
3372 * config/i386/predicates.md (address_no_seg_operand): Likewise.
3373 * config/s390/s390.c (s390_emit_prologue): Likewise.
3375 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
3378 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
3379 with a symbol for LRA.
3381 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
3383 PR rtl-optimization/82338
3384 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
3386 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
3388 * genmodes.c (calc_wider_mode): Suppress qsort macro.
3389 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
3390 (qsort_chk): Declare.
3391 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
3392 (qsort_chk): New function.
3394 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3396 PR tree-optimization/82337
3397 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
3398 phi definition if the PHI result appears in an abnormal PHI.
3399 (find_basis_for_base_expr): Don't record a basis if the LHS of the
3400 basis appears in an abnormal PHI.
3402 2017-09-29 Richard Biener <rguenther@suse.de>
3404 * graphite-isl-ast-to-gimple.c
3405 (translate_isl_ast_to_gimple::set_codegen_error): New function.
3406 (binary_op_to_tree): Use it.
3407 (get_rename_from_scev): Likewise.
3408 (copy_loop_phi_nodes): Likewise.
3409 (copy_bb_and_scalar_dependences): Likewise.
3410 (translate_pending_phi_nodes): Likewise.
3412 2017-09-29 Jakub Jelinek <jakub@redhat.com>
3415 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
3416 for movabsq $(i32 << shift), r64.
3418 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
3420 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
3421 index when encoding %esp as %rsp to avoid 0x67 prefix.
3423 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
3425 * config/i386/i386.md (*movsf_internal, *movdf_internal):
3426 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
3428 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
3430 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
3431 Extensions with more than 16 double VFP registers.
3432 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
3433 to_clear_mask and all code related to it. Replace the remaining
3434 entry by a sbitmap and adapt code accordingly.
3436 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
3438 * brig-builtins.def: Change pure attributes to const.
3440 2017-09-28 Joseph Myers <joseph@codesourcery.com>
3442 * config.gcc (default_gnu_indirect_function): Default to yes for
3443 sparc*-*-linux* with glibc.
3445 2017-09-28 Joseph Myers <joseph@codesourcery.com>
3447 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
3448 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
3449 when creating .init_array and .fini_array sections with priority
3452 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
3455 * config/aarch64/aarch64.c
3456 (aarch64_builtin_support_vector_misalignment): Always return false
3457 when misalignment is unknown.
3459 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
3461 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
3462 this function to return false if the definition used by the swap
3463 instruction is artificial, or if the memory address from which the
3464 constant value is loaded is not represented by a base address held
3465 in a register or if the base address register is a frame or stack
3466 pointer. Additionally, return false if the base address of the
3467 loaded constant is a SYMBOL_REF but is not considered to be a
3469 (replace_swapped_load_constant): New function.
3470 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
3471 loaded constant vector with a load of a swapped constant vector.
3473 2017-09-27 Carl Love <cel@us.ibm.com>
3475 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
3476 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
3477 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
3480 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
3482 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
3483 first, always call autopref_rank_data otherwise.
3485 2017-09-27 Richard Biener <rguenther@suse.de>
3487 * graphite-scop-detection.c (find_scop_parameters): Move
3488 loop bound handling ...
3489 (gather_bbs::before_dom_children): ... here, avoiding the need
3490 to build scop_info->loop_nest.
3491 (record_loop_in_sese): Remove.
3492 * sese.h (sese_info_t::loop_nest): Remove.
3493 * sese.c (new_sese_info): Do not allocate loop_nest.
3494 (free_sese_info): Do not free loop_nest.
3496 2017-09-27 Jakub Jelinek <jakub@redhat.com>
3499 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
3500 lhs from calls if the lhs has addressable type.
3502 2017-09-27 Richard Biener <rguenther@suse.de>
3504 * graphite.h (scop::max_alias_set): New member.
3505 * graphite-scop-detection.c: Remove references to non-existing
3506 --param in comments.
3507 (build_alias_sets): Record the maximum alias set used for drs.
3508 (build_scops): Support zero as unlimited for
3509 --param graphite-max-arrays-per-scop.
3510 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
3512 (build_poly_sr_1): ... here. Compute alias set based on the
3513 maximum alias set used for drs rather than
3514 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
3516 2017-09-27 Richard Biener <rguenther@suse.de>
3518 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
3519 --param loop-block-tile-size=0 to disable tiling.
3521 2017-09-27 Richard Biener <rguenther@suse.de>
3523 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
3524 (graphite-max-nb-scop-params): Document special value zero.
3525 * domwalk.h (dom_walker::STOP): New symbolical constant.
3526 (dom_walker::dom_walker): Add optional parameter for bb to
3528 (dom_walker::~dom_walker): Declare.
3529 (dom_walker::before_dom_children): Document STOP return value.
3530 (dom_walker::m_user_bb_to_rpo): New member.
3531 (dom_walker::m_bb_to_rpo): Likewise.
3532 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
3533 mapping here if not provided by the user.
3534 (dom_walker::~dom_walker): Free bb to RPO mapping if not
3535 provided by the user.
3536 (dom_walker::STOP): Define.
3537 (dom_walker::walk): Do not compute bb to RPO mapping here.
3538 Support STOP return value from before_dom_children to stop
3540 * graphite-optimize-isl.c (optimize_isl): If the schedule
3541 is the same still generate code if -fgraphite-identity
3542 or -floop-parallelize-all are given.
3543 * graphite-scop-detection.c: Include cfganal.h.
3544 (gather_bbs::gather_bbs): Get and pass through bb to RPO
3546 (gather_bbs::before_dom_children): Return STOP for BBs
3548 (build_scops): Compute bb to RPO mapping and pass it to
3549 the domwalk. Treat --param graphite-max-nb-scop-params=0
3550 as not limiting the number of params.
3551 * graphite.c (graphite_initialize): Remove limit on the
3552 number of basic-blocks in a function.
3553 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
3554 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
3555 default value of 10.
3557 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
3559 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
3560 Adjust code to eliminate needing to do the shift right 32-bits
3561 operation after XSCVDPSPN.
3563 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3565 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
3566 ((X / Y) != 0 -> X >= Y): Likewise.
3568 2017-09-26 Carl Love <cel@us.ibm.com>
3570 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
3571 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
3572 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
3573 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
3574 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
3575 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
3576 definitions and overloading.
3577 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
3578 statement for P9V_BUILTIN_XST_LEN_R.
3579 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
3580 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
3581 define_expand and define_insn for the instructions and builtins.
3582 * doc/extend.texi: Update the built-in documentation file for the new
3584 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
3585 define_insn for the instructions
3587 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
3590 * gcc/config/netbsd-protos.h: New file.
3591 * gcc/config/netbsd.c: New file.
3592 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
3593 * gcc/config/t-netbsd: New file.
3594 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
3595 (tmake_file) Add t-netbsd.
3596 (extra_objs) Add netbsd.o.
3598 2017-09-26 Janus Weil <janus@gcc.gnu.org>
3602 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
3604 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
3606 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
3607 sign extension from a vector register to a GPR by doing a 32-bit
3608 direct move and then an EXTSW.
3609 (extendsi<mode>2 splitter): Likewise.
3610 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
3611 right or vector extract after doing XSCVDPSPN. Use
3612 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
3614 (movdi_from_sf_zero_ext): Likewise.
3615 (reload_gpr_from_vsxsf): Likewise.
3616 (p8_mfvsrd_4_disf): Delete, no longer used.
3617 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
3618 then needing to move the SFmode to a GPR to use the XSCVDPSP
3619 instruction instead of FRSP and XSCVDPSPN.
3620 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
3621 it is adjacent to the other XSCVSPDP insns.
3622 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
3623 SFmode to be in traditional Altivec registers.
3624 (vsx_xscvdpspn): Eliminate useless alternative constraint.
3625 (vsx_xscvspdpn): Likewise.
3626 (vsx_xscvspdpn_scalar): Likewise.
3628 2017-09-26 Martin Jambor <mjambor@suse.cz>
3630 * tree-sra.c (compare_access_positions): Put integral types first,
3631 stabilize sorting of integral types, remove conditions putting
3632 non-full-precision integers last.
3633 (sort_and_splice_var_accesses): Disable scalarization if a
3634 non-integert would be represented by a non-full-precision integer.
3636 2017-09-26 Joseph Myers <joseph@codesourcery.com>
3638 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
3639 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
3640 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
3641 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
3642 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
3643 conditionals inside the function instead of around it. Call
3644 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
3645 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
3647 2017-09-26 Richard Biener <rguenther@suse.de>
3649 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
3651 (scop_detection::build_scop_breadth): ... this. Removed.
3652 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
3653 (scop_detection::harmful_stmt_in_bb): Likewise.
3654 (scop_detection::graphite_can_represent_stmt): Likewise.
3655 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
3656 (scop_detection::can_represent_loop): Remove recursion, fold in ...
3657 (scop_detection::can_represent_loop_1): ... this. Removed.
3658 (scop_detection::harmful_loop_in_region): Simplify after inlining
3659 the above and remove more quadraticness.
3660 (build_scops): Adjust.
3661 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
3664 2017-09-26 Jakub Jelinek <jakub@redhat.com>
3667 * config/i386/i386.c (ix86_print_operand_address_as): Only test
3668 REGNO (base) == SP_REG if base is a REG.
3671 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
3672 if it is different SSA_NAME.
3673 (optimize_range_tests_cmp_bitwise): New function.
3674 (optimize_range_tests): Call it.
3676 2017-09-26 Richard Biener <rguenther@suse.de>
3678 PR tree-optimization/82321
3679 * graphite.c (canonicalize_loop_closed_ssa): Properly check
3680 for the def being inside the loop.
3682 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3684 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
3686 * config/s390/s390-builtins.def: Fix constraint on op4.
3688 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3690 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
3691 independent expanders.
3692 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
3693 ("vec_ordered", "vec_unordered"): New expanders.
3695 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3697 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
3700 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3702 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
3703 vec_unpacks_lo_v16qi.
3704 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
3706 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3708 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
3709 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
3710 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
3712 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3714 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
3716 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
3718 ("*vec_slb<mode>"): New insn pattern.
3719 ("vec_shr_<mode>"): New expander.
3720 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
3721 and force the shift count operand to V16QImode.
3722 ("vec_srb<mode>"): Set shift count mode to V16QI.
3724 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3726 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
3727 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
3728 ("vec_widen_smult_hi_<mode>"): New expander definitions.
3730 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
3733 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
3735 2017-09-26 Richard Biener <rguenther@suse.de>
3737 PR tree-optimization/82320
3738 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
3741 2017-09-25 Jeff Law <law@redhat.com>
3743 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
3744 prototype for new argument.
3745 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
3746 mostly extracted from rs6000_emit_allocate_stack.
3747 (rs6000_emit_probe_stack_range_stack_clash): New function.
3748 (rs6000_emit_allocate_stack): Call
3749 rs6000_emit_probe_stack_range_stack_clash as needed.
3750 (rs6000_emit_probe_stack_range): Add additional argument
3751 to call to gen_probe_stack_range{si,di}.
3752 (output_probe_stack_range): New.
3753 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
3754 (output_probe_stack_range_stack_clash): New.
3755 (rs6000_emit_prologue): Emit notes into dump file as requested.
3756 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
3757 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
3758 Add additional operand and pass it to output_probe_stack_range.
3760 2017-09-25 Bin Cheng <bin.cheng@arm.com>
3762 PR tree-optimization/82163
3763 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
3764 (checking_verify_loop_closed_ssa): New parameter.
3765 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
3766 (check_loop_closed_ssa_stmt): Delete.
3767 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
3768 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
3769 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
3772 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
3774 * brig-builtins.def: Treat HSAIL barrier builtins as
3775 setjmp/longjump style functions.
3777 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
3779 * target.def (constant_alignment): New hook.
3780 * defaults.h (CONSTANT_ALIGNMENT): Delete.
3781 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
3782 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
3783 * doc/tm.texi: Regenerate.
3784 * targhooks.h (default_constant_alignment): Declare.
3785 (constant_alignment_word_strings): Likewise.
3786 * targhooks.c (default_constant_alignment): New function.
3787 (constant_alignment_word_strings): Likewise.
3788 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
3789 instead of CONSTANT_ALIGNMENT.
3790 * varasm.c (align_variable, get_variable_align, build_constant_desc)
3791 (force_const_mem): Likewise.
3792 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
3793 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
3794 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
3795 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3796 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
3798 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
3799 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3800 constant_alignment_word_strings.
3801 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
3802 (CONSTANT_ALIGNMENT): Likewise.
3803 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3804 (arm_constant_alignment): New function.
3805 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
3806 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3807 constant_alignment_word_strings.
3808 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
3809 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3810 constant_alignment_word_strings.
3811 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
3812 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3813 (cris_constant_alignment): New function.
3814 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
3815 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3816 (epiphany_constant_alignment): New function.
3817 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
3818 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3819 constant_alignment_word_strings.
3820 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
3821 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3822 constant_alignment_word_strings.
3823 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
3824 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3825 constant_alignment_word_strings.
3826 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
3827 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
3828 * config/i386/i386.c (ix86_constant_alignment): Make static.
3829 Use the same interface as the target hook.
3830 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3831 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
3832 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3833 constant_alignment_word_strings.
3834 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
3835 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
3836 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3837 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
3838 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3839 constant_alignment_word_strings.
3840 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
3841 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3842 constant_alignment_word_strings.
3843 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
3844 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3845 constant_alignment_word_strings.
3846 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
3847 * config/microblaze/microblaze.c (microblaze_constant_alignment):
3849 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3850 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
3851 * config/mips/mips.c (mips_constant_alignment): New function.
3852 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3853 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
3854 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
3855 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3856 (mmix_constant_alignment): Make static. Use the same interface
3858 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
3859 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3860 constant_alignment_word_strings.
3861 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
3862 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3863 constant_alignment_word_strings.
3864 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
3865 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3866 constant_alignment_word_strings.
3867 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
3868 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3869 (rs6000_constant_alignment): New function.
3870 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
3871 * config/riscv/riscv.c (riscv_constant_alignment): New function.
3872 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3873 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
3874 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3875 (rs6000_constant_alignment): New function.
3876 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
3877 * config/s390/s390.c (s390_constant_alignment): New function.
3878 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3879 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
3880 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3881 constant_alignment_word_strings.
3882 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
3883 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3884 (sparc_constant_alignment): New function.
3885 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
3886 * config/spu/spu.c (spu_constant_alignment): New function.
3887 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3888 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
3889 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3890 constant_alignment_word_strings.
3891 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
3892 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3893 constant_alignment_word_strings.
3894 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
3895 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
3896 constant_alignment_word_strings.
3897 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
3898 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3899 (visium_constant_alignment): New function.
3900 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
3901 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
3902 (xtensa_constant_alignment): New function.
3903 * system.h (CONSTANT_ALIGNMENT): Poison.
3905 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
3907 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
3908 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
3909 (rs6000_builtin_valid_without_lhs): New helper function.
3910 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3911 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
3913 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
3915 * target.h (vec_perm_indices): Use unsigned short rather than
3917 (auto_vec_perm_indices): Likewise.
3918 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
3919 Use unsigned int rather than unsigned char.
3920 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
3922 2017-09-25 Richard Biener <rguenther@suse.de>
3924 * cfgloop.h (sort_sibling_loops): Declare.
3925 * cfgloop.c (sort_sibling_loops_cmp): New helper.
3926 (sort_sibling_loops): New function sorting the sibling loop list
3928 * graphite.c (graphite_transform_loops): Sort sibling loops.
3930 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
3932 * target.def (vec_perm_const_ok): Change sel parameter to
3934 * optabs-query.c (can_vec_perm_p): Update accordingly.
3935 * doc/tm.texi: Regenerate.
3936 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
3937 auto_vec_perm_indices and remove separate nelt field.
3938 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
3939 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
3940 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
3941 (aarch64_expand_vec_perm_const): Update accordingly.
3942 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
3943 to vec_perm_indices.
3944 * config/arm/arm.c (expand_vec_perm_d): Change perm to
3945 auto_vec_perm_indices and remove separate nelt field.
3946 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
3947 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
3948 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
3950 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
3951 to vec_perm_indices.
3952 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
3953 sel to vec_perm_indices.
3954 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
3955 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
3956 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
3958 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
3961 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
3964 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
3965 on the FUNCTION_DECL function context if it has a DIE that is a
3968 2017-09-25 Richard Biener <rguenther@suse.de>
3970 PR tree-optimization/82285
3971 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
3974 2017-09-25 Tom de Vries <tom@codesourcery.com>
3978 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
3981 2017-09-25 Richard Biener <rguenther@suse.de>
3983 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
3984 ISL errors other than isl_error_quota happen. Dump if the
3985 schedule is the same.
3986 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
3987 errors instead of aborting inside ISL.
3989 2017-09-25 Iain Sandoe <iain@codesourcery.com>
3992 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
3993 of libgcc_eh for m64.
3994 * config/i386/darwin64.h: Likewise.
3996 2017-09-25 Richard Biener <rguenther@suse.de>
3999 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
4000 attribute for incomplete types nor twice for complete ones.
4002 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
4005 * config/i386/i386.c (ix86_print_operand_address_as): Encode
4006 %esp as %rsp to avoid 0x67 prefix if there is no index or base
4009 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
4012 * config/i386/i386.opt (mprefer-avx256): Use
4013 ix86_target_flags variable.
4014 * config/i386/i386.c (ix86_target_string): Move
4015 -mprefer-avx256 to flag2_opts.
4017 2017-09-22 Jakub Jelinek <jakub@redhat.com>
4020 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
4021 and x != -1 | y != -1 into (x & y) != -1.
4023 2017-09-22 Steve Ellcey <sellcey@cavium.com>
4025 * config.gcc: Add new case statement to set
4026 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
4027 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
4028 s390x-*-linux* case statements. Added aarch64 to the list of
4029 supported architectures.
4031 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
4033 PR tree-optimization/82289
4034 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
4035 STMT_VINFO_RELEVANT_P.
4037 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
4038 Alan Hayward <alan.hayward@arm.com>
4039 David Sherwood <david.sherwood@arm.com>
4041 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
4042 for VR_RANGE only; don't allow VR_ANTI_RANGE.
4043 (extract_range_from_binary_expr_1): Don't call
4044 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
4046 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
4047 Alan Hayward <alan.hayward@arm.com>
4048 David Sherwood <david.sherwood@arm.com>
4050 * target.def (preferred_vector_alignment): New hook.
4051 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
4053 * doc/tm.texi: Regenerate.
4054 * targhooks.h (default_preferred_vector_alignment): Declare.
4055 * targhooks.c (default_preferred_vector_alignment): New function.
4056 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
4058 (DR_TARGET_ALIGNMENT): New macro.
4059 (aligned_access_p): Update commentary.
4060 (vect_known_alignment_in_bytes): New function.
4061 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
4063 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
4064 Calculate the misalignment based on the target alignment rather than
4066 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
4067 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
4068 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
4069 the target alignment, rather than masking the element misalignment
4070 with the number of elements in a vector. Also use the target
4071 alignment when calculating the maximum number of peels.
4072 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
4073 instead of TYPE_ALIGN_UNIT.
4074 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
4075 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
4076 (vect_create_addr_base_for_vector_ref): Update call accordingly.
4077 (vect_create_data_ref_ptr): Likewise.
4078 (vect_setup_realignment): Realign by ANDing with
4079 -DR_TARGET_MISALIGNMENT.
4080 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
4081 the number of peels based on DR_TARGET_ALIGNMENT.
4082 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
4083 with the guaranteed alignment boundary when deciding whether
4085 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
4086 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
4087 (ensure_base_align): Remove stmt_info parameter. Get the
4088 target base alignment from DR_TARGET_ALIGNMENT.
4089 (vectorizable_store): Update call accordingly. Interpret
4090 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
4092 (vectorizable_load): Likewise.
4094 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
4095 Alan Hayward <alan.hayward@arm.com>
4096 David Sherwood <david.sherwood@arm.com>
4098 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
4099 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
4100 (vect_enhance_data_refs_alignment): Likewise.
4102 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
4104 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
4105 error. Only quit immediately if parsing is complete.
4106 (BEGIN): Initialize fatal_err and parse_done.
4107 (begin fpu, end fpu): Check number of arguments.
4108 (begin arch, end arch): Likewise.
4109 (begin cpu, end cpu): Likewise.
4110 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
4111 (optalias): Likewise.
4113 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
4115 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
4116 * config/arm/arm-isa.h: Delete. Move definitions to ...
4117 * arm-cpus.in: ... here. Use new feature and fgroup values.
4118 * config/arm/arm.c (arm_option_override): Use lower case for feature
4120 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
4121 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
4122 * config/arm/parsecpu.awk (END): Add new command 'isa'.
4124 (print_isa_bits_for): New function.
4125 (gen_isa): New function.
4126 (gen_comm_data): Use print_isa_bits_for.
4127 (define feature): New keyword.
4128 (define fgroup): New keyword.
4129 * config/arm/t-arm (TM_H): Remove.
4130 (GTM_H): Add arm-isa.h.
4131 (arm-isa.h): Add rule to generate file.
4132 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
4133 case for feature bit names.
4135 2017-09-22 Richard Biener <rguenther@suse.de>
4137 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
4139 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
4140 print of no dependency loops ...
4141 * graphite.c (graphite_transform_loops): ... here.
4142 (canonicalize_loop_closed_ssa_form): Work from inner to outer
4144 (same_close_phi_node, remove_duplicate_close_phi,
4145 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
4146 (canonicalize_loop_closed_ssa): ... here and simplify.
4147 * graphite-optimize-isl.c: Include tree-vectorizer.h.
4148 (optimize_isl): Use dump_printf_loc to tell when we stopped
4149 optimizing because of an ISL timeout.
4151 2017-09-22 Richard Biener <rguenther@suse.de>
4153 PR tree-optimization/82291
4154 * tree-if-conv.c (predicate_mem_writes): Make sure to
4155 remove writes in blocks predicated with false.
4157 2017-09-22 Richard Biener <rguenther@suse.de>
4159 * sese.c: Include cfganal.h.
4160 (if_region_set_false_region): Remove.
4161 (create_if_region_on_edge): Likewise.
4162 (move_sese_in_condition): Re-implement without destroying
4165 2017-09-22 Richard Biener <rguenther@suse.de>
4167 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
4168 Verify both BBs contain loop PHI nodes before dispatching to
4170 (graphite_regenerate_ast_isl): Do not recompute dominators,
4171 do not verify three times. Restructure for clarity.
4172 * graphite-scop-detection.c (same_close_phi_node,
4173 remove_duplicate_close_phi, make_close_phi_nodes_unique,
4174 defined_in_loop_p, canonicalize_loop_closed_ssa,
4175 canonicalize_loop_closed_ssa_form): Simplify, remove excess
4176 checking and SSA rewrite, move to ...
4177 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
4178 (graphite_initialize): Do not pass in ctx, do not reset the
4179 SCEV cache, compute only dominators.
4180 (graphite_transform_loops): Allocate ISL ctx after
4181 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
4182 Maintain post-dominators only around build_scops.
4183 * sese.c (if_region_set_false_region): Make static. Free
4184 and recompute dominators.
4185 (move_sese_in_condition): Assert we don't get called with
4186 post-dominators computed.
4187 * sese.h (if_region_set_false_region): Remove.
4189 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
4191 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
4192 mode attribute for TARGET_AVX512VL.
4194 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
4196 * config/i386/i386.opt (mprefer-avx256): New option.
4197 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
4199 (ix86_preferred_simd_mode): Return 256-bit AVX modes
4200 for TARGET_PREFER_AVX256.
4201 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
4203 2017-09-21 Jeff Law <law@redhat.com>
4205 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
4206 Fix dump output if the only stack space is for pushed registers.
4208 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4210 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
4213 2017-09-21 Martin Sebor <msebor@redhat.com>
4216 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
4217 code (in C++) or code that triggers warnings.
4219 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
4221 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
4223 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
4225 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
4226 * sched-rgn.c: Ditto.
4227 * sel-sched-ir.c: Ditto.
4229 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
4231 * toplev.h (set_random_seed): Adjust return type.
4232 * toplev.c (init_local_tick): Move eager initialization of random_seed
4233 to get_random_seed. Adjust comment.
4234 (init_random_seed): Inline to get_random_seed, delete.
4235 (get_random_seed): Initialize random_seed lazily.
4236 (set_random_seed): Do not return previous value.
4237 (print_switch_value): Do not call get_random_seed.
4239 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
4241 * cgraph.c (delete_function_version): New, broken out from...
4242 (cgraph_node::delete_function_version): ...here. Rename to
4243 cgraph_node::delete_function_version_by_decl. Update all uses.
4244 (cgraph_node::remove): Call delete_function_version.
4246 2017-09-21 Jakub Jelinek <jakub@redhat.com>
4249 * tree-inline.c (expand_call_inline): Emit clobber stmts for
4250 VAR_DECLs to which addressable non-volatile parameters are mapped
4251 and for id->retvar after the return value assignment. Clear
4252 id->retval and id->retbnd after inlining.
4254 2017-09-21 Richard Biener <rguenther@suse.de>
4256 PR tree-optimization/82276
4257 PR tree-optimization/82244
4258 * tree-vrp.c (build_assert_expr_for): Set
4259 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
4261 (remove_range_assertions): Revert earlier change.
4263 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
4266 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
4268 2017-09-21 Richard Biener <rguenther@suse.de>
4270 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
4271 Restore valid IL after code generation errors.
4272 * graphite.c (graphite_transform_loops): Diagnose code
4273 generation issues as MSG_MISSED_OPTIMIZATION and continue
4274 with processing SCOPs.
4276 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4277 Alan Hayward <alan.hayward@arm.com>
4278 David Sherwood <david.sherwood@arm.com>
4280 * calls.c (compute_argument_addresses): Use simplify_gen_binary
4281 rather than choosing between plus_constant and gen_rtx_<CODE>.
4282 * expr.c (emit_push_insn): Likewise.
4283 (expand_expr_real_2): Likewise.
4285 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4286 Alan Hayward <alan.hayward@arm.com>
4287 David Sherwood <david.sherwood@arm.com>
4289 * loop-unroll.c (split_iv): Call copy_rtx on the step.
4291 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4292 Alan Hayward <alan.hayward@arm.com>
4293 David Sherwood <david.sherwood@arm.com>
4295 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
4296 calling tree_to_uhwi.
4298 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4299 Alan Hayward <alan.hayward@arm.com>
4300 David Sherwood <david.sherwood@arm.com>
4302 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
4303 INTEGER_CST rather than a negative test for ADDR_EXPR.
4305 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
4306 Alan Hayward <alan.hayward@arm.com>
4307 David Sherwood <david.sherwood@arm.com>
4309 * tree-vrp.c (extract_range_from_binary_expr_1): Check
4310 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
4312 2017-09-21 Richard Biener <rguenther@suse.de>
4314 PR tree-optimization/71351
4315 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
4316 graphite_create_new_loop_guard): Remove, fold remaining parts
4318 (translate_isl_ast_node_for): ... here and simplify.
4320 2017-09-21 Jakub Jelinek <jakub@redhat.com>
4323 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
4324 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
4325 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
4326 alternative always use QI mode, for -Os imov (=R,R) alternative
4327 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
4330 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4331 Jeff Law <law@redhat.com>
4333 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
4334 (allocate_stack_space): New function, partially extracted from
4336 (s390_emit_prologue): Track offset to most recent stack probe.
4337 Code to allocate space moved into allocate_stack_space.
4338 Dump actions when no stack is allocated.
4339 (s390_prologue_plus_offset): New function.
4340 (s390_emit_stack_probe): Likewise.
4342 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
4344 * common.opt (Wa, Wl, Wp, g, gz=): Add
4346 (gno-column-info): Remove.
4347 (gcolumn-info): Drop RejectNegative.
4349 (gno-record-gcc-switches): Remove.
4350 (grecord-gcc-switches): Drop RejectNegative.
4351 (gno-split-dwarf): Remove.
4352 (gsplit-dwarf): Drop RejectNegative.
4353 (gno-strict-dwarf): Remove.
4354 (gstrict-dwarf): Drop RejectNegative.
4355 * config/darwin.opt (gfull, gused): Add RejectNegative.
4356 * dwarf2out.c (gen_producer_string): Drop
4357 gno-record-gcc-switches handler.
4358 * optc-gen.awk: Add g to prefixes with negative forms.
4359 * opts-common.c (remapping_prefix_p): New.
4360 (find_opt): Check it.
4361 (generate_canonical_option): Test g prefix.
4362 (option_map): Add -gno- mapping.
4363 (add_misspelling_candidates): Check remapping_prefix_p.
4365 2017-09-20 Jeff Law <law@redhat.com>
4367 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
4368 thinko in stack clash protection support.
4370 * explow.c (compute_stack_clash_protection_loop_data): Use
4371 CONST_INT_P instead of explicit test. Verify object is a
4372 CONST_INT_P before looking at INTVAL.
4373 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
4374 instead of explicit test.
4376 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
4379 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
4380 address instead of to r1 and r11.
4382 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
4384 * config.gcc: Support "knm".
4385 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
4386 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
4388 * config/i386/i386.c (m_KNM): Define.
4389 (processor_target_table): Add "knm".
4391 (ix86_option_override_internal): Add "knm".
4392 (ix86_issue_rate): Add PROCESSOR_KNM.
4393 (ix86_adjust_cost): Ditto.
4394 (ia32_multipass_dfa_lookahead): Ditto.
4395 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
4396 (fold_builtin_cpu): Add M_INTEL_KNM.
4397 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
4398 (processor_type): Add PROCESSOR_KNM.
4399 * config/i386/x86-tune.def: Add m_KNM.
4400 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
4402 2017-09-20 Richard Biener <rguenther@suse.de>
4404 PR tree-optimization/80213
4405 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
4406 are allowed in empty BBs as well.
4407 (canonicalize_loop_closed_ssa): Also look for other complex
4409 (scop_detection::get_sese): Include the loop-closed PHI block
4411 (scop_detection::merge_sese): Remove code adding extra blocks.
4412 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
4413 (build_scops): Assert the final returned scop is invalid.
4415 2017-09-20 Richard Biener <rguenther@suse.de>
4417 PR tree-optimization/82264
4418 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
4420 (vn_phi_lookup): Likewise.
4421 (vn_phi_insert): Likewise.
4423 2017-09-20 Jakub Jelinek <jakub@redhat.com>
4425 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
4426 that fits into uhwi or shwi, add DW_AT_const_value regardless
4427 of early_dwarf without going through RTL, using add_AT_unsigned
4430 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
4431 (DEBUG_LTO_ABBREV_SECTION): Likewise.
4432 (DEBUG_LTO_MACINFO_SECTION): Likewise.
4433 (DEBUG_MACRO_SECTION): Likewise.
4434 (DEBUG_LTO_MACRO_SECTION): Likewise.
4435 (DEBUG_STR_DWO_SECTION): Likewise.
4436 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
4437 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
4438 (DEBUG_LTO_DWO_LINE_SECTION): Define.
4439 (DEBUG_LTO_LINE_STR_SECTION): Define.
4440 (init_sections_and_labels): Initialize debug_line_str_section
4441 variable. Initialize debug_loc_section for -gdwarf-5 to
4442 DEBUG_LOCLISTS_SECTION. Formatting fixes.
4444 2017-09-20 Richard Biener <rguenther@suse.de>
4446 * graphite-sese-to-poly.c (extract_affine): Properly handle
4447 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
4449 2017-09-20 Richard Biener <rguenther@suse.de>
4451 PR tree-optimization/81373
4452 * graphite-scop-detection.c (build_cross_bb_scalars_def):
4453 Force SESE live-out defs to be handled even if they are
4456 2017-09-19 Jeff Law <law@redhat.com>
4458 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
4459 nothing for stack adjustments with REG_STACK_CHECK.
4460 * sched-deps.c (parse_add_or_inc): Reject insns with
4461 REG_STACK_CHECK from dependency breaking.
4462 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
4463 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
4464 * reg-notes.def (STACK_CHECK): New note.
4466 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
4467 (ix86_expand_prologue): Dump stack clash info as needed.
4468 Call ix86_adjust_stack_and_probe_stack_clash as needed.
4470 * function.c (dump_stack_clash_frame_info): New function.
4471 * function.h (dump_stack_clash_frame_info): Prototype.
4472 (enum stack_clash_probes): New enum.
4474 * config/alpha/alpha.c (alpha_expand_prologue): Also check
4475 flag_stack_clash_protection.
4476 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
4477 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
4478 (arm_frame_pointer_required): Likewise.
4479 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
4480 (ia64_expand_prologue): Likewise.
4481 * config/mips/mips.c (mips_expand_prologue): Likewise.
4482 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
4483 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
4484 (sparc_flat_expand_prologue): Likewise.
4485 * config/spu/spu.c (spu_expand_prologue): Likewise.
4487 * explow.c: Include "params.h".
4488 (anti_adjust_stack_and_probe_stack_clash): New function.
4489 (get_stack_check_protect): Likewise.
4490 (compute_stack_clash_protection_loop_data): Likewise.
4491 (emit_stack_clash_protection_loop_start): Likewise.
4492 (emit_stack_clash_protection_loop_end): Likewise.
4493 (allocate_dynamic_stack_space): Use get_stack_check_protect.
4494 Use anti_adjust_stack_and_probe_stack_clash.
4495 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
4496 (emit_stack_clash_protection_loop_start): Likewise.
4497 (emit_stack_clash_protection_loop_end): Likewise.
4498 * rtl.h (get_stack_check_protect): Prototype.
4499 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
4500 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
4501 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
4503 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
4505 * doc/tm.texi: Rebuilt.
4506 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
4507 get_stack_check_protect.
4508 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
4509 * config/arm/arm.c (arm_expand_prologue): Likewise.
4510 (arm_frame_pointer_required): Likewise.
4511 * config/i386/i386.c (ix86_expand_prologue): Likewise.
4512 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
4513 * config/mips/mips.c (mips_expand_prologue): Likewise.
4514 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
4515 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
4516 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
4517 (sparc_flat_expand_prologue): Likewise.
4519 * common.opt (-fstack-clash-protection): New option.
4520 * flag-types.h (enum stack_check_type): Note difference between
4521 -fstack-check= and -fstack-clash-protection.
4522 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
4523 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
4524 * toplev.c (process_options): Issue warnings/errors for cases
4525 not handled with -fstack-clash-protection.
4526 * doc/invoke.texi (-fstack-clash-protection): Document new option.
4527 (-fstack-check): Note additional problem with -fstack-check=generic.
4528 Note that -fstack-check is primarily for Ada and refer users
4529 to -fstack-clash-protection for stack-clash-protection.
4530 Document new params for stack clash protection.
4532 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
4534 * config/i386/i386.c (ix86_split_long_move): Do not handle
4535 address used for LEA in a special way.
4537 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
4539 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
4541 2017-09-19 Martin Sebor <msebor@redhat.com>
4544 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
4545 of incompatible types.
4547 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
4549 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
4550 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
4551 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4552 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
4554 2017-09-19 Richard Biener <rguenther@suse.de>
4556 PR tree-optimization/82244
4557 * tree-vrp.c (remove_range_assertions): Do not propagate
4558 a constant to abnormals but replace the assert with a copy.
4560 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
4562 PR rtl-optimization/57878
4563 PR rtl-optimization/68988
4564 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
4565 avoidance test involving non_reload_pseudos. Move frequency test
4566 below the general fragmentation avoidance test.
4568 2017-09-19 Richard Biener <rguenther@suse.de>
4570 PR tree-optimization/69728
4571 * graphite-sese-to-poly.c (schedule_error): New global.
4572 (add_loop_schedule): Handle empty domain by failing the
4574 (build_original_schedule): Handle schedule_error.
4576 2017-09-19 Richard Biener <rguenther@suse.de>
4578 * graphite-scop-detection.c (scop_detection::can_represent_loop):
4579 Do not iterate to sibling loops but only to siblings of inner
4582 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
4585 * config/m68k/m68k.md (moveq feeding equality comparison): Check
4586 that the registers are different.
4588 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
4590 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
4591 to processor_model and "amdfam17h" to arch_names_table.
4592 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
4594 2017-09-18 Jakub Jelinek <jakub@redhat.com>
4597 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
4599 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4600 Alan Hayward <alan.hayward@arm.com>
4601 David Sherwood <david.sherwood@arm.com>
4603 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
4605 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
4607 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
4608 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
4609 vect_schedule_slp_instance.
4610 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
4611 Update call to vect_slp_analyze_node_operations. Simplify return
4613 (vect_slp_analyze_bb_1): Update call accordingly.
4614 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
4615 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
4616 (vect_schedule_slp): Update call accordingly.
4618 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4619 Alan Hayward <alan.hayward@arm.com>
4620 David Sherwood <david.sherwood@arm.com>
4622 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
4623 with types that aren't in fact scalar.
4625 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4627 * tree-vect-slp.c (vect_record_max_nunits): New function,
4629 (vect_build_slp_tree_1): ...here.
4630 (vect_build_slp_tree_2): Call it for phis too.
4632 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4634 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
4635 to vect_get_vec_def_for_operand when getting the mask operand.
4637 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4638 Alan Hayward <alan.hayward@arm.com>
4639 David Sherwood <david.sherwood@arm.com>
4641 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
4644 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4645 Alan Hayward <alan.hayward@arm.com>
4646 David Sherwood <david.sherwood@arm.com>
4648 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
4649 calculation for vector booleans.
4651 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
4652 Alan Hayward <alan.hayward@arm.com>
4653 David Sherwood <david.sherwood@arm.com>
4655 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
4657 (vect_transform_stmt): ...here.
4658 (vect_analyze_stmt): Use it instead of calling
4659 vectorizable_live_operation directly.
4661 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
4663 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
4664 non-SIMT targets in acc vector loops.
4666 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
4668 * configure.ac: Add arc and check if assembler supports gdwarf2.
4669 * configure: Regenerate.
4671 2017-09-18 Richard Biener <rguenther@suse.de>
4673 PR tree-optimization/82220
4674 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
4675 epilogue niters from the min_profitable_iters compute.
4677 2017-09-18 Jakub Jelinek <jakub@redhat.com>
4680 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
4681 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
4682 (ix86_init_pic_reg): Revert 2017-09-01 changes.
4684 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
4687 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
4688 switching to a new text section.
4690 2017-09-18 Richard Biener <rguenther@suse.de>
4692 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
4694 (build_alias_set): Reject aliases with no access function.
4696 2017-09-18 Richard Biener <rguenther@suse.de>
4698 PR tree-optimization/79622
4699 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
4701 (build_cross_bb_scalars_use): Likewise.
4703 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
4705 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
4707 2017-09-18 Alan Modra <amodra@gmail.com>
4710 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
4711 stack_pointer_rtx for count 0. Update comments. Break up
4712 large rtl expression.
4714 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
4716 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
4717 Increase to 20 bytes.
4718 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
4719 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
4720 or avx version of the stub.
4722 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
4725 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
4726 compute the minimum stack alignment. Also update preferred stack
4727 boundary for leaf functions.
4729 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
4731 PR tree-optimization/82228
4732 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
4735 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
4737 * common/config/nds32/nds32-common.c
4738 (nds32_option_optimization_table): Refine formatting.
4739 (nds32_option_optimization_table): Use -fsched-pressure and
4740 -fomit-frame-pointer for specific optimization level.
4742 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
4744 * config/nds32/nds32.c: Refine formatting and comments.
4745 * config/nds32/nds32.h: Likewise.
4746 * config/nds32/nds32.md: Likewise.
4747 * config/nds32/nds32-cost.c: Likewise.
4748 * config/nds32/nds32-isr.c: Likewise.
4749 * config/nds32/nds32-md-auxiliary.c: Likewise.
4750 * config/nds32/nds32-multiple.md: Likewise.
4751 * config/nds32/nds32-predicates.c: Likewise.
4753 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
4754 Jakub Jelinek <jakub@redhat.com>
4756 Add support for -std=c++2a.
4757 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
4759 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
4761 2017-09-15 Steve Ellcey <sellcey@cavium.com>
4764 * doc/extend.texi (Common Function Attributes): Add
4765 references to ARM, AArch64, and S/390 specific attributes.
4766 (Function Specific Option Pragmas): Add AArch64 and S/390
4767 to list of back ends that support the target pragma.
4769 2017-09-15 Nathan Sidwell <nathan@acm.org>
4771 * doc/standards.texi: Fix C++17 description. Update URLs for
4774 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
4776 * common.opt (Wcast-align=strict): New warning option.
4777 * doc/invoke.texi: Document -Wcast-align=strict.
4779 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
4781 * cgraph.h (cgraph_thunk_info): Add comments.
4782 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
4783 assert for VIRTUAL_* arguments stricter.
4785 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
4787 PR tree-optimization/71026
4788 * match.pd: Move RDIV patterns from fold-const.c
4789 * fold-const.c (distribute_real_division): Removed.
4790 (fold_binary_loc): Remove calls to distribute_real_divison.
4792 2017-09-15 Jakub Jelinek <jakub@redhat.com>
4794 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
4795 c++1z and gnu++1z as deprecated. Change other references to
4796 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
4797 Change -Wc++1z-compat to -Wc++17-compat.
4798 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
4799 * dwarf2out.c (highest_c_language): Handle C++17.
4800 (gen_compile_unit_die): Likewise.
4802 2017-09-15 Jakub Jelinek <jakub@redhat.com>
4804 PR rtl-optimization/82192
4805 * combine.c (make_extraction): Don't look through non-paradoxical
4806 SUBREGs or TRUNCATE if pos + len is or might be bigger than
4809 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
4810 Alan Hayward <alan.hayward@arm.com>
4811 David Sherwood <david.sherwood@arm.com>
4813 * target.def (function_arg_offset): New hook.
4814 * targhooks.h (default_function_arg_offset): Declare.
4815 * targhooks.c (default_function_arg_offset): New function.
4816 * function.c (locate_and_pad_parm): Use
4817 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
4818 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
4819 (TARGET_FUNCTION_ARG_OFFSET): ...this.
4820 * doc/tm.texi: Regenerate.
4821 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
4822 * config/spu/spu.c (spu_function_arg_offset): New function.
4823 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
4824 * system.h (FUNCTION_ARG_OFFSET): Poison.
4826 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
4827 Alan Hayard <alan.hayward@arm.com>
4828 David Sherwood <david.sherwood@arm.com>
4830 * target.def (truly_noop_truncation): New hook.
4831 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
4832 than TRULY_NOOP_TRUNCATION.
4833 * hooks.h (hook_bool_uint_uint_true): Declare.
4834 * hooks.c (hook_bool_uint_uint_true): New function.
4835 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
4836 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
4837 * doc/tm.texi: Regenerate.
4838 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
4839 rather than TRULY_NOOP_TRUNCATION in comments.
4840 (simplify_comparison): Likewise.
4841 (record_truncated_value): Likewise.
4842 * expmed.c (extract_bit_field_1): Likewise.
4843 (extract_split_bit_field): Likewise.
4844 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
4845 instead of TRULY_NOOP_TRUNCATION.
4846 * function.c (assign_parm_setup_block): Likewise.
4847 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
4848 * rtlhooks.c: Include target.h.
4849 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
4850 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
4851 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
4852 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
4853 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
4854 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
4855 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
4856 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
4857 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
4858 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
4859 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
4860 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
4861 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
4862 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
4863 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
4864 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
4865 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
4866 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
4867 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
4868 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
4869 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
4870 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
4871 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
4872 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
4873 * config/mips/mips.c (mips_truly_noop_truncation): New function.
4874 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
4875 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
4876 rather than TRULY_NOOP_TRUNCATION in comments.
4877 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
4878 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
4879 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
4880 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
4881 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
4882 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
4883 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
4884 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
4885 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
4886 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
4887 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
4888 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
4889 rather than TRULY_NOOP_TRUNCATION in comments.
4890 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
4891 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
4892 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
4893 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
4894 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
4895 TRULY_NOOP_TRUNCATION condition.
4896 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
4897 (TRULY_NOOP_TRUNCATION): Delete.
4898 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
4899 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
4900 * config/spu/spu.c (spu_truly_noop_truncation): New function.
4901 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
4902 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
4903 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
4904 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
4905 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
4906 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
4907 rather than TRULY_NOOP_TRUNCATION in comments.
4908 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
4909 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
4910 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
4911 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
4912 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
4913 * system.h (TRULY_NOOP_TRUNCATION): Poison.
4915 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
4918 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
4919 (*cmp_ior): Likewise.
4920 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
4921 (*ior_scc_scc_cmp): Likewise.
4922 (*and_scc_scc): Likewise.
4923 (*and_scc_scc_cmp): Likewise.
4925 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
4926 Alan Hayard <alan.hayward@arm.com>
4927 David Sherwood <david.sherwood@arm.com>
4929 * target.def (can_change_mode_class): New hook.
4930 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
4931 (hard_regno_nregs): Likewise.
4932 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
4933 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
4934 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
4935 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
4936 (LOAD_EXTEND_OP): Update accordingly.
4937 * doc/tm.texi: Regenerate.
4938 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
4939 CANNOT_CHANGE_MODE_CLASS.
4940 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
4941 (REG_CAN_CHANGE_MODE_P): ...this new macro.
4942 * combine.c (simplify_set): Update accordingly.
4943 * emit-rtl.c (validate_subreg): Likewise.
4944 * recog.c (general_operand): Likewise.
4945 * regcprop.c (mode_change_ok): Likewise.
4946 * reload1.c (choose_reload_regs): Likewise.
4947 (inherit_piecemeal_p): Likewise.
4948 * rtlanal.c (simplify_subreg_regno): Likewise.
4949 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
4950 instead of CANNOT_CHANGE_MODE_CLASS.
4951 (reload_cse_simplify_operands): Likewise.
4952 * reload.c (push_reload): Use targetm.can_change_mode_class
4953 instead of CANNOT_CHANGE_MODE_CLASS.
4954 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
4955 REG_CANNOT_CHANGE_MODE_P.
4956 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4957 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
4958 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4959 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4960 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4961 (arm_can_change_mode_class): New function.
4962 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
4963 than CANNOT_CHANGE_MODE_CLASS in comments.
4964 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4965 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
4966 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
4967 (ix86_can_change_mode_class): ...this new function, inverting the
4968 sense of the return value.
4969 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4970 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4971 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4972 (ia64_can_change_mode_class): New function.
4973 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4974 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
4975 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
4976 (m32c_can_change_mode_class): ...this new function, inverting the
4977 sense of the return value.
4978 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4979 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4980 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
4981 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
4982 (mips_can_change_mode_class): ...this new function, inverting the
4983 sense of the return value.
4984 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4985 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4986 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4987 (msp430_can_change_mode_class): New function.
4988 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4989 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
4990 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4991 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4992 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
4993 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
4994 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
4995 (pa_cannot_change_mode_class): Replace with...
4996 (pa_can_change_mode_class): ...this new function, inverting the
4997 sense of the return value.
4998 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
4999 than CANNOT_CHANGE_MODE_CLASS in comments.
5000 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5001 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
5002 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5003 (pdp11_cannot_change_mode_class): Replace with...
5004 (pdp11_can_change_mode_class): ...this new function, inverting the
5005 sense of the return value.
5006 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5007 * config/powerpcspe/powerpcspe-protos.h
5008 (rs6000_cannot_change_mode_class_ptr): Delete.
5009 * config/powerpcspe/powerpcspe.c
5010 (rs6000_cannot_change_mode_class_ptr): Delete.
5011 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5012 (rs6000_option_override_internal): Assign to
5013 targetm.can_change_mode_class instead of
5014 rs6000_cannot_change_mode_class_ptr.
5015 (rs6000_cannot_change_mode_class): Replace with...
5016 (rs6000_can_change_mode_class): ...this new function, inverting the
5017 sense of the return value.
5018 (rs6000_debug_cannot_change_mode_class): Replace with...
5019 (rs6000_debug_can_change_mode_class): ...this new function.
5020 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5021 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
5022 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5023 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5024 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
5026 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
5027 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5028 (rs6000_option_override_internal): Assign to
5029 targetm.can_change_mode_class instead of
5030 rs6000_cannot_change_mode_class_ptr.
5031 (rs6000_cannot_change_mode_class): Replace with...
5032 (rs6000_can_change_mode_class): ...this new function, inverting the
5033 sense of the return value.
5034 (rs6000_debug_cannot_change_mode_class): Replace with...
5035 (rs6000_debug_can_change_mode_class): ...this new function.
5036 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5037 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
5038 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
5039 (s390_can_change_mode_class): ...this new function, inverting the
5040 sense of the return value.
5041 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5042 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5043 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
5044 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5045 (sh_cannot_change_mode_class): Replace with...
5046 (sh_can_change_mode_class): ...this new function, inverting the
5047 sense of the return value.
5048 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5049 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5050 (sparc_can_change_mode_class): New function.
5051 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5052 * config/spu/spu.c (spu_can_change_mode_class): New function.
5053 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5054 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
5055 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
5056 (visium_can_change_mode_class): New function.
5057 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
5059 2017-09-15 Richard Biener <rguenther@suse.de>
5061 PR tree-optimization/82217
5062 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
5063 but not undefined case.
5065 2017-09-15 Jakub Jelinek <jakub@redhat.com>
5068 * postreload.c (reload_cse_simplify_operands): Skip
5069 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
5071 2017-09-15 Richard Biener <rguenther@suse.de>
5073 PR tree-optimization/68823
5074 * graphite-scop-detection.c (build_alias_set): If we have a
5075 possible dependence check whether we can handle them by just
5076 looking at the DRs DR_ACCESS_FNs.
5077 (build_scops): If build_alias_set fails, fail the SCOP.
5079 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
5081 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
5082 to support float128 built-in functions that require the ISA 3.0
5084 (BU_FLOAT128_3_HW): Likewise.
5085 (SQRTF128): Add support for the IEEE 128-bit square root and fma
5087 (FMAF128): Likewise.
5089 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
5090 support for built-in functions that need the ISA 3.0 IEEE 128-bit
5091 floating point instructions.
5092 (rs6000_invalid_builtin): Likewise.
5093 (rs6000_builtin_mask_names): Likewise.
5094 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
5095 (RS6000_BTM_FLOAT128_HW): Likewise.
5096 (RS6000_BTM_COMMON): Likewise.
5097 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
5099 * doc/extend.texi (RS/6000 built-in functions): Document the
5100 IEEE 128-bit floating point square root and fused multiply-add
5103 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
5105 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
5106 reg (r2) isn't in the set of registers defined in the prologue.
5108 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5109 Alan Hayward <alan.hayward@arm.com>
5110 David Sherwood <david.sherwood@arm.com>
5112 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
5113 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
5114 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
5115 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
5116 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
5118 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
5119 max_vectorization_factor.
5120 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
5122 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5123 Alan Hayward <alan.hayward@arm.com>
5124 David Sherwood <david.sherwood@arm.com>
5126 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
5127 (vect_worthwhile_without_simd_p): Declare.
5128 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
5129 (vectorizable_reduction): Use it.
5130 * tree-vect-stmts.c (vectorizable_shift): Likewise.
5131 (vectorizable_operation): Likewise.
5133 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5134 Alan Hayward <alan.hayward@arm.com>
5135 David Sherwood <david.sherwood@arm.com>
5137 * tree-vectorizer.h (vect_get_num_copies): New function.
5138 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
5139 * tree-vect-loop.c (vectorizable_reduction): Likewise.
5140 (vectorizable_induction): Likewise.
5141 (vectorizable_live_operation): Likewise.
5142 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
5143 (vectorizable_bswap): Likewise.
5144 (vectorizable_call): Likewise.
5145 (vectorizable_conversion): Likewise.
5146 (vectorizable_assignment): Likewise.
5147 (vectorizable_shift): Likewise.
5148 (vectorizable_operation): Likewise.
5149 (vectorizable_store): Likewise.
5150 (vectorizable_load): Likewise.
5151 (vectorizable_condition): Likewise.
5152 (vectorizable_comparison): Likewise.
5153 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
5155 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5156 Alan Hayward <alan.hayward@arm.com>
5157 David Sherwood <david.sherwood@arm.com>
5159 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
5160 of vect_init_vector.
5162 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5163 Alan Hayward <alan.hayward@arm.com>
5164 David Sherwood <david.sherwood@arm.com>
5166 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
5167 an inline wrapper that provides a location.
5168 (gimple_build_vector): Likewise.
5169 * gimple-fold.c (gimple_build_vector_from_val): New function.
5170 (gimple_build_vector): Likewise.
5171 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
5172 functions to build the initial value. Always return a gimple value.
5173 (get_initial_defs_for_reduction): Likewise. Only compute
5175 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
5176 vect_init_vector on the results from get_initial_def(s)_for_reduction.
5177 (vectorizable_induction): Use gimple_build_vector rather than
5180 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5181 Alan Hayward <alan.hayward@arm.com>
5182 David Sherwood <david.sherwood@arm.com>
5184 * target.h (vec_perm_indices): New typedef.
5185 (auto_vec_perm_indices): Likewise.
5186 * optabs-query.h: Include target.h
5187 (can_vec_perm_p): Take a vec_perm_indices *.
5188 * optabs-query.c (can_vec_perm_p): Likewise.
5189 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
5190 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
5191 * tree-vect-generic.c (lower_vec_perm): Likewise.
5192 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
5193 (vect_grouped_load_supported): Likewise.
5194 (vect_shift_permute_load_chain): Likewise.
5195 (vect_permute_store_chain): Use auto_vec_perm_indices.
5196 (vect_permute_load_chain): Likewise.
5197 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
5198 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
5199 Update uses of can_vec_perm_p.
5200 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
5201 mode with a number of elements. Take a vec_perm_indices *.
5202 (vect_create_epilog_for_reduction): Update accordingly.
5203 Use auto_vec_perm_indices.
5204 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
5205 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
5206 (vect_transform_slp_perm_load): Likewise.
5207 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
5208 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
5209 (vect_gen_perm_mask_checked): Likewise.
5210 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
5211 (vect_gen_perm_mask_checked): Likewise.
5212 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
5213 (vectorizable_store): Likewise.
5214 (vectorizable_load): Likewise.
5215 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
5216 (vectorizable_bswap): Likewise.
5218 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5219 Alan Hayward <alan.hayward@arm.com>
5220 David Sherwood <david.sherwood@arm.com>
5222 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
5223 * tree.c (build_vector): Likewise.
5224 (build_vector_from_ctor): Update accordingly.
5225 (build_vector_from_val): Likewise.
5226 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
5227 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
5228 * tree-vect-generic.c (add_rshift): Likewise.
5229 (expand_vector_divmod): Likewise.
5230 (optimize_vector_constructor): Likewise.
5231 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
5232 (vect_transform_slp_perm_load): Likewise.
5233 (vect_schedule_slp_instance): Likewise.
5234 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
5235 (vectorizable_call): Likewise.
5236 (vect_gen_perm_mask_any): Likewise. Add elements in order.
5237 * expmed.c (make_tree): Likewise.
5238 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
5239 a vector passed to build_vector.
5240 (fold_convert_const): Likewise.
5241 (exact_inverse): Likewise.
5242 (fold_ternary_loc): Likewise.
5243 (fold_relational_const): Likewise.
5244 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
5245 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
5246 (const_unop): Likewise. Store the reduction accumulator in a
5247 variable rather than an array.
5248 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
5249 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
5250 the new vector, rather than constructing it after the input arrays.
5251 (native_interpret_vector): Use auto_vec<tree> when building
5252 a vector passed to build_vector. Add elements in order.
5253 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
5254 auto_vec<tree> when building a vector passed to build_vector.
5255 (vect_create_epilog_for_reduction): Likewise.
5256 (vectorizable_induction): Likewise.
5257 (get_initial_def_for_reduction): Likewise. Fix indentation of
5259 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
5261 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
5262 passed to build_vector.
5264 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
5265 Alan Hayward <alan.hayward@arm.com>
5266 David Sherwood <david.sherwood@arm.com>
5268 * tree-core.h (tree_base::u): Add an "nelts" field.
5269 (tree_vector): Use VECTOR_CST_NELTS as the length.
5270 * tree.c (tree_size): Likewise.
5271 (make_vector): Initialize VECTOR_CST_NELTS.
5272 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
5273 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
5274 TYPE_VECTOR_SUBPARTS.
5275 * expr.c (const_vector_mask_from_tree): Consistently use "units"
5276 as the number of units, setting it from VECTOR_CST_NELTS.
5277 (const_vector_from_tree): Likewise.
5278 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
5279 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
5280 (fold_negate_expr_1): Likewise.
5281 (fold_convert_const): Likewise.
5282 (const_binop): Likewise. Differentiate the number of output and
5284 (const_unop): Likewise.
5285 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
5286 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
5287 in cases that did the opposite.
5289 2017-09-14 Richard Biener <rguenther@suse.de>
5291 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
5294 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
5296 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
5298 2017-09-14 Jakub Jelinek <jakub@redhat.com>
5301 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
5302 if and where to split a bb, except for splitting before debug insn
5303 sequences followed by non-label real insn. Delete debug insns
5304 in between basic blocks.
5306 * combine.c (make_compound_operation_int): Formatting fixes.
5308 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
5309 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5310 * config/netbsd.h (LINK_EH_SPEC): Likewise.
5311 * config/sol2.h (LINK_EH_SPEC): Likewise.
5312 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5313 * config/s390/linux.h (LINK_SPEC): Likewise.
5314 * config/freebsd.h (LINK_EH_SPEC): Likewise.
5315 * config/openbsd.h (LINK_EH_SPEC): Likewise.
5316 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5317 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
5318 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
5319 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5320 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
5321 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
5323 2017-09-13 Jakub Jelinek <jakub@redhat.com>
5325 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
5327 (ENDFILE_LINUX_SPEC): Likewise.
5328 (LINK_EH_SPEC): Likewise.
5329 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
5330 (LINK_OS_LINUX_SPEC32): Likewise.
5331 (LINK_OS_LINUX_SPEC64): Likewise.
5332 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
5333 (LINK_OS_LINUX_SPEC): Likewise.
5335 2017-09-13 Martin Liska <mliska@suse.cz>
5338 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
5339 CASE_HIGH is NULL_TREE.
5341 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
5342 Alan Hayward <alan.hayward@arm.com>
5343 David Sherwood <david.sherwood@arm.com>
5345 * target.def (secondary_memory_needed): New hook.
5346 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
5347 instead of SECONDARY_MEMORY_NEEDED.
5348 (secondary_memory_needed_mode): Likewise.
5349 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
5350 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
5351 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
5352 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
5353 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
5354 * doc/tm.texi: Regenerate.
5355 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
5356 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
5357 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5358 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
5359 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
5360 * config/i386/i386.c (inline_secondary_memory_needed): Put the
5361 mode argument first and change the reg_class arguments to reg_class_t.
5362 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
5363 Make static. Update the call to inline_secondary_memory_needed.
5364 (ix86_register_move_cost): Update the call to
5365 inline_secondary_memory_needed.
5366 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5367 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
5369 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
5370 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
5372 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
5373 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
5374 * config/mips/mips.c (mips_secondary_memory_needed): Make static
5375 and match hook interface. Add comment from mips.h.
5376 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5377 * config/mmix/mmix.md (truncdfsf2): Refer to
5378 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
5380 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
5381 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
5382 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5383 (pa_secondary_memory_needed): New function.
5384 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
5385 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
5386 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5387 (pdp11_secondary_memory_needed): Make static and match hook interface.
5388 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
5389 * config/powerpcspe/powerpcspe-protos.h
5390 (rs6000_secondary_memory_needed_ptr): Delete.
5391 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
5393 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5394 (rs6000_option_override_internal): Assign to
5395 targetm.secondary_memory_needed rather than
5396 rs6000_secondary_memory_needed_ptr.
5397 (rs6000_secondary_memory_needed): Match hook interface.
5398 (rs6000_debug_secondary_memory_needed): Likewise.
5399 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
5400 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
5401 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
5402 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5403 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
5404 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
5406 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
5407 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5408 (rs6000_option_override_internal): Assign to
5409 targetm.secondary_memory_needed rather than
5410 rs6000_secondary_memory_needed_ptr.
5411 (rs6000_secondary_memory_needed): Match hook interface.
5412 (rs6000_debug_secondary_memory_needed): Likewise.
5413 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
5414 * config/s390/s390.c (s390_secondary_memory_needed): New function.
5415 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5416 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
5417 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
5418 (sparc_secondary_memory_needed): New function.
5419 * lra-constraints.c (check_and_process_move): Refer to
5420 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
5422 (curr_insn_transform): Likewise.
5423 (process_alt_operands): Use targetm.secondary_memory_needed
5424 instead of TARGET_SECONDARY_MEMORY_NEEDED.
5425 (check_secondary_memory_needed_p): Likewise.
5426 (choose_split_class): Likewise.
5427 * reload.c: Unconditionally include code that was previously
5428 conditional on SECONDARY_MEMORY_NEEDED.
5429 (push_secondary_reload): Use targetm.secondary_memory_needed
5430 instead of TARGET_SECONDARY_MEMORY_NEEDED.
5431 (push_reload): Likewise.
5432 * reload1.c: Unconditionally include code that was previously
5433 conditional on SECONDARY_MEMORY_NEEDED.
5434 (choose_reload_regs): Use targetm.secondary_memory_needed
5435 instead of TARGET_SECONDARY_MEMORY_NEEDED.
5436 (gen_reload): Likewise.
5437 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
5439 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
5440 Alan Hayward <alan.hayward@arm.com>
5441 David Sherwood <david.sherwood@arm.com>
5443 * target.def (secondary_memory_needed_mode): New hook:
5444 * targhooks.c (default_secondary_memory_needed_mode): Declare.
5445 * targhooks.h (default_secondary_memory_needed_mode): New function.
5446 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
5447 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
5448 * doc/tm.texi: Regenerate.
5449 * lra-constraints.c (check_and_process_move): Use
5450 targetm.secondary_memory_needed_mode instead of
5451 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
5452 (curr_insn_transform): Likewise.
5453 * reload.c (get_secondary_mem): Likewise.
5454 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
5455 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
5457 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
5458 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
5459 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
5460 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
5461 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
5463 * config/powerpcspe/powerpcspe-protos.h
5464 (rs6000_secondary_memory_needed_mode): Delete.
5465 * config/powerpcspe/powerpcspe.c
5466 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
5467 (rs6000_secondary_memory_needed_mode): Make static.
5468 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
5469 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
5471 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
5473 (rs6000_secondary_memory_needed_mode): Make static.
5474 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
5475 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
5476 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
5477 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
5478 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
5480 (sparc_secondary_memory_needed_mode): New function.
5481 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
5483 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
5485 * config/aarch64/constraints.md (Umq): New constraint.
5486 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
5488 (mov<mode>): Update condition.
5490 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5492 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
5493 when bitposition is the same.
5495 2017-09-13 Richard Biener <rguenther@suse.de>
5497 * dwarf2out.c (output_die_symbol): Remove.
5498 (output_die): Do not output a DIEs symbol.
5500 2017-09-13 Richard Biener <rguenther@suse.de>
5503 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
5504 default-def to avoid breaking iterator update with the weird
5505 interaction with cgraph_update_edges_for_call_stmt_node.
5507 2017-09-13 Richard Biener <rguenther@suse.de>
5509 * tree-cfg.c (verify_gimple_assign_binary): Add verification
5510 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
5511 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
5512 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
5514 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
5516 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
5517 Disable pc relative literal load irrespective of
5518 TARGET_FIX_ERR_A53_84341 for default.
5520 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
5522 * config/sparc/sparc.c (output_return): Output the source location of
5523 the insn in the delay slot, if any.
5524 (output_sibcall): Likewise.
5526 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
5529 * common.opt (-static-pie): New alias.
5530 (shared): Negate static-pie.
5531 (-no-pie): Update help text.
5533 (static-pie): New option.
5534 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
5535 -static-pie support.
5536 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
5537 (LINK_EH_SPEC): Likewise.
5538 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
5539 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
5540 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
5541 * gcc.c (LINK_COMMAND_SPEC): Likewise.
5542 (init_gcc_specs): Likewise.
5543 (init_spec): Likewise.
5544 (display_help): Update help message for -pie.
5545 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
5548 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
5550 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
5551 (movdi_aarch64): Likewise.
5552 (movti_aarch64): Likewise.
5554 2017-09-12 Simon Wright <simon@pushface.org>
5557 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
5558 calculation of the minor version, always output as 0.
5560 2017-09-12 Jakub Jelinek <jakub@redhat.com>
5563 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
5564 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
5565 on it early, rather than manual conversion late. For
5566 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
5567 instead of performing manual conversion.
5569 2017-09-12 Carl Love <cel@us.ibm.com>
5571 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
5572 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
5574 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
5575 VMULOSW): Add definitions.
5576 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5577 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
5578 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
5579 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
5580 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
5582 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
5584 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
5586 (movti_aarch64): Likewise.
5587 (movdf_aarch64): Likewise.
5588 (movtf_aarch64): Likewise.
5589 (load_pairdi): Likewise.
5590 (store_pairdi): Likewise.
5591 (load_pairdf): Likewise.
5592 (store_pairdf): Likewise.
5593 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
5594 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
5595 (ldr_got_small_<mode>): Likewise.
5596 (ldr_got_small_28k_<mode>): Likewise.
5597 (ldr_got_tiny): Likewise.
5598 * config/aarch64/iterators.md (ldst_sz): New.
5599 (ldpstp_sz): Likewise.
5600 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
5602 (thunderx_load): Split load_8 to load_16.
5603 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
5605 (thunderx2t99_storepair_basic): Split store_8 to store_16.
5606 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
5607 (xgene1_store_pair): Split store_8 to store_16.
5608 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
5609 (falkor_st_0_st_sd): Split store_8 to store_16.
5611 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
5613 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
5614 and store1/2/3/4 to store_4/8/12/16.
5615 * config/aarch64/aarch64.md: Update for rename.
5616 * config/arm/arm.md: Likewise.: Likewise.
5617 * config/arm/arm.c: Likewise.
5618 * config/arm/thumb1.md: Likewise.
5619 * config/arm/thumb2.md: Likewise.
5620 * config/arm/vfp.md: Likewise.
5621 * config/arm/arm-generic.md: Likewise.
5622 * config/arm/arm1020e.md: Likewise.
5623 * config/arm/arm1026ejs.md: Likewise.
5624 * config/arm/arm1136jfs.md: Likewise.
5625 * config/arm/arm926ejs.md: Likewise.
5626 * config/arm/cortex-a15.md: Likewise.
5627 * config/arm/cortex-a17.md: Likewise.
5628 * config/arm/cortex-a5.md: Likewise.
5629 * config/arm/cortex-a53.md: Likewise.
5630 * config/arm/cortex-a57.md: Likewise.
5631 * config/arm/cortex-a7.md: Likewise.
5632 * config/arm/cortex-a8.md: Likewise.
5633 * config/arm/cortex-a9.md: Likewise.
5634 * config/arm/cortex-m4.md: Likewise.
5635 * config/arm/cortex-m7.md: Likewise.
5636 * config/arm/cortex-r4.md: Likewise.
5637 * config/arm/exynos-m1.md: Likewise.
5638 * config/arm/fa526.md: Likewise.
5639 * config/arm/fa606te.md: Likewise.
5640 * config/arm/fa626te.md: Likewise.
5641 * config/arm/fa726te.md: Likewise.
5642 * config/arm/fmp626.md: Likewise.
5643 * config/arm/iwmmxt.md: Likewise.
5644 * config/arm/ldmstm.md: Likewise.
5645 * config/arm/marvell-pj4.md: Likewise.
5646 * config/arm/xgene1.md: Likewise.
5647 * config/aarch64/thunderx.md: Likewise.
5648 * config/aarch64/thunderx2t99.md: Likewise.
5649 * config/aarch64/falkor.md: Likewise.
5651 2017-09-12 Martin Liska <mliska@suse.cz>
5653 * attribs.c (private_lookup_attribute): New function.
5654 * attribs.h (private_lookup_attribute): Declared here.
5655 (lookup_attribute): Called from this place.
5657 2017-09-12 Richard Biener <rguenther@suse.de>
5659 PR tree-optimization/82157
5660 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
5661 stmts with side-effects.
5663 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5664 Alan Hayward <alan.hayward@arm.com>
5665 David Sherwood <david.sherwood@arm.com>
5667 * target.def (hard_regno_nregs): New hook.
5668 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
5669 * targhooks.h (default_hard_regno_nregs): Declare.
5670 * targhooks.c (default_hard_regno_nregs): New function.
5671 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
5672 (TARGET_HARD_REGNO_NREGS): ...this hook.
5673 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
5674 (CLASS_MAX_NREGS): Likewise.
5675 * doc/tm.texi: Regenerate.
5676 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
5677 instead of HARD_REGNO_NREGS.
5678 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
5679 HARD_REGNO_NREGS in the comment.
5680 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
5681 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
5682 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
5683 Return an unsigned int.
5684 (TARGET_HARD_REGNO_NREGS): Redefine.
5685 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
5686 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
5687 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
5688 (arc_hard_regno_nregs): New function.
5689 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
5690 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
5691 (arm_hard_regno_nregs): New function.
5692 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
5693 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
5694 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
5695 (TARGET_HARD_REGNO_NREGS): Redefine.
5696 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
5697 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
5698 (HARD_REGNO_NREGS): Delete.
5699 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
5700 (cr16_hard_regno_nregs): New function.
5701 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
5702 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
5703 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
5704 (cris_hard_regno_nregs): New function.
5705 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
5706 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
5707 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
5708 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
5709 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
5710 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
5711 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
5712 (frv_hard_regno_nregs): Make static. Take and return an
5714 (frv_class_max_nregs): Remove outdated copy of documentation.
5715 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
5716 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
5717 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
5718 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
5719 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
5720 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
5721 (TARGET_HARD_REGNO_NREGS): Redefine.
5722 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
5723 (CLASS_MAX_NREGS): Update comment.
5724 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
5725 (ia64_hard_regno_nregs): New function.
5726 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
5727 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
5728 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
5729 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
5730 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
5732 (m32c_hard_regno_nregs): Likewise. Make static.
5733 (TARGET_HARD_REGNO_NREGS): Redefine.
5734 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
5735 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
5736 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
5737 (m68k_hard_regno_nregs): New function.
5738 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
5739 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
5740 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
5741 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
5742 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
5743 Take and return an unsigned int.
5744 (TARGET_HARD_REGNO_NREGS): Redefine.
5745 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
5746 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
5747 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
5748 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
5749 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
5750 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
5751 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
5752 (msp430_hard_regno_nregs): Make static. Take and return an
5754 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
5755 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
5756 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
5757 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
5758 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
5759 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
5760 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
5761 (TARGET_HARD_REGNO_NREGS): Redefine.
5762 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
5763 (PA_HARD_REGNO_NREGS): ...this.
5764 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
5765 (PA_HARD_REGNO_NREGS): ...this.
5766 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
5767 (pa_hard_regno_nregs): New function.
5768 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
5769 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
5770 (pdp11_hard_regno_nregs): New function.
5771 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
5772 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
5773 (rs6000_hard_regno_nregs_hook): New function.
5774 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
5775 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
5776 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
5777 Take and return an unsigned int. Move earlier in file.
5778 (TARGET_HARD_REGNO_NREGS): Redefine.
5779 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
5780 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
5781 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
5782 (rl78_hard_regno_nregs): Make static. Take and return an
5784 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
5785 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
5786 (rs6000_hard_regno_nregs_hook): New function.
5787 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
5788 * config/rx/rx.c (rx_hard_regno_nregs): New function.
5789 (TARGET_HARD_REGNO_NREGS): Redefine.
5790 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
5791 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
5792 instead of HARD_REGNO_NREGS.
5793 (s390_hard_regno_nregs): New function.
5794 (s390_hard_regno_mode_ok): Add comment from s390.h.
5795 (TARGET_HARD_REGNO_NREGS): Redefine.
5796 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
5797 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
5798 (sh_hard_regno_nregs): New function.
5799 (sh_pass_in_reg_p): Use it.
5800 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
5801 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
5802 (sparc_hard_regno_nregs): New function.
5803 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
5804 * config/spu/spu.c (spu_hard_regno_nregs): New function.
5805 (spu_function_arg_advance): Use it, supplying a valid register number.
5806 (TARGET_HARD_REGNO_NREGS): Redefine.
5807 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
5808 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
5809 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
5810 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
5811 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
5812 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
5813 (CLASS_MAX_NREGS): Remove copy of old documentation.
5814 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
5815 (visium_hard_regno_nregs): New function.
5816 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
5817 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
5818 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
5819 xtensa_hard_regno_nregs): New function.
5820 * system.h (HARD_REGNO_NREGS): Poison.
5822 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5824 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
5825 hard_regno_nregs instead of HARD_REGNO_NREGS.
5826 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
5827 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
5828 (c6x_expand_epilogue): Likewise.
5829 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
5830 (frv_read_iacc_argument): Likewise.
5831 * config/sh/sh.c: Include regs.h.
5832 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
5833 (regs_used): Likewise.
5834 (output_stack_adjust): Likewise.
5835 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
5836 * expmed.c: Include regs.h.
5837 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
5838 * ree.c: Include regs.h.
5839 (combine_reaching_defs): Use hard_regno_nregs instead of
5841 (add_removable_extension): Likewise.
5843 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5845 * regs.h (hard_regno_nregs): Turn into a function.
5846 (end_hard_regno): Update accordingly.
5847 * caller-save.c (setup_save_areas): Likewise.
5848 (save_call_clobbered_regs): Likewise.
5849 (replace_reg_with_saved_mem): Likewise.
5850 (insert_restore): Likewise.
5851 (insert_save): Likewise.
5852 * combine.c (can_change_dest_mode): Likewise.
5853 (move_deaths): Likewise.
5854 (distribute_notes): Likewise.
5855 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
5856 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
5857 (rs6000_split_multireg_move): Likewise.
5858 (rs6000_register_move_cost): Likewise.
5859 (rs6000_memory_move_cost): Likewise.
5860 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
5861 (rs6000_split_multireg_move): Likewise.
5862 (rs6000_register_move_cost): Likewise.
5863 (rs6000_memory_move_cost): Likewise.
5864 * cselib.c (cselib_reset_table): Likewise.
5865 (cselib_lookup_1): Likewise.
5866 * emit-rtl.c (set_mode_and_regno): Likewise.
5867 * function.c (aggregate_value_p): Likewise.
5868 * ira-color.c (setup_profitable_hard_regs): Likewise.
5869 (check_hard_reg_p): Likewise.
5870 (calculate_saved_nregs): Likewise.
5871 (assign_hard_reg): Likewise.
5872 (improve_allocation): Likewise.
5873 (calculate_spill_cost): Likewise.
5874 * ira-emit.c (modify_move_list): Likewise.
5875 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
5876 (ira_hard_reg_in_set_p): Likewise.
5877 * ira.c (setup_reg_mode_hard_regset): Likewise.
5878 (clarify_prohibited_class_mode_regs): Likewise.
5879 (check_allocation): Likewise.
5880 * lra-assigns.c (find_hard_regno_for_1): Likewise.
5881 (lra_setup_reg_renumber): Likewise.
5882 (setup_try_hard_regno_pseudos): Likewise.
5883 (spill_for): Likewise.
5884 (assign_hard_regno): Likewise.
5885 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
5886 * lra-constraints.c (in_class_p): Likewise.
5887 (lra_constraint_offset): Likewise.
5888 (simplify_operand_subreg): Likewise.
5889 (lra_constraints): Likewise.
5890 (split_reg): Likewise.
5891 (split_if_necessary): Likewise.
5892 (invariant_p): Likewise.
5893 (inherit_in_ebb): Likewise.
5894 * lra-lives.c (process_bb_lives): Likewise.
5895 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
5896 (get_hard_regs): Likewise.
5897 (do_remat): Likewise.
5898 * lra-spills.c (assign_spill_hard_regs): Likewise.
5899 * mode-switching.c (create_pre_exit): Likewise.
5900 * postreload.c (reload_combine_recognize_pattern): Likewise.
5901 * recog.c (peep2_find_free_register): Likewise.
5902 * regcprop.c (kill_value_regno): Likewise.
5903 (set_value_regno): Likewise.
5904 (copy_value): Likewise.
5905 (maybe_mode_change): Likewise.
5906 (find_oldest_value_reg): Likewise.
5907 (copyprop_hardreg_forward_1): Likewise.
5908 * regrename.c (check_new_reg_p): Likewise.
5909 (regrename_do_replace): Likewise.
5910 * reload.c (push_reload): Likewise.
5911 (combine_reloads): Likewise.
5912 (find_dummy_reload): Likewise.
5913 (operands_match_p): Likewise.
5914 (find_reloads): Likewise.
5915 (find_equiv_reg): Likewise.
5916 (reload_adjust_reg_for_mode): Likewise.
5917 * reload1.c (count_pseudo): Likewise.
5918 (count_spilled_pseudo): Likewise.
5919 (find_reg): Likewise.
5920 (clear_reload_reg_in_use): Likewise.
5921 (free_for_value_p): Likewise.
5922 (allocate_reload_reg): Likewise.
5923 (choose_reload_regs): Likewise.
5924 (reload_adjust_reg_for_temp): Likewise.
5925 (emit_reload_insns): Likewise.
5926 (delete_output_reload): Likewise.
5927 * rtlanal.c (subreg_get_info): Likewise.
5928 * sched-deps.c (sched_analyze_reg): Likewise.
5929 * sel-sched.c (init_regs_for_mode): Likewise.
5930 (mark_unavailable_hard_regs): Likewise.
5931 (choose_best_reg_1): Likewise.
5932 (verify_target_availability): Likewise.
5933 * valtrack.c (dead_debug_insert_temp): Likewise.
5934 * var-tracking.c (track_loc_p): Likewise.
5935 (emit_note_insn_var_location): Likewise.
5936 * varasm.c (make_decl_rtl): Likewise.
5937 * reginfo.c (choose_hard_reg_mode): Likewise.
5938 (init_reg_modes_target): Refer directly to
5939 this_target_regs->x_hard_regno_nregs.
5941 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5943 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
5944 instead of hard_regno_nregs.
5946 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5948 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
5949 end_hard_regno instead of hard_regno_nregs.
5950 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
5951 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
5952 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
5953 * ira-color.c (improve_allocation): Likewise.
5954 * lra-assigns.c (find_hard_regno_for_1): Likewise.
5955 * lra-lives.c (mark_regno_live): Likewise.
5956 (mark_regno_dead): Likewise.
5957 * lra-remat.c (operand_to_remat): Likewise.
5958 * lra.c (collect_non_operand_hard_regs): Likewise.
5959 * postreload.c (reload_combine_note_store): Likewise.
5960 (move2add_valid_value_p): Likewise.
5961 * reload.c (regno_clobbered_p): Likewise.
5963 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5965 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
5967 * config/v850/v850.c (v850_reorg): Likewise.
5968 * reload.c (refers_to_regno_for_reload_p): Likewise.
5969 (find_equiv_reg): Likewise.
5970 * reload1.c (reload_reg_reaches_end_p): Likewise.
5972 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
5974 * caller-save.c (add_used_regs): Use REG_NREGS instead of
5976 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
5977 * config/arm/arm.c (output_move_neon): Likewise.
5978 (arm_attr_length_move_neon): Likewise.
5979 (neon_split_vcombine): Likewise.
5980 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
5981 (c6x_mark_reg_written): Likewise.
5982 (c6x_dwarf_register_span): Likewise.
5983 * config/i386/i386.c (ix86_save_reg): Likewise.
5984 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
5985 (rws_access_reg): Likewise.
5986 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
5987 * mode-switching.c (create_pre_exit): Likewise.
5988 * ree.c (combine_reaching_defs): Likewise.
5989 (add_removable_extension): Likewise.
5990 * regcprop.c (find_oldest_value_reg): Likewise.
5991 (copyprop_hardreg_forward_1): Likewise.
5992 * reload.c (reload_inner_reg_of_subreg): Likewise.
5993 (push_reload): Likewise.
5994 (combine_reloads): Likewise.
5995 (find_dummy_reload): Likewise.
5996 (reload_adjust_reg_for_mode): Likewise.
5997 * reload1.c (find_reload_regs): Likewise.
5998 (forget_old_reloads_1): Likewise.
5999 (reload_reg_free_for_value_p): Likewise.
6000 (reload_adjust_reg_for_temp): Likewise.
6001 (emit_reload_insns): Likewise.
6002 (delete_output_reload): Likewise.
6003 * sel-sched.c (choose_best_reg_1): Likewise.
6004 (choose_best_pseudo_reg): Likewise.
6006 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6007 Alan Hayward <alan.hayward@arm.com>
6008 David Sherwood <david.sherwood@arm.com>
6010 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
6011 * target.def (slow_unaligned_access): New hook.
6012 * targhooks.h (default_slow_unaligned_access): Declare.
6013 * targhooks.c (default_slow_unaligned_access): New function.
6014 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
6015 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
6016 * doc/tm.texi: Regenerate.
6017 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
6018 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
6019 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
6021 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
6022 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
6024 (rs6000_slow_unaligned_access): New function.
6025 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
6026 (expand_block_compare): Likewise.
6027 (expand_strn_compare): Likewise.
6028 (rs6000_rtx_costs): Likewise.
6029 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
6030 (riscv_slow_unaligned_access): Likewise.
6031 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
6032 (riscv_slow_unaligned_access_p): ...this and make static.
6033 (riscv_option_override): Update accordingly.
6034 (riscv_slow_unaligned_access): New function.
6035 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
6036 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
6037 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
6038 (rs6000_slow_unaligned_access): New function.
6039 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
6040 (rs6000_rtx_costs): Likewise.
6041 * config/rs6000/rs6000-string.c (expand_block_compare)
6042 (expand_strn_compare): Use targetm.slow_unaligned_access instead
6043 of SLOW_UNALIGNED_ACCESS.
6044 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
6045 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
6046 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
6047 of SLOW_UNALIGNED_ACCESS.
6048 * expmed.c (simple_mem_bitfield_p): Likewise.
6049 * expr.c (alignment_for_piecewise_move): Likewise.
6050 (emit_group_load_1): Likewise.
6051 (emit_group_store): Likewise.
6052 (copy_blkmode_from_reg): Likewise.
6053 (emit_push_insn): Likewise.
6054 (expand_assignment): Likewise.
6055 (store_field): Likewise.
6056 (expand_expr_real_1): Likewise.
6057 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
6058 * lra-constraints.c (simplify_operand_subreg): Likewise.
6059 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
6060 * gimple-ssa-store-merging.c: Likewise in block comment at start
6062 * tree-ssa-strlen.c: Include target.h.
6063 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
6064 of SLOW_UNALIGNED_ACCESS.
6065 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
6067 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6069 PR rtl-optimization/82185
6070 * expmed.c (emit_store_flag_int): Only test tem if it has been
6073 2017-09-12 Richard Biener <rguenther@suse.de>
6076 * match.pd ((FTYPE) N CMP CST): Fix typo.
6078 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
6080 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
6082 (mips_near_type_p): Add 'short_call' attribute as a synonym
6084 * doc/extend.texi (short_call): Document new function attribute.
6086 2017-09-12 Jakub Jelinek <jakub@redhat.com>
6089 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
6090 assertion check that in the condition.
6091 (get_atomic_generic_size): Likewise. Before testing if parameter
6092 has pointer type, if it has array type, call for C++
6093 default_conversion to perform array-to-pointer conversion.
6095 2017-09-12 Richard Biener <rguenther@suse.de>
6097 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
6098 for operations we cannot scalarize.
6100 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
6102 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
6103 vectors heap vectors. Clean up comments.
6104 Make visited_bbs a reference.
6105 (profitable_jump_thread_path): Make GC
6106 vectors heap vectors. Clean up comments.
6108 (convert_and_register_jump_thread_path): Make GC vectors heap
6110 (check_subpath_and_update_thread_path): Same. Clean up comments.
6111 Make visited_bbs a reference.
6112 (handle_phi): Abstract common code to to
6113 register_jump_thread_path_if_profitable.
6114 Rename VAR_BB to DEF_BB.
6116 Make GC vectors heap vectors.
6117 Make visited_bbs a reference.
6118 (handle_assignment): Same.
6119 (register_jump_thread_path_if_profitable): New.
6120 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
6122 Make GC vectors heap vectors. Clean up comments.
6123 Make visited_bbs a reference.
6124 (find_jump_threads_backwards): Make visited_bbs live in the stack.
6125 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
6128 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
6131 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
6132 words of E_DImode object are reachable by xtensa_uimm8x4 access.
6134 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
6136 Revert r251800 and r251799.
6138 2017-09-11 Martin Jambor <mjambor@suse.cz>
6141 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
6142 arguments in advance.
6143 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
6144 use it to find predecessor edges.
6145 (naive_outof_ssa): Collect vector of predecessors.
6147 2017-09-08 Jason Merrill <jason@redhat.com>
6149 PR c++/70029 - ICE with ref-qualifier and -flto
6150 * langhooks.h (struct lang_hooks_for_types): Add
6151 copy_lang_qualifiers.
6152 * attribs.c (build_type_attribute_qual_variant): Use it.
6153 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
6155 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
6156 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
6158 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
6161 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
6162 (*mulsi3_sp64): New instruction.
6163 (mulsi3): New expander.
6165 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
6167 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
6169 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6171 * sancov.c: Include memmodel.h.
6173 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
6176 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
6179 2017-09-07 Carl Love <cel@us.ibm.com>
6181 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
6182 the sldi instruction.
6184 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
6186 * sancov.c: Include tm_p.h.
6188 2017-09-07 Jakub Jelinek <jakub@redhat.com>
6191 * output.h (switch_to_other_text_partition): New declaration.
6192 * varasm.c (switch_to_other_text_partition): New function.
6193 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
6194 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
6195 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
6196 to the other text partition before emitting LCL label and switch back
6197 after emitting the word after it.
6199 2017-09-07 Richard Biener <rguenther@suse.de>
6201 * passes.def (pass_split_crit_edges): Remove instance before PRE.
6202 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
6203 critical edges here, after loop init.
6204 (pass_data_pre): Remove PROP_no_crit_edges flags.
6205 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
6206 for valueization of call args to avoid leaking VN_TOP.
6207 (visit_use): Assert we do not visit default defs.
6208 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
6209 Use error_mark_node to more easily detect leaking VN_TOP.
6210 All default-defs are varying, not VN_TOP. Mark them visited.
6211 (run_scc_vn): Make code match comment.
6213 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
6215 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
6216 OPTION_MASK_FLOAT128_KEYWORD.
6217 (POWERPC_MASKS): Likewise.
6218 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
6219 support for the -mfloat128-type option, and make -mfloat128
6220 default on PowerPC Linux systems. Define or undefine
6221 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
6222 Define __float128 to be __ieee128 if IEEE 128-bit support is
6223 enabled, or undefine it.
6224 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
6225 Delete defining __FLOAT128_TYPE__.
6226 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
6227 -mfloat128-type option and make -mfloat128 default on PowerPC
6229 (TARGET_FLOAT128_TYPE): Likewise.
6230 (-mfloat128-type): Likewise.
6231 * config/rs6000/rs6000.c (rs6000_option_override_internal):
6232 Delete the -mfloat128-type option and make -mfloat128 default on
6233 PowerPC Linux systems. Always use __ieee128 to be the keyword for
6234 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
6235 128-bit floating point is enabled. Change tests from using
6236 -mfloat128-type to -mfloat128.
6237 (rs6000_mangle_type): Use the correct mangling for the __float128
6238 type even if normal long double is restricted to 64-bits.
6239 (floatn_mode): Enable the _Float128 type by default on VSX Linux
6241 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
6242 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
6243 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
6245 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
6246 documentation for -mfloat128.
6248 2017-09-06 Olivier Hainque <hainque@adacore.com>
6250 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
6252 2017-09-06 Wish Wu <wishwu007@gmail.com>
6253 Jakub Jelinek <jakub@redhat.com>
6255 * asan.c (initialize_sanitizer_builtins): Add
6256 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
6257 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
6258 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
6259 BT_FN_VOID_UINT64_PTR variables.
6260 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
6261 (BT_FN_VOID_UINT16_UINT16): Likewise.
6262 (BT_FN_VOID_UINT32_UINT32): Likewise.
6263 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
6264 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
6265 (BT_FN_VOID_UINT64_PTR): Likewise.
6266 * common.opt (flag_sanitize_coverage): New variable.
6267 (fsanitize-coverage=trace-pc): Remove.
6268 (fsanitize-coverage=): Add.
6269 * flag-types.h (enum sanitize_coverage_code): New enum.
6270 * fold-const.c (fold_range_test): Disable non-short-circuit
6271 optimization if flag_sanitize_coverage.
6272 (fold_truth_andor): Likewise.
6273 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
6274 * opts.c (COVERAGE_SANITIZER_OPT): Define.
6275 (coverage_sanitizer_opts): New array.
6276 (get_closest_sanitizer_option): Add OPTS argument, handle also
6277 OPT_fsanitize_coverage_.
6278 (parse_sanitizer_options): Adjusted to also handle
6279 OPT_fsanitize_coverage_.
6280 (common_handle_option): Add OPT_fsanitize_coverage_.
6281 * sancov.c (instrument_comparison, instrument_switch): New function.
6282 (sancov_pass): Add trace-cmp support.
6283 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
6284 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
6285 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
6286 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
6287 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
6288 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
6289 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
6290 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
6291 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
6292 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
6294 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
6296 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
6297 error. Only quit immediately if parsing is complete.
6298 (BEGIN): Initialize fatal_err and parse_done.
6299 (begin fpu, end fpu): Check number of arguments.
6300 (begin arch, end arch): Likewise.
6301 (begin cpu, end cpu): Likewise.
6302 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
6303 (optalias): Likewise.
6305 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
6307 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
6308 * config/arm/arm-isa.h: Delete. Move definitions to ...
6309 * arm-cpus.in: ... here. Use new feature and fgroup values.
6310 * config/arm/arm.c (arm_option_override): Use lower case for feature
6312 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
6313 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
6314 * config/arm/parsecpu.awk (END): Add new command 'isa'.
6316 (print_isa_bits_for): New function.
6317 (gen_isa): New function.
6318 (gen_comm_data): Use print_isa_bits_for.
6319 (define feature): New keyword.
6320 (define fgroup): New keyword.
6321 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
6322 (arm-isa.h): Add rule to generate file.
6323 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
6324 case for feature bit names.
6326 2017-09-06 Richard Biener <rguenther@suse.de>
6328 * tree-ssa-pre.c (NECESSARY): Remove.
6329 (create_expression_by_pieces): Do not touch pass-local flags.
6330 (insert_into_preds_of_block): Likewise.
6331 (do_pre_regular_insertion): Likewise.
6332 (eliminate_insert): Likewise.
6333 (eliminate_dom_walker::before_dom_children): Likewise.
6334 (fini_eliminate): Do not look at inserted_exprs.
6335 (mark_operand_necessary): Remove.
6336 (remove_dead_inserted_code): Replace with simple work-list
6337 algorithm based on inserted_exprs and SSA uses.
6338 (pass_pre::execute): Re-order fini_eliminate and
6339 remove_dead_inserted_code.
6341 2017-09-06 Olivier Hainque <hainque@adacore.com>
6343 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
6344 for VxWorks 7. Adjust surrounding comments.
6346 2017-09-06 Richard Biener <rguenther@suse.de>
6348 * gimple-ssa-strength-reduction.c
6349 (find_candidates_dom_walker::before_dom_children): Also allow
6352 2017-09-06 Richard Biener <rguenther@suse.de>
6354 PR tree-optimization/82108
6355 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
6356 for gap in the non-permutation SLP case.
6358 2017-09-06 Martin Jambor <mjambor@suse.cz>
6360 PR tree-optimization/82078
6361 * tree-sra.c (sort_and_splice_var_accesses): Move call to
6362 add_access_to_work_queue...
6363 (build_accesses_from_assign): ...here.
6364 (propagate_all_subaccesses): Make sure racc is the group
6365 representative, if there is one.
6367 2017-09-06 Jakub Jelinek <jakub@redhat.com>
6370 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
6373 2017-09-06 Richard Biener <rguenther@suse.de>
6375 * gimple-ssa-strength-reduction.c
6376 (find_candidates_dom_walker::before_doom_children): Use a
6377 type and not a mode check.
6379 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
6382 * config/arm/predicates.md (arm_general_adddi_operand): Create new
6384 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
6386 2017-09-05 Jeff Law <law@redhat.com>
6388 PR tree-optimization/64910
6389 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
6390 cases where we have 3 or more operands.
6392 2017-09-05 Jakub Jelinek <jakub@redhat.com>
6395 * omp-low.c (lower_omp_for): Recompute tree invariant if
6396 gimple_omp_for_initial/final is ADDR_EXPR.
6399 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
6400 into gimple val before gimplification fo the COND_EXPR.
6402 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
6404 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
6405 REGION_COPY argument.
6406 (thread_through_all_blocks): Remove unused argument to
6407 duplicate_thread_path.
6409 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6410 Alan Hayward <alan.hayward@arm.com>
6411 David Sherwood <david.sherwood@arm.com>
6413 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
6414 Take a scalar_mode rather than a machine_mode.
6415 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
6416 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
6417 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
6418 (aarch64_gen_adjusted_ldpstp): Likewise.
6419 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
6421 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6422 Alan Hayward <alan.hayward@arm.com>
6423 David Sherwood <david.sherwood@arm.com>
6425 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
6426 Take a scalar_int_mode instead of a machine_mode.
6427 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
6428 (aarch64_output_scalar_simd_mov_immediate): Likewise.
6429 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
6430 (aarch64_simd_attr_length_rglist): Delete.
6431 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
6432 a scalar_int_mode instead of a machine_mode.
6433 (aarch64_add_offset): Likewise.
6434 (aarch64_internal_mov_immediate): Likewise
6435 (aarch64_add_constant_internal): Likewise.
6436 (aarch64_add_constant): Likewise.
6437 (aarch64_movw_imm): Likewise.
6438 (aarch64_rtx_arith_op_extract_p): Likewise.
6439 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
6440 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
6441 Remove assert that the mode isn't a vector.
6442 (aarch64_output_scalar_simd_mov_immediate): Likewise.
6443 (aarch64_expand_mov_immediate): Update calls after above changes.
6444 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
6445 (aarch64_and_bitmask_imm): Check for scalar integer modes.
6446 (aarch64_move_imm): Likewise.
6447 (aarch64_can_const_movi_rtx_p): Likewise.
6448 (aarch64_strip_extend): Likewise.
6449 (aarch64_extr_rtx_p): Likewise.
6450 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
6451 a CONST_INT when the mode parameter is VOIDmode.
6452 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
6454 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6456 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
6457 * stor-layout.c (bitwise_mode_for_mode): Likewise.
6458 (bitwise_type_for_mode): Update accordingly.
6460 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6462 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
6463 * stor-layout.c (mode_for_size_tree): Likewise.
6464 (mode_for_array): Update accordingly.
6465 (layout_decl): Likewise.
6466 (compute_record_mode): Likewise. Only set the mode once.
6468 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6470 * target.def (get_mask_mode): Change return type to opt_mode.
6472 * doc/tm.texi: Regenerate.
6473 * targhooks.h (default_get_mask_mode): Return an opt_mode.
6474 * targhooks.c (default_get_mask_mode): Likewise.
6475 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
6476 * optabs-query.c (can_vec_mask_load_store_p): Update use of
6477 targetm.get_mask_mode.
6478 * tree.c (build_truth_vector_type): Likewise.
6480 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6482 * machmode.h (mode_for_vector): Return an opt_mode.
6483 * stor-layout.c (mode_for_vector): Likewise.
6484 (mode_for_int_vector): Update accordingly.
6485 (layout_type): Likewise.
6486 * config/i386/i386.c (emit_memmov): Likewise.
6487 (ix86_expand_set_or_movmem): Likewise.
6488 (ix86_expand_vector_init): Likewise.
6489 (ix86_get_mask_mode): Likewise.
6490 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
6492 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
6493 * expmed.c (extract_bit_field_1): Likewise.
6494 * expr.c (expand_expr_real_2): Likewise.
6495 * optabs-query.c (can_vec_perm_p): Likewise.
6496 (can_vec_mask_load_store_p): Likewise.
6497 * optabs.c (expand_vec_perm): Likewise.
6498 * targhooks.c (default_get_mask_mode): Likewise.
6499 * tree-vect-stmts.c (vectorizable_store): Likewise.
6500 (vectorizable_load): Likewise.
6501 (get_vectype_for_scalar_type_and_size): Likewise.
6503 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6505 * machmode.h (mode_for_int_vector): New function.
6506 * stor-layout.c (mode_for_int_vector): Likewise.
6507 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
6508 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
6509 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
6510 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
6511 (s390_expand_vcond): Likewise.
6513 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6515 * machmode.h (opt_machine_mode): New type.
6516 (opt_mode<T>): Allow construction from anything that can be
6518 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
6519 (mode_for_size): Return an opt_machine_mode.
6520 * stor-layout.c (mode_for_size): Likewise.
6521 (mode_for_size_tree): Update call accordingly.
6522 (bitwise_mode_for_mode): Likewise.
6523 (make_fract_type): Likewise.
6524 (make_accum_type): Likewise.
6525 * caller-save.c (replace_reg_with_saved_mem): Update call
6527 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
6528 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
6529 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
6530 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
6531 * expmed.c (extract_bit_field_1): Likewise.
6532 * reload.c (get_secondary_mem): Likewise.
6533 * varasm.c (assemble_integer): Likewise.
6534 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
6537 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6539 * machmode.h (decimal_float_mode_for_size): New function.
6540 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
6541 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
6542 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
6543 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
6544 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
6545 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
6547 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6549 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
6550 (get_builtin_sync_mode): Likewise.
6551 (expand_ifn_atomic_compare_exchange): Likewise.
6552 (expand_builtin_atomic_clear): Likewise.
6553 (expand_builtin_atomic_test_and_set): Likewise.
6554 (fold_builtin_atomic_always_lock_free): Likewise.
6555 * calls.c (compute_argument_addresses): Likewise.
6556 (emit_library_call_value_1): Likewise.
6557 (store_one_arg): Likewise.
6558 * combine.c (combine_instructions): Likewise.
6559 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
6560 * config/arm/arm.c (arm_function_value): Likewise.
6561 (aapcs_allocate_return_reg): Likewise.
6562 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
6563 * config/i386/i386.c (construct_container): Likewise.
6564 (ix86_gimplify_va_arg): Likewise.
6565 (ix86_expand_sse_cmp): Likewise.
6566 (emit_memmov): Likewise.
6567 (emit_memset): Likewise.
6568 (expand_small_movmem_or_setmem): Likewise.
6569 (ix86_expand_pextr): Likewise.
6570 (ix86_expand_pinsr): Likewise.
6571 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
6572 * config/microblaze/microblaze.c (microblaze_block_move_straight):
6574 * config/mips/mips.c (mips_function_value_1) Likewise.
6575 (mips_block_move_straight): Likewise.
6576 (mips_expand_ins_as_unaligned_store): Likewise.
6577 * config/powerpcspe/powerpcspe.c
6578 (rs6000_darwin64_record_arg_advance_flush): Likewise.
6579 (rs6000_darwin64_record_arg_flush): Likewise.
6580 * config/rs6000/rs6000.c
6581 (rs6000_darwin64_record_arg_advance_flush): Likewise.
6582 (rs6000_darwin64_record_arg_flush): Likewise.
6583 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
6584 (sparc_function_value_1): Likewise.
6585 * config/spu/spu.c (adjust_operand): Likewise.
6586 (spu_emit_branch_or_set): Likewise.
6587 (arith_immediate_p): Likewise.
6588 * emit-rtl.c (gen_lowpart_common): Likewise.
6589 * expr.c (expand_expr_real_1): Likewise.
6590 * function.c (assign_parm_setup_block): Likewise.
6591 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
6592 * reload1.c (alter_reg): Likewise.
6593 * stor-layout.c (mode_for_vector): Likewise.
6594 (layout_type): Likewise.
6596 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
6598 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
6599 (spu_convert_move): Likewise.
6600 * lower-subreg.c (resolve_simple_move): Likewise.
6602 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6605 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
6606 define_insn to a define_expand.
6607 (altivec_vsum2sws_direct): New define_insn.
6608 (altivec_vsumsws): Convert from a define_insn to a define_expand.
6610 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
6612 * config/arm/arm.c (arm_option_params_internal): Improve setting of
6615 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
6620 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
6621 realign stack if stack alignment needed is less than incoming
6624 2017-09-05 Marek Polacek <polacek@redhat.com>
6627 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
6630 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
6632 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
6634 2017-09-05 Richard Biener <rguenther@suse.de>
6636 PR tree-optimization/82084
6637 * fold-const.c (can_native_encode_string_p): Handle wide characters.
6639 2017-09-05 Richard Biener <rguenther@suse.de>
6641 PR tree-optimization/82102
6642 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
6644 2017-09-05 Martin Liska <mliska@suse.cz>
6646 PR tree-optimization/82032
6647 * tree-cfg.c (generate_range_test): New function.
6648 * tree-cfg.h (generate_range_test): Declared here.
6649 * tree-cfgcleanup.c (convert_single_case_switch): New function.
6650 (cleanup_control_expr_graph): Use it.
6651 * tree-switch-conversion.c (try_switch_expansion): Remove
6653 (emit_case_nodes): Use generate_range_test.
6655 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
6658 * config/i386/i386.md (*<btsc><mode>_mask): Add
6659 TARGET_USE_BT to insn constraint.
6660 (*btr<mode>_mask): Ditto.
6662 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
6664 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
6665 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
6667 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
6670 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
6671 TARGET_NEON and TARGET_IWMMXT.
6672 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
6673 TARGET_NEON and TARGET_IWMMXT.
6674 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
6676 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
6678 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
6679 (ix86_rewrite_tls_address): Ditto.
6680 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
6681 (ix86_rewrite_tls_address_1): Ditto.
6682 (ix86_rewrite_tls_address): Ditto.
6683 * config/i386/predicates.md (tls_address_pattern): New predicate.
6684 * config/i386/i386.md (TLS address splitter): New splitter.
6686 2017-09-04 Richard Biener <rguenther@suse.de>
6688 PR tree-optimization/82084
6689 * fold-const.h (can_native_encode_string_p): Declare.
6690 * fold-const.c (can_native_encode_string_p): Factor out from ...
6691 (native_encode_string): ... here.
6692 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
6693 vectorizing stores from constants we later cannot handle.
6695 2017-09-04 Marek Polacek <polacek@redhat.com>
6698 * doc/invoke.texi: Update -Wtautological-compare documentation.
6700 2017-09-04 Jeff Law <law@redhat.com>
6702 PR tree-optimization/64910
6703 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
6704 swap the first and last operand if the last is a constant.
6706 2017-09-04 Marek Polacek <polacek@redhat.com>
6709 * convert.c (do_narrow): When sanitizing signed integer overflows,
6710 bail out for signed types.
6711 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
6713 2017-09-04 Richard Biener <rguenther@suse.de>
6715 PR tree-optimization/82060
6716 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
6717 Move devirtualization after stmt folding and before EH/AB/noreturn
6718 cleanup to get the stmt refs canonicalized. Use a bool instead
6719 of gimple_modified_p since that doesn't work for NOPs. Schedule
6720 NOPs generated by folding for removal.
6722 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6723 Alan Hayward <alan.hayward@arm.com>
6724 David Sherwood <david.sherwood@arm.com>
6726 * coretypes.h (pad_direction): New enum.
6727 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
6728 (FUNCTION_ARG_PADDING): Likewise.
6729 * target.def (function_arg_padding): New hook.
6730 * targhooks.h (default_function_arg_padding): Declare.
6731 * targhooks.c (default_function_arg_padding): New function.
6732 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
6733 (TARGET_FUNCTION_ARG_PADDING): ...this.
6734 * doc/tm.texi: Regenerate.
6735 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
6736 instead of direction.
6737 (compute_argument_addresses): Likewise.
6738 (load_register_parameters): Likewise.
6739 (emit_library_call_value_1): Likewise.
6740 (store_one_arg): Use targetm.calls.function_arg_padding instead
6741 of FUNCTION_ARG_PADDING.
6742 (must_pass_in_stack_var_size_or_pad): Likewise.
6743 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
6744 (emit_group_store): Likewise.
6745 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
6746 instead of FUNCTION_ARG_PADDING.
6747 (emit_push_insn): Likewise, and propagate enum change throughout
6749 * function.h (direction): Delete.
6750 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
6752 * function.c (assign_parm_find_stack_rtl): Likewise.
6753 (assign_parm_setup_block_p): Likewise.
6754 (assign_parm_setup_block): Likewise.
6755 (gimplify_parameters): Likewise.
6756 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
6757 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
6759 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
6760 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
6761 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
6762 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
6763 (aarch64_function_arg_padding): ...this new function.
6764 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
6765 (TARGET_FUNCTION_ARG_PADDING): Redefine.
6766 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
6767 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
6768 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
6769 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6770 (arm_pad_arg_upward): Replace with...
6771 (arm_function_arg_padding): ...this new function.
6772 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
6774 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
6775 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
6776 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6777 (ia64_hpux_function_arg_padding): Replace with...
6778 (ia64_function_arg_padding): ...this new function. Use pad_direction
6779 instead of direction. Check for TARGET_HPUX.
6780 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
6781 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6782 (iq2000_function_arg_padding): New function.
6783 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
6784 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
6785 (mips_function_arg_padding): ...this new function.
6786 (mips_pad_reg_upward): Update accordingly.
6787 (TARGET_FUNCTION_ARG_PADDING): Redefine.
6788 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
6789 targetm.calls.function_arg_padding.
6790 (FUNCTION_ARG_PADDING): Delete.
6791 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
6792 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
6793 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
6794 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
6795 (nios2_block_reg_padding): Return pad_direction instead of direction.
6796 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
6797 instead of direction.
6798 (nios2_function_arg_padding): Likewise. Make static.
6799 (TARGET_FUNCTION_ARG_PADDING): Redefine.
6800 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
6801 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
6802 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
6803 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6804 (pa_function_arg_padding): Make static. Return pad_direction instead
6806 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
6807 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
6808 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
6809 instead of direction. Use targetm.calls.function_arg_padding.
6810 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
6811 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
6812 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
6813 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
6814 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
6816 (function_arg_padding): Rename to...
6817 (rs6000_function_arg_padding): ...this. Make static. Return
6818 pad_direction instead of direction.
6819 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
6820 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
6821 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
6822 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
6823 instead of direction. Use targetm.calls.function_arg_padding.
6824 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
6825 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
6826 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
6827 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
6828 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6829 (function_arg_padding): Rename to...
6830 (rs6000_function_arg_padding): ...this. Make static. Return
6831 pad_direction instead of direction.
6832 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
6833 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
6834 * config/s390/s390.c (s390_function_arg_padding): New function.
6835 (TARGET_FUNCTION_ARG_PADDING): Redefine.
6836 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
6837 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
6838 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
6839 (function_arg_padding): Rename to...
6840 (sparc_function_arg_padding): ...this. Make static. Return
6841 pad_direction instead of direction.
6842 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
6843 * config/spu/spu.c (spu_function_arg_padding): New function.
6844 (TARGET_FUNCTION_ARG_PADDING): Redefine.
6845 * system.h (FUNCTION_ARG_PADDING): Poison.
6847 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
6848 Alan Hayward <alan.hayward@arm.com>
6849 David Sherwood <david.sherwood@arm.com>
6851 * target.def (modes_tieable_p): New hook.
6852 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
6853 (TARGET_MODES_TIEABLE_P): ...this.
6854 * doc/tm.texi.in: Regenerate.
6855 * hooks.h (hook_bool_mode_mode_true): Declare.
6856 * hooks.c (hook_bool_mode_mode_true): New function.
6857 * combine.c (subst): Use targetm.modes_tieable_p instead of
6859 * dse.c (find_shift_sequence): Likewise.
6860 * expmed.c (extract_low_bits): Likewise.
6861 * lower-subreg.c: Include target.h.
6862 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
6864 * rtlanal.c (rtx_cost): Likewise.
6865 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
6866 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
6867 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
6868 (TARGET_MODES_TIEABLE_P): Redefine.
6869 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
6870 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
6871 (TARGET_MODES_TIEABLE_P): Redefine.
6872 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
6873 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
6874 (arc_modes_tieable_p): New function.
6875 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
6876 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
6877 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
6878 (arm_modes_tieable_p): Make static.
6879 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
6880 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
6881 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
6882 (TARGET_MODES_TIEABLE_P): Redefine.
6883 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
6884 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
6885 (TARGET_MODES_TIEABLE_P): Redefine.
6886 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
6887 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
6888 (cr16_modes_tieable_p): New function.
6889 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
6890 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
6891 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
6892 (TRULY_NOOP_TRUNCATION): Update comment.
6893 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
6894 (TRULY_NOOP_TRUNCATION): Update comment.
6895 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
6896 (frv_modes_tieable_p): New function.
6897 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
6898 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
6899 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
6900 (TARGET_MODES_TIEABLE_P): Redefine.
6901 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
6902 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
6903 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
6904 (TARGET_MODES_TIEABLE_P): Redefine.
6905 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
6906 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
6907 (ia64_modes_tieable_p): New function.
6908 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
6909 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
6910 (iq2000_modes_tieable_p): New function.
6911 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
6912 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
6913 (lm32_modes_tieable_p): New function.
6914 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
6915 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
6916 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
6917 (TARGET_MODES_TIEABLE_P): Redefine.
6918 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
6919 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
6920 (m32r_modes_tieable_p): New function.
6921 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
6922 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
6923 (m68k_modes_tieable_p): New function.
6924 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
6925 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
6926 (mcore_modes_tieable_p): New function.
6927 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
6928 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
6930 (TARGET_MODES_TIEABLE_P): Redefine.
6931 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
6932 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
6933 * config/mips/mips.c (mips_modes_tieable_p): Make static.
6934 (TARGET_MODES_TIEABLE_P): Redefine.
6935 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
6936 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
6937 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
6938 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
6939 (mn10300_modes_tieable_p): ...this and make static.
6940 (TARGET_MODES_TIEABLE_P): Redefine.
6941 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
6942 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
6943 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
6944 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
6945 (msp430_modes_tieable_p): Make static.
6946 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
6947 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
6948 (TARGET_MODES_TIEABLE_P): Redefine.
6949 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
6950 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
6951 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
6952 (TARGET_MODES_TIEABLE_P): Redefine.
6953 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
6954 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
6955 * config/pa/pa.c (pa_modes_tieable_p): Make static.
6956 (TARGET_MODES_TIEABLE_P): Redefine.
6957 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
6958 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
6959 (pdp11_modes_tieable_p): New function.
6960 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
6961 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
6962 (rs6000_modes_tieable_p): New function.
6963 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
6964 * config/powerpcspe/powerpcspe.md: Update comment.
6965 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
6966 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
6967 (TARGET_MODES_TIEABLE_P): Redefine.
6968 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
6969 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
6970 (rl78_modes_tieable_p): New function.
6971 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
6972 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
6973 (rs6000_modes_tieable_p): New function.
6974 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
6975 * config/rs6000/rs6000.md: Update comment.
6976 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
6977 * config/rx/rx.c (rx_modes_tieable_p): New function.
6978 (TARGET_MODES_TIEABLE_P): Redefine.
6979 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
6980 * config/s390/s390.c (s390_modes_tieable_p): New function.
6981 (TARGET_MODES_TIEABLE_P): Redefine.
6982 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
6983 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
6984 (sh_modes_tieable_p): New function.
6985 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
6986 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
6987 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
6988 (sparc_modes_tieable_p): Make static.
6989 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
6990 * config/spu/spu.c (spu_modes_tieable_p): New function.
6991 (TARGET_MODES_TIEABLE_P): Redefine.
6992 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
6993 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
6994 (TARGET_MODES_TIEABLE_P): Redefine.
6995 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
6996 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
6997 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
6998 * config/v850/v850.c (v850_modes_tieable_p): New function.
6999 (TARGET_MODES_TIEABLE_P): Redefine.
7000 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
7001 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
7002 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
7003 (visium_modes_tieable_p): New function.
7004 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
7005 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
7006 (xtensa_modes_tieable_p): New function.
7007 * system.h (MODES_TIEABLE_P): Poison.
7009 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7010 Alan Hayward <alan.hayward@arm.com>
7011 David Sherwood <david.sherwood@arm.com>
7013 * target.def (hard_regno_mode_ok): New hook.
7014 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
7015 (TARGET_HARD_REGNO_MODE_OK): ...this.
7016 * doc/tm.texi.in: Regenerate.
7017 * hooks.h (hook_bool_uint_mode_true): Declare.
7018 * hooks.c (hook_bool_uint_mode_true): New function.
7019 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
7021 * genpreds.c (write_insn_preds_c): Add an include of target.h.
7022 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
7023 instead of HARD_REGNO_MODE_OK.
7024 * caller-save.c: Include target.h.
7025 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
7027 * combine.c (can_combine_p): Likewise.
7028 (combinable_i3pat): Likewise.
7029 (can_change_dest_mode): Likewise.
7030 * expr.c (init_expr_target): Likewise.
7031 (convert_move): Likewise.
7032 (convert_modes): Likewise.
7033 * ira.c (setup_prohibited_class_mode_regs): Likewise.
7034 (setup_prohibited_mode_move_regs): Likewise.
7035 * ira.h (target_ira): Likewise.
7036 * lra-assigns.c (find_hard_regno_for_1): Likewise.
7037 * lra-constraints.c (process_alt_operands): Likewise.
7038 (split_reg): Likewise.
7039 * recog.c (peep2_find_free_register): Likewise.
7040 * ree.c (combine_reaching_defs): Likewise.
7041 * regcprop.c (maybe_mode_change): Likewise.
7042 * reginfo.c (init_reg_sets_1): Likewise.
7043 (choose_hard_reg_mode): Likewise.
7044 (simplifiable_subregs): Likewise.
7045 * regrename.c (check_new_reg_p): Likewise.
7046 * reload.c (find_valid_class): Likewise.
7047 (find_valid_class_1): Likewise.
7048 (reload_inner_reg_of_subreg): Likewise.
7049 (push_reload): Likewise.
7050 (combine_reloads): Likewise.
7051 (find_dummy_reload): Likewise.
7052 (find_reloads): Likewise.
7053 * reload1.c (find_reg): Likewise.
7054 (set_reload_reg): Likewise.
7055 (allocate_reload_reg): Likewise.
7056 (choose_reload_regs): Likewise.
7057 (reload_adjust_reg_for_temp): Likewise.
7058 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
7059 (simplify_subreg_regno): Likewise.
7060 * sel-sched.c (init_regs_for_mode): Likewise.
7061 * varasm.c (make_decl_rtl): Likewise.
7062 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
7063 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
7065 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
7066 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
7067 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7068 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
7069 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
7070 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7071 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
7072 (arc_mode_class): Delete.
7073 (HARD_REGNO_MODE_OK): Delete.
7074 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7075 (arc_hard_regno_mode_ok): Rename old array to...
7076 (arc_hard_regno_mode_ok_modes): ...this.
7077 (arc_conditional_register_usage): Update accordingly.
7078 (arc_mode_class): Make static.
7079 (arc_hard_regno_mode_ok): New function.
7080 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
7081 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
7082 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7083 (arm_hard_regno_mode_ok): Make static.
7084 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
7086 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
7087 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
7088 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
7090 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7091 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
7092 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
7093 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
7094 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
7095 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7096 * config/bfin/predicates.md (valid_reg_operand): Use
7097 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
7098 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
7099 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
7100 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7101 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
7102 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
7103 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7104 (cr16_hard_regno_mode_ok): Make static and return a bool.
7105 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
7106 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7107 (cris_hard_regno_mode_ok): New function.
7108 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
7109 (epiphany_mode_class): Delete.
7110 (HARD_REGNO_MODE_OK): Delete.
7111 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
7112 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7113 (hard_regno_mode_ok): Rename to...
7114 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
7115 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
7116 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
7118 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
7119 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
7120 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7121 (frv_hard_regno_mode_ok): Make static and return a bool.
7122 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
7124 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
7125 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
7126 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
7127 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
7129 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7130 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
7131 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
7132 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
7134 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7135 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
7136 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7137 (ia64_hard_regno_mode_ok): New function.
7138 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
7139 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7140 (iq2000_hard_regno_mode_ok): New function.
7141 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
7142 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7143 (lm32_hard_regno_mode_ok): New function.
7144 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
7145 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
7146 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
7147 instead of HARD_REGNO_MODE_OK.
7148 (m32c_hard_regno_ok): Rename to...
7149 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
7150 (m32c_cannot_change_mode_class): Update accordingly.
7151 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7152 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
7153 (m32r_mode_class): Delete.
7154 (HARD_REGNO_MODE_OK): Delete.
7155 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7156 (m32r_hard_regno_mode_ok): Rename to...
7157 (m32r_hard_regno_modes): ...this.
7158 (m32r_mode_class): Make static.
7159 (m32r_hard_regno_mode_ok): New function.
7160 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
7161 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
7162 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7163 (m68k_hard_regno_mode_ok): Make static.
7164 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
7165 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7166 (mcore_hard_regno_mode_ok): New function.
7167 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
7168 (HARD_REGNO_MODE_OK): Delete.
7169 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
7171 (microblaze_hard_regno_mode_ok_p): ...this and make static.
7172 (microblaze_hard_regno_mode_ok): New function.
7173 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7174 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
7175 (mips_hard_regno_mode_ok): Delete.
7176 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
7177 (mips_hard_regno_mode_ok_p): ...this and make static.
7178 (mips_hard_regno_mode_ok_p): Rename to...
7179 (mips_hard_regno_mode_ok_uncached): ...this.
7180 (mips_hard_regno_mode_ok): New function.
7181 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
7182 of HARD_REGNO_MODE_OK.
7183 (mips_option_override): Update after above name changes.
7184 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7185 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
7186 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
7187 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
7188 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
7189 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7190 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
7191 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
7192 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
7193 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7194 (msp430_hard_regno_mode_ok): Make static and return a bool.
7195 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
7196 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
7197 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
7199 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7200 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
7201 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
7202 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
7203 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
7204 (PA_HARD_REGNO_MODE_OK): ...this
7205 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
7206 (PA_HARD_REGNO_MODE_OK): ...this.
7207 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7208 (pa_hard_regno_mode_ok): New function.
7209 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
7210 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7211 (pdp11_hard_regno_mode_ok): New function.
7212 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
7213 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
7215 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
7217 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7218 (rs6000_hard_regno_mode_ok): Rename to...
7219 (rs6000_hard_regno_mode_ok_uncached): ...this.
7220 (rs6000_init_hard_regno_mode_ok): Update accordingly.
7221 (rs6000_hard_regno_mode_ok): New function.
7222 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
7223 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
7224 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
7225 (riscv_hard_regno_mode_ok): ...this and make static.
7226 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7227 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
7228 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
7229 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7230 (rl78_hard_regno_mode_ok): Make static and return bool.
7231 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
7232 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
7234 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
7235 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7236 (rs6000_hard_regno_mode_ok): Rename to...
7237 (rs6000_hard_regno_mode_ok_uncached): ...this.
7238 (rs6000_init_hard_regno_mode_ok): Update accordingly.
7239 (rs6000_hard_regno_mode_ok): New function.
7240 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
7241 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
7242 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7243 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
7244 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
7245 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
7246 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7247 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
7248 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
7249 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7250 (sh_hard_regno_mode_ok): Make static.
7251 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
7252 instead of HARD_REGNO_MODE_OK.
7253 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
7254 (sparc_mode_class): Delete.
7255 (HARD_REGNO_MODE_OK): Delete.
7256 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7257 (hard_regno_mode_classes): Make static.
7258 (sparc_mode_class): Likewise.
7259 (sparc_hard_regno_mode_ok): New function.
7260 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
7261 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
7262 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
7264 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7265 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
7266 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
7267 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
7268 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
7269 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7270 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
7271 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
7272 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
7273 (visium_hard_regno_mode_ok): New function.
7274 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
7275 instead of HARD_REGNO_MODE_OK.
7276 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
7277 (HARD_REGNO_MODE_OK): Delete.
7278 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
7279 (xtensa_hard_regno_mode_ok_p): ...this and make static.
7280 (xtensa_option_override): Update accordingly.
7281 (TARGET_HARD_REGNO_MODE_OK): Redefine.
7282 (xtensa_hard_regno_mode_ok): New function.
7283 * system.h (HARD_REGNO_MODE_OK): Poison.
7285 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7286 Alan Hayward <alan.hayward@arm.com>
7287 David Sherwood <david.sherwood@arm.com>
7289 * target.def (hard_regno_call_part_clobbered): New hook.
7290 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
7291 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
7292 * doc/tm.texi: Regenerate.
7293 * hooks.h (hook_bool_uint_mode_false): Declare.
7294 * hooks.c (hook_bool_uint_mode_false): New function.
7295 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7296 * cselib.c (cselib_process_insn): Use
7297 targetm.hard_regno_call_part_clobbered instead of
7298 HARD_REGNO_CALL_PART_CLOBBERED.
7299 * ira-conflicts.c (ira_build_conflicts): Likewise.
7300 * ira-costs.c (ira_tune_allocno_costs): Likewise.
7301 * lra-constraints.c (need_for_call_save_p): Likewise.
7302 * lra-lives.c: Include target.h.
7303 (check_pseudos_live_through_calls): Use
7304 targetm.hard_regno_call_part_clobbered instead of
7305 HARD_REGNO_CALL_PART_CLOBBERED.
7306 * regcprop.c: Include target.h.
7307 (copyprop_hardreg_forward_1): Use
7308 targetm.hard_regno_call_part_clobbered instead of
7309 HARD_REGNO_CALL_PART_CLOBBERED.
7310 * reginfo.c (choose_hard_reg_mode): Likewise.
7311 * regrename.c (check_new_reg_p): Likewise.
7312 * reload.c (find_equiv_reg): Likewise.
7313 * reload1.c (emit_reload_insns): Likewise.
7314 * sched-deps.c (deps_analyze_insn): Likewise.
7315 * sel-sched.c (init_regs_for_mode): Likewise.
7316 (mark_unavailable_hard_regs): Likewise.
7317 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
7318 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7319 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
7321 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7322 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7323 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
7325 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
7327 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7328 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7329 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
7331 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7332 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7333 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
7335 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7336 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
7338 * config/powerpcspe/powerpcspe.c
7339 (rs6000_hard_regno_call_part_clobbered): New function.
7340 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7341 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7342 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
7344 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7345 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7346 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
7348 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
7349 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
7350 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
7352 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7353 Alan Hayward <alan.hayward@arm.com>
7354 David Sherwood <david.sherwood@arm.com>
7356 * rtl.h (subreg_memory_offset): Declare.
7357 * emit-rtl.c (subreg_memory_offset): New function.
7358 * expmed.c (store_bit_field_1): Use it.
7359 * expr.c (undefined_operand_subword_p): Likewise.
7360 * simplify-rtx.c (simplify_subreg): Likewise.
7362 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
7364 PR rtl-optimization/57448
7367 * optabs.c (expand_atomic_load): Place compiler memory barriers if
7368 using atomic_load pattern.
7369 (expand_atomic_store): Likewise.
7371 2017-09-04 Jakub Jelinek <jakub@redhat.com>
7374 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
7375 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
7376 handling. Use replace_call_with_value with NULL instead of
7377 gsi_replace, unlink_stmt_vdef and release_defs.
7379 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
7382 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
7384 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7387 * rtl.h (emit_library_call_value_1): Declare.
7388 (emit_library_call): Replace declaration with a series of overloads.
7389 Remove the parameter count argument.
7390 (emit_library_call_value): Likewise.
7391 * calls.c (emit_library_call_value_1): Make global. Replace varargs
7392 with an "rtx_mode_t *".
7393 (emit_library_call_value): Delete.
7394 (emit_library_call): Likewise.
7395 * asan.c (asan_emit_stack_protection): Update calls accordingly.
7396 (asan_emit_allocas_unpoison): Likewise.
7397 * builtins.c (expand_builtin_powi): Likewise.
7398 (expand_asan_emit_allocas_unpoison): Likewise.
7399 * cfgexpand.c (expand_main_function): Likewise.
7400 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
7401 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
7402 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
7403 * config/arm/arm.c (arm_trampoline_init): Likewise.
7404 (arm_call_tls_get_addr): Likewise.
7405 (arm_expand_divmod_libfunc): Likewise.
7406 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
7407 (smulsi3_highpart): Likewise.
7408 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
7409 (c6x_expand_compare): Likewise.
7410 (c6x_expand_movmem): Likewise.
7411 * config/frv/frv.c (frv_trampoline_init): Likewise.
7412 * config/i386/i386.c (ix86_trampoline_init): Likewise.
7413 (ix86_expand_divmod_libfunc): Likewise.
7414 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
7415 (ia64_expand_compare): Likewise.
7416 (ia64_profile_hook): Likewise.
7417 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
7418 (nonlocal_goto): Likewise.
7419 (restore_stack_nonlocal): Likewise.
7420 * config/m32r/m32r.c (block_move_call): Likewise.
7421 (m32r_trampoline_init): Likewise.
7422 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
7423 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
7424 (m68k_call_m68k_read_tp): Likewise.
7425 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
7426 (microblaze_expand_divide): Likewise.
7427 * config/mips/mips.h (mips_args): Likewise.
7428 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
7429 (MIPS_ICACHE_SYNC): Likewise.
7430 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
7431 (nios2_trampoline_init): Likewise.
7432 * config/pa/pa.c (hppa_tls_call): Likewise.
7433 (pa_trampoline_init): Likewise.
7434 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
7435 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
7436 (expand_strn_compare): Likewise.
7437 (rs6000_generate_compare): Likewise.
7438 (rs6000_expand_float128_convert): Likewise.
7439 (output_profile_hook): Likewise.
7440 (rs6000_trampoline_init): Likewise.
7441 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
7442 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
7443 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
7444 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
7445 (rs6000_generate_compare): Likewise.
7446 (rs6000_expand_float128_convert): Likewise.
7447 (output_profile_hook): Likewise.
7448 (rs6000_trampoline_init): Likewise.
7449 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
7450 * config/sh/sh.c (sh_trampoline_init): Likewise.
7451 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
7452 (sparc_emit_float_lib_cmp): Likewise.
7453 (sparc32_initialize_trampoline): Likewise.
7454 (sparc64_initialize_trampoline): Likewise.
7455 (sparc_profile_hook): Likewise.
7456 * config/spu/spu.c (ea_load_store): Likewise.
7457 * config/spu/spu.md (floatunssidf2): Likewise.
7458 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
7459 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
7460 * config/visium/visium.c (expand_block_move_4): Likewise.
7461 (expand_block_move_2): Likewise.
7462 (expand_block_move_1): Likewise.
7463 (expand_block_set_4): Likewise.
7464 (expand_block_set_2): Likewise.
7465 (expand_block_set_1): Likewise.
7466 (visium_trampoline_init): Likewise.
7467 (visium_profile_hook): Likewise.
7468 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
7469 (xtensa_setup_frame_addresses): Likewise.
7470 (xtensa_trampoline_init): Likewise.
7471 * except.c (sjlj_emit_function_enter): Likewise.
7472 (sjlj_emit_function_exit): Likewise.
7473 * explow.c (allocate_dynamic_stack_space): Likewise.
7474 (probe_stack_range): Likewise.
7475 * expr.c (convert_mode_scalar): Likewise.
7476 * optabs.c (expand_binop): Likewise.
7477 (expand_twoval_binop_libfunc): Likewise.
7478 (expand_unop): Likewise.
7479 (prepare_cmp_insn): Likewise.
7480 (prepare_float_lib_cmp): Likewise.
7481 (expand_float): Likewise.
7482 (expand_fix): Likewise.
7483 (expand_fixed_convert): Likewise.
7484 (maybe_emit_sync_lock_test_and_set): Likewise.
7485 (expand_atomic_compare_and_swap): Likewise.
7486 (expand_mem_thread_fence): Likewise.
7487 (expand_atomic_fetch_op): Likewise.
7489 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
7491 * doc/generic.texi (OpenACC): Adjust URL.
7492 * doc/invoke.texi (C Dialect Options): Ditto.
7494 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
7496 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
7497 predicate for operand 1. Add (m,<S>) constraint.
7498 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
7499 Prevent memory operand 1 with register operand 2.
7501 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
7503 PR rtl-optimization/82024
7504 * combine.c (try_combine): If the combination result is a PARALLEL,
7505 and we only need to retain the SET in there that would be placed
7506 at I2, check that we can place that at I3 instead, before doing so.
7508 2017-09-01 Jakub Jelinek <jakub@redhat.com>
7511 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
7513 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
7514 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
7517 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
7518 Jeff Law <law@redhat.com>
7520 * varasm.c (bss_initializer_p): Do not put constants into .bss
7521 (categorize_decl_for_section): Handle bss_initializer_p returning
7522 false when DECL_INITIAL is NULL.
7524 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7527 * config/s390/s390.c (s390_can_inline_p): New function.
7529 2017-09-01 Jeff Law <law@redhat.com>
7531 PR tree-optimization/82052
7532 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
7533 Always initialize the returned slot after a hash table miss
7534 when INSERT is true.
7536 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
7538 * config/s390/s390.md (mem_signal_fence): Remove.
7539 * doc/md.texi (mem_signal_fence): Remove.
7540 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
7542 * target-insns.def (mem_signal_fence): Remove.
7544 2017-09-01 Jakub Jelinek <jakub@redhat.com>
7547 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
7550 * asan.c (create_odr_indicator): Strip name encoding from assembler
7551 name before appending it after __odr_asan_.
7553 2017-09-01 Martin Liska <mliska@suse.cz>
7555 PR tree-optimization/82059
7556 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
7557 frequency only when an edge is redirected.
7559 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
7561 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
7562 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
7563 (arc_conditional_register_usage): Remove ARC600 lp_count
7565 (arc_file_start): Emit Tag_ARC_CPU_variation.
7566 (arc_can_use_doloop_p): New conditions to use ZOLs.
7567 (hwloop_fail): New function.
7568 (hwloop_optimize): Likewise.
7569 (hwloop_pattern_reg): Likewise.
7570 (arc_doloop_hooks): New struct, to be used with reorg_loops.
7571 (arc_reorg_loops): New function, calls reorg_loops.
7572 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
7573 (arc600_corereg_hazard): Remove ZOL checking, case handled by
7575 (arc_loop_hazard): Remove function, functionality moved into
7577 (arc_hazard): Remove arc_loop_hazard call.
7578 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
7579 into hwloop_optimize.
7580 (arc_label_align): Remove ZOL handling.
7581 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
7582 * config/arc/arc.md (doloop_begin): Remove pattern.
7583 (doloop_begin_i): Likewise.
7584 (doloop_end_i): Likewise.
7585 (doloop_fallback): Likewise.
7586 (doloop_fallback_m): Likewise.
7587 (doloop_end): Reimplement expand.
7588 (arc_lp): New pattern for LP instruction.
7589 (loop_end): New pattern.
7590 (loop_fail): Likewise.
7591 (decrement_and_branch_until_zero): Likewise.
7592 * config/arc/arc.opt (mlpc-width): New option.
7593 * doc/invoke.texi (mlpc-width): Document option.
7595 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
7597 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
7598 (arc_ccfsm_advance): Fix checking for delay slots.
7599 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
7601 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
7603 * config/arc/arc.md (movqi_insn): Add stores to save constant long
7605 (movhi_insn): Update store instruction constraint which are saving
7606 6-bit short immediates.
7607 (movsi_insn): Consider also short scaled load operations.
7608 (zero_extendhisi2_i): Use Usd constraint instead of T.
7609 (extendhisi2_i): Add q constraint.
7610 (arc_clzsi2): Add type and length attributes.
7611 (arc_ctzsi2): Likewise.
7612 * config/arc/constraints.md (Usc): Update constraint, the
7613 assembler can parse two relocations for a single instruction.
7615 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
7617 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
7618 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
7620 2017-08-31 Olivier Hainque <hainque@adacore.com>
7622 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
7623 match as powerpc-wrs-vxworks*.
7625 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
7627 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
7628 register constraint for by-element operand.
7629 (aarch64_mls_elt_merge<mode>): Likewise.
7631 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
7633 * config/arc/arc.c (arc_can_follow_jump): Check for short
7636 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
7638 * config.gcc: Use g.opt for arc.
7639 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
7640 functionality moved to ...
7641 (legitimate_scaled_address_p): New function, ...here.
7642 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
7643 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
7644 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
7646 (arc_override_options): Handle G option.
7647 (arc_output_pic_addr_const): Correct function definition.
7648 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
7649 (arc_decl_anon_ns_mem_p): Delete.
7650 (arc_in_small_data_p): Overhaul this function to take into
7651 consideration the value given via G option.
7652 (arc_rewrite_small_data_1): Renamed and corrected old
7653 arc_rewrite_small_data function.
7654 (arc_rewrite_small_data): New function.
7655 (small_data_pattern): Don't use pic_offset_table_rtx.
7656 * config/arc/arc.h (CC1_SPEC): Recognize G option.
7657 * config/arc/simdext.md (movmisalignv2hi): Use
7658 prepare_move_operands function.
7660 (movmisalign*): Likewise.
7661 * doc/invoke.texi (ARC options): Document -G option.
7663 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
7665 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
7667 * config/arc/arc.c (arc_print_operand): Output scalled address for
7668 sdata whenever is possible.
7669 (arc_in_small_data_p): Allow sdata for 64bit datum when double
7670 load/stores are available.
7671 (compact_sda_memory_operand): Check for the alignment required by
7672 code density instructions.
7673 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
7675 * config/arc/constraints.md (Usd): Update constraint.
7676 (Us0): New constraint.
7677 (Usc): Update constraint.
7679 2017-08-31 Richard Biener <rguenther@suse.de>
7682 * dwarf2out.c (dwarf2out_early_global_decl): Process each
7685 2017-08-31 Tamar Christina <tamar.christina@arm.com>
7687 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
7688 Resize type_signature.
7690 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
7691 Alan Hayward <alan.hayward@arm.com>
7692 David Sherwood <david.sherwood@arm.com>
7694 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
7695 subregs whose inner modes can be stored in GPRs.
7696 (aarch64_classify_index): Likewise.
7698 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
7699 Alan Hayward <alan.hayward@arm.com>
7700 David Sherwood <david.sherwood@arm.com>
7702 * config/aarch64/iterators.md (V_cmp_result): Rename to...
7703 (V_INT_EQUIV): ...this.
7704 (v_cmp_result): Rename to...
7705 (v_int_equiv): ...this.
7706 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
7707 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
7708 (copysign<mode>3): Likewise.
7709 (aarch64_simd_bsl<mode>_internal): Likewise.
7710 (aarch64_simd_bsl<mode>): Likewise.
7711 (vec_cmp<mode><mode>): Likewise.
7712 (vcond<mode><mode>): Likewise.
7713 (vcond<v_cmp_mixed><mode>): Likewise.
7714 (vcondu<mode><v_cmp_mixed>): Likewise.
7715 (aarch64_cm<optab><mode>): Likewise.
7716 (aarch64_cmtst<mode>): Likewise.
7717 (aarch64_fac<optab><mode>): Likewise.
7718 (vec_perm_const<mode>): Likewise.
7719 (vcond_mask_<mode><v_cmp_result>): Rename to...
7720 (vcond_mask_<mode><v_int_equiv>): ...this.
7721 (vec_cmp<mode><v_cmp_result>): Rename to...
7722 (vec_cmp<mode><v_int_equiv>): ...this.
7724 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
7725 Alan Hayward <alan.hayward@arm.com>
7726 David Sherwood <david.sherwood@arm.com>
7728 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
7730 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
7731 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
7732 (UNSPEC_LD4_DREG): New unspecs.
7733 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
7734 (aarch64_ld2<mode>_dreg_be): Replace with...
7735 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
7737 (aarch64_ld3<mode>_dreg_le)
7738 (aarch64_ld3<mode>_dreg_be): Replace with...
7739 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
7741 (aarch64_ld4<mode>_dreg_le)
7742 (aarch64_ld4<mode>_dreg_be): Replace with...
7743 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
7746 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7748 PR tree-optimization/81987
7749 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
7750 insert an initializer in a location not dominated by the stride
7753 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
7755 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
7756 on the entire header of the finally block in the fallthru case.
7758 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
7760 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
7762 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
7764 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
7765 rs6000_emit_move_from_cr and call renamed function.
7766 (rs6000_emit_prologue): Call renamed functions.
7767 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
7768 movesi_from_cr, remove volatile CRs.
7770 2017-08-30 Jon Beniston <jon@beniston.com>
7771 Richard Biener <rguenther@suse.de>
7773 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
7774 of VECTOR_MODE_P check.
7775 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
7776 element vector types.
7778 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7780 * df.h (df_read_modify_subreg_p): Remove in favor of...
7781 * rtl.h (read_modify_subreg_p): ...this new function. Take a
7782 const_rtx instead of an rtx.
7783 * cprop.c (local_cprop_find_used_regs): Update accordingly.
7784 * df-problems.c (df_word_lr_mark_ref): Likewise.
7785 * ira-lives.c (mark_pseudo_reg_live): Likewise.
7786 (mark_pseudo_reg_dead): Likewise.
7787 (mark_ref_dead): Likewise.
7788 * reginfo.c (init_subregs_of_mode): Likewise.
7789 * sched-deps.c (sched_analyze_1): Likewise.
7790 * df-scan.c (df_def_record_1): Likewise.
7791 (df_uses_record): Likewise.
7792 (df_read_modify_subreg_p): Remove in favor of...
7793 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
7794 const_rtx instead of an rtx.
7796 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7797 Alan Hayward <alan.hayward@arm.com>
7798 David Sherwood <david.sherwood@arm.com>
7800 * rtl.h (partial_subreg_p): New function.
7801 * caller-save.c (save_call_clobbered_regs): Use it.
7802 * calls.c (expand_call): Likewise.
7803 * combine.c (combinable_i3pat): Likewise.
7804 (simplify_set): Likewise.
7805 (make_extraction): Likewise.
7806 (make_compound_operation_int): Likewise.
7807 (gen_lowpart_or_truncate): Likewise.
7808 (force_to_mode): Likewise.
7809 (make_field_assignment): Likewise.
7810 (reg_truncated_to_mode): Likewise.
7811 (record_truncated_value): Likewise.
7812 (move_deaths): Likewise.
7813 * cse.c (record_jump_cond): Likewise.
7814 (cse_insn): Likewise.
7815 * cselib.c (cselib_lookup_1): Likewise.
7816 * expmed.c (extract_bit_field_using_extv): Likewise.
7817 * function.c (assign_parm_setup_reg): Likewise.
7818 * ifcvt.c (noce_convert_multiple_sets): Likewise.
7819 * ira-build.c (create_insn_allocnos): Likewise.
7820 * lra-coalesce.c (merge_pseudos): Likewise.
7821 * lra-constraints.c (match_reload): Likewise.
7822 (simplify_operand_subreg): Likewise.
7823 (curr_insn_transform): Likewise.
7824 * lra-lives.c (process_bb_lives): Likewise.
7825 * lra.c (new_insn_reg): Likewise.
7826 (lra_substitute_pseudo): Likewise.
7827 * regcprop.c (mode_change_ok): Likewise.
7828 (maybe_mode_change): Likewise.
7829 (copyprop_hardreg_forward_1): Likewise.
7830 * reload.c (push_reload): Likewise.
7831 (find_reloads): Likewise.
7832 (find_reloads_subreg_address): Likewise.
7833 * reload1.c (alter_reg): Likewise.
7834 (eliminate_regs_1): Likewise.
7835 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7837 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
7839 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
7840 back to if statements, including unpack.
7842 2017-08-30 Martin Liska <mliska@suse.cz>
7845 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
7847 (func_checker::compare_asm_inputs_outputs): ... this function.
7848 (func_checker::compare_gimple_asm): Use the function to compare
7849 also ASM constrains.
7850 * ipa-icf-gimple.h: Rename the function.
7852 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7853 Alan Hayward <alan.hayward@arm.com>
7854 David Sherwood <david.sherwood@arm.com>
7856 * coretypes.h (complex_mode): New type.
7857 * gdbhooks.py (build_pretty_printer): Handle it.
7858 * machmode.h (complex_mode): New class.
7859 (complex_mode::includes_p): New function.
7860 (is_complex_int_mode): Likewise.
7861 (is_complex_float_mode): Likewise.
7862 * genmodes.c (get_mode_class): Handle complex mode classes.
7863 * function.c (expand_function_end): Use is_complex_int_mode.
7865 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7866 Alan Hayward <alan.hayward@arm.com>
7867 David Sherwood <david.sherwood@arm.com>
7869 * coretypes.h (scalar_mode_pod): New typedef.
7870 * gdbhooks.py (build_pretty_printer): Handle it.
7871 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
7872 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
7873 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
7874 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
7877 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7878 Alan Hayward <alan.hayward@arm.com>
7879 David Sherwood <david.sherwood@arm.com>
7881 * machmode.h (mode_for_vector): Take a scalar_mode instead
7883 * stor-layout.c (mode_for_vector): Likewise.
7884 * explow.c (promote_mode): Use as_a <scalar_mode>.
7885 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
7887 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7888 Alan Hayward <alan.hayward@arm.com>
7889 David Sherwood <david.sherwood@arm.com>
7891 * target.def (preferred_simd_mode): Take a scalar_mode
7892 instead of a machine_mode.
7893 * targhooks.h (default_preferred_simd_mode): Likewise.
7894 * targhooks.c (default_preferred_simd_mode): Likewise.
7895 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
7896 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
7897 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
7898 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
7899 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
7900 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
7901 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
7902 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
7904 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
7905 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
7906 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
7907 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
7908 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
7909 * doc/tm.texi: Regenerate.
7910 * optabs-query.c (can_vec_mask_load_store_p): Return false for
7913 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7914 Alan Hayward <alan.hayward@arm.com>
7915 David Sherwood <david.sherwood@arm.com>
7917 * target.def (scalar_mode_supported_p): Take a scalar_mode
7918 instead of a machine_mode.
7919 * targhooks.h (default_scalar_mode_supported_p): Likewise.
7920 * targhooks.c (default_scalar_mode_supported_p): Likewise.
7921 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
7922 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
7923 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
7924 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
7925 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
7926 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
7927 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
7928 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
7929 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
7930 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
7931 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
7932 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
7934 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
7935 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
7936 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
7937 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
7938 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
7940 * doc/tm.texi: Regenerate.
7942 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7943 Alan Hayward <alan.hayward@arm.com>
7944 David Sherwood <david.sherwood@arm.com>
7946 * coretypes.h (opt_scalar_mode): New typedef.
7947 * gdbhooks.py (build_pretty_printers): Handle it.
7948 * machmode.h (mode_iterator::get_2xwider): Add overload for
7950 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
7952 * expr.c (convert_mode_scalar): Likewise.
7953 * omp-low.c (omp_clause_aligned_alignment): Likewise.
7954 * optabs.c (expand_float): Likewise.
7955 (expand_fix): Likewise.
7956 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
7958 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7959 Alan Hayward <alan.hayward@arm.com>
7960 David Sherwood <david.sherwood@arm.com>
7962 * optabs.c (expand_float): Explicitly check for scalars before
7963 using a branching expansion.
7964 (expand_fix): Likewise.
7966 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7967 Alan Hayward <alan.hayward@arm.com>
7968 David Sherwood <david.sherwood@arm.com>
7970 * expr.c (convert_mode): Split scalar handling out into...
7971 (convert_mode_scalar): ...this new function. Treat the modes
7974 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7975 Alan Hayward <alan.hayward@arm.com>
7976 David Sherwood <david.sherwood@arm.com>
7978 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
7980 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
7982 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7983 Alan Hayward <alan.hayward@arm.com>
7984 David Sherwood <david.sherwood@arm.com>
7986 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
7987 rather than a machine_mode.
7988 (fixed_from_string): Likewise.
7989 (fixed_convert): Likewise.
7990 (fixed_convert_from_int): Likewise.
7991 (fixed_convert_from_real): Likewise.
7992 (real_convert_from_fixed): Likewise.
7993 * fixed-value.c (fixed_from_double_int): Likewise.
7994 (fixed_from_string): Likewise.
7995 (fixed_convert): Likewise.
7996 (fixed_convert_from_int): Likewise.
7997 (fixed_convert_from_real): Likewise.
7998 (real_convert_from_fixed): Likewise.
7999 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
8001 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8002 Alan Hayward <alan.hayward@arm.com>
8003 David Sherwood <david.sherwood@arm.com>
8005 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
8006 of separate mode class checks. Do not allow vector modes here.
8007 (immed_wide_int_const): Use as_a <scalar_mode>.
8008 * explow.c (trunc_int_for_mode): Likewise.
8009 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
8010 (wi::shwi): Likewise.
8011 (wi::min_value): Likewise.
8012 (wi::max_value): Likewise.
8013 * dwarf2out.c (loc_descriptor): Likewise.
8014 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
8017 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8018 Alan Hayward <alan.hayward@arm.com>
8019 David Sherwood <david.sherwood@arm.com>
8021 * tree.h (SCALAR_TYPE_MODE): New macro.
8022 * expr.c (expand_expr_addr_expr_1): Use it.
8023 (expand_expr_real_2): Likewise.
8024 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
8025 (fold_convert_const_fixed_from_int): Likewise.
8026 (fold_convert_const_fixed_from_real): Likewise.
8027 (native_encode_fixed): Likewise
8028 (native_encode_complex): Likewise
8029 (native_encode_vector): Likewise.
8030 (native_interpret_fixed): Likewise.
8031 (native_interpret_real): Likewise.
8032 (native_interpret_complex): Likewise.
8033 (native_interpret_vector): Likewise.
8034 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
8035 (simd_clone_adjust_argument_types): Likewise.
8036 (simd_clone_init_simd_arrays): Likewise.
8037 (simd_clone_adjust): Likewise.
8038 * stor-layout.c (layout_type): Likewise.
8039 * tree.c (build_minus_one_cst): Likewise.
8040 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
8041 * tree-inline.c (estimate_move_cost): Likewise.
8042 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
8043 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
8044 (vectorizable_reduction): Likewise.
8045 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
8046 (vect_recog_mixed_size_cond_pattern): Likewise.
8047 (check_bool_pattern): Likewise.
8048 (adjust_bool_pattern): Likewise.
8049 (search_type_for_mask_1): Likewise.
8050 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
8051 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
8052 (vectorizable_load): Likewise.
8053 (vectorizable_store): Likewise.
8054 * ubsan.c (ubsan_encode_value): Likewise.
8055 * varasm.c (output_constant): Likewise.
8057 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8058 Alan Hayward <alan.hayward@arm.com>
8059 David Sherwood <david.sherwood@arm.com>
8061 * coretypes.h (scalar_mode): New class.
8062 * machmode.h (scalar_mode): Likewise.
8063 (scalar_mode::includes_p): New function.
8064 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
8065 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
8066 * genmodes.c (get_mode_class): Handle remaining scalar modes.
8067 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
8068 * expmed.c (store_bit_field_1): Likewise.
8069 (extract_bit_field_1): Likewise.
8070 * expr.c (write_complex_part): Likewise.
8071 (read_complex_part): Likewise.
8072 (emit_move_complex_push): Likewise.
8073 (expand_expr_real_2): Likewise.
8074 * function.c (assign_parm_setup_reg): Likewise.
8075 (assign_parms_unsplit_complex): Likewise.
8076 * optabs.c (expand_binop): Likewise.
8077 * rtlanal.c (subreg_get_info): Likewise.
8078 * simplify-rtx.c (simplify_immed_subreg): Likewise.
8079 * varasm.c (output_constant_pool_2): Likewise.
8081 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8082 Alan Hayward <alan.hayward@arm.com>
8083 David Sherwood <david.sherwood@arm.com>
8085 * expmed.c (extract_high_half): Use scalar_int_mode and remove
8087 (expmed_mult_highpart_optab): Likewise.
8088 (expmed_mult_highpart): Likewise.
8090 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8091 Alan Hayward <alan.hayward@arm.com>
8092 David Sherwood <david.sherwood@arm.com>
8094 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
8095 instead of a machine_mode.
8096 (builtin_memset_read_str): Likewise.
8097 * builtins.c (c_readstr): Likewise.
8098 (builtin_memcpy_read_str): Likewise.
8099 (builtin_strncpy_read_str): Likewise.
8100 (builtin_memset_read_str): Likewise.
8101 (builtin_memset_gen_str): Likewise.
8102 (expand_builtin_signbit): Use scalar_int_mode for local variables.
8103 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
8104 instead of a machine_mode.
8105 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
8107 (make_extraction): Likewise.
8108 (try_widen_shift_mode): Take and return scalar_int_modes instead
8110 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
8111 a scalar_int_mode instead of a machine_mode.
8112 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
8113 (avr_addr_space_pointer_mode): Likewise.
8114 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
8115 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
8116 (msp430_unwind_word_mode): Likewise.
8117 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
8118 (spu_addr_space_pointer_mode): Likewise.
8119 (spu_addr_space_address_mode): Likewise.
8120 (spu_libgcc_cmp_return_mode): Likewise.
8121 (spu_libgcc_shift_count_mode): Likewise.
8122 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
8123 (rl78_addr_space_pointer_mode): Likewise.
8124 (fl78_unwind_word_mode): Likewise.
8125 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
8127 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
8128 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
8129 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
8130 (mips_valid_pointer_mode): Likewise.
8131 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
8132 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
8133 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
8135 (ft32_addr_space_address_mode): Likewise.
8136 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
8137 scalar_int_mode instead of a machine_mode.
8138 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
8140 (m32c_addr_space_address_mode): Likewise.
8141 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
8142 (rs6000_eh_return_filter_mode): Likewise.
8143 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
8144 (rs6000_eh_return_filter_mode): Likewise.
8145 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
8146 (s390_libgcc_shift_count_mode): Likewise.
8147 (s390_unwind_word_mode): Likewise.
8148 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
8150 * target.def (mode_rep_extended): Likewise.
8151 (valid_pointer_mode): Likewise.
8152 (addr_space.valid_pointer_mode): Likewise.
8153 (eh_return_filter_mode): Return a scalar_int_mode rather than
8155 (libgcc_cmp_return_mode): Likewise.
8156 (libgcc_shift_count_mode): Likewise.
8157 (unwind_word_mode): Likewise.
8158 (addr_space.pointer_mode): Likewise.
8159 (addr_space.address_mode): Likewise.
8160 * doc/tm.texi: Regenerate.
8161 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
8163 (do_jump): Use scalar_int_mode for local variables.
8164 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
8165 rather than a machine_mode.
8166 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
8167 (scompare_loc_descriptor_wide): Likewise.
8168 (scompare_loc_descriptor_narrow): Likewise.
8169 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
8171 * except.c (sjlj_emit_dispatch_table): Likewise.
8172 (expand_builtin_eh_copy_values): Likewise.
8173 * explow.c (convert_memory_address_addr_space_1): Likewise.
8174 Take a scalar_int_mode rather than a machine_mode.
8175 (convert_memory_address_addr_space): Take a scalar_int_mode rather
8176 than a machine_mode.
8177 (memory_address_addr_space): Use scalar_int_mode for local variables.
8178 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
8179 rather than a machine_mode.
8180 * expmed.c (mask_rtx): Likewise.
8181 (init_expmed_one_conv): Likewise.
8182 (expand_mult_highpart_adjust): Likewise.
8183 (extract_high_half): Likewise.
8184 (expmed_mult_highpart_optab): Likewise.
8185 (expmed_mult_highpart): Likewise.
8186 (expand_smod_pow2): Likewise.
8187 (expand_sdiv_pow2): Likewise.
8188 (emit_store_flag_int): Likewise.
8189 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
8191 (extract_low_bits): Likewise.
8192 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
8194 * expr.c (pieces_addr::adjust): Likewise.
8195 (can_store_by_pieces): Likewise.
8196 (store_by_pieces): Likewise.
8197 (clear_by_pieces_1): Likewise.
8198 (expand_expr_addr_expr_1): Likewise.
8199 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
8200 (expand_expr_real_1): Likewise.
8201 (try_casesi): Likewise.
8202 * final.c (shorten_branches): Likewise.
8203 * fold-const.c (fold_convert_const_int_from_fixed): Change the
8204 type of "mode" to machine_mode.
8205 * internal-fn.c (expand_arith_overflow_result_store): Take a
8206 scalar_int_mode rather than a machine_mode.
8207 (expand_mul_overflow): Use scalar_int_mode for local variables.
8208 * loop-doloop.c (doloop_modify): Likewise.
8209 (doloop_optimize): Likewise.
8210 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
8211 than a machine_mode.
8212 (expand_doubleword_shift_condmove): Likewise.
8213 (expand_doubleword_shift): Likewise.
8214 (expand_doubleword_clz): Likewise.
8215 (expand_doubleword_popcount): Likewise.
8216 (expand_doubleword_parity): Likewise.
8217 (expand_absneg_bit): Use scalar_int_mode for local variables.
8218 (prepare_float_lib_cmp): Likewise.
8219 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
8220 rather than a machine_mode.
8221 (convert_memory_address_addr_space): Likewise.
8222 (get_mode_bounds): Likewise.
8223 (get_address_mode): Return a scalar_int_mode rather than a
8225 * rtlanal.c (get_address_mode): Likewise.
8226 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
8227 than a machine_mode.
8228 * targhooks.c (default_mode_rep_extended): Likewise.
8229 (default_valid_pointer_mode): Likewise.
8230 (default_addr_space_valid_pointer_mode): Likewise.
8231 (default_eh_return_filter_mode): Return a scalar_int_mode rather
8232 than a machine_mode.
8233 (default_libgcc_cmp_return_mode): Likewise.
8234 (default_libgcc_shift_count_mode): Likewise.
8235 (default_unwind_word_mode): Likewise.
8236 (default_addr_space_pointer_mode): Likewise.
8237 (default_addr_space_address_mode): Likewise.
8238 * targhooks.h (default_eh_return_filter_mode): Likewise.
8239 (default_libgcc_cmp_return_mode): Likewise.
8240 (default_libgcc_shift_count_mode): Likewise.
8241 (default_unwind_word_mode): Likewise.
8242 (default_addr_space_pointer_mode): Likewise.
8243 (default_addr_space_address_mode): Likewise.
8244 (default_mode_rep_extended): Take a scalar_int_mode rather than
8246 (default_valid_pointer_mode): Likewise.
8247 (default_addr_space_valid_pointer_mode): Likewise.
8248 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
8250 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
8251 rather than a machine_mode.
8252 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
8253 for local variables.
8254 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
8255 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
8256 than a machine_mode.
8258 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8259 Alan Hayward <alan.hayward@arm.com>
8260 David Sherwood <david.sherwood@arm.com>
8262 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
8263 the mode argument to scalar_int_mode.
8264 (do_jump_by_parts_zero_rtx): Likewise.
8265 (do_jump_by_parts_equality_rtx): Likewise.
8266 (do_jump_by_parts_greater): Take a mode argument.
8267 (do_jump_by_parts_equality): Likewise.
8268 (do_jump_1): Update calls accordingly.
8270 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8271 Alan Hayward <alan.hayward@arm.com>
8272 David Sherwood <david.sherwood@arm.com>
8274 * is-a.h (safe_dyn_cast): New function.
8275 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
8276 (jump_table_for_label): Likewise.
8277 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
8278 instead of an rtx_insn *.
8279 (shorten_branches): Use dyn_cast instead of LABEL_P and
8280 JUMP_TABLE_DATA_P. Use jump_table_for_label and
8281 rtx_jump_table_data::get_data_mode.
8282 (final_scan_insn): Likewise.
8284 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8285 Alan Hayward <alan.hayward@arm.com>
8286 David Sherwood <david.sherwood@arm.com>
8288 * combine.c (try_combine): Use is_a <scalar_int_mode> when
8289 trying to combine a full-register integer set with a subreg
8292 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8293 Alan Hayward <alan.hayward@arm.com>
8294 David Sherwood <david.sherwood@arm.com>
8296 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
8297 that is always either address_mode or pointer_mode.
8299 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8300 Alan Hayward <alan.hayward@arm.com>
8301 David Sherwood <david.sherwood@arm.com>
8303 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
8304 when the two are known to be equal.
8306 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8307 Alan Hayward <alan.hayward@arm.com>
8308 David Sherwood <david.sherwood@arm.com>
8310 * simplify-rtx.c (simplify_const_unary_operation): Use
8311 is_a <scalar_int_mode> instead of checking for a nonzero
8312 precision. Forcibly convert op_mode to a scalar_int_mode
8313 in that case. More clearly differentiate the operand and
8314 result modes and use the former when deciding what the value
8315 of a count-bits operation should be. Use is_int_mode instead
8316 of checking for a MODE_INT. Remove redundant check for whether
8317 this mode has a zero precision.
8319 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8320 Alan Hayward <alan.hayward@arm.com>
8321 David Sherwood <david.sherwood@arm.com>
8323 * optabs.c (widen_leading): Change the type of the mode argument
8324 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
8325 (widen_bswap): Likewise.
8326 (expand_parity): Likewise.
8327 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
8328 (expand_ffs): Likewise.
8329 (epand_unop): Check for scalar integer modes before calling the
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 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
8338 (expand_expr_real_1): Update call accordingly.
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 * expmed.c (store_bit_field_using_insv): Add op0_mode and
8345 value_mode arguments. Use scalar_int_mode internally.
8346 (store_bit_field_1): Rename the new integer mode from imode
8347 to op0_mode and use it instead of GET_MODE (op0). Update calls
8348 to store_split_bit_field, store_bit_field_using_insv and
8349 store_fixed_bit_field.
8350 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
8351 Use scalar_int_mode internally. Use a bit count rather than a mode
8352 when calculating the largest bit size for get_best_mode.
8353 Update calls to store_split_bit_field and store_fixed_bit_field_1.
8354 (store_fixed_bit_field_1): Add mode and value_mode arguments.
8355 Remove assertion that OP0 has a scalar integer mode.
8356 (store_split_bit_field): Add op0_mode and value_mode arguments.
8357 Update calls to extract_fixed_bit_field.
8358 (extract_bit_field_using_extv): Add an op0_mode argument.
8359 Use scalar_int_mode internally.
8360 (extract_bit_field_1): Rename the new integer mode from imode to
8361 op0_mode and use it instead of GET_MODE (op0). Update calls to
8362 extract_split_bit_field, extract_bit_field_using_extv and
8363 extract_fixed_bit_field.
8364 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
8365 to extract_split_bit_field and extract_fixed_bit_field_1.
8366 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
8367 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
8369 (extract_split_bit_field): Add an op0_mode argument. Update call
8370 to extract_fixed_bit_field.
8372 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8373 Alan Hayward <alan.hayward@arm.com>
8374 David Sherwood <david.sherwood@arm.com>
8376 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
8377 * explow.c (hard_function_value): Likewise.
8378 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
8379 convert_to_mode call outside the loop.
8380 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
8381 for the mode iterator. Require the mode specified by max_pieces
8383 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
8385 (copy_blkmode_to_reg): Likewise.
8386 (set_storage_via_setmem): Likewise.
8387 * optabs.c (prepare_cmp_insn): Likewise.
8388 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
8389 * stor-layout.c (finish_bitfield_representative): Likewise.
8391 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8392 Alan Hayward <alan.hayward@arm.com>
8393 David Sherwood <david.sherwood@arm.com>
8395 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
8396 * expr.c (convert_move): Use them.
8397 (convert_modes): Likewise.
8398 (store_expr_with_bounds): Likewise.
8400 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8401 Alan Hayward <alan.hayward@arm.com>
8402 David Sherwood <david.sherwood@arm.com>
8404 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
8405 parameter for the mode of "x". Remove the "known_x", "known_mode"
8406 and "known_ret" arguments. Change the type of the mode argument
8408 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
8409 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
8410 (reg_num_sign_bit_copies_for_combine): Likewise.
8411 * rtlanal.c (nonzero_bits1): Likewise.
8412 (num_sign_bit_copies1): Likewise.
8413 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
8414 (reg_num_sign_bit_copies_general): Likewise.
8415 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
8416 (reg_nonzero_bits_general): Likewise.
8418 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8419 Alan Hayward <alan.hayward@arm.com>
8420 David Sherwood <david.sherwood@arm.com>
8422 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
8423 than in subroutines. Return 1 for non-integer modes.
8424 (cached_num_sign_bit_copies): Change the type of the mode parameter
8426 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
8427 classes. Handle CONST_INT_P first and then check whether X also
8428 has a scalar integer mode. Check the same thing for inner registers
8429 of a SUBREG and for values that are being extended or truncated.
8431 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8432 Alan Hayward <alan.hayward@arm.com>
8433 David Sherwood <david.sherwood@arm.com>
8435 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
8436 in subroutines. Return the mode mask for non-integer modes.
8437 (cached_nonzero_bits): Change the type of the mode parameter
8439 (nonzero_bits1): Likewise. Remove early exit for other mode
8440 classes. Handle CONST_INT_P first and then check whether X
8441 also has a scalar integer mode.
8443 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8444 Alan Hayward <alan.hayward@arm.com>
8445 David Sherwood <david.sherwood@arm.com>
8447 * expr.c (widest_int_mode_for_size): Make the comment match the code.
8448 Return a scalar_int_mode and assert that the size is greater than
8450 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
8451 (op_by_pieces_d::op_by_pieces_d): Likewise.
8452 (op_by_pieces_d::run): Likewise.
8453 (can_store_by_pieces): Likewise.
8455 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8456 Alan Hayward <alan.hayward@arm.com>
8457 David Sherwood <david.sherwood@arm.com>
8459 * combine.c (extract_left_shift): Add a mode argument and update
8461 (make_compound_operation_int): Change the type of the mode parameter
8462 to scalar_int_mode and update the call to extract_left_shift.
8464 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8465 Alan Hayward <alan.hayward@arm.com>
8466 David Sherwood <david.sherwood@arm.com>
8468 * combine.c (simplify_and_const_int): Change the type of the mode
8469 parameter to scalar_int_mode.
8470 (simplify_and_const_int_1): Likewise. Update recursive call.
8472 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8473 Alan Hayward <alan.hayward@arm.com>
8474 David Sherwood <david.sherwood@arm.com>
8476 * combine.c (simplify_compare_const): Check that the mode is a
8477 scalar_int_mode (rather than VOIDmode) before testing its
8479 (simplify_comparison): Move COMPARISON_P handling out of the
8480 loop and restrict the latter part of the loop to scalar_int_modes.
8481 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
8482 and when considering SUBREG_REGs. Use is_int_mode instead of
8483 checking GET_MODE_CLASS against MODE_INT.
8485 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8486 Alan Hayward <alan.hayward@arm.com>
8487 David Sherwood <david.sherwood@arm.com>
8489 * combine.c (try_widen_shift_mode): Move check for equal modes to...
8490 (simplify_shift_const_1): ...here. Use scalar_int_mode for
8491 shift_unit_mode and for modes involved in scalar shifts.
8493 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8494 Alan Hayward <alan.hayward@arm.com>
8495 David Sherwood <david.sherwood@arm.com>
8497 * combine.c (force_int_to_mode): New function, split out from...
8498 (force_to_mode): ...here. Keep xmode up-to-date and use it
8499 instead of GET_MODE (x).
8501 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8502 Alan Hayward <alan.hayward@arm.com>
8503 David Sherwood <david.sherwood@arm.com>
8505 * optabs-query.h (extraction_insn::struct_mode): Change type to
8506 opt_scalar_int_mode and update comment.
8507 (extraction_insn::field_mode): Change type to scalar_int_mode.
8508 (extraction_insn::pos_mode): Likewise.
8509 * combine.c (make_extraction): Update accordingly.
8510 * optabs-query.c (get_traditional_extraction_insn): Likewise.
8511 (get_optab_extraction_insn): Likewise.
8512 * recog.c (simplify_while_replacing): Likewise.
8513 * expmed.c (narrow_bit_field_mem): Change the type of the mode
8514 parameter to opt_scalar_int_mode.
8516 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8517 Alan Hayward <alan.hayward@arm.com>
8518 David Sherwood <david.sherwood@arm.com>
8520 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
8521 to a scalar_int_mode instead of a machine_mode.
8522 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
8523 (get_best_mode): Return a boolean and use a pointer argument to store
8524 the selected mode. Replace the limit mode parameter with a bit limit.
8525 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
8526 for the values returned by bit_field_mode_iterator::next_mode.
8527 (store_bit_field): Update call to get_best_mode.
8528 (store_fixed_bit_field): Likewise.
8529 (extract_fixed_bit_field): Likewise.
8530 * expr.c (optimize_bitfield_assignment_op): Likewise.
8531 * fold-const.c (optimize_bit_field_compare): Likewise.
8532 (fold_truth_andor_1): Likewise.
8533 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
8534 Update for new type of m_mode.
8535 (get_best_mode): As above.
8537 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8538 Alan Hayward <alan.hayward@arm.com>
8539 David Sherwood <david.sherwood@arm.com>
8541 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
8542 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
8543 (store_bit_field): Check is_a <scalar_int_mode> before calling
8544 strict_volatile_bitfield_p.
8545 (extract_bit_field): Likewise.
8547 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8548 Alan Hayward <alan.hayward@arm.com>
8549 David Sherwood <david.sherwood@arm.com>
8551 * target.def (cstore_mode): Return a scalar_int_mode.
8552 * doc/tm.texi: Regenerate.
8553 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
8554 * targhooks.h (default_cstore_mode): Likewise.
8555 * targhooks.c (default_cstore_mode): Likewise, using a forced
8557 * expmed.c (emit_cstore): Expect the target of the cstore to be
8560 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8561 Alan Hayward <alan.hayward@arm.com>
8562 David Sherwood <david.sherwood@arm.com>
8564 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
8566 (niter_desc): Likewise mode.
8567 (iv_analyze): Add a mode parameter.
8569 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
8570 and change its type to scalar_int_mode.
8571 * loop-iv.c: Update commentary at head of file.
8572 (iv_constant): Pass the mode paraeter before the rtx it describes
8573 and change its type to scalar_int_mode. Remove VOIDmode handling.
8574 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
8575 (iv_extend): Likewise.
8576 (shorten_into_mode): Likewise.
8577 (iv_add): Use scalar_int_mode.
8578 (iv_mult): Likewise.
8579 (iv_shift): Likewise.
8580 (canonicalize_iv_subregs): Likewise.
8581 (get_biv_step_1): Pass the outer_mode parameter before the rtx
8582 it describes and change its mode to scalar_int_mode. Also change
8583 the type of the returned inner_mode to scalar_int_mode.
8584 (get_biv_step): Likewise, turning outer_mode from a pointer
8585 into a direct parameter. Update call to get_biv_step_1.
8586 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
8587 iv_constant and get_biv_step.
8588 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
8589 and change its type to scalar_int_mode. Don't initialise iv->mode
8590 to VOIDmode and remove later checks for its still being VOIDmode.
8591 Update calls to iv_analyze_op and iv_analyze_expr. Check
8592 is_a <scalar_int_mode> when changing the mode under consideration.
8593 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
8594 Update call to iv_analyze_expr.
8595 (iv_analyze_op): Add a mode parameter. Reject subregs whose
8596 inner register is not also a scalar_int_mode. Update call to
8598 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
8599 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
8600 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
8601 separate mode class checks. Update calls to iv_analyze. Remove
8602 fix-up of VOIDmodes after iv_analyze_biv.
8603 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
8604 don't have a scalar_int_mode. Update call to biv_p.
8606 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8607 Alan Hayward <alan.hayward@arm.com>
8608 David Sherwood <david.sherwood@arm.com>
8610 * cfgexpand.c (convert_debug_memory_address): Use
8611 as_a <scalar_int_mode>.
8612 * combine.c (expand_compound_operation): Likewise.
8613 (make_extraction): Likewise.
8614 (change_zero_ext): Likewise.
8615 (simplify_comparison): Likewise.
8616 * cse.c (cse_insn): Likewise.
8617 * dwarf2out.c (minmax_loc_descriptor): Likewise.
8618 (mem_loc_descriptor): Likewise.
8619 (loc_descriptor): Likewise.
8620 * expmed.c (init_expmed_one_mode): Likewise.
8621 (synth_mult): Likewise.
8622 (emit_store_flag_1): Likewise.
8623 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
8624 of a comparison with size.
8625 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
8626 (reduce_to_bit_field_precision): Likewise.
8627 * function.c (expand_function_end): Likewise.
8628 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
8629 * loop-doloop.c (doloop_modify): Likewise.
8630 * optabs.c (expand_binop): Likewise.
8631 (expand_unop): Likewise.
8632 (expand_copysign_absneg): Likewise.
8633 (prepare_cmp_insn): Likewise.
8634 (maybe_legitimize_operand): Likewise.
8635 * recog.c (const_scalar_int_operand): Likewise.
8636 * rtlanal.c (get_address_mode): Likewise.
8637 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8638 (simplify_cond_clz_ctz): Likewise.
8639 * tree-nested.c (get_nl_goto_field): Likewise.
8640 * tree.c (build_vector_type_for_mode): Likewise.
8641 * var-tracking.c (use_narrower_mode): Likewise.
8643 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8644 Alan Hayward <alan.hayward@arm.com>
8645 David Sherwood <david.sherwood@arm.com>
8647 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
8648 * builtins.c (expand_builtin_signbit): Use it.
8649 * cfgexpand.c (expand_debug_expr): Likewise.
8650 * dojump.c (do_jump): Likewise.
8651 (do_compare_and_jump): Likewise.
8652 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
8653 * expmed.c (make_tree): Likewise.
8654 * expr.c (expand_expr_real_2): Likewise.
8655 (expand_expr_real_1): Likewise.
8656 (try_casesi): Likewise.
8657 * fold-const-call.c (fold_const_call_ss): Likewise.
8658 * fold-const.c (unextend): Likewise.
8659 (extract_muldiv_1): Likewise.
8660 (fold_single_bit_test): Likewise.
8661 (native_encode_int): Likewise.
8662 (native_encode_string): Likewise.
8663 (native_interpret_int): Likewise.
8664 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
8665 * internal-fn.c (expand_addsub_overflow): Likewise.
8666 (expand_neg_overflow): Likewise.
8667 (expand_mul_overflow): Likewise.
8668 (expand_arith_overflow): Likewise.
8669 * match.pd: Likewise.
8670 * stor-layout.c (layout_type): Likewise.
8671 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
8672 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
8673 * tree-ssanames.c (get_range_info): Likewise.
8674 * tree-switch-conversion.c (array_value_type) Likewise.
8675 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
8676 (vect_recog_divmod_pattern): Likewise.
8677 (vect_recog_mixed_size_cond_pattern): Likewise.
8678 * tree-vrp.c (extract_range_basic): Likewise.
8679 (simplify_float_conversion_using_ranges): Likewise.
8680 * tree.c (int_fits_type_p): Likewise.
8681 * ubsan.c (instrument_bool_enum_load): Likewise.
8682 * varasm.c (mergeable_string_section): Likewise.
8683 (narrowing_initializer_constant_valid_p): Likewise.
8684 (output_constant): Likewise.
8686 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8687 Alan Hayward <alan.hayward@arm.com>
8688 David Sherwood <david.sherwood@arm.com>
8690 * machmode.h (NARROWEST_INT_MODE): New macro.
8691 * expr.c (alignment_for_piecewise_move): Use it instead of
8692 GET_CLASS_NARROWEST_MODE (MODE_INT).
8693 (push_block): Likewise.
8694 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
8696 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
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 * postreload.c (move2add_valid_value_p): Change the type of the
8703 mode parameter to scalar_int_mode.
8704 (move2add_use_add2_insn): Add a mode parameter and use it instead
8706 (move2add_use_add3_insn): Likewise.
8707 (reload_cse_move2add): Update accordingly.
8709 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8710 Alan Hayward <alan.hayward@arm.com>
8711 David Sherwood <david.sherwood@arm.com>
8713 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
8715 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
8716 * optabs.c (expand_unop): Likewise.
8718 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8719 Alan Hayward <alan.hayward@arm.com>
8720 David Sherwood <david.sherwood@arm.com>
8722 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
8723 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
8724 (popcount_loc_descriptor): Likewise.
8725 (bswap_loc_descriptor): Likewise.
8726 (rotate_loc_descriptor): Likewise.
8727 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
8728 calling the functions above.
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 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
8736 (try_combine): Likewise.
8737 (simplify_if_then_else): Likewise.
8738 * cse.c (cse_insn): Likewise.
8739 * dwarf2out.c (mem_loc_descriptor): Likewise.
8740 * emit-rtl.c (gen_lowpart_common): Likewise.
8741 * simplify-rtx.c (simplify_truncation): Likewise.
8742 (simplify_binary_operation_1): Likewise.
8743 (simplify_const_relational_operation): Likewise.
8744 (simplify_ternary_operation): Likewise.
8745 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
8747 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8748 Alan Hayward <alan.hayward@arm.com>
8749 David Sherwood <david.sherwood@arm.com>
8751 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
8752 * reload.c (push_reload): Likewise.
8753 (find_reloads): Likewise.
8755 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8756 Alan Hayward <alan.hayward@arm.com>
8757 David Sherwood <david.sherwood@arm.com>
8759 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
8760 (make_compound_operation_int): Likewise.
8761 (change_zero_ext): Likewise.
8762 * expr.c (convert_move): Likewise.
8763 (convert_modes): Likewise.
8764 * fwprop.c (forward_propagate_subreg): Likewise.
8765 * loop-iv.c (get_biv_step_1): Likewise.
8766 * optabs.c (widen_operand): Likewise.
8767 * postreload.c (move2add_valid_value_p): Likewise.
8768 * recog.c (simplify_while_replacing): Likewise.
8769 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8770 (simplify_binary_operation_1): Likewise. Remove redundant
8771 mode equality check.
8773 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8774 Alan Hayward <alan.hayward@arm.com>
8775 David Sherwood <david.sherwood@arm.com>
8777 * combine.c (combine_simplify_rtx): Add checks for
8778 is_a <scalar_int_mode>.
8779 (simplify_if_then_else): Likewise.
8780 (make_field_assignment): Likewise.
8781 (simplify_comparison): Likewise.
8782 * ifcvt.c (noce_try_bitop): Likewise.
8783 * loop-invariant.c (canonicalize_address_mult): Likewise.
8784 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8786 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8787 Alan Hayward <alan.hayward@arm.com>
8788 David Sherwood <david.sherwood@arm.com>
8790 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
8791 is_a <scalar_int_mode> instead of != BLKmode.
8793 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8794 Alan Hayward <alan.hayward@arm.com>
8795 David Sherwood <david.sherwood@arm.com>
8797 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
8798 instead of != VOIDmode.
8799 * combine.c (if_then_else_cond): Likewise.
8800 (change_zero_ext): Likewise.
8801 * dwarf2out.c (mem_loc_descriptor): Likewise.
8802 (loc_descriptor): Likewise.
8803 * rtlanal.c (canonicalize_condition): Likewise.
8804 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
8806 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8807 Alan Hayward <alan.hayward@arm.com>
8808 David Sherwood <david.sherwood@arm.com>
8810 * simplify-rtx.c (simplify_binary_operation_1): Use
8811 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
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 * wide-int.h (int_traits<unsigned char>) New class.
8818 (int_traits<unsigned short>) Likewise.
8819 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
8820 Use GET_MODE_UNIT_PRECISION and remove redundant test for
8822 * combine.c (set_nonzero_bits_and_sign_copies): Use
8823 is_a <scalar_int_mode>.
8824 (find_split_point): Likewise.
8825 (combine_simplify_rtx): Likewise.
8826 (simplify_logical): Likewise.
8827 (expand_compound_operation): Likewise.
8828 (expand_field_assignment): Likewise.
8829 (make_compound_operation): Likewise.
8830 (extended_count): Likewise.
8831 (change_zero_ext): Likewise.
8832 (simplify_comparison): Likewise.
8833 * dwarf2out.c (scompare_loc_descriptor): Likewise.
8834 (ucompare_loc_descriptor): Likewise.
8835 (minmax_loc_descriptor): Likewise.
8836 (mem_loc_descriptor): Likewise.
8837 (loc_descriptor): Likewise.
8838 * expmed.c (init_expmed_one_mode): Likewise.
8839 * lra-constraints.c (lra_constraint_offset): Likewise.
8840 * optabs.c (prepare_libcall_arg): Likewise.
8841 * postreload.c (move2add_note_store): Likewise.
8842 * reload.c (operands_match_p): Likewise.
8843 * rtl.h (load_extend_op): Likewise.
8844 * rtlhooks.c (gen_lowpart_general): Likewise.
8845 * simplify-rtx.c (simplify_truncation): Likewise.
8846 (simplify_unary_operation_1): Likewise.
8847 (simplify_binary_operation_1): Likewise.
8848 (simplify_const_binary_operation): Likewise.
8849 (simplify_const_relational_operation): Likewise.
8850 (simplify_subreg): Likewise.
8851 * stor-layout.c (bitwise_mode_for_mode): Likewise.
8852 * var-tracking.c (adjust_mems): Likewise.
8853 (prepare_call_arguments): Likewise.
8855 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8856 Alan Hayward <alan.hayward@arm.com>
8857 David Sherwood <david.sherwood@arm.com>
8859 * machmode.h (is_int_mode): New fuction.
8860 * combine.c (find_split_point): Use it.
8861 (combine_simplify_rtx): Likewise.
8862 (simplify_if_then_else): Likewise.
8863 (simplify_set): Likewise.
8864 (simplify_shift_const_1): Likewise.
8865 (simplify_comparison): Likewise.
8866 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
8867 * cse.c (notreg_cost): Likewise.
8868 (cse_insn): Likewise.
8869 * cselib.c (cselib_lookup_1): Likewise.
8870 * dojump.c (do_jump_1): Likewise.
8871 (do_compare_rtx_and_jump): Likewise.
8872 * dse.c (get_call_args): Likewise.
8873 * dwarf2out.c (rtl_for_decl_init): Likewise.
8874 (native_encode_initializer): Likewise.
8875 * expmed.c (emit_store_flag_1): Likewise.
8876 (emit_store_flag): Likewise.
8877 * expr.c (convert_modes): Likewise.
8878 (store_field): Likewise.
8879 (expand_expr_real_1): Likewise.
8880 * fold-const.c (fold_read_from_constant_string): Likewise.
8881 * gimple-ssa-sprintf.c (get_format_string): Likewise.
8882 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
8883 * optabs.c (expand_binop): Likewise.
8884 (expand_unop): Likewise.
8885 (expand_abs_nojump): Likewise.
8886 (expand_one_cmpl_abs_nojump): Likewise.
8887 * simplify-rtx.c (mode_signbit_p): Likewise.
8888 (val_signbit_p): Likewise.
8889 (val_signbit_known_set_p): Likewise.
8890 (val_signbit_known_clear_p): Likewise.
8891 (simplify_relational_operation_1): Likewise.
8892 * tree.c (vector_type_mode): Likewise.
8894 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8895 Alan Hayward <alan.hayward@arm.com>
8896 David Sherwood <david.sherwood@arm.com>
8898 * machmode.h (smallest_mode_for_size): Fix formatting.
8899 (smallest_int_mode_for_size): New function.
8900 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
8901 instead of smallest_mode_for_size.
8902 * combine.c (make_extraction): Likewise.
8903 * config/arc/arc.c (arc_expand_movmem): Likewise.
8904 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
8905 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
8906 * config/s390/s390.c (s390_expand_insv): Likewise.
8907 * config/sparc/sparc.c (assign_int_registers): Likewise.
8908 * config/spu/spu.c (spu_function_value): Likewise.
8909 (spu_function_arg): Likewise.
8910 * coverage.c (get_gcov_type): Likewise.
8911 (get_gcov_unsigned_t): Likewise.
8912 * dse.c (find_shift_sequence): Likewise.
8913 * expmed.c (store_bit_field_1): Likewise.
8914 * expr.c (convert_move): Likewise.
8915 (store_field): Likewise.
8916 * internal-fn.c (expand_arith_overflow): Likewise.
8917 * optabs-query.c (get_best_extraction_insn): Likewise.
8918 * optabs.c (expand_twoval_binop_libfunc): Likewise.
8919 * stor-layout.c (layout_type): Likewise.
8920 (initialize_sizetypes): Likewise.
8921 * targhooks.c (default_get_mask_mode): Likewise.
8922 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
8924 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8925 Alan Hayward <alan.hayward@arm.com>
8926 David Sherwood <david.sherwood@arm.com>
8928 * machmode.h (opt_mode::else_blk): New function.
8929 (int_mode_for_mode): Declare.
8930 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
8931 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
8933 * cfgexpand.c (expand_debug_expr): Likewise.
8934 * combine.c (gen_lowpart_or_truncate): Likewise.
8935 (gen_lowpart_for_combine): Likewise.
8936 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
8937 * config/avr/avr.c (avr_to_int_mode): Likewise.
8938 (avr_out_plus_1): Likewise.
8939 (avr_out_plus): Likewise.
8940 (avr_out_round): Likewise.
8941 * config/i386/i386.c (ix86_split_to_parts): Likewise.
8942 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
8943 (s390_expand_vcond): Likewise.
8944 * config/spu/spu.c (spu_split_immediate): Likewise.
8945 (spu_expand_mov): Likewise.
8946 * dse.c (get_stored_val): Likewise.
8947 * expmed.c (store_bit_field_1): Likewise.
8948 (convert_extracted_bit_field): Use int_mode_for_mode instead of
8950 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
8951 (extract_low_bits): Likewise.
8952 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
8953 handling rather than repeating the check.
8954 (emit_group_store): Likewise.
8955 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
8956 * optabs.c (expand_absneg_bit): Likewise.
8957 (expand_copysign_absneg): Likewise.
8958 (expand_copysign_bit): Likewise.
8959 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
8960 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
8961 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
8962 * var-tracking.c (prepare_call_arguments): Likewise.
8963 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
8964 int_mode_for_mode instead of mode_for_size.
8965 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
8967 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8968 Alan Hayward <alan.hayward@arm.com>
8969 David Sherwood <david.sherwood@arm.com>
8971 * machmode.h (int_mode_for_size): New function.
8972 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
8973 instead of mode_for_size.
8974 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
8976 * combine.c (expand_field_assignment): Use int_mode_for_size
8977 instead of mode_for_size.
8978 (make_extraction): Likewise.
8979 (simplify_shift_const_1): Likewise.
8980 (simplify_comparison): Likewise.
8981 * dojump.c (do_jump): Likewise.
8982 * dwarf2out.c (mem_loc_descriptor): Likewise.
8983 * emit-rtl.c (init_derived_machine_modes): Likewise.
8984 * expmed.c (flip_storage_order): Likewise.
8985 (convert_extracted_bit_field): Likewise.
8986 * expr.c (copy_blkmode_from_reg): Likewise.
8987 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
8988 * internal-fn.c (expand_mul_overflow): Likewise.
8989 * lower-subreg.c (simple_move): Likewise.
8990 * optabs-libfuncs.c (init_optabs): Likewise.
8991 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8992 * tree.c (vector_type_mode): Likewise.
8993 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
8994 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
8995 * tree-vect-generic.c (expand_vector_parallel): Likewise.
8996 * tree-vect-stmts.c (vectorizable_load): Likewise.
8997 (vectorizable_store): Likewise.
8999 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9000 Alan Hayward <alan.hayward@arm.com>
9001 David Sherwood <david.sherwood@arm.com>
9003 * coretypes.h (pod_mode): New type.
9004 (scalar_int_mode_pod): New typedef.
9005 * machmode.h (pod_mode): New class.
9006 (int_n_data_t::m): Change type to scalar_int_mode_pod.
9007 * genmodes.c (emit_mode_int_n): Update accordingly.
9008 * lower-subreg.h (target_lower_subreg): Change type to
9009 scalar_int_mode_pod.
9010 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
9011 scalar_int_mode_pod.
9013 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9014 Alan Hayward <alan.hayward@arm.com>
9015 David Sherwood <david.sherwood@arm.com>
9017 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
9018 machine_mode to scalar_int_mode.
9019 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
9020 (rs6000_option_override_internal): Remove cast to int.
9021 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
9022 machine_mode to scalar_int_mode.
9023 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
9024 (rs6000_option_override_internal): Remove cast to int.
9025 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
9026 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
9028 * config/s390/s390.c (s390_expand_builtin): Likewise.
9029 * coretypes.h (scalar_int_mode): New type.
9030 (opt_scalar_int_mode): New typedef.
9031 * machmode.h (scalar_int_mode): New class.
9032 (scalar_int_mode::includes_p): New function.
9033 (byte_mode): Change type to scalar_int_mode.
9034 (word_mode): Likewise.
9035 (ptr_mode): Likewise.
9036 * emit-rtl.c (byte_mode): Likewise.
9037 (word_mode): Likewise.
9038 (ptr_mode): Likewise.
9039 (init_derived_machine_modes): Update accordingly.
9040 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
9041 and MODE_PARTIAL_INT.
9042 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
9043 opt_scalar_int_mode.
9045 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9046 Alan Hayward <alan.hayward@arm.com>
9047 David Sherwood <david.sherwood@arm.com>
9049 * target.def (libgcc_floating_mode_supported_p): Take a
9051 * doc/tm.texi: Regenerate.
9052 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
9054 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
9055 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
9058 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9059 Alan Hayward <alan.hayward@arm.com>
9060 David Sherwood <david.sherwood@arm.com>
9062 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
9063 * doc/tm.texi: Regenerate.
9064 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
9065 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
9066 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
9067 * targhooks.h (default_floatn_mode): Likewise.
9068 * targhooks.c (default_floatn_mode): Likewise.
9069 * tree.c (build_common_tree_nodes): Update accordingly.
9071 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9072 Alan Hayward <alan.hayward@arm.com>
9073 David Sherwood <david.sherwood@arm.com>
9075 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
9076 (mode_iterator::iterate_p): Likewise.
9077 (mode_iterator::get_wider): Likewise.
9078 * expr.c (init_expr_target): Use opt_scalar_float_mode.
9080 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9081 Alan Hayward <alan.hayward@arm.com>
9082 David Sherwood <david.sherwood@arm.com>
9084 * coretypes.h (opt_scalar_float_mode): New typedef.
9085 * machmode.h (float_mode_for_size): New function.
9086 * emit-rtl.c (double_mode): Delete.
9087 (init_emit_once): Use float_mode_for_size.
9088 * stor-layout.c (layout_type): Likewise.
9089 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
9091 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9092 Alan Hayward <alan.hayward@arm.com>
9093 David Sherwood <david.sherwood@arm.com>
9095 * output.h (assemble_real): Take a scalar_float_mode.
9096 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
9097 * config/arm/arm.md (consttable_4): Likewise.
9098 (consttable_8): Likewise.
9099 (consttable_16): Likewise.
9100 * config/mips/mips.md (consttable_float): Likewise.
9101 * config/s390/s390.c (s390_output_pool_entry): Likewise.
9102 * varasm.c (assemble_real): Take a scalar_float_mode.
9103 (output_constant_pool_2): Update accordingly.
9104 (output_constant): Likewise.
9106 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9107 Alan Hayward <alan.hayward@arm.com>
9108 David Sherwood <david.sherwood@arm.com>
9110 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
9111 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
9112 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
9113 (native_encode_real): Likewise.
9114 (native_interpret_real): Likewise.
9115 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
9116 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
9118 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9119 Alan Hayward <alan.hayward@arm.com>
9120 David Sherwood <david.sherwood@arm.com>
9122 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
9123 <scalar_float_mode>. Simplify.
9124 (gen_extend_conv_libfunc): Likewise.
9126 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9127 Alan Hayward <alan.hayward@arm.com>
9128 David Sherwood <david.sherwood@arm.com>
9130 * coretypes.h (scalar_float_mode): New type.
9131 * machmode.h (mode_traits::from_int): Use machine_mode if
9132 USE_ENUM_MODES is defined.
9133 (is_a): New function.
9135 (dyn_cast): Likewise.
9136 (scalar_float_mode): New class.
9137 (scalar_float_mode::includes_p): New function.
9138 (is_float_mode): Likewise.
9139 * gdbhooks.py (MachineModePrinter): New class.
9140 (build_pretty_printer): Use it for scalar_float_mode.
9141 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
9142 (format_helper::format_helper): Turn into a template.
9143 * genmodes.c (get_mode_class): New function.
9144 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
9145 or machine_mode if none.
9146 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
9147 as_a <scalar_float_mode>.
9148 * dwarf2out.c (mem_loc_descriptor): Likewise.
9149 (insert_float): Likewise.
9150 (add_const_value_attribute): Likewise.
9151 * simplify-rtx.c (simplify_immed_subreg): Likewise.
9152 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
9153 (expand_unop): Update accordingly.
9154 (expand_abs_nojump): Likewise.
9155 (expand_copysign_absneg): Take a scalar_float_mode.
9156 (expand_copysign_bit): Likewise.
9157 (expand_copysign): Update accordingly.
9159 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9160 Alan Hayward <alan.hayward@arm.com>
9161 David Sherwood <david.sherwood@arm.com>
9163 * coretypes.h (opt_mode): New class.
9164 * machmode.h (opt_mode): Likewise.
9165 (opt_mode::else_void): New function.
9166 (opt_mode::require): Likewise.
9167 (opt_mode::exists): Likewise.
9168 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
9169 (GET_MODE_2XWIDER_MODE): Likewise.
9170 (mode_iterator::get_wider): Update accordingly.
9171 (mode_iterator::get_2xwider): Likewise.
9172 (mode_iterator::get_known_wider): Likewise, turning into a template.
9173 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
9174 forcing a wider mode to exist.
9175 * config/cr16/cr16.h (LONG_REG_P): Likewise.
9176 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
9177 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
9178 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
9179 * lower-subreg.c (init_lower_subreg): Likewise.
9180 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
9181 on the final iteration.
9182 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
9183 a wider mode exists before asking for a move pattern.
9184 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
9185 forcing a wider mode to exist.
9186 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
9187 returning false if no such mode exists.
9188 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
9189 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
9190 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
9191 Avoid checking for a MODE_INT if we already know the mode is not a
9193 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
9194 forcing a wider mode to exist.
9195 (expmed_mult_highpart_optab): Likewise.
9196 (expmed_mult_highpart): Likewise.
9197 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
9199 * lto-streamer-in.c (lto_input_mode_table): Likewise.
9200 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
9201 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
9202 * internal-fn.c (expand_mul_overflow): Update use of
9203 GET_MODE_2XWIDER_MODE.
9204 * omp-low.c (omp_clause_aligned_alignment): Likewise.
9205 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
9206 GET_MODE_WIDER_MODE.
9207 (convert_plusminus_to_widen): Likewise.
9208 * tree-switch-conversion.c (array_value_type): Likewise.
9209 * var-tracking.c (emit_note_insn_var_location): Likewise.
9210 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
9211 Return false inside rather than outside the loop if no wider mode
9213 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
9214 and GET_MODE_2XWIDER_MODE
9215 (can_compare_p): Use else_void.
9216 * gdbhooks.py (OptMachineModePrinter): New class.
9217 (build_pretty_printer): Use it for opt_mode.
9219 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9220 Alan Hayward <alan.hayward@arm.com>
9221 David Sherwood <david.sherwood@arm.com>
9223 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
9224 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
9226 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9227 Alan Hayward <alan.hayward@arm.com>
9228 David Sherwood <david.sherwood@arm.com>
9230 * machmode.h (mode_traits): New structure.
9231 (get_narrowest_mode): New function.
9232 (mode_iterator::start): Likewise.
9233 (mode_iterator::iterate_p): Likewise.
9234 (mode_iterator::get_wider): Likewise.
9235 (mode_iterator::get_known_wider): Likewise.
9236 (mode_iterator::get_2xwider): Likewise.
9237 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
9238 (FOR_EACH_MODE): Likewise.
9239 (FOR_EACH_MODE_FROM): Likewise.
9240 (FOR_EACH_MODE_UNTIL): Likewise.
9241 (FOR_EACH_WIDER_MODE): Likewise.
9242 (FOR_EACH_2XWIDER_MODE): Likewise.
9243 * builtins.c (expand_builtin_strlen): Use new mode iterators.
9244 * combine.c (simplify_comparison): Likewise
9245 * config/i386/i386.c (type_natural_mode): Likewise.
9246 * cse.c (cse_insn): Likewise.
9247 * dse.c (find_shift_sequence): Likewise.
9248 * emit-rtl.c (init_derived_machine_modes): Likewise.
9249 (init_emit_once): Likewise.
9250 * explow.c (hard_function_value): Likewise.
9251 * expmed.c (extract_fixed_bit_field_1): Likewise.
9252 (extract_bit_field_1): Likewise.
9253 (expand_divmod): Likewise.
9254 (emit_store_flag_1): Likewise.
9255 * expr.c (init_expr_target): Likewise.
9256 (convert_move): Likewise.
9257 (alignment_for_piecewise_move): Likewise.
9258 (widest_int_mode_for_size): Likewise.
9259 (emit_block_move_via_movmem): Likewise.
9260 (copy_blkmode_to_reg): Likewise.
9261 (set_storage_via_setmem): Likewise.
9262 (compress_float_constant): Likewise.
9263 * omp-low.c (omp_clause_aligned_alignment): Likewise.
9264 * optabs-query.c (get_best_extraction_insn): Likewise.
9265 * optabs.c (expand_binop): Likewise.
9266 (expand_twoval_unop): Likewise.
9267 (expand_twoval_binop): Likewise.
9268 (widen_leading): Likewise.
9269 (widen_bswap): Likewise.
9270 (expand_parity): Likewise.
9271 (expand_unop): Likewise.
9272 (prepare_cmp_insn): Likewise.
9273 (prepare_float_lib_cmp): Likewise.
9274 (expand_float): Likewise.
9275 (expand_fix): Likewise.
9276 (expand_sfix_optab): Likewise.
9277 * postreload.c (move2add_use_add2_insn): Likewise.
9278 * reg-stack.c (reg_to_stack): Likewise.
9279 * reginfo.c (choose_hard_reg_mode): Likewise.
9280 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
9281 * stor-layout.c (mode_for_size): Likewise.
9282 (smallest_mode_for_size): Likewise.
9283 (mode_for_vector): Likewise.
9284 (finish_bitfield_representative): Likewise.
9285 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
9286 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
9287 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
9288 * var-tracking.c (prepare_call_arguments): Likewise.
9290 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9291 Alan Hayward <alan.hayward@arm.com>
9292 David Sherwood <david.sherwood@arm.com>
9294 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
9295 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
9296 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
9297 * machmode.h (mode_size): Move earlier in file.
9298 (mode_precision): Likewise.
9299 (mode_inner): Likewise.
9300 (mode_nunits): Likewise.
9301 (mode_unit_size): Likewise.
9302 (unit_unit_precision): Likewise.
9303 (mode_wider): Likewise.
9304 (mode_2xwider): Likewise.
9305 (machine_mode): New class.
9306 (mode_to_bytes): New function.
9307 (mode_to_bits): Likewise.
9308 (mode_to_precision): Likewise.
9309 (mode_to_inner): Likewise.
9310 (mode_to_unit_size): Likewise.
9311 (mode_to_unit_precision): Likewise.
9312 (mode_to_nunits): Likewise.
9313 (GET_MODE_SIZE): Use mode_to_bytes.
9314 (GET_MODE_BITSIZE): Use mode_to_bits.
9315 (GET_MODE_PRECISION): Use mode_to_precision.
9316 (GET_MODE_INNER): Use mode_to_inner.
9317 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
9318 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
9319 (GET_MODE_NUNITS): Use mode_to_nunits.
9320 * system.h (ALWAYS_INLINE): New macro.
9321 * config/powerpcspe/powerpcspe-c.c
9322 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
9323 int for arg1_mode and arg2_mode.
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 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
9330 Prefix mode names with E_ in case statements.
9331 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9332 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
9333 (aarch64_split_simd_move): Likewise.
9334 (aarch64_gen_storewb_pair): Likewise.
9335 (aarch64_gen_loadwb_pair): Likewise.
9336 (aarch64_gen_store_pair): Likewise.
9337 (aarch64_gen_load_pair): Likewise.
9338 (aarch64_get_condition_code_1): Likewise.
9339 (aarch64_constant_pool_reload_icode): Likewise.
9340 (get_rsqrte_type): Likewise.
9341 (get_rsqrts_type): Likewise.
9342 (get_recpe_type): Likewise.
9343 (get_recps_type): Likewise.
9344 (aarch64_gimplify_va_arg_expr): Likewise.
9345 (aarch64_simd_container_mode): Likewise.
9346 (aarch64_emit_load_exclusive): Likewise.
9347 (aarch64_emit_store_exclusive): Likewise.
9348 (aarch64_expand_compare_and_swap): Likewise.
9349 (aarch64_gen_atomic_cas): Likewise.
9350 (aarch64_emit_bic): Likewise.
9351 (aarch64_emit_atomic_swap): Likewise.
9352 (aarch64_emit_atomic_load_op): Likewise.
9353 (aarch64_evpc_trn): Likewise.
9354 (aarch64_evpc_uzp): Likewise.
9355 (aarch64_evpc_zip): Likewise.
9356 (aarch64_evpc_ext): Likewise.
9357 (aarch64_evpc_rev): Likewise.
9358 (aarch64_evpc_dup): Likewise.
9359 (aarch64_gen_ccmp_first): Likewise.
9360 (aarch64_gen_ccmp_next): Likewise.
9361 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
9362 (alpha_emit_xfloating_libcall): Likewise.
9363 (emit_insxl): Likewise.
9364 (alpha_arg_type): Likewise.
9365 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
9366 (arc_preferred_simd_mode): Likewise.
9367 (arc_secondary_reload): Likewise.
9368 (get_arc_condition_code): Likewise.
9369 (arc_print_operand): Likewise.
9370 (arc_legitimate_constant_p): Likewise.
9371 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9372 * config/arc/arc.md (casesi_load): Likewise.
9373 (casesi_compact_jump): Likewise.
9374 * config/arc/predicates.md (proper_comparison_operator): Likewise.
9375 (cc_use_register): Likewise.
9376 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9377 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
9378 (arm_init_iwmmxt_builtins): Likewise.
9379 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
9380 (neon_expand_vector_init): Likewise.
9381 (arm_attr_length_move_neon): Likewise.
9382 (maybe_get_arm_condition_code): Likewise.
9383 (arm_emit_vector_const): Likewise.
9384 (arm_preferred_simd_mode): Likewise.
9385 (arm_output_iwmmxt_tinsr): Likewise.
9386 (thumb1_output_casesi): Likewise.
9387 (thumb2_output_casesi): Likewise.
9388 (arm_emit_load_exclusive): Likewise.
9389 (arm_emit_store_exclusive): Likewise.
9390 (arm_expand_compare_and_swap): Likewise.
9391 (arm_evpc_neon_vuzp): Likewise.
9392 (arm_evpc_neon_vzip): Likewise.
9393 (arm_evpc_neon_vrev): Likewise.
9394 (arm_evpc_neon_vtrn): Likewise.
9395 (arm_evpc_neon_vext): Likewise.
9396 (arm_validize_comparison): Likewise.
9397 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
9398 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
9399 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
9400 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
9401 (c6x_preferred_simd_mode): Likewise.
9402 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
9403 (epiphany_rtx_costs): Likewise.
9404 * config/epiphany/predicates.md (proper_comparison_operator):
9406 * config/frv/frv.c (condexec_memory_operand): Likewise.
9407 (frv_emit_move): Likewise.
9408 (output_move_single): Likewise.
9409 (output_condmove_single): Likewise.
9410 (frv_hard_regno_mode_ok): Likewise.
9411 (frv_matching_accg_mode): Likewise.
9412 * config/h8300/h8300.c (split_adds_subs): Likewise.
9413 (h8300_rtx_costs): Likewise.
9414 (h8300_print_operand): Likewise.
9415 (compute_mov_length): Likewise.
9416 (output_logical_op): Likewise.
9417 (compute_logical_op_length): Likewise.
9418 (compute_logical_op_cc): Likewise.
9419 (h8300_shift_needs_scratch_p): Likewise.
9420 (output_a_shift): Likewise.
9421 (compute_a_shift_length): Likewise.
9422 (compute_a_shift_cc): Likewise.
9423 (expand_a_rotate): Likewise.
9424 (output_a_rotate): Likewise.
9425 * config/i386/i386.c (classify_argument): Likewise.
9426 (function_arg_advance_32): Likewise.
9427 (function_arg_32): Likewise.
9428 (function_arg_64): Likewise.
9429 (function_value_64): Likewise.
9430 (ix86_gimplify_va_arg): Likewise.
9431 (ix86_legitimate_constant_p): Likewise.
9432 (put_condition_code): Likewise.
9433 (split_double_mode): Likewise.
9434 (ix86_avx256_split_vector_move_misalign): Likewise.
9435 (ix86_expand_vector_logical_operator): Likewise.
9436 (ix86_split_idivmod): Likewise.
9437 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
9438 (ix86_build_const_vector): Likewise.
9439 (ix86_build_signbit_mask): Likewise.
9440 (ix86_match_ccmode): Likewise.
9441 (ix86_cc_modes_compatible): Likewise.
9442 (ix86_expand_branch): Likewise.
9443 (ix86_expand_sse_cmp): Likewise.
9444 (ix86_expand_sse_movcc): Likewise.
9445 (ix86_expand_int_sse_cmp): Likewise.
9446 (ix86_expand_vec_perm_vpermi2): Likewise.
9447 (ix86_expand_vec_perm): Likewise.
9448 (ix86_expand_sse_unpack): Likewise.
9449 (ix86_expand_int_addcc): Likewise.
9450 (ix86_split_to_parts): Likewise.
9451 (ix86_vectorize_builtin_gather): Likewise.
9452 (ix86_vectorize_builtin_scatter): Likewise.
9453 (avx_vpermilp_parallel): Likewise.
9454 (inline_memory_move_cost): Likewise.
9455 (ix86_tieable_integer_mode_p): Likewise.
9456 (x86_maybe_negate_const_int): Likewise.
9457 (ix86_expand_vector_init_duplicate): Likewise.
9458 (ix86_expand_vector_init_one_nonzero): Likewise.
9459 (ix86_expand_vector_init_one_var): Likewise.
9460 (ix86_expand_vector_init_concat): Likewise.
9461 (ix86_expand_vector_init_interleave): Likewise.
9462 (ix86_expand_vector_init_general): Likewise.
9463 (ix86_expand_vector_set): Likewise.
9464 (ix86_expand_vector_extract): Likewise.
9465 (emit_reduc_half): Likewise.
9466 (ix86_emit_i387_round): Likewise.
9467 (ix86_mangle_type): Likewise.
9468 (ix86_expand_round_sse4): Likewise.
9469 (expand_vec_perm_blend): Likewise.
9470 (canonicalize_vector_int_perm): Likewise.
9471 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
9472 (expand_vec_perm_1): Likewise.
9473 (expand_vec_perm_interleave3): Likewise.
9474 (expand_vec_perm_even_odd_pack): Likewise.
9475 (expand_vec_perm_even_odd_1): Likewise.
9476 (expand_vec_perm_broadcast_1): Likewise.
9477 (ix86_vectorize_vec_perm_const_ok): Likewise.
9478 (ix86_expand_vecop_qihi): Likewise.
9479 (ix86_expand_mul_widen_hilo): Likewise.
9480 (ix86_expand_sse2_abs): Likewise.
9481 (ix86_expand_pextr): Likewise.
9482 (ix86_expand_pinsr): Likewise.
9483 (ix86_preferred_simd_mode): Likewise.
9484 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
9485 * config/i386/sse.md (*andnot<mode>3): Likewise.
9486 (<mask_codefor><code><mode>3<mask_name>): Likewise.
9487 (*<code><mode>3): Likewise.
9488 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
9489 (ia64_expand_atomic_op): Likewise.
9490 (ia64_arg_type): Likewise.
9491 (ia64_mode_to_int): Likewise.
9492 (ia64_scalar_mode_supported_p): Likewise.
9493 (ia64_vector_mode_supported_p): Likewise.
9494 (expand_vec_perm_broadcast): Likewise.
9495 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
9496 (iq2000_function_arg_advance): Likewise.
9497 (iq2000_function_arg): Likewise.
9498 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
9499 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
9500 (m68k_libcall_value): Likewise.
9501 (m68k_function_value): Likewise.
9502 (sched_attr_op_type): Likewise.
9503 * config/mcore/mcore.c (mcore_output_move): Likewise.
9504 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
9506 (microblaze_function_arg): Likewise.
9507 * config/mips/mips.c (mips16_build_call_stub): Likewise.
9508 (mips_print_operand): Likewise.
9509 (mips_mode_ok_for_mov_fmt_p): Likewise.
9510 (mips_vector_mode_supported_p): Likewise.
9511 (mips_preferred_simd_mode): Likewise.
9512 (mips_expand_vpc_loongson_even_odd): Likewise.
9513 (mips_expand_vec_unpack): Likewise.
9514 (mips_expand_vi_broadcast): Likewise.
9515 (mips_expand_vector_init): Likewise.
9516 (mips_expand_vec_reduc): Likewise.
9517 (mips_expand_msa_cmp): Likewise.
9518 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
9519 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
9520 (cc_flags_for_mode): Likewise.
9521 * config/msp430/msp430.c (msp430_print_operand): Likewise.
9522 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
9523 (nds32_output_casesi_pc_relative): Likewise.
9524 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9525 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
9526 (nvptx_gen_unpack): Likewise.
9527 (nvptx_gen_pack): Likewise.
9528 (nvptx_gen_shuffle): Likewise.
9529 (nvptx_gen_wcast): Likewise.
9530 (nvptx_preferred_simd_mode): Likewise.
9531 * config/pa/pa.c (pa_secondary_reload): Likewise.
9532 * config/pa/predicates.md (base14_operand): Likewise.
9533 * config/powerpcspe/powerpcspe-c.c
9534 (altivec_resolve_overloaded_builtin): Likewise.
9535 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
9537 (rs6000_preferred_simd_mode): Likewise.
9538 (output_vec_const_move): Likewise.
9539 (rs6000_expand_vector_extract): Likewise.
9540 (rs6000_split_vec_extract_var): Likewise.
9541 (reg_offset_addressing_ok_p): Likewise.
9542 (rs6000_legitimate_offset_address_p): Likewise.
9543 (rs6000_legitimize_address): Likewise.
9544 (rs6000_emit_set_const): Likewise.
9545 (rs6000_const_vec): Likewise.
9546 (rs6000_emit_move): Likewise.
9547 (spe_build_register_parallel): Likewise.
9548 (rs6000_darwin64_record_arg_recurse): Likewise.
9549 (swap_selector_for_mode): Likewise.
9550 (spe_init_builtins): Likewise.
9551 (paired_init_builtins): Likewise.
9552 (altivec_init_builtins): Likewise.
9553 (do_load_for_compare): Likewise.
9554 (rs6000_generate_compare): Likewise.
9555 (rs6000_expand_float128_convert): Likewise.
9556 (emit_load_locked): Likewise.
9557 (emit_store_conditional): Likewise.
9558 (rs6000_output_function_epilogue): Likewise.
9559 (rs6000_handle_altivec_attribute): Likewise.
9560 (rs6000_function_value): Likewise.
9561 (emit_fusion_gpr_load): Likewise.
9562 (emit_fusion_p9_load): Likewise.
9563 (emit_fusion_p9_store): Likewise.
9564 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
9565 (fusion_gpr_mem_load): Likewise.
9566 (fusion_addis_mem_combo_load): Likewise.
9567 (fusion_addis_mem_combo_store): Likewise.
9568 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
9569 (fusion_gpr_mem_load): Likewise.
9570 (fusion_addis_mem_combo_load): Likewise.
9571 (fusion_addis_mem_combo_store): Likewise.
9572 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9574 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
9575 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
9576 (rs6000_preferred_simd_mode): Likewise.
9577 (output_vec_const_move): Likewise.
9578 (rs6000_expand_vector_extract): Likewise.
9579 (rs6000_split_vec_extract_var): Likewise.
9580 (reg_offset_addressing_ok_p): Likewise.
9581 (rs6000_legitimate_offset_address_p): Likewise.
9582 (rs6000_legitimize_address): Likewise.
9583 (rs6000_emit_set_const): Likewise.
9584 (rs6000_const_vec): Likewise.
9585 (rs6000_emit_move): Likewise.
9586 (rs6000_darwin64_record_arg_recurse): Likewise.
9587 (swap_selector_for_mode): Likewise.
9588 (paired_init_builtins): Likewise.
9589 (altivec_init_builtins): Likewise.
9590 (rs6000_expand_float128_convert): Likewise.
9591 (emit_load_locked): Likewise.
9592 (emit_store_conditional): Likewise.
9593 (rs6000_output_function_epilogue): Likewise.
9594 (rs6000_handle_altivec_attribute): Likewise.
9595 (rs6000_function_value): Likewise.
9596 (emit_fusion_gpr_load): Likewise.
9597 (emit_fusion_p9_load): Likewise.
9598 (emit_fusion_p9_store): Likewise.
9599 * config/rx/rx.c (rx_gen_move_template): Likewise.
9600 (flags_from_mode): Likewise.
9601 * config/s390/predicates.md (s390_alc_comparison): Likewise.
9602 (s390_slb_comparison): Likewise.
9603 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
9604 (s390_vector_mode_supported_p): Likewise.
9605 (s390_cc_modes_compatible): Likewise.
9606 (s390_match_ccmode_set): Likewise.
9607 (s390_canonicalize_comparison): Likewise.
9608 (s390_emit_compare_and_swap): Likewise.
9609 (s390_branch_condition_mask): Likewise.
9610 (s390_rtx_costs): Likewise.
9611 (s390_secondary_reload): Likewise.
9612 (__SECONDARY_RELOAD_CASE): Likewise.
9613 (s390_expand_cs): Likewise.
9614 (s390_preferred_simd_mode): Likewise.
9615 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
9616 * config/sh/sh.c (sh_print_operand): Likewise.
9617 (dump_table): Likewise.
9618 (sh_secondary_reload): Likewise.
9619 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9620 * config/sh/sh.md (casesi_worker_1): Likewise.
9621 (casesi_worker_2): Likewise.
9622 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
9623 (fcc_comparison_operator): Likewise.
9624 * config/sparc/sparc.c (sparc_expand_move): Likewise.
9625 (emit_soft_tfmode_cvt): Likewise.
9626 (sparc_preferred_simd_mode): Likewise.
9627 (output_cbranch): Likewise.
9628 (sparc_print_operand): Likewise.
9629 (sparc_expand_vec_perm_bmask): Likewise.
9630 (vector_init_bshuffle): Likewise.
9631 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
9632 (spu_vector_mode_supported_p): Likewise.
9633 (spu_expand_insv): Likewise.
9634 (spu_emit_branch_or_set): Likewise.
9635 (spu_handle_vector_attribute): Likewise.
9636 (spu_builtin_splats): Likewise.
9637 (spu_builtin_extract): Likewise.
9638 (spu_builtin_promote): Likewise.
9639 (spu_expand_sign_extend): Likewise.
9640 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
9641 (tilegx_simd_int): Likewise.
9642 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
9643 (tilepro_simd_int): Likewise.
9644 * config/v850/v850.c (const_double_split): Likewise.
9645 (v850_print_operand): Likewise.
9646 (ep_memory_offset): Likewise.
9647 * config/vax/vax.c (vax_rtx_costs): Likewise.
9648 (vax_output_int_move): Likewise.
9649 (vax_output_int_add): Likewise.
9650 (vax_output_int_subtract): Likewise.
9651 * config/visium/predicates.md (visium_branch_operator): Likewise.
9652 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
9653 (visium_print_operand_address): Likewise.
9654 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
9655 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
9656 (xtensa_expand_conditional_branch): Likewise.
9657 (xtensa_copy_incoming_a7): Likewise.
9658 (xtensa_output_literal): Likewise.
9659 * dfp.c (decimal_real_maxval): Likewise.
9660 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
9662 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9663 Alan Hayward <alan.hayward@arm.com>
9664 David Sherwood <david.sherwood@arm.com>
9666 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
9667 (mode_nunits_inline): Likewise.
9668 (mode_inner_inline): Likewise.
9669 (mode_unit_size_inline): Likewise.
9670 (mode_unit_precision_inline): Likewise.
9671 (emit_insn_modes_h): Likewise. Also emit a #define of the
9673 (emit_mode_wider): Add an E_ prefix to mode names.
9674 (emit_mode_complex): Likewise.
9675 (emit_mode_inner): Likewise.
9676 (emit_mode_adjustments): Likewise.
9677 (emit_mode_int_n): Likewise.
9678 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
9679 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
9680 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
9681 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
9682 (CRC32_BUILTIN, ENTRY): Likewise.
9683 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
9684 (aarch64_pop_regs): Likewise.
9685 (aarch64_process_components): Likewise.
9686 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
9687 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
9688 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
9689 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
9690 * config/arm/arm.c (arm_init_libfuncs): Likewise.
9691 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
9693 * config/i386/i386-builtin.def (pcmpestr): Likewise.
9694 (pcmpistr): Likewise.
9695 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
9696 * config/mmix/mmix.c (mmix_output_condition): Likewise.
9697 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
9699 * config/rl78/rl78.c (mduc_regs): Likewise.
9700 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
9701 (htm_expand_builtin): Likewise.
9702 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
9703 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
9704 * config/xtensa/xtensa.c (print_operand): Likewise.
9705 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
9706 (NUM_MODE_VECTOR_INT): Likewise.
9707 * genoutput.c (null_operand): Likewise.
9708 (output_operand_data): Likewise.
9709 * genrecog.c (print_parameter_value): Likewise.
9710 * lra.c (debug_operand_data): Likewise.
9712 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9714 * dwarf2out.c (scompare_loc_descriptor_wide)
9715 (scompare_loc_descriptor_narrow): New functions, split out from...
9716 (scompare_loc_descriptor): ...here.
9717 * expmed.c (emit_store_flag_int): New function, split out from...
9718 (emit_store_flag): ...here.
9720 2017-08-30 Richard Biener <rguenther@suse.de>
9722 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
9723 (dwarf2out_early_finish): Move setting of AT_pubnames from
9724 early debug output to early finish.
9726 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
9728 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
9729 and -mdata-region to the assembler.
9731 2017-08-30 Richard Biener <rguenther@suse.de>
9733 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
9735 (gen_subprogram_die): Add DW_AT_object_pointer only early.
9736 (dwarf2out_early_global_decl): Only generate a DIE for the
9737 abstract origin if it doesn't already exist or is a declaration DIE.
9738 (resolve_addr): Do not add the linkage name twice when
9739 generating a stub DIE for the DW_TAG_GNU_call_site target.
9741 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9743 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9744 Use machine_mode rather than int for arg1_mode.
9746 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
9749 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
9750 that the second argument of the built-in functions to unpack
9751 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
9752 switch statement instead a lot of if statements.
9753 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
9754 Allow 64-bit values to be in Altivec registers as well as
9755 traditional floating point registers.
9756 (pack<mode>, FMOVE128_VSX iterator): Likewise.
9758 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
9760 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
9761 MAX_REGS_PER_ADDRESS == 1.
9763 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
9765 * config/i386/i386.opt (flag_fentry): Do not init to -1.
9766 * config/i386/i386.c (ix86_option_override_internal): Simplify
9767 setting of opts->x_flag_entry.
9769 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9770 Jakub Jelinek <jakub@redhat.com>
9771 Richard Biener <rguenther@suse.de>
9773 PR tree-optimization/81503
9774 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
9775 folded constant fits in the target type; reorder tests for clarity.
9777 2017-08-29 Martin Liska <mliska@suse.cz>
9779 * passes.def: Include pass_lower_switch.
9780 * stmt.c (dump_case_nodes): Remove and move to
9781 tree-switch-conversion.
9782 (case_values_threshold): Likewise.
9783 (expand_switch_as_decision_tree_p): Likewise.
9784 (emit_case_decision_tree): Likewise.
9785 (expand_case): Likewise.
9786 (balance_case_nodes): Likewise.
9787 (node_has_low_bound): Likewise.
9788 (node_has_high_bound): Likewise.
9789 (node_is_bounded): Likewise.
9790 (emit_case_nodes): Likewise.
9791 (struct simple_case_node): New struct.
9792 (add_case_node): Remove.
9793 (emit_case_dispatch_table): Use vector instead of case_list.
9794 (reset_out_edges_aux): Remove.
9795 (compute_cases_per_edge): Likewise.
9796 (expand_case): Build list of simple_case_node.
9797 (expand_sjlj_dispatch_table): Use it.
9798 * tree-switch-conversion.c (struct case_node): Moved from
9799 stmt.c and adjusted.
9800 (emit_case_nodes): Likewise.
9801 (node_has_low_bound): Likewise.
9802 (node_has_high_bound): Likewise.
9803 (node_is_bounded): Likewise.
9804 (case_values_threshold): Likewise.
9805 (reset_out_edges_aux): Likewise.
9806 (compute_cases_per_edge): Likewise.
9807 (add_case_node): Likewise.
9808 (dump_case_nodes): Likewise.
9809 (balance_case_nodes): Likewise.
9810 (expand_switch_as_decision_tree_p): Likewise.
9811 (emit_jump): Likewise.
9812 (emit_case_decision_tree): Likewise.
9813 (try_switch_expansion): Likewise.
9814 (do_jump_if_equal): Likewise.
9815 (emit_cmp_and_jump_insns): Likewise.
9816 (fix_phi_operands_for_edge): New function.
9817 (record_phi_operand_mapping): Likewise.
9818 (class pass_lower_switch): New pass.
9819 (pass_lower_switch::execute): New function.
9820 (make_pass_lower_switch): Likewise.
9821 (conditional_probability):
9822 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
9823 * tree-pass.h: Add make_pass_lower_switch.
9825 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
9828 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
9831 2017-08-29 Richard Biener <rguenther@suse.de>
9833 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
9834 we do not add a DW_AT_inline attribute twice.
9835 (gen_subprogram_die): Remove code setting DW_AT_inline on
9836 DECL_ABSTRACT_P nodes.
9838 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
9840 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
9841 calls to internal functions.
9842 (gimplify_modify_expr): Likewise.
9843 * tree-call-cdce.c (use_internal_fn): Likewise.
9844 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
9845 (convert_to_divmod): Set the nothrow flag.
9846 * tree-if-conv.c (predicate_mem_writes): Likewise.
9847 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
9848 (vectorizable_call): Likewise.
9849 (vectorizable_store): Likewise.
9850 (vectorizable_load): Likewise.
9851 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
9852 (vect_recog_mask_conversion_pattern): Likewise.
9854 2017-08-29 Martin Liska <mliska@suse.cz>
9857 * gcc.c (driver_handle_option): Add new argument.
9858 * opts-common.c (handle_option): Pass
9859 target_option_override_hook.
9860 * opts-global.c (lang_handle_option): Add new option.
9861 (set_default_handlers): Add new argument.
9862 (decode_options): Likewise.
9863 * opts.c (target_handle_option): Likewise.
9864 (common_handle_option): Call target_option_override_hook.
9865 * opts.h (struct cl_option_handler_func): Add hook for
9866 target option override.
9867 (struct cl_option_handlers): Likewise.
9868 (set_default_handlers): Add new argument.
9869 (decode_options): Likewise.
9870 (common_handle_option): Likewise.
9871 (target_handle_option): Likewise.
9872 * toplev.c (toplev::main): Pass targetm.target_option.override
9875 2017-08-29 Richard Biener <rguenther@suse.de>
9876 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
9878 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
9879 life to the active subtree.
9881 2017-08-28 Jeff Law <law@redhat.com>
9883 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
9884 derive_equivalences.
9885 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
9887 (edge_info::derive_equivalences): New private member function
9889 * tree-ssa-dom.c (class edge_info): Changed from a struct
9890 to a class. Add ctor/dtor, methods and data members.
9891 (edge_info::edge_info): Renamed from allocate_edge_info.
9892 Initialize additional members.
9893 (edge_info::~edge_info): New.
9894 (free_dom_edge_info): Delete the edge info.
9895 (record_edge_info): Use new class & associated member functions.
9896 Tighten forms for testing for edge equivalences.
9897 (record_temporary_equivalences): Iterate over the simple
9898 equivalences rather than assuming there's only one per edge.
9899 (cprop_into_successor_phis): Iterate over the simple
9900 equivalences rather than assuming there's only one per edge.
9901 (optimize_stmt): Use operand_equal_p rather than pointer
9902 equality for mini-DSE code.
9904 2017-08-28 Nathan Sidwell <nathan@acm.org>
9906 * gcc.c (execute): Fold SIGPIPE handling into switch
9907 statement. Adjust internal error message.
9909 2017-08-28 Richard Biener <rguenther@suse.de>
9912 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
9913 Do nothing for removed DIEs.
9915 2017-08-28 Richard Biener <rguenther@suse.de>
9917 PR tree-optimization/81977
9918 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
9921 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
9924 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
9925 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
9926 using targetm.gen_mem_thread_fence.
9928 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
9931 * config/i386/i386.md (*<btsc><mode>): Change operand 2
9932 predicate to register_operand. Reorder operands.
9933 (*btr<mode>): Ditto.
9934 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
9935 (*btr<mode>_mask): Ditto.
9937 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
9939 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
9940 * config/rs6000/xmmintrin.h: New file.
9941 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
9943 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9946 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
9947 parameter and_insn and return it.
9948 (recombine_lvx_pattern): Insert a copy to ensure availability of
9949 the base register of the copied masking operation at the point of
9950 the instruction replacement.
9951 (recombine_stvx_pattern): Likewise.
9953 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
9955 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
9956 undocumented switches.
9957 (-mpower9-dform-vector): Likewise.
9958 (-mpower9-dform): Likewise.
9959 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
9960 comments to delete references to -mpower9-dform* switches.
9961 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
9962 Delete reference to -mpower9-dform* switches, test for
9963 -mpower9-vector instead.
9964 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
9965 (OTHER_P9_VECTOR_MASKS): Likewise.
9966 (POWERPC_MASKS): Likewise.
9967 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
9968 tests against -mpower9-dform* to -mpower9-vector. Delete code
9969 that checked for -mpower9-dform* consistancy with other options.
9970 Add test for -mpower9-misc to enable other power9 switches.
9971 (rs6000_init_hard_regno_mode_ok): Likewise.
9972 (rs6000_option_override_internal): Likewise.
9973 (rs6000_emit_prologue): Likewise.
9974 (rs6000_emit_epilogue): Likewise.
9975 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
9976 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
9977 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
9979 (emit_fusion_p9_store): Likewise.
9980 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
9981 resetting these macros if the assembler does not support ISA 3.0
9983 (TARGET_P9_DFORM_VECTOR): Likewise.
9984 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
9985 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
9987 2017-08-25 Alan Modra <amodra@gmail.com>
9989 PR rtl-optimization/81747
9990 * cse.c (cse_extended_basic_block): Don't attempt to record
9991 equivalences for degenerate conditional branches that branch
9992 to their fall-through.
9994 2017-08-24 Martin Sebor <msebor@redhat.com>
9997 * gimple-fold.c (size_must_be_zero_p): New function.
9998 (gimple_fold_builtin_memory_op): Call it.
10000 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
10002 * config/rs6000/mm_malloc.h: New file.
10004 2017-08-24 Bin Cheng <bin.cheng@arm.com>
10006 PR tree-optimization/81913
10007 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
10008 analysis when either IVs in condition can wrap.
10010 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
10012 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
10013 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
10015 2017-08-24 Richard Biener <rguenther@suse.de>
10018 * targhooks.c (default_target_can_inline_p): Properly
10019 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
10020 is present and always compare.
10021 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
10022 infer -mfpmath=sse from TARGET_SSE_P.
10023 (ix86_can_inline_p): Properly use target_option_default_node when
10024 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
10026 2017-08-24 Richard Biener <rguenther@suse.de>
10029 * dwarf2out.c (output_die): Handle flag_generate_offload like
10031 (output_comp_unit): Likewise.
10032 (gen_array_type_die): Likewise.
10033 (dwarf2out_early_finish): Likewise.
10034 (note_variable_value_in_expr): Likewise.
10035 (dwarf2out_finish): Likewise. Adjust assert.
10036 * cgraphunit.c (symbol_table::compile): Move setting of
10037 flag_generate_offload earlier ...
10038 (symbol_table::finalize_compilation_unit): ... here, before
10039 early debug finalization.
10041 2017-08-24 Richard Biener <rguenther@suse.de>
10043 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
10044 and ipa-fnsummary.h.
10045 (ix86_can_inline_p): When ix86_fpmath flags do not match
10046 check whether the callee uses FP math at all.
10048 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
10050 PR middle-end/81931
10051 * tree-ssanames.c (get_nonzero_bits): Use element_precision
10052 instead of TYPE_PRECISION.
10054 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
10055 Alan Hayward <alan.hayward@arm.com>
10056 David Sherwood <david.sherwood@arm.com>
10058 * combine.c (make_extraction): Use subreg_offset_from_lsb.
10060 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
10062 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
10064 (ix86_frame::stack_realign_allocate): New field.
10065 (struct machine_frame_state): Modify comments.
10066 (machine_frame_state::sp_realigned_fp_end): New field.
10067 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
10068 layout calculation.
10069 (sp_valid_at): Add assertion to assure no attempt to access invalid
10070 offset of a realigned stack.
10071 (fp_valid_at): Likewise.
10072 (choose_baseaddr): Modify comments.
10073 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
10074 ix86_expand_prologue.
10075 (ix86_expand_prologue): Modify stack realignment and allocation.
10076 (ix86_expand_epilogue): Modify comments.
10077 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
10078 avx2_runtime, avx512f, and avx512f_runtime.
10080 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
10082 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
10083 (mstackrealign): Do not init to -1.
10084 * config/i386/i386.c (ix86_option_override_internal):
10085 Check opts_set, not opts when setting default value of
10086 opts->x_ix86_force_align_arg_pointer.
10088 2017-08-23 Richard Biener <rguenther@suse.de>
10090 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
10091 lang_hooks.decl_printable_name.
10092 * print-rtl-function.c (print_rtx_function): Likewise.
10093 * tree-pretty-print.c (dump_function_header): Likewise.
10095 2017-08-23 Richard Biener <rguenther@suse.de>
10098 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
10099 -g0 at compile-time.
10101 2017-08-23 Tamar Christina <tamar.christina@arm.com>
10103 PR middle-end/19706
10104 * doc/sourcebuild.texi (Other hardware attributes):
10107 2017-08-23 Tamar Christina <tamar.christina@arm.com>
10109 PR middle-end/19706
10110 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
10111 Fix single-use check.
10113 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
10115 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
10117 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
10119 * doc/install.texi: Modify to add more details on running selected
10122 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
10124 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
10125 is combined with -mabi=ms.
10126 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
10129 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10131 PR tree-optimization/81488
10132 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
10133 and cached_basis fields.
10134 (MAX_SPREAD): New constant.
10135 (alloc_cand_and_find_basis): Initialize new fields.
10136 (clear_visited): New function.
10137 (create_phi_basis_1): Rename from create_phi_basis, set visited
10138 and cached_basis fields.
10139 (create_phi_basis): New wrapper function.
10140 (phi_add_costs_1): Rename from phi_add_costs, add spread
10141 parameter, set visited field, short-circuit when limits reached.
10142 (phi_add_costs): New wrapper function.
10143 (record_phi_increments_1): Rename from record_phi_increments, set
10145 (record_phi_increments): New wrapper function.
10146 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
10147 (phi_incr_cost): New wrapper function.
10148 (all_phi_incrs_profitable_1): Rename from
10149 all_phi_incrs_profitable, set visited field.
10150 (all_phi_incrs_profitable): New wrapper function.
10152 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
10153 Alan Hayward <alan.hayward@arm.com>
10154 David Sherwood <david.sherwood@arm.com>
10156 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
10157 that takes the outer and inner modes.
10158 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
10159 comparison as the canonical test for a paradoxical subreg.
10160 * combine.c (simplify_set): Use paradoxical_subreg_p.
10161 (make_extraction): Likewise.
10162 (force_to_mode): Likewise.
10163 (rtx_equal_for_field_assignment_p): Likewise.
10164 (gen_lowpart_for_combine): Likewise.
10165 (simplify_comparison): Likewise.
10166 * cse.c (equiv_constant): Likewise.
10167 * expmed.c (store_bit_field_1): Likewise.
10168 * final.c (alter_subreg): Likewise.
10169 * fwprop.c (propagate_rtx): Likewise.
10170 (forward_propagate_subreg): Likewise.
10171 * ira-conflicts.c (ira_build_conflicts): Likewise.
10172 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
10173 * lra-constraints.c (curr_insn_transform): Likewise.
10174 (split_reg): Likewise.
10175 * lra-eliminations.c (move_plus_up): Likewise.
10176 (lra_eliminate_regs_1): Likewise.
10177 * recog.c (general_operand): Likewise.
10178 * ree.c (combine_reaching_defs): Likewise.
10179 * reload.c (push_reload): Likewise.
10180 (find_reloads): Likewise.
10181 * reload1.c (elimination_effects): Likewise.
10182 (compute_reload_subreg_offset): Likewise.
10183 (choose_reload_regs): Likewise.
10184 * rtlanal.c (subreg_lsb_1): Likewise.
10185 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10186 (simplify_subreg): Likewise.
10187 * var-tracking.c (track_loc_p): Likewise.
10188 * emit-rtl.c (byte_lowpart_offset): Likewise.
10189 (paradoxical_subreg_p): Delete out-of-line definition.
10191 2017-08-22 Jeff Law <law@redhat.com>
10193 PR tree-optimization/81741
10194 PR tree-optimization/71947
10195 * tree-ssa-dom.c: Include tree-inline.h.
10196 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
10197 equivalences if one is more expensive to compute than the other.
10198 * tree-ssa-scopedtables.h (class const_or_copies): Make
10199 record_const_or_copy_raw method private.
10200 (class avail_exprs_stack): New method simplify_binary_operation.
10201 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
10202 avail_exprs_stack::simplify_binary_operation as needed.
10203 (avail_exprs_stack::simplify_binary_operation): New function.
10205 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
10207 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
10208 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
10209 (DOT_SYMBOLS): Likewise.
10210 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
10211 (RELOCATABLE_NEEDS_FIXUP): Likewise.
10212 (RS6000_ABI_NAME): Likewise.
10213 (TARGET_CMODEL): Likewise.
10214 (TOC_SECTION_ASM_OP): Likewise.
10215 (SET_CMODEL): New macro.
10216 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
10218 2017-08-22 Richard Biener <rguenther@suse.de>
10220 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
10221 to allow for free-lang-data replacements similar to verify_type_variant.
10223 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
10224 Alan Hayward <alan.hayward@arm.com>
10225 David Sherwood <david.sherwood@arm.com>
10227 * config/aarch64/aarch64.md (casesi): Use DImode rather than
10228 VOIDmode for the LABEL_REF.
10230 2017-08-22 Richard Biener <rguenther@suse.de>
10232 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
10234 2017-08-22 Richard Biener <rguenther@suse.de>
10236 * common.opt (feliminate-dwarf2-dups): Ignore.
10237 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
10238 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
10239 same_die_p_wrap, compute_section_prefix,
10240 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
10241 (comdat_symbol_id, comdat_symbol_number): Likewise.
10242 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
10244 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
10245 (output_die): Mark unreachable path unreachable.
10246 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
10247 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
10248 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
10249 (dwarf2out_early_finish): Likewise.
10251 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
10253 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
10255 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
10258 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
10259 not VAR_P. Filter attribute warnings with OPT_Wattributes.
10260 (avr_attribute_table) <io, io_low, address>: Initialize
10261 .decl_required with true.
10263 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
10265 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
10266 undocumented debugging options.
10267 (-mvsx-scalar-double): Likewise.
10268 (-mallow-df-permute): Likewise.
10269 (-mvectorize-builtins): Likewise.
10270 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
10271 (rs6000_builtin_vectorized_function): Likewise.
10272 (rs6000_builtin_md_vectorized_function): Likewise.
10273 (rs6000_opt_vars): Likewise.
10275 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
10278 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
10279 (*btrq_imm): Rename from *btrq.
10280 (*btcq_imm): Rename from *btcq.
10281 (btsc): New code attribute.
10282 (*<btsc><mode>): New insn pattern.
10283 (*btr<mode>): Ditto.
10284 (*<btsc><mode>_mask): New insn_and_split pattern.
10285 (*btr<mode>_mask): Ditto.
10287 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10288 Alan Hayward <alan.hayward@arm.com>
10289 David Sherwood <david.sherwood@arm.com>
10291 * function.c (pad_below): Simplify padding calculation.
10293 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10294 Alan Hayward <alan.hayward@arm.com>
10295 David Sherwood <david.sherwood@arm.com>
10297 * target.def (function_prologue): Remove frame size argument.
10298 (function_epilogue): Likewise.
10299 * doc/tm.texi: Regenerate.
10300 * final.c (final_start_function): Update call to function_prologue.
10301 (final_end_function): Update call to function_epilogue.
10302 (default_function_pro_epilogue): Remove frame size argument.
10303 * output.h (default_function_pro_epilogue): Likewise.
10304 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
10305 (arm_output_function_prologue): Likewise.
10306 * config/frv/frv.c (frv_function_prologue): Likewise.
10307 (frv_function_epilogue): Likewise.
10308 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
10309 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
10310 (ia64_output_function_epilogue): Likewise.
10311 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
10312 (m32r_output_function_epilogue): Likewise.
10313 * config/microblaze/microblaze.c (microblaze_function_prologue)
10314 (microblaze_function_epilogue): Likewise.
10315 * config/mips/mips.c (mips_output_function_prologue): Likewise.
10316 (mips_output_function_epilogue): Likewise.
10317 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
10318 (mmix_target_asm_function_epilogue): Likewise.
10319 * config/msp430/msp430.c (msp430_start_function): Likewise.
10320 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
10321 (nds32_asm_function_epilogue): Likewise.
10322 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
10323 * config/pa/pa.c (pa_output_function_prologue): Likewise.
10324 (pa_output_function_epilogue): Likewise.
10325 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
10326 (rs6000_output_function_epilogue): Likewise.
10327 * config/rl78/rl78.c (rl78_start_function): Likewise.
10328 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
10329 (rs6000_output_function_epilogue): Likewise.
10330 * config/rx/rx.c (rx_output_function_prologue): Likewise.
10331 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
10332 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
10333 (sparc_asm_function_epilogue): Likewise.
10335 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10337 * tree.h (type_has_mode_precision_p): New function.
10338 * convert.c (convert_to_integer_1): Use it.
10339 * expr.c (expand_expr_real_2): Likewise.
10340 (expand_expr_real_1): Likewise.
10341 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
10342 * match.pd: Likewise.
10343 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
10344 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
10345 * tree-tailcall.c (process_assignment): Likewise.
10346 * tree-vect-loop.c (vectorizable_reduction): Likewise.
10347 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
10348 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
10349 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10350 (vectorizable_assignment): Likewise.
10351 (vectorizable_shift): Likewise.
10352 (vectorizable_operation): Likewise.
10353 * tree-vrp.c (register_edge_assert_for_2): Likewise.
10355 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
10357 * match.pd: Add pow (C, x) simplification.
10359 2017-08-21 Richard Biener <rguenther@suse.de>
10361 PR tree-optimization/81900
10362 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
10363 for blocks with abnormal predecessors.
10364 (compute_antic): Do not set visited flag prematurely.
10366 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
10369 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
10371 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10373 * stor-layout.h (vector_type_mode): Move to...
10374 * tree.h (vector_type_mode): ...here.
10375 * stor-layout.c (vector_type_mode): Move to...
10376 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
10378 2017-08-21 Richard Biener <rguenther@suse.de>
10380 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
10381 register_external_die hooks.
10382 (debug_false_tree_charstarstar_uhwistar): Declare.
10383 (debug_nothing_tree_charstar_uhwi): Likewise.
10384 * debug.c (do_nothing_debug_hooks): Adjust.
10385 (debug_false_tree_charstarstar_uhwistar): New do nothing.
10386 (debug_nothing_tree_charstar_uhwi): Likewise.
10387 * dbxout.c (dbx_debug_hooks): Adjust.
10388 (xcoff_debug_hooks): Likewise.
10389 * sdbout.c (sdb_debug_hooks): Likewise.
10390 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
10391 * dwarf2out.c (macinfo_label_base): New global.
10392 (dwarf2out_register_external_die): New function for the
10393 register_external_die hook.
10394 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
10395 (dwarf2_debug_hooks): Use them.
10396 (dwarf2_lineno_debug_hooks): Adjust.
10397 (struct die_struct): Add with_offset flag.
10398 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
10399 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
10400 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
10401 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
10402 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
10403 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
10404 defining section names for the early LTO debug variants.
10405 (reset_indirect_string): New helper.
10406 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
10407 (print_dw_val): Add support for offsetted symbol references.
10408 (get_ultimate_context): Split out from is_cxx.
10409 (is_cxx): Use get_ultimate_context.
10410 (is_fortran): Add decl overload.
10411 (compute_comp_unit_symbol): Split out worker from
10412 compute_section_prefix.
10413 (compute_section_prefix): Call compute_comp_unit_symbol and
10414 set comdat_type_p here.
10415 (output_die): Skip DIE symbol output for the LTO added one.
10416 Handle DIE symbol references with offset.
10417 (output_comp_unit): Guard section name mangling properly.
10418 For LTO debug sections emit a symbol at the section beginning
10419 which we use to refer to its DIEs.
10420 (add_abstract_origin_attribute): For DIEs registered via
10421 dwarf2out_register_external_die directly refer to the early
10422 DIE rather than indirectly through the shadow one we created.
10423 Remove obsolete call to dwarf2out_abstract_function for
10424 non-function/block origins.
10425 (gen_array_type_die): When generating early LTO debug do
10426 not emit DW_AT_string_length.
10427 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
10428 late when in LTO. As suggested place a gcc_unreachable for
10429 the DECL_ABSTRACT_P case.
10430 (gen_subprogram_die): Avoid another specification DIE
10431 for early built declarations/definitions for the late LTO case.
10432 (gen_variable_die): Add type references for late duplicated VLA dies
10434 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
10435 we have the abstract instance already.
10436 (process_scope_var): Adjust decl DIE contexts in LTO which
10437 first puts them in limbo.
10438 (gen_decl_die): Do not generate type DIEs late apart from
10439 types for VLAs or for decls we do not yet have a DIE. Do not
10440 call dwarf2out_abstract_function late.
10441 (dwarf2out_early_global_decl): Make sure to create DIEs
10442 for abstract instances of a decl first.
10443 (dwarf2out_late_global_decl): Adjust comment.
10444 (output_macinfo_op): With multiple macro sections use
10445 macinfo_label_base to distinguish labels.
10446 (output_macinfo): Likewise. Update macinfo_label_base.
10447 Pass in the line info label.
10448 (note_variable_value_in_expr): When generating LTO resolve
10449 all variable values here by generating DIEs as needed.
10450 (init_sections_and_labels): Add early LTO debug flag parameter
10451 and generate different sections and names if set. Add generation
10452 counter for the labels so we can have multiple of them.
10453 (reset_dies): Helper to allow DIEs to be output multiple times.
10454 (dwarf2out_finish): When outputting DIEs to the fat part of an
10455 LTO object first reset DIEs.
10456 (dwarf2out_early_finish): Output early DIEs when generating LTO.
10457 (modified_type_die): Check for decl_ultimate_origin being self
10459 (gen_type_die_with_usage): Likewise.
10460 (gen_typedef_die): Allow decl_ultimate_origin being self.
10461 (set_decl_abstract_flags): Remove.
10462 (set_block_abstract_flags): Likewise.
10463 (dwarf2out_abstract_function): Treat the early generated DIEs
10464 as the abstract copy and only add DW_AT_inline and
10465 DW_AT_artificial here and call set_decl_origin_self.
10466 If the DIE has an abstract origin don't do anything.
10467 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
10468 if we have none yet (Go fails to build one, PR78628).
10469 (variably_modified_type_p): Prevent endless recursion for Ada
10470 cyclic pointer types.
10471 * lto-streamer-in.c: Include debug.h.
10472 (dref_queue): New global.
10473 (lto_read_tree_1): Stream in DIE references.
10474 (lto_input_tree): Register DIE references.
10475 (input_function): Stream DECL_DEBUG_ARGS.
10476 * lto-streamer-out.c: Include debug.h.
10477 (lto_write_tree_1): Output DIE references.
10478 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
10479 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
10480 (output_function): Stream DECL_DEBUG_ARGS.
10481 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
10482 Stream DECL_ABSTRACT_ORIGIN.
10483 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
10484 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
10485 DECL_CONTEXT for file-scope decls.
10486 * lto-streamer.h (struct dref_entry): Declare.
10487 (dref_queue): Likewise.
10488 * cfgexpand.c (pass_expand::execute): Do not call the
10489 outlining_inline_function hook here.
10490 * lto-wrapper.c (debug_obj): New global.
10491 (tool_cleanup): Unlink it if required.
10492 (debug_objcopy): New function.
10493 (run_gcc): Handle early debug sections in the IL files by
10494 extracting them to separate files, partially linkin them and
10495 feeding the result back as result to the linker.
10496 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
10497 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
10498 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
10499 sections into a separate segment.
10500 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
10502 (darwin_asm_dwarf_section): Likewise.
10503 (darwin_asm_output_dwarf_offset): Likewise.
10504 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
10506 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10507 Alan Hayward <alan.hayward@arm.com>
10508 David Sherwood <david.sherwood@arm.com>
10510 * read-md.h (md_reader::record_potential_iterator_use): Replace
10511 pointer argument with an rtx and an index.
10512 * read-rtl.c (iterator_group::apply_iterator): Likewise.
10513 (apply_mode_iterator): Likewise.
10514 (apply_code_iterator): Likewise.
10515 (apply_int_iterator): Likewise.
10516 (apply_subst_iterator): Likewise.
10517 (record_iterator_use): Likewise.
10518 (record_attribute_use): Likewise.
10519 (md_reader::record_potential_iterator_use): Likewise. Update calls
10520 to record_iterator_use and apply_iterator.
10521 (iterator_use): Replace ptr with x and index.
10522 (attribute_use): Likewise.
10523 (apply_attribute_uses): Update calls to apply_iterator.
10524 (apply_iterators): Likewise. Update initialization of iterator_use.
10525 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
10526 and record_potential_iterator_use.
10527 (rtx_reader::read_rtx_operand): Likewise.
10529 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
10530 Alan Hayward <alan.hayward@arm.com>
10531 David Sherwood <david.sherwood@arm.com>
10533 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
10536 2017-08-21 Richard Biener <rguenther@suse.de>
10538 PR middle-end/81884
10539 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
10540 at struct end conservatively when comparing common bases.
10542 2017-08-21 Richard Biener <rguenther@suse.de>
10544 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
10545 (mem_ref_in_stmt): Remove.
10546 (determine_max_movement): Use ref index to get at the reference.
10547 (invariantness_dom_walker::before_dom_children): Deal with
10548 lim data already initialized.
10549 (gather_mem_refs_stmt): Initialize lim data and record ref index.
10551 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
10553 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
10554 (TARGET_ISA_ROUND): Ditto.
10555 (TARGET_ROUND): Ditto.
10556 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
10557 * config/i386/i386.md: Ditto.
10558 * config/i386/sse.md: Ditto.
10559 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
10560 with OPTION_MASK_ISA_SSE4_1.
10562 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
10565 * doc/extend.texi (x86 Built-in Functions): Correct the name of
10566 __builtin_ia32_lzcnt_u16.
10568 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
10571 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
10572 (rs6000_set_current_function): Rewrite function to use it.
10574 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
10577 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
10578 and TYPE_WARN_IF_NOT_ALIGN.
10579 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
10580 (handle_warn_if_not_align): New.
10581 (place_union_field): Call handle_warn_if_not_align.
10582 (place_field): Call handle_warn_if_not_align.
10583 Copy TYPE_WARN_IF_NOT_ALIGN.
10584 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
10585 (layout_type): Likewise.
10586 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
10588 (tree_decl_common): Add warn_if_not_align.
10589 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
10590 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
10591 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
10592 (DECL_WARN_IF_NOT_ALIGN): Likewise.
10593 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
10594 * doc/extend.texi: Document warn_if_not_aligned attribute.
10595 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
10597 2017-08-17 Martin Liska <mliska@suse.cz>
10600 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
10601 (get_data_dependence): Use it as pointer type.
10602 (distribute_loop): Likewise.
10604 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10606 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
10607 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
10608 (p8_vmrgow_<mode>_direct): New define_insn.
10609 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
10610 handle endianness for vmrgew and vmrgow permute patterns.
10612 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
10614 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
10615 * config/rs6000/rs6000-cpus.def: Remove comment.
10616 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
10617 (POWERPC_MASKS): Likewise.
10618 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
10619 use of TARGET_VSX_TIMODE.
10620 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
10621 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
10622 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
10623 (rs6000_option_override_internal): Remove dead code.
10624 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
10625 (rs6000_legitimize_reload_address): Likewise.
10626 (rs6000_legitimate_address_p): Likewise.
10627 (rs6000_opt_masks): Delete "vsx-timode".
10628 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
10629 from function comment.
10630 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
10631 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
10632 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
10634 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
10635 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
10636 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
10637 (VSX_TI): Likewise.
10639 (define_peephole2): Likewise.
10641 2017-08-17 Martin Sebor <msebor@redhat.com>
10644 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
10645 past the end of an array.
10646 (test_pp_format): Add test cases.
10648 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
10650 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
10651 missing ECF_NOTHROW flags.
10653 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
10656 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
10657 operands residing in integer registers.
10658 (*vsx_le_perm_load_<mode>): Likewise.
10659 (*vsx_le_perm_store_<mode>): Likewise.
10660 (define_peephole2): Add peepholes to optimize the above.
10662 2017-08-17 Marek Polacek <polacek@redhat.com>
10664 PR middle-end/81814
10665 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
10666 to mimic what shorten_compare did. Change the return type to bool.
10667 (fold_cond_expr_with_comparison): Update call to
10668 operand_equal_for_comparison_p.
10669 (fold_ternary_loc): Likewise.
10671 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
10673 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
10675 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
10677 2017-08-17 Richard Biener <rguenther@suse.de>
10679 * tree-ssa-structalias.c (solve_graph): When propagating
10680 to successors update the graphs succ edges and avoid duplicate work.
10682 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
10685 * config/i386/i386.c (ix86_option_override_internal): Save target
10686 specific options after ix86_stack_protector_guard_reg was changed.
10688 2017-08-17 Richard Biener <rguenther@suse.de>
10690 PR tree-optimization/81827
10691 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
10692 (new_var_info): Initialize it conservatively.
10693 (get_call_vi): Mark register vars.
10694 (new_scalar_tmp_constraint_exp): Likewise.
10695 (handle_rhs_call): Likewise.
10696 (handle_const_call): Likewise.
10697 (create_function_info_for): Likewise.
10698 (solve_constraints): Sort varinfos to separate register from
10699 non-register vars to pack points-to solution bitmaps during
10702 2017-08-17 Marek Polacek <polacek@redhat.com>
10704 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
10706 2017-08-17 Richard Biener <rguenther@suse.de>
10708 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
10709 to true when overflow is undefined and we saturated the result.
10711 2017-08-17 Alan Modra <amodra@gmail.com>
10714 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
10715 Don't use store multiple if only one reg needs saving.
10716 (interesting_frame_related_regno): New function.
10717 (rs6000_frame_related): Don't emit frame info for regs that
10719 (rs6000_emit_epilogue): Likewise.
10721 2017-08-16 Nathan Sidwell <nathan@acm.org>
10723 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
10724 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
10725 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
10726 * tree.c (free_lang_data_in_type): Use else-if chain. Always
10727 clear TYPE_LANG_1. Remove obsolete member-function stripping.
10728 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
10729 (verify_type): Adjust for TYPE_BINFO move.
10730 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
10731 process TYPE_BINFO directly.
10732 (hash_tree): Likewise.
10733 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
10735 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
10738 2017-08-16 David Malcolm <dmalcolm@redhat.com>
10740 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
10742 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
10745 * config/i386/i386.md (*anddi_1_btr): Change predicates of
10746 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
10747 Add ix86_binary_operator_ok to insn constraint.
10748 (*iordi_1_bts): Ditto.
10749 (*xordi_1_btc): Ditto.
10750 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
10751 Update corresponding peephole2 pattern.
10755 2017-08-16 Bin Cheng <bin.cheng@arm.com>
10757 PR tree-optimization/81832
10758 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
10759 copy loop header which has IFN_LOOP_DIST_ALIAS call.
10761 2017-08-16 Marek Polacek <polacek@redhat.com>
10764 * fold-const.c (fold_indirect_ref_1): Restore original behavior
10765 regarding size_zero_node.
10767 2017-08-16 Martin Liska <mliska@suse.cz>
10770 * config.gcc: Respect previously set extra_objs in case
10773 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
10775 PR tree-optimization/81835
10776 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
10777 the phi SSA_NAME. Check that the condition in a COND_EXPR does
10778 not depend on the phi.
10780 2017-08-16 Alan Modra <amodra@gmail.com>
10782 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
10785 2017-08-16 Alan Modra <amodra@gmail.com>
10787 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
10788 (save_reg_p): ..into this. Update all callers.
10789 (first_reg_to_save): Simplify.
10791 2017-08-16 Alan Modra <amodra@gmail.com>
10793 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
10796 2017-08-15 Joseph Myers <joseph@codesourcery.com>
10800 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
10801 constant count if that count is less than 32.
10803 2017-08-15 Nathan Sidwell <nathan@acm.org>
10805 * gcc.c (execute): Emit friendlier message if inferior is killed
10806 by an external cause.
10808 2017-08-15 Richard Biener <rguenther@suse.de>
10810 PR tree-optimization/81790
10811 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
10812 CONSTRUCTORs from simplifying and VN.
10814 2017-08-14 Martin Sebor <msebor@redhat.com>
10816 * builtin-attrs.def: Add comments.
10818 2017-08-14 Martin Sebor <msebor@redhat.com>
10821 * doc/extend.texi (attribute nonstring): Document new attribute.
10823 2017-08-14 Martin Sebor <msebor@redhat.com>
10826 * tree-diagnostic.c (default_tree_printer): Handle %G.
10827 * gimple-pretty-print.h (percent_G_format): Declare new function.
10828 * gimple-pretty-print.c (percent_G_format): Define.
10829 * tree-pretty-print.c (percent_K_format): Add argument.
10831 2017-08-14 Martin Sebor <msebor@redhat.com>
10833 PR translation/79998
10834 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
10835 Remove a stray space.
10837 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
10840 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
10841 (*iordi_1_bts): Ditto.
10842 (*xordi_1_btc): Ditto.
10844 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10847 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
10848 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10850 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
10851 quoted strings, and make more translator-friendly.
10852 (darwin_rs6000_override_options): Likewise.
10853 (rs6000_option_override_internal): Likewise.
10854 (rs6000_return_in_memory): Fix overlong line.
10855 (init_cmulative_args): Use quoted strings, and make more
10856 translator-friendly.
10857 (rs6000_pass_by_reference): Fix overlong line.
10858 (def_builtin): Use quoted strings.
10859 (altivec_expand_predicate_builtin): Use quoted strings, and make
10860 more translator-friendly.
10861 (htm_expand_builtin): Use quoted strings.
10862 (cpu_expand_builtin): Use quoted strings, and make more
10863 translator-friendly.
10864 (altivec_expand_builtin): Likewise.
10865 (paired_expand_predicate_builtin): Likewise.
10866 (rs6000_invalid_builtin): Likewise.
10867 (builtin_function_type): Use quoted strings.
10868 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
10869 more translator-friendly.
10870 (rs6000_trampoline_init): Likewise.
10871 (rs6000_handle_altivec_attribute): Likewise.
10872 (rs6000_inner_target_options): Use quoted strings.
10873 (rs6000_disable_incompatible_switches): Likewise.
10874 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
10875 strings, and make more translator-friendly.
10876 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
10878 2017-08-14 Bin Cheng <bin.cheng@arm.com>
10880 PR tree-optimization/81799
10881 * tree-loop-distribution.c (version_loop_by_alias_check): Force
10882 cond_expr to simple gimple operand.
10884 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
10886 PR middle-end/46932
10887 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
10889 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
10893 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
10894 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
10895 TARGET_GASISR_PROLOGUES.
10896 * config/avr/avr.c (avr_option_override): Same.
10897 (avr_pass_pre_proep::execute): Same.
10899 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
10902 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
10903 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
10905 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
10907 * config/i386/i386.md (*load_tp_<mode>): Redefine as
10908 define_insn_and_split. Split to a memory load from 0 in
10909 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
10910 using PTR mode iterator.
10911 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
10912 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
10913 (*add_tp_<mode>): Redefine as define_insn_and_split.
10914 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
10915 address space. Merge with *add_tp_x32 using PTR mode iterator.
10916 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
10917 Split to an add with a memory load from 0 in
10918 DEFAULT_TLS_SEG_REG address space.
10920 2017-08-12 Andrew Pinski <apinski@cavium.com>
10922 * config/aarch64/aarch64-option-extensions.def (rdma):
10923 Fix feature string to what Linux prints out in /proc/cpuinfo.
10925 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
10928 * dwarf2out.c (modified_type_die): For C typedef types that have
10929 an ultimate origin, process the ultimate origin instead of the
10931 (gen_typedef_die): Assert that input DECLs have no ultimate
10933 (gen_type_die_with_usage): For typedef variants that have an
10934 ultimate origin, just call gen_decl_die on the original DECL.
10935 (process_scope_var): Avoid creating DIEs for local typedefs and
10936 concrete static variables.
10938 2017-08-12 Alan Modra <amodra@gmail.com>
10942 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
10943 match gnu-user.h startfile.
10944 (ENDFILE_LINUX_SPEC): Similarly.
10946 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
10949 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
10951 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
10953 2017-08-11 Tamar Christina <tamar.christina@arm.com>
10954 * config/aarch64/aarch64.md (mov<mode>): Change.
10955 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
10956 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
10957 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
10959 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
10961 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
10962 for storage order barriers.
10964 2017-08-11 Martin Liska <mliska@suse.cz>
10967 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
10968 variables of void type.
10970 2017-08-11 Martin Liska <mliska@suse.cz>
10972 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
10973 TARGET_SUPPORTS_ALIASES.
10974 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
10975 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
10976 (optimize_weakref): Likewise.
10977 * symtab.c (symtab_node::noninterposable_alias): Likewise.
10978 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
10979 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
10981 2017-08-11 Martin Liska <mliska@suse.cz>
10984 * config/i386/i386.c (make_resolver_func): Do complete
10985 refactoring of the function.
10987 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
10990 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
10991 * config/i386/i386.c (ix86_stack_protect_guard): Use
10992 ix86_stack_protect_guard_symbol_str to generate varible declaration.
10993 * doc/invoke.texi (x86 Options): Document
10994 -mstack-protector-guard-symbol= option.
10996 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
10998 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
10999 * config/i386/i386.c (ix86_split_stack_guard): New function.
11000 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
11001 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
11002 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
11003 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
11004 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
11005 (split_stack_space_check): Call ix86_split_stack_guard.
11007 2017-08-10 Martin Sebor <msebor@redhat.com>
11009 * print-tree.c (print_node): Print location using the established
11011 Replace spaces with colons.
11012 (debug_raw, debug): Ditto.
11014 2017-08-10 Martin Sebor <msebor@redhat.com>
11017 * pretty-print.c (pp_format): Correct the handling of %s precision.
11019 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
11022 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
11024 (ix86_finalize_stack_frame_flags): This. Also clear
11025 frame_pointer_needed if -fno-omit-frame-pointer is used without
11027 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
11028 with ix86_finalize_stack_frame_flags.
11029 (ix86_expand_epilogue): Likewise.
11030 (ix86_expand_split_stack_prologue): Likewise.
11031 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
11033 2017-08-10 Martin Liska <mliska@suse.cz>
11036 * c-attribs.c (handle_target_attribute):
11037 Report warning for an empty string argument of target attribute.
11039 2017-08-09 Jakub Jelinek <jakub@redhat.com>
11042 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
11044 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
11045 or DECL_NONLOCAL labels.
11046 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
11047 or DECL_NONLOCAL labels here.
11049 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
11051 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
11052 to indicate when early gimple folding has been disabled.
11053 (rs6000_gimple_fold_builtin): Add debug content.
11054 (rs6000_invalid_builtin): Fix whitespace.
11055 (rs6000_expand_builtin): Fix whitespace.
11056 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
11058 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
11061 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
11062 SAVE_MULTIPLE if not all the registers that saves, should be saved.
11064 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
11066 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
11067 (qdf24xx): Likewise.
11068 * config/aarch64/aarch64.md: Include falkor.md.
11069 * config/aarch64/falkor.md: New.
11071 2017-08-09 Marek Polacek <polacek@redhat.com>
11074 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
11075 * diagnostic.c (emit_diagnostic): Add a comment.
11076 (emit_diagnostic_valist): New function.
11078 2017-08-09 Marek Polacek <polacek@redhat.com>
11081 * input.c (make_location): New overload.
11082 * input.h (make_location): Declare.
11084 2017-08-08 Alan Modra <amodra@gmail.com>
11085 H.J. Lu <hongjiu.lu@intel.com>
11088 * gcc.c (NO_PIE_SPEC): Delete.
11089 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
11091 (LINK_PIE_SPEC): ..to here.
11092 (LINK_COMMAND_SPEC): Support -no-pie.
11093 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
11094 chain of crtbegin*.o selection, update for PIE_SPEC changes and
11096 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
11097 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
11098 (ENDFILE_CRTEND_SPEC): Similarly.
11100 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
11103 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
11104 (mstack-protector-guard-offset=): Ditto.
11105 * config/i386/i386.c (ix86_option_override): Handle
11106 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
11108 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
11109 ix86_stack_protect_guard_offset variables.
11110 (TARGET_STACK_PROTECT_GUARD): Always define.
11111 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
11112 and -mstack-protector-guard-offset= options.
11114 2017-08-08 Bin Cheng <bin.cheng@arm.com>
11116 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
11117 boundary case for the last candidate.
11119 2017-08-08 Bin Cheng <bin.cheng@arm.com>
11121 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
11122 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
11124 2017-08-08 Tamar Christina <tamar.christina@arm.com>
11126 PR middle-end/19706
11127 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
11128 * config/aarch64/aarch64-builtins.c
11129 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
11130 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
11131 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
11133 2017-08-08 Tamar Christina <tamar.christina@arm.com>
11134 Andrew Pinski <pinskia@gmail.com>
11136 PR middle-end/19706
11137 * internal-fn.def (XORSIGN): New.
11138 * optabs.def (xorsign_optab): New.
11139 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
11140 (convert_expand_mult_copysign): New.
11141 (pass_optimize_widening_mul::execute): Call
11142 convert_expand_mult_copysign.
11144 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11146 PR tree-optimization/81354
11147 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
11148 Insert on edges rather than explicitly creating landing pads.
11149 (analyze_candidates_and_replace): Commit edge inserts.
11151 2017-08-08 Richard Biener <rguenther@suse.de>
11153 PR middle-end/81719
11154 * tree-ssa-loop-niter.c: Include tree-dfa.h.
11155 (expand_simple_operations): Also look through ADDR_EXPRs with
11156 MEM_REF bases treating them as POINTER_PLUS_EXPR.
11158 2017-08-08 Richard Biener <rguenther@suse.de>
11160 PR tree-optimization/81723
11161 * tree-vect-slp.c (struct bst_traits): New hash traits.
11162 (bst_fail): New global.
11163 (vect_build_slp_tree_2): New worker, split out from ...
11164 (vect_build_slp_tree): ... this now wrapping it with using
11165 bst_fail set to cache SLP tree build fails. Properly handle
11167 (vect_analyze_slp_instance): Allocate and free bst_fail.
11169 2017-08-08 Martin Liska <mliska@suse.cz>
11172 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
11173 LABEL_DECLs that can be from a different function.
11175 2017-08-08 Bin Cheng <bin.cheng@arm.com>
11177 PR tree-optimization/81744
11178 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
11179 loop's number of iterations.
11181 2017-08-08 Martin Liska <mliska@suse.cz>
11183 * asan.c: Include header files.
11184 * attribs.c (build_decl_attribute_variant): New function moved
11186 (build_type_attribute_qual_variant): Likewise.
11187 (cmp_attrib_identifiers): Likewise.
11188 (simple_cst_list_equal): Likewise.
11189 (omp_declare_simd_clauses_equal): Likewise.
11190 (attribute_value_equal): Likewise.
11191 (comp_type_attributes): Likewise.
11192 (build_type_attribute_variant): Likewise.
11193 (lookup_ident_attribute): Likewise.
11194 (remove_attribute): Likewise.
11195 (merge_attributes): Likewise.
11196 (merge_type_attributes): Likewise.
11197 (merge_decl_attributes): Likewise.
11198 (merge_dllimport_decl_attributes): Likewise.
11199 (handle_dll_attribute): Likewise.
11200 (attribute_list_equal): Likewise.
11201 (attribute_list_contained): Likewise.
11202 * attribs.h (lookup_attribute): New function moved from tree.[ch].
11203 (lookup_attribute_by_prefix): Likewise.
11204 * bb-reorder.c: Include header files.
11205 * builtins.c: Likewise.
11206 * calls.c: Likewise.
11207 * cfgexpand.c: Likewise.
11208 * cgraph.c: Likewise.
11209 * cgraphunit.c: Likewise.
11210 * convert.c: Likewise.
11211 * dwarf2out.c: Likewise.
11212 * final.c: Likewise.
11213 * fold-const.c: Likewise.
11214 * function.c: Likewise.
11215 * gimple-expr.c: Likewise.
11216 * gimple-fold.c: Likewise.
11217 * gimple-pretty-print.c: Likewise.
11218 * gimple.c: Likewise.
11219 * gimplify.c: Likewise.
11220 * hsa-common.c: Likewise.
11221 * hsa-gen.c: Likewise.
11222 * internal-fn.c: Likewise.
11223 * ipa-chkp.c: Likewise.
11224 * ipa-cp.c: Likewise.
11225 * ipa-devirt.c: Likewise.
11226 * ipa-fnsummary.c: Likewise.
11227 * ipa-inline.c: Likewise.
11228 * ipa-visibility.c: Likewise.
11230 * lto-cgraph.c: Likewise.
11231 * omp-expand.c: Likewise.
11232 * omp-general.c: Likewise.
11233 * omp-low.c: Likewise.
11234 * omp-offload.c: Likewise.
11235 * omp-simd-clone.c: Likewise.
11236 * opts-global.c: Likewise.
11237 * passes.c: Likewise.
11238 * predict.c: Likewise.
11239 * sancov.c: Likewise.
11240 * sanopt.c: Likewise.
11241 * symtab.c: Likewise.
11242 * toplev.c: Likewise.
11243 * trans-mem.c: Likewise.
11244 * tree-chkp.c: Likewise.
11245 * tree-eh.c: Likewise.
11246 * tree-into-ssa.c: Likewise.
11247 * tree-object-size.c: Likewise.
11248 * tree-parloops.c: Likewise.
11249 * tree-profile.c: Likewise.
11250 * tree-ssa-ccp.c: Likewise.
11251 * tree-ssa-live.c: Likewise.
11252 * tree-ssa-loop.c: Likewise.
11253 * tree-ssa-sccvn.c: Likewise.
11254 * tree-ssa-structalias.c: Likewise.
11255 * tree-ssa.c: Likewise.
11256 * tree-streamer-in.c: Likewise.
11257 * tree-vectorizer.c: Likewise.
11258 * tree-vrp.c: Likewise.
11259 * tsan.c: Likewise.
11260 * ubsan.c: Likewise.
11261 * varasm.c: Likewise.
11262 * varpool.c: Likewise.
11263 * tree.c: Remove functions moved to attribs.[ch].
11264 * tree.h: Likewise.
11265 * config/aarch64/aarch64.c: Add attrs.h header file.
11266 * config/alpha/alpha.c: Likewise.
11267 * config/arc/arc.c: Likewise.
11268 * config/arm/arm.c: Likewise.
11269 * config/avr/avr.c: Likewise.
11270 * config/bfin/bfin.c: Likewise.
11271 * config/c6x/c6x.c: Likewise.
11272 * config/cr16/cr16.c: Likewise.
11273 * config/cris/cris.c: Likewise.
11274 * config/darwin.c: Likewise.
11275 * config/epiphany/epiphany.c: Likewise.
11276 * config/fr30/fr30.c: Likewise.
11277 * config/frv/frv.c: Likewise.
11278 * config/ft32/ft32.c: Likewise.
11279 * config/h8300/h8300.c: Likewise.
11280 * config/i386/winnt.c: Likewise.
11281 * config/ia64/ia64.c: Likewise.
11282 * config/iq2000/iq2000.c: Likewise.
11283 * config/lm32/lm32.c: Likewise.
11284 * config/m32c/m32c.c: Likewise.
11285 * config/m32r/m32r.c: Likewise.
11286 * config/m68k/m68k.c: Likewise.
11287 * config/mcore/mcore.c: Likewise.
11288 * config/microblaze/microblaze.c: Likewise.
11289 * config/mips/mips.c: Likewise.
11290 * config/mmix/mmix.c: Likewise.
11291 * config/mn10300/mn10300.c: Likewise.
11292 * config/moxie/moxie.c: Likewise.
11293 * config/msp430/msp430.c: Likewise.
11294 * config/nds32/nds32-isr.c: Likewise.
11295 * config/nds32/nds32.c: Likewise.
11296 * config/nios2/nios2.c: Likewise.
11297 * config/nvptx/nvptx.c: Likewise.
11298 * config/pa/pa.c: Likewise.
11299 * config/pdp11/pdp11.c: Likewise.
11300 * config/powerpcspe/powerpcspe.c: Likewise.
11301 * config/riscv/riscv.c: Likewise.
11302 * config/rl78/rl78.c: Likewise.
11303 * config/rx/rx.c: Likewise.
11304 * config/s390/s390.c: Likewise.
11305 * config/sh/sh.c: Likewise.
11306 * config/sol2.c: Likewise.
11307 * config/sparc/sparc.c: Likewise.
11308 * config/spu/spu.c: Likewise.
11309 * config/stormy16/stormy16.c: Likewise.
11310 * config/tilegx/tilegx.c: Likewise.
11311 * config/tilepro/tilepro.c: Likewise.
11312 * config/v850/v850.c: Likewise.
11313 * config/vax/vax.c: Likewise.
11314 * config/visium/visium.c: Likewise.
11315 * config/xtensa/xtensa.c: Likewise.
11317 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
11320 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
11321 constraints since the -mupper-regs-* switches have been
11323 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
11324 into a vector from a double word element that was extracted from
11325 another vector, and eliminate extra XXPERMDI instructions.
11326 (vsx_concat_<mode>_2): Likewise.
11327 (vsx_concat_<mode>_3): Likewise.
11328 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
11329 concat to allow optimizing inserts from previous extracts.
11331 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
11333 * config/i386/i386.c (ix86_stack_protect_guard): Generate
11334 memory reference to a SSP offset in TLS address space.
11335 (ix86_print_operand) <case '@'>: Remove.
11336 (ix86_print_operand_punct_valid_p): Remove '@' code.
11337 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
11338 UNSPEC_SP_TLS_TEST.
11339 (stack_tls_protect_set_<mode>): Remove.
11340 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
11341 (stack_tls_protect_test_<mode>): Remove.
11342 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
11344 2017-08-07 Olivier Hainque <hainque@adacore.com>
11347 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
11349 2017-08-07 Douglas Rupp <rupp@adacore.com>
11351 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
11352 variable was referenced as multidir in command.
11354 2017-08-07 Jakub Jelinek <jakub@redhat.com>
11357 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
11360 2017-08-07 Martin Liska <mliska@suse.cz>
11362 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
11363 * config/rl78/rl78.c: Add include of attribs.h.
11364 * config/sh/sh.c: Likewise.
11365 * config/v850/v850.c: Likewise.
11367 2017-08-07 Tom de Vries <tom@codesourcery.com>
11369 PR middle-end/78266
11370 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
11372 2017-08-07 Martin Liska <mliska@suse.cz>
11374 * config/mips/mips.c: Include attribs.h.
11376 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
11379 * doc/invoke.texi: Document change in behvaior for -Ofast for
11382 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
11384 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
11386 (aarch64_pop_regs): Likewise.
11387 (aarch64_gen_load_pair): Likewise.
11388 (aarch64_save_callee_saves): Likewise.
11389 (aarch64_restore_callee_saves): Likewise.
11391 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
11393 * config/i386/i386.c: Revert the last change.
11395 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
11398 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
11400 (ix86_finalize_stack_frame_flags): This. Also clear
11401 frame_pointer_needed if -fno-omit-frame-pointer is used without
11403 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
11404 with ix86_finalize_stack_frame_flags.
11405 (ix86_expand_epilogue): Likewise.
11406 (ix86_expand_split_stack_prologue): Likewise.
11408 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
11411 * config/i386/i386.c (get_builtin_code_for_version): Set priority
11412 to P_AES for Westmere.
11414 2017-08-07 Jonathan Yong <10walls@gmail.com>
11416 * config/i386/mingw.opt (fset-stack-executable): Removed.
11417 * config/i386/cygming.opt (fset-stack-executable): Moved
11419 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
11421 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
11423 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
11425 2017-08-07 Marek Polacek <polacek@redhat.com>
11427 PR middle-end/81737
11428 * fold-const.c (fold_indirect_ref_1): Check type_domain.
11430 2017-08-07 Martin Liska <mliska@suse.cz>
11432 * attribs.h (canonicalize_attr_name): New function.
11433 (cmp_attribs): Move from c-format.c and adjusted.
11434 (is_attribute_p): Moved from tree.h.
11435 * tree-inline.c: Add new includes.
11436 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
11437 (private_is_attribute_p): Remove.
11438 (private_lookup_attribute): Likewise.
11439 (private_lookup_attribute_by_prefix): Simplify.
11440 (remove_attribute): Use is_attribute_p.
11441 * tree.h: Remove removed declarations.
11443 2017-08-07 Jakub Jelinek <jakub@redhat.com>
11445 PR middle-end/81698
11446 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
11447 instead of computing it in the function. Formatting fix.
11448 (expand_case): Don't rely on default_edge being the first edge,
11449 clear it if removing it, pass default_edge to
11450 emit_case_dispatch_table.
11451 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
11454 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
11456 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
11457 insn in the function, emit NOP after the insn.
11459 2017-08-06 Tom de Vries <tom@codesourcery.com>
11461 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
11464 2017-08-06 Tom de Vries <tom@codesourcery.com>
11466 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
11469 2017-08-04 Yury Gribov <tetra2005@gmail.com>
11471 PR tree-optimization/57371
11472 * match.pd: New pattern.
11474 2017-08-04 Marek Polacek <polacek@redhat.com>
11476 PR middle-end/81695
11477 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
11478 perform the computation in offset_int.
11480 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
11482 PR tree-optimization/81136
11483 * tree-vectorizer.h: Include tree-hash-traits.h.
11484 (vec_base_alignments): New typedef.
11485 (vec_info): Add a base_alignments field.
11486 (vect_record_base_alignments): Declare.
11487 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
11489 (DR_IS_CONDITIONAL_IN_STMT): New macro.
11490 (create_data_ref): Add an is_conditional_in_stmt argument.
11491 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
11492 the is_conditional_in_stmt field.
11493 (data_ref_loc): Add an is_conditional_in_stmt field.
11494 (get_references_in_stmt): Set the is_conditional_in_stmt field.
11495 (find_data_references_in_stmt): Update call to create_data_ref.
11496 (graphite_find_data_references_in_stmt): Likewise.
11497 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
11498 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
11499 (vect_record_base_alignment): New function.
11500 (vect_record_base_alignments): Likewise.
11501 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
11502 for nested statements even if we fail to compute a misalignment.
11503 Use pooled base alignments for unconditional references.
11504 (vect_find_same_alignment_drs): Compare base addresses instead
11506 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
11507 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
11509 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
11511 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
11512 Add an explicit name for the enum. Use auto_vec for slp_instances
11513 and grouped_stores.
11514 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
11516 (_bb_vec_info): Add a constructor and destructor.
11517 (vinfo_for_stmt): Return NULL for uids of -1 as well.
11518 (destroy_loop_vec_info): Delete.
11519 (vect_destroy_datarefs): Likewise.
11520 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
11521 (vec_info::vec_info): New function.
11522 (vec_info::~vec_info): Likewise.
11523 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
11524 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
11525 destroy_loop_vec_info.
11526 * tree-vect-loop.c (new_loop_vec_info): Replace with...
11527 (_loop_vec_info::_loop_vec_info): ...this.
11528 (destroy_loop_vec_info): Replace with...
11529 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
11530 the stmt_vec_infos. Leave handling of vec_info information to its
11531 destructor. Remove explicit vector releases.
11532 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
11533 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
11534 * tree-vect-slp.c (new_bb_vec_info): Replace with...
11535 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
11536 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
11537 (destroy_bb_vec_info): Replace with...
11538 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
11539 information to its destructor.
11540 (vect_slp_analyze_bb_1): Use new and delete instead of
11541 new_bb_vec_info and destroy_bb_vec_info.
11542 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
11545 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
11547 * tree-data-ref.h (subscript): Add access_fn field.
11548 (data_dependence_relation): Add could_be_independent_p.
11549 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
11550 (same_access_functions): Move to tree-data-ref.c.
11551 * tree-data-ref.c (ref_contains_union_access_p): New function.
11552 (access_fn_component_p): Likewise.
11553 (access_fn_components_comparable_p): Likewise.
11554 (dr_analyze_indices): Add a reference to access_fn_component_p.
11555 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
11557 (constant_access_functions): Likewise.
11558 (add_other_self_distances): Likewise.
11559 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
11560 (initialize_data_dependence_relation): Use XCNEW and remove
11561 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
11562 of access functions that have the same type. Allow the
11563 subsequence to end with different bases in some circumstances.
11564 Record the chosen access functions in SUB_ACCESS_FN.
11565 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
11566 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
11567 (subscript_dependence_tester_1): Likewise dra and drb.
11568 (build_classic_dist_vector): Update calls accordingly.
11569 (subscript_dependence_tester): Likewise.
11570 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
11571 DDR_COULD_BE_INDEPENDENT_P.
11572 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
11573 comp_alias_ddrs instead of may_alias_ddrs.
11574 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
11576 (vect_analyze_data_ref_dependence): Use it if
11577 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
11578 distance vectors if that fails.
11579 (dependence_distance_ge_vf): New function.
11580 (vect_prune_runtime_alias_test_list): Use it. Don't clear
11581 LOOP_VINFO_MAY_ALIAS_DDRS.
11583 2017-08-04 Richard Biener <rguenther@suse.de>
11585 PR middle-end/81705
11586 * fold-const.c (fold_binary_loc): Properly restrict
11587 minus_var0 && minus_var1 case when associating undefined overflow
11590 2017-08-04 Tom de Vries <tom@codesourcery.com>
11592 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
11594 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11596 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11597 Don't start diagnostic messages with a capital letter.
11598 * config/rs6000/rs6000.c (rs6000_option_override_internal):
11600 (rs6000_invalid_builtin): Likewise.
11601 (rs6000_trampoline_init): Likewise.
11603 2017-08-03 Jakub Jelinek <jakub@redhat.com>
11606 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
11607 after setting changeable df flags.
11609 2017-08-03 Richard Biener <rguenther@suse.de>
11611 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
11612 up if the use is in USE - X.
11614 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
11616 * toplev.c (dumpfile.h): New include.
11617 (internal_error_reentered): New static function. Use it...
11618 (internal_error_function): ...here to handle reentered internal_error.
11620 2017-08-03 Richard Biener <rguenther@suse.de>
11622 PR middle-end/81148
11623 * fold-const.c (split_tree): Add minus_var and minus_con
11624 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
11625 here but always use minus_*.
11626 (associate_trees): Assert we never associate with MINUS_EXPR
11627 and NULL first operand. Do not recurse for PLUS_EXPR operands
11628 when associating as MINUS_EXPR either.
11629 (fold_binary_loc): Track minus_var and minus_con.
11631 2017-08-03 Tom de Vries <tom@codesourcery.com>
11634 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
11635 ACCEL_COMPILER, apply finish_options on
11636 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
11638 2017-08-03 Tom de Vries <tom@codesourcery.com>
11641 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
11642 function_entry_patch_area_size > 0.
11644 2017-08-03 Jakub Jelinek <jakub@redhat.com>
11647 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
11648 instead of 10??LU, perform unit multiplication in wide_int,
11649 don't change alloc_object_size_limit if the limit is larger
11652 PR tree-optimization/81655
11653 PR tree-optimization/81588
11654 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
11655 the case when ranges[i].low and high are 1 for unsigned type with
11658 PR middle-end/81052
11659 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
11660 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
11662 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
11664 * tree-vrp.h: Add include guard.
11666 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
11669 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
11670 (ud2): New insn pattern.
11671 * config/i386/i386.c (ix86_expand_epilogue):
11672 For naked functions, generate ud2 instead of trap insn.
11674 2017-08-02 Marek Polacek <polacek@redhat.com>
11677 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
11679 2017-08-02 Tom de Vries <tom@codesourcery.com>
11680 Cesar Philippidis <cesar@codesourcery.com>
11682 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
11683 Add missing edge probabilities.
11685 2017-08-02 Tamar Christina <tamar.christina@arm.com>
11687 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
11688 Correct endianness.
11690 2017-08-02 Jakub Jelinek <jakub@redhat.com>
11692 PR middle-end/79499
11693 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
11694 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
11695 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
11696 prologue_seq sequences - if any.
11698 2017-08-02 Richard Biener <rguenther@suse.de>
11700 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
11701 via vectors if supported, integer extracts via punning if supported
11702 or otherwise vector extracts.
11704 2017-08-02 Richard Biener <rguenther@suse.de>
11706 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
11708 (bitmap_insert_into_set): ... this.
11710 2017-08-02 Richard Biener <rguenther@suse.de>
11712 PR tree-optimization/81633
11714 2015-08-17 Alan Hayward <alan.hayward@arm.com>
11716 PR tree-optimization/71752
11717 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
11719 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
11721 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
11722 (machine_function::call_ms2sysv_pad_out): Remove field.
11723 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
11724 (ix86_compute_frame_layout): Likewise.
11726 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
11729 * config/i386/i386.c (ix86_set_func_type): Disallow naked
11730 attribute with interrupt attribute.
11732 2017-08-01 Andrew Pinski <apinski@cavium.com>
11734 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
11735 BIT_INSERT_EXPR's operand 1
11736 to see if the types precision matches.
11738 2017-08-01 Martin Liska <mliska@suse.cz>
11740 PR middle-end/70140
11741 * builtins.c (expand_builtin_memcpy_args): Remove.
11742 (expand_builtin_memcpy): Call newly added function
11743 expand_builtin_memory_copy_args.
11744 (expand_builtin_memcpy_with_bounds): Likewise.
11745 (expand_builtin_mempcpy): Remove last argument.
11746 (expand_builtin_mempcpy_with_bounds): Likewise.
11747 (expand_builtin_memory_copy_args): New function created from
11748 expand_builtin_mempcpy_args with small modifications.
11749 (expand_builtin_mempcpy_args): Remove.
11750 (expand_builtin_stpcpy): Remove unused argument.
11751 (expand_builtin): Likewise.
11752 (expand_builtin_with_bounds): Likewise.
11754 2017-08-01 Martin Liska <mliska@suse.cz>
11757 Make mempcpy more optimal (PR middle-end/70140).
11759 2017-08-01 Jakub Jelinek <jakub@redhat.com>
11762 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
11763 __builtin_vec_cmpne verify both arguments are compatible vectors
11764 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
11765 verify arg1_type is a pointer or array type. For __builtin_vec_st,
11766 move computation of aligned to after checking the argument types.
11770 * config/rs6000/vsx.md (vextract_fp_from_shorth,
11771 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
11774 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
11775 Doug Rupp <rupp@adacore.com>
11776 Olivier Hainque <hainque@adacore.com>
11778 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
11779 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
11781 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
11782 for TARGET_OS_CPP_BUILTIN.
11783 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
11784 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
11786 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
11787 passing required abi options to the assembler for EABI configurations.
11788 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
11789 of .text.hot and .text.unlikely sections for kernel modules when
11790 using ARM style exceptions.
11791 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
11792 options. Add EXTRA_CC1_SPEC.
11793 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
11795 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
11797 (ARM_TARGET2_DWARF_FORMAT): Define.
11798 * config/arm/t-vxworks: Adjust multilib control to removal of the
11799 Diab command line options.
11801 2017-08-01 Martin Liska <mliska@suse.cz>
11803 PR gcov-profile/81561
11804 * gcov.c (unblock): Make unblocking safe as we need to preserve
11805 index correspondence of blocks and block_lists.
11807 2017-08-01 Richard Biener <rguenther@suse.de>
11809 PR tree-optimization/81181
11810 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
11811 (compute_antic): ... end of iteration here.
11813 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
11815 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
11816 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
11817 (ftree-slp-vectorize): Likewise.
11818 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
11819 can no longer be set independent of flag_tree_loop_vectorize.
11820 * omp-general.c (emp_max_vf): Likewise.
11821 * opts.c (enable_fdo_optimizations): Remove references to
11822 flag_tree_vectorize, these are now implicit.
11823 (common_handle_option): Remove handling for OPT_ftree_vectorize,
11824 and leave it for the options machinery.
11826 2017-08-01 Martin Liska <mliska@suse.cz>
11828 PR middle-end/70140
11829 * builtins.c (expand_builtin_memcpy_args): Remove.
11830 (expand_builtin_memcpy): Call newly added function
11831 expand_builtin_memory_copy_args.
11832 (expand_builtin_memcpy_with_bounds): Likewise.
11833 (expand_builtin_mempcpy): Remove last argument.
11834 (expand_builtin_mempcpy_with_bounds): Likewise.
11835 (expand_builtin_memory_copy_args): New function created from
11836 expand_builtin_mempcpy_args with small modifications.
11837 (expand_builtin_mempcpy_args): Remove.
11838 (expand_builtin_stpcpy): Remove unused argument.
11839 (expand_builtin): Likewise.
11840 (expand_builtin_with_bounds): Likewise.
11842 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
11845 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
11846 print "ds:" only for immediates in generic address space.
11848 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
11851 * config/i386/i386.c (ix86_funciton_naked): New prototype.
11852 (ix86_function_ok_for_sibcall): Return false for naked functions.
11854 2017-08-01 Richard Biener <rguenther@suse.de>
11856 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
11857 (compute_antic): Seed worklist with exit block predecessors.
11858 * cfganal.c (dfs_find_deadend): For a cycle return the source
11859 of the edge closing it.
11861 2017-08-01 Tamar Christina <tamar.christina@arm.com>
11863 * config/aarch64/aarch64.c
11864 (aarch64_can_const_movi_rtx_p): Move 0 check.
11866 2017-08-01 Bin Cheng <bin.cheng@arm.com>
11868 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
11869 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
11871 * match.pd: Ditto in address comparison pattern.
11873 2017-08-01 Bin Cheng <bin.cheng@arm.com>
11875 PR tree-optimization/81627
11876 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
11877 closed ssa form for store-store chain.
11879 2017-08-01 Bin Cheng <bin.cheng@arm.com>
11881 PR tree-optimization/81620
11882 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
11883 for store-store chain.
11885 2017-08-01 Jakub Jelinek <jakub@redhat.com>
11887 PR tree-optimization/81588
11888 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
11889 ranges[i].in_p, invert comparison code ccode. For >/>=,
11890 swap rhs1 and rhs2 and comparison code unconditionally,
11891 for </<= don't do that. Don't swap rhs1/rhs2 again if
11892 ranges[i].in_p, instead invert comparison code ccode if
11893 opcode or oe->rank is BIT_IOR_EXPR.
11896 * optabs.def (vec_extract_optab, vec_init_optab): Change from
11897 a direct optab to conversion optab.
11898 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
11899 with GET_MODE_INNER as last argument instead of optab_handler.
11900 * expmed.c (extract_bit_field_1): Likewise. Use vector from
11901 vector extraction if possible and optab is available.
11902 * expr.c (store_constructor): Use convert_optab_handler instead
11903 of optab_handler. Use vector initialization from smaller
11904 vectors if possible and optab is available.
11905 * tree-vect-stmts.c (vectorizable_load): Likewise.
11906 * doc/md.texi (vec_extract, vec_init): Document that the optabs
11907 now have two modes.
11908 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
11909 of vec_init from half-sized vectors with the same element mode.
11910 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
11911 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
11912 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
11913 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
11914 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
11915 after mode in gen_vec_extract* calls.
11916 (vec_extract<mode>): Renamed to ...
11917 (vec_extract<mode><ssescalarmodelower>): ... this.
11918 (vec_extract<mode><ssehalfvecmodelower>): New expander.
11919 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
11920 element mode after mode in gen_vec_init* calls.
11921 (VEC_INIT_HALF_MODE): New mode iterator.
11922 (vec_init<mode>): Renamed to ...
11923 (vec_init<mode><ssescalarmodelower>): ... this.
11924 (vec_init<mode><ssehalfvecmodelower>): New expander.
11925 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
11926 (vec_extractv2sfsf): ... this.
11927 (vec_initv2sf): Renamed to ...
11928 (vec_initv2sfsf): ... this.
11929 (vec_extractv2si): Renamed to ...
11930 (vec_extractv2sisi): ... this.
11931 (vec_initv2si): Renamed to ...
11932 (vec_initv2sisi): ... this.
11933 (vec_extractv4hi): Renamed to ...
11934 (vec_extractv4hihi): ... this.
11935 (vec_initv4hi): Renamed to ...
11936 (vec_initv4hihi): ... this.
11937 (vec_extractv8qi): Renamed to ...
11938 (vec_extractv8qiqi): ... this.
11939 (vec_initv8qi): Renamed to ...
11940 (vec_initv8qiqi): ... this.
11941 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
11942 (vec_init<mode>): Renamed to ...
11943 (vec_init<mode><VEC_base_l>): ... this.
11944 (vec_extract<mode>): Renamed to ...
11945 (vec_extract<mode><VEC_base_l>): ... this.
11946 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
11947 (vec_initv2sfsf): ... this.
11948 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
11949 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
11950 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
11951 element mode after mode in gen_vec_init* calls.
11952 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
11953 (vec_init<mode><Vel>): ... this.
11954 (vec_extract<mode>): Renamed to ...
11955 (vec_extract<mode><Vel>): ... this.
11956 * config/aarch64/iterators.md (Vel): New mode attribute.
11957 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
11958 Add element mode after mode in gen_vec_extract* calls.
11959 * config/s390/vector.md (non_vec_l): New mode attribute.
11960 (vec_extract<mode>): Renamed to ...
11961 (vec_extract<mode><non_vec_l>): ... this.
11962 (vec_init<mode>): Renamed to ...
11963 (vec_init<mode><non_vec_l>): ... this.
11964 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
11965 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
11967 * config/arm/iterators.md (V_elem_l): New mode attribute.
11968 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
11969 (vec_extract<mode><V_elem_l>): ... this.
11970 (vec_extractv2di): Renamed to ...
11971 (vec_extractv2didi): ... this.
11972 (vec_init<mode>): Renamed to ...
11973 (vec_init<mode><V_elem_l>): ... this.
11974 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
11975 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
11976 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
11977 Add element mode after gen_vec_extract* calls.
11978 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
11979 (vec_init<mode><unitmode>): ... this.
11980 (vec_extract<mode>): Renamed to ...
11981 (vec_extract<mode><unitmode>): ... this.
11982 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
11983 (vec_init<mode><unitmode>): ... this.
11984 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
11985 (vec_initv2sfsf): ... this.
11986 (vec_extractv2sf): Renamed to ...
11987 (vec_extractv2sfsf): ... this.
11988 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
11989 Add element mode after gen_vec_extract* calls.
11990 * config/mips/mips.md (unitmode): New mode iterator.
11991 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
11992 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
11993 * config/spu/spu.md (inner_l): New mode attribute.
11994 (vec_init<mode>): Renamed to ...
11995 (vec_init<mode><inner_l>): ... this.
11996 (vec_extract<mode>): Renamed to ...
11997 (vec_extract<mode><inner_l>): ... this.
11998 * config/sparc/sparc.md (veltmode): New mode iterator.
11999 (vec_init<VMALL:mode>): Renamed to ...
12000 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
12001 * config/ia64/vect.md (vec_initv2si): Renamed to ...
12002 (vec_initv2sisi): ... this.
12003 (vec_initv2sf): Renamed to ...
12004 (vec_initv2sfsf): ... this.
12005 (vec_extractv2sf): Renamed to ...
12006 (vec_extractv2sfsf): ... this.
12007 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
12008 (vec_init<mode>): Renamed to ...
12009 (vec_init<mode><VEC_base_l>): ... this.
12010 (vec_extract<mode>): Renamed to ...
12011 (vec_extract<mode><VEC_base_l>): ... this.
12012 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
12013 (vec_initv2sfsf): ... this.
12014 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
12015 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
12016 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
12017 gen_vec_init* calls.
12019 2017-08-01 Richard Biener <rguenther@suse.de>
12021 PR tree-optimization/81297
12022 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
12023 TREE_OVERFLOW from INTEGER_CSTs.
12025 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
12027 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
12029 2017-07-31 Carl Love <cel@us.ibm.com>
12031 * config/rs6000/rs6000-c: Add support for built-in functions
12032 vector signed char vec_xl_be (signed long long, signed char *);
12033 vector unsigned char vec_xl_be (signed long long, unsigned char *);
12034 vector signed int vec_xl_be (signed long long, signed int *);
12035 vector unsigned int vec_xl_be (signed long long, unsigned int *);
12036 vector signed long long vec_xl_be (signed long long, signed long long *);
12037 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
12038 vector signed short vec_xl_be (signed long long, signed short *);
12039 vector unsigned short vec_xl_be (signed long long, unsigned short *);
12040 vector double vec_xl_be (signed long long, double *);
12041 vector float vec_xl_be (signed long long, float *);
12042 * config/rs6000/altivec.h (vec_xl_be): Add #define.
12043 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
12044 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
12046 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
12047 (altivec_expand_builtin): Add switch statement to call
12048 altivec_expand_xl_be for each builtin.
12049 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
12050 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
12051 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
12052 __builtin_vsx_le_be_v16qi.
12053 * doc/extend.texi: Update the built-in documentation file for the
12054 new built-in functions.
12056 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
12059 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
12061 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
12063 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12065 * config.gcc: Add z14.
12066 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
12067 CPU model numbers for z13s and z14.
12068 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
12070 * config/s390/s390-opts.h (enum processor_type): Rename
12071 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
12072 * config/s390/s390.c (processor_table): Add field for CPU name to
12073 be passed to Binutils.
12074 (s390_asm_output_machine_for_arch): Use the new field in
12075 processor_table for Binutils.
12076 (s390_expand_builtin): Replace arch12 with z14.
12077 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
12078 (s390_get_sched_attrmask): Likewise.
12079 (s390_get_unit_mask): Likewise.
12080 * config/s390/s390.opt: Add z14 to processor_type enum.
12082 2017-07-31 Martin Jambor <mjambor@suse.cz>
12085 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
12086 regardless of optimization level.
12088 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
12089 Martin Liska <mliska@suse.cz>
12091 * predict.def: Remove old comment and adjust probability.
12092 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
12093 PREDICT statements.
12095 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
12098 * config/i386/i386.c (ix86_function_naked): New function.
12099 (ix86_can_use_return_insn_p): Return false for naked functions.
12100 (ix86_expand_prologue): Skip prologue for naked functions.
12101 (ix86_expand_epilogue): Skip epilogue for naked functions
12102 and emit trap instruction.
12103 (ix86_warn_func_return): New function.
12104 (ix86_attribute_table): Add "naked" attribute specification.
12105 (TARGET_WARN_FUNC_RETURN): Define.
12106 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
12108 2017-07-31 Martin Liska <mliska@suse.cz>
12110 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
12111 (dump_gimple_bb_header): Always dump BB info.
12112 (pp_cfg_jump): Do not append info about BB when dumping a jump.
12114 2017-07-31 Martin Liska <mliska@suse.cz>
12117 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
12118 also with current_function_decl non-null equality.
12120 2017-07-31 Jakub Jelinek <jakub@redhat.com>
12123 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
12124 change type to the element type, instead add eltype variable and
12125 use it where we are interested in the element type.
12127 PR tree-optimization/81603
12128 * ipa-polymorphic-call.c
12129 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
12130 offset arithmetic in offset_int, bail out if the resulting bit offset
12131 doesn't fit into shwi.
12133 2017-07-31 Martin Liska <mliska@suse.cz>
12135 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
12136 (gimplify_save_expr): Fix comment.
12138 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
12141 * config/i386/i386.c (ix86_function_arg): Update arguments for
12143 (ix86_compute_frame_layout): Set the initial stack offset to
12144 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
12145 INCOMING_FRAME_SP_OFFSET.
12146 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
12147 stack before exception handler returns.
12148 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
12149 the 'ERROR_CODE' for exception handler.
12151 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
12153 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
12154 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
12155 (ASM_OUTPUT_REG_POP): Ditto.
12156 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
12157 instead of asm_fprintf to output pure string.
12159 2017-07-29 Jakub Jelinek <jakub@redhat.com>
12161 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
12162 to imported_module_or_decl hook.
12163 (debug_nothing_tree_tree_tree_bool): Remove.
12164 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
12165 * debug.c (do_nothing_debug_hooks): Use
12166 debug_nothing_tree_tree_tree_bool_bool instead of
12167 debug_nothing_tree_tree_tree_bool.
12168 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
12169 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
12170 * sdbout.c (sdb_debug_hooks): Likewise.
12171 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
12172 (gen_namespace_die): Add DW_AT_export_symbols attribute if
12174 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
12175 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
12176 attribute, don't add anything.
12178 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12180 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
12181 (fold_build2_stat_loc): Likewise.
12182 (fold_build3_stat_loc): Likewise.
12183 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
12184 (fold_build1_loc): Remove macro.
12185 (fold_build2_loc): Likewise.
12186 (fold_build3_loc): Likewise.
12188 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12190 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
12191 (gimple_build_debug_bind_source_stat): Likewise.
12192 * gimple.h (gimple_build_debug_bind): Remove macro.
12193 (gimple_build_debug_bind_source): Likewise.
12195 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12197 * bitmap.c (bitmap_alloc): Adjust.
12198 (bitmap_gc_alloc): Likewise.
12199 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
12201 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12203 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
12204 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
12205 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
12206 (bitmap_gc_alloc_stat): Likewise.
12207 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
12209 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12211 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
12212 * rtl.h (shallow_copy_rtx): Remove macro.
12214 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12216 * emit-rtl.c (gen_raw_REG): Adjust.
12217 * gengenrtl.c (gendef): Likewise.
12218 * rtl.c (rtx_alloc_stat): Remove _stat from name.
12219 * rtl.h (rtx_alloc): Remove macro.
12221 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12223 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
12224 (build_tree_list_stat): Likewise.
12225 * tree.h (build_tree_list): Remove macro.
12226 (build_tree_list_vec): Likewise.
12228 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12230 * tree.c (make_vector_stat): Remove _stat from name.
12231 (build_vector_stat): Likewise.
12232 * tree.h (make_vector_stat): Remove macro.
12233 (build_vector_stat): Likewise.
12235 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12237 * tree.h (build_var_debug_value): Remove prototype.
12239 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12241 * tree.c (tree_cons_stat): Remove _stat from name.
12242 * tree.h (tree_cons): Remove macro.
12244 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12246 * tree.c (build_vl_exp_stat): Remove _stat from name.
12247 * tree.h (build_vl_exp): Remove macro.
12249 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12251 * tree.c (build_decl_stat): Remove _stat from name.
12252 * tree.h (build_decl): Remove macro.
12254 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12256 * gimple.c (gimple_build_with_ops_stat): Adjust.
12257 (gimple_alloc_stat): Remove _stat from name.
12258 * gimple.h (gimple_alloc): Remove macro.
12260 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12262 * tree.c (make_tree_vec_stat): Remove _stat from name.
12263 (grow_tree_vec_stat): Likewise.
12264 * tree.h (make_tree_vec_stat): Adjust prototype.
12265 (grow_tree_vec_stat): Likewise.
12266 (make_tree_vec): Remove macro.
12267 (grow_tree_vec): Likewise.
12269 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12271 * fold-const.c (fold_build1_stat_loc): Adjust.
12272 (fold_build2_stat_loc): Likewise.
12273 (fold_build3_stat_loc): Likewise.
12274 * tree.c (build0_stat): Remove _stat from name.
12275 (build1_stat): Likewise.
12276 (build2_stat): Likewise.
12277 (build3_stat): Likewise.
12278 (build4_stat): Likewise.
12279 (build5_stat): Likewise.
12280 * tree.h (build1_loc): Remove macro, and rename _stat function
12282 (build2_loc): Likewise.
12283 (build3_loc): Likewise.
12284 (build4_loc): Likewise.
12285 (build5_loc): Likewise.
12287 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12289 * tree.c (make_int_cst_stat): Remove _stat from name.
12290 * tree.h (make_int_cst_stat): Adjust prototype.
12291 (make_int_cst): Remove macro.
12293 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12295 * tree.c (make_tre_binfo_stat): Remove _stat from name.
12296 * tree.h (make_tree_binfo_stat): Adjust prototype.
12297 (make_tree_binfo): Remove.
12299 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12301 * tree.c (copy_node_stat): Rename to copy_node.
12302 (build_distinct_type_copy): Adjust.
12303 * tree.h (copy_node_stat): Adjust prototype.
12304 (copy_node): Remove macro.
12306 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12308 * tree.c (make_node_stat): rename to make_node.
12309 (build_tree_list_stat): Adjust.
12310 (build0_stat): Likewise.
12311 (build2_stat): Likewise.
12312 (build3_stat): Likewise.
12313 (build4_stat): Likewise.
12314 (build5_stat): Likewise.
12315 (build_decl_stat): Likewise.
12316 * tree.h (make_node_stat): Adjust prototype.
12317 (make_node): remove macro.
12319 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
12321 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
12322 (PPC_FEATURE2_SCV): Likewise.
12323 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
12325 2017-07-28 Tamar Christina <tamar.christina@arm.com>
12327 * config/aarch64/aarch64.c
12328 (aarch64_internal_mov_immediate): Add new special pattern.
12329 * config/aarch64/aarch64.md (*movdi_aarch64):
12330 Add reg/32bit const mov case.
12332 2017-07-28 Tamar Christina <tamar.christina@arm.com>
12333 Richard Sandiford <richard.sandiford@linaro.org>
12335 * config/aarch64/aarch64.md (mov<mode>): Generalize.
12336 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
12337 Add integer and movi cases.
12338 (movi-split-hf-df-sf split, fp16): New.
12339 (enabled): Added TARGET_FP_F16INST.
12340 * config/aarch64/iterators.md (GPF_HF): New.
12341 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
12343 2017-07-28 Tamar Christina <tamar.christina@arm.com>
12345 * config/aarch64/aarch64.c
12346 (aarch64_simd_container_mode): Add prototype.
12347 (aarch64_expand_mov_immediate): Add HI support.
12348 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
12349 (aarch64_can_const_movi_rtx_p): New.
12350 (aarch64_preferred_reload_class):
12351 Remove restrictions of using FP registers for certain SIMD operations.
12352 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
12353 (aarch64_valid_floating_const): Add integer move validation.
12354 (aarch64_simd_imm_scalar_p): Remove.
12355 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
12356 (aarch64_legitimate_constant_p): Expand list of supported cases.
12357 * config/aarch64/aarch64-protos.h
12358 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
12359 (aarch64_reinterpret_float_as_int): New.
12360 (aarch64_simd_imm_scalar_p): Remove.
12361 * config/aarch64/constraints.md (Uvi): New.
12362 (Dd): Split into Ds and new Dd.
12363 * config/aarch64/aarch64.md (*movsi_aarch64):
12365 (*movdi_aarch64): Add SIMD mov case.
12367 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12369 * tree-predcom.c: (struct chain): Handle store-store chain in which
12370 stores for elimination only store loop invariant values.
12371 (execute_pred_commoning_chain): Ditto.
12372 (prepare_initializers_chain_store_elim): Ditto.
12373 (prepare_finalizers): Ditto.
12374 (is_inv_store_elimination_chain): New function.
12375 (initialize_root_vars_store_elim_1): New function.
12377 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12379 * tree-predcom.c: Revise general description of the pass.
12380 (enum chain_type): New enum type for store elimination.
12381 (struct chain): New field supporting store elimination.
12382 (struct component): Ditto.
12383 (dump_chain): Dump store-stores chain.
12384 (release_chain): Release resources.
12385 (split_data_refs_to_components): Compute and create component
12386 contains only stores for elimination.
12387 (get_chain_last_ref_at): New function.
12388 (make_invariant_chain): Initialization.
12389 (make_rooted_chain): Specify chain type in parameter and record it.
12390 (add_looparound_copies): Skip for store-stores chain.
12391 (determine_roots_comp): Compute type of chain and pass it to
12393 (initialize_root_vars_store_elim_2): New function.
12394 (finalize_eliminated_stores): New function.
12395 (remove_stmt): Handle store for elimination.
12396 (execute_pred_commoning_chain): Execute predictive commoning on
12397 store-store chains.
12398 (determine_unroll_factor): Skip unroll for store-stores chain.
12399 (prepare_initializers_chain_store_elim): New function.
12400 (prepare_initializers_chain): Hanlde store-store chain.
12401 (prepare_finalizers_chain, prepare_finalizers): New function.
12402 (tree_predictive_commoning_loop): Return integer value indicating
12403 if loop is unrolled or lcssa form is corrupted.
12404 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
12406 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12408 * tree-predcom.c (initialize_root): Delete.
12409 (execute_pred_commoning_chain): Initialize root vars and replace
12410 reference of non-combined chain directly, rather than call above
12413 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12415 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
12416 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
12418 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12420 * tree-predcom.c (struct chain): New field init_seq.
12421 (release_chain): Release init_seq.
12422 (prepare_initializers_chain): Record intialization stmts in above
12424 (insert_init_seqs): New function.
12425 (tree_predictive_commoning_loop): Call insert_init_seqs.
12427 2017-07-28 Bin Cheng <bin.cheng@arm.com>
12429 * tree-predcom.c (determine_roots_comp): Skip trivial components.
12431 2017-07-28 Richard Biener <rguenther@suse.de>
12433 * match.pd: Remove superfluous :c.
12434 * genmatch.c (simplify::id): Add member.
12435 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
12437 (current_id): New global.
12438 (dt_node::parent): Move from ...
12439 (dt_operand::parent): ... here. Add for_id member.
12440 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
12441 (decision_tree::find_node): Relax order requirement when
12442 merging DT_TRUE nodes to ones inbetween the current simplify
12443 and the one we try to merge with. Add diagnostic whenever
12444 we need to enforce pattern order by not merging.
12445 (decision_tree::insert): Set current_id.
12446 (decision_tree::print_node): Dump parent node and for_id.
12447 (parser::last_id): Add member.
12448 (parser::push_simplify): Assign unique id.
12449 (parser::parser): Initialize last_id.
12451 2017-07-28 Martin Liska <mliska@suse.cz>
12454 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
12455 gimple_build_debug_bind.
12457 2017-07-28 Richard Biener <rguenther@suse.de>
12459 PR tree-optimization/81502
12460 * match.pd: Add pattern combining BIT_INSERT_EXPR with
12462 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
12464 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
12465 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
12466 for BIT_FIELD_REF args.
12467 * fold-const.c (make_bit_field_ref): Likewise.
12468 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
12470 2017-07-28 Jakub Jelinek <jakub@redhat.com>
12473 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
12474 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
12475 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
12476 Or it into SANITIZER_UNDEFINED.
12477 * ubsan.c: Include gimple-fold.h and varasm.h.
12478 (ubsan_expand_ptr_ifn): New function.
12479 (instrument_pointer_overflow): New function.
12480 (maybe_instrument_pointer_overflow): New function.
12481 (instrument_object_size): Formatting fix.
12482 (pass_ubsan::execute): Call instrument_pointer_overflow
12483 and maybe_instrument_pointer_overflow.
12484 * internal-fn.c (expand_UBSAN_PTR): New function.
12485 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
12486 * sanitizer.def (__ubsan_handle_pointer_overflow,
12487 __ubsan_handle_pointer_overflow_abort): New builtins.
12488 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
12489 * internal-fn.def (UBSAN_PTR): New internal function.
12490 * opts.c (sanitizer_opts): Add pointer-overflow.
12491 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
12492 * fold-const.c (build_range_check): Compute pointer range check in
12493 integral type if pointer arithmetics would be needed. Formatting
12496 2017-07-28 Martin Liska <mliska@suse.cz>
12499 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
12500 parameters that are of a variable-length.
12502 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
12504 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
12506 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
12507 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
12508 (CRT_CALL_STATIC_FUNCTION): Likewise.
12509 (ASM_DEFAULT_SPEC): New define.
12510 (ASM_SPEC32): Likewise.
12511 (ASM_SPEC64): Likewise.
12512 (ASM_SPEC_COMMON): Likewise.
12513 (ASM_SPEC): Likewise.
12514 (INVALID_64BIT): Likewise.
12515 (LINK_OS_DEFAULT_SPEC): Likewise.
12516 (LINK_OS_SPEC32): Likewise.
12517 (LINK_OS_SPEC64): Likewise.
12518 (POWERPC_LINUX): Likewise.
12519 (PTRDIFF_TYPE): Likewise.
12520 (RESTORE_FP_PREFIX): Likewise.
12521 (RESTORE_FP_SUFFIX): Likewise.
12522 (SAVE_FP_PREFIX): Likewise.
12523 (SAVE_FP_SUFFIX): Likewise.
12524 (SIZE_TYPE): Likewise.
12525 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
12526 (TARGET_64BIT): Likewise.
12527 (TARGET_64BIT): Likewise.
12528 (TARGET_AIX): Likewise.
12529 (WCHAR_TYPE_SIZE): Likewise.
12530 (WCHAR_TYPE): Undefine.
12531 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
12532 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
12533 (CPP_OS_RTEMS_SPEC): Delete.
12534 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
12535 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
12537 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
12539 2017-07-28 Jakub Jelinek <jakub@redhat.com>
12541 PR tree-optimization/81578
12542 * tree-parloops.c (build_new_reduction): Bail out if
12543 reduction_code isn't one of the standard OpenMP reductions.
12544 Move the details printing after that decision.
12546 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
12548 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
12549 related to reload_in_progress.
12550 (splat_input_operand): Likewise.
12551 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
12553 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
12555 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
12556 (TARGET_INSTANTIATE_DECLS): Likewise.
12557 (legitimate_indexed_address_p): Delete reload_in_progress code.
12558 (rs6000_debug_legitimate_address_p): Likewise.
12559 (rs6000_eliminate_indexed_memrefs): Likewise.
12560 (rs6000_emit_le_vsx_store): Likewise.
12561 (rs6000_emit_move_si_sf_subreg): Likewise.
12562 (rs6000_emit_move): Likewise.
12563 (register_to_reg_type): Likewise.
12564 (rs6000_pre_atomic_barrier): Likewise.
12565 (rs6000_machopic_legitimize_pic_address): Likewise.
12566 (rs6000_allocate_stack_temp): Likewise.
12567 (rs6000_address_for_fpconvert): Likewise.
12568 (rs6000_address_for_altivec): Likewise.
12569 (rs6000_secondary_memory_needed_rtx): Delete function.
12570 (rs6000_check_sdmode): Likewise.
12571 (rs6000_alloc_sdmode_stack_slot): Likewise.
12572 (rs6000_instantiate_decls): Likewise.
12573 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
12574 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
12575 Delete reload_in_progress.
12576 (*vec_reload_and_plus_<mptrsize>): Likewise.
12577 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
12578 (vsx_div_v2di): Likewise.
12579 (vsx_udiv_v2di): Likewise.
12581 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
12583 * config/rs6000/rs6000.opt (mlra): Replace with stub.
12584 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
12585 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
12586 (rs6000_debug_reg_global): Delete print of LRA status.
12587 (rs6000_option_override_internal): Delete dead LRA related code.
12588 (rs6000_lra_p): Delete function.
12589 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
12591 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
12593 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
12594 * config/riscv/rtems.h: New file.
12596 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12597 Sudakshina Das <sudi.das@arm.com>
12599 * config/aarch64/aarch64.md
12600 (define_split for and<mode>3nr_compare): Move
12601 non aarch64_logical_operand to a register.
12602 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
12603 register immediate operand to a register.
12604 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
12606 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
12608 PR middle-end/81564
12609 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
12611 2017-07-27 Richard Biener <rguenther@suse.de>
12613 PR tree-optimization/81573
12614 PR tree-optimization/81494
12615 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
12616 multi defuse cycle case.
12618 2017-07-27 Richard Biener <rguenther@suse.de>
12620 PR tree-optimization/81571
12621 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
12624 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
12626 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
12627 earlier and only if MASK_FPU is set. Adjust formatting.
12629 2017-07-27 Martin Liska <mliska@suse.cz>
12631 * opt-functions.awk: Add validation of value of Init.
12632 * optc-gen.awk: Pass new argument.
12634 2017-07-27 Martin Liska <mliska@suse.cz>
12636 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
12637 Fix wrong condition.
12639 2017-07-27 Martin Liska <mliska@suse.cz>
12641 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
12642 BBs and edges seen by autoFDO.
12644 2017-07-27 Richard Biener <rguenther@suse.de>
12646 PR tree-optimization/81502
12647 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
12648 with incompatible but same sized type.
12649 (execute_update_addresses_taken): Likewise.
12651 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
12653 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
12654 flag_tree_loop_vectorize rather than flag_tree_vectorize.
12656 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12659 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
12660 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
12661 Change s_operand to memory_operand.
12663 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
12665 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
12666 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
12667 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
12668 Emit instructions rather than returning an expression. Handle TFmode
12669 and KFmode by casting to TImode.
12670 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
12671 (rs6000_emit_le_vsx_store): Likewise.
12672 * config/rs6000/vsx.md (VSX_TI): New iterator.
12673 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
12674 (*vsx_le_undo_permute_<mode>): Likewise.
12675 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
12676 emit the split sequence.
12677 (*vsx_le_perm_store_<mode>): Likewise.
12679 2017-07-27 Jakub Jelinek <jakub@redhat.com>
12681 PR tree-optimization/81555
12682 PR tree-optimization/81556
12683 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
12684 if true, force CHANGED for the recursive invocation.
12685 (reassociate_bb): Remember original length of ops array, pass
12686 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
12688 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
12689 attributes for noipa attribute. For naked attribute use
12690 lookup_attribute first before lookup_attribute_spec.
12691 * final.c (rest_of_handle_final): Disable IPA RA for functions with
12693 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
12694 for functions with noipa attribute.
12695 (cgraph_externally_visible_p): Return true for functions with noipa
12697 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
12698 for functions with noipa attribute.
12699 * doc/extend.texi: Document noipa function attribute.
12700 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
12701 also for functions with noipa attribute.
12702 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
12704 2017-07-26 Andrew Pinski <apinski@cavium.com>
12706 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
12707 vec_unalign_load_cost and vec_unalign_store_cost.
12709 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
12711 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
12712 -mvsx-small-integer option.
12713 (ISA_3_0_MASKS_IEEE): Likewise.
12714 (OTHER_VSX_VECTOR_MASKS): Likewise.
12715 (POWERPC_MASKS): Likewise.
12716 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
12717 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
12718 code, only testing for DImode being allowed in non-VSX floating
12720 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
12721 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
12723 (rs6000_option_override_internal): Delete -mvsx-small-integer.
12724 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
12725 TARGET_P8_VECTOR test.
12726 (rs6000_secondary_reload_simple_move): Likewise.
12727 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
12728 since TARGET_P9_VECTOR was already tested.
12729 (rs6000_opt_masks): Remove -mvsx-small-integer.
12730 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
12731 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
12733 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
12734 test for TARGET_VEXTRACTUB was used, and that uses
12736 (p9 extract splitter): Likewise.
12737 (vsx_extract_<mode>_di_p9): Likewise.
12738 (vsx_extract_<mode>_store_p9): Likewise.
12739 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
12740 for TARGET_P9_VECTOR was used. Delete code that is now dead with
12741 the elimination of TARGET_VSX_SMALL_INTEGER.
12742 (vsx_extract_<mode>_p8): Likewise.
12743 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
12744 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
12745 (vsx_set_<mode>_p9): Likewise.
12746 (vsx_set_v4sf_p9): Likewise.
12747 (vsx_set_v4sf_p9_zero): Likewise.
12748 (vsx_insert_extract_v4sf_p9): Likewise.
12749 (vsx_insert_extract_v4sf_p9_2): Likewise.
12750 * config/rs6000/rs6000.md (sign extend splitter): Change
12751 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
12752 (floatsi<mode>2_lfiwax_mem): Likewise.
12753 (floatunssi<mode>2_lfiwzx_mem): Likewise.
12754 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
12755 since a test for TARGET_P9_VECTOR was used.
12756 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
12757 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
12758 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
12759 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
12760 TARGET_P8_VECTOR test.
12761 (fix_trunc<mode>si2_stfiwx): Likewise.
12762 (fix_trunc<mode>si2_internal): Likewise.
12763 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
12764 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
12766 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
12767 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
12768 TARGET_P8_VECTOR test.
12769 (fixuns_trunc<mode>si2_stfiwx): Likewise.
12770 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
12771 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
12773 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
12774 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
12775 since a test for TARGET_P9_VECTOR was used.
12776 (splitter for loading small constants): Likewise.
12778 2017-07-26 Andrew Pinski <apinski@cavium.com>
12780 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
12783 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
12786 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
12787 (fp_valid_at): Likewise.
12789 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
12791 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
12792 (qdf24xx_addrcost_table): Likewise.
12793 (cortexa57_tunings): Update to use generic_branch_cost.
12794 (cortexa72_tunings): Likewise.
12795 (cortexa73_tunings): Likewise.
12796 (qdf24xx_tunings): Likewise.
12798 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
12800 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
12801 (thunderx2t99_branch_cost): Likewise.
12802 (cortexa35_tunings): Update to use generic_branch_cost.
12803 (cortexa53_tunings): Likewise.
12804 (cortexa57_tunings): Likewise.
12805 (cortexa72_tunings): Likewise.
12806 (cortexa73_tunings): Likewise.
12807 (thunderx2t99_tunings): Likewise.
12809 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
12811 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
12812 (sparc_option_override): Honour MASK_FSMULD.
12813 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
12814 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
12815 * config/sparc/sparc.opt (mfsmuld): New option.
12816 * doc/invoke.texi (mfsmuld): Document option.
12818 2017-07-26 Marek Polacek <polacek@redhat.com>
12820 PR middle-end/70992
12821 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
12823 2017-07-26 Richard Biener <rguenther@suse.de>
12825 * gimple-match-head.c (do_valueize): Return OP if valueize
12827 (get_def): New helper to get at the def stmt of a SSA name
12828 if valueize allows.
12829 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
12830 do_valueize to get at the def stmt.
12831 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
12833 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
12835 PR middle-end/46932
12836 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
12838 2017-07-26 Martin Liska <mliska@suse.cz>
12841 * function.c (expand_function_start): Make expansion of
12842 nonlocal_goto_save_area after parm_birth_insn.
12844 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
12846 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
12847 from all CPU target flags enable members.
12849 2017-07-26 Richard Biener <rguenther@suse.de>
12851 * genmatch.c (dt_simplify::gen): Make iterator vars const.
12852 (decision_tree::gen): Make 'type' const.
12853 (write_predicate): Likewise.
12855 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
12857 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
12858 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
12859 (rs6000_option_override_internal): Likewise.
12860 (rs6000_expand_vector_set): Likewise.
12861 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
12862 (TARGET_UPPER_REGS_SF): Likewise.
12863 (TARGET_UPPER_REGS_DI): Likewise.
12864 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
12865 (TARGET_DIRECT_MOVE_64BIT): Likewise.
12866 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
12867 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
12868 (Splitters for DI constants in Altivec registers): Likewise.
12869 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
12870 (vsx_set_v4sf_p9): Likewise.
12871 (vsx_set_v4sf_p9_zero): Likewise.
12872 (vsx_insert_extract_v4sf_p9): Likewise.
12873 (vsx_insert_extract_v4sf_p9_2): Likewise.
12875 2017-07-25 Carl Love <cel@us.ibm.com>
12877 * doc/extend.texi: Update the built-in documentation file for the
12878 existing built-in functions
12879 vector signed char vec_cnttz (vector signed char);
12880 vector unsigned char vec_cnttz (vector unsigned char);
12881 vector signed short vec_cnttz (vector signed short);
12882 vector unsigned short vec_cnttz (vector unsigned short);
12883 vector signed int vec_cnttz (vector signed int);
12884 vector unsigned int vec_cnttz (vector unsigned int);
12885 vector signed long long vec_cnttz (vector signed long long);
12886 vector unsigned long long vec_cnttz (vector unsigned long long);
12888 2017-07-25 Andrew Pinski <apinski@cavium.com>
12890 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
12891 accesses where the use is for the first operand of a BIT_INSERT.
12893 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
12896 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
12897 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
12899 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
12901 * config/i386/gstabs.h: Delete.
12902 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
12904 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
12906 * config/i386/i386.c (ix86_decompose_address): Do not check for
12907 register RTX when looking at index_reg or base_reg.
12908 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
12910 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
12912 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
12913 to update EH info here.
12915 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
12917 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
12919 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
12921 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
12923 2017-07-25 Torsten Duwe <duwe@suse.de>
12925 * common.opt: Introduce -fpatchable-function-entry
12926 command line option, and its variables function_entry_patch_area_size
12927 and function_entry_patch_area_start.
12928 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
12929 including a two-value parser.
12930 * target.def (print_patchable_function_entry): New target hook.
12931 * targhooks.h (default_print_patchable_function_entry): New function.
12932 * targhooks.c (default_print_patchable_function_entry): Likewise.
12933 * toplev.c (process_options): Switch off IPA-RA if
12934 patchable function entries are being generated.
12935 * varasm.c (assemble_start_function): Look at the
12936 patchable-function-entry command line switch and current
12937 function attributes and maybe generate NOP instructions by
12938 calling the print_patchable_function_entry hook.
12939 * doc/extend.texi: Document patchable_function_entry attribute.
12940 * doc/invoke.texi: Document -fpatchable_function_entry
12941 command line option.
12942 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
12944 * doc/tm.texi: Re-generate.
12946 2017-07-25 Jakub Jelinek <jakub@redhat.com>
12949 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
12950 TARGET_AVX512DQ rather than TARGET_AVX512BW.
12952 2017-07-25 Tamar Christina <tamar.christina@arm.com>
12954 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
12956 2017-07-25 Richard Biener <rguenther@suse.de>
12958 PR tree-optimization/81455
12959 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
12960 not walk in cycles when looking for guards.
12962 2017-07-25 Richard Biener <rguenther@suse.de>
12964 PR tree-optimization/81529
12965 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
12966 when optimizing backedge uses.
12968 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
12970 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
12972 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
12973 to dl_section_ref. On AIX, append an expression to subtract
12974 the size of the section length to dl_section_ref.
12976 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
12978 * configure.ac: If any of the config.* scripts fail, exit 1.
12979 * configure: Regenerate.
12981 2017-07-25 Richard Biener <rguenther@suse.de>
12983 PR middle-end/81546
12984 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
12985 of immediate uses, be more verbose on errors.
12987 2017-07-25 Richard Biener <rguenther@suse.de>
12989 PR tree-optimization/81510
12990 * tree-vect-loop.c (vect_is_simple_reduction): When the
12991 reduction stmt is not inside the loop bail out.
12993 2017-07-25 Richard Biener <rguenther@suse.de>
12995 PR tree-optimization/81303
12996 * tree-vect-loop-manip.c (vect_loop_versioning): Build
12997 profitability check against LOOP_VINFO_NITERSM1.
12999 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
13001 * domwalk.c (cmp_bb_postorder): Simplify.
13002 (sort_bbs_postorder): New function. Use it...
13003 (dom_walker::walk): ...here to optimize common cases.
13005 2017-07-25 Martin Liska <mliska@suse.cz>
13008 * ipa-visibility.c (function_and_variable_visibility): Make the
13009 redirection just on target that supports aliasing.
13010 Fix GNU coding style.
13012 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
13015 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
13017 * config/arm/rtems.h: Likewise.
13018 * config/bfin/rtems.h: Likewise.
13019 * config/i386/rtemself.h: Likewise.
13020 * config/lm32/rtems.h: Likewise.
13021 * config/m32c/rtems.h: Likewise.
13022 * config/m68k/rtemself.h: Likewise.
13023 * config/microblaze/rtems.h: Likewise.
13024 * config/mips/rtems.h: Likewise.
13025 * config/moxie/rtems.h: Likewise.
13026 * config/nios2/rtems.h: Likewise.
13027 * config/powerpcspe/rtems.h: Likewise.
13028 * config/rs6000/rtems.h: Likewise.
13029 * config/rtems.h: Likewise.
13030 * config/sh/rtems.h: Likewise.
13031 * config/sh/rtemself.h: Likewise.
13032 * config/sparc/rtemself.h: Likewise.
13034 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
13037 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
13038 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
13039 * tree-ssa-structalias.c (alias_get_name): Same.
13041 2017-07-25 Bin Cheng <bin.cheng@arm.com>
13044 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
13045 instructions if no du chain is found.
13047 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
13049 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
13051 2017-07-25 Richard Biener <rguenther@suse.de>
13053 PR middle-end/81505
13054 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
13057 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
13059 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
13060 upper-regs options.
13061 (ISA_2_7_MASKS_SERVER): Likewise.
13062 (ISA_3_0_MASKS_IEEE): Likewise.
13063 (OTHER_P8_VECTOR_MASKS): Likewise.
13064 (OTHER_VSX_VECTOR_MASKS): Likewise.
13065 (POWERPC_MASKS): Likewise.
13066 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
13067 duplicate list of options.
13068 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
13069 explicit -mupper-regs options.
13070 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
13071 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
13072 alias for -mupper-regs-df.
13073 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
13074 (rs6000_init_hard_regno_mode_ok): Likewise.
13075 (rs6000_option_override_internal): Likewise.
13076 (rs6000_opt_masks): Likewise.
13077 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
13078 options in terms of whether -mvsx or -mpower8-vector was used.
13079 (TARGET_UPPER_REGS_DI): Likewise.
13080 (TARGET_UPPER_REGS_SF): Likewise.
13081 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
13082 -mupper-regs-* options.
13084 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
13086 * passes.c (emergency_dump_function): Print some empty lines and a
13087 header before the RTL dump.
13089 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
13091 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
13093 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
13096 * config/aarch64/aarch64.c (aarch64_classify_symbol):
13097 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
13099 2017-07-24 Carl Love <cel@us.ibm.com>
13101 * config/rs6000/rs6000-c.c: Add support for built-in functions
13102 vector float vec_extract_fp32_from_shorth (vector unsigned short);
13103 vector float vec_extract_fp32_from_shortl (vector unsigned short);
13104 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
13105 vec_extract_fp_from_shortl): Add defines for the two builtins.
13106 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
13107 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
13109 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
13110 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
13111 * doc/extend.texi: Update the built-in documentation file for the
13112 new built-in function.
13114 2017-07-24 Jakub Jelinek <jakub@redhat.com>
13117 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
13119 * doc/generic.texi: Likewise.
13120 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
13121 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
13123 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
13125 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
13126 (aarch64_mls_elt_merge<mode>): Likewise.
13128 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
13130 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
13131 having __cxa_atexit.
13133 2017-07-23 Michael Collison <michael.collison@arm.com>
13135 * config/arm/arm.c (arm_option_override): Deprecate
13136 use of -mstructure-size-boundary.
13137 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
13138 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
13140 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13143 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
13144 Reduce cost estimate for direct moves.
13146 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
13149 * config/i386/i386.c (ix86_option_override_internal): Disable
13150 BMI, BMI2 and TBM instructions for -m16.
13152 2017-07-21 Carl Love <cel@us.ibm.com>
13154 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13155 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
13156 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
13157 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
13158 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
13159 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
13160 VMULOSW): New enum "unspec" values.
13161 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
13162 altivec_vmulosw): New patterns.
13163 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
13164 VMULOSW): Add definitions.
13166 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
13168 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
13169 (qdf24xx): Likewise.
13170 * config/aarch64/aarch64-options-extensions.def (rdma); New.
13171 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
13172 (AARCH64_FL_V8_1): Renumber.
13173 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
13174 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
13175 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
13176 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
13177 rdma to feature modifiers list.
13179 2017-07-21 Yury Gribov <tetra2005@gmail.com>
13181 PR middle-end/56727
13182 * ipa-visibility (function_and_variable_visibility): Convert
13183 recursive PLT call to direct call if appropriate.
13185 2017-07-21 Andrew Pinski <apinski@cavium.com>
13187 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
13188 operand 1 to see if the types precision matches.
13189 * fold-const.c (operand_equal_p): Likewise.
13191 2017-07-21 Richard Biener <rguenther@suse.de>
13193 PR tree-optimization/81303
13194 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
13195 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
13196 (vect_peeling_hash_get_lowest_cost): Adjust.
13197 (vect_enhance_data_refs_alignment): Likewise. Use
13198 vect_get_peeling_costs_all_drs to compute the penalty for no
13199 peeling to match up costs.
13201 2017-07-21 Richard Biener <rguenther@suse.de>
13203 PR tree-optimization/81500
13204 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
13205 we didn't identify a reduction path.
13207 2017-07-21 Tom de Vries <tom@codesourcery.com>
13208 Cesar Philippidis <cesar@codesourcery.com>
13210 PR gcov-profile/81442
13211 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
13214 2017-07-21 Tom de Vries <tom@codesourcery.com>
13217 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
13219 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
13220 nvptx_override_options_after_change.
13222 2017-07-21 Ulrich Drepper <drepper@redhat.com>
13224 * dwarf2out.c (output_file_names): Avoid double testing for
13225 dwarf_version >= 5.
13227 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
13229 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
13231 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
13233 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
13235 (try_crossjump_to_edge): Do not punt on partitioned functions.
13237 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
13239 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
13240 Put all BBs reachable only via paths crossing cold region to cold
13242 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
13244 2016-07-21 Richard Biener <rguenther@suse.de>
13246 PR tree-optimization/81303
13247 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
13248 into account prologue and epilogue iterations when raising
13249 min_profitable_iters to sth at least covering one vector iteration.
13251 2017-07-21 Tamar Christina <tamar.christina@arm.com>
13253 * config/arm/arm.c (arm_test_cpu_arch_dat):
13256 2017-07-20 Nathan Sidwell <nathan@acm.org>
13258 Remove TYPE_METHODS.
13259 * tree.h (TYPE_METHODS): Delete.
13260 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
13261 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
13262 (dbxout_type_methods): Scan TYPE_FIELDS.
13263 (dbxout_type): Don't check TYPE_METHODS here.
13264 * function.c (use_register_for_decl): Always ignore register for
13265 class types when not optimizing.
13266 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
13267 * tree.c (free_lang_data_in_type): Stitch out member functions and
13268 templates from TYPE_FIELDS.
13269 (build_distinct_type_copy, verify_type_variant,
13270 verify_type): Member fns are on TYPE_FIELDS.
13271 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
13272 * tree-pretty-print.c (dump_generic_node): Likewise.
13274 2017-07-20 Jakub Jelinek <jakub@redhat.com>
13277 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
13278 V2TImode and V4TImode.
13279 (ix86_expand_vector_extract): Likewise.
13280 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
13281 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
13282 (ssescalarmode): Handle V4TImode and V2TImode.
13283 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
13284 (*vec_extractv2ti, *vec_extractv4ti): New insns.
13285 (VEXTRACTI128_MODE): New mode iterator.
13286 (splitter for *vec_extractv?ti first element): New.
13287 (VEC_INIT_MODE): New mode iterator.
13288 (vec_init<mode>): Consolidate 3 expanders into one using
13289 VEC_INIT_MODE mode iterator.
13291 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
13293 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
13294 non_spilled_static_chain_regno_p.
13296 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
13298 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
13300 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
13302 * bb-reorder.c (connect_traces): Allow copying of blocks within
13305 2017-07-20 Richard Biener <rguenther@suse.de>
13307 * gimple.h (gimple_phi_result): Add gphi * overload.
13308 (gimple_phi_result_ptr): Likewise.
13309 (gimple_phi_arg): Likewise. Adjust index assert to only
13310 allow actual argument accesses rather than all slots available
13312 (gimple_phi_arg_def): Add gphi * overload.
13313 * tree-phinodes.c (make_phi_node): Initialize only actual
13315 (resize_phi_node): Clear memory not covered by old node,
13316 do not initialize excess argument slots.
13317 (reserve_phi_args_for_new_edge): Initialize new argument slot
13320 2017-07-20 Bin Cheng <bin.cheng@arm.com>
13322 PR tree-optimization/81388
13324 2016-07-21 Bin Cheng <bin.cheng@arm.com>
13326 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
13327 by removing computation of may_be_zero.
13329 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
13330 Tom de Vries <tom@codesourcery.com>
13332 PR middle-end/81030
13333 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
13334 when gimple level profile disagrees with what RTL expander did.
13336 2017-07-20 Richard Biener <rguenther@suse.de>
13338 PR tree-optimization/61171
13339 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
13340 (vect_analyze_stmt): Add slp instance parameter.
13341 (vectorizable_reduction): Likewise.
13342 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
13343 (vect_is_simple_reduction): Deal with chains not detected
13344 as SLP reduction chain, specifically not properly associated
13345 chains containing a mix of plus/minus.
13346 (get_reduction_op): Remove.
13347 (get_initial_defs_for_reduction): Simplify, pass in whether
13348 this is a reduction chain, pass in the SLP node for the PHIs.
13349 (vect_create_epilog_for_reduction): Get the SLP instance as
13351 (vectorizable_reduction): Get the SLP instance as arg.
13352 During analysis remember the SLP node with the PHIs in the
13353 instance. Simplify getting at the vectorized reduction PHIs.
13354 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
13355 through SLP instance.
13356 (vect_slp_analyze_operations): Likewise.
13357 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
13358 (vect_transform_stmt): Likewise.
13360 2017-07-20 Tom de Vries <tom@codesourcery.com>
13362 PR tree-optimization/81489
13363 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
13364 read of phi arg location to before loop that modifies phi.
13366 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13368 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
13371 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
13373 PR middle-end/81331
13374 * except.c (execute): Fix ordering issue.
13376 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
13378 PR rtl-optimization/81423
13379 * combine.c (make_compound_operation_int): Don't try to optimize
13380 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
13382 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
13384 PR rtl-optimization/81423
13385 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
13386 with a constant that is -1 in the truncated to mode.
13388 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
13390 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
13391 (determine_unlikely_bbs): ... here.
13392 * predict.h (propagate_unlikely_bbs_forward): Declare.
13393 * cfgexpand.c (pass_expand::execute): Use it.
13394 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
13396 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
13397 propagate_unlikely_bbs_forward.
13399 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
13401 PR middle-end/81331
13402 * except.c (maybe_add_nop_after_section_switch): New function.
13405 2017-07-19 Tom de Vries <tom@codesourcery.com>
13407 * gimple.h (gimple_phi_set_arg): Make assert more strict.
13409 2017-07-19 Tom de Vries <tom@codesourcery.com>
13411 * gimple.h (gimple_phi_arg): Make assert more strict.
13413 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
13415 * config.gcc (powerpc*-*-*): Add mmintrin.h.
13416 * config/rs6000/mmintrin.h: New file.
13417 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
13419 2017-07-19 Jakub Jelinek <jakub@redhat.com>
13421 PR tree-optimization/81346
13422 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
13424 2017-07-19 Tom de Vries <tom@codesourcery.com>
13426 * config/nvptx/nvptx.md (VECIM): Add V2DI.
13428 2017-07-19 Tom de Vries <tom@codesourcery.com>
13430 * config/nvptx/nvptx-modes.def: Add V2DImode.
13431 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
13432 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
13433 (nvptx_output_mov_insn): Handle lack of mov.b128.
13434 (nvptx_print_operand): Handle 'H' and 'L' codes.
13435 (nvptx_vector_mode_supported): Allow V2DImode.
13436 (nvptx_preferred_simd_mode): New function.
13437 (nvptx_data_alignment): New function.
13438 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
13439 nvptx_preferred_simd_mode.
13440 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
13442 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
13444 2017-07-19 Tom de Vries <tom@codesourcery.com>
13446 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
13447 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
13448 (nvptx_vector_mode_supported): New function. Allow V2SImode.
13449 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
13450 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
13451 (mov<VECIM>_insn): New define_insn.
13452 (define_expand "mov<VECIM>): New define_expand.
13454 2017-07-19 Tom de Vries <tom@codesourcery.com>
13456 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
13458 2017-07-19 Jakub Jelinek <jakub@redhat.com>
13460 PR tree-optimization/81346
13461 * fold-const.h (fold_div_compare, range_check_type): Declare.
13462 * fold-const.c (range_check_type): New function.
13463 (build_range_check): Use range_check_type.
13464 (fold_div_compare): No longer static, rewritten into
13465 a match.pd helper function.
13466 (fold_comparison): Don't call fold_div_compare here.
13467 * match.pd (X / C1 op C2): New optimization using fold_div_compare
13468 as helper function.
13470 2017-07-19 Nathan Sidwell <nathan@acm.org>
13472 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
13473 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
13474 * tree.c (find_decls_types_r, verify_type): Use
13475 TYPE_{MIN,MAX}_VALUE_RAW.
13476 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
13477 (hash_tree): Likewise.
13478 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
13480 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
13483 2017-07-18 Tom de Vries <tom@codesourcery.com>
13485 PR middle-end/81464
13486 * omp-expand.c (expand_omp_for_static_chunk): Handle
13487 equal-argument loop exit phi.
13489 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
13492 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
13493 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
13494 operand 2 predicate.
13495 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
13496 operand 2 predicate.
13497 (ror,rol -> rorx splitters): Use const_int_operand as
13498 operand 2 predicate.
13500 2017-06-18 Richard Biener <rguenther@suse.de>
13502 PR tree-optimization/81410
13503 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
13504 the gap in the ! slp_perm SLP case after each group.
13506 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
13508 PR middle-end/81463
13509 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
13512 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
13514 PR middle-end/81462
13515 * predict.c (set_even_probabilities): Cleanup; do not affect
13516 probabilities that are already known.
13517 (combine_predictions_for_bb): Call even when count is set.
13519 2017-07-18 Nathan Sidwell <nathan@acm.org>
13521 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
13524 2017-07-18 Bin Cheng <bin.cheng@arm.com>
13527 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
13528 optimization for loop niter analysis.
13530 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
13533 * config/avr/avr.c (avr_optimize_casesi): Don't use
13534 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
13536 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
13538 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
13539 body_cost_vec from _vect_peel_extended_info.
13540 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
13541 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
13544 2017-07-18 Bin Cheng <bin.cheng@arm.com>
13546 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
13548 2017-07-18 Richard Biener <rguenther@suse.de>
13550 PR tree-optimization/80620
13551 PR tree-optimization/81403
13552 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
13553 info when re-using a VN table entry.
13555 2017-07-18 Richard Biener <rguenther@suse.de>
13557 PR tree-optimization/81418
13558 * tree-vect-loop.c (vectorizable_reduction): Properly compute
13559 vectype_in. Verify that with lane-reducing reduction operations
13560 we have a single def-use cycle.
13562 2017-07-17 Carl Love <cel@us.ibm.com>
13564 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
13566 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13567 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
13568 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
13569 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
13570 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
13571 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
13572 VMULOSW): New enum "unspec" values.
13573 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
13574 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
13575 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
13576 altivec_vmulosw): New patterns.
13577 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
13578 VMULOSW): Add definitions.
13579 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
13581 * config/alpha/alpha.c: Include predict.h.
13583 2017-07-17 Yury Gribov <tetra2005@gmail.com>
13585 * tree-vrp.c (compare_assert_loc): Fix comparison function
13586 to return predictable results.
13588 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
13590 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
13592 (subdi3): Likewise.
13593 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
13594 * doc/invoke.texi (mexpand-adddi): Update text.
13596 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
13598 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
13599 that also clobbers the CC register. The old expand code is moved
13601 (*arc_clzsi2): ... here.
13602 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
13603 the CC register. The old expand code is moved to ...
13604 (arc_ctzsi2): ... here.
13606 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
13608 * config/arc/arc.opt (mindexed-loads): Use initial value
13609 TARGET_INDEXED_LOADS_DEFAULT.
13610 (mauto-modify-reg): Use initial value
13611 TARGET_AUTO_MODIFY_REG_DEFAULT.
13612 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
13613 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
13614 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
13615 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
13617 2017-07-17 Martin Liska <mliska@suse.cz>
13620 * opts.c (finish_options): Do not allow -fgnu-tm
13621 w/ -fsanitize={kernel-,}address. Say sorry.
13623 2017-07-17 Bin Cheng <bin.cheng@arm.com>
13626 * tree-loop-distribution.c (classify_partition): Only assert on
13627 numer of iterations.
13628 (merge_dep_scc_partitions): Delete prameter. Update function call.
13629 (distribute_loop): Remove code handling loop with unknown niters.
13630 (pass_loop_distribution::execute): Skip loop with unknown niters.
13632 2017-07-17 Bin Cheng <bin.cheng@arm.com>
13635 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
13636 function sort_partitions_by_post_order.
13638 2017-07-17 Bin Cheng <bin.cheng@arm.com>
13640 PR tree-optimization/81374
13641 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
13642 the max index of basic blocks, rather than number of basic blocks.
13644 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
13646 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
13648 (arc_legitimate_pic_operand_p): Likewise.
13649 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
13651 (arc_needs_pcl_p): Likewise.
13652 (arc_legitimate_pc_offset_p): Likewise.
13653 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
13654 function is also used in constrains.md.
13655 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
13656 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
13657 PLUS. Only return true/false in known cases, otherwise assert.
13658 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
13659 is already called in arc_legitimate_constant_p.
13660 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
13662 (LEGITIMATE_PIC_OPERAND_P): Use
13663 arc_raw_symbolic_reference_mentioned_p function.
13664 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
13669 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
13670 Andrew Burgess <andrew.burgess@embecosm.com>
13672 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
13673 (arc_return_address_register): New function.
13674 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
13675 (arc_handle_fndecl_attribute): Add naked attribute.
13676 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
13677 (TARGET_WARN_FUNC_RETURN): Likewise.
13678 (arc_allocate_stack_slots_for_args): New function.
13679 (arc_warn_func_return): Likewise.
13680 (machine_function): Change type fn_type.
13681 (arc_compute_function_type): Consider new naked function type,
13682 change function return type.
13683 (arc_must_save_register): Adapt to handle new
13684 arc_compute_function_type's return type.
13685 (arc_expand_prologue): Likewise.
13686 (arc_expand_epilogue): Likewise.
13687 (arc_return_address_regs): Delete.
13688 (arc_return_address_register): New function.
13689 (arc_epilogue_uses): Use above function.
13690 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
13691 (arc_function_type): Change encoding, add naked type.
13692 (ARC_INTERRUPT_P): Change to handle the new encoding.
13693 (ARC_FAST_INTERRUPT_P): Likewise.
13694 (ARC_NORMAL_P): Define.
13695 (ARC_NAKED_P): Likewise.
13696 (arc_compute_function_type): Delete prototype.
13697 * config/arc/arc.md (in_ret_delay_slot): Use
13698 arc_return_address_register function.
13699 (simple_return): Likewise.
13700 (p_return_i): Likewise.
13702 2017-07-17 Jakub Jelinek <jakub@redhat.com>
13704 PR tree-optimization/81428
13705 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
13706 can't be built for those types.
13708 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
13710 Remove stuff dead since r239246.
13712 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
13713 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
13714 (avr_inform_devices): Remove dead stuff.
13716 2017-07-17 Tamar Christina <tamar.christina@arm.com>
13718 * config/arm/arm_neon.h: Fix softp typo.
13720 2017-07-17 Jakub Jelinek <jakub@redhat.com>
13722 PR tree-optimization/81365
13723 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
13724 aggregate moves onto bb predecessor edges, make sure there are no
13725 loads that could alias the lhs in between the start of bb and the
13728 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
13731 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
13732 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
13733 [LSHIFTRT, outer_code = TRUNCATE]: Same.
13735 2017-07-17 Jakub Jelinek <jakub@redhat.com>
13737 PR tree-optimization/81396
13738 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
13739 (init_symbolic_number): Initialize it to 1.
13740 (perform_symbolic_merge): Add n_ops from both operands into the new
13742 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
13743 without base_addr as useless if they need more than one operation.
13744 (bswap_replace): Handle !bswap case for NULL base_addr.
13746 2017-07-17 Tom de Vries <tom@codesourcery.com>
13749 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
13752 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
13754 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
13755 conditional builtin define __FIX_LEON3FT_B2BST.
13757 2017-07-17 Daniel Cederman <cederman@gaisler.com>
13759 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
13760 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
13763 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
13765 PR rtl-optimization/81424
13766 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
13767 to remove potential trapping from operands if -fnon-call-exceptions.
13769 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13771 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
13772 profile_proability for scalling.
13773 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
13775 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13777 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
13779 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13781 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
13782 fixpoint arithmetics.
13784 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13786 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
13787 fixpoint arithmetics.
13789 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13791 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
13792 fixpoint arithmetics.
13794 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
13796 * profile-count.h (profile_probability::from_reg_br_prob_note,
13797 profile_probability::to_reg_br_prob_note): New functions.
13798 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
13799 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
13800 * predict.c (probability_reliable_p): Update.
13801 (edge_probability_reliable_p): Update.
13802 (br_prob_note_reliable_p): Update.
13803 (invert_br_probabilities): Update.
13804 (add_reg_br_prob_note): New function.
13805 (combine_predictions_for_insn): Update.
13806 * asan.c (asan_clear_shadow): Update.
13807 * cfgbuild.c (compute_outgoing_frequencies): Update.
13808 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
13809 (update_br_prob_note): Update.
13810 (rtl_verify_edges): Update.
13811 (purge_dead_edges): Update.
13812 (fixup_reorder_chain): Update.
13813 * emit-rtl.c (try_split): Update.
13814 * ifcvt.c (cond_exec_process_insns): Update.
13815 (cond_exec_process_if_block): Update.
13816 (dead_or_predicable): Update.
13817 * internal-fn.c (expand_addsub_overflow): Update.
13818 (expand_neg_overflow): Update.
13819 (expand_mul_overflow): Update.
13820 * loop-doloop.c (doloop_modify): Update.
13821 * loop-unroll.c (compare_and_jump_seq): Update.
13822 * optabs.c (emit_cmp_and_jump_insn_1): Update.
13823 * predict.h: Update.
13824 * reorg.c (mostly_true_jump): Update.
13826 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
13827 * config/alpha/alpha.c (emit_unlikely_jump): Update.
13828 * config/arc/arc.c: (emit_unlikely_jump): Update.
13829 * config/arm/arm.c: (emit_unlikely_jump): Update.
13830 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
13831 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
13832 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
13833 (ix86_print_operand): Update.
13834 (ix86_split_fp_branch): Update.
13835 (predict_jump): Update.
13836 * config/ia64/ia64.c (ia64_print_operand): Update.
13837 * config/mmix/mmix.c (mmix_print_operand): Update.
13838 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
13839 (rs6000_expand_split_stack_prologue): Update.
13840 * config/rs6000/rs6000.c: Update.
13841 * config/s390/s390.c (s390_expand_vec_strlen): Update.
13842 (s390_expand_vec_movstr): Update.
13843 (s390_expand_cs_tdsi): Update.
13844 (s390_expand_split_stack_prologue): Update.
13845 * config/sh/sh.c (sh_print_operand): Update.
13846 (expand_cbranchsi4): Update.
13847 (expand_cbranchdi4): Update.
13848 * config/sparc/sparc.c (output_v9branch): Update.
13849 * config/spu/spu.c (get_branch_target): Update.
13850 (ea_load_store_inline): Update.
13851 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
13852 * config/tilepro/tilepro.c: Update.
13854 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
13856 * gimplify.c (mostly_copy_tree_r): Revert latest change.
13857 (gimplify_save_expr): Likewise.
13859 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
13861 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
13863 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
13865 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
13867 * timevar.def (TV_IPA_FNSUMMARY): Define.
13869 2017-07-16 Daniel Cederman <cederman@gaisler.com>
13871 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
13872 to back store errata sensitive sequence from being generated.
13873 (sqrtdf2_fix): Likewise.
13875 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
13877 * tree-ssa-threadupdate.c (compute_path_counts,
13878 update_joiner_offpath_counts): Use profile_probability.
13880 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
13883 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
13885 * config/arm/arm-c.c (arm_cpu_builtins): Define
13886 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
13888 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
13890 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13891 array entries to represent __ieee128 versions of the
13892 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
13893 scalar_extract_sig, and scalar_insert_exp built-in functions.
13894 (altivec_resolve_overloaded_builtin): Add special case handling
13895 for the __builtin_scalar_insert_exp function, as represented by
13896 the P9V_BUILTIN_VEC_VSIEDP constant.
13897 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
13898 exponent support for __ieee128 argument.
13899 (VSESQP): Add scalar extract signature support for __ieee128
13901 (VSTDCNQP): Add scalar test negative support for __ieee128
13903 (VSIEQP): Add scalar insert exponent support for __int128 argument
13904 with __ieee128 result.
13905 (VSIEQPF): Add scalar insert exponent support for __ieee128
13906 argument with __ieee128 result.
13907 (VSTDCQP): Add scalar test data class support for __ieee128
13909 (VSTDCNQP): Add overload support for scalar test negative with
13910 __ieee128 argument.
13911 (VSTDCQP): Add overload support for scalar test data class
13912 __ieee128 argument.
13913 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
13914 UNSPEC_VSX_SXSIGDP.
13915 (UNSPEC_VSX_SIEXPQP): New constant.
13916 (xsxexpqp): New insn for VSX scalar extract exponent quad
13918 (xsxsigqp): New insn for VSX scalar extract significand quad
13920 (xsiexpqpf): New insn for VSX scalar insert exponent quad
13921 precision with floating point argument.
13922 (xststdcqp): New expand for VSX scalar test data class quad
13924 (xststdcnegqp): New expand for VSX scalar test negative quad
13926 (xststdcqp): New insn to match expansions for VSX scalar test data
13927 class quad precision and VSX scalar test negative quad precision.
13928 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
13929 special case operand checking to enforce that second operand of
13930 VSX scalar test data class with quad precision argument is a 7-bit
13932 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
13933 prototypes and descriptions of __ieee128 versions of
13934 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
13935 scalar_test_data_class, and scalar_test_neg built-in functions.
13937 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13939 PR tree-optimization/81162
13940 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
13941 replace a negate with an add.
13943 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
13945 * doc/invoke.texi (arm/-mcpu): Document +crypto.
13947 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
13949 * config/arm/arm-c.c (arm_cpu_builtins): Define
13950 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
13952 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
13954 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
13955 (armv8-r): Set ARM Cortex-R52 as default CPU.
13956 * config/arm/arm-tables.opt: Regenerate.
13957 * config/arm/arm-tune.md: Regenerate.
13958 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
13960 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
13961 extension for -mcpu=cortex-r52.
13963 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
13965 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
13966 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
13967 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
13968 (fp-armv8): Define it as FP_ARMv8 only.
13969 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
13970 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
13972 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
13973 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
13974 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
13975 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
13976 than TARGET_FPU_ARMV8.
13977 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
13978 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
13979 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
13981 * config/arm/neon.md (neon_vrint): Likewise.
13982 (neon_vcvt): Likewise.
13983 (neon_<fmaxmin_op><mode>): Likewise.
13984 (<fmaxmin><mode>3): Likewise.
13985 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
13986 * config/arm/predicates.md (arm_cond_move_operator): Check against
13987 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
13989 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
13991 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
13992 to top of function.
13994 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13996 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
13997 loop in comment with memset.
13999 2017-07-14 Martin Liska <mliska@suse.cz>
14001 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
14002 * dwarf2out.c (is_java): Remove the function.
14003 (output_pubname): Remove usage of the function.
14004 (lower_bound_default): Remove usage of DW_LANG_Java.
14005 (gen_compile_unit_die): Likewise.
14006 * gcc.c: Remove compiler defaults for .java and .zip files.
14007 * gimple-expr.c (remove_suffix): Change as there's no longer
14008 extension than 4-letter one.
14009 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
14010 (gimplify_save_expr): Likewise.
14011 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
14012 as it's possible even for other languages than Java.
14013 * langhooks.h (struct lang_hooks): Remove Java from a comment.
14014 * lto-opts.c (lto_write_options): Remove reference to Java.
14015 * opts.c (strip_off_ending): Update file extension handling.
14016 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
14017 * tree-eh.c (lower_resx): Likewise.
14018 * tree.c (free_lang_data_in_type): Remove dead code.
14019 (find_decls_types_r): Likewise.
14020 (build_common_builtin_nodes): Remove Java from a comment.
14021 (verify_type): Remove dead code.
14022 * varasm.c (assemble_external): Remove Java from a comment.
14024 2017-07-14 Martin Liska <mliska@suse.cz>
14026 * opts.c (finish_options): Add quotes.
14027 (common_handle_option): Likewise.
14029 2017-07-14 Martin Liska <mliska@suse.cz>
14031 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
14032 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
14033 Remove N_SO_PASCAL.
14034 * dwarf2out.c (lower_bound_default): Do not handle
14036 (gen_compile_unit_die): Likewise.
14037 * gcc.c: Remove default extension binding for GNU Pascal.
14038 * stmt.c: Remove Pascal language from a comment.
14039 * xcoffout.c: Likewise.
14041 2017-07-13 David Malcolm <dmalcolm@redhat.com>
14044 * diagnostic-show-locus.c (fixit_cmp): New function.
14045 (layout::layout): Sort m_fixit_hints.
14046 (column_range::column_range): Assert that the values are valid.
14047 (struct char_span): New struct.
14048 (correction::overwrite): New method.
14049 (struct source_line): New struct.
14050 (line_corrections::add_hint): Add assertions. Reimplement memcpy
14051 calls in terms of classes source_line and char_span, and
14052 correction::overwrite.
14053 (selftest::test_overlapped_fixit_printing_2): New function.
14054 (selftest::diagnostic_show_locus_c_tests): Call it.
14056 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
14058 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
14059 early if there is no lhs.
14061 2017-07-13 Martin Liska <mliska@suse.cz>
14063 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
14064 (gen_reference_type_die): Likewise.
14065 * stor-layout.c: Remove Pascal-related comment.
14067 2017-07-13 Martin Liska <mliska@suse.cz>
14069 * opts.c (finish_options): Add quotes to error messages.
14070 (parse_sanitizer_options): Likewise.
14072 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
14074 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
14076 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
14078 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
14080 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
14082 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
14084 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
14086 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
14089 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
14090 provides the hardware capability bits, define the macro
14091 __BUILTIN_CPU_SUPPORTS__.
14092 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
14093 if GLIBC does not provide the hardware capability bits. Add a
14094 gcc_unreachable call if the built-in cpu function is neither
14095 __builtin_cpu_is nor __builtin_cpu_supports.
14096 (rs6000_get_function_versions_dispatcher): Change the warning
14097 that an old GLIBC is used which does not export the capability
14098 bits to be an error.
14099 * doc/extend.texi (target_clones attribute): Document the
14100 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
14101 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
14102 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
14103 the macros defined by GCC if the newer GLIBC is available.
14105 2017-07-12 Jeff Law <law@redhat.com>
14107 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
14108 remaining includes slightly.
14109 * config/riscv/riscv-builtins.c: Include profile-count.h.
14111 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
14114 * config/avr/avr.c (avr_set_current_function): In diagnostic
14115 messages: Quote keywords and (parts of) identifiers.
14116 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
14119 2017-07-12 Carl Love <cel@us.ibm.com>
14121 * config/rs6000/rs6000-c.c: Add support for built-in functions
14122 vector bool char vec_revb (vector bool char);
14123 vector bool short vec_revb (vector short char);
14124 vector bool int vec_revb (vector bool int);
14125 vector bool long long vec_revb (vector bool long long);
14126 * doc/extend.texi: Update the built-in documentation file for the
14127 new built-in functions.
14129 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14131 * config/s390/s390.md: Remove movcc splitter.
14133 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14135 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
14136 load/store on condition.
14138 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
14141 * config/avr/avr.c (avr_encode_section_info)
14142 [progmem && !TREE_READONLY]: Error if progmem object needs
14145 2017-07-11 Michael Collison <michael.collison@arm.com>
14147 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
14150 2017-07-11 Carl Love <cel@us.ibm.com>
14152 * config/rs6000/rs6000-c.c: Add support for builtins
14153 vector unsigned int vec_parity_lsbb (vector signed int);
14154 vector unsigned int vec_parity_lsbb (vector unsigned int);
14155 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
14156 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
14157 vector unsigned long long vec_parity_lsbb (vector signed long long);
14158 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
14159 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
14160 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
14161 * doc/extend.texi: Update the built-in documentation file for the
14162 new built-in functions.
14164 2017-07-11 David Malcolm <dmalcolm@redhat.com>
14166 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
14167 (layout::m_primary_loc): New field.
14168 (layout::layout): Initialize new field. Move location filtering
14169 logic from here to...
14170 (layout::maybe_add_location_range): ...this new method. Add
14171 support for filtering to just the lines already specified by other
14173 (layout::will_show_line_p): New method.
14174 (gcc_rich_location::add_location_if_nearby): New method.
14175 (selftest::test_add_location_if_nearby): New test function.
14176 (selftest::diagnostic_show_locus_c_tests): Call it.
14177 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
14180 2017-07-11 Tom de Vries <tom@codesourcery.com>
14182 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
14183 (bb_first_real_insn): New function.
14184 (nvptx_single): Add extra initialization of broadcasted condition
14187 2017-07-11 Nathan Sidwell <nathan@acm.org>
14189 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
14191 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
14193 * doc/extend.texi (AVR Function Attributes): Remove weblink to
14194 Binutils doc as TEXI will mess them up.
14195 * doc/invoke.texi (AVR Options): Same here.
14197 2017-07-11 Daniel Cederman <cederman@gaisler.com>
14199 * config/sparc/sparc.opt (mfix-ut700): New option.
14200 (mfix-gr712rc): Likewise.
14201 (sparc_fix_b2bst): New variable.
14202 * doc/invoke.texi (SPARC options): Document them.
14203 (ARM options): Fix warnings.
14204 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
14205 instructions to prevent sequences that can trigger the store-store
14206 errata for certain LEON3FT processors.
14207 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
14208 (sparc_option_override): Set sparc_fix_b2bst appropriately.
14209 * config/sparc/sparc.md (fix_b2bst): New attribute.
14210 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
14212 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
14215 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
14217 (*rsqrtsf2_sse): Ditto.
14219 (div<mode>3): Macroize insn from divdf3 and divsf3
14220 using MODEF mode iterator.
14222 2017-07-10 Martin Sebor <msebor@redhat.com>
14224 PR tree-optimization/80397
14225 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
14226 instead of testing for equality to INTEGER_TYPE.
14228 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
14230 * config.gcc: Remove uclibc from arc target spec.
14232 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
14234 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
14236 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
14239 * lto-wrapper.c (remove_option): New function.
14240 (merge_and_complain): Merge PIC/PIE options more realistically.
14242 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
14244 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
14248 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
14249 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
14250 * config.in: Regenerate.
14251 * configure: Regenerate.
14252 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
14253 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
14254 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
14255 (TARGET_GASISR_PROLOGUES): ...target mask.
14256 * common/config/avr/avr-common.c
14257 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
14258 Set -mgas-isr-prologues.
14259 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
14260 INSERT_PASS_BEFORE for it.
14261 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
14262 * config/avr/avr.c (avr_option_override)
14263 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
14264 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
14265 (avr_attribute_table) <no_gccisr>: Add new function attribute.
14266 (avr_set_current_function) <is_no_gccisr>: Init machine field.
14267 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
14269 (make_avr_pass_pre_proep): New function.
14270 (emit_push_sfr) <treg>: Add argument to function and use it
14271 instead of TMP_REG.
14272 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
14273 and set machine->gasisr.yes.
14274 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
14275 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
14276 __gcc_isr.n_pushed to .L__stack_usage.
14277 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
14278 (avr_asm_final_postscan_insn): ...this new static function.
14279 * config/avr/avr.h (machine_function)
14280 <is_no_gccisr, use_L__stack_usage>: New fields.
14281 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
14282 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
14283 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
14284 (gasisr, *gasisr): New expander and insn.
14285 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
14286 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
14287 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
14289 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
14291 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
14294 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
14296 Move jump-tables out of .text again.
14299 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
14300 (ASM_OUTPUT_ADDR_VEC): New function.
14301 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
14302 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
14303 INSN_ADDRESSes as asm comment.
14304 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
14305 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
14306 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
14307 * config/avr/avr.md (*tablejump): Adjust comment.
14308 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
14309 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
14311 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
14312 (avr_output_addr_vec): New proto.
14313 (avr_log_t) <insn_addresses>: New field.
14315 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
14318 * config/i386/i386.c (ix86_function_arg_advance): Set
14319 outgoing_args_on_stack to true if there are outgoing arguments
14321 (ix86_function_arg): Likewise.
14322 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
14323 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
14324 * config/i386/i386.h (machine_function): Add
14325 outgoing_args_on_stack.
14327 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
14329 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
14330 supporting pthreds.
14331 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
14333 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
14335 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
14336 (REAL_H): Remove $(MACHMODE_H).
14337 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
14339 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
14340 $(MACHMODE_H) and double-int.h.
14341 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
14343 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
14346 2017-07-07 Andrew Pinski <apinski@cavium.com>
14348 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
14349 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
14351 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
14353 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
14354 Add warning if GCC was not configured to link against a GLIBC that
14355 exports the hardware capability bits.
14356 (make_resolver_func): Make resolver function private and not a
14357 COMDAT function. Create the name with clone_function_name instead
14358 of make_unique_name.
14361 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
14362 correct operand in doing the split.
14364 2017-07-07 Carl Love <cel@us.ibm.com>
14366 * config/rs6000/rs6000-c: Add support for built-in function
14367 vector unsigned short vec_pack_to_short_fp32 (vector float,
14369 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
14370 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
14371 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
14372 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
14373 (convert_4f32_8i16): Add define_expand.
14374 * doc/extend.texi: Update the built-in documentation file for the
14375 new built-in function.
14377 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14379 * config/sparc/m8.md: New file.
14380 * config/sparc/sparc.md: Include m8.md.
14382 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14384 * config/sparc/sparc.opt: New option -mvis4b.
14385 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
14386 (sparc_option_override): Handle VIS4B.
14387 (enum sparc_builtins): Define
14388 SPARC_BUILTIN_DICTUNPACK{8,16,32},
14389 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
14390 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
14391 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
14392 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
14393 (check_constant_argument): New function.
14394 (sparc_vis_init_builtins): Define builtins
14395 __builtin_vis_dictunpack{8,16,32},
14396 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
14397 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
14398 __builtin_vis_fpcmpde{8,16,32}shl and
14399 __builtin_vis_fpcmpur{8,16,32}shl.
14400 (sparc_expand_builtin): Check that the constant operands to
14401 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
14402 constant and in range.
14403 * config/sparc/sparc-c.c (sparc_target_macros): Handle
14405 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
14406 (SPARC_IMM5_P): Likewise.
14407 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
14408 (enabled): Handle vis4b.
14409 (UNSPEC_DICTUNPACK): New unspec.
14410 (UNSPEC_FPCMPSHL): Likewise.
14411 (UNSPEC_FPUCMPSHL): Likewise.
14412 (UNSPEC_FPCMPDESHL): Likewise.
14413 (UNSPEC_FPCMPURSHL): Likewise.
14414 (cpu_feature): New CPU feature `vis4b'.
14415 (dictunpack{8,16,32}): New insns.
14416 (FPCSMODE): New mode iterator.
14417 (fpcscond): New code iterator.
14418 (fpcsucond): Likewise.
14419 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
14420 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
14421 (fpcmpde{8,16,32}{si,di}shl): Likewise.
14422 (fpcmpur{8,16,32}{si,di}shl): Likewise.
14423 * config/sparc/constraints.md: Define constraints `q' for unsigned
14424 2-bit integer constants and `t' for unsigned 5-bit integer
14426 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
14428 (imm5_operand_dictunpack16): Likewise.
14429 (imm5_operand_dictunpack32): Likewise.
14430 (imm2_operand): Likewise.
14431 * doc/invoke.texi (SPARC Options): Document -mvis4b.
14432 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
14433 ditunpack* and fpcmp*shl builtins.
14435 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14437 * config.gcc: Handle m8 in --with-{cpu,tune} options.
14438 * config.in: Add HAVE_AS_SPARC6 define.
14439 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
14441 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
14443 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
14444 (CPP_CPU_SPEC): Handle m8.
14445 (ASM_CPU_SPEC): Likewise.
14446 * config/sparc/sparc-opts.h (enum processor_type): Add
14448 * config/sparc/sparc.c (m8_costs): New struct.
14449 (sparc_option_override): Handle TARGET_CPU_m8.
14450 (sparc32_initialize_trampoline): Likewise.
14451 (sparc64_initialize_trampoline): Likewise.
14452 (sparc_issue_rate): Likewise.
14453 (sparc_register_move_cost): Likewise.
14454 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
14455 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
14456 (ASM_CPU64_DEFAULT_SPEC): Likewise.
14457 (CPP_CPU_SPEC): Handle M8.
14458 (ASM_CPU_SPEC): Likewise.
14459 (AS_M8_FLAG): Define.
14460 * config/sparc/sparc.md: Add m8 to the cpu attribute.
14461 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
14462 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
14464 * configure: Regenerate.
14465 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
14468 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14470 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
14472 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
14473 ("*movdi_insn_sp32"): Do not set v3pipe.
14474 ("*movsi_insn"): Likewise.
14475 ("*movdi_insn_sp64"): Likewise.
14476 ("*movsf_insn"): Likewise.
14477 ("*movdf_insn_sp32"): Likewise.
14478 ("*movdf_insn_sp64"): Likewise.
14479 ("*zero_extendsidi2_insn_sp64"): Likewise.
14480 ("*sign_extendsidi2_insn"): Likewise.
14481 ("*mov<VM32:mode>_insn"): Likewise.
14482 ("*mov<VM64:mode>_insn_sp64"): Likewise.
14483 ("*mov<VM64:mode>_insn_sp32"): Likewise.
14484 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
14485 ("<vlop:code><VL:mode>3"): Likewise.
14486 ("*not_<vlop:code><VL:mode>3"): Likewise.
14487 ("*nand<VL:mode>_vis"): Likewise.
14488 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
14489 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
14490 ("one_cmpl<VL:mode>2"): Likewise.
14491 ("faligndata<VM64:mode>_vis"): Likewise.
14492 ("alignaddrsi_vis"): Likewise.
14493 ("alignaddrdi_vis"): Likweise.
14494 ("alignaddrlsi_vis"): Likewise.
14495 ("alignaddrldi_vis"): Likewise.
14496 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
14497 ("bmaskdi_vis"): Likewise.
14498 ("bmasksi_vis"): Likewise.
14499 ("bshuffle<VM64:mode>_vis"): Likewise.
14500 ("cmask8<P:mode>_vis"): Likewise.
14501 ("cmask16<P:mode>_vis"): Likewise.
14502 ("cmask32<P:mode>_vis"): Likewise.
14503 ("pdistn<P:mode>_vis"): Likewise.
14504 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
14506 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14508 * config/sparc/sparc.md ("subtype"): New insn attribute.
14509 ("*wrgsr_sp64"): Set insn subtype.
14510 ("*rdgsr_sp64"): Likewise.
14511 ("alignaddrsi_vis"): Likewise.
14512 ("alignaddrdi_vis"): Likewise.
14513 ("alignaddrlsi_vis"): Likewise.
14514 ("alignaddrldi_vis"): Likewise.
14515 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
14516 ("fexpand_vis"): Likewise.
14517 ("fpmerge_vis"): Likewise.
14518 ("faligndata<VM64:mode>_vis"): Likewise.
14519 ("bshuffle<VM64:mode>_vis"): Likewise.
14520 ("cmask8<P:mode>_vis"): Likewise.
14521 ("cmask16<P:mode>_vis"): Likewise.
14522 ("cmask32<P:mode>_vis"): Likewise.
14523 ("fchksm16_vis"): Likewise.
14524 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
14525 ("fmean16_vis"): Likewise.
14526 ("fp<plusminus_insn>64_vis"): Likewise.
14527 ("<plusminus_insn>v8qi3"): Likewise.
14528 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
14529 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
14530 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
14531 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
14532 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
14533 ("*movqi_insn"): Likewise.
14534 ("*movhi_insn"): Likewise.
14535 ("*movsi_insn"): Likewise.
14536 ("movsi_pic_gotdata_op"): Likewise.
14537 ("*movdi_insn_sp32"): Likewise.
14538 ("*movdi_insn_sp64"): Likewise.
14539 ("movdi_pic_gotdata_op"): Likewise.
14540 ("*movsf_insn"): Likewise.
14541 ("*movdf_insn_sp32"): Likewise.
14542 ("*movdf_insn_sp64"): Likewise.
14543 ("*zero_extendhisi2_insn"): Likewise.
14544 ("*zero_extendqihi2_insn"): Likewise.
14545 ("*zero_extendqisi2_insn"): Likewise.
14546 ("*zero_extendqidi2_insn"): Likewise.
14547 ("*zero_extendhidi2_insn"): Likewise.
14548 ("*zero_extendsidi2_insn_sp64"): Likewise.
14549 ("ldfsr"): Likewise.
14550 ("prefetch_64"): Likewise.
14551 ("prefetch_32"): Likewise.
14552 ("tie_ld32"): Likewise.
14553 ("tie_ld64"): Likewise.
14554 ("*tldo_ldub_sp32"): Likewise.
14555 ("*tldo_ldub1_sp32"): Likewise.
14556 ("*tldo_ldub2_sp32"): Likewise.
14557 ("*tldo_ldub_sp64"): Likewise.
14558 ("*tldo_ldub1_sp64"): Likewise.
14559 ("*tldo_ldub2_sp64"): Likewise.
14560 ("*tldo_ldub3_sp64"): Likewise.
14561 ("*tldo_lduh_sp32"): Likewise.
14562 ("*tldo_lduh1_sp32"): Likewise.
14563 ("*tldo_lduh_sp64"): Likewise.
14564 ("*tldo_lduh1_sp64"): Likewise.
14565 ("*tldo_lduh2_sp64"): Likewise.
14566 ("*tldo_lduw_sp32"): Likewise.
14567 ("*tldo_lduw_sp64"): Likewise.
14568 ("*tldo_lduw1_sp64"): Likewise.
14569 ("*tldo_ldx_sp64"): Likewise.
14570 ("*mov<VM32:mode>_insn"): Likewise.
14571 ("*mov<VM64:mode>_insn_sp64"): Likewise.
14572 ("*mov<VM64:mode>_insn_sp32"): Likewise.
14574 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14576 * config/sparc/sparc.md ("type"): New insn type viscmp.
14577 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
14579 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
14580 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
14581 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
14582 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
14584 ("n7_vis_logical_11cycle"): Likewise.
14585 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
14586 * config/sparc/niagara2.md ("niag3_vis": Likewise.
14587 * config/sparc/niagara.md ("niag_vis"): Likewise.
14588 * config/sparc/ultra3.md ("us3_fga"): Likewise.
14589 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
14591 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
14593 * config/sparc/sparc.md: New instruction type `bmask'.
14594 (bmaskdi_vis): Use the `bmask' type.
14595 (bmasksi_vis): Likewise.
14596 * config/sparc/ultra3.md (us3_array): Likewise.
14597 * config/sparc/niagara7.md (n7_array): Likewise.
14598 * config/sparc/niagara4.md (n4_array): Likewise.
14599 * config/sparc/niagara2.md (niag2_vis): Likewise.
14600 (niag3_vis): Likewise.
14601 * config/sparc/niagara.md (niag_vis): Likewise.
14603 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
14605 * ipa-comdats.c: Remove optimize check from gate.
14606 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
14607 for functions not optimized.
14608 (ipa_fn_summary_read): Skip optimize check.
14609 (ipa_fn_summary_write): Likewise.
14610 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
14612 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
14614 (can_inline_edge_p): Check flag_pcc_struct_return for match.
14615 (check_callers): Give up on caller which is not optimized.
14616 (inline_small_functions): Likewise.
14617 (ipa_inline): Do not give up when not optimizing.
14618 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
14619 away unoptimizes cdtors.
14620 (whole_program_function_and_variable_visibility): Do
14621 ipa_discover_readonly_nonaddressable_vars in LTO mode.
14622 * ipa.c (process_references): Do not check optimize.
14623 (symbol_table::remove_unreachable_nodes): Update optimize check.
14624 (set_writeonly_bit): Update optimize check.
14625 (pass_ipa_cdtor_merge::gate): Do not check optimize.
14626 (pass_ipa_single_use::gate): Remove.
14628 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
14630 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
14631 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
14632 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
14633 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
14634 permute_load, permute_store, adjust_extract, adjust_splat,
14635 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
14636 replace_swap_with_copy, dump_swap_insn_table,
14637 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
14638 recombine_lvx_pattern, recombine_stvx_pattern,
14639 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
14640 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
14641 to file rs6000-p8swap.c.
14642 * config/rs6000/rs6000-p8swap.c: New file.
14643 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
14644 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
14645 and rs6000*-*-* targets.
14647 2017-07-06 David Malcolm <dmalcolm@redhat.com>
14649 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
14651 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
14653 * lto-wrapper.c (merge_and_complain): Do not merge
14654 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
14655 fsigned_zeros, ftrapping_math, fwrapv.
14656 (append_compiler_options): Do not track these options.
14657 (append_linker_options): Likewie
14659 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
14661 * cgraphunit.c (cgraph_node::finalize_function): When
14662 !flag_toplevel_reorde set no_reorder flag.
14663 (varpool_node::finalize_decl): Likewise.
14664 (symbol_table::compile): Drop no toplevel reorder path.
14666 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
14668 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
14669 edges; zero probability is not better than uninitialized.
14671 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
14673 * asan.h (asan_sanitize_allocas_p): Declare.
14674 * asan.c (asan_sanitize_allocas_p): New function.
14675 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
14676 (handle_builtin_alloca): Likewise.
14677 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
14678 if !asan_sanitize_allocas_p.
14679 * params.def (asan-instrument-allocas): Add new option.
14680 * params.h (ASAN_PROTECT_ALLOCAS): Define.
14681 * opts.c (common_handle_option): Disable allocas sanitization for
14684 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
14686 * asan.c: Include gimple-fold.h.
14687 (get_last_alloca_addr): New function.
14688 (handle_builtin_stackrestore): Likewise.
14689 (handle_builtin_alloca): Likewise.
14690 (asan_emit_allocas_unpoison): Likewise.
14691 (get_mem_refs_of_builtin_call): Add new parameter, remove const
14692 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
14693 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
14694 (instrument_builtin_call): Pass gimple iterator to
14695 get_mem_refs_of_builtin_call.
14696 (last_alloca_addr): New global.
14697 * asan.h (asan_emit_allocas_unpoison): Declare.
14698 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
14699 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
14700 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
14701 if function calls alloca.
14702 * gimple-fold.c (replace_call_with_value): Remove static keyword.
14703 * gimple-fold.h (replace_call_with_value): Declare.
14704 * internal-fn.c: Include asan.h.
14705 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
14706 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
14708 2017-07-06 David Malcolm <dmalcolm@redhat.com>
14710 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
14711 (C_SELFTEST_FLAGS): New.
14712 (CPP_SELFTEST_FLAGS): New.
14713 (SELFTEST_DEPS): New, from deps of s-selftest.
14714 (C_SELFTEST_DEPS): New, from deps of s-selftest.
14715 (CPP_SELFTEST_DEPS): New.
14716 (selftest): Add dependency on s-selftest-c++.
14717 (s-selftest): Rename to...
14718 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
14719 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
14720 than SELFTEST_FLAGS.
14721 (selftest-gdb): Rename to...
14722 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
14724 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
14725 (selftest-valgrind): Rename to...
14726 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
14728 (selftest-valgrind): Reintroduce as an alias for
14729 selftest-c-valgrind.
14730 (s-selftest-c++): New.
14731 (selftest-c++-gdb): New.
14732 (selftest-c++-valgrind): New.
14734 2017-07-06 Olivier Hainque <hainque@adacore.com>
14736 * gcc.c (process_command): When deciding if undefined variables
14737 should be ignored when processing specs, accept "gcc -v" as well.
14739 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
14741 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
14742 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
14744 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
14746 * config/arm/arm-cpus.in (armv8-r): Add new entry.
14747 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
14748 * config/arm/arm-tables.opt: Regenerate.
14749 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
14751 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
14753 2017-07-06 Carl Love <cel@us.ibm.com>
14755 * ChangeLog: Clean up from mid air collision
14757 2017-07-06 Carl Love <cel@us.ibm.com>
14759 * config/rs6000/rs6000-c.c: Add support for built-in functions
14760 vector signed int vec_subc (vector signed int, vector signed int);
14761 vector signed __int128 vec_subc (vector signed __int128,
14762 vector signed __int128);
14763 vector unsigned __int128 vec_subc (vector unsigned __int128,
14764 vector unsigned __int128);
14765 vector signed int vec_sube (vector signed int, vector signed int,
14766 vector signed int);
14767 vector unsigned int vec_sube (vector unsigned int,
14768 vector unsigned int,
14769 vector unsigned int);
14770 vector signed __int128 vec_sube (vector signed __int128,
14771 vector signed __int128,
14772 vector signed__int128);
14773 vector unsigned __int128 vec_sube (vector unsigned __int128,
14774 vector unsigned __int128,
14775 vector unsigned __int128);
14776 vector signed int vec_subec (vector signed int, vector signed int,
14777 vector signed int);
14778 vector unsigned int vec_subec (vector unsigned int,
14779 vector unsigned int,
14780 vector unsigned int);
14781 vector signed __int128 vec_subec (vector signed __int128,
14782 vector signed __int128,
14783 vector signed__int128);
14784 vector unsigned __int128 vec_subec (vector unsigned __int128,
14785 vector unsigned __int128,
14786 vector unsigned __int128);
14787 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
14788 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
14789 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
14790 BU_ALTIVEC_OVERLOAD_X definitions.
14791 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
14792 * doc/extend.texi: Update the built-in documentation file for the new
14793 built-in functions.
14795 2017-07-06 David Malcolm <dmalcolm@redhat.com>
14798 * diagnostic-show-locus.c (layout::layout): Use start and finish
14799 spelling location for the start and finish of each range.
14800 * genmatch.c (linemap_client_expand_location_to_spelling_point):
14801 Add unused aspect param.
14802 * input.c (expand_location_1): Add "aspect" param, and use it
14803 to access the correct part of the location.
14804 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
14806 (expand_location_to_spelling_point): Likewise.
14807 (linemap_client_expand_location_to_spelling_point): Add "aspect"
14808 param, and pass it to expand_location_1.
14810 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
14812 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
14813 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
14814 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
14815 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
14816 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
14817 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
14818 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
14819 _mm_maskz_getmant_ss): New intrinsics.
14820 (__builtin_ia32_getexpss128_mask): Changed to ...
14821 __builtin_ia32_getexpss128_round ... this.
14822 (__builtin_ia32_getexpsd128_mask): Changed to ...
14823 __builtin_ia32_getexpsd128_round ... this.
14824 * config/i386/i386-builtin-types.def
14825 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
14826 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
14827 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
14828 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
14829 __builtin_ia32_getmantss_mask_round): New builtins.
14830 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
14831 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
14832 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
14833 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
14834 * config/i386/sse.md
14835 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
14836 avx512f_sgetexp<mode><mask_scalar_name>
14837 <round_saeonly_scalar_name> ... this.
14838 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
14839 %0, %1, %2<round_saeonly_op3>}): Changed to ...
14840 vgetexp<ssescalarmodesuffix>
14841 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
14842 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
14843 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
14844 avx512f_vgetmant<mode><mask_scalar_name>
14845 <round_saeonly_scalar_name> ... this.
14846 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
14847 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
14848 vgetmant<ssescalarmodesuffix>
14849 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
14850 %0<mask_scalar_operand4>, %1, %2
14851 <round_saeonly_scalar_mask_op4>, %3} ... this.
14852 * config/i386/subst.md (mask_scalar_operand4,
14853 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
14854 round_saeonly_scalar_nimm_predicate): New subst attributes.
14856 2017-07-06 Julia Koval <julia.koval@intel.com>
14858 * config/i386/i386.c (ix86_erase_embedded_rounding):
14859 Remove code for old rounding pattern.
14861 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
14863 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
14865 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
14867 * doc/sourcebuild.texi (Test Directives, Variants of
14868 dg-require-support): Add documentation for dg-require-stack-check.
14870 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
14872 * config/i386/subst.md (mask_scalar, round_scalar,
14873 round_saeonly_scalar): New meta-templates.
14874 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
14875 round_scalar_mask_operand3, round_scalar_mask_op3,
14876 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
14877 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
14878 round_saeonly_scalar_constraint,
14879 round_saeonly_scalar_prefix): New subst attribute.
14880 * config/i386/sse.md
14881 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
14882 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
14883 <round_scalar_name> ... this.
14884 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
14885 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
14886 <round_scalar_name> ... this.
14887 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
14888 <sse>_vm<code><mode>3<mask_scalar_name>
14889 <round_saeonly_scalar_name> ... this.
14890 (v<plusminus_mnemonic><ssescalarmodesuffix>
14891 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
14892 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
14893 v<plusminus_mnemonic><ssescalarmodesuffix>
14894 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
14895 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
14896 (v<multdiv_mnemonic><ssescalarmodesuffix>
14897 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
14898 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
14899 v<multdiv_mnemonic><ssescalarmodesuffix>
14900 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
14901 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
14902 (v<maxmin_float><ssescalarmodesuffix>
14903 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
14904 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
14905 v<maxmin_float><ssescalarmodesuffix>
14906 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
14907 %0<mask_scalar_operand3>, %1, %<iptr>2
14908 <round_saeonly_scalar_mask_op3>} ... this.
14910 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
14912 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
14913 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
14915 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
14916 Alan Hayward <alan.hayward@arm.com>
14917 David Sherwood <david.sherwood@arm.com>
14919 * combine.c (simplify_if_then_else): Remove "enum" before
14921 * compare-elim.c (can_eliminate_compare): Likewise.
14922 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
14924 (aarch64_lookup_simd_builtin_type): Likewise.
14925 (aarch64_simd_builtin_type): Likewise.
14926 (aarch64_init_simd_builtin_types): Likewise.
14927 (aarch64_simd_expand_args): Likewise.
14928 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
14930 (aarch64_reverse_mask): Likewise.
14931 (aarch64_simd_emit_reg_reg_move): Likewise.
14932 (aarch64_gen_adjusted_ldpstp): Likewise.
14933 (aarch64_ccmp_mode_to_code): Likewise.
14934 (aarch64_operands_ok_for_ldpstp): Likewise.
14935 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
14936 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
14938 (aarch64_min_divisions_for_recip_mul): Likewise.
14939 (aarch64_reassociation_width): Likewise.
14940 (aarch64_get_condition_code_1): Likewise.
14941 (aarch64_simd_emit_reg_reg_move): Likewise.
14942 (aarch64_simd_attr_length_rglist): Likewise.
14943 (aarch64_reverse_mask): Likewise.
14944 (aarch64_operands_ok_for_ldpstp): Likewise.
14945 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
14946 (aarch64_gen_adjusted_ldpstp): Likewise.
14947 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
14949 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
14950 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
14951 (arm_lookup_simd_builtin_type): Likewise.
14952 (arm_simd_builtin_type): Likewise.
14953 (arm_init_simd_builtin_types): Likewise.
14954 (arm_expand_builtin_args): Likewise.
14955 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
14956 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
14957 (ft32_setup_incoming_varargs): Likewise.
14958 (ft32_function_arg): Likewise.
14959 (ft32_function_arg_advance): Likewise.
14960 (ft32_pass_by_reference): Likewise.
14961 (ft32_arg_partial_bytes): Likewise.
14962 (ft32_valid_pointer_mode): Likewise.
14963 (ft32_addr_space_pointer_mode): Likewise.
14964 (ft32_addr_space_legitimate_address_p): Likewise.
14965 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
14967 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
14968 (ix86_emit_outlined_ms2sysv_restore): Likewise.
14969 (iamcu_alignment): Likewise.
14970 (canonicalize_vector_int_perm): Likewise.
14971 (ix86_noce_conversion_profitable_p): Likewise.
14972 (ix86_mpx_bound_mode): Likewise.
14973 (ix86_operands_ok_for_move_multiple): Likewise.
14974 * config/microblaze/microblaze-protos.h
14975 (microblaze_expand_conditional_branch_reg): Likewise.
14976 * config/microblaze/microblaze.c
14977 (microblaze_expand_conditional_branch_reg): Likewise.
14978 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
14980 (rs6000_reassociation_width): Likewise.
14981 (rs6000_invalid_binary_op): Likewise.
14982 (fusion_p9_p): Likewise.
14983 (emit_fusion_p9_load): Likewise.
14984 (emit_fusion_p9_store): Likewise.
14985 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
14987 (riscv_hard_regno_mode_ok_p): Likewise.
14988 (riscv_address_insns): Likewise.
14989 (riscv_split_symbol): Likewise.
14990 (riscv_legitimize_move): Likewise.
14991 (riscv_function_value): Likewise.
14992 (riscv_hard_regno_nregs): Likewise.
14993 (riscv_expand_builtin): Likewise.
14994 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
14995 (riscv_build_integer): Likewise.
14996 (riscv_split_integer): Likewise.
14997 (riscv_legitimate_constant_p): Likewise.
14998 (riscv_cannot_force_const_mem): Likewise.
14999 (riscv_regno_mode_ok_for_base_p): Likewise.
15000 (riscv_valid_base_register_p): Likewise.
15001 (riscv_valid_offset_p): Likewise.
15002 (riscv_valid_lo_sum_p): Likewise.
15003 (riscv_classify_address): Likewise.
15004 (riscv_legitimate_address_p): Likewise.
15005 (riscv_address_insns): Likewise.
15006 (riscv_load_store_insns): Likewise.
15007 (riscv_force_binary): Likewise.
15008 (riscv_split_symbol): Likewise.
15009 (riscv_force_address): Likewise.
15010 (riscv_legitimize_address): Likewise.
15011 (riscv_move_integer): Likewise.
15012 (riscv_legitimize_const_move): Likewise.
15013 (riscv_legitimize_move): Likewise.
15014 (riscv_address_cost): Likewise.
15015 (riscv_subword): Likewise.
15016 (riscv_output_move): Likewise.
15017 (riscv_canonicalize_int_order_test): Likewise.
15018 (riscv_emit_int_order_test): Likewise.
15019 (riscv_function_arg_boundary): Likewise.
15020 (riscv_pass_mode_in_fpr_p): Likewise.
15021 (riscv_pass_fpr_single): Likewise.
15022 (riscv_pass_fpr_pair): Likewise.
15023 (riscv_get_arg_info): Likewise.
15024 (riscv_function_arg): Likewise.
15025 (riscv_function_arg_advance): Likewise.
15026 (riscv_arg_partial_bytes): Likewise.
15027 (riscv_function_value): Likewise.
15028 (riscv_pass_by_reference): Likewise.
15029 (riscv_setup_incoming_varargs): Likewise.
15030 (riscv_print_operand): Likewise.
15031 (riscv_elf_select_rtx_section): Likewise.
15032 (riscv_save_restore_reg): Likewise.
15033 (riscv_for_each_saved_reg): Likewise.
15034 (riscv_register_move_cost): Likewise.
15035 (riscv_hard_regno_mode_ok_p): Likewise.
15036 (riscv_hard_regno_nregs): Likewise.
15037 (riscv_class_max_nregs): Likewise.
15038 (riscv_memory_move_cost): Likewise.
15039 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
15040 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
15041 (rl78_addr_space_address_mode): Likewise.
15042 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15044 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
15045 (rs6000_reassociation_width): Likewise.
15046 (rs6000_invalid_binary_op): Likewise.
15047 (fusion_p9_p): Likewise.
15048 (emit_fusion_p9_load): Likewise.
15049 (emit_fusion_p9_store): Likewise.
15050 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
15051 (ok_for_simple_move_operands): Likewise.
15052 (ok_for_simple_move_strict_operands): Likewise.
15053 (ok_for_simple_arith_logic_operands): Likewise.
15054 (visium_legitimize_reload_address): Likewise.
15055 (visium_select_cc_mode): Likewise.
15056 (output_cbranch): Likewise.
15057 (visium_split_double_move): Likewise.
15058 (visium_expand_copysign): Likewise.
15059 (visium_expand_int_cstore): Likewise.
15060 (visium_expand_fp_cstore): Likewise.
15061 * config/visium/visium.c (visium_pass_by_reference): Likewise.
15062 (visium_function_arg): Likewise.
15063 (visium_function_arg_advance): Likewise.
15064 (visium_libcall_value): Likewise.
15065 (visium_setup_incoming_varargs): Likewise.
15066 (visium_legitimate_constant_p): Likewise.
15067 (visium_legitimate_address_p): Likewise.
15068 (visium_legitimize_address): Likewise.
15069 (visium_secondary_reload): Likewise.
15070 (visium_register_move_cost): Likewise.
15071 (visium_memory_move_cost): Likewise.
15072 (prepare_move_operands): Likewise.
15073 (ok_for_simple_move_operands): Likewise.
15074 (ok_for_simple_move_strict_operands): Likewise.
15075 (ok_for_simple_arith_logic_operands): Likewise.
15076 (visium_function_value_1): Likewise.
15077 (rtx_ok_for_offset_p): Likewise.
15078 (visium_legitimize_reload_address): Likewise.
15079 (visium_split_double_move): Likewise.
15080 (visium_expand_copysign): Likewise.
15081 (visium_expand_int_cstore): Likewise.
15082 (visium_expand_fp_cstore): Likewise.
15083 (visium_split_cstore): Likewise.
15084 (visium_select_cc_mode): Likewise.
15085 (visium_split_cbranch): Likewise.
15086 (output_cbranch): Likewise.
15087 (visium_print_operand_address): Likewise.
15088 * expmed.c (flip_storage_order): Likewise.
15089 * expmed.h (emit_cstore): Likewise.
15090 (flip_storage_order): Likewise.
15091 * genrecog.c (validate_pattern): Likewise.
15092 * hsa-gen.c (gen_hsa_addr): Likewise.
15093 * internal-fn.c (expand_arith_overflow): Likewise.
15094 * ira-color.c (allocno_copy_cost_saving): Likewise.
15095 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15096 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
15097 (process_invariant_for_inheritance): Likewise.
15098 * lra-eliminations.c (move_plus_up): Likewise.
15099 * omp-low.c (lower_oacc_reductions): Likewise.
15100 * simplify-rtx.c (simplify_subreg): Likewise.
15101 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
15102 (TARGET_CHKP_BOUND_MODE): Likewise..
15103 * targhooks.c (default_chkp_bound_mode): Likewise.
15104 (default_setup_incoming_vararg_bounds): Likewise.
15105 * targhooks.h (default_chkp_bound_mode): Likewise.
15106 (default_setup_incoming_vararg_bounds): Likewise.
15107 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
15108 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
15109 (have_whole_vector_shift): Likewise.
15110 * tree-vect-stmts.c (vectorizable_load): Likewise.
15111 * doc/tm.texi: Regenerate.
15113 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
15115 Graceful degrade if Binutils PR21472 is not available.
15118 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
15119 .rodata in flash test fails.
15120 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
15121 * confgure: Regenerate.
15122 * config.in: Regenerate.
15123 * config/avr/avr.c (avr_asm_named_section)
15124 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
15125 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
15126 (avr_asm_init_sections): Same.
15128 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15130 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
15131 (fma<VH:mode>4_intrinsic): Likewise.
15132 (*fmsub<VCVTF:mode>4): Likewise.
15133 (*fmsub<VH:mode>4_intrinsic): Likewise.
15135 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
15138 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
15139 Don't depend on "optimize > 0".
15140 (out_movhi_r_mr, out_movqi_mr_r): Same.
15141 (out_movhi_mr_r, out_movqi_r_mr): Same.
15142 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
15143 io_address_operand on "optimize > 0".
15145 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15147 * tree-loop-distribution.c: Add general explanantion on the pass.
15148 (generate_loops_for_partition): Mark distributed loop.
15149 (pg_add_dependence_edges): New parameter. Handle alias data
15150 dependence specially and record it in the parameter if asked.
15151 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
15152 (init_partition_graph_vertices, add_partition_graph_edge): New.
15153 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
15154 (free_partition_graph_vdata, build_partition_graph): New.
15155 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
15156 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
15157 (data_ref_segment_size, latch_dominated_by_data_ref): New.
15158 (compute_alias_check_pairs, version_loop_by_alias_check): New.
15159 (version_for_distribution_p, finalize_partitions): New.
15160 (distribute_loop): Handle alias data dependence specially. Factor
15161 out loop fusion code as functions and call these functions.
15163 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15165 * tree-loop-distribution.c (classify_partition): New parameter and
15166 better handle reduction statement.
15167 (rdg_build_partitions): Revise comment.
15168 (distribute_loop): Compute statements in all partitions and pass it
15169 to classify_partition.
15171 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15173 * tree-loop-distribution.c (enum partition_type): New.
15174 (struct partition): New field type.
15175 (partition_merge_into): Add parameter. Update partition type.
15176 (data_dep_in_cycle_p, update_type_for_merge): New functions.
15177 (build_rdg_partition_for_vertex): Compute partition type.
15178 (rdg_build_partitions): Dump partition type.
15179 (distribute_loop): Update calls to partition_merge_into.
15181 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15183 * tree-loop-distribution.c (struct ddr_hasher): New.
15184 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
15186 (classify_partition): Call get_data_dependence.
15187 (pg_add_dependence_edges): Ditto.
15188 (distribute_loop): Release data dependence hash table.
15190 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15192 * tree-loop-distribution.c (ref_base_address): Delete.
15193 (similar_memory_accesses): Rename ...
15194 (share_memory_accesses): ... to this. Check if partitions access
15195 the same memory reference.
15196 (distribute_loop): Call share_memory_accesses.
15198 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15200 * tree-loop-distribution.c (struct partition): New field recording
15201 its data reference.
15202 (partition_alloc, partition_free): Init and release data refs.
15203 (partition_merge_into): Merge data refs.
15204 (build_rdg_partition_for_vertex): Collect data refs for partition.
15205 (pg_add_dependence_edges): Change parameters from vector to bitmap.
15207 (distribute_loop): Remve data refs from vertice data of partition
15210 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15212 * tree-loop-distribution.c (params.h): Include header file.
15213 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
15214 (datarefs_vec): New global var.
15215 (create_rdg_vertices): Use datarefs_vec directly.
15216 (free_rdg): Don't free data references.
15217 (build_rdg): Update use. Don't free data references.
15218 (distribute_loop): Compute global variable for data references.
15219 Bail out if there are too many data references.
15221 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15223 * tree-loop-distribution.c (loop_nest): New global var.
15224 (build_rdg): Use loop directly, rather than loop nest.
15225 (pg_add_dependence_edges): Remove loop nest parameter. Use global
15227 (distribute_loop): Compute global variable loop nest. Update use.
15229 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15231 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
15232 (partition_merge_into): New parameter. Dump reason for fusion.
15233 (distribute_loop): Update use of partition_merge_into.
15235 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15237 * tree-loop-distribution.c (bb_top_order_index): New.
15238 (bb_top_order_index_size, bb_top_order_cmp): New.
15239 (stmts_from_loop): Use topological order.
15240 (pass_loop_distribution::execute): Compute and release topological
15241 order for basic blocks.
15243 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15245 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
15248 2017-07-05 Bin Cheng <bin.cheng@arm.com>
15250 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
15251 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
15252 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
15253 * internal-fn.def (LOOP_DIST_ALIAS): New.
15254 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
15255 (fold_loop_internal_call): ... this.
15256 (vect_loop_dist_alias_call): New function.
15257 (set_uid_loop_bbs): Call fold_loop_internal_call.
15258 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
15261 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
15264 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
15265 Require dead FLAGS_REG at the beginning of a peephole.
15267 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
15270 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
15271 arguments in the call to __builtin_ia32_sbb_u32.
15272 (_subborrow_u64): Swap _X and _Y arguments in the call to
15273 __builtin_ia32_sbb_u64.
15275 2017-07-04 Jakub Jelinek <jakub@redhat.com>
15278 * tree-vrp.c (compare_assert_loc): Turn into a function template
15279 with stable template parameter. Only test if a->e is NULL,
15280 !a->e == !b->e has been verified already. Use e == NULL or
15281 e != NULL instead of e or ! e tests. If stable is true, don't use
15282 iterative_hash_expr, on the other side allow a or b or both NULL
15283 and sort the NULLs last.
15284 (process_assert_insertions): Sort using compare_assert_loc<false>
15285 instead of compare_assert_loc, later sort using
15286 compare_assert_loc<true> before calling process_assert_insertions_for
15287 in a loop. Use break instead of continue once seen NULL pointer.
15289 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
15291 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
15292 Cortex-R7 and Cortex-R8 processors.
15294 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
15296 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
15297 uninitialized while src is not.
15299 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
15301 * common/config/arm/arm-common.c: Adjust include path for
15303 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
15304 (arm-cpu.h): Create in build directory. Adjust dependency rules.
15305 (arm-cpu-data.h): Likewise.
15306 (arm-cpu-cdata.h): Likewise.
15307 * config/arm/arm-cpu.h: Delete.
15308 * config/arm/arm-cpu-cdata.h: Delete.
15309 * config/arm/arm-cpu-data.h: Delete.
15311 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
15313 * config/arm/arm-cpus.in (cortex-a55): New.
15314 (cortex-a75): Likewise.
15315 (cortex-a75.cortex-a55): Likewise.
15316 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
15318 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
15319 * config/arm/arm-cpu-cdata.h: Regenerate.
15320 * config/arm/arm-cpu-data.h: Regenerate.
15321 * config/arm/arm-cpu.h: Regenerate.
15322 * config/arm/arm-tables.opt: Regenerate.
15323 * config/arm/arm-tune.md: Regenerate.
15325 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
15327 * haifa-sched.c (sched_create_recovery_edges): Update profile.
15329 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
15331 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
15334 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
15336 PR tree-optimization/81292
15337 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
15338 full_string_p, also call adjust_related_strinfos if the adjustment
15339 is simple, otherwise invalidate related strinfos.
15341 2017-07-04 Martin Liska <mliska@suse.cz>
15344 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
15345 newly created variable as DECL_IGNORED_P.
15347 2017-07-04 Martin Liska <mliska@suse.cz>
15350 * ipa-inline.c (inline_small_functions):
15351 Use xstrdup_for_dump.
15353 2017-07-04 Tom de Vries <tom@codesourcery.com>
15355 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
15357 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
15360 * config/darwin.c (darwin_function_switched_text_sections):
15363 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
15365 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
15367 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
15369 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
15371 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15373 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
15374 min_profitable_iters, and th as inclusive lower bounds.
15375 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
15376 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
15377 for min_profitable_iters and min_profitable_estimate.
15378 (vect_transform_loop): Treat th as an inclusive lower bound.
15379 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
15381 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
15384 * config/darwin.c (darwin_function_switched_text_sections):
15385 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
15386 in two pieces, and suppress the use of buf.
15388 2017-07-03 Nathan Sidwell <nathan@acm.org>
15390 * hash-table.h (hash_table_mod1): Fix indentation.
15392 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15394 PR middle-end/81290
15395 * predict.c (force_edge_cold): Be more careful about propagation
15397 * profile-count.h (profile_probability::guessed,
15398 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
15400 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
15402 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
15404 * doc/invoke.texi (rcpc architecture extension): Document it.
15406 2017-07-03 Richard Biener <rguenther@suse.de>
15408 PR tree-optimization/60510
15409 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
15410 the scalar reduction PHI and use it.
15411 (vectorizable_reduction): Properly guard the single_defuse_cycle
15412 path for non-SLP reduction chains where we cannot use it.
15413 Rework reduc_def/index and vector type deduction. Rework
15414 vector operand gathering during reduction op code-gen.
15415 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
15416 chains dissolve the chain and leave it to non-SLP reduction
15419 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15421 * tree-data-ref.h (dr_alignment): Declare.
15422 * tree-data-ref.c (dr_alignment): New function.
15423 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
15424 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
15426 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
15428 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15430 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
15431 and base_misalignment fields.
15432 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
15433 * tree-data-ref.c: Include builtins.h.
15434 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
15435 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
15436 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
15437 * tree-vect-data-refs.c: Include tree-cfg.h.
15438 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
15439 fields instead of calculating an alignment here.
15440 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
15441 innermost_loop_behavior fields.
15443 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15445 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
15447 (DR_STEP_ALIGNMENT): New macro.
15448 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
15449 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
15450 (create_data_ref): Print it.
15451 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
15452 to tell whether the step preserves vector (mis)alignment.
15453 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
15454 Move the check for an integer step and generalise to all INTEGER_CST.
15455 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
15456 Print the outer step alignment.
15458 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15460 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
15461 with offset_alignment.
15462 (DR_ALIGNED_TO): Delete.
15463 (DR_OFFSET_ALIGNMENT): New macro.
15464 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
15465 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
15466 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
15467 (create_data_ref): Likewise.
15468 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
15469 (vect_analyze_data_refs): Likewise.
15470 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
15471 creating dummy innermost behavior.
15473 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15475 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
15476 with a "innermost_loop_behavior *" and refeence tree.
15477 * tree-data-ref.c (dr_analyze_innermost): Likewise.
15478 (create_data_ref): Update call accordingly.
15479 * tree-predcom.c (find_looparound_phi): Likewise.
15481 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15483 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
15484 fields with dr_wrt_vec_loop.
15485 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
15486 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
15487 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
15488 (vect_dr_behavior): New function.
15489 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
15490 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
15491 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
15492 track whether the step preserves the misalignment.
15493 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
15494 Use vect_dr_behavior.
15495 (vect_setup_realignment): Update call accordingly.
15496 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
15497 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
15498 call to vect_create_addr_base_for_vector_ref.
15499 (vect_create_cond_for_align_checks): Likewise.
15500 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
15501 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
15502 (vect_recog_mask_conversion_pattern): Likewise.
15503 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
15504 (new_stmt_vec_info): Remove redundant zeroing.
15506 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
15508 * common/config/arm/arm-common.c (arm_be8_option): New function.
15509 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
15510 (ISA_ARMv6): Add isa_bit_be8.
15511 * config/arm/arm.h (arm_be8_option): Add prototype.
15512 (BE8_SPEC_FUNCTION): New define.
15513 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
15514 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
15515 (mlittle-endian): Similarly.
15516 (mbe8, mbe32): New options.
15517 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
15518 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
15520 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15522 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
15524 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15526 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
15527 (cleanup_tree_cfg_bb): Use it.
15528 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
15530 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
15532 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15535 * loop-doloop.c (add_test): Update profile.
15537 2017-07-03 Martin Liska <mliska@suse.cz>
15540 * sanopt.c (rewrite_usage_of_param): New function.
15541 (sanitize_rewrite_addressable_params): Likewise.
15542 (pass_sanopt::execute): Call rewrite_usage_of_param.
15544 2017-07-03 Richard Biener <rguenther@suse.de>
15546 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
15547 back to using VIEW_CONVERT_EXPR.
15549 2017-07-03 Martin Liska <mliska@suse.cz>
15552 * doc/extend.texi: Document when a resolver function is
15553 generated for target_clones.
15555 2017-07-03 Martin Liska <mliska@suse.cz>
15557 * asan.c (asan_emit_stack_protection): Unpoison just red zones
15558 and shadow memory of auto variables which are subject of
15559 use-after-scope sanitization.
15560 (asan_expand_mark_ifn): Add do set only when is_poison.
15562 2016-07-03 Richard Biener <rguenther@suse.de>
15564 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
15566 (vect_force_simple_reduction): Record reduction def -> phi mapping.
15567 (vectorizable_reduction): Perform reduction PHI creation when
15568 visiting a reduction PHI and adjust and simplify code generation
15569 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
15570 (vect_transform_loop): Visit reduction PHIs.
15571 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
15572 defs into the SLP tree.
15573 (vect_build_slp_tree): Reduction defs terminate the recursion.
15574 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
15576 (vect_get_vec_defs_for_stmt_copy): Export.
15577 (vect_get_vec_defs): Likewise.
15578 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
15580 (vect_get_vec_defs_for_stmt_copy): Declare.
15581 (vect_get_vec_defs): Likewise.
15583 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
15585 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
15586 parameter with a "loop" parameter and use it instead of the
15587 loop containing DR_STMT. Don't check simple_iv when doing
15588 BB analysis. Describe the two analysis modes in the comment.
15590 2017-07-03 Tom de Vries <tom@codesourcery.com>
15592 PR tree-optimization/69468
15593 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
15594 (find_same_succ_bb): Handle ignore_edge_flags.
15596 2017-07-03 Tom de Vries <tom@codesourcery.com>
15598 PR tree-optimization/81192
15599 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
15601 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
15603 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
15605 2017-07-03 Tom de Vries <tom@codesourcery.com>
15607 PR tree-optimization/81192
15608 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
15609 BB_SAME_SUCC (bb) == NULL.
15611 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15613 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
15616 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15618 * dumpfile.c: Include profile-count.h
15619 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
15621 (insert_cond_bb): Update profile.
15622 * tree-cfg.h (insert_cond_bb): Update prototype.
15623 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
15624 * tree-dump.c: Do not include tree-cfg.
15626 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15628 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
15630 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15632 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
15635 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
15637 * tree-complex.c (expand_complex_div_wide): update profile.
15639 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
15640 Alan Hayward <alan.hayward@arm.com>
15641 David Sherwood <david.sherwood@arm.com>
15643 * Makefile.in (MACHMODE_H): Remove insn-modes.h
15644 (CORETYPES_H): New define.
15645 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
15646 (insn-modes-inline.h, s-modes-inline-h): New rules.
15647 (generated_files): Add insn-modes-inline.h.
15648 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
15649 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
15650 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
15651 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
15652 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
15653 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
15654 (build/gencodes.o, build/genconditions.o): Likewise.
15655 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
15656 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
15657 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
15658 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
15659 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
15660 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
15661 * coretypes.h: Include everything up to real.h for generators.
15662 Include insn-modes.h first. Include wide-int-print.h after
15663 wide-int.h. Include insn-modes-inline.h and then machmode.h.
15664 * machmode.h: Don't include insn-modes.h here.
15665 * function-tests.c: Remove includes of signop.h, machmode.h,
15666 double-int.h and wide-int.h.
15668 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
15670 * optc-save-gen.awk: Likewise.
15671 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
15672 * godump.c: Remove include of wide-int-print.h.
15673 * pretty-print.h: Likewise.
15674 * wide-int-print.cc: Likewise.
15675 * wide-int.cc: Likewise.
15676 * hash-map-tests.c: Remove include of signop.h.
15677 * hash-set-tests.c: Likewise.
15678 * rtl-tests.c: Likewise.
15679 * mkconfig.sh: Remove include of machmode.h.
15680 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
15682 (emit_insn_modes_inline_h): ...this new function. Emit the code
15683 into an insn-modes-inline.h header file, adding appropriate
15684 include guards and end comments.
15685 (emit_insn_modes_c_header): Remove include of machmode.h.
15686 (emit_min_insn_modes_c_header): Include coretypes.h rather than
15688 (main): Handle -i flag and call emit_insn_modes_inline_h when
15691 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
15693 * tree-ssa-strlen.c (strinfo): Rename the length field to
15694 nonzero_chars. Add a full_string_p field.
15695 (compare_nonzero_chars, zero_length_string_p): New functions.
15696 (get_addr_stridx): Add an offset_out parameter.
15697 Use compare_nonzero_chars.
15698 (get_stridx): Update accordingly. Use compare_nonzero_chars.
15699 (new_strinfo): Update after above changes to strinfo.
15700 (set_endptr_and_length): Set full_string_p.
15701 (get_string_length): Update after above changes to strinfo.
15702 (unshare_strinfo): Update call to new_strinfo.
15703 (maybe_invalidate): Likewise.
15704 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
15705 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
15706 as a uhwi instead of an shwi. Update after above changes to
15707 strinfo and new_strinfo.
15708 (zero_length_string): Assert that chainsi contains full strings.
15709 Use zero_length_string_p. Update call to new_strinfo.
15710 (adjust_related_strinfos): Update after above changes to strinfo.
15711 Copy full_string_p from origsi.
15712 (adjust_last_stmt): Use zero_length_string_p.
15713 (handle_builtin_strlen): Update after above changes to strinfo and
15714 new_strinfo. Install the lhs as the string length if the previous
15715 entry didn't describe a full string.
15716 (handle_builtin_strchr): Update after above changes to strinfo
15718 (handle_builtin_strcpy): Likewise.
15719 (handle_builtin_strcat): Likewise.
15720 (handle_builtin_malloc): Likewise.
15721 (handle_pointer_plus): Likewise.
15722 (handle_builtin_memcpy): Likewise. Track nonzero characters
15723 that aren't necessarily followed by a nul terminator.
15724 (handle_char_store): Likewise.
15726 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
15728 PR tree-optimization/80769
15729 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
15730 for malloc and calloc. Document the new invariant that all related
15731 strinfos have delayed lengths or none do.
15732 (verify_related_strinfos): Move earlier in file.
15733 (set_endptr_and_length): New function, split out from...
15734 (get_string_length): ...here. Also set the lengths of related
15736 (zero_length_string): Assert that chainsi has known (rather than
15738 (adjust_related_strinfos): Likewise.
15740 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
15742 PR tree-optimization/81136
15743 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
15744 assert that two references with the same misalignment have the same
15745 compile-time misalignment if those compile-time misalignments
15748 2017-07-01 Andi Kleen <ak@linux.intel.com>
15750 * print-tree.c (print_node): Print all attributes.
15752 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15754 * cfg.c (scale_bbs_frequencies): New function.
15755 * cfg.h (scale_bbs_frequencies): Declare it.
15756 * cfgloopanal.c (single_likely_exit): Cleanup.
15757 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
15759 (scale_loop_profile): Likewise.
15760 (loop_version): Likewise.
15761 (create_empty_loop_on_edge): Update.
15762 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
15763 scale_loop_frequencies, scale_loop_profile, loopify,
15764 loop_version): Update prototypes.
15765 * modulo-sched.c (sms_schedule): Update.
15766 * predict.c (unlikely_executed_edge_p): Also check probability.
15767 (probably_never_executed_edge_p): Fix typo.
15768 * tree-if-conv.c (version_loop_for_if_conversion): Update.
15769 * tree-parloops.c (gen_parallel_loop): Update.
15770 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
15771 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
15772 * tree-ssa-loop-split.c (split_loop): Update.
15773 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
15774 * tree-vect-loop-manip.c (vect_do_peeling): Update.
15775 (vect_loop_versioning): Update.
15776 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
15778 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15780 * trans-mem.c (split_bb_make_tm_edge): Update profile.
15782 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15784 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
15785 to keep profile consistent.
15787 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15789 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
15790 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
15791 * profile-count.h (max_safe_multiplier): Make unsigned.
15792 (profile_count::guessed_zero): New.
15794 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15796 * bb-reorder.c (fix_up_crossing_landing_pad,
15797 fix_crossing_conditional_branches): Use make_single_succ_edge
15798 to keep profile consistent.
15800 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
15802 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
15805 2017-07-01 Jakub Jelinek <jakub@redhat.com>
15808 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
15809 the right scopes, make sure cond_jump isn't preserved between multiple
15810 iterations. Search for fallthru edge whenever there are 3+ edges and
15811 use find_fallthru_edge for it.
15813 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15815 Patch by Alexander Monakov <amonakov@ispras.ru>
15816 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
15817 probabilities consistently.
15819 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15821 * pa.c (pa_expand_compare_and_swap_loop): Update call of
15822 emit_cmp_and_jump_insns.
15824 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15827 * tree-inline.c (expand_call_inline): Combine profile statuses.
15829 2017-06-30 Andrew Pinski <apinski@cavium.com>
15831 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
15832 fold_stmt returned true.
15834 2017-06-30 Nathan Sidwell <nathan@acm.org>
15836 * ggc.h (empty_string): Delete.
15837 * cfgexpand.c (expand_asm_stmt): Use plain "".
15838 * optabs.c (expand_asm_memory_barrier): Likewise.
15839 * stringpool.c (empty_string): Delete.
15840 (digit_vector, digit_string): Delete.
15841 (ggc_alloc_string): Use plain "", don't optimize single digit
15842 strings. Use ggc_alloc_atomic.
15844 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
15846 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
15847 comparison set and one other set, use the cost of the non-comparison
15850 2017-06-30 Nathan Sidwell <nathan@acm.org>
15852 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
15855 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
15857 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
15858 loops. Remove now unneeded calls to gimple_switch_set_label() that
15859 just set removed labels to NULL_TREE.
15861 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
15863 * tree-ssanames.c (set_range_info_raw): Abstract from ...
15864 (set_range_info): ...here. Only call set_range_info_raw if domain
15866 (set_nonzero_bits): Call set_range_info_raw.
15867 * tree-ssanames.h (set_range_info_raw): New.
15869 2017-06-30 Jakub Jelinek <jakub@redhat.com>
15872 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
15873 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
15874 of nonimmediate_operand and <store_mask_constraint> instead of m
15875 for the input operand. For V8FI iterator, always split if input
15876 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
15877 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
15878 <store_mask_predicate> instead of register_operand and
15879 <store_mask_constraint> instead of v for the input operand. Make
15880 sure both operands aren't MEMs for if not <mask_applied>.
15882 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
15884 * lto-wrapper.c (copy_file) Close both file descriptors before
15887 2017-06-30 Martin Liska <mliska@suse.cz>
15890 * multiple_target.c (create_dispatcher_calls): Make ifunc
15891 also for function that don't have calls or are not referenced.
15893 2017-06-30 Richard Biener <rguenther@suse.de>
15895 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
15896 analyze the first scalar stmt. Move vector type computation
15897 for the BB case here from ...
15898 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
15899 live operation processing in the SLP case properly.
15901 2017-06-30 Richard Biener <rguenther@suse.de>
15903 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
15905 2017-06-30 Martin Liska <mliska@suse.cz>
15908 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
15909 before BUILT_IN_UNWIND_RESUME when ASAN is used.
15911 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
15913 * doc/invoke.texi (AArch64): Add missing options and remove redundant
15916 2017-06-30 Richard Biener <rguenther@suse.de>
15918 PR tree-optimization/81249
15919 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
15920 condition reduction result to original scalar type.
15922 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15924 * profile-count.h (enum profile_quality): Fix typos and whitespace
15927 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15929 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
15930 type for branch probabilities.
15932 2017-06-29 Julian Brown <julian@codesourcery.com>
15933 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15935 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
15936 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
15937 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
15938 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
15940 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15942 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
15943 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
15944 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
15945 CC usage from generic code to here.
15946 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
15947 CC usage into the target macros.
15949 2017-06-29 Maya Rashish <coypu@sdf.org>
15951 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
15954 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
15956 * arm/arm-builtins.c: Include profile-count.h
15957 * except.c (sjlj_emit_function_enter): Use
15958 profile_probability::unlikely.
15960 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15962 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
15963 and tocrel_offset be pointer args rather than implicitly using
15965 (legitimate_constant_pool_address_p, rs6000_emit_move,
15966 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
15967 tocrel_offset and use in toc_relative_expr_p call.
15968 (print_operand, print_operand_address): Use static tocrel_base_oac
15969 and tocrel_offset_oac.
15970 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
15973 2017-06-29 Maya Rashish <coypu@sdf.org>
15975 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
15977 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
15979 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
15980 objects, take into account only the alignment of 'op0' and 'mode1' if
15983 2017-06-29 Steve Ellcey <sellcey@cavium.com>
15985 * ccmp.c (ccmp_tree_comparison_p): New function.
15986 (ccmp_candidate_p): Update to use above function.
15987 (get_compare_parts): New function.
15988 (expand_ccmp_next): Update to use new functions.
15989 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
15991 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
15992 take mode as argument.
15993 * ccmp.h (expand_ccmp_expr): Add mode as argument.
15994 * expr.c (expand_expr_real_1): Pass mode as argument.
15996 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
15998 * combine.c (combine_instructions): Print insns to dump_file, together
16001 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
16003 * asan.c (asan_emit_stack_protection): Update.
16004 (create_cond_insert_point): Update.
16005 * auto-profile.c (afdo_propagate_circuit): Update.
16006 * basic-block.h (struct edge_def): Turn probability to
16007 profile_probability.
16008 (EDGE_FREQUENCY): Update.
16009 * bb-reorder.c (find_traces_1_round): Update.
16010 (better_edge_p): Update.
16011 (sanitize_hot_paths): Update.
16012 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
16013 (make_single_succ_edge): Update.
16014 (check_bb_profile): Update.
16015 (dump_edge_info): Update.
16016 (update_bb_profile_for_threading): Update.
16017 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
16018 probabilitycount to 0.
16019 * cfgbuild.c (compute_outgoing_frequencies): Update.
16020 * cfgcleanup.c (try_forward_edges): Update.
16021 (outgoing_edges_match): Update.
16022 (try_crossjump_to_edge): Update.
16023 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
16024 (expand_gimple_tailcall): Update.
16025 (construct_init_block): Use make_single_succ_edge.
16026 (construct_exit_block): Use make_single_succ_edge.
16027 * cfghooks.c (verify_flow_info): Update.
16028 (redirect_edge_succ_nodup): Update.
16029 (split_edge): Update.
16030 (account_profile_record): Update.
16031 * cfgloopanal.c (single_likely_exit): Update.
16032 * cfgloopmanip.c (scale_loop_profile): Update.
16033 (set_zero_probability): Remove.
16034 (duplicate_loop_to_header_edge): Update.
16035 * cfgloopmanip.h (loop_version): Update prototype.
16036 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
16037 (force_nonfallthru_and_redirect): Update.
16038 (update_br_prob_note): Update.
16039 (rtl_verify_edges): Update.
16040 (purge_dead_edges): Update.
16041 (rtl_lv_add_condition_to_bb): Update.
16042 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
16043 * cgraphunit.c (init_lowered_empty_function): Update.
16044 (cgraph_node::expand_thunk): Update.
16045 * cilk-common.c: Include profile-count.h
16046 * dojump.c (inv): Remove.
16047 (jumpifnot): Update.
16048 (jumpifnot_1): Update.
16049 (do_jump_1): Update.
16051 (do_jump_by_parts_greater_rtx): Update.
16052 (do_compare_rtx_and_jump): Update.
16053 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
16054 do_jump_1. do_compare_rtx_and_jump): Update prototype.
16055 * dwarf2cfi.c: Include profile-count.h
16056 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
16057 (sjlj_emit_dispatch_table): Likewise.
16058 * explow.c: Include profile-count.h
16059 * expmed.c (emit_store_flag_force): Update.
16060 (do_cmp_and_jump): Update.
16061 * expr.c (compare_by_pieces_d::generate): Update.
16062 (compare_by_pieces_d::finish_mode): Update.
16063 (emit_block_move_via_loop): Update.
16064 (store_expr_with_bounds): Update.
16065 (store_constructor): Update.
16066 (expand_expr_real_2): Update.
16067 (expand_expr_real_1): Update.
16068 * expr.h (try_casesi, try_tablejump): Update prototypes.
16069 * gimple-pretty-print.c (dump_probability): Update.
16070 (dump_profile): New.
16071 (dump_gimple_label): Update.
16072 (dump_gimple_bb_header): Update.
16073 * graph.c (draw_cfg_node_succ_edges): Update.
16074 * hsa-gen.c (convert_switch_statements): Update.
16075 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
16076 (find_if_case_1): Update.
16077 (find_if_case_2): Update.
16078 * internal-fn.c (expand_arith_overflow_result_store): Update.
16079 (expand_addsub_overflow): Update.
16080 (expand_neg_overflow): Update.
16081 (expand_mul_overflow): Update.
16082 (expand_vector_ubsan_overflow): Update.
16083 * ipa-cp.c (good_cloning_opportunity_p): Update.
16084 * ipa-split.c (split_function): Use make_single_succ_edge.
16085 * ipa-utils.c (ipa_merge_profiles): Update.
16086 * loop-doloop.c (add_test): Update.
16087 (doloop_modify): Update.
16088 * loop-unroll.c (compare_and_jump_seq): Update.
16089 (unroll_loop_runtime_iterations): Update.
16090 * lra-constraints.c (lra_inheritance): Update.
16091 * lto-streamer-in.c (input_cfg): Update.
16092 * lto-streamer-out.c (output_cfg): Update.
16093 * mcf.c (adjust_cfg_counts): Update.
16094 * modulo-sched.c (sms_schedule): Update.
16095 * omp-expand.c (expand_omp_for_init_counts): Update.
16096 (extract_omp_for_update_vars): Update.
16097 (expand_omp_ordered_sink): Update.
16098 (expand_omp_for_ordered_loops): Update.
16099 (expand_omp_for_generic): Update.
16100 (expand_omp_for_static_nochunk): Update.
16101 (expand_omp_for_static_chunk): Update.
16102 (expand_cilk_for): Update.
16103 (expand_omp_simd): Update.
16104 (expand_omp_taskloop_for_outer): Update.
16105 (expand_omp_taskloop_for_inner): Update.
16106 * omp-simd-clone.c (simd_clone_adjust): Update.
16107 * optabs.c (expand_doubleword_shift): Update.
16108 (expand_abs): Update.
16109 (emit_cmp_and_jump_insn_1): Update.
16110 (expand_compare_and_swap_loop): Update.
16111 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
16112 * predict.c (predictable_edge_p): Update.
16113 (edge_probability_reliable_p): Update.
16114 (set_even_probabilities): Update.
16115 (combine_predictions_for_insn): Update.
16116 (combine_predictions_for_bb): Update.
16117 (propagate_freq): Update.
16118 (estimate_bb_frequencies): Update.
16119 (force_edge_cold): Update.
16120 * profile-count.c (profile_count::dump): Add missing space into dump.
16121 (profile_count::debug): Add newline.
16122 (profile_count::differs_from_p): Explicitly convert to unsigned.
16123 (profile_count::stream_in): Update.
16124 (profile_probability::dump): New member function.
16125 (profile_probability::debug): New member function.
16126 (profile_probability::differs_from_p): New member function.
16127 (profile_probability::differs_lot_from_p): New member function.
16128 (profile_probability::stream_in): New member function.
16129 (profile_probability::stream_out): New member function.
16130 * profile-count.h (profile_count_quality): Rename to ...
16131 (profile_quality): ... this one.
16132 (profile_probability): New.
16133 (profile_count): Update.
16134 * profile.c (compute_branch_probabilities): Update.
16135 * recog.c (peep2_attempt): Update.
16136 * sched-ebb.c (schedule_ebbs): Update.
16137 * sched-rgn.c (find_single_block_region): Update.
16138 (compute_dom_prob_ps): Update.
16139 (schedule_region): Update.
16140 * sel-sched-ir.c (compute_succs_info): Update.
16141 * stmt.c (struct case_node): Update.
16142 (do_jump_if_equal): Update.
16143 (get_outgoing_edge_probs): Update.
16144 (conditional_probability): Update.
16145 (emit_case_dispatch_table): Update.
16146 (expand_case): Update.
16147 (expand_sjlj_dispatch_table): Update.
16148 (emit_case_nodes): Update.
16149 * targhooks.c: Update.
16150 * tracer.c (better_p): Update.
16151 (find_best_successor): Update.
16152 * trans-mem.c (expand_transaction): Update.
16153 * tree-call-cdce.c: Update.
16154 * tree-cfg.c (gimple_split_edge): Upate.
16155 (move_sese_region_to_fn): Upate.
16156 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
16157 * tree-eh.c (lower_resx): Upate.
16158 (cleanup_empty_eh_move_lp): Upate.
16159 * tree-if-conv.c (version_loop_for_if_conversion): Update.
16160 * tree-inline.c (copy_edges_for_bb): Update.
16161 (copy_cfg_body): Update.
16162 * tree-parloops.c (gen_parallel_loop): Update.
16163 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
16164 (gimple_gen_time_profiler): Update.
16165 * tree-ssa-dce.c (remove_dead_stmt): Update.
16166 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
16167 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
16168 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
16169 (unloop_loops): Update.
16170 (try_peel_loop): Update.
16171 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
16172 * tree-ssa-loop-split.c (connect_loops): Update.
16173 (split_loop): Update.
16174 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
16175 (hoist_guard): Update.
16176 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
16177 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
16178 (value_replacement): Update.
16179 * tree-ssa-reassoc.c (branch_fixup): Update.
16180 * tree-ssa-tail-merge.c (replace_block_by): Update.
16181 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
16182 (create_edge_and_update_destination_phis): Update.
16183 (compute_path_counts): Update.
16184 (recompute_probabilities): Update.
16185 (update_joiner_offpath_counts): Update.
16186 (freqs_to_counts_path): Update.
16187 (duplicate_thread_path): Update.
16188 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
16189 (struct switch_conv_info): Update.
16190 (gen_inbound_check): Update.
16191 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
16192 (vect_do_peeling): Update.
16193 (vect_loop_versioning): Update.
16194 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
16195 (optimize_mask_stores): Update.
16196 * ubsan.c (ubsan_expand_null_ifn): Update.
16197 * value-prof.c (gimple_divmod_fixed_value): Update.
16198 (gimple_divmod_fixed_value_transform): Update.
16199 (gimple_mod_pow2): Update.
16200 (gimple_mod_pow2_value_transform): Update.
16201 (gimple_mod_subtract): Update.
16202 (gimple_mod_subtract_transform): Update.
16203 (gimple_ic): Update.
16204 (gimple_stringop_fixed_value): Update.
16205 (gimple_stringops_transform): Update.
16206 * value-prof.h: Update.
16208 2017-06-29 Carl Love <cel@us.ibm.com>
16210 * config/rs6000/rs6000-c.c: Add support for built-in functions
16211 vector signed int vec_signed (vector float);
16212 vector signed long long vec_signed (vector double);
16213 vector signed int vec_signed2 (vector double, vector double);
16214 vector signed int vec_signede (vector double);
16215 vector signed int vec_signedo (vector double);
16216 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
16217 instruction generator.
16218 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
16219 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
16220 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
16222 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
16223 vunsignede_v2df): Add define_expands.
16224 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
16225 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
16226 VEC_UNSIGNEDO): Add definitions.
16227 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
16228 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
16229 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
16230 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
16231 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
16232 * config/rs6000/altivec.h (vec_signed, vec_signed2,
16233 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
16234 vec_unsignede, vec_unsignedo): Add builtin defines.
16235 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
16237 * doc/extend.texi: Update the built-in documentation file for the
16238 new built-in functions.
16240 2017-06-29 Richard Biener <rguenther@suse.de>
16242 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
16243 reduction chains to LOOP_VINFO_REDUCTIONS.
16244 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
16245 SLP reductions after processing reduction chains.
16247 2017-06-29 Nathan Sidwell <nathan@acm.org>
16249 * builtins.c (fold_builtin_FUNCTION): Use
16250 lang_hooks.decl_printable_name.
16252 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
16254 PR middle-end/81194
16255 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
16256 with only one label.
16257 * stmt.c (expand_case): Assert NCASES is greater than one.
16259 2017-06-29 Richard Biener <rguenther@suse.de>
16261 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
16263 (group_case_labels): Likewise.
16264 (find_taken_edge): Push sanity checking on val to workers...
16265 (find_taken_edge_cond_expr): ... here
16266 (find_taken_edge_switch_expr): ... and here, handle cases
16267 with just a default label.
16268 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
16269 (group_case_labels): Likewise.
16270 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
16271 group_case_labels does anything cleanup the CFG again.
16273 2017-06-29 Bin Cheng <bin.cheng@arm.com>
16275 PR tree-optimization/81196
16276 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
16277 exit condition comparing two IVs.
16279 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
16281 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
16282 profile to the dummy entry at the end of the list of architectures.
16283 * config/arm/arm-cpu-cdata.h: Regenerated.
16285 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16286 Michael Collison <michael.collison@arm.com>
16289 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
16291 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
16292 (*aarch64_reg_<mode>3_minus_mask): New pattern.
16293 (*aarch64_<optab>_reg_di3_mask2): New pattern.
16294 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
16295 of shift when the shift amount is masked with constant equal to
16296 the size of the mode.
16297 * config/aarch64/predicates.md (subreg_lowpart_operator): New
16300 2017-06-29 Martin Liska <mliska@suse.cz>
16302 * config/i386/i386.opt: Change range from [1,5] to [0,5].
16304 2017-06-29 Yury Gribov <tetra2005@gmail.com>
16307 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
16310 (edge_growth_cache_entry::edge_growth_cache_entry): New
16312 (reset_edge_growth_cache): Update to use constructor.
16314 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16316 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
16317 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
16318 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
16320 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
16322 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
16323 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
16325 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
16327 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
16328 (*-linux-uclibc*): Add t-uclibc tmake_file.
16329 * config/t-musl: New.
16330 * config/t-uclibc: New.
16332 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
16334 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
16336 (gen_comm_data): Emit architectural setting of arch_prof.
16337 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
16339 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
16340 (armv8-m.base, armv8-m.main): Likewise.
16341 * arm-protos.h (arm_build_target): Add profile field.
16342 (arch_option): Likewise.
16343 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
16345 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
16346 arm_active_target.profile.
16348 2017-06-28 Richard Biener <rguenther@suse.de>
16350 PR middle-end/81227
16351 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
16352 TYPE_OVERFLOW_WRAPS.
16353 * match.pd (negate_expr_p): Likewise.
16354 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
16355 fold_build2, not fold_binary.
16357 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
16359 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
16360 Convert memory address to Pmode.
16361 (aarch64_print_operand): Assert MEM operands are always Pmode.
16363 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
16366 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
16367 Remove redundant if.
16368 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
16369 * config/arm/aarch-common-protos.h
16370 (aarch_forward_to_shift_is_not_shifted_re): Remove.
16371 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
16373 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
16376 * multiple_target.c (create_dispatcher_calls): Set the default
16377 clone to be static, not public.
16379 2017-06-28 Richard Biener <rguenther@suse.de>
16381 * tree-vect-loop.c (vectorizable_reduction): Move special
16382 cond reduction IV var creation ...
16383 (vect_create_epilog_for_reduction): ... here. Remove induction_index
16384 parameter. Use STMT_VINFO_VECTYPE.
16385 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
16388 2017-06-28 Martin Liska <mliska@suse.cz>
16391 * ipa-visibility.c (non_local_p): Handle visibility.
16393 2017-06-28 Martin Liska <mliska@suse.cz>
16396 * common.opt: Add IntegerRange to various options.
16397 * opt-functions.awk (integer_range_info): New function.
16398 * optc-gen.awk: Add integer_range_info to cl_options struct.
16399 * opts-common.c (decode_cmdline_option): Handle
16400 CL_ERR_INT_RANGE_ARG.
16401 (cmdline_handle_error): Likewise.
16402 * opts.c (print_filtered_help): Show valid interval in
16403 when --help is provided.
16404 * opts.h (struct cl_option): Add range_min and range_max fields.
16405 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
16407 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
16409 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
16410 (x * C EQ/NE y * C): New transformation.
16412 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
16414 * genmultilib (combination_space): Accept '+' in option names.
16416 2017-06-28 Martin Liska <mliska@suse.cz>
16419 * asan.c (instrument_derefs): Bail out inner references
16420 that are hard register variables.
16422 2017-06-28 Jakub Jelinek <jakub@redhat.com>
16425 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
16426 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
16428 2017-06-28 Richard Biener <rguenther@suse.de>
16430 * tree-vectorizer.h (vect_get_vec_defs): Remove.
16431 (vect_get_slp_defs): Adjust.
16432 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
16434 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
16436 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
16437 get_initial_defs_for_reduction instead of vect_get_vec_defs.
16438 (vectorizable_reduction): Adjust.
16439 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
16441 (vect_get_slp_defs): Likewise.
16442 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
16443 (vectorizable_bswap): Adjust.
16444 (vectorizable_call): Likewise.
16445 (vectorizable_conversion): Likewise.
16446 (vectorizable_assignment): Likewise.
16447 (vectorizable_shift): Likewise.
16448 (vectorizable_operation): Likewise.
16449 (vectorizable_store): Likewise.
16450 (vectorizable_condition): Likewise.
16451 (vectorizable_comparison): Likewise.
16453 2017-06-28 Michael Collison <michael.collison@arm.com>
16456 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
16458 (arm_gen_movmemqi): Removed unused variable 'i'.
16459 Convert 'for' loop into 'while' loop.
16460 (arm_expand_prologue): Remove last unnecessary set of insn.
16461 (thumb_pop): Remove unused variable 'pushed_words'.
16462 (thumb_exit): Remove last unnecessary set of regs_to_pop.
16464 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16466 * config/s390/predicates.md: Use s390_rel_address_ok_p.
16467 * config/s390/s390-protos.h: Add prototype of
16468 s390_rel_address_ok_p.
16469 * config/s390/s390.c (s390_got_symbol): New function.
16470 (s390_rel_address_ok_p): New function.
16471 (legitimize_pic_address): Use s390_rel_address_ok_p.
16472 (s390_load_got): Use s390_got_symbol.
16473 (s390_option_override): Issue error if
16474 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
16475 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
16477 * config/s390/s390.opt: New option mpic-data-is-text-relative.
16479 2017-06-27 Andrew Pinski <apinski@cavium.com>
16481 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
16482 (X * copysign (1.0, X)): New pattern.
16483 (X * copysign (1.0, -X)): New pattern.
16484 (copysign (-1.0, CST)): New pattern.
16486 2017-06-27 Joseph Myers <joseph@codesourcery.com>
16488 * genmultilib (combination_space): Remove variable.
16489 Validate reuse rules against regular expression for any sequence
16490 of multilib options in any order.
16492 2017-06-27 Michael Collison <michael.collison@arm.com>
16494 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
16495 call aarch64_split_simd_combine.
16496 * (aarch64_combine_internal<mode>): Delete pattern.
16497 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
16498 Allow register and subreg operands.
16500 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16502 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
16503 specific need, just fallback on defaults.
16504 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
16506 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16507 Olivier Hainque <hainque@adacore.com>
16509 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
16511 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
16512 targets. Pick a default if no particular attempt applied.
16513 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
16516 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16518 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
16519 (x86_64-wrs-vxworks7): Likewise.
16521 2017-06-27 Marek Polacek <polacek@redhat.com>
16524 * ubsan.c (instrument_null): Check get_base_address's result for null.
16526 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
16528 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
16530 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
16532 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
16533 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
16534 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
16535 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
16536 New function types.
16537 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
16538 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
16539 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
16540 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
16541 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
16542 BUILT_IN_FEUPDATEENV): New builtins.
16543 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
16544 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
16545 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
16546 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
16548 (builtin_structptr_types): Adjust size.
16549 * tree.c (builtin_structptr_types): Add four entries.
16551 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16552 Olivier Hainque <hainque@adacore.com>
16554 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
16555 (TLS_SYM): New local macro, forcing reference to __tls__ on
16556 link command lines for VxWorks 7 RTPs, triggering initialization
16558 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
16559 OS features TLS support, true for RTPs on VxWorks 7.
16560 * config/vxworks.c (vxworks_override_options): Setup emutls
16563 2017-06-27 Jakub Jelinek <jakub@redhat.com>
16565 * predict.c (test_prediction_value_range): Use -1U instead of -1
16566 to avoid narrowing conversion warning.
16567 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
16568 to avoid narrowing conversion warning.
16569 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
16571 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
16573 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
16575 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
16576 64bit configurations.
16577 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
16578 (SIZE_TYPE): Likewise.
16579 * config/vxworks.c (vxworks_emutls_var_fields): Use
16580 long_unsigned_type_node instead of unsigned_type_node as the offset
16581 field type, which is "pointer" mode in emutls.c.
16583 2017-06-27 Jakub Jelinek <jakub@redhat.com>
16586 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
16588 PR middle-end/81207
16589 * gimple-fold.c (replace_call_with_call_and_fold): Handle
16590 gimple_vuse copying separately from gimple_vdef copying.
16592 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16594 * value-prof.c (free_hist): Remove call to memset and the enclosing if
16597 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
16598 Olivier Hainque <hainque@adacore.com>
16600 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
16601 for all vxworks7 targets.
16602 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
16603 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
16604 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
16605 variations for VX6/VX7 and 32/64bits later on in ...
16606 (VXWORKS_LIB_SPEC): Leverage new macros.
16607 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
16608 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
16610 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
16612 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
16613 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
16615 2017-06-26 Carl Love <cel@us.ibm.com>
16617 * config/rs6000/rs6000-c.c: Add support for built-in functions
16618 vector bool char vec_reve (vector bool char);
16619 vector signed char vec_reve (vector signed char);
16620 vector unsigned char vec_reve (vector unsigned char);
16621 vector bool int vec_reve (vector bool int);
16622 vector signed int vec_reve (vector signed int);
16623 vector unsigned int vec_reve (vector unsigned int);
16624 vector bool long long vec_reve (vector bool long long);
16625 vector signed long long vec_reve (vector signed long long);
16626 vector unsigned long long vec_reve (vector unsigned long long);
16627 vector bool short vec_reve (vector bool short);
16628 vector signed short vec_reve (vector signed short);
16629 vector double vec_reve (vector double);
16630 vector float vec_reve (vector float);
16631 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
16632 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
16633 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
16634 (altivec_vreve): New pattern.
16635 * config/rs6000/altivec.h (vec_reve): New define.
16636 * doc/extend.texi (vec_rev): Update the built-in documentation file
16637 for the new built-in functions.
16639 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16641 PR tree-optimization/71815
16642 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
16644 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
16646 (slsr_process_phi): Likewise.
16647 (replace_uncond_cands_and_profitable_phis): Don't replace a
16648 multiply candidate with a stride of 1 (copy or cast).
16649 (phi_incr_cost): Call uses_consumed_by_stmt rather than
16651 (lowest_cost_path): Likewise.
16652 (total_savings): Likewise.
16654 2017-06-26 Richard Biener <rguenther@suse.de>
16657 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
16658 Use def_builtin_pure for all gather builtins.
16660 2017-06-26 Richard Biener <rguenther@suse.de>
16662 PR tree-optimization/81203
16663 * tree-tailcall.c (find_tail_calls): Do not move stmts into
16664 non-dominating BBs.
16666 2017-06-26 Marek Polacek <polacek@redhat.com>
16669 * doc/invoke.texi: Document -Wmultistatement-macros.
16671 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
16673 * doc/sourcebuild.texi (ARM-specific attributes): Document new
16674 arm_neon_ok_no_float_abi effective target.
16676 2017-06-26 Richard Biener <rguenther@suse.de>
16678 PR tree-optimization/80928
16679 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
16680 (copy_bbs): Set BB_DUPLICATED flag early.
16681 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
16683 (execute_on_shrinking_pred): Likewise.
16684 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
16685 BB_DUPLICATED blocks.
16686 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
16687 iterate over all PHIs considering removal of *gsi.
16689 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
16691 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
16694 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16696 * config/rs6000/rs6000-string.c: (expand_block_clear,
16697 do_load_for_compare, select_block_compare_mode,
16698 compute_current_alignment, expand_block_compare,
16699 expand_strncmp_align_check, expand_strn_compare,
16700 expand_block_move, rs6000_output_load_multiple)
16701 Move functions related to string/block move/compare
16702 to a separate file.
16703 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
16704 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
16705 for this function which is now used in two files.
16706 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
16707 * config.gcc: Add rs6000-string.o to extra_objs for
16708 targets powerpc*-*-* and rs6000*-*-*.
16710 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
16713 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
16714 32-bit, since indexed is not valid for DImode.
16715 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
16716 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
16717 (define_peephole2 for Altivec d-form load): Add 32-bit support.
16718 (define_peephole2 for Altivec d-form store): Likewise.
16721 * multiple_target.c (create_dispatcher_calls): Only create the
16722 dispatcher call if the function is the default clone of a
16723 versioned function.
16725 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
16727 PR middle-end/80902
16728 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
16729 a call, force the call to not be a tail call.
16731 2017-06-23 Jeff Law <law@redhat.com>
16733 * doc/contrib.texi: Add entry for Steven Pemberton's work on
16736 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
16738 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
16739 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
16740 handling for early expansion of vector shifts (sl,sr,sra,rl).
16741 (builtin_function_type): Add vector shift right instructions
16742 to the unsigned argument list.
16744 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
16746 rtl-optimizatoin/79286
16747 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
16748 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
16749 trap. PIC register plus a const unspec without offset can never trap.
16751 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
16753 * tree.h (builtin_structptr_type): New type.
16754 (builtin_structptr_types): Declare new array.
16755 * tree.c (builtin_structptr_types): New array.
16756 (free_lang_data, build_common_tree_nodes): Use it.
16758 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
16761 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
16764 2017-06-22 Matt Turner <mattst88@gmail.com>
16766 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
16767 Lake models to skylake case. Assume skylake for unknown
16768 models with clflushopt.
16770 2017-06-22 Jeff Law <law@redhat.com>
16772 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
16773 frame sizes that do not satisfy aarch64_uimm12_shift.
16775 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
16777 * profile-count.h (apply_probability,
16778 apply_scale, probability_in): Fix checks for zero.
16780 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16782 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
16783 * doc/cppdiropts.texi (-I @var{dir}): Document it.
16785 2016-06-22 Richard Biener <rguenther@suse.de>
16787 * tree-vect-loop.c (vect_model_reduction_cost): Handle
16788 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
16789 REDUC_MAX_EXPR support.
16790 (vectorizable_reduction): Likewise.
16791 (vect_create_epilog_for_reduction): Likewise.
16793 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
16795 * match.pd (A / (1 << B) -> A >> B): New.
16796 * generic-match-head.c: Include optabs-tree.h.
16797 * gimple-match-head.c: Likewise.
16798 * optabs-tree.h (target_supports_op_p): New.
16799 * optabs-tree.c (target_supports_op_p): New.
16801 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16803 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
16804 $gcc_cv_ld --help output.
16805 (gcc_cv_ld_demangle): Likewise.
16806 (gcc_cv_ld_eh_frame_hdr): Likewise.
16807 (gcc_cv_ld_pie): Likewise.
16808 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
16809 (gcc_cv_ld_buildid): Likewise.
16810 (gcc_cv_ld_sysroot): Likewise.
16811 (ld_bndplt_support): Likewise.
16812 (ld_pushpopstate_support): Likewise.
16813 * configure: Regenerate.
16814 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
16816 2017-06-21 Jakub Jelinek <jakub@redhat.com>
16819 * config/i386/sse.md (round<mode>2): Renumber match_dup and
16820 operands indexes to avoid gap between operands and match_dups.
16822 2017-06-21 Andrew Pinski <apinski@cavium.com>
16824 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
16825 Increment Arith_shift and Arith_shift_reg by 1.
16826 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
16828 * config/aarch64/aarch64.c (thunderx_tunings): Enable
16829 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
16830 (aarch64_strip_extend): Add new argument and test for it.
16831 (aarch64_cheap_mult_shift_p): New function.
16832 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
16833 add a cost if it is true.
16834 Update calls to aarch64_strip_extend.
16835 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
16837 2017-06-21 Andrew Pinski <apinski@cavium.com>
16839 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
16841 (thunderxt88): Likewise.
16842 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
16843 (thunderx_prefetch_tune): New variable.
16844 (thunderx2t99_prefetch_tune): Update for the correct values.
16845 (thunderxt88_tunings): New variable.
16846 (thunderx_tunings): Use thunderx_prefetch_tune instead of
16847 generic_prefetch_tune.
16848 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
16850 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16852 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
16853 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
16854 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
16855 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
16856 (aarch64_atomic_cas<mode>, GPI): Likewise.
16858 2017-06-21 Martin Liska <mliska@suse.cz>
16860 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
16861 statements on cold and hot labels.
16862 * predict.c (tree_estimate_probability_bb): Remove the
16863 prediction from this place.
16865 2017-06-21 Martin Liska <mliska@suse.cz>
16867 PR tree-optimization/79489
16868 * gimplify.c (maybe_add_early_return_predict_stmt): New
16870 (gimplify_return_expr): Call the function.
16871 * predict.c (tree_estimate_probability_bb): Remove handling
16873 * predict.def: Update comment about early return predictor.
16874 * gimple-predict.h (is_gimple_predict): New function.
16875 * predict.def: Change default value of early return to 66.
16876 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
16878 * passes.def: Put pass_strip_predict_hints to the beginning of
16881 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
16883 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
16884 FUNCTION_DECL declarations.
16885 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
16887 (dwaf2out_decl): Likewise.
16888 * godump.c (go_early_global_decl): Skip call to the real debug hook
16889 for FUNCTION_DECL declarations.
16890 * passes.c (rest_of_decl_compilation): Skip call to the
16891 early_global_decl debug hook for FUNCTION_DECL declarations, unless
16892 -fdump-go-spec is passed.
16894 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
16896 * config/i386/i386.c (struct builtin_isa): New field pure_p.
16897 Reorder for compactness.
16898 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
16899 (def_builtin_pure, def_builtin_pure2): New functions.
16900 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
16902 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
16904 * match.pd (nop_convert): New predicate.
16905 ((A +- CST1) +- CST2): Allow some NOP conversions.
16907 2017-06-21 Jakub Jelinek <jakub@redhat.com>
16910 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
16911 with ctors/dtors if GOVD_SHARED is set.
16913 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
16915 * config/aarch64/aarch64.md (movti_aarch64):
16916 Emit mov rather than orr.
16917 (movtf_aarch64): Likewise.
16918 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
16919 Emit mov rather than orr.
16921 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
16923 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
16924 Swap alternatives, make integer dup more expensive.
16926 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
16928 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
16929 Return true for non-tls symbols.
16931 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
16933 * config/aarch64/aarch64-cores.def (cortex-a55): New.
16934 (cortex-a75): Likewise.
16935 (cortex-a75.cortex-a55): Likewise.
16936 * config/aarch64/aarch64-tune.md: Regenerate.
16937 * doc/invoke.texi (-mtune): Document new values for -mtune.
16939 2017-06-21 Tom de Vries <tom@codesourcery.com>
16941 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
16942 stack_size feature.
16943 (Effective-Target Keywords, Other attributes): Suggest using
16944 dg-add-options stack_size feature to get stack limit in stack_size
16945 effective target documentation.
16947 2017-06-21 Julian Brown <julian@codesourcery.com>
16948 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16950 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
16951 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
16952 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
16954 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
16955 attribute type list for neon_multiply.
16956 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
16957 attribute type list for neon_multiply.
16958 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
16959 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
16960 attribute type list for neon_multiply.
16961 * config/arm/types.md (crypto_pmull): Add.
16962 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
16963 attribute type list.
16965 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
16967 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
16970 2017-06-20 Jakub Jelinek <jakub@redhat.com>
16972 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
16973 to make sure not to dereference a NULL cost_classes_ptr pointer.
16975 2017-06-20 Carl Love <cel@us.ibm.com>
16977 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16978 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16979 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16980 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16981 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16982 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16983 VMULOSW): New enum "unspec" values.
16984 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
16985 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
16986 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16987 altivec_vmulosw): New patterns.
16988 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16989 VMULOSW): Add definitions.
16991 2017-06-20 Julia Koval <julia.koval@intel.com>
16993 * config/i386/i386.c: Fix rounding expand for new pattern.
16994 * config/i386/subst.md: Fix pattern (parallel -> unspec).
16996 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
16998 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
16999 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
17001 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
17003 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
17006 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
17008 * config/aarch64/aarch64-cores.def: Rearrange to sort by
17009 architecture, then by implementer ID.
17010 * config/aarch64/aarch64-tune.md: Regenerate.
17012 2017-06-20 Richard Biener <rguenther@suse.de>
17014 PR middle-end/81097
17015 * fold-const.c (split_tree): Fold to type before negating.
17017 2017-06-20 David Malcolm <dmalcolm@redhat.com>
17019 * diagnostic-show-locus.c
17020 (selftest::test_fixit_deletion_affecting_newline): New function.
17021 (selftest::diagnostic_show_locus_c_tests): Call it.
17023 2017-06-20 Andreas Schwab <schwab@suse.de>
17026 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
17029 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
17031 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
17032 __ARM_FEATURE_COPROC according to support.
17034 2017-06-20 Jakub Jelinek <jakub@redhat.com>
17036 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
17037 Rewritten to avoid overflow for > 32-bit pointers.
17040 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
17041 by removing enum keyword.
17042 (ubsan_type_descriptor): Likewise. Formatting fix.
17045 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
17046 splitter): Require TARGET_SSE2 in the condition.
17048 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
17051 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
17052 for doing vector set of SFmode on ISA 3.0.
17053 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
17054 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
17056 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
17057 SFmode value into a V4SF variable that was extracted from another
17058 V4SF variable without converting the element to double precision
17059 and back to single precision vector format.
17060 (vsx_insert_extract_v4sf_p9_2): Likewise.
17062 2017-06-19 Jakub Jelinek <jakub@redhat.com>
17064 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
17065 in UWHI to avoid undefined overflow.
17068 * ubsan.h (enum ubsan_encode_value_phase): New.
17069 (ubsan_encode_value): Change second argument to
17070 enum ubsan_encode_value_phase with default value of
17071 UBSAN_ENCODE_VALUE_GENERIC.
17072 * ubsan.c (ubsan_encode_value): Change second argument to
17073 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
17074 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
17075 create_tmp_var_raw instead of create_tmp_var and use a
17077 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
17078 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
17079 ubsan_encode_value callers.
17082 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
17083 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
17084 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
17086 2017-06-19 Richard Biener <rguenther@suse.de>
17088 PR middle-end/81118
17089 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
17090 estimates if we changed anything.
17092 2017-06-19 Richard Biener <rguenther@suse.de>
17094 PR tree-optimization/80887
17095 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
17096 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
17097 simplified lookups, then reset mprts_hook.
17098 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
17100 (try_to_simplify): Likewise.
17102 2017-06-19 Martin Liska <mliska@suse.cz>
17105 * gimplify.c (gimplify_switch_expr):
17106 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
17108 2017-06-19 Martin Liska <mliska@suse.cz>
17110 * doc/install.texi: Document that PGO runs in 4 stages.
17112 2017-06-19 Martin Liska <mliska@suse.cz>
17115 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
17116 to dispatcher function name.
17117 * multiple_target.c (replace_function_decl): New function.
17118 (create_dispatcher_calls): Redirect both edges and references.
17120 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
17122 * profile-count.c (profile_count::dump): Dump quality.
17123 (profile_count::differs_from_p): Update for unsigned val.
17124 * profile-count.h (profile_count_quality): New enum.
17125 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
17127 2017-06-19 Richard Biener <rguenther@suse.de>
17129 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
17130 struct function as arg.
17131 (estimate_numbers_of_iterations): Export overload with loop arg.
17132 (free_numbers_of_iterations_estimates_loop): Use an overload of
17133 free_numbers_of_iterations_estimates instead.
17134 * tree-cfg.c (remove_bb): Adjust.
17135 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
17136 * tree-parloops.c (gen_parallel_loop): Likewise.
17137 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
17139 (tree_unroll_loops_completely): Likewise.
17140 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
17141 Use an overload instead and export.
17142 (estimated_loop_iterations): Adjust.
17143 (max_loop_iterations): Likewise.
17144 (likely_max_loop_iterations): Likewise.
17145 (estimate_numbers_of_iterations): Take struct function as arg
17147 (loop_exits_before_overflow): Adjust.
17148 (free_numbers_of_iterations_estimates_loop): Use an overload.
17149 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
17150 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
17152 2017-06-19 Richard Biener <rguenther@suse.de>
17155 * ipa-prop.c (find_constructor_constant_at_offset): Handle
17156 RANGE_EXPR conservatively.
17158 2017-06-16 Carl Love <cel@us.ibm.com>
17160 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17161 definitions for vec_float, vec_float2, vec_floato,
17162 vec_floate built-ins.
17163 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
17164 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
17166 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
17167 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
17168 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
17169 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
17170 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
17171 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
17172 vec_floato): Add builtin defines.
17173 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
17174 Update the built-in documentation file for the new built-in
17177 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17179 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
17180 (mthumb): Mark as the negative of -marm.
17182 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17184 * doc/invoke.texi (ARM Options, -mcpu): Document supported
17186 (ARM Options, -mtune): Document that this accepts the same
17187 extension options as -mcpu.
17188 (ARM Options, -mfpu): Document addition of -mfpu=auto.
17190 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17192 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
17193 permitted extensions.
17195 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17197 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
17198 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
17199 (armv8-m.main): Add option +nodsp.
17200 * config/arm/arm-cpu-cdata.h: Regenerated.
17202 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17204 * config/arm/t-fuchsia: New file.
17205 * config.gcc (arm*-*-fuchsia*): Use it.
17207 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17209 * config/arm/t-symbian: Rewrite for new option infrastructure.
17211 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17213 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
17214 (MULTILIB_REQUIRED): Likewise.
17216 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17218 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
17219 (MULTILIB_RESUE): Likewise.
17220 (MULTILIB_MATCHES): Likewise.
17221 (MULTLIB_REQUIRED): Likewise.
17223 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17225 * config/arm/t-rtems: Rewrite for new option framework.
17227 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17229 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
17230 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
17231 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
17232 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
17233 * config/arm/t-multilib: ... here.
17234 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
17235 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
17236 armv7-a and armv8*-a when A-profile libraries have not been built.
17237 * config/arm/t-rmprofile: Rewrite.
17239 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17241 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
17242 with a backslash. Remove the backslash after substituting unescaped
17244 * doc/fragments.texi (MULTILIB_REUSE): Document it.
17246 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17248 * config.gcc: (arm*-*-*): When building a-profile libraries, force
17249 the driver to pass through the default setting of -mfloat-abi.
17250 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
17252 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
17253 (all_feat_combs): New rule.
17254 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
17256 * config/arm/t-aprofile: Rewrite.
17258 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17260 * config/arm/arm.h (FPUTYPE_AUTO): Define.
17261 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
17262 fpu is not specified by the user/command-line.
17263 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
17264 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
17265 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
17266 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
17267 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
17268 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
17270 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17272 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
17273 * config/arm/t-arm-elf: Rewritten.
17275 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17277 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
17278 have some floating-point instructions.
17279 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
17280 (TARGET_MAYBE_HARD_FLOAT): New macro.
17281 * config/arm/arm-builtins.c (arm_init_builtins): Use
17282 TARGET_MAYBE_HARD_FLOAT.
17283 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
17285 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17287 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
17288 (configargs.h): Include it.
17289 (arm_print_hint_for_fpu_option): New function.
17290 (arm_parse_fpu_option): New function.
17291 (candidate_extension): New class.
17292 (arm_canon_for_multilib): New function.
17293 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
17294 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
17295 (ARCH_CANONICAL_SPECS): New macro.
17296 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
17298 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17300 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
17301 are set after handling multilib fragments. Set target_cpu_default2
17304 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17306 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
17308 (arm*-*-*): Set target_cpu_default2 to a quoted string.
17309 * config/arm/parsecpu.awk (check_cpu): Validate any extension
17311 (check_arch): Likewise.
17312 * config/arm/arm.c (arm_configure_build_target): Handle
17313 TARGET_CPU_DEFAULT being a string constant. Scan any feature
17314 options in the default.
17316 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17318 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
17319 when an option is an alias of another.
17320 * config/arm/parsecpu.awk (optalias): New parser token.
17321 (gen_comm_data): Mark non-alias options as such. Emit entries
17322 for extension aliases.
17323 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
17324 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
17325 (armv6kz, armv6zk, armv6t2): Likewise.
17326 (armv7): Make vfpv3-d16 an alias.
17327 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
17329 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
17330 Sort in canonical order.
17331 (armv8-a): Sort in canonical order.
17332 (armv8.1-a, armv8.2-a): Likewise.
17333 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
17335 (cortex-a9): Sort in canonical order.
17336 * config/arm/arm.c (selftests.h): Include it.
17337 (arm_test_cpu_arch_data): New function.
17338 (arm_run_self_tests): New function.
17339 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
17340 (targetm): Move declaration to the end of the file.
17341 * arm-cpu-cdata.h: Regenerated.
17343 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17345 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
17346 call to target_mode_check describing the type of option passed.
17347 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
17348 (arm_target_thumb_only): Use arm_parse_arch_option_name or
17349 arm_parse_cpu_option_name to match parameters against list of
17351 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
17352 arm_arch_core_flags data structure.
17353 * config/arm/arm-cpu_cdata.h: Regenerated.
17355 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17357 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
17359 (arm_print_hint_for_cpu_option): Likewise.
17360 (arm_print_hint_for_arch_option): Likewise.
17361 (arm_parse_cpu_option_name): Likewise.
17362 (arm_parse_arch_option_name): Likewise.
17363 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
17364 of entries in the all_fpus list.
17365 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
17366 (arm_parse_cpu_option_name): Declare.
17367 (arm_parse_arch_option_name): Declare.
17368 (arm_parse_option_features): Declare.
17369 (arm_intialize_isa): Declare.
17370 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
17372 (gen_comm_data): ... here. Make definitions non-static.
17373 * config/arm/arm-cpu-data.h: Regenerated.
17374 * config/arm/arm-cpu-cdata.h: Regenerated.
17376 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17378 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
17379 (cpu_arch_extension): New structure.
17380 (cpu_arch_option, arch_option, cpu_option): New structures.
17381 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
17382 architecture types.
17383 (gen_data): Generate new format data tables.
17384 * config/arm/arm.c (cpu_tune): New structure.
17385 (cpu_option, processors): Delete.
17386 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
17387 (arm_print_hint_for_cpu_option): ... this and ...
17388 (arm_print_hint_for_arch_option): ... this.
17389 (arm_parse_arch_cpu_name): Delete. Replace with ...
17390 (arm_parse_cpu_option_name): ... this and ...
17391 (arm_parse_arch_option_name): ... this.
17392 (arm_unrecognized_feature): Change type of target parameter to
17394 (arm_parse_arch_cpu_features): Delete. Replace with ...
17395 (arm_parse_option_features): ... this.
17396 (arm_configure_build_target): Rework to use new configuration data
17398 (arm_print_tune_info): Rework for new configuration data tables.
17399 * config/arm/arm-cpu-data.h: Regenerated.
17400 * config/arm/arm-cpu.h: Regenerated.
17402 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17404 * Makefile.in (OBJS): Move sbitmap.o from here ...
17405 (OBJS-libcommon): ... to here.
17407 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17409 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
17410 (ISA_ALL_CRYPTO): New macro.
17411 (ISA_ALL_SIMD): New macro
17412 (ISA_ALL_FP): New macro.
17413 * config/arm/arm.c (fpu_bitlist): Update initializer.
17414 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
17416 (arm9e): Add fpu. Add option for nofp
17417 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
17418 (arm926ej-s, arm1026ej-s): Likewise.
17419 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
17420 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
17421 neon-fp16, neon-vfpv4, nofp and nosimd.
17422 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
17423 (cortex-a8): Add fpu. Add option for nofp.
17424 (cortex-a9): Add fpu. Add options for nosimd and nofp.
17425 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
17426 (cortex-r4f): Add fpu.
17427 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
17428 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
17430 (cortex-r8): Likewise.
17431 (cortex-m4): Add fpu. Add option for nofp.
17432 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
17433 (cortex-a17.cortex-a7): Likewise.
17434 (cortex-a32): Add fpu. Add options for crypto and nofp.
17435 (cortex-a35, cortex-a53): Likewise.
17436 (cortex-a57): Add fpu. Add option for crypto.
17437 (cortex-a72, cortex-a73): Likewise.
17438 (exynos-m1): Likewise.
17439 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
17440 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
17441 (cortex-m33): Add fpu. Add option for nofp.
17442 * config/arm/arm-cpu-cdata.h: Regenerated
17443 * config/arm/arm-cpu-data.h: Regenerated.
17445 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17447 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
17448 (armv5te, armv5tej): Likewise.
17449 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
17450 (armv7): Add options fp and vfpv3-d16.
17451 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
17452 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
17454 (armv7ve): Likewise.
17455 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
17456 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
17457 (armv8-a): Add nocrypto option.
17458 (armv8.1-a, armv8.2-a): Likewise.
17459 (armv8-m.main): add options fp, fp.dp and nofp.
17461 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17463 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
17465 (armv8-a+crc): Delete.
17466 (armv8.1-a): Add options simd, crypto and nofp.
17467 (armv8.2-a): Add options fp16, simd, crypto and nofp.
17468 (armv8.2-a+fp16): Delete.
17469 (armv8-m.main): Add option dsp.
17470 (armv8-m.main+dsp): Delete.
17471 (cortex-a8): Add fpu. Add nofp option.
17472 (cortex-a9): Add fpu. Add nofp and nosimd options.
17473 * config/arm/parsecpu.awk (gen_data): Generate option tables and
17474 link to main cpu and architecture data structures.
17475 (gen_comm_data): Only put isa attributes from the main architecture
17477 (option): New statement for architecture and CPU entries.
17478 * arm.c (struct cpu_option): New structure.
17479 (struct processors): Add entry for options.
17480 (arm_unrecognized_feature): New function.
17481 (arm_parse_arch_cpu_name): Ignore any characters after the first
17483 (arm_parse_arch_cpu_feature): New function.
17484 (arm_configure_build_target): Separate out any CPU and architecture
17485 features and parse separately. Don't error out if -mfpu=auto is
17486 used with only an architecture string.
17487 (arm_print_asm_arch_directives): New function.
17488 (arm_file_start): Call it.
17489 * config/arm/arm-cpu-cdata.h: Regenerated.
17490 * config/arm/arm-cpu-data.h: Likewise.
17491 * config/arm/arm-tables.opt: Likewise.
17493 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17495 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
17496 assembler when it is not -mfpu=auto.
17498 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17500 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
17501 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
17502 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
17503 (ASM_CPU_SPEC): Rewrite.
17504 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
17505 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
17506 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
17507 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
17508 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
17509 copied string is NUL-terminated. Also strip any characters prefixed
17511 (arm_rewrite_selected_arch): New function.
17512 (arm_rewrite_march): New function.
17514 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
17516 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
17517 (x_arm_cpu_string, x_arm_tune_string): Likewise.
17518 (march, mcpu, mtune): Convert to string-based options.
17519 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
17520 (arm_parse_arch_cpu_name): New function.
17521 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
17522 identify selected architecture or CPU.
17523 (arm_option_save): New function.
17524 (TARGET_OPTION_SAVE): Redefine.
17525 (arm_option_restore): Restore string options.
17526 (arm_option_print): Print string options.
17528 2017-06-16 Martin Sebor <msebor@redhat.com>
17530 PR tree-optimization/80933
17531 PR tree-optimization/80934
17532 * builtins.c (fold_builtin_3): Do not handle bcmp here.
17533 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
17534 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
17535 (gimple_fold_builtin): Call them.
17537 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17539 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
17540 as unlikely; update profile.
17542 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17544 * predict.c (force_edge_cold): Handle declaring edges impossible
17547 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17549 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
17551 (try_unroll_loop_completely): Fix reporting.
17553 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17555 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
17557 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
17560 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
17561 if given a non-constant argument for an intrinsic which requires a
17564 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17566 * profile.c (compare_freqs): New function.
17567 (branch_prob): Sort edge list.
17568 (find_spanning_tree): Assume that the list is priority sorted.
17570 2017-06-16 Richard Biener <rguenther@suse.de>
17572 PR tree-optimization/81090
17573 * passes.def (pass_record_bounds): Remove.
17574 * tree-pass.h (make_pass_record_bounds): Likewise.
17575 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
17576 make_pass_record_bounds): Likewise.
17577 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
17578 not free niter estimates at the beginning but at the end.
17579 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
17581 2017-06-16 Richard Biener <rguenther@suse.de>
17583 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
17584 initializer to workaround ICE in host GCC 4.8.
17586 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
17588 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
17590 (clone_inlined_nodes): Update.
17592 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17594 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
17595 prefetch settings, and enable prefetching by default at -O3.
17597 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17599 * config/aarch64/aarch64.c (aarch64_override_options_internal):
17600 Set flag_prefetch_loop_arrays according to tuning data.
17602 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17604 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
17605 New tune structure.
17606 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
17607 [Unrelated to main purpose of the patch] Place the pointer field last
17608 to enable type checking errors when tune structure are wrongly merged.
17609 * config/aarch64/aarch64.c (generic_prefetch_tune,)
17610 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
17611 (thunderx2t99_prefetch_tune): New tune constants.
17612 (tune_params *_tunings): Update all tunings (no functional change).
17613 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
17614 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
17615 from tunings structures.
17617 2017-06-16 Jakub Jelinek <jakub@redhat.com>
17620 * ubsan.c (instrument_null): Add T argument, use it instead
17621 of computing it based on IS_LHS.
17622 (instrument_object_size): Likewise.
17623 (pass_ubsan::execute): Adjust instrument_null and
17624 instrument_object_size callers to pass gimple_get_lhs or
17625 gimple_assign_rhs1 result to it. Use instrument_null instead of
17626 calling get_base_address and instrument_mem_ref. Handle
17627 aggregate call arguments for object-size sanitization.
17629 2017-06-16 Yury Gribov <tetra2005@gmail.com>
17631 PR tree-optimization/81089
17632 * tree-vrp.c (is_masked_range_test): Validate operands of
17635 2017-06-15 Martin Sebor <msebor@redhat.com>
17638 * dumpfile.c (dump_register): Avoid calling memset to initialize
17639 a class with a default ctor.
17640 * gcc.c (struct compiler): Remove const qualification.
17641 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
17642 * hash-table.h: Ditto.
17643 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
17645 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
17646 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
17648 * params.h (struct param_info): Make struct members non-const.
17649 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
17650 with default initialization.
17651 * vec.h (vec_copy_construct, vec_default_construct): New helper
17653 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
17654 with vec_copy_construct.
17655 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
17656 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
17657 * doc/invoke.texi (-Wclass-memaccess): Document.
17659 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17661 * emit-rtl.h (is_leaf): Update comment about local
17662 register allocator.
17664 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
17667 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
17668 for a variable to have a section before checking if the section has a
17670 Set section to.persistent if persistent attribute is set.
17671 Warn if .persistent attribute is used on an automatic variable.
17673 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
17675 PR rtl-optimization/80474
17676 * reorg.c (update_block): Do not ignore instructions in a delay slot.
17678 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
17680 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
17683 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
17685 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
17686 (casesi): Emit bounds checking as RTL.
17687 (casesi_internal_mips16_<mode>): Remove bounds checking.
17689 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
17691 * config/xtensa/xtensa.c (xtensa_option_override): Append
17692 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
17693 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
17694 xtensa_doloop_hooks): Define unconditionally.
17695 (xtensa_reorg_loops): Only call reorg_loops in the presence of
17697 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
17698 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
17699 for it in xtensa_option_override.
17700 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
17701 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
17703 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
17705 * doc/cppopts.texi: Document '-' special value to -MF.
17707 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
17709 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
17710 (cortex_a53_fconst): Likewise.
17711 (cortex_a53_fpmul): Likewise.
17712 (cortex_a53_f_load_64): Likewise.
17713 (cortex_a53_f_load_many): Likewise.
17714 (cortex_a53_advsimd_alu): Likewise.
17715 (cortex_a53_advsimd_alu_q): Likewise.
17716 (cortex_a53_advsimd_mul): Likewise.
17717 (cortex_a53_advsimd_mul_q): Likewise.
17718 (fpmac bypass): Add new bypass for fpmac-fpmac case.
17719 Add missing fmul, r2f_cvt and fconst cases.
17721 2017-06-14 Richard Biener <rguenther@suse.de>
17723 PR middle-end/81088
17724 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
17726 (fold_binary_loc): When associating do not treat pre-existing
17727 TREE_OVERFLOW on literal constants as a reason to allow
17728 TREE_OVERFLOW on associated literal constants.
17730 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
17732 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
17733 (MASK_FEATURES): New macro.
17734 * config/sparc/sparc.c (sparc_option_override): Remove the special
17735 handling of -mfpu and generalize it to all MASK_FEATURES switches.
17737 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
17739 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
17740 a division of 0 if non-call exceptions are enabled.
17742 2017-06-14 Andrew Pinski <apinski@cavium.com>
17743 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17746 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
17747 Improve vector initialization code gen for only variable case.
17749 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
17751 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
17753 2017-06-14 Richard Biener <rguenther@suse.de>
17755 PR tree-optimization/81083
17756 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
17759 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17761 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
17762 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
17763 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
17764 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
17765 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
17766 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
17768 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17770 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
17771 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
17773 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17775 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
17777 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17779 * config/rs6000/t-rtems: Don't handle SPE.
17781 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17783 * config/rs6000/t-linux: Don't handle SPE.
17785 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17787 * config/rs6000/eabispe.h: Delete file.
17789 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17791 * config/rs6000/t-spe: Delete file.
17793 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17795 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
17796 (rs6000_legitimate_offset_address_p): Return false for anything in
17797 V2SImode or V2SFmode.
17799 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
17801 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
17802 except V2SF and V2SI. Rearrange the vector modes, and add comments.
17803 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
17805 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
17806 (rs6000_legitimate_offset_address_p): Ditto.
17807 (rs6000_emit_move): Ditto.
17808 (rs6000_init_builtins): Remove V4HI_type_node.
17810 2017-06-13 Martin Liska <mliska@suse.cz>
17813 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
17814 (gate_asan): Likewise.
17815 * asan.h (asan_no_sanitize_address_p): Remove the function.
17816 (sanitize_flags_p): New function.
17817 * builtins.def: Fix coding style.
17818 * common.opt: Use renamed enum value.
17819 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
17820 * doc/extend.texi: Document no_sanitize attribute.
17821 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
17822 to SANITIZE_UNDEFINED_NONDEFAULT.
17823 * gcc.c (sanitize_spec_function): Use the renamed enum value.
17824 * gimple-fold.c (optimize_atomic_compare_exchange_p):
17825 Use sanitize_flags_p.
17826 * gimplify.c (gimplify_function_tree): Likewise.
17827 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
17828 * opts.c (parse_no_sanitize_attribute): New function.
17829 (common_handle_option): Use renamed enum value.
17830 * opts.h (parse_no_sanitize_attribute): Declare.
17831 * tree.c (sanitize_flags_p): New function.
17832 * tree.h: Declared here.
17833 * tsan.c: Use sanitize_flags_p.
17834 * ubsan.c (ubsan_expand_null_ifn): Likewise.
17835 (instrument_mem_ref): Likewise.
17836 (instrument_bool_enum_load): Likewise.
17837 (do_ubsan_in_current_function): Remove the function.
17838 (pass_ubsan::execute): Use sanitize_flags_p.
17839 * ubsan.h: Remove do_ubsan_in_current_function
17840 * tree-cfg.c (print_no_sanitize_attr_value): New function.
17841 (dump_function_to_file): Use it here.
17843 2017-06-13 Martin Jambor <mjambor@suse.cz>
17845 PR tree-optimization/80803
17846 PR tree-optimization/81063
17847 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
17848 (propagate_subaccesses_across_link): Enqueue subtree whenever
17849 necessary instead of relying on the caller.
17851 2017-06-13 Martin Jambor <mjambor@suse.cz>
17853 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
17854 that have a first_link.
17855 (sort_and_splice_var_accesses): Do not check first_link before
17857 (subtree_mark_written_and_enqueue): Likewise.
17858 (propagate_all_subaccesses): Likewise and do not stop at first
17859 parent with a first_link.
17861 2017-06-13 Martin Jambor <mjambor@suse.cz>
17863 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
17866 2017-06-13 Yury Gribov <tetra2005@gmail.com>
17868 * match.pd: New pattern.
17870 2017-06-13 Yury Gribov <tetra2005@gmail.com>
17872 * tree-vrp.c (is_masked_range_test): New function.
17873 (register_edge_assert_for): Determine ranges for
17876 2017-06-13 Yury Gribov <tetra2005@gmail.com>
17878 PR tree-optimization/67328
17879 * fold-const.c (maskable_range_p): New function.
17880 (build_range_check): Generate bittests if possible.
17882 2017-06-13 Martin Liska <mliska@suse.cz>
17884 * gimple-pretty-print.c (dump_probability): Add new argument.
17885 (dump_edge_probability): Dump both probability and count.
17886 (dump_gimple_label): Likewise.
17887 (dump_gimple_bb_header): Likewise.
17889 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
17892 * config/avr/avr-devices.c: Fix indentation.
17893 * config/avr/gen-avr-mmcu-specs.c: Dito.
17895 2017-06-13 Richard Biener <rguenther@suse.de>
17897 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
17898 instead get vector type from stmt_info.
17899 (vectorizable_reduction): Adjust. Remove dead code.
17901 2017-06-13 Richard Biener <rguenther@suse.de>
17903 PR middle-end/81065
17904 * fold-const.c (extract_muldiv_1): Remove bogus distribution
17905 case of C * (x * C2 + C3).
17906 (fold_addr_of_array_ref_difference): Properly fold index difference.
17908 2017-06-12 David S. Miller <davem@davemloft.net>
17911 * config/sparc/sparc.md (return expander): Emit frame blockage if
17912 function uses alloca.
17914 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
17916 * combine.c (make_field_assignment): Check len rather than the mode
17917 precision when calling force_to_mode.
17919 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
17921 Support multilibs and devices that see flash in RAM address range.
17924 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
17925 (avr_mcu_t) <flash_pm_offset>: New field.
17926 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
17927 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
17928 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
17929 (AVR_TINY_PM_OFFSET): Remove macro.
17930 * config/avr/avr.opt (-mshort-calls): New option.
17931 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
17932 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
17933 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
17934 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
17935 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
17936 instead of avr_arch->have_jmp_call.
17937 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
17938 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
17939 avr_arch->flash_pm_offset to define.
17940 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
17941 new field flash_pm_offset. Add entry for avrxmega3.
17942 (avr_texinfo): Add entry for avrxmega3.
17943 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
17944 attiny212, attiny214,
17945 attiny412, attiny414, attiny416, attiny417,
17946 attiny814, attiny816, attiny817,
17947 attiny1614, attiny1616, attiny1617,
17948 attiny3214, attiny3216, attiny3217.
17949 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
17950 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
17951 (avr_print_operand_address) [AVR_TINY]: Same.
17952 (avr_asm_init_sections) <readonly_data_section>: Only patch
17953 callback if avr_arch->flash_pm_offset = 0.
17954 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
17955 for rodata if avr_arch->flash_pm_offset != 0.
17956 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
17957 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
17958 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
17959 (m_options): Append opt_rcall.
17960 (m_dirnames): Append dir_rcall.
17961 * config/avr/t-multilib: Regenerate.
17963 * configure.ac [target=avr]: Check whether avrxmega3 default
17964 linker description file works as needed.
17965 * configure: Regenerate.
17966 * doc/avr-mmcu.texi: Regenerate.
17967 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
17968 <__AVR_ARCH__>: Document avrxmega3 and 103.
17969 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
17970 <__AVR_SHORT_CALLS__>: Document it.
17971 <__AVR_PM_BASE_ADDRESS__>: Document it.
17972 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
17973 (AVR Variable Attributes) <progmem>: Document this is
17974 not needed for avrxmega3.
17975 (AVR Named Address Spaces) <__flash>: Dito.
17977 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
17979 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
17981 2017-06-12 Doug Rupp <rupp@adacore.com>
17983 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
17984 Append vxworks-stdint.h to the tm_file list.
17985 * config/vxworks-stdint.h: New file.
17987 2017-06-12 Martin Liska <mliska@suse.cz>
17989 PR tree-optimization/81041
17990 * tree-profile.c (gimple_gen_ic_func_profiler):
17991 Create an extra BB in profile-generate
17992 (gimple_gen_time_profiler): Likewise.
17994 2017-06-12 Jakub Jelinek <jakub@redhat.com>
17996 PR tree-optimization/81003
17997 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
17998 (update_range_test): Use it instead of force_gimple_operand_gsi.
18000 2017-06-12 Richard Biener <rguenther@suse.de>
18002 PR tree-optimization/81053
18003 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
18004 with backedge value not defined in loop. Simplify def stmt
18007 2017-06-11 Tom de Vries <tom@codesourcery.com>
18010 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
18012 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
18013 nvptx_cannot_force_const_mem.
18015 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
18017 * opts.c (finish_options): Move test for flag_split_stack after
18018 it has been initialized.
18020 2017-06-11 Jason Merrill <jason@redhat.com>
18022 * tree.h (id_equal): New.
18023 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
18024 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
18025 instead of strcmp of IDENTIFIER_POINTER.
18027 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
18029 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
18030 (mark_all_inlined_calls_cdtor): Fix formating.
18031 (inline_transform): Rescale profile before inlining.
18033 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
18035 * cgraph.h (cgraph_edge::clone): Update prototype.
18036 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
18037 (cgraph_node::create_clone): Update.
18038 (cgraph_node::create_version_clone): Update.
18039 * tree-inline.c (copy_bb): Update.
18040 (expand_call_inline): Update.
18042 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
18044 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
18045 factored out from ...
18046 (rs6000_emit_prologue): ... here.
18048 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
18050 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
18051 factored out from ...
18052 (rs6000_emit_prologue): ... here.
18054 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
18056 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
18058 (handle_missing_profiles): Fix computation of tp_first_run.
18059 (counts_to_freqs): Do not touch freqs when count is 0.
18061 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
18063 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
18066 2017-06-10 Tom de Vries <tom@codesourcery.com>
18068 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
18069 attributes): Document signal effective target.
18071 2017-06-10 Tom de Vries <tom@codesourcery.com>
18073 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
18074 Document effective target stack_size.
18076 2017-06-09 David Malcolm <dmalcolm@redhat.com>
18078 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
18079 to the edit_context if they can be auto-applied.
18081 2017-06-9 Ian Lance Taylor <iant@golang.org>
18083 * opts.c (finish_options): If -fsplit-stack, disable implicit
18084 -forder-blocks-and-partition.
18085 * doc/invoke.texi (Optimize Options): Document that when using
18086 -fsplit-stack -forder-blocks-and-partition is not implicitly
18089 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
18091 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
18092 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
18093 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
18094 * builtins.def (abort, trap, unreachable): Declare cold.
18095 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
18096 * tree-core.h (ECF_COLD): New.
18097 * tree.c (set_call_expr_flags): Handle ECF_COLD.
18098 (build_common_builtin_nodes): Mark unreachable and abort as cold.
18100 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
18102 * predict.c (unlikely_executed_stmt_p): Cleanup.
18104 2017-06-09 Richard Biener <rguenther@suse.de>
18106 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
18107 model if the ref is always written to.
18109 2017-06-09 Tamar Christina <tamar.christina@arm.com>
18111 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
18113 2017-06-09 Tamar Christina <tamar.christina@arm.com>
18115 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
18118 2017-06-09 Tom de Vries <tom@codesourcery.com>
18121 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
18122 "target cannot support label values" when encountering LABEL_REF.
18124 2017-06-09 Martin Liska <mliska@suse.cz>
18126 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
18127 (gimple_gen_ic_func_profiler): Emit direct comparison
18128 of __gcov_indirect_call_callee with NULL.
18129 (gimple_gen_time_profiler): Change probability from
18130 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
18132 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
18134 * profile.c (edge_gcov_counts): Turn to pointer.
18135 (compute_branch_probabilities, compute_branch_probabilities): Update.
18136 (branch_prob): Do not clear edge_gcov_count.
18137 * profile.h (edge_gcov_counts): Turn to pointer.
18138 (edge_gcov_count): Update.
18140 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
18142 * gimple.h (gimple_check_failed): Mark cold.
18144 2017-06-09 Richard Biener <rguenther@suse.de>
18146 PR tree-optimization/66623
18147 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
18148 refactor check_reduction into two parts, properly computing
18149 whether we have to check reduction validity for outer loop
18152 2017-06-09 Richard Biener <rguenther@suse.de>
18154 PR tree-optimization/79483
18155 * graphite-scop-detection.c (order): New global.
18156 (get_order): Compute bb to order mapping that satisfies code
18157 generation constraints.
18158 (cmp_pbbs): New helper.
18159 (build_scops): Start domwalk at entry block, sort generated
18162 2017-06-09 Richard Biener <rguenther@suse.de>
18164 PR middle-end/81007
18165 * ipa-polymorphic-call.c
18166 (ipa_polymorphic_call_context::restrict_to_inner_class):
18167 Skip FIELD_DECLs with error_mark_node type.
18168 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
18171 2017-06-09 Martin Liska <mliska@suse.cz>
18173 * predict.c (struct branch_predictor): New struct.
18174 (test_prediction_value_range): New test.
18175 (predict_c_tests): New function.
18176 * selftest-run-tests.c (selftest::run_tests): Run the function.
18177 * selftest.h: Declare new tests.
18179 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
18182 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
18183 gen_add3_insn did not fail.
18184 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
18185 r0, construct that number in a temporary reg and add that reg to r0.
18186 If asked to put the result in r0 as well, fail.
18188 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
18190 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
18191 for early expansion of vec_eqv.
18193 2017-06-08 Jakub Jelinek <jakub@redhat.com>
18195 PR middle-end/81005
18196 * ubsan.c (instrument_null): Avoid pointless code temporary.
18197 (pass_ubsan::execute): Instrument aggregate arguments of calls.
18199 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
18203 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
18206 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
18209 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18211 * predict.c (unlikely_executed_edge_p): Move ahead.
18212 (probably_never_executed_edge_p): Use it.
18214 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
18216 PR middle-end/79988
18217 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
18218 gimple_call_builtin_p call.
18220 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18222 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
18223 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
18224 rtl_check_failed_type2, rtl_check_failed_code1,
18225 rtl_check_failed_code2, rtl_check_failed_code_mode,
18226 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
18227 rtvec_check_failed_bounds, rtl_check_failed_flag,
18228 _fatal_insn_not_found, _fatal_insn): Likewise.
18229 * tree.h (tree_contains_struct_check_failed,
18230 tree_check_failed, tree_not_check_failed,
18231 tree_class_check_failed, tree_range_check_failed,
18232 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
18233 tree_vec_elt_check_failed, phi_node_elt_check_failed,
18234 tree_operand_check_failed, omp_clause_check_failed,
18235 omp_clause_operand_check_failed, omp_clause_range_check_failed):
18238 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18240 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
18241 flag_branch_probabilities.
18242 * ipa-inline.c (edge_badness): Likewise.
18243 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
18244 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
18245 * predict.c (maybe_hot_frequency_p): Likewise.
18246 (probably_never_executed): Likewise.
18247 * sched-ebb.c (schedule_ebbs): Likewise.
18248 * sched-rgn.c (find_single_block_region): Likewise.
18249 * tracer.c (tail_duplicate): Likewise.
18251 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18253 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
18254 longer requires x_flag_profile_use.
18256 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18258 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
18259 instead of flag_reorder_blocks_and_partition.
18260 * dbxout.c (dbxout_function_end): Likewise.
18261 * dwarf2out.c (gen_subprogram_die): Likewise.
18262 * haifa-sched.c (sched_create_recovery_edges): Likewise.
18263 * hw-doloop.c (reorg_loops): Likewise.
18264 * varasm.c (assemble_start_function,
18265 assemble_end_function): Likewise.
18266 (decide_function_section): Do not check for
18267 flag_reorder_blocks_and_partition.
18269 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
18271 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
18273 (chkp_get_hard_register_fake_addr_expr): Ditto.
18274 (chkp_build_addr_expr): Add check for hard reg case.
18275 (chkp_parse_array_and_component_ref): Ditto.
18276 (chkp_find_bounds_1): Ditto.
18277 (chkp_process_stmt): Don't generate bounds store for
18280 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18282 * predict.c (maybe_hot_bb_p): Do not check profile status.
18283 (maybe_hot_edge_p): Likewise.
18284 (probably_never_executed): Check for zero counts even if profile
18286 (unlikely_executed_edge_p): New function.
18287 (unlikely_executed_stmt_p): New function.
18288 (unlikely_executed_bb_p): New function.
18289 (set_even_probabilities): Use unlikely predicates.
18290 (combine_predictions_for_bb): Likewise.
18291 (predict_paths_for_bb): Likewise.
18292 (predict_paths_leading_to_edge): Likewise.
18293 (determine_unlikely_bbs): New function.
18294 (estimate_bb_frequencies): Use it.
18295 (compute_function_frequency): Use zero counts even if profile is
18297 * profile-count.h: Fix typo.
18299 2017-08-08 Julia Koval <julia.koval@intel.com>
18301 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
18302 _mm512_mask_cvtsepi16_storeu_epi8,
18303 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
18304 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
18305 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
18306 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
18307 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
18308 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
18309 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
18310 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
18311 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
18312 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
18313 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
18314 __builtin_ia32_pmovuswb256mem_mask,
18315 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
18316 __builtin_ia32_pmovwb512mem_mask): New builtins.
18318 2017-08-08 Julia Koval <julia.koval@intel.com>
18320 PR target/73350,80862
18321 * config/i386/subst.md (round): Fix round pattern.
18322 * config/i386/i386.c (ix86_erase_embedded_rounding):
18323 Fix erasing rounding for the fixed pattern.
18325 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
18327 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
18329 2017-06-08 Martin Liska <mliska@suse.cz>
18331 PR gcov-profile/80911
18332 * gcov.c (block_info::block_info): New constructor.
18334 2017-06-07 Carl Love <cel@us.ibm.com>
18336 * config/rs6000/rs6000-c: The return type of the following
18337 built-in functions was implemented as int not long long. Fix sign
18338 of return value for the unsigned version of vec_mulo and vec_mule.
18339 vector unsigned long long vec_bperm (vector unsigned long long,
18340 vector unsigned char)
18341 vector signed long long vec_mule (vector signed int,
18343 vector unsigned long long vec_mule (vector unsigned int,
18344 vector unsigned int)
18345 vector signed long long vec_mulo (vector signed int,
18347 vector unsigned long long vec_mulo (vector unsigned int,
18348 vector unsigned int)
18349 * doc/extend.texi: Fix the documentation for the built-in
18352 2017-06-07 Carl Love <cel@us.ibm.com>
18355 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
18358 2017-06-07 Carl Love <cel@us.ibm.com>
18360 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
18361 support, Generate doublehv for signed int/float for BE case only.
18363 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
18365 * doc/invoke.texi (mcx16): Rewrite.
18367 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18369 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
18370 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
18371 *mov<mode>_softfloat, and an anonymous splitter): Use
18372 nonimmediate_operand instead of rs6000_nonimmediate_operand.
18374 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18376 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
18378 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
18379 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
18380 (rs6000_debug_reg_global): Adjust.
18381 (rs6000_init_hard_regno_mode_ok): Adjust.
18382 (rs6000_dbx_register_number): Adjust.
18383 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
18384 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
18385 Remove SPE_ACC and SPEFSCR.
18386 (REG_ALLOC_ORDER): Ditto.
18387 (FRAME_POINTER_REGNUM): Change to 111.
18388 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
18389 (REG_CLASS_NAMES): Ditto.
18390 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
18391 (REGISTER_NAMES): Ditto.
18392 (ADDITIONAL_REG_NAMES): Ditto.
18393 (rs6000_reg_names): Ditto.
18394 * config/rs6000/rs6000.md: Renumber some register number
18397 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18399 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
18401 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
18402 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
18404 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
18405 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
18406 Delete the SPE high registers.
18407 (REG_ALLOC_ORDER): Ditto.
18408 (enum reg_class): Remove SPE_HIGH_REGS.
18409 (REG_CLASS_NAMES): Ditto.
18410 (REG_CLASS_CONTENTS): Delete the SPE high registers.
18411 (REGISTER_NAMES): Ditto.
18412 (rs6000_reg_names): Ditto.
18413 * doc/tm.texi.in: Remove SPE as example.
18414 * doc/tm.texi: Regenerate.
18416 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18418 * config/rs6000/8540.md (ppc8540_brinc): Delete.
18419 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
18420 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
18421 * config/rs6000/rs6000.md (type): Remove "brinc".
18423 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18425 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
18426 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
18427 * config/rs6000/linuxspe.h: Delete file.
18428 * config/rs6000/rs6000.md: Don't include spe.md.
18429 * config/rs6000/spe.h: Delete file.
18430 * config/rs6000/spe.md: Delete file.
18431 * config/rs6000/t-rs6000: Remove spe.md.
18433 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18435 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
18436 (reg_or_none500mem_operand): Delete.
18437 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
18438 instead of reg_or_none500mem_operand.
18440 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18442 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
18443 handling of SPE flags.
18444 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
18446 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18448 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
18450 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
18451 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
18452 paired_divv2sf3): Similar.
18453 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
18454 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
18455 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
18457 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
18458 Rename the paired_* instruction patterns.
18459 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
18461 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
18462 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
18463 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
18464 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
18465 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
18466 PAIRED_VECTOR_MODE.
18467 (struct machine_function): Delete field spe_insn_chain_scanned_p.
18468 (spe_func_has_64bit_regs_p): Delete.
18469 (spe_expand_predicate_builtin): Delete.
18470 (spe_expand_evsel_builtin): Delete.
18471 (TARGET_DWARF_REGISTER_SPAN): Do not define.
18472 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
18473 (invalid_e500_subreg): Delete.
18474 (rs6000_legitimize_address): Always force_reg op2 as well, for
18475 paired single memory accesses.
18476 (rs6000_member_type_forces_blk): Delete.
18477 (rs6000_spe_function_arg): Delete.
18478 (rs6000_expand_unop_builtin): Delete SPE handling.
18479 (rs6000_expand_binop_builtin): Ditto.
18480 (spe_expand_stv_builtin): Delete.
18481 (bdesc_2arg_spe): Delete.
18482 (spe_expand_builtin): Delete.
18483 (spe_expand_predicate_builtin): Delete.
18484 (spe_expand_evsel_builtin): Delete.
18485 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
18486 (spe_init_builtins): Delete.
18487 (spe_func_has_64bit_regs_p): Delete.
18488 (savres_routine_name): Delete "info" parameter. Adjust callers.
18489 (rs6000_emit_stack_reset): Ditto.
18490 (rs6000_dwarf_register_span): Delete.
18491 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
18492 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
18493 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
18495 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
18497 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
18498 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
18499 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
18500 mulv2sf3, divv2sf3): Delete expanders.
18502 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18504 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
18506 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18508 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
18509 * config/rs6000/rs6000.c: Ditto.
18511 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18513 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
18514 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
18515 comparison_operator.
18517 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18519 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
18520 * config/rs6000/rs6000.opt: Ditto.
18521 * config/rs6000/t-rtems: Ditto.
18523 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18525 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
18526 TARGET_E500_SINGLE by 0, simplify.
18527 * config/rs6000/rs6000.c: Ditto.
18528 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
18529 (spe_build_register_parallel): Delete.
18530 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
18531 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
18532 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
18533 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
18534 (E500_CONVERT): Delete.
18535 * config/rs6000/spe.md: Remove many patterns and all define_constants.
18537 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
18539 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
18540 * config/rs6000/dfp.md: Ditto.
18541 (negdd2, *negdd2_fpr): Merge.
18542 (absdd2, *absdd2_fpr): Merge.
18543 (negtd2, *negtd2_fpr): Merge.
18544 (abstd2, *abstd2_fpr): Merge.
18545 * config/rs6000/e500.h: Delete file.
18546 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
18547 TARGET_FPRS by 1 and simplify.
18548 * config/rs6000/rs6000-c.c: Ditto.
18549 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
18550 TARGET_DF_SPE by 0.
18551 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
18553 * config/rs6000/rs6000.md: Ditto.
18554 (floatdidf2, *floatdidf2_fpr): Merge.
18555 (move_from_CR_gt_bit): Delete.
18556 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
18557 (E500_CR_IOR_COMPARE): Delete.
18558 (All patterns that require !TARGET_FPRS): Delete.
18559 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
18561 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18563 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
18565 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18567 * graphds.c (add_edge): Intitialize edge's attached data.
18568 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
18569 pointer parameter. Call pointed function on each edge during
18570 graph traversing. Skip traversing the edge when the function
18572 (graphds_dfs, graphds_scc): Ditto.
18573 (for_each_edge): New parameter. Pass the new parameter to callback
18575 * graphds.h (skip_edge_callback): New function pointer type.
18576 (graphds_dfs, graphds_scc): New function pointer parameter.
18577 (graphds_edge_callback, for_each_edge): New parameter.
18579 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18581 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
18582 out code checking if runtime alias check is possible to below ...
18583 Call the new function.
18584 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
18585 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
18587 2017-06-07 Marek Polacek <polacek@redhat.com>
18590 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
18591 TYPE_OVERFLOW_WRAPS check.
18593 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18595 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
18596 if versioning is required.
18597 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
18598 peeling with the check for versioning.
18600 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18602 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
18603 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
18604 Set true to new parameter if new ssa variable is defined.
18605 (vect_gen_vector_loop_niters): Refactor. Set range information
18606 for the new vector loop bound variable.
18607 (vect_do_peeling): Ditto.
18609 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18611 * tree-affine.c (ssa.h): Include header file.
18612 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
18613 has wrapping overflow behavior.
18615 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18617 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
18619 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18621 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
18622 (tree_to_aff_combination): ... here.
18624 2017-06-07 Bin Cheng <bin.cheng@arm.com>
18626 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
18627 reg_pressure model function.
18628 (ivopts_global_cost_for_size): Delete.
18629 (determine_set_costs, iv_ca_recount_cost): Call new model function
18630 ivopts_estimate_reg_pressure.
18632 2017-06-07 Tamar Christina <tamar.christina@arm.com>
18634 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
18635 expensive than udiv. Remove floating point cases from mod.
18637 2017-06-07 Tamar Christina <tamar.christina@arm.com>
18639 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
18640 Increase idiv cost.
18642 2017-06-07 Tamar Christina <tamar.christina@arm.com>
18644 * config/aarch64/aarch64.md
18645 (copysignsf3): Fix mask generation.
18647 2017-06-07 Jakub Jelinek <jakub@redhat.com>
18649 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
18651 (class dump_manager): Add register_dumps method.
18652 * dumpfile.c: Include langhooks.h.
18653 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
18654 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
18655 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
18656 (dump_manager::dump_register): Start with 512 entries instead of 32.
18657 (dump_manager::register_dumps): New method.
18658 * toplev.c (general_init): Instead of invoking register_dumps
18659 langhook, invoke register_dumps method on the dump manager.
18660 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
18663 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
18665 * doc/md.texi: Clarify the restrictions on a define_insn condition.
18666 Say that # requires an associated define_split to exist, and that
18667 the define_split must be suitable for use after register allocation.
18669 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
18671 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
18672 (compute_outgoing_frequencies): Also initialize zero counts.
18673 (find_many_sub_basic_blocks): Do not produce uninitialized profile
18674 around loops; preserve more of profile when nothing changes.
18676 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
18678 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
18680 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
18681 * config/arm/arm-cpu-cdata.h: Regenerate.
18682 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
18683 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
18684 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
18685 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
18686 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
18688 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
18690 * config/arm/t-rmprofile: Likewise.
18691 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
18693 2017-06-06 David S. Miller <davem@davemloft.net>
18696 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
18697 blockage if function uses alloca.
18699 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
18701 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
18702 New "uid" fields to hold pretty-print IDs of group and ref.
18703 Memory references are now identified as <group_id>:<ref_id>
18704 instead of using [random] addresses.
18705 (dump_mem_details): Simplify, no functional change.
18706 (dump_mem_ref): Simplify and make output more concise.
18707 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
18708 (find_or_create_group): Initialize group uid.
18709 (record_ref): Initialize ref uid. Improve debug output.
18710 (prune_group_by_reuse, should_issue_prefetch_p,)
18711 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
18712 (mark_nontemporal_store, determine_loop_nest_reuse):
18713 Improve debug output.
18715 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
18717 * dbgcnt.def (prefetch): New debug counter.
18718 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
18719 (schedule_prefetches): Stop issueing prefetches if debug counter
18722 2017-06-06 Tom de Vries <tom@codesourcery.com>
18724 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
18725 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
18727 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18729 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
18730 Use aarch64_reg_or_zero predicate for operand 4.
18731 (aarch64_compare_and_swap<mode> define_insn_and_split):
18732 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
18733 (aarch64_store_exclusive<mode>): Likewise for operand 2.
18735 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
18737 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
18738 (arm_compute_save_core_reg_mask): This.
18739 (thumb1_compute_save_reg_mask): Rename into ...
18740 (thumb1_compute_save_core_reg_mask): This.
18741 (arm_compute_save_reg0_reg12_mask): Adapt comment.
18742 (arm_compute_frame_layout): Likewise.
18744 2017-06-06 Richard Biener <rguenther@suse.de>
18746 PR tree-optimization/80974
18747 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
18748 keep or clear leaders SSA info.
18750 2017-06-06 Tom de Vries <tom@codesourcery.com>
18752 * config/nvptx/nvptx.c (split_mode_p): New function.
18753 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
18755 2017-06-06 Tom de Vries <tom@codesourcery.com>
18757 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
18759 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
18762 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
18765 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
18767 * shrink-wrap.c (handle_simple_exit): Update profile.
18768 (try_shrink_wrapping): Upate profile.
18770 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
18772 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
18773 (tree_guess_outgoing_edge_probabilities): New.
18774 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
18775 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
18777 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
18779 * ipa-split.c (split_function): Initialize return bb profile.
18781 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
18783 * profile.c (compute_branch_probabilities): Also initialize
18784 EXIT_BLOCK profile.
18786 2017-06-06 Richard Biener <rguenther@suse.de>
18788 PR tree-optimization/80928
18789 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
18790 (vect_analyze_loop_operations): Properly guard analysis for
18792 (vect_transform_loop): Likewise.
18793 (vect_analyze_loop_2): Also reset SLP type on PHIs.
18794 (vect_model_induction_cost): Do not cost for pure SLP.
18795 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
18796 of induction in inner loop vectorization.
18797 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
18798 (vect_get_and_check_slp_defs): Handle vect_induction_def.
18799 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
18801 (vect_analyze_slp_cost_1): Cost induction.
18802 (vect_detect_hybrid_slp_stmts): Handle PHIs.
18803 (vect_get_slp_vect_defs): Likewise.
18804 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
18805 (vect_transform_stmt): Handle SLP reductions.
18806 * tree-vectorizer.h (vectorizable_induction): Adjust.
18808 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
18810 * config/rs6000/rs6000.c (make_resolver_func): Update
18811 init_lowered_empty_function call.
18813 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
18815 * doc/invoke.texi: Document the -fprofile-abs-path option.
18816 * common.opt (fprofile-abs-path): New option.
18817 * gcov-io.h (gcov_write_filename): Declare.
18818 * gcov-io.c (gcov_write_filename): New function.
18819 * coverage.c (coverage_begin_function): Use gcov_write_filename.
18820 * profile.c (output_location): Likewise.
18822 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
18824 * shring-wrap.c: Revert accidental commit.
18826 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
18828 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
18830 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
18832 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
18834 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
18835 profile in callgraph edge.
18836 * profile-count.h (apply_probability): If THIS is 0, then result is 0
18837 (apply_scale): Likewise.
18838 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
18839 Also scale profile when inlining function with zero profile.
18840 (initialize_cfun): Update exit block profile even when it is zero.
18841 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
18842 when profile is read.
18844 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
18846 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
18847 (CLONE_*): New constants to define the processors we can generate
18848 code for with the target_clone attribute.
18849 (rs6000_clone_map): New array to identify which clone processors
18850 the current program is running on.
18851 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
18852 target_clone attribute.
18853 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
18854 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
18855 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
18856 (cpu_expand_builtin): Add support for target_clone attribute.
18857 (rs6000_valid_attribute_p): Allow "default" attribute.
18858 (get_decl_name): New debug function to simplify printing the
18859 current function name in debugging statements.
18860 (rs6000_clone_priority): New functions to support the target_clone
18861 attribute, and be able to generate code to switch between ISA 2.05
18862 through ISA 3.0 (power6 through power9).
18863 (rs6000_compare_version_priority): Likewise.
18864 (rs6000_get_function_versions_dispatcher): Likewise.
18865 (make_resolver_func): Likewise.
18866 (add_condition_to_bb): Likewise.
18867 (dispatch_function_versions): Likewise.
18868 (rs6000_generate_version_dispatcher_body): Likewise.
18869 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
18870 (fusion_gpr_load_p): Fix a spacing issue.
18871 * doc/extend.texi (Common Function Attributes): Document that the
18872 PowerPC supports the target_clone attribute.
18874 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
18876 * config/arm/arm.h: explain F symbol found in description of ARM
18877 register allocation in its legend.
18879 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
18881 * config/mips/frame-header-opt.c: Include profile-count.h.
18882 * config/riscv/riscv.c: Include profile-count.h
18884 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
18886 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
18888 (sm_set_flag_if_changed): Add bbs field.
18889 (execute_sm_if_changed_flag_set): Pass BBS.
18890 (execute_sm): Update.
18892 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18894 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
18897 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18899 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
18900 (peephole2): New peephole2 to emit the above.
18901 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
18903 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18905 * config/aarch64/aarch64.c (define_peephole2 above
18906 *sub_<shift>_<mode>): New peephole.
18908 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
18910 * config/i386/i386.c (make_resolver_func): Update.
18911 * Makefile.in: Add profile-count.h and profile-count.o
18912 * auto-profile.c (afdo_indirect_call): Update to new API.
18913 (afdo_set_bb_count): Update.
18914 (afdo_propagate_edge): Update.
18915 (afdo_propagate_circuit): Update.
18916 (afdo_calculate_branch_prob): Update.
18917 (afdo_annotate_cfg): Update.
18918 * basic-block.h: Include profile-count.h
18919 (struct edge_def): Turn count to profile_count.
18920 (struct basic_block_def): Likewie.
18921 (REG_BR_PROB_BASE): Move to profile-count.h
18922 (RDIV): Move to profile-count.h
18923 * bb-reorder.c (max_entry_count): Turn to profile_count.
18924 (find_traces): Update.
18925 (rotate_loop):Update.
18926 (connect_traces):Update.
18927 (sanitize_hot_paths):Update.
18928 * bt-load.c (migrate_btr_defs): Update.
18929 * cfg.c (RDIV): Remove.
18930 (init_flow): Use alloc_block.
18931 (alloc_block): Uninitialize count.
18932 (unchecked_make_edge): Uninitialize count.
18933 (check_bb_profile): Update.
18934 (dump_edge_info): Update.
18935 (dump_bb_info): Update.
18936 (update_bb_profile_for_threading): Update.
18937 (scale_bbs_frequencies_int): Update.
18938 (scale_bbs_frequencies_gcov_type): Update.
18939 (scale_bbs_frequencies_profile_count): New.
18940 * cfg.h (update_bb_profile_for_threading): Update.
18941 (scale_bbs_frequencies_profile_count): Declare.
18942 * cfgbuild.c (compute_outgoing_frequencies): Update.
18943 (find_many_sub_basic_blocks): Update.
18944 * cfgcleanup.c (try_forward_edges): Update.
18945 (try_crossjump_to_edge): Update.
18946 * cfgexpand.c (expand_gimple_tailcall): Update.
18947 (construct_exit_block): Update.
18948 * cfghooks.c (verify_flow_info): Update.
18949 (dump_bb_for_graph): Update.
18950 (split_edge): Update.
18951 (make_forwarder_block): Update.
18952 (duplicate_block): Update.
18953 (account_profile_record): Update.
18954 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
18955 (get_estimated_loop_iterations): Update.
18956 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
18957 (single_likely_exit): Update.
18958 * cfgloopmanip.c (scale_loop_profile): Update.
18960 (set_zero_probability): Update.
18961 (lv_adjust_loop_entry_edge): Update.
18962 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
18963 (purge_dead_edges): Update.
18964 (rtl_account_profile_record): Update.
18965 * cgraph.c (cgraph_node::create): Uninitialize count.
18966 (symbol_table::create_edge): Uninitialize count.
18967 (cgraph_update_edges_for_call_stmt_node): Update.
18968 (cgraph_edge::dump_edge_flags): Update.
18969 (cgraph_node::dump): Update.
18970 (cgraph_edge::maybe_hot_p): Update.
18971 * cgraph.h: Include profile-count.h
18972 (create_clone), create_edge, create_indirect_edge): Update.
18973 (cgraph_node): Turn count to profile_count.
18974 (cgraph_edge0: Likewise.
18975 (make_speculative, clone): Update.
18976 (create_edge): Update.
18977 (init_lowered_empty_function): Update.
18978 * cgraphclones.c (cgraph_edge::clone): Update.
18979 (duplicate_thunk_for_node): Update.
18980 (cgraph_node::create_clone): Update.
18981 * cgraphunit.c (cgraph_node::analyze): Update.
18982 (cgraph_node::expand_thunk): Update.
18983 * final.c (dump_basic_block_info): Update.
18984 * gimple-streamer-in.c (input_bb): Update.
18985 * gimple-streamer-out.c (output_bb): Update.
18986 * graphite.c (print_global_statistics): Update.
18987 (print_graphite_scop_statistics): Update.
18988 * hsa-brig.c: Include basic-block.h.
18989 * hsa-dump.c: Include basic-block.h.
18990 * hsa-gen.c (T sum_slice): Update.
18991 (convert_switch_statements):Update.
18992 * hsa-regalloc.c: Include basic-block.h.
18993 * ipa-chkp.c (chkp_produce_thunks): Update.
18994 * ipa-cp.c (struct caller_statistics): Update.
18995 (init_caller_stats): Update.
18996 (gather_caller_stats): Update.
18997 (ipcp_cloning_candidate_p): Update.
18998 (good_cloning_opportunity_p): Update.
18999 (get_info_about_necessary_edges): Update.
19000 (dump_profile_updates): Update.
19001 (update_profiling_info): Update.
19002 (update_specialized_profile): Update.
19003 (perhaps_add_new_callers): Update.
19004 (decide_about_value): Update.
19005 (ipa_cp_c_finalize): Update.
19006 * ipa-devirt.c (struct odr_type_warn_count): Update.
19007 (struct decl_warn_count): Update.
19008 (struct final_warning_record): Update.
19009 (possible_polymorphic_call_targets): Update.
19010 (ipa_devirt): Update.
19011 * ipa-fnsummary.c (redirect_to_unreachable): Update.
19012 * ipa-icf.c (sem_function::merge): Update.
19013 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
19014 * ipa-inline.c (compute_uninlined_call_time): Update.
19015 (compute_inlined_call_time): Update.
19016 (want_inline_small_function_p): Update.
19017 (want_inline_self_recursive_call_p): Update.
19018 (edge_badness): Update.
19019 (lookup_recursive_calls): Update.
19020 (recursive_inlining): Update.
19021 (inline_small_functions): Update.
19022 (dump_overall_stats): Update.
19023 (dump_inline_stats): Update.
19024 * ipa-profile.c (ipa_profile_generate_summary): Update.
19025 (ipa_propagate_frequency): Update.
19026 (ipa_profile): Update.
19027 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
19028 * ipa-utils.c (ipa_merge_profiles): Update.
19029 * loop-doloop.c (doloop_modify): Update.
19030 * loop-unroll.c (report_unroll): Update.
19031 (unroll_loop_runtime_iterations): Update.
19032 * lto-cgraph.c (lto_output_edge): Update.
19033 (lto_output_node): Update.
19034 (input_node): Update.
19035 (input_edge): Update.
19036 (merge_profile_summaries): Update.
19037 * lto-streamer-in.c (input_cfg): Update.
19038 * lto-streamer-out.c (output_cfg): Update.
19039 * mcf.c (create_fixup_graph): Update.
19040 (adjust_cfg_counts): Update.
19041 (sum_edge_counts): Update.
19042 * modulo-sched.c (sms_schedule): Update.
19043 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
19044 * predict.c (maybe_hot_count_p): Update.
19045 (probably_never_executed): Update.
19046 (dump_prediction): Update.
19047 (combine_predictions_for_bb): Update.
19048 (propagate_freq): Update.
19049 (handle_missing_profiles): Update.
19050 (counts_to_freqs): Update.
19051 (rebuild_frequencies): Update.
19052 (force_edge_cold): Update.
19053 * predict.h: Include profile-count.h
19054 (maybe_hot_count_p, counts_to_freqs): UPdate.
19055 * print-rtl-function.c: Do not include cfg.h
19056 * print-rtl.c: Include basic-block.h
19057 * profile-count.c: New file.
19058 * profile-count.h: New file.
19059 * profile.c (is_edge_inconsistent): Update.
19060 (correct_negative_edge_counts): Update.
19061 (is_inconsistent): Update.
19062 (set_bb_counts): Update.
19063 (read_profile_edge_counts): Update.
19064 (compute_frequency_overlap): Update.
19065 (compute_branch_probabilities): Update; Initialize and deinitialize
19067 (branch_prob): Update.
19068 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
19069 (edge_gcov_count): New.
19070 (bb_gcov_count): New.
19071 * shrink-wrap.c (try_shrink_wrapping): Update.
19072 * tracer.c (better_p): Update.
19073 * trans-mem.c (expand_transaction): Update.
19074 (ipa_tm_insert_irr_call): Update.
19075 (ipa_tm_insert_gettmclone_call): Update.
19076 * tree-call-cdce.c: Update.
19077 * tree-cfg.c (gimple_duplicate_sese_region): Update.
19078 (gimple_duplicate_sese_tail): Update.
19079 (gimple_account_profile_record): Update.
19080 (execute_fixup_cfg): Update.
19081 * tree-inline.c (copy_bb): Update.
19082 (copy_edges_for_bb): Update.
19083 (initialize_cfun): Update.
19084 (freqs_to_counts): Update.
19085 (copy_cfg_body): Update.
19086 (expand_call_inline): Update.
19087 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
19088 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
19089 (try_unroll_loop_completely): Update.
19090 (try_peel_loop): Update.
19091 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19092 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
19093 * tree-ssa-loop-split.c (connect_loops): Update.
19094 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
19095 * tree-ssa-reassoc.c (branch_fixup): Update.
19096 * tree-ssa-tail-merge.c (replace_block_by): Update.
19097 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
19098 (compute_path_counts): Update.
19099 (update_profile): Update.
19100 (recompute_probabilities): Update.
19101 (update_joiner_offpath_counts): Update.
19102 (estimated_freqs_path): Update.
19103 (freqs_to_counts_path): Update.
19104 (clear_counts_path): Update.
19105 (ssa_fix_duplicate_block_edges): Update.
19106 (duplicate_thread_path): Update.
19107 * tree-switch-conversion.c (case_bit_test_cmp): Update.
19108 (struct switch_conv_info): Update.
19109 * tree-tailcall.c (decrease_profile): Update.
19110 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
19111 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19112 * value-prof.c (check_counter): Update.
19113 (gimple_divmod_fixed_value): Update.
19114 (gimple_mod_pow2): Update.
19115 (gimple_mod_subtract): Update.
19116 (gimple_ic_transform): Update.
19117 (gimple_stringop_fixed_value): Update.
19118 * value-prof.h (gimple_ic): Update.
19120 2017-06-02 Carl Love <cel@us.ibm.com>
19122 * config/rs6000/rs6000-c: Add support for built-in functions
19123 vector double vec_doublee (vector signed int);
19124 vector double vec_doublee (vector unsigned int);
19125 vector double vec_doublee (vector float);
19126 vector double vec_doubleh (vector signed int);
19127 vector double vec_doubleh (vector unsigned int);
19128 vector double vec_doubleh (vector float);
19129 vector double vec_doublel (vector signed int);
19130 vector double vec_doublel (vector unsigned int);
19131 vector double vec_doublel (vector float);
19132 vector double vec_doubleo (vector signed int);
19133 vector double vec_doubleo (vector unsigned int);
19134 vector double vec_doubleo (vector float);.
19135 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
19136 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
19138 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
19139 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
19140 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
19142 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
19143 vec_doublel, vec_doubleh.
19144 * doc/extend.texi: Update the built-in documentation file for the
19145 new built-in functions.
19147 2017-06-02 David Malcolm <dmalcolm@redhat.com>
19150 * ipa-inline-analysis.c (free_growth_caches): Set
19151 edge_removal_hook_holder to NULL after removing it.
19153 2017-06-02 Sudakshina Das <sudi.das@arm.com>
19155 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
19156 comparision with zero.
19158 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
19159 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
19160 for early expansion of vec_min and vec_max builtins.
19161 (builtin_function_type): Add min/max unsigned variants to those
19162 identified as having unsigned arguments.
19164 2017-06-02 Olivier Hainque <hainque@adacore.com>
19166 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
19168 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19170 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
19171 Use VALL_F16 iterator rather than VALL.
19173 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19175 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
19176 Emit CBNZ inside loop when doing a strong exchange and comparing
19177 against zero. Generate the CC flags after the loop.
19179 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
19181 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
19182 (dl_section_ref): New.
19183 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
19184 On AIX, append an expression to subtract the size of the
19185 section length to dl_section_ref.
19187 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
19189 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
19190 for early expansion of vector absolute builtins.
19192 2017-06-02 Richard Biener <rguenther@suse.de>
19194 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
19195 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
19197 2017-06-02 Richard Biener <rguenther@suse.de>
19199 PR tree-optimization/80948
19200 * tree-tailcall.c (find_tail_calls): Track stmts to move in
19201 stmt order as well.
19203 2017-06-02 Richard Biener <rguenther@suse.de>
19205 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
19207 * tree-vect-stmts.c (process_use): Do not mark backedge defs
19208 for inductions as relevant.
19210 2017-06-02 Richard Biener <rguenther@suse.de>
19212 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
19213 (vectorizable_induction): ... this. Remove dead code.
19215 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
19217 * builtins. (expand_builtin_alloca): Remove second parameter and
19218 infer its value from the first parameter instead.
19219 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
19221 2017-06-02 Jakub Jelinek <jakub@redhat.com>
19223 PR rtl-optimization/80903
19224 * loop-doloop.c (add_test): Unshare sequence.
19226 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
19228 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
19230 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
19232 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
19234 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
19235 xlogue_layout::get_instance, logue_layout::xlogue_layout,
19236 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
19237 (xlogue_layout::get_stub_rtx): Make static.
19238 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
19239 (xlogue_layout::compute_stub_managed_regs): Rename to...
19240 (xlogue_layout::count_stub_managed_regs): ...this.
19241 (xlogue_layout::is_stub_managed_reg): New function.
19242 (xlogue_layout::m_stub_names): Rename to...
19243 (xlogue_layout::s_stub_names): ...this, make static.
19244 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
19245 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
19246 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
19247 xlogue_layout::s_stub_names): Instantiate statics.
19248 (stub_managed_regs): Remove.
19249 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
19250 (disable_call_ms2sysv_xlogues): Rename to...
19251 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
19252 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
19254 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
19255 change after reload_completed.
19256 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
19258 (ix86_expand_prologue): Likewise.
19259 (ix86_expand_epilogue): Likewise.
19260 (ix86_expand_split_stack_prologue): Likewise.
19261 (ix86_compute_frame_layout): Remove frame parameter ...
19262 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
19263 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
19265 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
19266 (ix86_frame): Move from here ...
19267 * config/i386/i386.h (ix86_frame): ... to here.
19268 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
19269 complete ix86_frame data structure instead. Remove some_ld_name.
19271 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
19273 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
19274 symbols that hold a DECL_VALUE_EXPR.
19276 2017-06-01 Martin Jambor <mjambor@suse.cz>
19278 PR tree-optimization/80898
19279 * tree-sra.c (process_subtree_disqualification): Removed.
19280 (disqualify_candidate): Do not acll
19281 process_subtree_disqualification.
19282 (subtree_mark_written_and_enqueue): New function.
19283 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
19284 RHS has been disqualified and re-queue LHS if necessary. Apart
19285 from that, ignore disqualified RHS.
19287 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19289 * config/s390/s390.c (s390_emit_epilogue): Disable early return
19290 address fetch for z10 or later.
19292 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19294 * config/arc/arc.md (tst_movb): Add guard when splitting.
19296 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19298 * config/arc/arc.c (arc_can_eliminate): Test against
19299 arc_frame_pointer_needed.
19301 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19303 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
19304 to prevent store reordering.
19305 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
19306 (type): Add block type.
19307 (stack_tie): Define special instruction to be used in
19310 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19312 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
19313 constraint. It is not valid for the pattern.
19314 (noncommutative_binary_comparison): Likewise.
19316 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19318 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
19321 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19323 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
19324 be used by the reg-alloc.
19326 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19328 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
19329 reg-alloc when having mul64 or mul32x16 instructions.
19330 (mulsidi3): Likewise.
19331 (umulsidi3): Likewise.
19332 (mulsi32x16): New pattern.
19333 (mulsi64): Likewise.
19334 (mulsidi64): Likewise.
19335 (umulsidi64): Likewise.
19336 (MUL32x16_REG): Define.
19337 (mul64_600): Use MUL32x16_REG.
19338 (mac64_600): Likewise.
19339 (umul64_600): Likewise.
19340 (umac64_600): Likewise.
19342 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
19344 * config/arc/arc.md (mulsi3_700): Make it commutative.
19346 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
19348 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
19350 (*sign_extendsidi2_insn): Likewise for movstosw.
19352 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
19354 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
19355 the type of the input discriminant value. Convert the
19356 discriminant value of signedness vary.
19358 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
19360 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
19361 Add to -Wall section.
19363 2017-06-01 Richard Biener <rguenther@suse.de>
19365 PR middle-end/66313
19366 * fold-const.c (fold_plusminus_mult_expr): If the factored
19367 factor may be zero use a wrapping type for the inner operation.
19368 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
19369 and handle moved defs.
19370 (process_assignment): Properly guard the unary op case. Return a
19371 tri-state indicating that moving the stmt before the call may allow
19372 to continue. Pass through to_move.
19373 (find_tail_calls): Handle moving unrelated defs before
19376 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
19379 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
19380 splitter result in the canonical way.
19382 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
19384 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
19385 also for 32bit target. Update insn attributes.
19386 (zero-extendsidi2 splitter): Allow all registers for operand 1.
19388 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
19390 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
19391 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
19392 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
19393 (_mm_maskz_min_ss): New intrinsics.
19395 2017-05-31 Martin Liska <mliska@suse.cz>
19397 * tree-vect-loop.c (vect_create_epilog_for_reduction):
19398 Change comment style to one we normally use.
19399 (vectorizable_reduction): Likewise.
19400 (vectorizable_induction): Likewise.
19401 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
19402 (vectorizable_call): Likewise.
19403 (vectorizable_simd_clone_call): Likewise.
19404 (vectorizable_conversion): Likewise.
19405 (vectorizable_assignment): Likewise.
19406 (vectorizable_shift): Likewise.
19407 (vectorizable_operation): Likewise.
19408 (vectorizable_store): Likewise.
19409 (vectorizable_load): Likewise.
19410 * tree-vectorizer.h: Likewise.
19412 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
19414 * passes.c (emergency_dump_function): New.
19415 * tree-pass.h (emergency_dump_function): Declare.
19416 * plugin.c (plugins_internal_error_function): Remove.
19417 * plugin.h (plugins_internal_error_function): Remove declaration.
19418 * toplev.c (internal_error_function): New static function. Use it...
19419 (general_init): ...here.
19421 2017-05-31 Graham Markall <graham.markall@embecosm.com>
19423 * config/arc/arc.c (arc_print_operand): Handle constant operands.
19424 (arc_rtx_costs): Add costs for new patterns.
19425 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
19426 * config/arc/predicates.md: Add _1_2_3_operand predicate.
19428 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
19430 * tree-ssa-strlen.c (get_next_strinfo): New function.
19431 (get_stridx_plus_constant): Use it.
19432 (zero_length_string): Likewise.
19433 (adjust_related_strinfos): Likewise.
19434 (adjust_last_stmt): Likewise.
19436 2017-05-31 Richard Biener <rguenther@suse.de>
19439 * config/i386/i386.c (ix86_expand_builtin): Remove assert
19440 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
19442 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
19444 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
19445 loop_vinfo argument and use of dependence distance vectors.
19446 Check instead whether the two references differ only in their
19447 initial value and assume that they have the same alignment if the
19448 difference is a multiple of the vector alignment.
19449 (vect_analyze_data_refs_alignment): Update call accordingly.
19451 2017-05-31 Martin Liska <mliska@suse.cz>
19454 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
19456 2017-05-31 Bin Cheng <bin.cheng@arm.com>
19458 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
19459 (create_intersect_range_checks): Move from ...
19460 * tree-data-ref.c (create_intersect_range_checks_index)
19461 (create_intersect_range_checks): ... to here.
19462 (create_runtime_alias_checks): New function factored from ...
19463 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
19464 here. Call above function.
19465 * tree-data-ref.h (create_runtime_alias_checks): New function.
19467 2017-05-31 Bin Cheng <bin.cheng@arm.com>
19469 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
19470 segment length for dr_b and compute it in wide_int.
19472 2017-05-31 Richard Biener <rguenther@suse.de>
19474 PR tree-optimization/80906
19475 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
19476 and pass through iv_map.
19477 (copy_bb_and_scalar_dependences): Adjust.
19478 (translate_pending_phi_nodes): Likewise.
19479 (copy_loop_close_phi_args): Handle code-generating IVs instead
19482 2017-05-30 David Malcolm <dmalcolm@redhat.com>
19484 * diagnostic-color.c (color_dict): Add "type-diff".
19485 (parse_gcc_colors): Update comment.
19486 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
19487 -fdiagnostics-show-template-tree and -fno-elide-type.
19488 (GCC_COLORS): Add type-diff to example.
19490 (-fdiagnostics-show-template-tree): New.
19491 (-fno-elide-type): New.
19492 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
19493 the pp_format_decoder callback. Call any m_format_postprocessor's
19495 (pretty_printer::pretty_printer): Initialize
19496 m_format_postprocessor.
19497 (pretty_printer::~pretty_printer): Delete any
19498 m_format_postprocessor.
19499 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
19500 (class format_postprocessor): New class.
19501 (struct pretty_printer::format_decoder): Document the new parameters.
19502 (struct pretty_printer::m_format_postprocessor): New field.
19503 * tree-diagnostic.c (default_tree_printer): Update for new
19504 bool and const char ** params.
19505 * tree-diagnostic.h (default_tree_printer): Likewise.
19507 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
19509 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
19510 (lwa_operand): Delete rs6000_gen_cell_microcode test.
19511 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
19512 rs6000_gen_cell_microcode code.
19513 (rs6000_final_prescan_insn): Delete.
19514 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
19515 "warn-cell-microcode" entries.
19516 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
19517 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
19518 throughout. Change cc_reg_not_micro_cr0_operand to
19519 cc_reg_not_cr0_operand throughout.
19520 (*extendhi<mode>2_noload): Delete.
19521 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
19522 (mwarn-cell-microcode): Delete.
19523 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
19524 -mgen-cell-microcode and -mwarn-cell-microcode.
19526 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
19529 * config/i386/constraints.md (Yd): New constraint.
19531 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
19532 and (?Yd, r) alternatives. Update insn attributes.
19533 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
19534 and (?*Yd, r) alternatives. Update insn attributes.
19535 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
19537 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
19539 * gimplify.c (gimplify_modify_expr): Don't create a
19540 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
19543 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
19545 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
19547 2017-05-30 Richard Biener <rguenther@suse.de>
19549 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
19550 and reduc_def fields.
19551 (STMT_VINFO_REDUC_TYPE): New define.
19552 (STMT_VINFO_REDUC_DEF): Likewise.
19553 (vect_force_simple_reduction): Adjust prototype.
19554 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
19555 (vect_is_simple_reduction): Remove check_reduction argument.
19556 (vect_force_simple_reduction): Adjust and set
19557 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
19558 (vectorizable_reduction): Do not re-do reduction analysis
19559 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
19560 * tree-parloops.c (gather_scalar_reductions): Adjust.
19562 2017-05-30 Richard Biener <rguenther@suse.de>
19564 PR middle-end/80901
19565 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
19568 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
19570 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
19571 Introduce unknown_misalignment parameter and remove vf.
19572 (vect_peeling_hash_get_lowest_cost):
19573 Pass unknown_misalignment parameter.
19574 (vect_enhance_data_refs_alignment):
19575 Fix unsupportable data ref treatment.
19577 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
19579 * tree-vect-data-refs.c (vect_get_data_access_cost):
19580 Workaround for SLP handling.
19581 (vect_enhance_data_refs_alignment):
19582 Compute costs for doing no peeling at all, compare to the best
19583 peeling costs so far and avoid peeling if cheaper.
19585 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
19587 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
19588 Return peeling info and set costs to zero for unlimited cost
19590 (vect_enhance_data_refs_alignment): Also inspect all datarefs
19591 with unknown misalignment. Compute and costs for unknown
19592 misalignment, compare them to the costs for known misalignment
19593 and choose the cheapest for peeling.
19595 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
19597 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
19598 (vect_get_peeling_costs_all_drs): Create function.
19599 (vect_peeling_hash_get_lowest_cost):
19600 Use vect_get_peeling_costs_all_drs.
19601 (vect_peeling_supportable): Create function.
19602 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
19604 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
19606 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
19607 DR_HAS_NEGATIVE_STEP.
19608 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
19609 (vect_enhance_data_refs_alignment): Use.
19610 (vect_duplicate_ssa_name_ptr_info): Use.
19611 * tree-vectorizer.h (dr_misalignment): Use.
19612 (known_alignment_for_access_p): Use.
19614 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19617 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
19619 (has_section_name): New function.
19621 2017-05-30 Martin Liska <mliska@suse.cz>
19624 * auto-profile.c (get_function_decl_from_block): Fix
19627 2017-05-30 Richard Biener <rguenther@suse.de>
19629 PR middle-end/80876
19630 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
19632 2017-05-30 Martin Liska <mliska@suse.cz>
19634 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
19635 * dumpfile.h (struct dump_file_info): Remove ctors.
19637 2017-05-30 Martin Liska <mliska@suse.cz>
19639 * predict.def: Fix GNU coding style.
19641 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
19643 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
19644 Mark 'to' argument with ATTRIBUTE_UNUSED.
19646 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
19648 * config/xtensa/xtensa.c (xtensa_emit_call): Use
19649 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
19650 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
19653 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
19655 * doc/install.texi (Options specification): Restore entry of
19656 --enable-sjlj-exceptions.
19658 2017-05-27 Michael Eager <eager@eagercon.com>
19661 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
19663 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
19665 * config/microblaze/microblaze.h
19666 (FIXED_REGISTERS): Update in macro.
19667 (CALL_USED_REGISTERS): Update in macro.
19669 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
19671 * doc/install.texi: Add links to macOS binary distributions.
19673 2017-05-27 Jakub Jelinek <jakub@redhat.com>
19677 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
19679 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
19681 2017-05-26 Martin Liska <mliska@suse.cz>
19683 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
19685 2017-05-26 Martin Liska <mliska@suse.cz>
19687 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
19688 always leading ';; '.
19689 (dump_bb_info): Likewise.
19690 (brief_dump_cfg): Likewise.
19691 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
19692 * dumpfile.c: Remove usage of TDF_VERBOSE.
19693 * dumpfile.h (enum dump_kind): Likewise.
19694 (dump_gimple_bb_header): Do not use TDF_COMMENT.
19695 * print-tree.c (debug_verbose): Remove.
19696 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
19697 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
19698 * tree-diagnostic.c (default_tree_printer): Replace
19699 TDF_DIAGNOSTIC with TDF_SLIM.
19701 2017-05-26 Bin Cheng <bin.cheng@arm.com>
19703 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
19704 in parameter loop, rather than loop_vinfo.
19705 (create_intersect_range_checks): Ditto.
19706 (vect_create_cond_for_alias_checks): Update call to above functions.
19708 2017-05-26 Bin Cheng <bin.cheng@arm.com>
19710 PR tree-optimization/80815
19711 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
19712 for merging runtime alias checks. Handle negative DR_STEPs.
19714 2017-05-26 Bin Cheng <bin.cheng@arm.com>
19716 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
19718 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
19719 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
19720 out code pruning runtime alias checks.
19721 * tree-data-ref.c (prune_runtime_alias_test_list): New function
19722 factored out from above.
19723 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
19725 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
19727 (prune_runtime_alias_test_list): New decalaration.
19729 2017-05-26 Bin Cheng <bin.cheng@arm.com>
19731 * tree-vect-data-refs.c (compare_tree): Rename and move ...
19732 * tree-data-ref.c (data_ref_compare_tree): ... to here.
19733 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
19734 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
19735 (operator==, comp_dr_with_seg_len_pair): Ditto.
19736 (vect_prune_runtime_alias_test_list): Ditto.
19738 2017-05-26 Martin Liska <mliska@suse.cz>
19741 * params.def: Bound partial-inlining-entry-probability param.
19743 2017-05-26 Marek Polacek <polacek@redhat.com>
19746 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
19749 2017-05-26 Richard Biener <rguenther@suse.de>
19751 PR tree-optimization/80842
19752 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
19755 2017-05-26 Richard Biener <rguenther@suse.de>
19757 PR tree-optimization/80844
19758 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
19760 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
19762 * doc/md.texi (Machine Constraints): Update x86 family
19763 machine constraints section to match 'config/i386/constraints.md'.
19765 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
19767 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
19769 2017-05-25 Nathan Sidwell <nathan@acm.org>
19771 * doc/invoke.texi (--enable-languages): Update documentation.
19773 2017-05-25 Martin Liska <mliska@suse.cz>
19775 * dumpfile.c: Add TDF_FOLDING.
19776 * dumpfile.h (enum dump_kind): Likewise.
19777 * genmatch.c (dt_simplify::gen_1): Use it.
19779 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
19781 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
19783 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
19785 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
19786 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
19788 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
19790 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
19791 * match.pd (X == C): Rewrite it here.
19792 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
19793 with_certain_nonzero_bits2): New predicates.
19794 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
19796 2017-05-24 Nathan Sidwell <nathan@acm.org>
19798 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
19801 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
19804 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
19806 * config/powerpcspe: New port. Files are copied from the rs6000
19807 port, with "rs6000" in filenames replaced by "powerpcspe".
19809 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
19811 PR rtl-optimization/80754
19812 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
19814 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
19816 * config/sparc/sparc.md (length): Return the correct value for -mflat
19817 sibcalls to match output_sibcall.
19819 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
19823 * config/rs6000/rs6000.c (struct machine_function): Add new field
19825 (rs6000_get_separate_components): Init that field, use it.
19826 (rs6000_components_for_bb): Use the field.
19828 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
19830 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
19832 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
19834 PR middle-end/80823
19835 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
19837 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19840 * config/s390/s390.c (s390_check_qrst_address): Check incoming
19841 address against address_operand predicate.
19842 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
19844 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
19846 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
19847 parameters passed indirectly.
19849 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
19851 * config/i386/i386.md (*movdi_internal): Remove SSE4
19852 alternative 18 (?r, *v). Update insn attributes.
19853 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
19854 Update insn attributes.
19855 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
19856 Update insn attributes.
19857 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
19858 alternative 1 (r, v). Remove isa attribute.
19859 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
19860 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
19861 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
19863 2017-05-23 Tom de Vries <tom@codesourcery.com>
19865 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
19868 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
19870 * cgraphunit.c (symbol_table::process_new_functions): Update.
19871 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
19872 (inline_generate_summary): Rename to ...
19873 (ipa_fn_summary_generate): ... this one.
19874 (inline_read_summary): Rename to ...
19875 (ipa_fn_summary_read): ... this one.
19876 (inline_write_summary): Rename to ...
19877 (ipa_fn_summary_write): ... this one.
19878 (inline_free_summary): Rename to ...
19879 (ipa_free_fn_summary): ... this one.
19880 (pass_data_local_fn_summary, pass_local_fn_summary,
19881 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
19882 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
19883 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
19884 make_pass_ipa_fn_summary): New.
19885 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
19886 inline_write_summary, inline_free_summary): Remove.
19887 (ipa_free_fn_summary) : New.
19888 * ipa-inline.c (ipa_inline): Update.
19889 (pass_ipa_inline): Do not generate summaries.
19890 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
19892 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
19893 and add pass_ipa_fn_summary.
19894 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
19896 (make_pass_inline_parameters): Remove.
19898 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
19900 * omp-low.c (struct omp_context): Remove "default_kind" member.
19903 * omp-offload.c (execute_oacc_device_lower): Remove the
19904 parallelism dimensions function attributes for unparallelized
19905 OpenACC kernels constructs.
19907 2017-05-23 Martin Liska <mliska@suse.cz>
19909 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
19911 (cgraph_edge::make_speculative): Likewise.
19912 (cgraph_edge::resolve_speculation): Likewise.
19913 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
19914 (cgraph_node::dump): Likewise.
19915 * cgraph.h: Likewise.
19916 * cgraphunit.c (analyze_functions): Likewise.
19917 (symbol_table::compile): Likewise.
19918 * ipa-cp.c (print_all_lattices): Likewise.
19919 (determine_versionability): Likewise.
19920 (initialize_node_lattices): Likewise.
19921 (ipcp_verify_propagated_values): Likewise.
19922 (estimate_local_effects): Likewise.
19923 (update_profiling_info): Likewise.
19924 (create_specialized_node): Likewise.
19925 (perhaps_add_new_callers): Likewise.
19926 (decide_about_value): Likewise.
19927 (decide_whether_version_node): Likewise.
19928 (identify_dead_nodes): Likewise.
19929 (ipcp_store_bits_results): Likewise.
19930 * ipa-devirt.c (dump_targets): Likewise.
19931 (ipa_devirt): Likewise.
19932 * ipa-icf.c (sem_item::dump): Likewise.
19933 (sem_function::equals): Likewise.
19934 (sem_variable::equals): Likewise.
19935 (sem_item_optimizer::read_section): Likewise.
19936 (sem_item_optimizer::execute): Likewise.
19937 (congruence_class::dump): Likewise.
19938 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
19939 (dump_inline_summary): Likewise.
19940 (estimate_node_size_and_time): Likewise.
19941 (inline_analyze_function): Likewise.
19942 * ipa-inline-transform.c (inline_call): Likewise.
19943 * ipa-inline.c (report_inline_failed_reason): Likewise.
19944 (want_early_inline_function_p): Likewise.
19945 (edge_badness): Likewise.
19946 (update_edge_key): Likewise.
19947 (inline_small_functions): Likewise.
19948 * ipa-profile.c (ipa_profile): Likewise.
19949 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
19950 (ipa_make_edge_direct_to_target): Likewise.
19951 (remove_described_reference): Likewise.
19952 (ipa_impossible_devirt_target): Likewise.
19953 (propagate_controlled_uses): Likewise.
19954 (ipa_print_node_params): Likewise.
19955 (ipcp_transform_function): Likewise.
19956 * ipa-pure-const.c (pure_const_read_summary): Likewise.
19957 (propagate_pure_const): Likewise.
19958 * ipa-reference.c (generate_summary): Likewise.
19959 (read_write_all_from_decl): Likewise.
19960 (propagate): Likewise.
19961 (ipa_reference_read_optimization_summary): Likewise.
19962 * ipa-utils.c (ipa_merge_profiles): Likewise.
19963 * ipa.c (walk_polymorphic_call_targets): Likewise.
19964 (symbol_table::remove_unreachable_nodes): Likewise.
19965 (ipa_single_use): Likewise.
19966 * passes.c (execute_todo): Likewise.
19967 * predict.c (drop_profile): Likewise.
19968 * symtab.c (symtab_node::get_dump_name): New function.
19969 (symtab_node::dump_name): Likewise.
19970 (symtab_node::dump_asm_name): Likewise.
19971 (symtab_node::dump_references): Likewise.
19972 (symtab_node::dump_referring): Likewise.
19973 (symtab_node::dump_base): Likewise.
19974 (symtab_node::debug_symtab): Likewise.
19975 * tree-sra.c (convert_callers_for_node): Likewise.
19976 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
19977 * value-prof.c (init_node_map): Likewise.
19979 2017-05-23 Martin Liska <mliska@suse.cz>
19981 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
19982 and symtab_node::debug_symtab to symbol_table::debug.
19983 * cgraphunit.c (analyze_functions): Use the renamed function.
19984 (symbol_table::compile): Likewise.
19985 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
19986 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
19987 * passes.c (execute_todo): Likewise.
19988 * symtab.c (symbol_table::dump): New function.
19989 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
19991 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
19993 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
19994 that nonconst implies exec.
19996 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
19998 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
19999 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
20000 (inline_edge_summary_vec): Turn into ...
20001 (ipa_call_summaries): ... this one.
20002 (redirect_to_unreachable, edge_set_predicate,
20003 evaluate_properties_for_edge, inline_summary_alloc,
20004 reset_ipa_call_summary, reset_inline_summary,
20005 inline_summary_t::duplicate): Update.
20006 (inline_edge_duplication_hook): Turn to ...
20007 (ipa_call_summary_t::duplicate): ... this one.
20008 (inline_edge_removal_hook): Turn to ...
20009 (ipa_call_summary_t::remove): ... this one.
20010 (dump_inline_edge_summary): Turn to ...
20011 (dump_ipa_call_summary): ... this one.
20012 (estimate_function_body_sizes): Update.
20013 (inline_update_callee_summaries): Update.
20014 (remap_edge_change_prob): Update.
20015 (remap_edge_summaries): Update.
20016 (inline_merge_summary): Update.
20017 (do_estimate_edge_time): Update.
20018 (inline_generate_summary): Update.
20019 (inline_read_section): Update.
20020 (inline_read_summary): Update.
20021 (inline_free_summary): Update.
20022 * ipa-inline.c (can_inline_edge_p): Update.
20023 (compute_inlined_call_time): Update.
20024 (want_inline_small_function_p): Update.
20025 (edge_badness): Update.
20026 (early_inliner): Update.
20027 * ipa-inline.h (inline_edge_summary): Turn to ...
20028 (ipa_call_summary): ... this one.
20029 (ipa_call_summary_t): New class.
20030 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
20031 (ipa_call_summaries): New.
20032 (inline_edge_summary): Remove.
20033 (estimate_edge_growth): Update.
20034 * ipa-profile.c (ipa_propagate_frequency_1): Update.
20035 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
20036 * ipa-split.c (execute_split_functions): Update.
20037 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
20039 2017-05-23 Tom de Vries <tom@codesourcery.com>
20041 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
20042 attributes): Document rdrand effective target.
20044 2017-05-23 Tom de Vries <tom@codesourcery.com>
20046 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
20047 attributes): Sort alphabetically.
20049 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
20051 * config/avr/genmultilib.awk: Use gsub instead of gensub.
20053 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
20056 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
20057 V2DF/V2DI splat into two separate patterns, one that handles
20058 registers, and the other that only handles memory. Drop support
20059 for splatting from a GPR on ISA 2.07 and then splitting the
20060 splat into direct move and splat.
20061 (vsx_splat_<mode>_reg): Likewise.
20062 (vsx_splat_<mode>_mem): Likewise.
20064 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
20066 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
20068 2017-05-22 Jakub Jelinek <jakub@redhat.com>
20070 PR middle-end/80809
20071 * omp-low.c (finish_taskreg_remap): New function.
20072 (finish_taskreg_scan): If unit size of ctx->record_type
20073 is non-constant, unshare the size expression and replace
20074 decls in it with possible outer var refs.
20076 PR middle-end/80809
20077 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
20078 GOVD_SHARED rather than GOVD_PRIVATE with it.
20079 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
20080 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
20082 PR middle-end/80853
20083 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
20084 as last argument to build_outer_var_ref for pointer bases of array
20085 section reductions.
20087 2017-05-19 Martin Sebor <msebor@redhat.com>
20089 * print-tree.c (print_node): Print DECL_READ_P flag.
20091 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
20093 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
20094 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
20095 * cgraph.c: Likewise.
20096 * cgraphunit.c: Likewise.
20097 * gengtype.c: Likewise.
20098 * ipa-cp.c: Likewise.
20099 * ipa-devirt.c: Likewise.
20100 * ipa-icf.c: Likewise.
20101 * ipa-predicate.c: Likewise.
20102 * ipa-profile.c: Likewise.
20103 * ipa-prop.c: Likewise.
20104 * ipa-split.c: Likewise.
20106 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
20107 edge_predicate_pool, dump_inline_hints,
20108 inline_summary::account_size_time, redirect_to_unreachable,
20109 edge_set_predicate, set_hint_predicate,
20110 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
20111 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
20112 inline_summary_t::remove, remap_hint_predicate_after_duplication,
20113 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
20114 ipa_call_summary_t::remove, initialize_growth_caches,
20115 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
20116 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
20117 mark_modified, unmodified_parm_1, unmodified_parm,
20118 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
20119 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
20120 compute_bb_predicates, will_be_nonconstant_expr_predicate,
20121 will_be_nonconstant_predicate, record_modified_bb_info,
20122 get_minimal_bb, record_modified, param_change_prob,
20123 phi_result_unknown_predicate, predicate_for_phi_result,
20124 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
20125 estimate_function_body_sizes, compute_inline_parameters,
20126 compute_inline_parameters_for_curren, pass_data_inline_parameters,
20127 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
20128 inline_update_callee_summaries, remap_edge_change_prob,
20129 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
20130 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
20131 inline_analyze_function, inline_summary_t::insert,
20132 inline_generate_summary, read_ipa_call_summary, inline_read_section,
20133 inline_read_summary, write_ipa_call_summary, inline_write_summary,
20134 inline_free_summary): Move to ipa-fnsummary.h
20135 (predicate_t): Remove.
20136 * ipa-fnsummary.c: New file.
20137 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
20138 (enum inline_hints_vals, inline_hints, agg_position_info,
20139 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
20140 inline_summaries, ipa_call_summary, ipa_call_summary_t,
20141 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
20142 dump_inline_summary, dump_inline_hints, inline_generate_summary,
20143 inline_read_summary, inline_write_summary, inline_free_summary,
20144 inline_analyze_function, initialize_inline_failed,
20145 inline_merge_summary, inline_update_overall_summary,
20146 compute_inline_parameters): Move to ipa-fnsummary.h
20147 * ipa-fnsummary.h: New file.
20148 * ipa-inline-transform.h: Include ipa-inline.h.
20149 * ipa-inline.c: LIkewise.
20151 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
20153 * ipa-inline.c (edge_badness): Use inlined_time instead of
20154 inline_summaries->get.
20156 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
20158 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
20160 2017-05-22 Nathan Sidwell <nathan@acm.org>
20162 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
20163 (fdump-lang): Document 'raw' option.
20164 * dumpfile.h (TDI_tu): Delete.
20165 * dumpfile.c (dump_files): Remove translation-unit.
20166 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
20168 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
20170 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
20171 command option from $(AWK) call.
20172 * config/avr/genmultilib.awk: Simplify and rewrite so that it
20173 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
20174 [FORMAT]: Remove handling of variable.
20175 * config/avr/t-multilib: Regenerate.
20177 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
20179 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
20181 (dump_inline_summary): Do not print self_time.
20182 (estimate_function_body_sizes): Do not set self_time.
20183 (compute_inline_parameters): Likewise.
20184 (inline_read_section, inline_write_summary): Do not stream self_time.
20185 * ipa-inline.h (inline_summary): Drop self_time.
20187 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
20189 * ipa-inline-analysis.c (account_size_time): Rename to ...
20190 (inline_summary::account_size_time): ... this one.
20191 (reset_ipa_call_summary): Turn to ...
20192 (ipa_call_summary::reset): ... this one.
20193 (reset_inline_summary): Turn to ...
20194 (inline_summary::reset): ... this one.
20195 (inline_summary_t::remove): Update.
20196 (inline_summary_t::duplicate): Update.
20197 (ipa_call_summary_t::remove): Update.
20198 (dump_inline_summary): Update.
20199 (estimate_function_body_sizes): Update.
20200 (compute_inline_parameters): Update.
20201 (estimate_node_size_and_time): Update.
20202 (inline_merge_summary): Update.
20203 (inline_update_overall_summary): Update.
20204 (inline_read_section): Update.
20205 (inline_write_summary): Update.
20206 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
20207 add account_size_time and reset member functions.
20208 (ipa_call_summary): Add reset function.
20209 * ipa-predicate.h (predicate::operator &): Constify.
20211 2017-05-22 Richard Biener <rguenther@suse.de>
20213 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
20215 2017-05-19 Jason Merrill <jason@redhat.com>
20217 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
20219 2017-05-19 Marek Polacek <polacek@redhat.com>
20222 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
20223 TYPE_OVERFLOW_WRAPS checks.
20225 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
20227 * tree-core.h (enum omp_clause_default_kind): Add
20228 "OMP_CLAUSE_DEFAULT_PRESENT".
20229 * tree-pretty-print.c (dump_omp_clause): Handle it.
20230 * gimplify.c (enum gimplify_omp_var_data): Add
20231 "GOVD_MAP_FORCE_PRESENT".
20232 (gimplify_adjust_omp_clauses_1): Map it to
20233 "GOMP_MAP_FORCE_PRESENT".
20234 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
20236 * gimplify.c (oacc_default_clause): Clarify.
20238 2017-05-19 Nathan Sidwell <nathan@acm.org>
20240 LANG_HOOK_REGISTER_DUMPS
20241 * toplev.c (general_init): Call register dump lang hook.
20242 * doc/invoke.texi: Document -fdump-lang option family.
20243 * dumpfile.c (dump_files): Remove class dump here.
20244 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
20245 * dumpfile.h (tree_dump_index): Remove TDI_class.
20246 * langhooks-def.h (lhd_register_dumps): Declare.
20247 (LANG_HOOKS_REGISTER_DUMPS): Define.
20248 (LANG_HOOKS_INITIALIZER): Add it.
20249 * langhooks.c (lhd_register_dumps): Define.
20250 * langhooks.h (struct lang_hooks): Add register_dumps.
20252 2017-05-19 Nathan Sidwell <nathan@acm.org>
20254 * context.h (context::set_passes): New.
20255 * context.c (context::context): Do not create pass manager.
20256 * toplev.c (general_init): Create pass manager here.
20258 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
20260 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
20261 use this splitter if two add or or instructions would also work for
20262 the constant we want to generate.
20264 2017-05-19 Richard Biener <rguenther@suse.de>
20267 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
20268 predicate evaluation.
20270 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
20272 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
20274 * ipa-inline.c (want_inline_small_function_p): Do not cast to
20277 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
20279 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
20280 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
20281 (inline_edge_summary_vec): Turn into ...
20282 (ipa_call_summaries): ... this one.
20283 (redirect_to_unreachable, edge_set_predicate,
20284 evaluate_properties_for_edge, inline_summary_alloc,
20285 reset_ipa_call_summary, reset_inline_summary,
20286 inline_summary_t::duplicate): Update.
20287 (inline_edge_duplication_hook): Turn to ...
20288 (ipa_call_summary_t::duplicate): ... this one.
20289 (inline_edge_removal_hook): Turn to ...
20290 (ipa_call_summary_t::remove): ... this one.
20291 (dump_inline_edge_summary): Turn to ...
20292 (dump_ipa_call_summary): ... this one.
20293 (estimate_function_body_sizes): Update.
20294 (inline_update_callee_summaries): Update.
20295 (remap_edge_change_prob): Update.
20296 (remap_edge_summaries): Update.
20297 (inline_merge_summary): Update.
20298 (do_estimate_edge_time): Update.
20299 (inline_generate_summary): Update.
20300 (inline_read_section): Update.
20301 (inline_read_summary): Update.
20302 (inline_free_summary): Update.
20303 * ipa-inline.c (can_inline_edge_p): Update.
20304 (compute_inlined_call_time): Update.
20305 (want_inline_small_function_p): Update.
20306 (edge_badness): Update.
20307 (early_inliner): Update.
20308 * ipa-inline.h (inline_edge_summary): Turn to ...
20309 (ipa_call_summary): ... this one.
20310 (ipa_call_summary_t): New class.
20311 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
20312 (ipa_call_summaries): New.
20313 (inline_edge_summary): Remove.
20314 (estimate_edge_growth): Update.
20315 * ipa-profile.c (ipa_propagate_frequency_1): Update.
20316 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
20317 * ipa-split.c (execute_split_functions): Update.
20318 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
20320 2017-05-19 Richard Biener <rguenther@suse.de>
20322 PR middle-end/80764
20323 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
20325 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
20327 * config/rs6000/rs6000.c (struct machine_function): Add field
20328 fpr_is_wrapped_separately.
20329 (rs6000_get_separate_components): Use 64 components. Handle the
20330 new FPR components.
20331 (rs6000_components_for_bb): Handle the FPR components.
20332 (rs6000_emit_prologue_components): Handle the FPR components.
20333 (rs6000_emit_epilogue_components): Handle the FPR components.
20334 (rs6000_set_handled_components): Handle the FPR components.
20335 (rs6000_emit_prologue): Don't output prologue code for those FPRs
20336 that are already separately shrink-wrapped.
20337 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
20338 that are already separately shrink-wrapped.
20340 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
20343 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
20346 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
20347 (define_peephole2 for Altivec d-form load): Add peepholes to catch
20348 cases where the register allocator uses a move and an offsettable
20349 memory operation to/from a FPR register on ISA 2.06/2.07.
20350 (define_peephole2 for Altivec d-form store): Likewise.
20352 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
20355 * config/i386/mmx.md (*mov<mode>_internal): Enable
20356 alternatives 11, 12, 13 and 14 also for 32bit targets.
20357 Remove alternatives 15, 16, 17 and 18.
20358 * config/i386/sse.md (vec_concatv2di): Change
20359 alternative (!x, *y) to (x, ?!*Yn).
20361 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
20363 * dumpfile.h (enum dump_kind): Remove stray comma.
20365 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
20367 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
20368 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
20369 predicate::num_conditions
20370 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
20371 (CHANGED): turn into predicate::changed.
20372 (agg_position_info): Move to ipa-predicate.h
20373 (add_condition, predicate::add_clause, predicate::operator &=,
20374 predicate::or_with, predicate::evaluate, predicate::probability,
20375 dump_condition, dump_clause, predicate::dump,
20376 predicate::remap_after_duplication, predicate::remap_after_inlining,
20377 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
20378 (evaluate_conditions_for_known_args): Update.
20379 (set_cond_stmt_execution_predicate): Update.
20380 * ipa-inline.h: Include ipa-predicate.h
20381 (condition, inline_param_summary, conditions, agg_position_info,
20382 predicate): Move to ipa-predicate.h
20383 * ipa-predicate.c: New file.
20384 * ipa-predicate.h: New file.
20386 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
20388 * final.c (leaf_function_p): Check we are not in a sequence.
20390 2017-05-18 Martin Liska <mliska@suse.cz>
20392 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
20393 * dumpfile.c (dump_register): Use new enum dump_kind.
20394 (get_dump_file_name): Likewise.
20395 (dump_enable_all): Likewise.
20396 (dump_switch_p_1): Likewise.
20397 (enable_rtl_dump_file): Remove usage of TDF_RTL.
20398 * dumpfile.h (enum dump_kind): New enum type.
20399 (struct dump_file_info): Create constructor and
20400 format fields and comments.
20401 * passes.c (pass_manager::register_one_dump_file):
20403 * statistics.c (statistics_early_init): Likewise.
20404 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
20405 TDF_TREE with TDF_SLIM.
20406 (gather_memory_references_ref): Likewise.
20408 2017-05-18 Martin Liska <mliska@suse.cz>
20410 * vec.h (struct vnull): Use it.
20412 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
20414 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
20415 (true_predicate, false_predicate, true_predicate_p,
20416 false_predicate_p): Remove.
20417 (single_cond_predicate, not_inlined_predicate): Turn to member function
20419 (add_condition): Update.
20420 (add_clause): Turn to...
20421 (predicate::add_clause): ... this one; update; allow passing NULL
20423 (and_predicates): Turn to ...
20424 (predicate::operator &=): ... this one.
20425 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
20426 (or_predicates): Turn to ...
20427 (predicate::or_with): ... this one.
20428 (evaluate_predicate): Turn to ...
20429 (predicate::evaluate): ... this one.
20430 (predicate_probability): Turn to ...
20431 (predicate::probability): ... this one.
20432 (dump_condition): Update.
20433 (dump_predicate): Turn to ...
20434 (predicate::dump): ... this one.
20435 (account_size_time): Update.
20436 (edge_set_predicate): Update.
20437 (set_hint_predicate): UPdate.
20438 (evaluate_conditions_for_known_args): Update.
20439 (evaluate_properties_for_edge): Update.
20440 (remap_predicate_after_duplication): Turn to...
20441 (predicate::remap_after_duplication): ... this one.
20442 (remap_hint_predicate_after_duplication): Update.
20443 (inline_summary_t::duplicate): UPdate.
20444 (dump_inline_edge_summary): Update.
20445 (dump_inline_summary): Update.
20446 (set_cond_stmt_execution_predicate): Update.
20447 (set_switch_stmt_execution_predicate): Update.
20448 (compute_bb_predicates): Update.
20449 (will_be_nonconstant_expr_predicate): Update.
20450 (will_be_nonconstant_predicate): Update.
20451 (phi_result_unknown_predicate): Update.
20452 (predicate_for_phi_result): Update.
20453 (array_index_predicate): Update.
20454 (estimate_function_body_sizes): Update.
20455 (estimate_node_size_and_time): Update.
20456 (estimate_ipcp_clone_size_and_time): Update.
20457 (remap_predicate): Rename to ...
20458 (predicate::remap_after_inlining): ... this one.
20459 (remap_hint_predicate): Update.
20460 (inline_merge_summary): Update.
20461 (inline_update_overall_summary): Update.
20462 (estimate_size_after_inlining): Update.
20463 (read_predicate): Rename to ...
20464 (predicate::stream_in): ... this one.
20465 (read_inline_edge_summary): Update.
20466 (write_predicate): Rename to ...
20467 (predicate::stream_out): ... this one.
20468 (write_inline_edge_summary): Update.
20469 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
20470 (clause_t): Turn to uint32_t
20471 (predicate): Turn to class; implement constructor and operators
20473 (size_time_entry): Update.
20474 (inline_summary): Update.
20475 (inline_edge_summary): Update.
20477 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
20479 * fold-const.c (fold_binary_loc): Move transformation...
20480 * match.pd (C - X CMP X): ... here.
20482 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
20484 * config/sparc/sparc.c (sparc_option_override): Set function
20485 alignment for -mcpu=niagara7 to 64 to match the I$ line.
20486 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
20488 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
20490 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
20492 2017-05-18 Marek Polacek <polacek@redhat.com>
20495 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
20496 (pass_ubsan::execute): Call gimple_assign_single_p instead of
20497 gimple_assign_load_p.
20499 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
20501 PR middle-end/80692
20502 * real.c (do_compare): Give decimal_do_compare preference over
20503 comparing just the signs.
20505 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
20507 * doc/md.texi (Canonicalization of Instructions): Describe the
20508 canonical form of instructions that inherently set a condition
20511 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
20513 PR middle-end/80775
20514 * tree-cfg.c: Move deletion of unreachable case statements to after
20515 the merging of consecutive case labels.
20517 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
20519 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
20520 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
20521 restoring of callee-saved registers.
20523 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
20525 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
20526 * config/visium/visium.c (single_set_and_flags): Likewise.
20527 * config/visium/visium.md (Substitutions): Likewise.
20529 2017-05-17 Martin Liska <mliska@suse.cz>
20531 * cfg.c: Introduce dump_flags_t type and
20532 use it instead of int type.
20534 * cfghooks.c: Likewise.
20535 * cfghooks.h (struct cfg_hooks): Likewise.
20536 * cfgrtl.c: Likewise.
20537 * cfgrtl.h: Likewise.
20538 * cgraph.c (cgraph_node::get_body): Likewise.
20539 * coretypes.h: Likewise.
20540 * domwalk.c: Likewise.
20541 * domwalk.h: Likewise.
20542 * dumpfile.c (struct dump_option_value_info): Likewise.
20543 (dump_enable_all): Likewise.
20544 (dump_switch_p_1): Likewise.
20545 (opt_info_switch_p): Likewise.
20546 * dumpfile.h (enum tree_dump_index): Likewise.
20547 (struct dump_file_info): Likewise.
20548 * genemit.c: Likewise.
20549 * generic-match-head.c: Likewise.
20550 * gengtype.c (open_base_files): Likewise.
20551 * gimple-pretty-print.c: Likewise.
20552 * gimple-pretty-print.h: Likewise.
20553 * graph.c (print_graph_cfg): Likewise.
20554 * graphite-scop-detection.c (dot_all_sese): Likewise.
20555 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
20556 * loop-unroll.c (report_unroll): Likewise.
20557 * passes.c (pass_manager::register_one_dump_file): Likewise.
20558 * print-tree.c: Likewise.
20559 * statistics.c: Likewise.
20560 * tree-cfg.c: Likewise.
20561 * tree-cfg.h: Likewise.
20562 * tree-dfa.c: Likewise.
20563 * tree-dfa.h: Likewise.
20564 * tree-dump.c (dump_function): Likewise.
20565 * tree-dump.h (struct dump_info): Likewise.
20566 * tree-pretty-print.c: Likewise.
20567 * tree-pretty-print.h: Likewise.
20568 * tree-ssa-live.c: Likewise.
20569 * tree-ssa-live.h: Likewise.
20570 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
20571 * tree-vect-loop.c: Likewise.
20572 * tree-vect-slp.c: Likewise.
20574 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
20575 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20577 PR tree-optimization/80457
20578 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
20579 of all arguments to a statement as scalar_to_vec operations.
20580 (vectorizable_call): Adjust call to vect_model_simple_cost for
20582 (vectorizable_conversion): Likewise.
20583 (vectorizable_assignment): Likewise.
20584 (vectorizable_shift): Likewise.
20585 (vectorizable_operation): Likewise.
20586 (vectorizable_comparison): Likewise.
20587 (vect_is_simple_cond): Record the def types for operands.
20588 (vectorizable_condition): Likewise, call vect_model_simple_cost.
20589 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
20590 for statement argument count.
20592 2017-05-16 Carl Love <cel@us.ibm.com>
20594 * config/rs6000/rs6000-c: Add support for built-in functions
20595 vector unsigned long long vec_bperm (vector unsigned long long,
20596 vector unsigned char)
20597 vector signed long long vec_mule (vector signed int,
20599 vector unsigned long long vec_mule (vector unsigned int,
20600 vector unsigned int)
20601 vector signed long long vec_mulo (vector signed int,
20603 vector unsigned long long vec_mulo (vector unsigned int,
20604 vector unsigned int)
20605 vector signed char vec_sldw (vector signed char,
20606 vector signed char,
20608 vector unsigned char vec_sldw (vector unsigned char,
20609 vector unsigned char,
20611 vector signed short vec_sldw (vector signed short,
20612 vector signed short,
20614 vector unsigned short vec_sldw (vector unsigned short,
20615 vector unsigned short,
20617 vector signed int vec_sldw (vector signed int,
20620 vector unsigned int vec_sldw (vector unsigned int,
20621 vector unsigned int,
20623 vector signed long long vec_sldw (vector signed long long,
20624 vector signed long long,
20626 vector unsigned long long vec_sldw (vector unsigned long long,
20627 vector unsigned long long,
20629 * config/rs6000/rs6000-c: Add support for built-in functions
20630 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
20631 * config/rs6000/altivec.h: Add defintion for vec_sldw.
20632 * doc/extend.texi: Update the built-in documentation for the
20633 new built-in functions.
20635 2017-05-16 Marek Polacek <polacek@redhat.com>
20639 * tree.c (save_expr): Don't fold the expression.
20641 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
20643 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
20644 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
20645 and (?*y,m). Update insn attributes.
20647 2017-05-16 Martin Liska <mliska@suse.cz>
20649 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
20650 flags argument of print_gimple_stmt, print_gimple_expr,
20651 print_generic_stmt and print_generic_expr.
20652 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
20653 * coretypes.h: Likewise.
20654 * except.c (dump_eh_tree): Likewise.
20655 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
20656 * gimple-pretty-print.h: Likewise.
20657 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
20658 (backprop::push_to_worklist): Likewise.
20659 (backprop::pop_from_worklist): Likewise.
20660 (backprop::process_use): Likewise.
20661 (backprop::intersect_uses): Likewise.
20662 (note_replacement): Likewise.
20663 * gimple-ssa-store-merging.c
20664 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
20665 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
20666 (pass_store_merging::execute): Likewise.
20667 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
20668 (ssa_base_cand_dump_callback): Likewise.
20669 (dump_incr_vec): Likewise.
20670 (replace_refs): Likewise.
20671 (replace_mult_candidate): Likewise.
20672 (create_add_on_incoming_edge): Likewise.
20673 (create_phi_basis): Likewise.
20674 (insert_initializers): Likewise.
20675 (all_phi_incrs_profitable): Likewise.
20676 (introduce_cast_before_cand): Likewise.
20677 (replace_one_candidate): Likewise.
20678 * gimplify.c (gimplify_expr): Likewise.
20679 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
20680 (set_rename): Likewise.
20681 (rename_uses): Likewise.
20682 (copy_loop_phi_nodes): Likewise.
20683 (add_close_phis_to_merge_points): Likewise.
20684 (copy_loop_close_phi_args): Likewise.
20685 (copy_cond_phi_args): Likewise.
20686 (graphite_copy_stmts_from_block): Likewise.
20687 (translate_pending_phi_nodes): Likewise.
20688 * graphite-poly.c (print_pdr): Likewise.
20689 (dump_gbb_cases): Likewise.
20690 (dump_gbb_conditions): Likewise.
20691 (print_scop_params): Likewise.
20692 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
20693 (build_cross_bb_scalars_use): Likewise.
20694 (gather_bbs::before_dom_children): Likewise.
20695 * hsa-dump.c (dump_hsa_immed): Likewise.
20696 * ipa-cp.c (print_ipcp_constant_value): Likewise.
20697 (get_replacement_map): Likewise.
20698 * ipa-inline-analysis.c (dump_condition): Likewise.
20699 (estimate_function_body_sizes): Likewise.
20700 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
20701 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
20702 * ipa-prop.c (ipa_dump_param): Likewise.
20703 (ipa_print_node_jump_functions_for_edge): Likewise.
20704 (ipa_modify_call_arguments): Likewise.
20705 (ipa_modify_expr): Likewise.
20706 (ipa_dump_param_adjustments): Likewise.
20707 (ipa_dump_agg_replacement_values): Likewise.
20708 (ipcp_modif_dom_walker::before_dom_children): Likewise.
20709 * ipa-pure-const.c (check_stmt): Likewise.
20710 (pass_nothrow::execute): Likewise.
20711 * ipa-split.c (execute_split_functions): Likewise.
20712 * omp-offload.c (dump_oacc_loop_part): Likewise.
20713 (dump_oacc_loop): Likewise.
20714 * trans-mem.c (tm_log_emit): Likewise.
20715 (tm_memopt_accumulate_memops): Likewise.
20716 (dump_tm_memopt_set): Likewise.
20717 (dump_tm_memopt_transform): Likewise.
20718 * tree-cfg.c (gimple_verify_flow_info): Likewise.
20719 (print_loop): Likewise.
20720 * tree-chkp-opt.c (chkp_print_addr): Likewise.
20721 (chkp_gather_checks_info): Likewise.
20722 (chkp_get_check_result): Likewise.
20723 (chkp_remove_check_if_pass): Likewise.
20724 (chkp_use_outer_bounds_if_possible): Likewise.
20725 (chkp_reduce_bounds_lifetime): Likewise.
20726 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
20727 (chkp_mark_completed_bounds): Likewise.
20728 (chkp_register_incomplete_bounds): Likewise.
20729 (chkp_mark_invalid_bounds): Likewise.
20730 (chkp_maybe_copy_and_register_bounds): Likewise.
20731 (chkp_build_returned_bound): Likewise.
20732 (chkp_get_bound_for_parm): Likewise.
20733 (chkp_build_bndldx): Likewise.
20734 (chkp_get_bounds_by_definition): Likewise.
20735 (chkp_generate_extern_var_bounds): Likewise.
20736 (chkp_get_bounds_for_decl_addr): Likewise.
20737 * tree-chrec.c (chrec_apply): Likewise.
20738 * tree-data-ref.c (dump_data_reference): Likewise.
20739 (dump_subscript): Likewise.
20740 (dump_data_dependence_relation): Likewise.
20741 (analyze_overlapping_iterations): Likewise.
20742 * tree-inline.c (expand_call_inline): Likewise.
20743 (tree_function_versioning): Likewise.
20744 * tree-into-ssa.c (dump_defs_stack): Likewise.
20745 (dump_currdefs): Likewise.
20746 (dump_names_replaced_by): Likewise.
20747 (dump_update_ssa): Likewise.
20748 (update_ssa): Likewise.
20749 * tree-object-size.c (pass_object_sizes::execute): Likewise.
20750 * tree-parloops.c (build_new_reduction): Likewise.
20751 (try_create_reduction_list): Likewise.
20752 (ref_conflicts_with_region): Likewise.
20753 (oacc_entry_exit_ok_1): Likewise.
20754 (oacc_entry_exit_single_gang): Likewise.
20755 * tree-pretty-print.h: Likewise.
20756 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
20757 (get_scalar_evolution): Likewise.
20758 (add_to_evolution): Likewise.
20759 (get_loop_exit_condition): Likewise.
20760 (analyze_evolution_in_loop): Likewise.
20761 (analyze_initial_condition): Likewise.
20762 (analyze_scalar_evolution): Likewise.
20763 (instantiate_scev): Likewise.
20764 (number_of_latch_executions): Likewise.
20765 (gather_chrec_stats): Likewise.
20766 (final_value_replacement_loop): Likewise.
20767 (scev_const_prop): Likewise.
20768 * tree-sra.c (dump_access): Likewise.
20769 (disqualify_candidate): Likewise.
20770 (create_access): Likewise.
20771 (reject): Likewise.
20772 (maybe_add_sra_candidate): Likewise.
20773 (create_access_replacement): Likewise.
20774 (analyze_access_subtree): Likewise.
20775 (analyze_all_variable_accesses): Likewise.
20776 (sra_modify_assign): Likewise.
20777 (initialize_constant_pool_replacements): Likewise.
20778 (find_param_candidates): Likewise.
20779 (decide_one_param_reduction): Likewise.
20780 (replace_removed_params_ssa_names): Likewise.
20781 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
20782 * tree-ssa-copy.c (dump_copy_of): Likewise.
20783 (copy_prop_visit_cond_stmt): Likewise.
20784 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
20785 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
20786 (record_equivalences_from_stmt): Likewise.
20787 * tree-ssa-dse.c (compute_trims): Likewise.
20788 (delete_dead_call): Likewise.
20789 (delete_dead_assignment): Likewise.
20790 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
20791 (forward_propagate_into_cond): Likewise.
20792 (pass_forwprop::execute): Likewise.
20793 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
20794 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
20796 (move_computations_worker): Likewise.
20797 (execute_sm): Likewise.
20798 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
20799 (remove_exits_and_undefined_stmts): Likewise.
20800 (remove_redundant_iv_tests): Likewise.
20801 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
20802 (adjust_iv_update_pos): Likewise.
20803 * tree-ssa-math-opts.c (bswap_replace): Likewise.
20804 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
20805 (value_replacement): Likewise.
20806 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
20807 * tree-ssa-pre.c (print_pre_expr): Likewise.
20808 (get_representative_for): Likewise.
20809 (create_expression_by_pieces): Likewise.
20810 (insert_into_preds_of_block): Likewise.
20811 (eliminate_insert): Likewise.
20812 (eliminate_dom_walker::before_dom_children): Likewise.
20813 (eliminate): Likewise.
20814 (remove_dead_inserted_code): Likewise.
20815 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
20816 * tree-ssa-reassoc.c (get_rank): Likewise.
20817 (eliminate_duplicate_pair): Likewise.
20818 (eliminate_plus_minus_pair): Likewise.
20819 (eliminate_not_pairs): Likewise.
20820 (undistribute_ops_list): Likewise.
20821 (eliminate_redundant_comparison): Likewise.
20822 (update_range_test): Likewise.
20823 (optimize_range_tests_var_bound): Likewise.
20824 (optimize_vec_cond_expr): Likewise.
20825 (rewrite_expr_tree): Likewise.
20826 (rewrite_expr_tree_parallel): Likewise.
20827 (linearize_expr): Likewise.
20828 (break_up_subtract): Likewise.
20829 (linearize_expr_tree): Likewise.
20830 (attempt_builtin_powi): Likewise.
20831 (attempt_builtin_copysign): Likewise.
20832 (transform_stmt_to_copy): Likewise.
20833 (transform_stmt_to_multiply): Likewise.
20834 (dump_ops_vector): Likewise.
20835 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
20836 (print_scc): Likewise.
20837 (set_ssa_val_to): Likewise.
20838 (visit_reference_op_store): Likewise.
20839 (visit_use): Likewise.
20840 (sccvn_dom_walker::before_dom_children): Likewise.
20841 (run_scc_vn): Likewise.
20842 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
20844 (expr_hash_elt::print): Likewise.
20845 (const_and_copies::pop_to_marker): Likewise.
20846 (const_and_copies::record_const_or_copy_raw): Likewise.
20847 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
20848 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
20849 (dump_predicates): Likewise.
20850 (find_uninit_use): Likewise.
20851 (warn_uninitialized_phi): Likewise.
20852 (pass_late_warn_uninitialized::execute): Likewise.
20853 * tree-ssa.c (verify_vssa): Likewise.
20854 (verify_ssa): Likewise.
20855 (maybe_optimize_var): Likewise.
20856 * tree-vrp.c (dump_value_range): Likewise.
20857 (dump_all_value_ranges): Likewise.
20858 (dump_asserts_for): Likewise.
20859 (register_edge_assert_for_2): Likewise.
20860 (vrp_visit_cond_stmt): Likewise.
20861 (vrp_visit_switch_stmt): Likewise.
20862 (vrp_visit_stmt): Likewise.
20863 (vrp_visit_phi_node): Likewise.
20864 (simplify_cond_using_ranges_1): Likewise.
20865 (fold_predicate_in): Likewise.
20866 (evrp_dom_walker::before_dom_children): Likewise.
20867 (evrp_dom_walker::push_value_range): Likewise.
20868 (evrp_dom_walker::pop_value_range): Likewise.
20869 (execute_early_vrp): Likewise.
20871 2017-05-16 Richard Biener <rguenther@suse.de>
20873 * dwarf2out.c (loc_list_from_tree_1): Do not create
20874 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
20876 2017-05-16 Richard Biener <rguenther@suse.de>
20878 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
20880 (note_variable_value_in_expr): If we resolved the decl ref
20881 do not push to the stack.
20883 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
20885 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
20886 operations in fast-math mode.
20887 (vaddq_f16): Likewise.
20888 (vmul_f16): Likewise.
20889 (vmulq_f16): Likewise.
20890 (vsub_f16): Likewise.
20891 (vsubq_f16): Likewise.
20892 * config/arm/neon.md (add<mode>3): New.
20894 (fma:<VH:mode>3): New. Also remove outdated comment.
20897 2017-05-16 Martin Liska <mliska@suse.cz>
20901 * ipa-devirt.c (warn_types_mismatch): Fix typo.
20902 (odr_types_equivalent_p): Likewise.
20904 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
20906 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
20908 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
20911 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
20912 non-interunit SSE move alternatives with '?'.
20913 (zero-extendsidi peephole2): New peephole to skip intermediate
20914 general register in SSE zero-extend sequence.
20916 2017-05-15 Jeff Law <law@redhat.com>
20918 * reorg.c (relax_delay_slots): Create a new variable to hold
20919 the temporary target rather than clobbering TARGET_LABEL.
20921 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
20922 missing argument to extract_bit_field call.
20923 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
20925 2017-05-15 Martin Liska <mliska@suse.cz>
20928 * gcc.c (process_command): Do not allow empty argument of -o option.
20930 2017-05-15 Renlin Li <renlin.li@arm.com>
20932 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
20933 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
20934 * config/aarch64/constraints.md (Usf): Add long call check.
20935 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
20936 (call_value): Likewise.
20937 (sibcall): Likewise.
20938 (sibcall_value): Likewise.
20940 (call_value_insn): New.
20941 (sibcall_insn): Update rtx pattern.
20942 (sibcall_value_insn): Likewise.
20943 (call_internal): Remove.
20944 (call_value_internal): Likewise.
20945 (sibcall_internal): Likewise.
20946 (sibcall_value_internal): Likewise.
20947 (call_reg): Likewise.
20948 (call_symbol): Likewise.
20949 (call_value_reg): Likewise.
20950 (call_value_symbol): Likewise.
20952 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
20955 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
20957 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
20959 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
20960 compatible with CCGOCmode and with CCZmode.
20962 2017-05-14 Martin Sebor <msebor@redhat.com>
20964 PR middle-end/77671
20965 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
20966 (gimple_fold_builtin_snprintf): Same.
20967 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
20968 (gimple_fold_builtin_snprintf): Same.
20969 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
20970 of character types.
20971 (is_call_safe): New function.
20972 (try_substitute_return_value): Call it.
20973 (try_simplify_call): New function.
20974 (pass_sprintf_length::handle_gimple_call): Call it.
20976 2017-05-14 Martin Sebor <msebor@redhat.com>
20978 PR middle-end/80669
20979 * builtins.c (expand_builtin_stpncpy): Simplify.
20981 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
20983 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
20984 * config/i386/i386.h
20985 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
20986 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
20987 (struct machine_function): Add new members call_ms2sysv,
20988 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
20989 (struct machine_frame_state): New fields sp_realigned and
20990 sp_realigned_offset.
20991 * config/i386/i386.c
20992 (enum xlogue_stub): New enum.
20993 (enum xlogue_stub_sets): New enum.
20994 (class xlogue_layout): New class.
20995 (struct ix86_frame): New fields stack_realign_allocate_offset,
20996 stack_realign_offset and outlined_save_offset. Modify comments to
20997 detail stack layout when using out-of-line stubs.
20998 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
20999 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
21000 -mcall-ms2sysv-xlogues.
21001 (stub_managed_regs): New static variable.
21002 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
21003 registers managed by out-of-line stub.
21004 (disable_call_ms2sysv_xlogues): New function.
21005 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
21006 m->call_ms2sysv when appropriate and compute frame layout for
21008 (sp_valid_at, fp_valid_at): New inline functions.
21009 (choose_basereg): New function.
21010 (choose_baseaddr): Add align parameter, use choose_basereg and modify
21012 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
21013 Use align parameter of choose_baseaddr to generated aligned SSE movs
21015 (pro_epilogue_adjust_stack): Modify to track
21016 machine_frame_state::sp_realigned.
21017 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
21018 (ix86_nsaved_sseregs): Likewise.
21019 (ix86_emit_save_regs): Likewise.
21020 (ix86_emit_save_regs_using_mov): Likewise.
21021 (ix86_emit_save_sse_regs_using_mov): Likewise.
21022 (get_scratch_register_on_entry): Likewise.
21023 (gen_frame_set): New function.
21024 (gen_frame_load): Likewise.
21025 (gen_frame_store): Likewise.
21026 (emit_outlined_ms2sysv_save): Likewise.
21027 (emit_outlined_ms2sysv_restore): Likewise.
21028 (ix86_expand_prologue): Modify stack re-alignment code and call
21029 emit_outlined_ms2sysv_save when appropriate.
21030 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
21031 parameter rtx_insn *insn, which allows the function to be used to only
21032 generate the notes.
21033 (ix86_expand_epilogue): Modify validity checks of frame and stack
21034 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
21035 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
21036 * config/i386/predicates.md
21037 (save_multiple): New predicate.
21038 (restore_multiple): Likewise.
21039 * config/i386/sse.md
21040 (save_multiple<mode>): New pattern.
21041 (save_multiple_realign<mode>): Likewise.
21042 (restore_multiple<mode>): Likewise.
21043 (restore_multiple_and_return<mode>): Likewise.
21044 (restore_multiple_leave_return<mode>): Likewise.
21045 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
21047 2017-05-14 Julia Koval <julia.koval@intel.com>
21049 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
21050 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
21051 (__builtin_ia32_xsetbv): New builtins.
21052 * config/i386/i386.c (ix86_expand_special_args_builtin):
21054 (ix86_expand_builtin): Special expand for new intrinsics.
21055 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
21056 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
21057 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
21059 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21061 * cfganal.c (inverted_post_order_compute): Change argument type
21063 * cfganal.h (inverted_post_order_compute): Adjust prototype.
21064 * df-core.c (rest_of_handle_df_initialize): Adjust.
21065 (rest_of_handle_df_finish): Likewise.
21066 (df_analyze_1): Likewise.
21067 (df_analyze): Likewise.
21068 (loop_inverted_post_order_compute): Change argument to be a vec *.
21069 (df_analyze_loop): Adjust.
21070 (df_get_n_blocks): Likewise.
21071 (df_get_postorder): Likewise.
21072 * df.h (struct df_d): Change field to be a vec.
21073 * lcm.c (compute_laterin): Adjust.
21074 (compute_available): Likewise.
21075 * lra-lives.c (lra_create_live_ranges_1): Likewise.
21076 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
21077 * tree-ssa-pre.c (compute_antic): Likewise.
21079 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21081 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
21082 (depth_first_search::depth_first_search): Change structure init
21083 function to this constructor.
21084 (depth_first_search::add_bb): Rename function to this member.
21085 (depth_first_search::execute): Likewise.
21086 (flow_dfs_compute_reverse_finish): Adjust.
21088 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21090 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
21091 (longest_simple_path): Likewise.
21092 * shrink-wrap.c (spread_components): Likewise.
21093 (disqualify_problematic_components): Likewise.
21094 (emit_common_heads_for_components): Likewise.
21095 (emit_common_tails_for_components): Likewise.
21096 (insert_prologue_epilogue_for_components): Likewise.
21098 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21100 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
21103 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21105 * df-core.c (df_set_blocks): Start using auto_bitmap.
21106 (df_compact_blocks): Likewise.
21107 * df-problems.c (df_rd_confluence_n): Likewise.
21108 * df-scan.c (df_insn_rescan_all): Likewise.
21109 (df_process_deferred_rescans): Likewise.
21110 (df_update_entry_block_defs): Likewise.
21111 (df_update_exit_block_uses): Likewise.
21112 (df_entry_block_bitmap_verify): Likewise.
21113 (df_exit_block_bitmap_verify): Likewise.
21114 (df_scan_verify): Likewise.
21115 * lra-constraints.c (lra_constraints): Likewise.
21116 (undo_optional_reloads): Likewise.
21117 (lra_undo_inheritance): Likewise.
21118 * lra-remat.c (calculate_gen_cands): Likewise.
21119 (do_remat): Likewise.
21120 * lra-spills.c (assign_spill_hard_regs): Likewise.
21121 (spill_pseudos): Likewise.
21122 * tree-ssa-pre.c (bitmap_set_and): Likewise.
21123 (bitmap_set_subtract_values): Likewise.
21125 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21127 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
21128 management with auto_bitmap.
21129 (fix_inter_tick): Likewise.
21130 (fix_recovery_deps): Likewise.
21131 * ira.c (add_store_equivs): Likewise.
21132 (find_moveable_pseudos): Likewise.
21133 (split_live_ranges_for_shrink_wrap): Likewise.
21134 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
21135 (rtx_reuse_manager::seen_def_p): Likewise.
21136 (rtx_reuse_manager::set_seen_def): Likewise.
21137 * print-rtl.h (class rtx_reuse_manager): Likewise.
21139 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21141 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
21143 (migrate_btr_def): Likewise.
21144 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
21145 * df-core.c (loop_post_order_compute): Likewise.
21146 (loop_inverted_post_order_compute): Likewise.
21147 * hsa-common.h: Likewise.
21148 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
21149 * init-regs.c (initialize_uninitialized_regs): Likewise.
21150 * ipa-inline.c (resolve_noninline_speculation): Likewise.
21151 (inline_small_functions): Likewise.
21152 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
21153 * ira.c (combine_and_move_insns): Likewise.
21154 (build_insn_chain): Likewise.
21155 * loop-invariant.c (find_invariants): Likewise.
21156 * lower-subreg.c (propagate_pseudo_copies): Likewise.
21157 * predict.c (tree_predict_by_opcode): Likewise.
21158 (predict_paths_leading_to): Likewise.
21159 (predict_paths_leading_to_edge): Likewise.
21160 (estimate_loops_at_level): Likewise.
21161 (estimate_loops): Likewise.
21162 * shrink-wrap.c (try_shrink_wrapping): Likewise.
21163 (spread_components): Likewise.
21164 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
21165 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
21166 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
21167 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
21168 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
21169 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
21170 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
21171 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
21172 (mark_threaded_blocks): Likewise.
21173 (thread_through_all_blocks): Likewise.
21174 * tree-ssa.c (verify_ssa): Likewise.
21175 (execute_update_addresses_taken): Likewise.
21176 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
21178 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21180 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
21182 (post_order_compute): Likewise.
21183 (inverted_post_order_compute): Likewise.
21184 (pre_and_rev_post_order_compute_fn): Likewise.
21186 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21188 * genrecog.c (int_set::int_set): Explicitly construct our
21189 auto_vec base class.
21190 * vec.h (auto_vec::auto_vec): New constructor.
21192 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21194 * bitmap.h (class auto_bitmap): New constructor taking
21195 bitmap_obstack * argument.
21197 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
21199 * bitmap.h (class auto_bitmap): Change type of m_bits to
21200 bitmap_head, and adjust ctor / dtor and member operators.
21202 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
21204 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
21205 when returned register mode doesn't match original mode.
21207 2017-05-12 Jeff Law <law@redhat.com>
21208 Jakub Jelinek <jakub@redhat.com>
21210 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
21211 we look for cc setter after the compare-elim changes.
21212 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
21213 within the vector to match what compare-elim now expects.
21214 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
21215 (xorsi3_flags, one_cmplsi2_flags): Likewise.
21217 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
21218 after the compare-elim changes.
21219 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
21220 the vector to match what compare-elim now expects.
21221 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
21222 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
21223 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
21224 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
21225 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
21227 * config/visium/visium.c (single_set_and_flags): Fix where
21228 we look for cc setter after the compare-elim changes.
21229 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
21230 with the vector to match what compare-elim now expects.
21231 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
21232 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
21233 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
21234 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
21235 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
21236 (neg<mode>2_insn_set_overflow): Likewise.
21238 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
21240 PR middle-end/79794
21241 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
21242 maybe_expand_insn call, set ops[0].target. If still set after call,
21243 set alt_rtl. Add extra arg to recursive calls.
21244 (extract_bit_field): Add alt_rtl argument. Pass to
21246 * expmed.h (extract_bit_field): Fix prototype.
21247 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
21248 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
21249 to extract_bit_field_calls.
21250 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
21251 Pass alt_rtl to extract_bit_field calls.
21252 * calls.c (store_unaligned_arguments_into_psuedos)
21253 load_register_parameters): Pass extra NULL to extract_bit_field calls.
21254 * optabs.c (maybe_legitimize_operand): Clear op->target when call
21256 * optabs.h (struct expand_operand): Add target bitfield.
21258 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
21260 * compare-elim.c (try_eliminate_compare): Canonicalize
21261 operation with embedded compare to
21262 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
21263 (set (reg) (operation)].
21265 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
21267 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
21270 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
21271 cost of adding a carry flag for ADC instruction.
21272 [case MINUS]: Ignore the cost of subtracting a carry flag
21273 for SBB instruction.
21275 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
21277 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
21279 * config/rs6000/bmiintrin.h: New file.
21280 * config/rs6000/bmi2intrin.h: New file.
21281 * config/rs6000/x86intrin.h: New file.
21283 2017-05-12 Jeff Law <law@redhat.com>
21285 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
21288 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
21290 PR middle-end/80707
21291 * tree-cfg.c: Remove cfg edges of unreachable case statements.
21293 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
21295 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
21296 early expansion of vector divide builtins.
21297 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
21298 builtins identified as having unsigned arguments.
21300 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
21302 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
21303 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
21304 expansion of vector logical operations (and, andc, or, xor,
21307 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
21309 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
21310 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
21312 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
21314 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
21315 early GIMPLE expansion of vector multiplies.
21317 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
21319 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
21320 TARGET_HAVE_MOVT conditional.
21321 (movt splitter): Likewise.
21323 2017-05-12 Richard Biener <rguenther@suse.de>
21325 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
21326 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
21327 Fold all stmts not inplace.
21329 2017-05-12 Richard Biener <rguenther@suse.de>
21331 PR tree-optimization/80713
21332 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
21333 inserted_exprs bit for not removed stmts.
21335 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
21337 PR middle-end/69921
21338 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
21339 parallelized" attribute for parallelized OpenACC kernels.
21340 * omp-offload.c (execute_oacc_device_lower): Use it.
21342 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
21343 Set "oacc kernels" attribute.
21344 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
21345 parameter. Adjust all users.
21346 (oacc_fn_attrib_kernels_p): Remove function.
21347 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
21348 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
21349 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
21350 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
21351 assert "oacc kernels" attribute is set.
21353 2017-05-11 Carl Love <cel@us.ibm.com>
21355 * config/rs6000/rs6000-c: Add support for built-in functions
21356 vector unsigned char vec_popcnt (vector signed char)
21357 vector unsigned char vec_popcnt (vector unsigned char)
21358 vector unsigned short vec_popcnt (vector signed short)
21359 vector unsigned short vec_popcnt (vector unsigned short)
21360 vector unsigned int vec_popcnt (vector signed int)
21361 vector unsigned int vec_popcnt (vector unsigned int)
21362 vector unsigned long long vec_popcnt (vector signed long long)
21363 vector unsigned long long vec_popcnt (vector unsigned long long)
21364 vector signed long long vec_slo (vector signed long long,
21365 vector signed char)
21366 vector signed long long vec_slo (vector signed long long,
21367 vector unsigned char)
21368 vector unsigned long long vec_slo (vector unsigned long long,
21369 vector signed char)
21370 vector unsigned long long vec_slo (vector unsigned long long,
21371 vector unsigned char)
21372 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
21373 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
21374 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
21375 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
21376 * doc/extend.texi: Update the built-in documentation file for the
21377 new built-in functions.
21379 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
21381 * attribs.h (sorted_attr_string): Move machine independent
21382 functions for target clone support from the i386 port to common
21383 code. Rename ix86_function_versions to common_function_versions.
21384 Rename make_name to make_unique_name.
21385 (common_function_versions): Likewise.
21386 (make_unique_name): Likewise.
21387 (make_dispatcher_decl): Likewise.
21388 (is_function_default_version): Likewise.
21389 * attribs.c (attr_strcmp): Likewise.
21390 (sorted_attr_string): Likewise.
21391 (common_function_versions): Likewise.
21392 (make_unique_name): Likewise.
21393 (make_dispatcher_decl): Likewise.
21394 (is_function_default_version): Likewise.
21395 * config/i386/i386.c (attr_strcmp): Likewise.
21396 (sorted_attr_string): Likewise.
21397 (ix86_function_versions): Likewise.
21398 (make_name): Likewise.
21399 (make_dispatcher_decl): Likewise.
21400 (is_function_default_version): Likewise.
21401 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
21403 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21406 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
21407 Account for direct move costs for vec_construct of integer
21410 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
21413 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
21414 (UNSPEC_STX_ATOMIC): Ditto.
21415 (loaddi_via_sse): New insn.
21416 (storedi_via_sse): Ditto.
21417 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
21418 Update corresponding peephole2 patterns.
21419 (atomic_storedi_fpu): Ditto.
21421 2017-05-11 Julia Koval <julia.koval@intel.com>
21423 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
21424 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
21426 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
21427 (__builtin_ia32_rsqrt14ss_mask): New builtins.
21428 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
21430 2017-05-11 Nathan Sidwell <nathan@acm.org>
21432 * graphite-poly.c: Include dumpfile.h.
21434 * dumpfle.h (dump_function): Declare here ...
21435 * tree-dump.h (dump_function): ... not here.
21436 * dumpfile.c: #include tree-cfg.h.
21437 (dump_function): Move here from ...
21438 * tree-dump.c (dump_function): ... here.
21439 * gimplify.c: #include splay-tree.h, not tree-dump.h.
21440 * graphite-poly.c: Don't include tree-dump.h.
21441 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
21442 * print-tree.c: Likewise.
21443 * stor-layout.c: Likewise.
21444 * tree-nested.c: Likewise.
21446 * dumpfile.c (dump_start): Use TDF_FLAGS.
21447 (dump_enable_all): Fix TDF_KIND check thinko.
21449 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
21451 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21452 array entries to represent two legal parameterizations of the
21453 overloaded __builtin_cmpb function, as represented by the
21454 P6_OV_BUILTIN_CMPB constant.
21455 (altivec_resolve_overloaded_builtin): Add special case handling
21456 for the __builtin_cmpb function, as represented by the
21457 P6_OV_BUILTIN_CMPB constant.
21458 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
21459 (BU_P6_64BIT_2): New macro.
21460 (BU_P6_OVERLOAD_2): New macro
21461 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
21462 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
21463 (CMPB): Add overload support to represent both 32-bit and 64-bit
21464 compare-bytes function.
21465 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
21466 support for TARGET_CMPB.
21467 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
21468 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
21469 documentation of the __builtin_cmpb overloaded built-in function.
21471 2017-05-11 Richard Biener <rguenther@suse.de>
21473 PR tree-optimization/80705
21474 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
21475 bases are not vectorizable.
21477 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21479 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
21480 when counting register pressure.
21482 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21484 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
21485 (iv_ca_more_deps): Renamed to ...
21486 (iv_ca_compare_deps): ... this.
21487 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
21489 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21491 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
21493 (determine_group_iv_costs): ... here.
21494 (find_inv_vars_cb): Record inv var if it's not recorded before.
21496 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21498 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
21499 (get_shiftadd_cost): Ditto.
21501 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21503 * tree-ssa-address.c: Include header file.
21504 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
21506 (add_to_parts): Refactor.
21507 (addr_to_parts): New parameter. Update use of move_hint_to_base.
21508 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
21511 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21513 PR tree-optimization/53090
21514 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
21516 (extract_cond_operands): Detect condition with IV on both sides
21517 and return COMP_IV_EXPR_2.
21518 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
21519 (rewrite_use_compare): Simplify by removing call to function
21520 extract_cond_operands.
21522 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21524 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
21525 (extract_cond_operands): Detect condition comparing against non-
21526 invariant bound and return appropriate enum value.
21527 (find_interesting_uses_cond): Update use of extract_cond_operands.
21528 Handle its return value accordingly.
21529 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
21531 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21533 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
21534 nonlinear iv_use computation in loop invariant sensitive way.
21536 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21538 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
21539 (find_iv_candidates): Call relate_compare_use_with_all_cands.
21541 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21543 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
21544 (dump_cand): Support iv_cand.inv_exprs.
21545 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
21547 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
21550 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21552 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
21554 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
21555 as local function. Include necessary header files.
21556 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
21558 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21560 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
21562 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21564 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
21565 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
21566 RSHIFT_EXPR and BIT_NOT_EXPR.
21568 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21570 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
21571 (adjust_setup_cost): New parameter supporting round up adjustment.
21572 (struct address_cost_data): Delete.
21573 (force_expr_to_var_cost): Don't bound cost with spill_cost.
21574 (split_address_cost, ptr_difference_cost): Delete.
21575 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
21576 (struct ainc_cost_data): New struct.
21577 (get_address_cost_ainc): New function.
21578 (get_address_cost, get_computation_cost): Reimplement.
21579 (determine_group_iv_cost_address): Record inv_expr for all uses of
21581 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
21582 (iv_ca_has_deps): Reimplemented to ...
21583 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
21585 (iv_ca_extend): Call iv_ca_more_deps.
21587 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21589 * tree-ssa-address.c (struct mem_address): Move to header file.
21590 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
21591 * tree-ssa-address.h (struct mem_address): Move from C file.
21592 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
21594 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21596 * tree-affine.h (aff_combination_type): New interface.
21597 (aff_combination_zero_p): Remove static.
21598 (aff_combination_const_p): New interface.
21599 (aff_combination_singleton_var_p): New interfaces.
21601 2017-05-11 Richard Biener <rguenther@suse.de>
21603 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
21604 Skip unreachable blocks and destinations.
21605 (eliminate): Move stmt removal and fixup ...
21606 (fini_eliminate): ... here. Skip inserted exprs.
21607 (pass_pre::execute): Move fini_pre after fini_eliminate.
21608 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
21609 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
21610 PRE to get rid of dead code that has invalid SSA form and
21611 split critical edges again.
21613 2017-05-11 Bin Cheng <bin.cheng@arm.com>
21615 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
21617 2017-05-11 Richard Biener <rguenther@suse.de>
21619 * passes.c (execute_function_todo): Verify loops if they are
21620 said to be up-to-date.
21621 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
21622 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
21624 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
21627 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
21628 handle calling assemble_external ourself.
21631 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
21632 modes with zero size. Enhance comment.
21634 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21636 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
21637 built-ins for vec_xl and vec_xst with short and char pointer
21640 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
21642 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
21643 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
21644 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
21645 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
21646 (_mm_maskz_min_round_ss): New intrinsics.
21647 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
21648 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
21649 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
21650 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
21651 (__builtin_ia32_minss_mask_round): New builtins.
21652 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
21653 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
21654 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
21656 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
21657 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
21659 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
21662 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
21664 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
21665 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
21666 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
21667 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
21668 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
21669 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
21670 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
21671 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
21672 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
21673 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
21674 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
21675 (__builtin_ia32_mulss_mask_round): New builtins.
21676 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
21677 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
21678 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
21680 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
21681 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
21683 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
21686 2017-05-10 Julia Koval <julia.koval@intel.com>
21688 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
21689 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
21690 (_mm256_setr_m128i): New intrinsics.
21692 2017-05-10 Julia Koval <julia.koval@intel.com>
21694 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
21695 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
21696 (_mm_maskz_rcp14_ss): New intrinsics.
21697 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
21698 (__builtin_ia32_rcp14ss_mask): New builtins.
21699 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
21701 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
21703 PR tree-optimization/51513
21704 * tree-cfg.c (gimple_seq_unreachable_p): New function.
21705 (assert_unreachable_fallthru_edge_p): Use it.
21706 (group_case_labels_stmt): Likewise.
21707 * tree-cfg.h: Prototype it.
21708 * stmt.c: Include cfghooks.h and tree-cfg.h.
21709 (emit_case_dispatch_table) <gap_label>: New local variable.
21710 Use it to fill dispatch table gaps.
21711 Test for default_label before updating probabilities.
21712 (expand_case) <default_label>: Remove unneeded initialization.
21713 Test for unreachable default case statement and remove its edge.
21714 Set default_label accordingly.
21715 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
21717 2017-05-10 Carl Love <cel@us.ibm.com>
21719 * config/rs6000/rs6000-c: Add support for built-in functions
21720 vector signed char vec_neg (vector signed char)
21721 vector signed short int vec_neg (vector short int)
21722 vector signed int vec_neg (vector signed int)
21723 vector signed long long vec_neg (vector signed long long)
21724 vector float vec_neg (vector float)
21725 vector double vec_neg (vector double)
21726 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
21728 * config/rs6000/altivec.h: Add define for vec_neg
21729 * doc/extend.texi: Update the built-in documentation for the
21730 new built-in functions.
21732 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21734 PR tree-optimization/77644
21735 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
21737 2017-05-10 Nathan Sidwell <nathan@acm.org>
21739 * dumpfile.h (TDI_lang_all): New.
21740 (TDF_KIND): New. Renumber others
21741 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
21743 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
21745 (get_dump_file_name): Adjust suffix generation.
21746 (dump_enable_all): Use TDF_KIND.
21747 * doc/invoke.texi (-fdump-lang-all): Document.
21749 * dumpfile.h: Tabify.
21751 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
21754 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
21755 Move member access before delete.
21757 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
21759 * tree-inline.c (expand_call_inline): Split block at stmt
21762 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
21765 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
21766 are now unused after splitting mov{sf,sd}_hardfloat.
21767 (f32_lr2): Likewise.
21768 (f32_lm): Likewise.
21769 (f32_lm2): Likewise.
21770 (f32_li): Likewise.
21771 (f32_li2): Likewise.
21772 (f32_lv): Likewise.
21773 (f32_sr): Likewise.
21774 (f32_sr2): Likewise.
21775 (f32_sm): Likewise.
21776 (f32_sm2): Likewise.
21777 (f32_si): Likewise.
21778 (f32_si2): Likewise.
21779 (f32_sv): Likewise.
21780 (f32_dm): Likewise.
21781 (f32_vsx): Likewise.
21782 (f32_av): Likewise.
21783 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
21784 For movsf, order stores so the VSX stores occur before the GPR
21785 store which encourages the register allocator to use a traditional
21786 FPR instead of a GPR. For movsd, order the stores so that the GPR
21787 store comes before the VSX stores to allow the power6 to work.
21788 This is due to the power6 not having a 32-bit integer store
21789 instruction from a FPR.
21790 (movsf_hardfloat): Likewise.
21791 (movsd_hardfloat): Likewise.
21793 2017-05-09 Martin Sebor <msebor@redhat.com>
21795 PR translation/80280
21796 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
21799 PR translation/80280
21800 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
21801 data member added in r247778.
21802 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
21804 2017-05-09 Nathan Sidwell <nathan@acm.org>
21806 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
21808 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
21811 2017-05-09 Marek Polacek <polacek@redhat.com>
21813 * doc/invoke.texi: Fix typo.
21815 2017-05-09 Richard Biener <rguenther@suse.de>
21817 * tree-vrp.c (vrp_val_is_max): Adjust comment.
21818 (vrp_val_is_min): Likewise.
21819 (set_value_range_to_value): Likewise.
21820 (set_value_range_to_nonnegative): Likewise.
21821 (gimple_assign_nonzero_p): Likewise.
21822 (gimple_stmt_nonzero_p): Likewise.
21823 (vrp_int_const_binop): Likewise. Remove unreachable case.
21824 (adjust_range_with_scev): Adjust comments.
21825 (compare_range_with_value): Likewise.
21826 (extract_range_from_phi_node): Likewise.
21827 (test_for_singularity): Likewise.
21829 2017-05-09 Richard Biener <rguenther@suse.de>
21831 * tree-vrp.c (get_single_symbol): Add assert that we don't
21832 get overflowed constants as invariant part.
21833 (compare_values_warnv): Add comment before the TREE_NO_WARNING
21834 checks. Use wi::cmp instead of recursing for integer constants.
21835 (compare_values): Just ignore whether we assumed undefined
21836 overflow instead of failing the compare.
21837 (extract_range_for_var_from_comparison_expr): Add comment before the
21838 TREE_NO_WARNING sets.
21839 (test_for_singularity): Likewise.
21840 (extract_range_from_comparison): Do not disable optimization
21841 when we assumed undefined overflow.
21842 (extract_range_basic): Remove init of unused var.
21844 2017-05-09 Richard Biener <rguenther@suse.de>
21846 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
21847 (extract_range_from_multiplicative_op_1): Adjust.
21848 (extract_range_from_binary_expr_1): Use int_const_binop.
21850 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
21853 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
21854 rs6000_store_data_bypass_p in seven define_bypass directives and
21855 in several comments.
21856 * config/rs6000/rs6000-protos.h: Add prototype for
21857 rs6000_store_data_bypass_p function.
21858 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
21859 function implements slightly different (rs6000-specific) semantics
21860 than store_data_bypass_p, returning false rather than aborting
21861 with assertion error when arguments do not satisfy the
21862 requirements of store data bypass.
21863 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
21864 rs6000_store_data_bypass_p.
21866 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
21868 * config/xtensa/xtensa-protos.h
21869 (xtensa_initial_elimination_offset): New declaration.
21870 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
21871 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
21872 macro definition, add case for FRAME_POINTER_REGNUM when
21873 FRAME_GROWS_DOWNWARD.
21874 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
21875 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
21876 xtensa_initial_elimination_offset.
21878 2017-05-08 Nathan Sidwell <nathan@acm.org>
21880 * doc/invoke.texi: Alphabetize -fdump options.
21882 2017-05-08 Martin Sebor <msebor@redhat.com>
21884 PR translation/80280
21885 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
21887 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
21889 * target.def (compute_frame_layout): New optional target hook.
21890 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
21891 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
21892 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
21894 * reload1.c (verify_initial_elim_offsets): Likewise.
21895 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
21896 (use_simple_return_p): Call arm_compute_frame_layout if needed.
21897 (arm_get_frame_offsets): Split up into this ...
21898 (arm_compute_frame_layout): ... and this function.
21900 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
21902 * config/aarch64/constraints.md (Usa): New constraint.
21903 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
21905 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
21907 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
21908 with_multilib_list after it has been checked.
21910 2017-05-08 Richard Biener <rguenther@suse.de>
21912 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
21913 (bitmap_set_subtract_values): Likewise.
21915 2017-05-08 Richard Biener <rguenther@suse.de>
21917 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
21918 (gimple_assign_nonzero): ... this and remove strict_overflow_p
21920 (gimple_stmt_nonzero_warnv_p): Rename to ...
21921 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
21923 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
21924 (extract_range_basic): Adjust, do not disable propagation on
21925 strict overflow sensitive simplification.
21926 (vrp_visit_cond_stmt): Likewise.
21928 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
21930 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
21931 body size unconditionally.
21933 2017-05-07 Jeff Law <law@redhat.com>
21936 2017-05-06 Jeff Law <law@redhat.com>
21937 PR tree-optimization/78496
21938 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
21941 PR tree-optimization/78496
21942 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
21943 (simplify_stmt_using_ranges): Call it.
21944 (vrp_dom_walker::before_dom_children): Extract equivalences
21945 from an ASSERT_EXPR with an equality comparison against a
21948 2017-05-06 Jeff Law <law@redhat.com>
21950 PR tree-optimization/78496
21951 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
21954 PR tree-optimization/78496
21955 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
21956 (simplify_stmt_using_ranges): Call it.
21957 (vrp_dom_walker::before_dom_children): Extract equivalences
21958 from an ASSERT_EXPR with an equality comparison against a
21961 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
21963 * lra-constraints.c (lra_copy_reg_equiv): New function.
21964 (split_reg): Use it to copy equivalence information from the
21965 original register to the spill register.
21967 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
21969 PR rtl-optimization/75964
21970 * simplify-rtx.c (simplify_const_relational_operation): Remove
21971 invalid handling of comparisons of integer ABS.
21973 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
21975 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
21976 initialize to zero.
21977 (init_regs): Remove declaration.
21978 (function_arg_advance_32): Initialize error_p as boolean variable.
21980 2017-05-05 Nathan Sidwell <nathan@acm.org>
21982 * store-motion.c (remove_reachable_equiv_notes): Reformat long
21983 lines. Use for (;;).
21985 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21987 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
21988 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
21989 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
21990 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
21991 VF=2 that require versioning.
21993 2017-05-05 David Malcolm <dmalcolm@redhat.com>
21995 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
21998 2017-05-05 David Malcolm <dmalcolm@redhat.com>
22000 * diagnostic.h (diagnostic_override_option_index): Convert from
22001 macro to inline function.
22003 2017-05-05 David Malcolm <dmalcolm@redhat.com>
22005 * diagnostic.c (last_module_changed_p): New function.
22006 (set_last_module): New function.
22007 (diagnostic_report_current_module): Convert macro usage to
22008 the above functions.
22009 * diagnostic.h (diagnostic_context::last_module): Strengthen
22010 from const line_map * to const line_map_ordinary *.
22011 (diagnostic_last_module_changed): Delete macro.
22012 (diagnostic_set_last_module): Delete macro.
22014 2017-05-05 David Malcolm <dmalcolm@redhat.com>
22016 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
22017 with diagnostic_report_diagnostic.
22018 (diagnostic_n_impl_richloc): Likewise.
22019 * diagnostic.h (report_diagnostic): Delete macro.
22020 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
22021 with diagnostic_report_diagnostic.
22022 * substring-locations.c (format_warning_va): Likewise.
22024 2017-05-05 David Malcolm <dmalcolm@redhat.com>
22026 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
22027 save/restor of format_spec. Move option-printing code to...
22028 (print_option_information): ...this new function, and
22029 reimplement by simply printing to the pretty_printer,
22030 rather than appending to the format string.
22032 2017-05-05 David Malcolm <dmalcolm@redhat.com>
22034 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
22035 handling logic into...
22036 (update_effective_level_from_pragmas): ...this new function.
22038 2017-05-04 Andrew Waterman <andrew@sifive.com>
22040 * config/riscv/riscv.opt (mstrict-align): New option.
22041 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
22042 (SLOW_UNALIGNED_ACCESS): Define.
22043 (riscv_slow_unaligned_access): Declare.
22044 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
22046 (riscv_slow_unaligned_access): New variable.
22047 (rocket_tune_info): Set slow_unaligned_access to true.
22048 (optimize_size_tune_info): Set slow_unaligned_access to false.
22049 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
22050 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
22051 (riscv_option_override): Set riscv_slow_unaligned_access.
22052 * doc/invoke.texi: Add -mstrict-align to RISC-V.
22054 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
22056 * config/riscv/riscv.md: Unify indentation.
22058 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22063 * config/rs6000/rs6000.md (u code attribute): Add FIX and
22065 (extendsi<mode>2): Add support for doing sign extension via
22066 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
22067 don't have ISA 3.0 instructions.
22068 (extendsi<mode>2 splitter): Likewise.
22069 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
22070 generate the normal insns since SImode can now go in vector
22071 registers. Disallow the special UNSPECs needed for previous
22072 machines to hide SImode being used. Add new insns
22073 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
22074 (fix_trunc<mode>si2_stfiwx): Likewise.
22075 (fix_trunc<mode>si2_internal): Likewise.
22076 (fixuns_trunc<mode>si2): Likewise.
22077 (fixuns_trunc<mode>si2_stfiwx): Likewise.
22078 (fctiw<u>z_<mode>_smallint): Likewise.
22079 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
22080 of floating point to 32-bit integer from doing a direct move to
22081 the GPR registers to do a store.
22082 (fctiwz_<mode>): Break long line.
22084 2017-05-05 Bin Cheng <bin.cheng@arm.com>
22086 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
22087 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
22088 (addr_list, addr_offset_valid_p): New.
22089 (split_address_groups): Check offset validity with above function.
22090 (gt-tree-ssa-loop-ivopts.h): Include header file.
22092 2017-05-05 Nathan Sidwell <nathan@acm.org>
22094 * config.gcc (arm*-*-*): Add missing 'fi'.
22096 2017-05-05 Steve Ellcey <sellcey@cavium.com>
22098 * doc/invoke.texi (-fopt-info): Explicitly say order of options
22099 included in -fopt-info does not matter.
22100 * doc/optinfo.texi (-fopt-info): Fix description of default
22101 behavour. Explicitly say order of options included in -fopt-info
22104 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
22106 * config.gcc: Allow combinations of aprofile and rmprofile values for
22107 --with-multilib-list.
22108 * config/arm/t-multilib: New file.
22109 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
22110 variables. Remove setting of ISA and floating-point ABI in
22111 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
22112 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
22113 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
22114 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
22116 * config/arm/t-rmprofile: Likewise except for the matches changes.
22117 * doc/install.texi (--with-multilib-list): Document the combination of
22118 aprofile and rmprofile values and warn about pitfalls in doing that.
22120 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
22122 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
22123 (movdi_aarch64): Likewise.
22125 2017-05-05 Jakub Jelinek <jakub@redhat.com>
22127 PR tree-optimization/80632
22128 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
22130 (build_arrays): Initialize it for virtual phis.
22131 (fix_phi_nodes): Use it for virtual phis.
22133 PR tree-optimization/80558
22134 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
22135 [x, y] op z into [x op, y op z] for op & or | if conditions
22138 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
22139 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
22142 * config/arm/arm.md (use_literal_pool): Remove.
22143 (64-bit immediate split): No longer takes cost into consideration
22144 if arm_disable_literal_pool is enabled.
22145 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
22146 used when arm_disable_literal_pool is enabled.
22147 (arm_max_const_double_inline_cost): Remove use of
22148 arm_disable_literal_pool.
22149 (push_minipool_fix): Add assert.
22150 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
22151 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
22152 (no_literal_pool_sf_immediate): New.
22154 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22156 PR tree-optimization/80613
22157 * tree-ssa-dce.c (propagate_necessity): Remove cases for
22158 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
22160 2017-05-05 Richard Biener <rguenther@suse.de>
22162 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
22164 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
22166 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
22167 of this flag from insn conditions due to removal from r247495.
22169 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
22171 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
22173 (arm_early_store_addr_dep_ptr): Likewise.
22174 * config/arm/aarch-common-protos.h
22175 (arm_early_load_addr_dep_ptr): Add prototype.
22176 (arm_early_store_addr_dep_ptr): Likewise.
22177 * config/arm/cortex-a53.md: Add new bypasses.
22179 2017-05-05 Jakub Jelinek <jakub@redhat.com>
22181 * tree.c (next_type_uid): Change type to unsigned.
22182 (type_hash_canon): Decrement back next_type_uid if
22183 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
22184 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
22187 2017-05-04 Martin Sebor <msebor@redhat.com>
22189 * builtins.c: Fix a trivial typo in a comment.
22191 PR middle-end/79234
22192 * builtins.c (check_sizes): Adjust to handle reading past the end.
22193 Avoid printing excessive upper bound of ranges. Use %E to print
22194 tree nodes instead of converting them to %wu.
22195 (expand_builtin_memchr): New function.
22196 (compute_dest_size): Rename...
22197 (compute_objsize): ...to this.
22198 (expand_builtin_memcpy): Adjust.
22199 (expand_builtin_mempcpy): Adjust.
22200 (expand_builtin_strcat): Adjust.
22201 (expand_builtin_strcpy): Adjust.
22202 (check_strncat_sizes): Adjust.
22203 (expand_builtin_strncat): Adjust.
22204 (expand_builtin_strncpy): Adjust and simplify.
22205 (expand_builtin_memset): Adjust.
22206 (expand_builtin_bzero): Adjust.
22207 (expand_builtin_memcmp): Adjust.
22208 (expand_builtin): Handle memcmp.
22209 (maybe_emit_chk_warning): Check strncat just once.
22211 2017-05-04 Martin Sebor <msebor@redhat.com>
22213 PR preprocessor/79214
22214 PR middle-end/79222
22215 PR middle-end/79223
22216 * builtins.c (check_sizes): Add inlining context and issue
22217 warnings even when -Wno-system-headers is set.
22218 (check_strncat_sizes): Same.
22219 (expand_builtin_strncat): Same.
22220 (expand_builtin_memmove): New function.
22221 (expand_builtin_stpncpy): Same.
22222 (expand_builtin): Handle memmove and stpncpy.
22224 2017-05-04 Bin Cheng <bin.cheng@arm.com>
22226 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
22227 which is not used any more.
22229 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
22231 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
22233 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
22235 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
22236 (cortexa53_tunings): Likewise.
22237 (cortexa57_tunings): Likewise.
22238 (cortexa72_tunings): Likewise.
22239 (cortexa73_tunings): Likewise.
22241 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
22243 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
22244 Set loop alignment to 8.
22246 2017-05-04 Martin Sebor <msebor@redhat.com>
22248 PR translation/80280
22249 * builtins.c (expand_builtin_object_size): Add missing quoting to
22250 %D and like directives.
22251 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
22252 (hsa_type_for_tree_type): Same.
22253 (verify_function_arguments): Same.
22254 * symtab.c (symbol_table::change_decl_assembler_name): Same.
22255 * varasm.c (get_section): Same.
22258 2017-05-04 Martin Sebor <msebor@redhat.com>
22260 PR translation/80280
22261 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
22263 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
22265 * config/aarch64/aarch64.c (generic_addrcost_table):
22266 Change HI/TI mode setting.
22268 2017-05-04 Martin Jambor <mjambor@suse.cz>
22270 PR tree-optimization/80622
22271 * tree-sra.c (comes_initialized_p): New function.
22272 (build_accesses_from_assign): Only set write lazily when
22273 comes_initialized_p is false.
22274 (analyze_access_subtree): Use comes_initialized_p.
22275 (propagate_subaccesses_across_link): Assert !comes_initialized_p
22276 instead of testing for PARM_DECL.
22278 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22280 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
22281 constraint on operand 0 to allow more general addressing modes.
22282 Adjust output template.
22283 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
22285 * config/aarch64/aarch64-protos.h
22286 (aarch64_address_valid_for_prefetch_p): Declare prototype.
22287 * config/aarch64/constraints.md (Dp): New address constraint.
22288 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
22291 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
22293 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
22294 update use of estimate_ipcp_clone_size_and_time.
22295 (estimate_local_effects): Update use of
22296 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
22297 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
22298 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
22299 Return nonspecialized time.
22301 2017-05-04 Richard Biener <rguenther@suse.de>
22303 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
22304 for the last VUSE which def dominates the PHI. Directly call
22306 (get_continuation_for_phi_1): Remove.
22308 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
22310 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
22311 to explain the use of truncating division. Cap the number of
22312 iterations to the maximum given by nb_iterations_upper_bound,
22315 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
22317 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
22318 * configure: Regenerate.
22319 * config.in: Regenerate.
22320 * config/i386/driver-mingw32.c: new file.
22321 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
22322 * config.host: Link driver-mingw32.o on MinGW host.
22323 * doc/install.texi: Document new --enable-mingw-wildcard configure
22326 2017-05-04 Marek Polacek <polacek@redhat.com>
22328 PR tree-optimization/80612
22329 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
22331 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
22332 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
22334 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
22335 (movt splitter): Likewise.
22336 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
22337 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
22338 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
22339 block for Thumb-1 with MOVT.
22340 (thumb2_legitimate_address_p): Move code block ...
22341 (can_avoid_literal_pool_for_label_p): ... into this new function.
22342 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
22344 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
22345 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
22346 "M-profile targets with the MOVT instruction".
22348 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
22350 * config/arm/arm-builtins.c (arm_init_builtins): Rename
22351 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
22352 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
22354 2017-05-04 Martin Liska <mliska@suse.cz>
22356 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
22357 variable cond_code.
22359 2017-05-04 Richard Biener <rguenther@suse.de>
22361 * tree.c (array_at_struct_end_p): Handle arrays at struct
22362 end with flexarrays more conservatively. Refactor and treat
22363 arrays of arrays or aggregates more strict. Fix
22364 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
22365 * tree.c (array_at_struct_end_p): Adjust prototype.
22366 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
22367 * gimple-fold.c (get_range_strlen): Likewise.
22368 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
22370 2017-05-04 Richard Biener <rguenther@suse.de>
22372 PR tree-optimization/31130
22373 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
22375 (supports_overflow_infinity): Likewise.
22376 (is_negative_overflow_infinity): Likewise.
22377 (is_positive_overflow_infinity): Likewise.
22378 (is_overflow_infinity): Likewise.
22379 (stmt_overflow_infinity): Likewise.
22380 (overflow_infinity_range_p): Likewise.
22381 (usable_range_p): Remove as always returning true.
22382 (make_overflow_infinity): Remove.
22383 (negative_overflow_infinity): Likewise.
22384 (positive_overflow_infinity): Likewise.
22385 (avoid_overflow_infinity): Likewise.
22386 (set_value_range): Adjust accordingly.
22387 (set_value_range_to_nonnegative): Likewise, remove now unused
22388 overflow_infinity arg.
22389 (vrp_operand_equal_p): Adjust.
22390 (update_value_range): Likewise.
22391 (range_int_cst_singleton_p): Likewise.
22392 (operand_less_p): Likewise.
22393 (compare_values_warnv): Likewise.
22394 (extract_range_for_var_from_comparison_expr): Likewise.
22395 (vrp_int_const_binop): Likewise.
22396 (zero_nonzero_bits_from_vr): Likewise.
22397 (extract_range_from_multiplicative_op_1): Likewise.
22398 (extract_range_from_binary_expr_1): Likewise.
22399 (extract_range_from_unary_expr): Likewise.
22400 (extract_range_from_comparison): Likewise.
22401 (extract_range_basic): Likewise.
22402 (adjust_range_with_scev): Likewise.
22403 (compare_ranges): Likewise.
22404 (compare_range_with_value): Likewise.
22405 (dump_value_range): Likewise.
22406 (test_for_singularity): Likewise, remove strict_overflow_p parameter
22408 (simplify_cond_using_ranges): Adjust.
22410 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
22412 * brig-builtins.def: Added a builtin for class_f64.
22413 * builtin-types.def: Added a builtin type needed by class_f64.
22415 2017-05-03 Jason Merrill <jason@redhat.com>
22417 * timevar.def: Add TV_CONSTEXPR.
22419 2017-05-03 David Malcolm <dmalcolm@redhat.com>
22421 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
22423 2017-05-03 Martin Jambor <mjambor@suse.cz>
22425 * ipa-prop.c (ipa_update_after_lto_read): Removed.
22426 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
22427 * ipa-cp.c (ipcp_propagate_stage): Do not call
22428 ipa_update_after_lto_read.
22429 * ipa-inline.c (ipa_inline): Likewise.
22431 2017-05-03 Martin Jambor <mjambor@suse.cz>
22433 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
22434 tag. Added a default constructor and a destructor.
22435 (ipa_edge_args_sum_t): New class;
22436 (ipa_edge_args_sum): Declare.
22437 (ipa_edge_args_vector): Remove declaration.
22438 (IPA_EDGE_REF): Use ipa_edge_args_sum.
22439 (ipa_free_edge_args_substructures): Remove declaration.
22440 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
22441 (ipa_edge_args_info_available_for_edge_p): Likewise.
22442 * ipa-prop.c (ipa_edge_args_vector): Removed.
22443 (edge_removal_hook_holder): Likewise.
22444 (edge_duplication_hook_holder): Likewise.
22445 (ipa_edge_args_sum): New variable.
22446 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
22447 ipa_edge_args_vector.
22448 (ipa_free_edge_args_substructures): Likewise.
22449 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
22450 ipa_edge_args_vector.
22451 (ipa_edge_removal_hook): Turned into method
22452 ipa_edge_args_sum_t::remove.
22453 (ipa_edge_duplication_hook): Turned into method
22454 ipa_edge_args_sum_t::duplicate.
22455 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
22456 registering edge hooks.
22457 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
22458 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
22459 ipa_edge_args_sum instead of ipa_edge_args_vector.
22460 * ipa-profile.c (ipa_profile): Likewise.
22462 2017-05-03 Martin Jambor <mjambor@suse.cz>
22464 * symbol-summary.h (function_summary): New method exists.
22465 (function_summary::symtab_removal): Deallocate through release.
22466 (call_summary): New class.
22467 (gt_ggc_mx): New overload.
22468 (gt_pch_nx): Likewise.
22469 (gt_pch_nx): Likewise.
22471 2017-05-03 Jeff Law <law@redhat.com>
22473 PR tree-optimization/78496
22474 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
22475 from simplify_cond_using_ranges. Split off code to walk
22476 backwards through casts into ...
22477 (simplify_cond_using_ranges_2): New function.
22478 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
22479 (execute_vrp): After identifying jump threads, call
22480 simplify_cond_using_ranges_2.
22482 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
22485 * ipa-inline.h (inline_summary): Add ctor.
22486 (create_ggc): Do not use ggc_cleared_alloc.
22488 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
22489 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22491 * gcc.c (handle_braces): Support escaping in switch matching
22493 * doc/invoke.texi (Spec Files): Document it.
22494 Remove superfluous @code markup in items.
22496 2017-05-03 David Malcolm <dmalcolm@redhat.com>
22498 * diagnostic-show-locus.c (struct column_range): New struct.
22499 (get_affected_columns): New function.
22500 (get_printed_columns): New function.
22501 (struct correction): New struct.
22502 (correction::ensure_capacity): New function.
22503 (correction::ensure_terminated): New function.
22504 (struct line_corrections): New struct.
22505 (line_corrections::~line_corrections): New dtor.
22506 (line_corrections::add_hint): New function.
22507 (layout::print_trailing_fixits): Reimplement in terms of the new
22509 (selftest::test_overlapped_fixit_printing): New function.
22510 (selftest::diagnostic_show_locus_c_tests): Call it.
22512 2017-05-03 Nathan Sidwell <nathan@acm.org>
22514 Canonicalize canonical type hashing
22515 * tree.h (type_hash_canon_hash): Declare.
22516 * tree.c (type_hash_list, attribute_hash_list): Move into
22517 type_hash_canon_hash.
22518 (build_type_attribute_qual_variant): Break out hash code calc into
22519 type_hash_canon_hash.
22520 (type_hash_canon_hash): New. Generic type hash computation.
22521 (build_range_type_1, build_array_type_1, build_function_type,
22522 build_method_type_directly, build_offset_type, build_complex_type,
22523 make_vector_type): Call it.
22525 2017-05-03 Richard Biener <rguenther@suse.de>
22527 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
22528 When all DRs have unknown misaligned do not always peel
22529 when there is a store but apply the same costing model as if
22530 there were only loads.
22532 2017-05-03 Richard Biener <rguenther@suse.de>
22535 PR tree-optimization/80492
22536 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
22537 compare_base_decls returning dont-know properly.
22539 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
22541 * config/arm/iterators.md (CCSI): New mode iterator.
22542 (arch): New mode attribute.
22543 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
22544 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
22545 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
22546 code iterator for success result mode.
22547 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
22548 the corresponding new insn generators.
22550 2017-05-03 Bin Cheng <bin.cheng@arm.com>
22553 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22554 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
22556 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
22558 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
22559 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
22560 (DDR_A): Wrap DDR argument in brackets.
22561 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
22562 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
22563 (DDR_REVERSED_P): Likewise.
22565 2017-05-03 Jakub Jelinek <jakub@redhat.com>
22567 PR tree-optimization/79472
22568 * tree-switch-conversion.c (struct switch_conv_info): Add
22569 contiguous_range and default_case_nonstandard fields.
22570 (collect_switch_conv_info): Compute contiguous_range and
22571 default_case_nonstandard fields, don't clear final_bb if
22572 contiguous_range and only the default case doesn't have the required
22574 (check_all_empty_except_final): Set default_case_nonstandard instead
22575 of failing if contiguous_range and the default case doesn't have empty
22577 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
22578 and only the default case doesn't have the required constants. Skip
22580 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
22581 if default_case_nonstandard.
22582 (build_constructors): Build constant 1 just once. Assert that default
22583 values aren't inserted in between cases if contiguous_range. Skip
22585 (build_arrays): Skip virtual phis.
22586 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
22587 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
22588 Handle virtual phis.
22589 (gen_inbound_check): Handle default_case_nonstandard case.
22590 (process_switch): Adjust check_final_bb caller. Call
22591 gather_default_values with the first non-default case instead of
22592 default case if default_case_nonstandard.
22594 2017-05-02 Nathan Sidwell <nathan@acm.org>
22596 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
22597 check. Fix formatting.
22599 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
22601 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
22602 errors when comparing specialized and unspecialized times.
22604 2017-05-02 David Malcolm <dmalcolm@redhat.com>
22606 * diagnostic-show-locus.c
22607 (layout::should_print_annotation_line_p): Make private.
22608 (layout::print_annotation_line): Make private.
22609 (layout::annotation_line_showed_range_p): Make private.
22610 (layout::show_ruler): Make private.
22611 (layout::print_source_line): Make private. Pass in line and
22612 line_width, rather than calling location_get_source_line. Drop
22614 (layout::print_leading_fixits): New method.
22615 (layout::print_any_fixits): Rename to...
22616 (layout::print_trailing_fixits): ...this, and make private.
22617 Don't print newline fixits.
22618 (diagnostic_show_locus): Move logic for printing one row into...
22619 (layout::print_line): ...this new function. Move the
22620 location_get_source_line call and error-handling from
22621 print_source_line to here. Call print_leading_fixits, and rename
22622 print_any_fixits to print_trailing_fixits.
22623 (selftest::test_fixit_insert_containing_newline): Update now that
22624 newlines are partially supported.
22625 (selftest::test_fixit_insert_containing_newline_2): New test.
22626 (selftest::test_fixit_replace_containing_newline): Update comments.
22627 (selftest::diagnostic_show_locus_c_tests): Call the new test.
22628 * edit-context.c (class added_line): New class.
22629 (class edited_line): Describe newline handling in comment.
22630 (edited_line::actually_edited_p): New method.
22631 (edited_line::print_content): Delete redundant decl.
22632 (edited_line::m_predecessors): New field.
22633 (edited_file::print_content): Call edited_line::print_content.
22634 (edited_file::print_diff): Update to support newlines.
22635 (edited_file::print_diff_hunk): Likewise.
22636 (edited_file::print_run_of_changed_lines): New function.
22637 (edited_file::print_diff_line): Convert to...
22638 (print_diff_line): ...this.
22639 (edited_file::get_effective_line_count): New function.
22640 (edited_line::edited_line): Initialize new field m_predecessors.
22641 (edited_line::~edited_line): Clean up m_predecessors.
22642 (edited_line::apply_fixit): Handle newlines.
22643 (edited_line::get_effective_line_count): New function.
22644 (edited_line::print_content): New function.
22645 (edited_line::print_diff_lines): New function.
22646 (selftest::test_applying_fixits_insert_containing_newline): New
22648 (selftest::test_applying_fixits_replace_containing_newline): New
22650 (selftest::insert_line): New function.
22651 (selftest::test_applying_fixits_multiple_lines): Add example of
22653 (selftest::edit_context_c_tests): Call the new tests.
22655 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22657 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
22658 parameter cand. Update dump information.
22659 (get_computation_cost): Update uses.
22661 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22663 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
22664 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
22665 (get_computation_at, rewrite_use_address): Update use of
22666 get_computation_aff.
22668 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22670 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
22671 (get_computation): Delete.
22672 (get_computation_cost): Implement like get_computation_cost_at.
22673 Use get_computation_at.
22674 (get_computation_cost_at): Delete.
22675 (rewrite_use_nonlinear_expr): Use get_computation_at.
22676 (rewrite_use_compare, remove_unused_ivs): Ditto.
22678 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22680 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
22682 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22684 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
22685 (ivopts_global_cost_for_size): Rename parameter and update uses.
22686 (iv_ca_recount_cost): Update uses.
22687 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
22688 candidates seperately in n_invs and n_cands.
22689 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
22691 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22693 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
22694 (find_inv_vars_cb): New.
22695 (find_depends): Renamed to ...
22696 (find_inv_vars): ... this.
22697 (add_candidate_1, force_var_cost): Call find_inv_vars.
22698 (split_address_cost, determine_group_iv_cost_cond): Ditto.
22700 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22702 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
22703 inv_vars. Add inv_exprs.
22704 (struct iv_cand): Rename depends_on to inv_vars.
22705 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
22706 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
22707 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
22708 (dump_cand): Dump inv_vars.
22709 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
22710 (record_invariant, find_depends, add_candidate_1): Ditto.
22711 (set_group_iv_cost, force_var_cost): Ditto.
22712 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
22713 (get_computation_cost_at, get_computation_cost): Ditto.
22714 (determine_group_iv_cost_generic): Ditto.
22715 (determine_group_iv_cost_address): Ditto.
22716 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
22717 (determine_group_iv_costs): Ditto.
22718 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
22719 (iv_ca_set_remove_invariants): Renamed to ...
22720 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
22721 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
22722 (iv_ca_set_add_invariants): Renamed to ...
22723 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
22724 (iv_ca_set_cp): Use iv_ca_set_add_invs.
22725 (iv_ca_has_deps): Support inv_vars and inv_exprs.
22726 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
22727 (create_new_ivs): Remove useless dump.
22729 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22731 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
22733 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
22734 (iv_ca_set_no_cp, create_new_iv): Ditto.
22736 2017-05-02 Bin Cheng <bin.cheng@arm.com>
22738 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
22740 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
22742 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
22743 function tree_check2.
22745 2017-05-02 Martin Liska <mliska@suse.cz>
22747 * doc/gcov.texi: Add missing preposition.
22748 * gcov.c (function_info::function_info): Properly fill up
22749 all member variables.
22751 2017-05-02 Tamar Christina <tamar.christina@arm.com>
22753 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
22755 2017-05-02 Tamar Christina <tamar.christina@arm.com>
22757 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
22759 2017-05-02 Martin Liska <mliska@suse.cz>
22762 * lto-streamer-in.c (lto_read_tree_1): Remove
22763 LTO_STREAMER_DEBUG.
22764 * lto-streamer.c (struct tree_hash_entry): Likewise.
22765 (struct tree_entry_hasher): Likewise.
22766 (tree_entry_hasher::hash): Likewise.
22767 (tree_entry_hasher::equal): Likewise.
22768 (lto_streamer_init): Likewise.
22769 (lto_orig_address_map): Likewise.
22770 (lto_orig_address_get): Likewise.
22771 (lto_orig_address_remove): Likewise.
22772 * lto-streamer.h: Likewise.
22773 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
22774 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
22776 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
22778 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
22779 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
22780 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
22781 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
22782 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
22783 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
22784 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
22785 (mm_maskz_sub_ss): New intrinsics.
22786 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
22787 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
22788 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
22789 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
22790 (__builtin_ia32_subss_mask_round): New builtins.
22791 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
22792 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
22793 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
22795 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
22796 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
22798 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
22801 2017-05-02 Martin Jambor <mjambor@suse.cz>
22803 PR tree-optimization/78687
22804 * tree-sra.c (access): New field parent.
22805 (process_subtree_disqualification): New function.
22806 (disqualify_candidate): Call it.
22807 (build_accesses_from_assign): Reset write flag if creating an
22809 (build_access_subtree): Fill in parent field and also prpagate
22810 down grp_write flag.
22811 (create_artificial_child_access): New parameter set_grp_write, set
22812 grp_write to its value.
22813 (propagate_subaccesses_across_link): Also propagate grp_write flag
22815 (propagate_all_subaccesses): Push the closest parent back to work
22816 queue if add_access_to_work_queue returned true.
22818 2017-05-02 Richard Biener <rguenther@suse.de>
22820 * common.opt (fstrict-overflow): Alias negative to fwrapv.
22821 * doc/invoke.texi (fstrict-overflow): Remove all traces of
22822 -fstrict-overflow documentation.
22823 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
22824 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
22825 flag_strict_overflow.
22826 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
22827 * lto-opts.c (lto_write_options): Do not stream it.
22828 * lto-wrapper.c (merge_and_complain): Do not handle it.
22829 * opts.c (default_options_table): Do not set -fstrict-overflow.
22830 (finish_options): Likewise do not clear it when sanitizing.
22831 * simplify-rtx.c (simplify_const_relational_operation): Do not
22832 test flag_strict_overflow.
22834 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
22836 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
22837 using enabled attribute.
22838 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
22839 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
22840 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
22841 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
22842 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
22843 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
22844 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
22845 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
22846 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
22847 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
22849 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
22851 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
22853 2017-05-02 Richard Biener <rguenther@suse.de>
22855 PR tree-optimization/80591
22857 2017-04-10 Richard Biener <rguenther@suse.de>
22859 * tree-ssa-structalias.c (find_func_aliases): Properly handle
22862 2017-05-02 Richard Biener <rguenther@suse.de>
22864 PR tree-optimization/80549
22865 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
22866 (cleanup_tree_cfg_noloop): Create forwarders to known loop
22867 headers if they do not have a preheader.
22869 2017-05-02 Martin Liska <mliska@suse.cz>
22872 * common.opt: Fix typo.
22873 * doc/invoke.texi: Likewise.
22875 2017-05-01 Jan Beulich <jbeulich@suse.com>
22877 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
22878 swapping, add (x,x,m,x,n) alternative.
22880 2017-05-01 Nathan Sidwell <nathan@acm.org>
22882 * calls.c (combine_pending_stack_adjustment_and_call): Remove
22883 unnecessary unadjusted_alignment check.
22885 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
22888 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
22890 * gimplify.c (gimplify_cilk_detach): New function.
22891 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
22892 * tree-core.h: Document EXPR_CILK_SPAWN.
22893 * tree.h (EXPR_CILK_SPAWN): Define.
22895 2017-05-01 David Malcolm <dmalcolm@redhat.com>
22897 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
22898 to use new fixit_hint representation, using the "replace" logic.
22899 (get_line_span_for_fixit_hint): Likewise.
22900 (layout::print_any_fixits): Likewise.
22901 (selftest::test_one_liner_many_fixits): Rename to...
22902 (selftest::test_one_liner_many_fixits_1): ...this, and update
22903 comment and expected output to reflect that the multiple fix-it
22904 hints are now consolidated into one insertion.
22905 (selftest::test_one_liner_many_fixits_2): New test.
22906 (selftest::test_diagnostic_show_locus_one_liner): Update for
22908 (selftest::test_fixit_consolidation): Update for fix-it API
22910 * diagnostic.c (print_parseable_fixits): Likewise.
22911 * edit-context.c (edited_line::m_line_events): Convert from
22912 auto_vec <line_event *> to auto_vec <line_event>.
22913 (class line_event): Convert from abstract base class to a concrete
22914 class, taking over the role of replace_event.
22915 (class insert_event): Delete.
22916 (class replace_event): Rename to class line_event. Convert to
22918 (edit_context::add_fixits): Reimplement.
22919 (edit_context::apply_insert): Delete.
22920 (edit_context::apply_replace): Rename to...
22921 (edit_context::apply_fixit): ...this. Convert to half-open range.
22922 (edited_file::apply_insert): Delete.
22923 (edited_file::apply_replace): Rename to...
22924 (edited_file::apply_fixit): ...this.
22925 (edited_line::~edited_line): Drop deletion of events.
22926 (edited_line::apply_insert): Delete.
22927 (edited_line::apply_replace): Rename to...
22928 (edited_line::apply_fixit): ...this. Convert to half-open range.
22929 Update for change to type of m_line_events.
22930 * edit-context.h (edit_context::apply_insert): Delete.
22931 (edit_context::apply_replace): Rename to...
22932 (edit_context::apply_fixit): ...this.
22934 2017-05-01 Martin Sebor <msebor@redhat.com>
22936 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
22939 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
22942 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
22943 __get_cpuid_max returns 0.
22944 (__get_cpuid_count): Ditto.
22946 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
22948 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
22949 replacement expression is another instance of one of its arguments.
22951 2017-05-01 Jakub Jelinek <jakub@redhat.com>
22954 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
22955 check for stack push/pop autoinc.
22956 * config/i386/i386.c (ix86_agi_dependent): Return false
22957 if the only reason why modified_in_p returned true is that
22958 addr is SP based and set_insn is a push or pop.
22960 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
22962 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
22965 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
22968 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
22969 (account_size_time): Use two predicates - exec_pred and
22971 (evaluate_conditions_for_known_args): Compute both clause and
22973 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
22974 (inline_summary_t::duplicate): Update.
22975 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
22977 (compute_inline_parameters): Likewise.
22978 (estimate_edge_size_and_time): Update caluclation of time.
22979 (estimate_node_size_and_time): Compute both time and nonspecialized
22981 (estimate_ipcp_clone_size_and_time): Update.
22982 (inline_merge_summary): Update.
22983 (do_estimate_edge_time): Update.
22984 (do_estimate_edge_size): Update.
22985 (do_estimate_edge_hints): Update.
22986 (inline_read_section, inline_write_summary): Stream both new predicates.
22987 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
22989 (compute_inlined_call_time): Cleanup.
22990 (big_speedup_p): Update.
22991 (edge_badness): Update.
22992 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
22993 (size_time_entry): Replace predicate by exec_predicate and
22994 nonconst_predicate.
22995 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
22996 (estimate_edge_time): Return also nonspec_time.
22997 (reset_edge_growth_cache): Update.
22999 2017-04-29 Jakub Jelinek <jakub@redhat.com>
23001 PR rtl-optimization/80491
23002 * ifcvt.c (noce_process_if_block): When looking for x setter
23003 with missing else_bb, don't check only the insn right before
23004 cond_earliest, but look for the last insn that x is modified in
23005 within the same bb.
23007 PR rtl-optimization/80491
23008 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
23010 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
23012 PR tree-optimization/80487
23013 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
23015 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23017 PR tree-optimization/79697
23018 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
23019 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
23020 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
23022 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
23023 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
23025 2017-04-28 Martin Sebor <msebor@redhat.com>
23027 PR tree-optimization/80523
23028 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
23029 (init_target_to_host_charmap, target_to_host, target_strtol10): New
23031 (maybe_warn, format_directive, parse_directive): Use new functions.
23032 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
23034 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
23036 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
23038 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
23040 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
23041 target_header_dir): Set correctly.
23042 * configure: Regenerated.
23043 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
23044 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
23045 instead of SYSTEM_HEADER_DIR.
23047 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
23049 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
23050 (estimate_local_effects): Likewise.
23051 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
23052 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
23053 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
23054 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
23055 do_estimate_edge_time, estimate_edge_time): Likewise.
23056 * ipa-inline-analysis.c (estimate_node_size_and_time,
23057 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
23058 (estimate_time_after_inlining): Remove.
23060 2017-04-28 Martin Liska <mliska@suse.cz>
23062 * doc/gcov.texi: Enhance documentation of gcov.
23064 2017-04-28 Martin Liska <mliska@suse.cz>
23066 * doc/gcov.texi: Sort options in alphabetic order.
23067 * doc/gcov-dump.texi: Likewise.
23068 * doc/gcov-tool.texi: Likewise.
23069 * gcov.c (print_usage): Likewise.
23070 * gcov-dump.c (print_usage): Likewise.
23071 * gcov-tool.c (print_merge_usage_message): Likewise.
23072 (print_rewrite_usage_message): Likewise.
23073 (print_overlap_usage_message): Likewise.
23075 2017-04-28 Martin Liska <mliska@suse.cz>
23077 PR gcov-profile/53915
23078 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
23080 2017-04-28 Martin Liska <mliska@suse.cz>
23082 PR gcov-profile/79891
23083 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
23084 is marked by compiler as living on a line.
23085 (get_cycles_count): Remove usage of the union.
23086 (output_intermediate_file): Likewise.
23087 (find_source): Fix GNU coding style.
23088 (accumulate_line_counts): Remove old non-all block mode.
23089 (output_lines): Remove usage of the union.
23090 * profile.c (output_location): Include all BBs, even if
23091 belonging to a same line (and file) as a previous BB.
23093 2017-04-28 Martin Liska <mliska@suse.cz>
23095 * gcov.c (process_args): Handle new argument 'w'.
23096 (read_graph_file): Assign ID to BBs.
23097 (output_branch_count): Display BB # if verbose flag is set.
23098 (output_lines): Likewise for arcs.
23099 (print_usage): Add '--verbose' option help.
23100 * doc/gcov.texi: Document --verbose (-w) option.
23102 2017-04-28 Martin Liska <mliska@suse.cz>
23104 * gcov.c (struct block_location_info): New struct.
23105 (process_file): Fill up the new structure.
23106 (read_graph_file): Replace usage of encoding by the newly added
23108 (add_line_counts): Likewise.
23109 (accumulate_line_counts): Remove usage of the union.
23110 (function_info::function_info): New function.
23111 (function_info::~function_info): Likewise.
23112 (process_file): Call delete instead of release_function.
23113 (release_function): Release the function.
23114 (release_structures): Call delete instead of release_function.
23115 (solve_flow_graph): Replace usage of num_blocks.
23116 (find_exception_blocks): Likewise.
23117 (output_lines): Fix GNU coding style.
23119 2017-04-28 Martin Liska <mliska@suse.cz>
23122 * coverage.c (coverage_remove_note_file): New function.
23123 * coverage.h: Declare the function.
23124 * toplev.c (finalize): Clean if an error has been seen.
23126 2017-04-28 Martin Liska <mliska@suse.cz>
23128 PR gcov-profile/80031
23129 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
23130 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
23131 * gcov.c (read_graph_file): Read just number of blocks.
23132 * profile.c (branch_prob): Do not stream 0 flags per a basic
23135 2017-04-28 Martin Liska <mliska@suse.cz>
23137 * gcov-dump.c (tag_*): Add new argument to declarations.
23138 (dump_gcov_file): Likewise.
23139 (tag_blocks): Add and use new argument depth.
23140 (tag_arcs): Likewise.
23141 (tag_lines): Likewise.
23142 (tag_counters): Likewise.
23143 (tag_summary): Likewise.
23144 (dump_working_sets): Use depth to do a proper indentation.
23146 2017-04-28 Jakub Jelinek <jakub@redhat.com>
23149 * cgraph.h (symtab_node::debug_symtab): No longer inline.
23150 * symtab.c (symtab_node::debug_symtab): Move definition here.
23152 2017-04-28 Richard Biener <rguenther@suse.de>
23154 * lto-streamer.h (LTO_major_version): Bump to 7.
23156 2017-04-28 Richard Biener <rguenther@suse.de>
23158 * tree-vrp.c (assert_info): New struct.
23159 (add_assert_info): New helper.
23160 (register_edge_assert_for_2): Refactor to add asserts to a vector
23162 (register_edge_assert_for_1): Likewise.
23163 (register_edge_assert_for): Likewise.
23164 (finish_register_edge_assert_for): New helper actually registering
23165 asserts where live on edge.
23166 (find_conditional_asserts): Adjust.
23167 (find_switch_asserts): Likewise.
23168 (evrp_dom_walker::try_find_new_range): Generalize.
23169 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
23171 2017-04-27 Marek Polacek <polacek@redhat.com>
23174 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
23175 arg10 and arg11 to itype.
23177 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
23179 * doc/extend.texi (Object Size Checking): Improve grammar.
23181 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
23184 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
23185 that the logic for permitting reciprocal estimates matches that
23188 2017-04-27 Jakub Jelinek <jakub@redhat.com>
23191 * tree.c (type_cache_hasher::equal): Only compare
23192 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
23193 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
23194 non-aggregate element types.
23195 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
23196 about the flag on ARRAY_TYPEs in the comment, formatting fix.
23198 2017-04-27 Richard Biener <rguenther@suse.de>
23200 PR middle-end/80533
23201 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
23202 stripping ARRAY_REFs from MEM_EXPR make sure we're not
23203 keeping a reference to a trailing array.
23205 2017-04-27 Richard Biener <rguenther@suse.de>
23207 PR middle-end/80539
23208 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
23209 being in loop-closed SSA form conservatively.
23210 (chrec_fold_multiply_poly_poly): Likewise.
23212 2017-04-27 Tamar Christina <tamar.christina@arm.com>
23214 PR middle-end/79665
23215 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
23216 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
23218 2017-04-27 Jakub Jelinek <jakub@redhat.com>
23221 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
23222 (aarch64_function_arg_alignment): Return unsigned int again, but still
23223 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
23224 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
23225 Don't emit -Wpsabi note.
23226 (aarch64_function_arg_boundary): Likewise.
23227 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
23230 2017-04-26 Nathan Sidwell <nathan@acm.org>
23232 * tree.h (crc32_unsigned_n): Declare.
23233 (crc32_unsigned, crc32_unsigned): Make inline.
23234 * tree.c (crc32_unsigned_bits): Replace with ...
23235 (crc32_unsigned_n): ... this.
23236 (crc32_unsigned, crc32_byte): Remove.
23237 (crc32_string): Remove unnecessary braces.
23239 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
23241 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
23242 * ipa-inline-analysis.c (MAX_TIME): Remove.
23243 (account_size_time): Use sreal for time.
23244 (dump_inline_summary): Update.
23245 (estimate_function_body_sizes): Update.
23246 (estimate_edge_size_and_time): Update.
23247 (estimate_calls_size_and_time): Update.
23248 (estimate_node_size_and_time): Update.
23249 (inline_merge_summary): Update.
23250 (inline_update_overall_summary): Update.
23251 (estimate_time_after_inlining): Update.
23252 (inline_read_section): Update.
23253 (inline_write_summary): Update.
23254 * ipa-inline.c (compute_uninlined_call_time): Update.
23255 (compute_inlined_call_time): Update.
23256 (recursive_inlining): Update.
23257 (inline_small_functions): Update.
23258 (dump_overall_stats): Update.
23259 * ipa-inline.h: Include sreal.h.
23260 (size_time_entry): Turn time to sreal.
23261 (inline_summary): Turn self_time nad time to sreal.
23263 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
23265 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
23267 (sreal::stream_out, sreal::stream_in): New.
23268 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
23270 2017-04-25 Jakub Jelinek <jakub@redhat.com>
23272 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
23275 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
23278 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
23280 (dimode_scalar_chain::compute_convert_gain): Ditto.
23281 (dimode_scalar_chain::make_vector_copies): Ditto.
23282 (dimode_scalar_chain::convert_reg): Ditto.
23283 (dimode_scalar_chain::convert_insn): Ditto.
23284 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
23285 (VI248_AVX512BW_1): New mode iterator.
23286 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
23287 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
23290 2017-04-25 Martin Sebor <msebor@redhat.com>
23292 PR tree-optimization/80497
23293 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
23294 constants are representable in HOST_WIDE_INT.
23295 (parse_directive): Ditto.
23297 2017-04-25 Martin Sebor <msebor@redhat.com>
23300 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
23301 (new_zero_array): Adjust signature.
23302 (dom_info::dom_init): Used unsigned rather that size_t.
23303 (dom_info::dom_info): Same.
23305 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23306 Jakub Jelinek <jakub@redhat.com>
23309 * config/arm/arm.c: Include gimple.h.
23310 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
23311 returns negative, increment ncrn only if it returned positive.
23312 (arm_needs_doubleword_align): Return int instead of bool,
23313 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
23314 members, but if there is any such non-FIELD_DECL
23315 > PARM_BOUNDARY aligned decl, return -1 instead of false.
23316 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
23317 returns negative, increment nregs only if it returned positive.
23318 (arm_setup_incoming_varargs): Likewise.
23319 (arm_function_arg_boundary): Emit -Wpsabi note if
23320 arm_needs_doubleword_align returns negative, return
23321 DOUBLEWORD_ALIGNMENT only if it returned positive.
23323 2017-04-25 Marek Polacek <polacek@redhat.com>
23326 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
23327 first argument to type.
23329 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
23332 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
23333 type checks to test for compatibility instead of equality.
23335 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23336 Jakub Jelinek <jakub@redhat.com>
23339 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
23341 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
23342 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
23343 the alignment computation, but return their maximum in warn_alignment.
23344 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
23345 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
23347 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
23348 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
23351 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23353 * config/arc/simdext.md (dmpyh): Fix typo.
23355 2017-04-25 Richard Biener <rguenther@suse.de>
23357 PR tree-optimization/80492
23358 * alias.c (compare_base_decls): Handle registers with asm
23359 specification conservatively.
23360 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
23361 compare_base_decls returning dont-know properly.
23363 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23365 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
23366 (legitimate_offset_address_p): New function.
23367 (arc_legitimate_address_p): Use above function.
23369 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23371 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
23373 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23375 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
23376 ACCH registers whenever they are available.
23378 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23380 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
23381 double regs fix when not used.
23383 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23385 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
23387 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
23388 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
23390 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23392 * config/arc/arc.c (arc_output_addsi): Check for h-register class
23393 when emitting short ADD instructions.
23395 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
23397 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
23399 (cmpsi_cc_c_insn): Likewise.
23400 (cbranchsi4_scratch): Compute proper instruction length using
23401 compact_hreg_operand.
23402 * config/arc/predicates.md (compact_hreg_operand): New predicate.
23404 2017-04-25 Richard Biener <rguenther@suse.de>
23406 PR middle-end/80509
23407 * passes.c (pass_manager::pass_manager): Initialize
23408 m_name_to_pass_map.
23410 2017-04-25 Richard Biener <rguenther@suse.de>
23412 PR tree-optimization/79201
23413 * tree-ssa-sink.c (statement_sink_location): Handle calls.
23415 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23418 * config/s390/vector.md: Split MEM->GPR vector moves for
23419 non-s_operand addresses.
23421 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23424 * config/s390/predicates.md (reload_const_wide_int_operand): New
23426 * config/s390/s390.md ("movti"): Remove d/P alternative.
23427 ("movti_bigconst"): New pattern definition.
23429 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
23432 * s390-protos.h (s390_expand_cs_hqi): Removed.
23433 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
23434 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
23435 modes as well as CCZ1mode and CCZmode.
23436 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
23437 signature of s390_emit_compare_and_swap.
23438 (s390_expand_cs_hqi): Likewise, make static.
23439 (s390_expand_cs_tdsi): Generate an explicit compare before trying
23440 compare-and-swap, in some cases.
23441 (s390_expand_cs): Wrapper function.
23442 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
23444 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
23445 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
23446 patterns for small and large integers. Forbid symref memory operands.
23447 Move expander to s390.c. Require cc register.
23448 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
23449 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
23450 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
23451 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
23452 symref memory operands. Remove CC mode and call s390_match_ccmode
23454 ("atomic_exchange<mode>"): Allow and implement all integer modes.
23456 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
23458 * config/s390/s390.md (define_peephole2): New peephole to help
23459 combining the load-and-test pattern with volatile memory.
23461 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
23463 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
23464 with CCZmode for TARGET_Z196.
23466 2017-04-25 Jakub Jelinek <jakub@redhat.com>
23468 PR rtl-optimization/80501
23469 * combine.c (make_compound_operation_int): Set subreg_code to SET
23470 even for AND with mask of the sign bit of mode.
23472 PR rtl-optimization/80500
23473 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
23474 sum's initial value.
23476 2017-04-25 Julian Brown <julian@codesourcery.com>
23477 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23479 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
23481 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
23483 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
23485 2017-04-25 Julian Brown <julian@codesourcery.com>
23486 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23488 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
23489 (thunderx2t99_sha): New Reservation.
23491 2017-04-25 Julian Brown <julian@codesourcery.com>
23492 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23494 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
23495 type for 1-element load.
23497 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
23499 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
23501 2017-04-24 Martin Jambor <mjambor@suse.cz>
23503 PR tree-optimization/80293
23504 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
23505 char arrays not totally scalarizable if it is false.
23506 (analyze_all_variable_accesses): Pass correct value in the new
23507 parameter. Add a statistics counter.
23509 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
23511 PR middle-end/79931
23512 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
23514 2017-04-24 Richard Biener <rguenther@suse.de>
23516 PR tree-optimization/80494
23517 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
23518 out for complex types.
23520 2017-04-24 Richard Biener <rguenther@suse.de>
23522 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
23523 * tree-ssa-sccvn.c (print_scc): Print SCC size.
23524 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
23525 (DFS): Adjust and never fail.
23526 (sccvn_dom_walker::fail): Remove.
23527 (sccvn_dom_walker::before_dom_children): Adjust.
23528 (run_scc_vn): Likewise and never fail.
23529 * tree-ssa-pre.c (pass_pre::execute): Adjust.
23530 (pass_fre::execute): Likewise.
23532 2017-04-24 Richard Biener <rguenther@suse.de>
23534 PR tree-optimization/79725
23535 * tree-ssa-sink.c (statement_sink_location): Return whether
23536 failure reason was zero uses. Move that check later.
23537 (sink_code_in_bb): Deal with zero uses by removing the stmt
23540 2017-04-24 Richard Biener <rguenther@suse.de>
23543 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
23544 pointer-based references.
23546 2017-04-24 Richard Biener <rguenther@suse.de>
23549 * pass_manager.h (pass_manager::operator new): Remove.
23550 (pass_manager::operator delete): Likewise.
23551 * passes.c (pass_manager::operator new): Remove.
23552 (pass_manager::operator delete): Likewise.
23553 (pass_manager::pass_manager): Zero individual pass members.
23555 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
23558 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
23559 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
23560 Check "XEXP (src, 1)" operand here.
23561 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
23562 Check "XEXP (src, 1)" operand here.
23563 (dimode_scalar_chain::make_vector_copies): Detect count register
23564 of a shift instruction. Zero extend count register from QImode
23565 to DImode to satisfy vector shift pattern count operand predicate.
23566 Substitute vector shift count operand with a DImode copy.
23567 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
23570 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
23572 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
23573 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
23574 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
23575 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
23576 (UNSPEC_NOREX_MEM): Remove definition.
23578 2017-04-21 Richard Biener <rguenther@suse.de>
23580 PR tree-optimization/79547
23581 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
23582 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
23583 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
23584 without any constraints.
23586 2017-04-21 Richard Biener <rguenther@suse.de>
23588 PR tree-optimization/78847
23589 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
23591 2017-04-21 Richard Biener <rguenther@suse.de>
23593 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
23594 (build_distinct_type_copy): Likewise.
23595 (build_variant_type_copy): Likewise.
23596 * tree.c (build_qualified_type): Pass down mem-stat info.
23597 (build_distinct_type_copy): Likewise.
23598 (build_variant_type_copy): Likewise.
23600 2017-04-21 Richard Biener <rguenther@suse.de>
23602 PR tree-optimization/80237
23603 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
23605 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
23606 for a simplified result.
23608 2016-04-21 Richard Biener <rguenther@suse.de>
23610 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
23611 sth as strict as a simple_iv but a chrec without symbols and an
23612 operand defined in the loop we are peeling (and not some subloop).
23613 (propagate_constants_for_unrolling): Propagate all constants.
23615 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
23618 * config/i386/i386.c (print_reg): Remove assert for disalowed
23619 regno values, call output_operand_lossage instead.
23621 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
23624 * config/i386/constraints.md (Yc): New register constraint.
23625 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
23626 Use Yc constraint for alternative 2 of operand 0. Remove
23627 preferred_for_speed attribute.
23629 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
23631 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
23632 lastprivate clauses in SIMT case.
23634 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
23636 * doc/invoke.texi (-Wextra-semi): Document new warning option.
23638 2017-04-20 Richard Biener <rguenther@suse.de>
23640 PR tree-optimization/57796
23641 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
23642 as N scalar stores.
23643 (vect_model_load_cost): Cost gathers as N scalar loads.
23645 2017-04-20 Richard Biener <rguenther@suse.de>
23647 * ggc-page.c (ggc_allocated_p): Rename to ...
23648 (safe_lookup_page_table_entry): ... this and return the lookup
23650 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
23652 2017-04-20 Richard Biener <rguenther@suse.de>
23654 PR tree-optimization/80453
23655 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
23656 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
23657 from the conditions.
23658 (vn_phi_eq): Pass them down.
23659 (vn_phi_lookup): Record them.
23660 (vn_phi_insert): Likewise.
23662 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
23664 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
23665 uninitialized variable warning to avoid buffer overrun.
23667 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
23670 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
23671 is suppressed for '{ 0 }' in C.
23673 2017-04-20 Jakub Jelinek <jakub@redhat.com>
23675 * BASE-VER: Set to 8.0.0.
23677 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
23679 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
23680 priority .init_array and .fini_array section with SECTION_NOTYPE
23683 2017-04-20 Jakub Jelinek <jakub@redhat.com>
23685 PR middle-end/80423
23686 * tree.h (build_array_type): Add typeless_storage default argument.
23687 * tree.c (type_cache_hasher::equal): Also compare
23688 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
23689 (build_array_type): Add typeless_storage argument, set
23690 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
23692 (build_nonshared_array_type): Adjust build_array_type_1 caller.
23693 (build_array_type): Likewise. Add typeless_storage argument.
23695 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
23696 Jakub Jelinek <jakub@redhat.com>
23698 PR tree-optimization/80426
23699 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
23700 operation on symbolic operands, also compute the overflow for the
23701 invariant part when the operation degenerates into a negation.
23703 2017-04-19 Jakub Jelinek <jakub@redhat.com>
23706 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
23707 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
23710 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
23712 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
23715 * config/avr/avr.c (tree.h): Include it.
23716 (cgraph.h): Include it.
23717 (avr_encode_section_info): Don't warn for uninitialized progmem
23718 variable if it's just an alias.
23720 2017-04-19 Richard Biener <rguenther@suse.de>
23723 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
23724 when needed by AutoPGO.
23726 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
23729 * doc/lto.texi: Remove an extra 'that'.
23731 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
23733 PR rtl-optimization/80429
23734 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
23735 are only used in debug insns.
23737 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
23738 Vladimir Makarov <vmakarov@redhat.com>
23740 * config/sparc/predicates.md (input_operand): Add comment. Return
23741 true for any memory operand when LRA is in progress.
23742 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
23744 2017-04-18 Jeff Law <law@redhat.com>
23747 * mips.md ({return,simple_return}_internal): Do not overwrite
23750 2017-04-18 Jakub Jelinek <jakub@redhat.com>
23752 PR tree-optimization/80443
23753 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
23754 instead of adding 1, subtract -1 and similarly instead of subtracting
23757 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
23759 PR rtl-optimization/80357
23760 * haifa-sched.c (tmp_bitmap): New variable.
23761 (model_recompute): Handle duplicate use records.
23762 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
23763 (free_global_sched_pressure_data): Free it.
23765 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
23768 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
23769 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
23770 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
23771 instead of SYSTEM_HEADER_DIR.
23773 2017-04-18 Jeff Law <law@redhat.com>
23775 PR middle-end/80422
23776 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
23777 predecessors after walking up the insn chain.
23779 2017-04-18 Jakub Jelinek <jakub@redhat.com>
23782 * dwarf2out.c (modified_type_die): Try harder not to emit internal
23783 sizetype type into debug info.
23785 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
23788 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
23789 unneeded test for TARGET_UPPER_REGS_SF.
23790 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
23792 2017-04-18 Jakub Jelinek <jakub@redhat.com>
23795 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
23796 instead of gsi_after_labels.
23798 2017-04-18 Jeff Law <law@redhat.com>
23800 * regcprop.c (maybe_mode_change): Avoid creating copies of the
23804 2017-04-13 Jeff Law <law@redhat.com>
23805 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
23806 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
23808 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
23811 * config/avr/avr.c (intl.h): Include it.
23812 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
23814 2017-04-18 Martin Liska <mliska@suse.cz>
23816 PR gcov-profile/78783
23817 * gcov-tool.c (gcov_output_files): Validate that destination
23818 file is either removed by the tool or by a user.
23820 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
23821 Guy Benyei <guybe@mellanox.com>
23823 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
23824 block, and do not negate it, the stored id is already negative.
23826 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
23828 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
23830 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
23833 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
23834 masks of options that should be turned off if the VSX vector
23835 options are turned off.
23836 (OTHER_P8_VECTOR_MASKS): Likewise.
23837 (OTHER_VSX_VECTOR_MASKS): Likewise.
23838 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
23839 rs6000_disable_incompatible_switches to validate no type switches
23841 (rs6000_incompatible_switch): New function to disallow turning on
23842 other vector options if -mno-vsx, -mno-power8-vector, or
23843 -mno-power9-vector are specified.
23845 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
23847 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
23849 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
23851 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
23852 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
23853 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
23854 (ARG_POINTER_CFA_OFFSET): Likewise.
23856 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
23858 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
23859 conditions to take advantage of various optimizations.
23861 2017-04-13 Jeff Law <law@redhat.com>
23863 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
23864 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
23865 (zero_extendsidi2_dext): Likewise.
23867 2017-04-13 Jakub Jelinek <jakub@redhat.com>
23870 * fold-const.c (fold_ternary_loc): Revert
23871 use op0 instead of fold_convert_loc (loc, type, arg0) part of
23874 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
23876 PR rtl-optimization/80343
23877 * lra-remat.c (update_scratch_ops): Assign original hard reg to
23878 new scratch pseudo.
23880 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
23883 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
23884 to ubsan_encode_value.
23886 2017-04-13 Jeff Law <law@redhat.com>
23888 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
23889 appearing in DEBUG_INSNs.
23891 2017-04-13 Martin Liska <mliska@suse.cz>
23893 PR gcov-profile/80413
23894 * gcov-io.c (gcov_write_string): Copy to buffer just when
23895 allocated size is greater than zero.
23897 2017-04-13 Jakub Jelinek <jakub@redhat.com>
23900 * dwarf2out.c (decls_for_scope): Ignore declarations of
23901 current_function_decl in BLOCK_NONLOCALIZED_VARS.
23903 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
23906 * ipa-visibility.c (non_local_p): Fix typos.
23907 (localize_node): When localizing symbol in same comdat group,
23908 dissolve the group only when we know external symbols are going
23910 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
23912 2017-04-12 Jakub Jelinek <jakub@redhat.com>
23914 PR tree-optimization/79390
23915 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
23916 order does not result in usable sequence, retry with reversed operand
23922 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
23923 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
23924 op0 instead of fold_convert_loc (loc, type, arg0).
23926 2017-04-12 Jeff Law <law@redhat.com>
23928 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
23929 has a delay slot in the generated code.
23931 * config/cris/cris.md (cris_preferred_reload_class): Return
23932 GENNONACR_REGS rather than GENERAL_REGS.
23934 2017-04-12 Jakub Jelinek <jakub@redhat.com>
23937 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
23938 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
23939 signedness of the result type.
23941 2017-04-12 Richard Biener <rguenther@suse.de>
23942 Jeff Law <law@redhat.com>
23944 PR tree-optimization/80359
23945 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
23946 trim stores to TARGET_MEM_REFs.
23948 2017-04-12 Richard Biener <rguenther@suse.de>
23950 PR tree-optimization/79390
23951 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
23952 threading case even more.
23954 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
23957 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
23958 for quad_address_p for TImode, instead of just not indexed_address.
23960 2017-04-12 Richard Biener <rguenther@suse.de>
23961 Bernd Edlinger <bernd.edlinger@hotmail.de>
23963 PR middle-end/79671
23964 * alias.c (component_uses_parent_alias_set_from): Handle
23965 TYPE_TYPELESS_STORAGE.
23966 (get_alias_set): Likewise.
23967 * tree-core.h (tree_type_common): Add typeless_storage flag.
23968 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
23969 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
23970 for types containing members with TYPE_TYPELESS_STORAGE.
23971 (place_field): Likewise.
23972 (layout_type): Likewise for ARRAY_TYPE.
23973 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
23974 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
23975 TYPE_TYPELESS_STORAGE.
23976 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
23978 2017-04-12 Jakub Jelinek <jakub@redhat.com>
23981 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
23982 first argument to type.
23984 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23988 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
23989 CONST0_RTX (mode) rather than const0_rtx where appropriate.
23990 (rs6000_expand_binop_builtin): Likewise.
23991 (rs6000_expand_ternop_builtin): Likewise; also add missing
23992 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
23993 vshasigma built-ins.
23994 * doc/extend.texi: Document that vec_xxpermdi's third argument
23995 must be a constant.
23997 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
23999 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
24000 Use shift_const cost parameter when calculating gain of STV shifts.
24002 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
24004 PR rtl-optimization/70478
24005 * lra-constraints.c (process_alt_operands): Check memory for
24006 disfavoring memory insn operand.
24008 2017-04-11 Jakub Jelinek <jakub@redhat.com>
24010 PR middle-end/80100
24011 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
24012 left shift in unsigned HOST_WIDE_INT type.
24014 PR rtl-optimization/80385
24015 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
24016 (not (neg X)) into (plus X -1) for complex or non-integral modes.
24019 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
24020 if they have any depend clauses.
24022 2017-04-11 Martin Liska <mliska@suse.cz>
24025 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
24026 * ipa-split.c (split_function): Create a local comdat symbol
24027 if caller is in a comdat group.
24029 2017-04-11 Martin Liska <mliska@suse.cz>
24032 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
24035 2017-04-11 Martin Sebor <msebor@redhat.com>
24037 PR middle-end/80364
24038 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
24039 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
24041 (directive::set_width, directive::set_precision, format_character):
24043 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
24046 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
24049 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
24050 conflict, set target->arch_name instead of target->cpu_name.
24052 2017-04-11 Richard Biener <rguenther@suse.de>
24054 PR tree-optimization/80374
24055 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
24056 build_zero_cst, remove fold_convertible_p check again.
24058 2017-04-11 Martin Liska <mliska@suse.cz>
24061 * ubsan.c (instrument_object_size): Do not instrument register
24064 2017-04-11 Jakub Jelinek <jakub@redhat.com>
24067 * config/i386/i386-builtin-types.def
24068 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
24069 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
24070 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
24071 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
24072 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
24073 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
24074 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
24075 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
24076 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
24077 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
24078 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
24079 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
24080 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
24081 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
24082 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
24083 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
24084 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
24085 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
24086 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
24087 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
24088 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
24089 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
24090 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
24091 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
24092 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
24093 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
24094 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
24095 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
24096 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
24097 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
24098 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
24099 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
24100 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
24101 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
24102 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
24103 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
24104 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
24105 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
24106 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
24107 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
24108 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
24109 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
24110 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
24111 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
24112 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
24113 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
24115 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
24116 flag to second_arg_count, handle 4 argument function type _COUNT
24117 aliases, handle second_arg_count on second argument rather than last.
24119 2017-04-10 Jeff Law <law@redhat.com>
24121 PR tree-optimization/80374
24122 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
24123 record anything if we can not convert integer_zero_node to the
24126 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
24129 * config/rs6000/rs6000.c (rs6000_option_override_internal):
24130 Enhance special handling given to the TARGET_P9_MINMAX option in
24131 relation to certain other options.
24133 2017-04-10 Bin Cheng <bin.cheng@arm.com>
24135 PR tree-optimization/80153
24136 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
24137 remove POINTER_PLUS_EXPR's base part directly, rather than through
24140 2017-04-10 Richard Biener <rguenther@suse.de>
24141 Bin Cheng <bin.cheng@arm.com>
24143 PR tree-optimization/80153
24144 * tree-affine.c (aff_combination_to_tree): Get base pointer from
24145 the first element of pointer type aff_tree. Build result expr in
24147 (add_elt_to_tree): Convert to type unconditionally. Remove other
24148 fold_convert calls.
24149 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
24150 (rewrite_use_nonlinear_expr): Check invariant using iv information.
24152 2017-04-10 Richard Biener <rguenther@suse.de>
24154 * tree-ssa-structalias.c (find_func_aliases): Properly handle
24157 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
24159 PR rtl-optimization/70478
24160 * lra-constraints.c (curr_small_class_check): New.
24161 (update_and_check_small_class_inputs): New.
24162 (process_alt_operands): Update curr_small_class_check. Disfavor
24163 alternative insn memory operands. Check available regs for small
24166 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
24169 * config/mips/mips.opt (-mvirt): Update description.
24170 * doc/invoke.texi (-mvirt): Likewise.
24172 2017-04-10 Richard Biener <rguenther@suse.de>
24174 PR middle-end/80362
24175 * fold-const.c (fold_binary_loc): Look at unstripped ops when
24176 looking for NEGATE_EXPR in -A / -B to A / B folding.
24178 2017-04-10 Martin Liska <mliska@suse.cz>
24180 PR gcov-profile/80224
24181 * gcov.c (print_usage): Fix usage string.
24182 (get_gcov_intermediate_filename): Remove.
24183 (output_gcov_file): Use both for normal and intermediate format.
24184 (generate_results): Do not initialize special file for
24185 intermediate format.
24187 2017-04-10 Richard Biener <rguenther@suse.de>
24189 PR tree-optimization/80304
24190 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
24193 2017-04-10 Nathan Sidwell <nathan@acm.org>
24196 * config/rs6000/rs6000.c (rs6000_vector_type): New.
24197 (rs6000_init_builtins): Use it.
24199 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24201 * config/arm/arm.md (<mrc>): Add mode to SET source.
24202 (<mrrc>): Likewise.
24204 2017-04-10 Richard Biener <rguenther@suse.de>
24206 PR middle-end/80344
24207 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
24209 2017-04-10 Jakub Jelinek <jakub@redhat.com>
24212 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
24213 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
24214 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
24215 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
24216 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
24217 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
24218 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
24219 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
24220 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
24221 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
24222 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
24223 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
24224 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
24225 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
24226 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
24227 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
24228 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
24229 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
24230 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
24231 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
24232 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
24233 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
24234 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
24236 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
24238 PR rtl-optimization/70478
24239 * lra-constraints.c: Reverse the last patch.
24241 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
24243 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
24244 Add comment for WCHAR_T.
24246 2017-04-08 Martin Liska <mliska@suse.cz>
24249 2017-04-07 Martin Liska <mliska@suse.cz>
24252 * ipa-split.c (split_function): Add function part to a same comdat
24255 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24258 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
24260 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
24262 * rs6000/rs6000.c (vec_load_pendulum): Rename...
24263 (vec_pairing): ...to this.
24264 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
24265 (rs6000_sched_init): Adjust for name change.
24266 (struct rs6000_sched_context): Likewise.
24267 (rs6000_init_sched_context): Likewise.
24268 (rs6000_set_sched_context): Likewise.
24270 2017-04-07 Jakub Jelinek <jakub@redhat.com>
24276 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
24278 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
24279 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
24280 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
24282 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
24284 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
24286 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
24288 PR rtl-optimization/70703
24289 * ira-color.c (update_conflict_hard_regno_costs): Use
24290 int64_t instead of HOST_WIDE_INT.
24292 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
24294 PR rtl-optimization/70478
24295 * lra-constraints.c (process_alt_operands): Disfavor alternative
24296 insn memory operands.
24298 2017-04-07 Jeff Law <law@redhat.com>
24300 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
24301 CALL and NOTE_INSN_CALL_ARG_LOCATION.
24303 2017-04-07 Martin Liska <mliska@suse.cz>
24306 * config/aarch64/aarch64.c (aarch64_process_target_attr):
24307 Show error message instead of an ICE.
24309 2017-04-07 Martin Liska <mliska@suse.cz>
24312 * ipa-split.c (split_function): Add function part to a same comdat
24315 2017-04-07 Richard Biener <rguenther@suse.de>
24317 PR middle-end/80341
24318 * tree.c (get_unwidened): Also handle ! for_type case for
24320 * convert.c (do_narrow): Split out from ...
24321 (convert_to_integer_1): ... here. Do not pass final truncation
24322 type to get_unwidened for TRUNC_DIV_EXPR.
24324 2017-04-07 Richard Biener <rguenther@suse.de>
24326 * tree-affine.c (wide_int_ext_for_comb): Take type rather
24328 (aff_combination_const): Adjust.
24329 (aff_combination_scale): Likewise.
24330 (aff_combination_add_elt): Likewise.
24331 (aff_combination_add_cst): Likewise.
24332 (aff_combination_convert): Likewise.
24333 (add_elt_to_tree): Likewise. Remove unused argument.
24334 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
24336 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
24338 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
24340 * config/arm/arm.c (arm_default_short_enums): Use
24341 ARM_DEFAULT_SHORT_ENUMS.
24342 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
24344 2017-04-06 Jakub Jelinek <jakub@redhat.com>
24347 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
24348 members with redundant out-of-class redeclaration.
24350 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
24353 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
24354 * config/i386/i386.md (*zero_extendsidi2):
24355 Add (?*x,*x) and (?*v,*v) alternatives.
24357 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
24360 * config/i386/i386.c (ix86_expand_builtin)
24361 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
24362 mode from insn data. Convert operands to insn operand mode.
24363 Copy operands that don't satisfy insn predicate to a register.
24365 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
24367 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
24370 2017-04-06 Richard Biener <rguenther@suse.de>
24372 PR tree-optimization/80334
24373 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
24374 preserve alignment of accesses.
24376 2017-04-06 Richard Biener <rguenther@suse.de>
24378 PR tree-optimization/80262
24379 * tree-sra.c (build_ref_for_offset): Preserve address-space
24381 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
24382 Drop useless address-space information on MEM_REF offsets.
24384 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
24386 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
24388 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
24390 PR rtl-optimization/70703
24391 * ira-color.c (update_conflict_hard_regno_costs): Use
24392 HOST_WIDE_INT instead of long.
24394 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
24397 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
24398 not defined for x86_64 target. Add -mmmx target option when __SSE2__
24400 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
24401 for x86_64 target. Handle -m3dnowa option.
24403 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
24405 PR rtl-optimization/70703
24406 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
24407 (update_conflict_hard_regno_costs): Use long instead of unsigned
24408 arithmetic for cost calculation.
24410 2017-04-05 Jakub Jelinek <jakub@redhat.com>
24411 Bernd Edlinger <bernd.edlinger@hotmail.de>
24414 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
24417 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
24420 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
24421 ptr_mode with Pmode throughout.
24422 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
24423 into probe_stack_range and use DImode.
24425 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
24428 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
24429 call_eh_return is true.
24431 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24433 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
24434 Initialize last_match_fntype_index.
24436 2017-04-05 Jakub Jelinek <jakub@redhat.com>
24439 * tree-nvr.c: Include internal-fn.h.
24440 (pass_return_slot::execute): Ignore internal calls without
24443 2017-04-04 Jakub Jelinek <jakub@redhat.com>
24444 Richard Biener <rguenther@suse.de>
24447 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
24448 captures used multiple times, except for the last use.
24449 * generic-match-head.c: Include gimplify.h.
24451 2017-04-04 Jakub Jelinek <jakub@redhat.com>
24453 PR tree-optimization/79390
24454 * target.h (struct noce_if_info): Declare.
24455 * targhooks.h (default_noce_conversion_profitable_p): Declare.
24456 * target.def (noce_conversion_profitable_p): New target hook.
24457 * ifcvt.h (struct noce_if_info): New type, moved from ...
24458 * ifcvt.c (struct noce_if_info): ... here.
24459 (noce_conversion_profitable_p): Renamed to ...
24460 (default_noce_conversion_profitable_p): ... this. No longer
24462 (noce_try_store_flag_constants, noce_try_addcc,
24463 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
24464 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
24465 instead of noce_conversion_profitable_p.
24466 * config/i386/i386.c: Include ifcvt.h.
24467 (ix86_option_override_internal): Don't override
24468 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
24469 (ix86_noce_conversion_profitable_p): New function.
24470 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
24471 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
24472 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
24473 * doc/tm.texi: Regenerated.
24475 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24477 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
24480 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24483 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
24484 instructions for small multiply cores.
24486 2017-04-04 Jeff Law <law@redhat.com>
24488 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
24490 (mips_expand_vec_perm_const): Initialize elements in orig_perm
24491 that are not set by the loop over the elements.
24493 2017-04-04 Jakub Jelinek <jakub@redhat.com>
24496 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
24497 int mode, convert_modes it to mode as unsigned, otherwise use
24498 lowpart_subreg to mode rather than SImode.
24499 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
24500 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
24501 Use DImode instead of SImode for the shift count operand.
24502 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
24505 2017-04-04 Richard Biener <rguenther@suse.de>
24507 PR middle-end/80281
24508 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
24509 arithmetic done for the negate or the plus. Simplify.
24510 (A - (-B) -> A + B): Likewise.
24511 * fold-const.c (split_tree): Make sure to not negate pointers.
24513 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
24515 PR rtl-optimization/60818
24516 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
24517 a compare of comparisons with the thing compared if this results
24518 in a different machine mode.
24520 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
24522 * alias.c (base_alias_check): Fix typo in comment.
24523 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
24524 * cgraphunit.c (symbol_table::compile): Likewise.
24525 * collect2.c (maybe_run_lto_and_relink): Likewise.
24526 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
24527 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
24528 * config/avr/avr.c (avr_map_op_t): Likewise.
24529 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
24530 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
24531 * config/epiphany/epiphany.md (movcc): Likewise.
24532 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
24533 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
24535 * config/mips/mips.c (mips_save_restore_reg): Likewise.
24536 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
24537 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
24538 * config/sh/sh.c (sh_rtx_costs): Likewise.
24539 * fold-const.c (fold_truth_andor): Likewise.
24540 * genautomata.c (collapse_flag): Likewise.
24541 * gengtype.h (struct type::u::s): Likewise.
24542 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
24543 * input.c (FORMAT_AMOUNT): Likewise.
24544 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
24545 (known_aggs_to_agg_replacement_list): Likewise.
24546 * ipa-inline-analysis.c: Likewise.
24547 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
24548 * ipa-polymorphic-call.c
24549 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
24550 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
24551 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
24553 * modulo-sched.c (apply_reg_moves): Likewise.
24554 * omp-expand.c (build_omp_regions_1): Likewise.
24555 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
24556 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
24557 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
24558 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
24559 * value-prof.c: Likewise.
24560 * var-tracking.c (val_reset): Likewise.
24562 2017-04-03 Richard Biener <rguenther@suse.de>
24564 PR tree-optimization/80275
24565 * fold-const.c (split_address_to_core_and_offset): Handle
24568 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
24570 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
24571 descriptors is at least equal to that of functions.
24573 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
24575 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
24577 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
24580 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
24581 (mov<IMOD4:mode>): New expander.
24582 (*mov<IMOD4:mode>_internal): New insn and split pattern.
24584 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
24586 PR rtl-optimization/79405
24587 * fwprop.c (propagations_left): New variable.
24588 (forward_propagate_into): Decrement it.
24589 (fwprop_init): Initialize it.
24590 (fw_prop): If the variable has reached zero, stop propagating.
24591 (fwprop_addr): Ditto.
24593 2017-03-31 Jakub Jelinek <jakub@redhat.com>
24596 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
24597 a FUNCTION_DECL, pass it as decl instead of origin to
24600 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
24602 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
24605 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
24608 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
24609 TARGET_VSX_SMALL_INTEGER.
24611 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24613 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
24614 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
24616 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
24618 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
24619 extraction from odd-numbered MSA register.
24621 2017-03-31 Jakub Jelinek <jakub@redhat.com>
24623 PR middle-end/80173
24624 * expmed.c (store_bit_field_1): Don't attempt to create
24625 a word subreg out of hard registers wider than word if they
24626 have HARD_REGNO_NREGS of 1 for their mode.
24628 PR middle-end/80163
24629 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
24630 conversions to integer types wider than word and pointer.
24633 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
24634 (rtx_equal_for_cselib_p): Pass 0 to it.
24635 * cselib.c (cselib_hasher::equal): Likewise.
24636 (rtx_equal_for_cselib_1): Add depth argument. If depth
24637 is 128, don't look up VALUE locs and punt. Increment
24638 depth in recursive calls when walking VALUE locs.
24640 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
24642 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
24643 (make_gcov_file_name): Use the canonical path name for generating
24645 (read_line): Fix handling of files with ascii null bytes.
24647 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
24649 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
24650 to initialise a vector register instead
24651 of using a const_int.
24653 2017-03-30 Jakub Jelinek <jakub@redhat.com>
24655 PR translation/80189
24656 * gimplify.c (omp_default_clause): Use %qs instead of %s in
24657 diagnostic messages.
24659 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
24662 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
24663 (dfp_diex_<mode>): Update mode of operand 1.
24664 * doc/extend.texi (dxex, dxexq): Document change to return type.
24665 (diex, diexq): Document change to argument type.
24667 2017-03-30 Martin Jambor <mjambor@suse.cz>
24670 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
24671 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
24672 it reflects the signature changes performed at the callee side.
24673 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
24674 to cgraph_build_function_type_skip_args.
24675 (build_function_decl_skip_args): Adjust call to the above function.
24677 2017-03-30 Jakub Jelinek <jakub@redhat.com>
24680 * config/i386/sse.md
24681 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
24682 register as dest whenever it is a MEM not rtx_equal_p to the
24683 corresponding dup operand, and when forcing into reg move the
24684 reg into the memory afterwards.
24685 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
24686 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
24687 for the force_reg mode.
24688 (avx512vl_vextractf128<mode>): Use register as dest either
24689 always when a MEM, or when it is a MEM not rtx_equal_p to the
24690 corresponding dup operand, or even not when it is a CONST_VECTOR
24691 depending on the mode and lo vs. hi.
24692 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
24694 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
24695 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
24696 Likewise. Require that operands[2] is even.
24697 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
24698 Remove extraneous parens. Require that operands[2] is a multiple
24700 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
24701 operands[0] is a MEM if <mask_applied>, the predicates/constraints
24702 disallow memory then.
24704 2017-03-30 Richard Biener <rguenther@suse.de>
24706 PR tree-optimization/77498
24707 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
24708 to non-constants over backedges.
24710 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
24712 PR rtl-optimization/80233
24713 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
24714 as last_combined_insn. Do not test for BARRIER_P separately.
24716 2017-03-29 Andreas Schwab <schwab@suse.de>
24719 * calls.c (prepare_call_address): Convert funexp to Pmode before
24720 copying to temp reg.
24722 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24724 PR tree-optimization/80158
24725 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
24726 Handle possible future case of more than one alternate
24728 (replace_rhs_if_not_dup): Likewise.
24729 (replace_one_candidate): Likewise.
24731 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
24733 PR rtl-optimization/80193
24734 * ira.c (ira): Do not check allocation for LRA.
24736 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
24738 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
24739 (nvptx_output_simt_exit): Declare.
24740 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
24741 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
24742 (init_softstack_frame): Move initialization of crtl->is_leaf to...
24743 (nvptx_declare_function_name): ...here. Emit declaration of local
24744 memory space buffer for omp_simt_enter insn.
24745 (nvptx_output_unisimt_switch): New.
24746 (nvptx_output_softstack_switch): New.
24747 (nvptx_output_simt_enter): New.
24748 (nvptx_output_simt_exit): New.
24749 * config/nvptx/nvptx.h (struct machine_function): New fields
24750 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
24751 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
24752 (UNSPECV_SIMT_EXIT): Ditto.
24753 (omp_simt_enter_insn): New insn.
24754 (omp_simt_enter): New expansion.
24755 (omp_simt_exit): New insn.
24756 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
24758 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
24759 (expand_GOMP_SIMT_ENTER_ALLOC): New.
24760 (expand_GOMP_SIMT_EXIT): New.
24761 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
24762 (GOMP_SIMT_ENTER_ALLOC): Ditto.
24763 (GOMP_SIMT_EXIT): Ditto.
24764 * target-insns.def (omp_simt_enter): New insn.
24765 (omp_simt_exit): Ditto.
24766 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
24768 (lower_rec_simd_input_clauses): Implement SIMT privatization.
24769 (lower_rec_input_clauses): Likewise.
24770 (lower_lastprivate_clauses): Handle SIMT privatization.
24772 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
24773 (ompdevlow_adjust_simt_enter): New.
24774 (find_simtpriv_var_op): New.
24775 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
24776 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
24778 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
24779 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
24780 (copy_decl_for_dup_finish): Ditto.
24782 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
24784 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
24787 * config/i386/i386.c (ix86_option_override_internal): Always
24788 allow -mpreferred-stack-boundary=3 for 64-bit targets.
24790 2017-03-28 Bin Cheng <bin.cheng@arm.com>
24792 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
24794 2017-03-28 Bin Cheng <bin.cheng@arm.com>
24796 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
24797 mark new edge's irreducible flag accordign to it.
24798 (vect_do_peeling): Check loop preheader edge's irreducible flag
24799 and pass it to function slpeel_add_loop_guard.
24801 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
24803 PR tree-optimization/80218
24804 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
24805 Update block frequencies and counts.
24807 2017-03-28 Richard Biener <rguenther@suse.de>
24809 PR tree-optimization/78644
24810 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
24811 of a simplification result we may not use it at all.
24813 2017-03-28 Richard Biener <rguenther@suse.de>
24816 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
24817 without arguments, generate default definition of a SSA name.
24819 2017-03-28 Richard Biener <rguenther@suse.de>
24821 PR middle-end/80222
24822 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
24823 TYPE_REF_CAN_ALIAS_ALL references.
24824 * fold-const.c (fold_indirect_ref_1): Likewise.
24826 2017-03-28 Martin Liska <mliska@suse.cz>
24829 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
24830 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
24832 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
24833 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
24835 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
24836 (EXTRA_SPECS): Define.
24837 (SUBTARGET_EXTRA_SPECS): Likewise.
24838 (SUBTARGET_CPP_SPEC): Likewise.
24839 * config/arc/elf.h (EXTRA_SPECS): Renamed to
24840 SUBTARGET_EXTRA_SPECS.
24841 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
24843 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
24845 * config/arc/simdext.md (vst64_insn): Update pattern.
24846 (vld32wh_insn): Likewise.
24847 (vld32wl_insn): Likewise.
24848 (vld64_insn): Likewise.
24849 (vld32_insn): Likewise.
24851 2017-03-28 Marek Polacek <polacek@redhat.com>
24854 * fold-const.c (fold_comparison): Use protected_set_expr_location
24855 instead of SET_EXPR_LOCATION.
24857 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
24859 * tree.c (add_expr): Avoid name lookup warning.
24861 2017-03-27 Jeff Law <law@redhat.com>
24863 PR tree-optimization/80216
24864 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
24865 function name. Limit recursion depth.
24866 (record_temporary_equivalences): Corresponding changes.
24868 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
24870 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
24873 2017-03-27 Jakub Jelinek <jakub@redhat.com>
24876 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
24878 * cfgcleanup.c (reg_note_cfa_p): New array.
24879 (insns_have_identical_cfa_notes): New function.
24880 (old_insns_match_p): Don't cross-jump in between /f
24881 and non-/f instructions. If both i1 and i2 are frame related,
24882 verify all CFA notes, their order and content.
24884 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
24887 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
24888 HImode and SImode with zero extend to DImode to one insn.
24889 (bswap<mode>2_extenddi): Likewise.
24890 (bswapsi2_extenddi): Likewise.
24891 (bswaphi2_extendsi): Likewise.
24892 (bswaphi2): Combine bswap HImode and SImode into one insn.
24893 Separate memory insns from swapping register.
24894 (bswapsi2): Likewise.
24895 (bswap<mode>2): Likewise.
24896 (bswaphi2_internal): Delete, no longer used.
24897 (bswapsi2_internal): Likewise.
24898 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
24899 store, and gpr<-gpr swap insns.
24900 (bswap<mode>2_store): Likewise.
24901 (bswaphi2_reg): Register only splitter, combine with the splitter.
24902 (bswaphi2 splitter): Likewise.
24903 (bswapsi2_reg): Likewise.
24904 (bswapsi2 splitter): Likewise.
24905 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
24906 the insns into load, store, and register/register insns.
24907 (bswapdi2_ldbrx): Likewise.
24908 (bswapdi2_load): Likewise.
24909 (bswapdi2_store): Likewise.
24910 (bswapdi2_reg): Likewise.
24912 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
24914 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
24915 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
24917 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
24920 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
24922 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
24923 special handling for target option conflicts between dform
24924 options (-mpower9-dform, -mpower9-dform-vector,
24925 -mpower9-dform-scalar) and -mno-direct-move.
24927 2017-03-27 Richard Biener <rguenther@suse.de>
24929 PR tree-optimization/80181
24930 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
24932 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24934 * config/arc/predicates.md (move_double_src_operand): Replace the
24935 call to move_double_src_operand with a call to address_operand.
24937 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24939 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
24940 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
24941 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
24943 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24945 * config/arc/predicates.md (long_immediate_loadstore_operand):
24946 Consider scaled addresses cases.
24948 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
24950 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
24951 restored when in interrupt.
24952 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
24953 doesn't have delay slot.
24955 2017-03-27 Richard Biener <rguenther@suse.de>
24958 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
24959 inlined thunk clones.
24961 2017-03-27 Jakub Jelinek <jakub@redhat.com>
24964 * asan.c (instrument_derefs): Copy over last operand from
24965 original COMPONENT_REF to the new COMPONENT_REF with
24966 DECL_BIT_FIELD_REPRESENTATIVE.
24967 * ubsan.c (instrument_object_size): Likewise.
24969 2017-03-27 Richard Biener <rguenther@suse.de>
24971 PR tree-optimization/80170
24972 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
24973 sure DR/SCEV didnt fold in constants we do not see when looking
24974 at the reference base alignment.
24976 2017-03-27 Richard Biener <rguenther@suse.de>
24978 PR middle-end/80171
24979 * gimple-fold.c (fold_ctor_reference): Properly guard against
24980 NULL return value from canonicalize_constructor_val.
24982 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
24985 * config/i386/i386.c (ix86_expand_builtin)
24986 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
24987 flags reg setting and flags reg using instructions.
24988 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
24989 clobbering instructions to zero extend op2.
24991 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
24993 * doc/install.texi (Configuration) <--with-aix-soname>:
24994 Update link to AIX ld.
24996 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
24998 PR rtl-optimization/80160
24999 PR rtl-optimization/80159
25000 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
25001 reg_alternate_class into account.
25003 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
25006 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
25007 to consider in curr_insn_transform.
25009 2017-03-24 Jakub Jelinek <jakub@redhat.com>
25011 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
25012 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
25013 and emit_mode_inner.
25015 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25017 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
25018 argument to the overloaded builtin variants. Use the new flag to
25019 deprecate certain builtin variants.
25020 * config/s390/s390-builtin-types.def: Add new builtin types.
25021 * config/s390/s390-builtins.h: Support new flags field for
25022 overloaded builtins.
25023 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
25024 (s390_macro_to_expand): Enable vector float data type.
25025 (s390_cpu_cpp_builtins_internal): Indicate support of the new
25026 builtins by incrementing the __VEC__ version number.
25027 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
25029 (s390_resolve_overloaded_builtin): Emit error messages depending
25030 on the builtin flags.
25031 * config/s390/s390.c (s390_expand_builtin): Support additional
25032 flags argument. Change error message to match the messages
25033 emitted in s390-c.c.
25034 * config/s390/s390.md: New UNSPEC_* constants.
25035 (op_type): Add new instruction types.
25036 * config/s390/vecintrin.h: Add new builtins and test data class
25038 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
25039 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
25040 (VEC_INEXACT, VEC_NOINEXACT): New constants.
25041 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
25042 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
25043 ("vec_mergel<mode>"): V_HW -> VEC_HW.
25045 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
25046 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
25047 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
25048 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
25050 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
25051 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
25052 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
25053 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
25055 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
25056 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
25057 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
25058 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
25059 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
25060 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
25061 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
25063 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
25064 ("vec_scatter_element<V_HW_4:mode>_DI")
25065 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
25066 ("vec_fpint<mode>", "vflls")
25067 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
25068 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
25069 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
25070 ("*vec_cmphe<mode>_cc"): ... these.
25072 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
25073 mode constant instead of magic value.
25075 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25077 * config/s390/s390.c (s390_expand_vec_compare): Support other
25078 vector floating point modes than just V2DF.
25079 (s390_expand_vcond): Likewise.
25080 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
25081 (s390_cannot_change_mode_class): Prevent mode changes between TF
25082 and V1TF in vector registers.
25083 * config/s390/s390.md (DF, SF): New mode attributes.
25084 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
25085 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
25086 SFmode support for VRs.
25087 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
25089 (VFT, VF_HW): New mode iterators.
25090 (vw, sdx): New mode attributes.
25091 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
25092 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
25093 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
25094 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
25095 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
25096 also the new vector floating point modes. Renaming to ...
25098 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
25099 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
25100 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
25101 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
25102 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
25103 ("vec_unordered<mode>"): ... these.
25105 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
25106 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
25107 ("*vec_extendv2df"): New insn definitions.
25109 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25111 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
25112 ("mulditi3_2", "*muldi3_sign"): New patterns.
25113 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
25114 rename the pattern definition.
25116 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25118 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
25120 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
25122 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25124 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
25125 instruction if possible.
25126 * config/s390/vector.md (vec_halfnumelts): New mode
25128 ("*vec_vllezlf<mode>"): New pattern.
25130 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25132 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
25133 ("popcountv4si2", "popcountv2di2"): Rename to ...
25134 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
25135 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
25137 ("popcount<mode>2_vxe"): New pattern.
25139 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25141 * common/config/s390/s390-common.c (processor_flags_table): Add
25143 * config.gcc: Add arch12.
25144 * config/s390/driver-native.c (s390_host_detect_local_cpu):
25145 Default to arch12 for unknown CPU model numbers.
25146 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
25147 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
25148 PROCESSOR_max sanity check.
25149 * config/s390/s390-opts.h (enum processor_type): Add
25151 * config/s390/s390.c (processor_table): Add arch12.
25152 (s390_expand_builtin): Add check for B_VXE flag.
25153 (s390_issue_rate): Add PROCESSOR_ARCH12.
25154 (s390_get_sched_attrmask): Likewise.
25155 (s390_get_unit_mask): Likewise.
25156 (s390_sched_score): Enable z13 scheduling for arch12.
25157 (s390_sched_reorder): Likewise.
25158 (s390_sched_variable_issue): Likewise.
25159 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
25161 (s390_tune_attr): Use z13 scheduling also for arch12.
25162 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
25163 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
25164 (TARGET_VXE_P): New macros.
25165 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
25166 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
25167 * config/s390/s390.opt: Add arch12 as processor_type.
25169 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25171 * config/s390/s390.md
25172 ("fixuns_truncdddi2", "fixuns_trunctddi2")
25173 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
25174 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
25176 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
25177 Rename expanders to ...
25179 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
25180 ("fixuns_truncdddi2_emu"): ... these.
25182 ("fixuns_trunc<mode>si2_emu"): New expander.
25184 ("*fixuns_truncdfdi2_z13"): Rename to ...
25185 ("*fixuns_truncdfdi2_vx"): ... this.
25187 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25189 * config/s390/2964.md: Remove the single element vector compare
25190 instructions which are no longer used.
25191 * config/s390/s390.c (s390_select_ccmode): Remove handling of
25193 (s390_canonicalize_comparison): Remove handling of DFmode
25195 (s390_expand_vec_compare_scalar): Remove function.
25196 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
25197 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
25199 ("*cmp<mode>_ccs"): Add wfcdb instruction.
25201 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25203 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
25205 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
25206 will anyway by matched by mov<mode>_64dfp.
25208 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25210 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
25211 vlef/vstef. Add missing operand to vleif.
25213 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25215 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
25216 pair for all vector types with 64 bit elements.
25217 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
25218 * config/s390/vector.md (V_HW_64): ... here.
25219 (V_128_NOSINGLE): New mode iterator.
25220 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
25221 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
25222 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
25223 ("*vec_load_pairv2di"): Change to ...
25224 ("*vec_load_pair<mode>"): ... this one.
25226 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25228 * config/s390/constraints.md: Add comments.
25229 (jKK): Reject element sizes > 8 bytes.
25230 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
25232 * config/s390/s390.md: Add the s_operand checks formerly in
25233 s390_split_ok_p to various splitters where they are still
25235 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
25236 for 128 bit vectors. Plus two splitters.
25238 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25240 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
25243 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25246 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
25247 error if the boundary argument is not constant.
25249 2017-03-24 Jakub Jelinek <jakub@redhat.com>
25251 PR rtl-optimization/80112
25252 * loop-doloop.c (doloop_condition_get): Don't check condition
25253 if cmp isn't SET with IF_THEN_ELSE src.
25255 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25257 PR tree-optimization/80158
25258 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
25259 replacing a candidate statement, also replace it for the
25260 candidate's alternate interpretation.
25261 (replace_rhs_if_not_dup): Likewise.
25262 (replace_one_candidate): Likewise.
25264 2017-03-24 Richard Biener <rguenther@suse.de>
25266 PR tree-optimization/80167
25267 * graphite-isl-ast-to-gimple.c
25268 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
25270 (translate_isl_ast_to_gimple::get_rename): Likewise.
25272 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
25274 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
25275 handling of certain combinations of target options, including the
25276 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
25277 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
25279 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25282 * config/arm/arm.md (*load_multiple): Add reload_completed to
25283 matching condition.
25285 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25286 Richard Biener <rguenth@suse.de>
25288 PR tree-optimization/79908
25289 PR tree-optimization/80136
25290 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
25291 been cast away, gimplify_and_add suffices.
25293 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
25295 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
25297 2017-03-23 Richard Biener <rguenther@suse.de>
25299 PR tree-optimization/80032
25300 * gimplify.c (gimple_push_cleanup): Forced unconditional
25301 cleanups still have to go to the conditional_cleanups
25304 2017-03-22 Jakub Jelinek <jakub@redhat.com>
25306 PR tree-optimization/80072
25307 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
25309 (next_operand_entry_id): Change type to unsigned int.
25310 (sort_by_operand_rank): Make sure to return the right return value
25311 even if unsigned fields are bigger than INT_MAX.
25312 (struct oecount): Change cnt and id type to unsigned int.
25313 (oecount_hasher::equal): Formatting fix.
25314 (oecount_cmp): Make sure to return the right return value
25315 even if unsigned fields are bigger than INT_MAX.
25316 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
25319 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
25320 TREE_READONLY on result if writing it more than once.
25323 * doc/invoke.texi (-fsanitize=thread): Document that with
25324 -fnon-call-exceptions atomics are not able to throw
25328 * tsan.c: Include tree-eh.h.
25329 (instrument_builtin_call): Call maybe_clean_eh_stmt or
25330 maybe_clean_or_replace_eh_stmt where needed.
25331 (instrument_memory_accesses): Add cfg_changed argument.
25332 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
25333 if it returned true.
25334 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
25336 PR rtl-optimization/63191
25337 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
25338 wrapper function, moved the whole old content into ...
25339 (ix86_delegitimize_address_1): ... this. New inline function.
25340 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
25341 true as last argument instead of ix86_delegitimize_address.
25343 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
25345 * config/aarch64/aarch64.c (generic_branch_cost): Copy
25346 cortexa57_branch_cost.
25348 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
25350 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
25352 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25355 * doc/md.texi (Constraints): Document wA constraint.
25356 * config/rs6000/constraints.md (wA): New.
25357 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
25358 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
25359 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
25360 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
25362 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
25365 * gimplify.c (is_oacc_declared): New function.
25366 (oacc_default_clause): Use it to set default flags for acc declared
25367 variables inside parallel regions.
25368 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
25369 declared variables.
25370 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
25371 declare attribute to any decl as necessary.
25373 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
25376 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
25377 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
25378 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
25379 (arm_arch_lpae): This.
25380 * config/arm/arm.c (arm_arch7ve): Rename into ...
25381 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
25382 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
25385 2017-03-22 Martin Liska <mliska@suse.cz>
25388 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
25389 error message instead of an ICE.
25391 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25393 * doc/extend.texi (6.11 Additional Floating Types): Revise.
25395 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
25397 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
25399 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
25402 2017-03-21 Martin Sebor <msebor@redhat.com>
25404 * doc/extend.texi: Use "cannot" instead of "can't."
25405 * doc/hostconfig.texi: Same.
25406 * doc/install.texi: Same.
25407 * doc/invoke.texi: Same.
25408 * doc/loop.texi: Same.
25409 * doc/md.texi: Same.
25410 * doc/objc.texi: Same.
25411 * doc/rtl.texi: Same.
25412 * doc/tm.texi: Same.
25413 * doc/tm.texi.in: Same.
25414 * doc/trouble.texi: Same.
25416 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
25419 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
25420 (collect_checksum_attributes): Set it.
25421 (die_checksum_ordered): Use it.
25423 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25425 PR tree-optimization/79908
25426 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
25427 change: For a VA_ARG whose LHS has been cast away, use
25428 force_gimple_operand to construct the side effects.
25430 2017-03-21 David Malcolm <dmalcolm@redhat.com>
25432 PR translation/80001
25433 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
25434 more amenable to translation.
25435 (oacc_loop_auto_partitions): Likewise.
25437 2017-03-21 Marek Polacek <polacek@redhat.com>
25438 Martin Sebor <msebor@redhat.com>
25440 PR tree-optimization/80109
25441 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
25442 on INTEGRAL_TYPE_P.
25444 2017-03-21 Jakub Jelinek <jakub@redhat.com>
25445 Segher Boessenkool <segher@kernel.crashing.org>
25448 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
25449 check reg_used_between_p between insn and one of succ or succ2
25450 depending on if succ is artificial insn not inserted into insn
25453 2017-03-21 Martin Liska <mliska@suse.cz>
25455 PR gcov-profile/80081
25456 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
25457 * doc/gcc.texi: Include gcov-dump stuff.
25458 * doc/gcov-dump.texi: New file.
25460 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
25462 PR rtl-optimization/79150
25463 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
25464 conditional jump, if the jump is the last insn of the loop.
25466 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25467 Richard Biener <rguenth@suse.de>
25469 PR tree-optimization/79908
25470 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
25471 been cast away, use force_gimple_operand to construct the side
25474 2017-03-21 Martin Liska <mliska@suse.cz>
25476 PR libfortran/79956
25477 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
25480 2017-03-21 Brad Spengler <spender@grsecurity.net>
25483 * plugin.c (htab_hash_plugin): New function.
25484 (add_new_plugin): Use it and adjust.
25485 (parse_plugin_arg_opt): Adjust.
25486 (init_one_plugin): Likewise.
25488 2017-03-21 Richard Biener <rguenther@suse.de>
25490 PR tree-optimization/80032
25491 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
25492 if set force the cleanup to happen unconditionally.
25493 (gimplify_target_expr): Push inserted clobbers with force_uncond
25494 to avoid them being removed by control-dependent DCE.
25496 2017-03-21 Richard Biener <rguenther@suse.de>
25498 PR tree-optimization/80122
25499 * tree-inline.c (copy_bb): Do not expans va-arg packs or
25500 va_arg_pack_len when the inlined call stmt requires pack
25502 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
25504 2017-03-21 Jakub Jelinek <jakub@redhat.com>
25507 * tsan.c (instrument_builtin_call): If the memory model argument
25508 is not a constant, assume it is valid.
25511 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
25514 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
25516 PR rtl-optimization/79910
25517 * combine.c (can_combine_p): Do not allow combining an I0 or I1
25518 if its dest is used by an insn before I2 (other than the combined
25519 insns themselves, which are properly handled already).
25521 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
25524 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
25526 * combine.c (record_used_regs): New static function.
25527 (try_combine): Handle situations where there is an additional
25528 instruction between I2 and I3 which needs to have a LOG_LINK
25532 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
25534 * combine.c (try_combine): Delete redundant i1 test. Call
25535 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
25537 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25540 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
25541 alternatives 13/14.
25543 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25545 PR tree-optimization/80054
25546 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
25547 the optimization if a PHI or any of its arguments is not dominated
25548 by the candidate's basis. Use gphi* rather than gimple* as
25550 (replace_profitable_candidates): Clean up a gimple* variable that
25551 should be a gphi* variable.
25553 2017-03-20 Martin Sebor <msebor@redhat.com>
25556 * doc/extend.texi (attribute constructor): Document present limitation.
25558 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
25561 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
25562 __POWER9_VECTOR__ #ifdef control, change template definition to
25563 use Power9-specific built-in function.
25564 (vec_any_eq): Likewise.
25565 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
25566 to control outcomes from this test.
25567 (vector_ae_<mode>p): For VEC_F modes, likewise.
25569 2017-03-20 Ian Lance Taylor <iant@google.com>
25571 * config/i386/i386.c (ix86_function_regparm): Save an extra
25572 register for -fsplit-stack with DECL_STATIC_CHAIN.
25574 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
25577 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
25578 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
25580 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
25582 * config/riscv/riscv.c (riscv_print_operand): Use "fence
25584 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
25587 2017-03-20 Marek Polacek <polacek@redhat.com>
25590 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
25592 2017-03-20 Richard Biener <rguenther@suse.de>
25594 PR tree-optimization/80113
25595 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
25596 allocate extra SSA name for PHI def.
25597 (add_close_phis_to_outer_loops): Likewise.
25598 (add_close_phis_to_merge_points): Likewise.
25599 (copy_loop_close_phi_args): Likewise.
25600 (copy_cond_phi_nodes): Likewise.
25602 2017-03-20 Martin Liska <mliska@suse.cz>
25604 PR middle-end/79753
25605 * tree-chkp.c (chkp_build_returned_bound): Do not build
25606 returned bounds for a LHS that's not a BOUNDED_P type.
25608 2017-03-20 Martin Liska <mliska@suse.cz>
25612 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
25613 COMPLEX_CST and VECTOR_CST.
25615 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25618 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
25619 target operand. A new splitter adds the clobber statement in case
25620 the target operand is dead anyway.
25622 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
25624 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
25625 to age-old versions of binutils and glibc.
25627 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
25629 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
25631 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
25633 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
25635 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
25637 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
25638 requirement for binutils 2.13.
25640 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
25642 * combine.c (try_combine): Delete redundant i1 test. Call
25643 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
25645 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
25647 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
25648 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
25650 <riscv64-*-elf>: Re-arrange section
25651 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
25652 <riscv32-*-linux>: Likewise.
25653 <riscv64-*-elf>: Likewise
25654 <riscv64-*-linux>: Likewise.
25656 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
25659 * aarch64.opt(verbose-cost-dump): Fix typo.
25661 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
25664 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
25665 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
25667 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
25669 * reload.c (find_reloads): When reloading a nonoffsettable address,
25670 use RELOAD_OTHER for it and its address reloads.
25672 PR rtl-optimization/79910
25673 * combine.c (record_used_regs): New static function.
25674 (try_combine): Handle situations where there is an additional
25675 instruction between I2 and I3 which needs to have a LOG_LINK
25678 2017-03-17 Jeff Law <law@redhat.com>
25680 PR tree-optimization/71437
25681 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
25682 conditional in the hash table first.
25683 (vrp_dom_walker::before_dom_children): Extract condition from
25684 ASSERT_EXPR. Record condition, its inverion and any implied
25685 conditions as well.
25687 2017-03-17 Marek Polacek <polacek@redhat.com>
25688 Markus Trippelsdorf <markus@trippelsdorf.de>
25690 PR tree-optimization/80079
25691 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
25694 2017-03-17 Richard Biener <rguenther@suse.de>
25696 PR middle-end/80075
25697 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
25698 Properly verify the LHS before the RHS possibly claims to be
25700 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
25703 2017-03-17 Martin Jambor <mjambor@suse.cz>
25705 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
25706 (List of -O2 options): Likewise.
25707 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
25710 2017-03-17 Tom de Vries <tom@codesourcery.com>
25712 * gcov-dump.c (print_usage): Print bug_report_url.
25714 2017-03-17 Richard Biener <rguenther@suse.de>
25716 PR middle-end/80050
25717 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
25718 (parser::peek): Likewise.
25720 2017-03-17 Richard Biener <rguenther@suse.de>
25722 PR tree-optimization/80048
25723 * sese.c (free_sese_info): Properly release rename_map and
25724 copied_bb_map elements.
25726 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
25728 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
25729 Add linked-list forward and backlinks. Insert on
25730 construction, remove on destruction.
25731 (class pass_store_merging): Add m_stores_head field.
25732 (pass_store_merging::terminate_and_process_all_chains):
25733 Iterate over m_stores_head list.
25734 (pass_store_merging::terminate_all_aliasing_chains):
25736 (pass_store_merging::execute): Check for debug stmts first.
25737 Push new chains onto the m_stores_head stack.
25739 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
25742 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
25743 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
25744 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
25746 2017-03-16 Jeff Law <law@redhat.com>
25748 PR tree-optimization/71437
25749 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
25750 member function. Implementation moved into after_dom_children
25751 member function and into the threader's thread_outgoing_edges
25753 (dom_opt_dom_walker::after_dom_children): Simplify by moving
25754 some code into new thread_outgoing_edges.
25755 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
25756 definition. Simplify marker handling (do it here). Assume we always
25757 have the available expression and the const/copies tables.
25758 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
25760 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
25761 * tree-vrp.c (equiv_stack): No longer file scoped.
25762 (vrp_dom_walker): New class.
25763 (vrp_dom_walker::before_dom_children): New member function.
25764 (vrp_dom_walker::after_dom_children): Likewise.
25765 (identify_jump_threads): Setup domwalker. Use it rather than
25766 walking edges in a random order by hand. Simplify setup/finalization.
25767 (finalize_jump_threads): Remove.
25768 (vrp_finalize): Do not call identify_jump_threads here.
25769 (execute_vrp): Do it here instead and call thread_through_all_blocks
25772 PR tree-optimization/71437
25773 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
25775 (simplify_stmt_for_jump_threading): Add basic_block argument. All
25777 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
25778 (dom_opt_dom_walker::thread_across_edge): Remove
25779 handle_dominating_asserts argument. All callers changed.
25780 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
25781 changes. Remove calls to lhs_of_dominating_assert. Other
25782 uses of handle_dominating_asserts turn into unconditional code
25783 (simplify_control_stmt_condition_1): Likewise.
25784 (simplify_control_stmt_condition): Likewise.
25785 (thread_through_normal_block, thread_across_edge): Likewise.
25786 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
25787 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
25788 object if it is not an SSA_NAME.
25789 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
25790 before calling into the VRP specific simplifiers.
25791 (identify_jump_threads): Remove handle_dominating_asserts
25794 2017-03-16 Jakub Jelinek <jakub@redhat.com>
25797 * tree-diagnostic.c (default_tree_printer): No longer static.
25798 * tree-diagnostic.h (default_tree_printer): New prototype.
25800 2017-03-16 Tamar Christina <tamar.christina@arm.com>
25802 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
25803 Change ins into fmov.
25805 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25807 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
25808 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
25809 Use h_con constraint for operand 1.
25810 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
25811 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
25813 2017-03-15 Jeff Law <law@redhat.com>
25815 PR tree-optimization/71437
25816 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
25817 (record_temporary_equivalences): Use it.
25819 PR tree-optimization/71437
25820 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
25821 tree-ssa-scopedtables.
25822 (lookup_avail_expr, build_and_record_new_cond): Likewise.
25823 (record_conditions, record_cond, vuse_eq): Likewise.
25824 (record_edge_info): Adjust to API tweak of record_conditions.
25825 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
25826 (record_temporary_equivalences, optimize_stmt): Likewise.
25827 (eliminate_redundant_computations): Likewise.
25828 (record_equivalences_from_stmt): Likewise.
25829 * tree-ssa-scopedtables.c: Include options.h and params.h.
25830 (vuse_eq): New function, moved from tree-ssa-dom.c
25831 (build_and_record_new_cond): Likewise.
25832 (record_conditions): Likewise. Accept vector of conditions rather
25833 than edge_equivalence structure for first argument.
25834 for the first argument.
25835 (avail_exprs_stack::lookup_avail_expr): New member function, moved
25836 from tree-ssa-dom.c.
25837 (avail_exprs_stack::record_cond): Likewise.
25838 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
25839 from tree-ssa-dom.c.
25840 (avail_exprs_stack): Add new member functions lookup_avail_expr
25842 (record_conditions): Declare.
25844 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
25847 * lra-constraints.c (process_alt_operands): Increase reject for
25848 reloading an input/output operand.
25850 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
25853 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
25854 insns to convert from signed/unsigned char/short to IEEE 128-bit
25856 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
25858 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
25861 * config/i386/i386.c (ix86_vector_duplicate_value): Create
25862 subreg of inner mode for values already in registers.
25864 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
25866 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
25867 iteration reg is used after the loop.
25869 2017-03-14 Martin Sebor <msebor@redhat.com>
25871 PR tree-optimization/79800
25872 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
25873 precision in negative-positive range.
25874 (format_floating): Call non-const overload with adjusted precision.
25876 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
25879 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
25882 2017-03-14 Martin Sebor <msebor@redhat.com>
25884 PR middle-end/80020
25885 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
25886 * builtins.def (aligned_alloc): Use it.
25889 * Makefile.in (GTFILES): Add calls.c.
25890 * calls.c: Include "gt-calls.h".
25892 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
25894 PR rtl-optimization/79728
25895 * regs.h (struct target_regs): New field
25896 x_contains_allocatable_regs_of_mode.
25897 (contains_allocatable_regs_of_mode): New macro.
25898 * reginfo.c (init_reg_sets_1): Initialize it, and change
25899 contains_reg_of_mode so it includes global regs as well.
25900 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
25901 rather than contains_regs_of_mode.
25903 2017-03-14 Martin Liska <mliska@suse.cz>
25905 * doc/invoke.texi: Document options that can't be combined with
25906 -fcheck-pointer-bounds.
25908 2017-03-14 Martin Liska <mliska@suse.cz>
25910 PR middle-end/79831
25911 * doc/invoke.texi (-Wchkp): Document the option.
25913 2017-03-14 Martin Liska <mliska@suse.cz>
25915 * Makefile.in: Install gcov-dump.
25917 2017-03-14 Martin Liska <mliska@suse.cz>
25919 * multiple_target.c (expand_target_clones): Bail out for
25920 an invalid attribute.
25922 2017-03-14 Richard Biener <rguenther@suse.de>
25924 * alias.c (struct alias_set_entry): Pack properly.
25925 * cfgloop.h (struct loop): Likewise.
25926 * cse.c (struct set): Likewise.
25927 * ipa-utils.c (struct searchc_env): Likewise.
25928 * loop-invariant.c (struct invariant): Likewise.
25929 * lra-remat.c (struct cand): Likewise.
25930 * recog.c (struct change_t): Likewise.
25931 * rtl.h (struct address_info): Likewise.
25932 * symbol-summary.h (function_summary): Likewise.
25933 * tree-loop-distribution.c (struct partition): Likewise.
25934 * tree-object-size.c (struct object_size_info): Likewise.
25935 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
25936 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
25937 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
25938 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
25939 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
25940 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
25941 (struct _stmt_vec_info): Likewise.
25943 2017-03-14 Martin Liska <mliska@suse.cz>
25946 * multiple_target.c (create_dispatcher_calls): Check that
25947 a target can create a function dispatcher.
25949 2017-03-14 Martin Liska <mliska@suse.cz>
25952 * multiple_target.c (expand_target_clones): Drop local.local
25953 flag for default implementation.
25955 2017-03-14 Richard Biener <rguenther@suse.de>
25957 PR tree-optimization/80030
25958 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
25960 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
25962 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
25963 gcc_fallthrough() instead of __attribute__((fallthrough));
25965 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
25967 * doc/gcc.texi: Remove "up" link to (DIR).
25968 * doc/gccint.texi: Ditto.
25970 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
25972 * doc/install.texi (Specific) <avr>: Remove reference to
25975 2017-03-13 Jeff Law <law@redhat.com>
25977 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
25978 attribute rather than comments.
25980 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
25981 match_scratch operand is highest.
25983 2017-03-13 Martin Liska <mliska@suse.cz>
25985 PR middle-end/78339
25986 * ipa-pure-const.c (warn_function_noreturn): If the declarations
25987 is a CHKP clone, use original declaration.
25989 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25991 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
25992 (arc_conditional_register_usage): Use a different allocation order
25993 when optimizing for size.
25994 * common/config/arc/arc-common.c (arc_option_optimization_table):
25995 Section anchors default on when optimizing for size.
25997 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
25999 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
26001 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
26003 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
26004 * config/arc/arc.md (cpu_facility): Add cd variant.
26005 (*movqi_insn): Add code density variant.
26006 (*movhi_insn): Likewise.
26007 (*movqi_insn): Likewise.
26008 (*addsi3_mixed): Likewise.
26009 (subsi3_insn): Likewise.
26011 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
26013 * config/arc/arc.md (movsi_cond_exec): Update constraint.
26015 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
26017 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
26018 expressions with MINUS and UNARY ops.
26020 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26023 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
26025 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
26026 between vec_select and vector argument.
26027 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
26028 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
26029 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
26030 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
26031 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
26032 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
26034 2017-03-13 Richard Biener <rguenther@suse.de>
26037 * params.def (vect-max-peeling-for-alignment): Fix typo.
26039 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
26041 * doc/install.texi (Specific) <mips-*-*>: Remove description of
26042 issue that only occurred with binutils below 2.18.
26044 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
26046 * doc/install.texi (Specific) <cris-axis-elf>: No longer
26047 refer to binutils 2.11/2.12 minimum.
26049 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
26051 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
26052 ftp.kernel.org and simplify binutils requirement.
26054 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
26056 * doc/invoke.texi (Warning Options): Fix spelling of link-time
26058 (Optimize Options): Ditto. Also remove redundancy.
26060 2017-03-10 David Malcolm <dmalcolm@redhat.com>
26062 PR translation/79848
26063 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
26065 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
26066 to G_ to avoid double translation.
26068 2017-03-10 David Malcolm <dmalcolm@redhat.com>
26070 PR translation/79923
26071 * auto-profile.c (get_combined_location): Convert leading
26072 character of diagnostics to lower case and remove trailing period.
26073 (read_profile): Likewise for various diagnostics.
26074 * config/arm/arm.c (arm_option_override): Remove trailing period
26075 from various diagnostics.
26076 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
26077 (msp430_expand_delay_cycles): Likewise.
26079 2017-03-10 David Malcolm <dmalcolm@redhat.com>
26082 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
26083 full command-line argument, rather than just "str".
26084 (aarch64_validate_march): Likewise.
26085 (aarch64_validate_mtune): Likewise.
26087 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
26089 PR rtl-optimization/78911
26090 * lra-assigns.c (must_not_spill_p): New function.
26091 (spill_for): Use it.
26093 2017-03-10 Jakub Jelinek <jakub@redhat.com>
26095 PR tree-optimization/79981
26096 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
26097 ATOMIC_COMPARE_EXCHANGE ifn result.
26098 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
26099 IFN_ATOMIC_COMPARE_EXCHANGE.
26101 2017-03-10 David Malcolm <dmalcolm@redhat.com>
26104 * opts.c (parse_sanitizer_options): Add missing question mark to
26105 "did you mean" message.
26107 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26109 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
26111 (VMULEUH_UNS): Likewise.
26112 (VMULOUB_UNS): Likewise.
26113 (VMULOUH_UNS): Likewise.
26114 * config/rs6000/rs6000.c (builtin_function_type): Remove
26115 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
26117 2017-03-10 David Malcolm <dmalcolm@redhat.com>
26120 * read-rtl-function.c (function_reader::read_rtx_operand): Update
26121 x with result of extra_parsing_for_operand_code_0.
26122 (function_reader::extra_parsing_for_operand_code_0): Convert
26123 return type from void to rtx, returning x. When reading
26124 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
26125 larger size containing struct block_symbol.
26127 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
26129 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
26130 -mfloat128-hardware without -m64.
26132 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
26135 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
26136 entries to the case statement that marks unsigned arguments to
26137 overloaded functions.
26139 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
26141 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
26142 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
26144 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
26147 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
26148 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
26150 2017-03-10 Martin Liska <mliska@suse.cz>
26154 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
26155 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
26158 2017-03-10 Olivier Hainque <hainque@adacore.com>
26160 * tree-switch-conversion (array_value_type): Start by resetting
26161 candidate type to it's main variant.
26163 2017-03-10 Jakub Jelinek <jakub@redhat.com>
26165 PR rtl-optimization/79909
26166 * combine.c (try_combine): Use simplify_replace_rtx on individual
26167 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
26168 of the whole CALL_INSN_FUNCTION_USAGE.
26170 PR tree-optimization/79972
26171 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
26172 get_range_info on SSA_NAMEs. Formatting fixes.
26174 2017-03-10 Richard Biener <rguenther@suse.de>
26175 Jakub Jelinek <jakub@redhat.com>
26177 PR tree-optimization/77975
26178 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
26179 edge to be constant.
26180 (get_val_for): For constant x return it. Formatting fix.
26181 (loop_niter_by_eval): Avoid pointless looping if the next iteration
26182 would use the same bases as the current one.
26184 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26186 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
26187 instead of vec_select for V1TImode.
26188 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
26190 (VSX_LE_128): Add V1TI to this mode iterator.
26191 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
26192 (*vsx_le_perm_store_<mode>): Likewise.
26193 (pre-reload splitter for VSX stores): Likewise.
26194 (post-reload splitter for VSX stores): Likewise.
26195 (*vsx_xxpermdi2_le_<mode>): Likewise.
26196 (*vsx_lxvd2x2_le_<mode>): Likewise.
26197 (*vsx_stxvd2x2_le_<mode>): Likewise.
26199 2017-03-09 Michael Eager <eager@eagercon.com>
26201 Correct failures with --enable-checking=yes,rtl.
26203 * config/microblaze/microblaze.c (microblaze_expand_shift):
26204 Replace GET_CODE test with CONST_INT_P and INTVAL test with
26205 test for const0_rtx.
26206 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
26207 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
26209 2017-03-09 Richard Biener <rguenther@suse.de>
26211 PR tree-optimization/79977
26212 * graphite-scop-detection.c (scop_detection::merge_sese):
26213 Handle the case of extra exits to blocks dominating the entry.
26215 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
26217 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
26220 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
26222 PR rtl-optimization/79949
26223 * lra-constraints.c (process_alt_operands): Check memory when
26224 trying to predict a cycle. Print about the overall increase.
26226 2017-03-09 Richard Biener <rguenther@suse.de>
26228 PR middle-end/79971
26229 * gimple-expr.c (useless_type_conversion_p): Preserve
26230 TYPE_SATURATING for fixed-point types.
26232 2017-03-09 Richard Biener <rguenther@suse.de>
26235 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
26236 alignment of BLKmode params.
26238 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26241 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
26242 (VALL_NO_V2Q): Likewise.
26244 * config/aarch64/aarch64-simd.md
26245 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
26247 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
26248 VALL_NO_V2Q mode iterator.
26249 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
26251 2017-03-09 Martin Liska <mliska@suse.cz>
26253 PR tree-optimization/79631
26254 * tree-chkp-opt.c (chkp_is_constant_addr): Call
26255 tree_int_cst_sign_bit just for INTEGER constants.
26257 2017-03-09 Martin Liska <mliska@suse.cz>
26261 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
26264 2017-03-09 Marek Polacek <polacek@redhat.com>
26267 * tree.c (inchash::add_expr): Handle TREE_VEC.
26269 2017-03-09 Martin Liska <mliska@suse.cz>
26272 (chkp_narrow_size_and_offset): New function.
26273 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
26274 (void chkp_parse_bit_field_ref): New function.
26275 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
26276 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
26278 2017-03-09 Martin Liska <mliska@suse.cz>
26281 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
26282 (chkp_find_bounds_1): Remove gcc_unreachable.
26284 2017-03-09 Jakub Jelinek <jakub@redhat.com>
26287 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
26288 BUILT_IN_SYNC*, determine the access type from the size suffix and
26289 always build a MEM_REF with that type. Handle forgotten
26290 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
26293 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
26294 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
26295 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
26296 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
26297 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
26298 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
26299 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
26300 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
26301 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
26302 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
26303 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
26304 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
26305 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
26306 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
26307 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
26308 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
26309 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
26310 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
26311 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
26312 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
26313 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
26314 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
26315 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
26316 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
26317 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
26318 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
26319 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
26320 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
26321 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
26322 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
26323 definitions outside of __OPTIMIZE__ guarded section.
26326 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
26327 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
26328 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
26329 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
26332 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26334 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
26335 ("vfenez<mode>"): Add missing constraints.
26337 2017-03-08 Martin Sebor <msebor@redhat.com>
26340 * config/nds32/nds32.c (nds32_option_override):
26341 Fix misspelled diagnostic.
26343 2017-03-08 Jakub Jelinek <jakub@redhat.com>
26346 * gimplify.c (gimplify_omp_for): Replace index var in outer
26347 taskloop statement with an artificial variable and add
26348 OMP_CLAUSE_PRIVATE clause for it.
26350 2017-03-08 Richard Biener <rguenther@suse.de>
26352 PR tree-optimization/79955
26353 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
26354 for accesses that are completely outside of the variable.
26356 2017-03-08 Andrew Haley <aph@redhat.com>
26358 PR tree-optimization/79943
26359 * tree-ssa-loop-split.c (compute_new_first_bound): When
26360 calculating the new upper bound, (END-BEG) should be added, not
26363 2017-03-08 Jakub Jelinek <jakub@redhat.com>
26365 * config/avr/avr.md (setmemhi): Make sure match_dup
26366 operand number comes before match_scratch.
26368 2017-03-08 Richard Biener <rguenther@suse.de>
26370 PR tree-optimization/79920
26371 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
26372 with ncopies == 1 to ...
26373 (vect_transform_slp_perm_load): ... here. Properly compute
26374 all element loads by iterating VF times over the group. Do
26375 not handle ncopies (computed in a broken way) in
26376 vect_create_mask_and_perm.
26378 2017-03-08 Jakub Jelinek <jakub@redhat.com>
26381 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
26382 is a uniform vector, use uniform_vector_p return value instead of
26383 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
26385 2017-03-07 Marek Polacek <polacek@redhat.com>
26387 PR middle-end/79809
26388 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
26389 (alloca_call_type): Likewise.
26391 2017-03-07 Martin Liska <mliska@suse.cz>
26393 * gcov.c (process_args): Put comment to correct location.
26395 2017-03-07 Martin Liska <mliska@suse.cz>
26397 PR middle-end/68270
26398 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
26399 Use array_at_struct_end_p instead of DECL_CHAIN (field).
26400 (chkp_narrow_bounds_for_field): Likewise.
26401 (chkp_parse_array_and_component_ref): Pass one more argument to
26404 2017-03-07 Richard Biener <rguenther@suse.de>
26406 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
26409 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
26411 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
26412 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
26414 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26417 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
26418 to end of description.
26419 (PARAM_MAX_STORES_TO_MERGE): Likewise.
26421 2017-03-07 Jakub Jelinek <jakub@redhat.com>
26423 PR rtl-optimization/79901
26424 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
26426 (*avx512f_<code><mode>3<mask_name>): ... this.
26427 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
26428 iterator instead of VI8_AVX2_AVX512BW.
26430 PR rtl-optimization/79901
26431 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
26432 min/max expander, expand it using expand_vec_cond_expr.
26435 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
26438 2017-03-06 Jakub Jelinek <jakub@redhat.com>
26441 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
26442 to void * for PCH reasons.
26443 * dwarf2out.c (output_loc_operands, output_die): Cast
26444 v.val_vec.array to unsigned char *.
26446 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
26449 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
26452 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
26454 PR rtl-optimization/79571
26455 * lra-constraints.c (process_alt_operands): Calculate static
26456 reject and subtract it from overall when only addresses will be
26459 2017-03-06 Julia Koval <julia.koval@intel.com>
26462 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
26463 incoming stack boundary to 128 for 64-bit targets.
26465 2017-03-06 Richard Biener <rguenther@suse.de>
26467 PR tree-optimization/79894
26468 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
26469 to NULL after folding it.
26471 2017-03-06 Richard Biener <rguenther@suse.de>
26473 PR tree-optimization/79824
26474 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
26475 check disabling peeling for gaps.
26477 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
26479 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
26480 attributes): Document gettimeofday.
26482 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
26484 * config/s390/s390.c (s390_option_override_internal): Set
26485 PARAM_MIN_VECT_LOOP_BOUND
26487 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
26489 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
26490 * config/s390/s390.md: Likewise.
26492 2017-03-06 Jakub Jelinek <jakub@redhat.com>
26495 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
26496 (<avx2_avx512>_perm<mode>): Rename to ...
26497 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
26499 (<avx512>_perm<mode>_mask): Rename to ...
26500 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
26502 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
26503 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
26504 instead of VI8F_256_512.
26505 (avx512f_perm<mode>): New define_expand.
26506 (avx512f_perm<mode>_mask): Likewise.
26507 (avx512f_perm<mode>_1<mask_name>): New define_insn.
26508 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
26510 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
26512 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
26513 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
26515 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
26517 2017-03-06 Martin Liska <mliska@suse.cz>
26520 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
26521 when having a SSA NAME w/o VAR_DECL assigned to it.
26523 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
26525 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
26526 msa_dpsub_<su>_d): Fix MODE for vec_select.
26528 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
26530 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
26532 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
26534 2017-03-06 Richard Biener <rguenther@suse.de>
26536 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
26537 * plugin.c (register_plugin_info): Likewise.
26538 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
26540 2017-03-05 Jakub Jelinek <jakub@redhat.com>
26542 * config/i386/sse.md (sse_storehps, sse_storelps,
26543 avx_<castmode><avxsizesuffix>_<castmode>,
26544 avx512f_<castmode><avxsizesuffix>_<castmode>,
26545 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
26546 in condition that at least one operand is not a MEM.
26548 2017-03-03 Jakub Jelinek <jakub@redhat.com>
26550 PR middle-end/79805
26551 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
26552 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
26554 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
26555 gimple_call_nothrow_p flag based on whether original builtin can throw.
26556 If it can, emit following stmts on the fallthrough edge.
26557 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
26558 don't create new bb if inserting just debug stmts on the edge, try to
26559 insert them on the fallthru bb or just reset debug stmts.
26561 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
26564 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
26565 restore recog_data (including the operand rtxes inside it) around
26566 the call to get_insn_template.
26568 2017-03-03 Martin Sebor <msebor@redhat.com>
26570 PR tree-optimization/79699
26571 * context.c (context::~context): Free MPFR caches to avoid
26572 a memory leak on program exit.
26574 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26576 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
26577 Use wide_int::ulow () instead of .elt (0).
26579 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
26581 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
26582 (*pushxf): Limit oF constraint to 32bit targets and add oC
26583 constraint for 64bit targets.
26584 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
26585 (*pushdf): Change rmF constraint to rmC.
26587 2017-03-03 Martin Liska <mliska@suse.cz>
26589 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
26590 Remove unused variable.
26592 2017-03-03 Jakub Jelinek <jakub@redhat.com>
26595 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
26596 is a memory operand, increase num_memory.
26597 (ix86_expand_args_builtin): Likewise.
26599 2017-03-03 Jan Hubicka <jh@suse.cz>
26602 * ipa-devirt.c (maybe_record_node): Properly handle
26603 __cxa_pure_virtual visibility.
26605 2017-03-03 Martin Liska <mliska@suse.cz>
26607 PR tree-optimization/79803
26608 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
26610 (pass_loop_prefetch::execute): Disabled optimization if an
26611 assumption about L1 cache size is not met.
26613 2017-03-03 Martin Liska <mliska@suse.cz>
26615 PR rtl-optimization/79574
26616 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
26617 (hash_scan_set): Likewise.
26618 (dump_hash_table): Likewise.
26619 (hoist_code): Likewise.
26621 2017-03-03 Richard Biener <rguenther@suse.de>
26623 * fixed-value.c (fixed_from_string): Restore use of elt (1)
26624 in place of uhigh ().
26625 (fixed_convert_from_real): Likewise.
26627 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
26630 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
26632 2017-03-03 Richard Biener <rguenther@suse.de>
26634 PR middle-end/79818
26635 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
26636 TYPE_OVERFLOW_UNDEFINED check.
26638 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26640 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
26642 (vector_ae_<mode>_p): Likewise.
26643 (vector_nez_<mode>_p): Likewise.
26644 (vector_ne_v2di_p): Likewise.
26645 (vector_ae_v2di_p): Likewise.
26646 (vector_ne_<mode>_p): Likewise.
26647 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
26649 (vsx_tsqrt<mode>2_fe): Likewise.
26651 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
26654 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
26656 2017-03-02 Jakub Jelinek <jakub@redhat.com>
26658 PR rtl-optimization/79780
26659 * cprop.c (one_cprop_pass): When second and further conditional trap
26660 in a single basic block is turned into an unconditional trap, turn it
26661 into a deleted note to avoid RTL verification failures.
26663 2017-03-02 Richard Biener <rguenther@suse.de>
26665 * fold-const.c (const_binop): Use ulow () instead of elt (0).
26667 2017-03-02 Richard Biener <rguenther@suse.de>
26669 PR tree-optimization/79345
26671 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
26672 param and abort the walk, returning -1 if it is hit.
26673 (walk_aliased_vdefs): Take a limit param and pass it on.
26674 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
26675 defaulting to 0 and return a signed int.
26676 * tree-ssa-uninit.c (struct check_defs_data): New struct.
26677 (check_defs): New helper.
26678 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
26679 about uninitialized memory.
26680 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
26681 bogus uninitialized warning.
26682 (fixed_convert_from_real): Likewise.
26684 2017-03-02 Bin Cheng <bin.cheng@arm.com>
26686 PR tree-optimization/66768
26687 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
26688 iv_use if base object can't be determined.
26690 2017-03-02 Jakub Jelinek <jakub@redhat.com>
26692 PR tree-optimization/79345
26693 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
26694 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
26695 (get_pattern_stats): Initialize it.
26696 * genemit.c (gen_expand): Verify match_scratch numbers come after
26697 match_operand/match_dup numbers.
26698 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
26699 match_scratch numbers.
26700 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
26702 * config/s390/s390.md (trunctdsd2): Likewise.
26704 2017-03-02 Richard Biener <rguenther@suse.de>
26706 * wide-int.h (wide_int_storage::operator=): Implement in terms
26709 2017-03-02 Richard Biener <rguenther@suse.de>
26711 PR tree-optimization/79777
26712 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
26713 the to insert expression to sth existing.
26715 2017-03-01 Martin Sebor <msebor@redhat.com>
26717 PR middle-end/79692
26718 * gimple-ssa-sprintf.c
26719 (directive::known_width_and_precision): New function.
26720 (format_integer): Use it.
26721 (get_mpfr_format_length): Consider the full range of precision
26722 when computing %g output with the # flag. Set the likely byte
26723 count to 3 rather than 1 when precision is indeterminate.
26724 (format_floating): Correct the lower bound of precision.
26726 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26728 * doc/invoke.texi: Document default code model for 64-bit Linux.
26730 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26733 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
26734 udiv rather than div since input pattern is unsigned.
26736 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
26738 * config/i386/i386.c (print_reg): Warn for values of
26739 unsupported size in integer register.
26741 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
26744 * config/rs6000/predicates.md (current_file_function_operand): Do
26745 not allow self calls to be local if the function is replaceable.
26747 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
26750 * config/rs6000/altivec.h (vec_ctz and others): Change the
26751 preprocessor macro that controls conditional compilation from
26752 _ARCH_PWR9 to __POWER9_VECTOR__.
26753 (vec_all_ne): Change parameterization of __altivec_scalar_pred
26754 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
26755 control (instead of _ARCH_PWR9 control) so that template
26756 definition uses power9-specific function.
26757 (vec_any_eq): Likewise.
26758 (vec_all_ne): Change macro definition to use a power9-specific
26759 expansion under #ifdef __POWER9_VECTOR__ control (instead of
26760 _ARCH_PWR9 control).
26761 (vec_any_eq) Likewise.
26762 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
26763 expansion for CMPNEF to remove support for xvcmpnesp instruction.
26764 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
26765 support for xvcmpnedp instruction.
26766 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
26767 macro expansion so that Power9 implementation of vec_all_ne does
26768 not use the AltiVec predicate framework.
26769 (VCMPNEH_P): Likewise.
26770 (VCMPNEW_P): Likewise.
26771 (VCMPNED_P): Likewise.
26772 (VCMPNEFP_P): Likewise.
26773 (VCMPNEDP_P): Likewise.
26774 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
26775 implementation of vec_any_eq to not use AltiVec predicate
26777 (VCMPAEH_P): Likewise.
26778 (VCMPAEW_P): Likewise.
26779 (VCMPAED_P): Likewise.
26780 (VCMPAEFP_P): Likewise.
26781 (VCMPAEDP_P): Likewise.
26782 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
26783 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
26784 not use the AltiVec predicate framework.
26785 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
26786 of vec_any_eq to not use AltiVec predicate framework.
26787 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
26788 support for predefined __POWER9_VECTOR__ macro to indicate that
26789 Power9 instruction selection is enabled.
26790 (altivec_overloaded_builtins): Remove extraneous
26791 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
26792 function argument types RS6000_BTI_bool_V16QI and
26793 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
26794 entry for overloaded function argument types RS6000_BTI_bool_V4SI
26795 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
26796 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
26797 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
26798 Power9 for implementations of vec_cmpne. Change the signature for
26799 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
26800 (representing vec_all_ne) to remove the previously described first
26801 argument of type RS6000_BTI_INTSI, as this was an artifact of
26802 reliance on the AltiVec predicate framework, which is no longer
26803 used in the implementation of these functions. Add
26804 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
26805 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
26806 since, unlike the AltiVec predicate framework implementation, we
26807 do not share function descriptors between vec_alle and vec_anyeq.
26808 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
26809 set of modes that receive special treatment even when
26810 TARGET_P9_VECTOR is true. The special treatment emits code that
26811 does not depend on Power9 instructions.
26812 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
26813 define_expand to not rely on AltiVec predicate framework.
26814 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
26816 (vector_ne_v2di_p): Change this define_expand to not rely on
26817 AltiVec predicate framework.
26818 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
26820 (vector_ne_<mode>_p): Change this define_expand to not rely on
26821 AltiVec predicate framework.
26822 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
26824 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
26825 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
26826 define_insn pattern.
26827 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
26828 define_insn pattern because the xvcmpne<VSs>. instruction is not
26830 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
26831 instruction is not supported.
26833 2017-03-01 Jakub Jelinek <jakub@redhat.com>
26835 * config/nvptx/nvptx.c: Include intl.h.
26837 2017-03-01 Martin Jambor <mjambor@suse.cz>
26840 * ipa-prop.h (ipa_bits): Removed field known.
26841 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
26842 to pointers. Adjusted their comments to warn about their sharing.
26843 (ipcp_transformation_summary): Change bits to a vector of pointers.
26844 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
26845 (ipa_get_ipa_bits_for_value): Declare.
26846 * tree-vrp.h (value_range): Mark as GTY((for_user)).
26847 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
26848 (ipa_bits_hash_table): Likewise.
26849 (ipa_vr_ggc_hash_traits): Likewise.
26850 (ipa_vr_hash_table): Likewise.
26851 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
26852 being pointers and vr_known being removed.
26853 (ipa_set_jf_unknown): Likewise.
26854 (ipa_get_ipa_bits_for_value): New function.
26855 (ipa_set_jfunc_bits): Likewise.
26856 (ipa_get_value_range): New overloaded functions.
26857 (ipa_set_jfunc_vr): Likewise.
26858 (ipa_compute_jump_functions_for_edge): Use the above functions to
26859 construct bits and vr parts of jump functions.
26860 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
26861 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
26863 (ipcp_grow_transformations_if_necessary): Also allocate
26864 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
26866 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
26867 them. Fix too long lines.
26868 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
26869 vr_known being removed.
26870 (ipa_read_jump_function): Use new setter functions to construct bits
26871 and vr parts of jump functions or set them to NULL.
26872 (write_ipcp_transformation_info): Adjust for bits being pointers.
26873 (read_ipcp_transformation_info): Likewise.
26874 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
26876 Include gt-ipa-prop.h.
26877 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
26879 (ipcp_store_bits_results): Likewise.
26880 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
26881 Do not write to existing jump functions but use a temporary instead.
26883 2017-03-01 Jakub Jelinek <jakub@redhat.com>
26886 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
26887 attempt to use its first operand as BIT_FIELD_REF base.
26889 2017-03-01 Richard Biener <rguenther@suse.de>
26891 PR middle-end/79721
26892 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
26893 interpolating formula in wrapping arithmetic.
26894 (chrec_apply): Convert chrec_evaluate return value to wanted type.
26896 2017-03-01 Jakub Jelinek <jakub@redhat.com>
26898 PR tree-optimization/79734
26899 * tree-vect-generic.c (expand_vector_condition): Optimize
26900 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
26901 Handle VEC_COND_EXPR where comparison has different inner width from
26902 type's inner width.
26904 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
26906 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
26907 markup, and similar issues. Remove @opindex entries for things
26908 that aren't options. Add missing -mmpy-option entries.
26910 2017-02-28 Jakub Jelinek <jakub@redhat.com>
26912 PR tree-optimization/79737
26913 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
26914 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
26915 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
26916 instead of byte_size. Formatting fix.
26917 (shift_bytes_in_array_right): Formatting fix.
26919 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
26922 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
26923 condition on optimize for the leaf function test.
26925 2017-02-28 Martin Liska <mliska@suse.cz>
26928 * read-rtl-function.c (function_reader::handle_unknown_directive):
26929 Bail out when one uses -flto.
26931 2017-02-28 Martin Liska <mliska@suse.cz>
26933 * common.opt: Replace space with tabular for options of <number>
26935 * config/i386/i386.opt: Show <number> value for
26936 -mlarge-data-threshold.
26937 * opts.c (print_filtered_help): Do not display number in hexadecimal
26940 2017-02-28 Martin Liska <mliska@suse.cz>
26942 * common.opt: Fix --help=option -Q for options which are of
26945 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
26947 * config/i386/i386.c (print_reg): Error out for values
26948 of 8-bit size in invalid integer register.
26950 2017-02-28 Martin Sebor <msebor@redhat.com>
26952 PR tree-optimization/79691
26953 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
26955 2017-02-28 Jakub Jelinek <jakub@redhat.com>
26958 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
26959 gcc_unreachable with output_operand_lossage.
26961 2017-02-28 Richard Biener <rguenther@suse.de>
26963 PR tree-optimization/79740
26964 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
26966 (visit_nary_op): Insert the nary into the hashtable if we
26967 pattern-matched sth.
26968 * tree-ssa-pre.c (eliminate_insert): Robustify.
26970 2017-02-28 Richard Biener <rguenther@suse.de>
26972 PR middle-end/79731
26973 * fold-const.c (decode_field_reference): Reject out-of-bound
26976 2017-02-28 Jakub Jelinek <jakub@redhat.com>
26978 * config/i386/i386.c: Include intl.h.
26979 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
26980 instead of just cond ? "..." : "...".
26981 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
26982 * coverage.c (read_counts_file): Likewise.
26983 * omp-offload.c: Include intl.h.
26984 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
26985 of just cond ? "..." : "...".
26986 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
26987 of just cond ? "..." : "...".
26989 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
26992 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
26994 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
26995 'tune for' CPU name.
26996 * config/arm/arm-cpu-data.h: Regenerated.
26998 2017-02-28 Richard Biener <rguenther@suse.de>
27000 PR tree-optimization/79732
27001 * tree-inline.c (expand_call_inline): Do not shadow var.
27003 2017-02-28 Richard Biener <rguenther@suse.de>
27005 PR tree-optimization/79723
27006 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
27007 address-space properly.
27009 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
27011 * doc/optinfo.texi (Optimization groups): Fix option used for
27013 * doc/invoke.texi (-fopt-info): Document "omp".
27014 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
27015 (OPTGROUP_ALL): Add OPTGROUP_OMP.
27016 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
27017 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
27018 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
27020 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
27022 * dumpfile.c (optgroup_options): Instead of "openmp", associate
27023 OPTGROUP_OMP with "omp".
27025 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
27028 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
27029 for arithmetic shift of unsigned V2DI.
27031 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
27033 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
27034 arc/linux.h headers.
27035 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
27036 (LINK_SPEC): Likewise.
27037 (ARC_TLS_EXTRA_START_SPEC): Likewise.
27038 (EXTRA_SPECS): Likewise.
27039 (STARTFILE_SPEC): Likewise.
27040 (ENDFILE_SPEC): Likewise.
27041 (LIB_SPEC): Likewise.
27042 (TARGET_SDATA_DEFAULT): Likewise.
27043 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
27044 (MULTILIB_DEFAULTS): Likewise.
27045 (DWARF2_UNWIND_INFO): Likewise.
27046 * config/arc/big.h: New file.
27047 * config/arc/elf.h: Likewise.
27048 * config/arc/linux.h: Likewise.
27049 * config/arc/t-uClibc: Remove.
27051 2017-02-27 Bin Cheng <bin.cheng@arm.com>
27053 PR tree-optimization/77536
27054 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
27055 (tree_transform_and_unroll_loop): Use above function to compute the
27056 estimated niter of unrolled loop and use it when scaling profile.
27057 Also use count info rather than frequency if it's non-zero.
27058 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
27059 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
27060 (vect_transform_loop): Call above function.
27062 2017-02-27 Richard Biener <rguenther@suse.de>
27064 PR tree-optimization/45397
27065 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
27066 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
27067 (visit_nary_op): Add pattern matching for CSEing sign-changed
27068 or truncated operations with wider ones.
27070 2017-02-27 Richard Biener <rguenther@suse.de>
27072 PR tree-optimization/79690
27073 * tree-vect-stmts.c (vectorizable_store): Use vector type
27074 built from the DR with address-space.
27076 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
27078 * doc/invoke.texi (Optimize Options): Refine the description
27079 of asan-use-after-return.
27081 2017-02-25 Alan Modra <amodra@gmail.com>
27083 PR rtl-optimization/79584
27084 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
27085 base, not ad->base_term, the reg within base. Remove assertion
27086 that ad->base == ad->base_term. Replace gen_int_mode using
27087 bogus mode with const0_rtx.
27089 2017-02-25 Jakub Jelinek <jakub@redhat.com>
27091 PR middle-end/79396
27092 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
27093 FMA_EXPR like tcc_binary or tcc_unary.
27095 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
27098 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
27100 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
27101 (output_loc_operands): Handle DW_OP_call_ref and
27102 DW_OP_GNU_variable_value.
27103 (struct variable_value_struct): New type.
27104 (struct variable_value_hasher): Likewise.
27105 (variable_value_hash): New variable.
27106 (string_types): Remove.
27107 (copy_loc_descr): New function.
27108 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
27109 (prepend_loc_descr_to_each): New function.
27110 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
27111 instead of add_loc_descr_to_each if the first argument is single
27112 location list and the second has multiple.
27113 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
27114 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
27115 when looking for variable value which doesn't have other location info.
27116 (loc_list_from_tree): Formatting fix.
27117 (gen_array_type_die): Simplify DW_AT_string_length handling.
27118 (adjust_string_types): Remove.
27119 (gen_subprogram_die): Don't call adjust_string_types nor test/set
27120 string_types. Call resolve_variable_values.
27121 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
27122 (resolve_addr_in_expr): Likewise. Add A argument.
27123 (copy_deref_exprloc): Remove deref argument. Adjust for the
27124 original expression being DW_OP_GNU_variable_value with optionally
27125 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
27126 optionally after it.
27127 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
27128 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
27129 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
27130 (variable_value_hasher::hash, variable_value_hasher::equal): New
27132 (resolve_variable_value_in_expr, resolve_variable_value,
27133 resolve_variable_values, note_variable_value_in_expr,
27134 note_variable_value): New functions.
27135 (dwarf2out_early_finish): Call note_variable_value on all toplevel
27138 2017-02-24 Jakub Jelinek <jakub@redhat.com>
27141 * opts.h (handle_generated_option): Add GENERATED_P argument.
27142 * opts-common.c (handle_option): Adjust function comment.
27143 (handle_generated_option): Add GENERATED_P argument, pass it to
27145 (control_warning_option): Pass false to handle_generated_option
27147 * opts.c (maybe_default_option): Pass true to handle_generated_option
27149 * optc-gen.awk: Likewise.
27151 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27153 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
27154 a REG, look at the REG it is a SUBREG of.
27155 (splitter for cmpeqsi_t): Ditto.
27157 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27159 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
27160 the special USEs with the pattern of the insn, not the insn itself.
27162 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
27165 * doc/invoke.texi: Document -mload-store-pairs.
27167 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27168 Sandra Loosemore <sandra@codesourcery.com>
27170 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
27171 argument isn't a CONST_INT.
27172 (nios2_alternate_compare_const): Assert op is a CONST_INT.
27173 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
27174 (nios2_validate_compare): Bypass alternate compare logic if *op2
27175 is not a CONST_INT.
27176 (ldstwm_operation_p): Return false if first_base is not a REG or
27177 if first_offset is not a CONST_INT.
27179 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27181 * config/cris/cris.md: Use correct operand in a define_peephole2.
27183 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27185 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
27187 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
27189 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
27190 this_insn if it is an INSN or JUMP_INSN.
27191 (force_offsettable): Look at base, not at addr.
27192 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
27193 on things that aren't necessarily CONST_INTs.
27195 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
27197 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
27198 -mfpmath=sse is the default also for x86-32 targets with SSE2
27199 instruction set when @option{-ffast-math} is enabled
27201 2017-02-24 Jeff Law <law@redhat.com>
27203 PR rtl-optimizatoin/79286
27204 * ira.c (update_equiv_regs): Drop may_trap_p exception to
27207 2017-02-24 Richard Biener <rguenther@suse.de>
27209 PR tree-optimization/79389
27210 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
27213 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
27215 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
27216 function comment to reflect reality.
27217 (loop_exits_before_overflow): Fix typo in function description.
27219 2017-02-24 Richard Biener <rguenther@suse.de>
27221 PR tree-optimization/79389
27222 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
27223 properly that a threading opportunity exists. Detect conditional
27224 copy/constant propagation opportunities.
27226 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
27228 * config/visium/visium.md (type): Add trap.
27229 (b): New mode attribute.
27230 (*btst): Rename into...
27231 (*btst<mode>): ...this and adjust.
27232 (*cbranchsi4_btst_insn): Rename into...
27233 (*cbranch<mode>4_btst_insn): ...this and adjust.
27234 (trap): New define_insn.
27236 2017-02-23 Jakub Jelinek <jakub@redhat.com>
27238 PR tree-optimization/79389
27239 * ifcvt.c (struct noce_if_info): Add rev_cond field.
27240 (noce_reversed_cond_code): New function.
27241 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
27242 reversed_comparison_code. Formatting fix.
27243 (noce_try_store_flag): Test rev_cond != NULL in addition to
27244 reversed_comparison_code.
27245 (noce_try_store_flag_constants): Likewise.
27246 (noce_try_store_flag_mask): Likewise.
27247 (noce_try_addcc): Use rev_cond if non-NULL instead of
27248 reversed_comparison_code.
27249 (noce_try_cmove_arith): Likewise. Formatting fixes.
27250 (noce_try_minmax, noce_try_abs): Clear rev_cond.
27251 (noce_find_if_block): Initialize rev_cond.
27252 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
27253 instead of false as last argument never attempt to reverse it
27256 2017-02-23 Bin Cheng <bin.cheng@arm.com>
27258 PR tree-optimization/79663
27259 * tree-predcom.c (combine_chains): Process refs in reverse order
27260 only for ZERO length chains, and add explaining comment.
27262 2017-02-23 Jeff Law <law@redhat.com>
27264 PR tree-optimization/79578
27265 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
27266 in call to operand_equal_p.
27268 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
27271 * config/i386/cpuid.h: Fix another undefined behavior.
27273 2017-02-23 Richard Biener <rguenther@suse.de>
27275 PR tree-optimization/79683
27276 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
27277 vector types for data-refs.
27279 2017-02-23 Martin Liska <mliska@suse.cz>
27281 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
27283 2017-02-23 Jakub Jelinek <jakub@redhat.com>
27285 PR middle-end/79665
27286 * internal-fn.c (get_range_pos_neg): Moved to ...
27287 * tree.c (get_range_pos_neg): ... here. No longer static.
27288 * tree.h (get_range_pos_neg): New prototype.
27289 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
27290 are known to be in between 0 and signed maximum inclusive, try to
27291 expand both unsigned and signed divmod and use the cheaper one from
27294 2017-02-22 Jeff Law <law@redhat.com>
27296 PR tree-optimization/79578
27297 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
27298 to compare base operands.
27300 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
27303 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
27304 gpc_reg_operand instead of fpr_reg_operand.
27306 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
27308 * config/mips/mips.c (mips_return_in_memory): Force FP
27309 vector types to be returned in memory for o32 ABI.
27311 2017-02-22 Jakub Jelinek <jakub@redhat.com>
27313 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
27314 instead of DW_TAG_member for static data member declarations and don't
27315 set no_linkage_name for static inline data members.
27316 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
27319 2017-02-22 Martin Liska <mliska@suse.cz>
27321 * doc/invoke.texi: Replace inequality signs with square brackets
27324 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27326 PR tree-optimization/68644
27327 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
27329 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
27332 * lra-constraints.c (simplify_operand_subreg): Handle
27333 WORD_REGISTER_OPERATIONS targets.
27335 2017-02-22 Jakub Jelinek <jakub@redhat.com>
27338 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
27339 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
27340 elimination by swapping fld*.
27342 2017-02-22 Richard Biener <rguenther@suse.de>
27344 PR tree-optimization/79673
27345 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
27346 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
27347 irrelevant address-space qualifiers and avoiding a
27348 ADDR_SPACE_CONVERT_EXPR from fold_convert.
27350 2017-02-22 Richard Biener <rguenther@suse.de>
27352 PR tree-optimization/79666
27353 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
27354 to not symbolically negate if that may introduce undefined
27357 2017-02-22 Martin Liska <mliska@suse.cz>
27360 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
27361 * data-streamer-out.c (streamer_write_gcov_count_stream):
27363 * value-prof.c (stream_out_histogram_value): Make assert more
27364 precise based on type of counter.
27366 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
27369 * config/i386/i386.md (standard_x87sse_constant_load splitter):
27370 Use nonimmediate_operand instead of memory_operand for operand 1.
27371 (float-extend standard_x87sse_constant_load splitter): Ditto.
27373 2017-02-21 Jeff Law <law@redhat.com>
27375 PR tree-optimization/79621
27376 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
27377 blocks with edges to themselves.
27379 2017-02-21 Jakub Jelinek <jakub@redhat.com>
27382 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
27383 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
27384 Use gimple_call_builtin_p.
27387 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
27388 on temporarily removed DEBUG_INSNs.
27390 PR tree-optimization/79649
27391 * tree-loop-distribution.c (classify_partition): Give up on
27392 non-generic address space loads/stores.
27394 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
27396 * doc/loop.texi (Loop manipulation): Remove nonexistent
27397 tree_ssa_loop_version from the documentation.
27398 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
27400 2017-02-21 Jakub Jelinek <jakub@redhat.com>
27403 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
27404 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
27405 * config/rs6000/rs6000.c: Include except.h.
27406 (rs6000_expand_split_stack_prologue): Call
27407 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
27409 2017-02-21 Martin Jambor <mjambor@suse.cz>
27412 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
27413 have been analyzed.
27415 2017-02-21 Martin Jambor <mjambor@suse.cz>
27417 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
27418 for backward compatibility only.
27419 * doc/invoke.texi (Option Summary): Remove all references to
27420 -fipa-cp-alignment.
27422 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
27426 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27428 * lra-constraints.c (curr_insn_transform): Handle
27429 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
27431 2017-02-21 Martin Liska <mliska@suse.cz>
27433 * config/i386/i386.opt: Replace -masm-dialect with -masm.
27435 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
27437 PR translation/79638
27438 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
27440 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
27443 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
27444 (arm_function_ok_for_sibcall): Return false for an indirect call by
27445 descriptor if all the argument registers are used.
27446 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
27447 alignment of the function.
27449 2017-02-21 Jakub Jelinek <jakub@redhat.com>
27451 PR tree-optimization/61441
27452 * simplify-rtx.c (simplify_const_unary_operation): For
27453 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
27454 the sNaN unmodified.
27456 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
27458 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
27459 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
27460 instead of SYSTEM_HEADER_DIR.
27462 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
27463 Martin Liška <mliska@suse.cz>
27465 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
27466 Fix typos and grammar, use active voice, and clarify.
27468 2017-02-20 Marek Polacek <polacek@redhat.com>
27470 PR middle-end/79537
27471 * gimplify.c (gimplify_expr): Handle unused *&&L;.
27474 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
27476 2017-02-20 Jakub Jelinek <jakub@redhat.com>
27479 * config/i386/i386.c (ix86_expand_builtin): Handle
27480 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
27481 ix86_builtins_isa[fcode].isa as a requirement of those
27482 flags and any other flag in the bitmask.
27483 (ix86_init_mmx_sse_builtins): Use 0 instead of
27484 ~OPTION_MASK_ISA_64BIT as mask.
27485 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
27486 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
27487 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
27488 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
27490 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27493 * lra-constraints.c (split_reg): Check requested split mode
27494 is supported by the register.
27496 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27498 * lra-constraints.c (simplify_operand_subreg): Remove early
27501 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27504 * lra-constraints.c (curr_insn_transform): Tighten condition
27505 for converting SUBREG reloads from OP_OUT to OP_INOUT.
27507 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
27510 * lra-constraints.c (curr_insn_transform): Handle
27511 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
27513 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
27516 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
27518 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
27520 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
27523 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
27526 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27528 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
27531 2017-02-18 Jakub Jelinek <jakub@redhat.com>
27533 * final.c (last_columnnum, override_columnnum): New variables.
27534 (final_start_function): Set last_columnnum, pass it to begin_prologue
27535 hook and pass 0 to dwarf2out_begin_prologue.
27536 (final_scan_insn): Update override_columnnum. Pass last_columnnum
27537 to source_line debug hook.
27538 (notice_source_line): Compute last_columnnum and for debug_column_info
27539 return true on column changes.
27540 * debug.h (struct gcc_debug_hooks): Add column argument to
27541 source_line and begin_prologue hooks.
27542 (debug_nothing_int_charstar_int_bool): Remove prototype.
27543 (debug_nothing_int_int_charstar,
27544 debug_nothing_int_int_charstar_int_bool): New prototypes.
27545 (dwarf2out_begin_prologue): Add column argument.
27546 * debug.c (do_nothing_debug_hooks): Adjust source_line and
27547 begin_prologue hooks.
27548 (debug_nothing_int_charstar_int_bool): Remove.
27549 (debug_nothing_int_int_charstar,
27550 debug_nothing_int_int_charstar_int_bool): New functions.
27551 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
27552 through to dwarf2out_source_line.
27553 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
27554 (dwarf2out_source_line): Add column argument, emit it if requested.
27555 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
27557 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
27558 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
27559 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
27560 through to dwarf2out_begin_prologue.
27561 (vmsdbgout_source_line): Add column argument, pass it through to
27562 dwarf2out_source_line.
27563 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
27564 dbxout_source_line caller.
27565 (dbxout_source_line): Add column argument.
27567 * common.opt (gno-column-info, gcolumn-info): New options.
27568 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
27569 (check_die): Also test for multiple DW_AT_decl_column attributes.
27570 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
27571 DW_AT_decl_column if requested.
27572 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
27574 (gen_variable_die): Likewise.
27575 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
27576 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
27579 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
27580 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
27581 (ix86_handle_option): Handle OPT_m3dnowa.
27582 * doc/invoke.texi (-m3dnowa): Document.
27583 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
27584 -m3dnowa instead of -m3dnow -march=athlon.
27587 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
27588 instead of gcc_assert for K, r and R code checks. Formatting fixes.
27590 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27593 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
27594 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
27595 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
27596 generator for vsx_xxpermdi_<mode>_be.
27597 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
27598 force big-endian semantics.
27599 (vsx_xxpermdi_<mode>_be): New define_expand with same
27600 implementation as previous version of vsx_xxpermdi_<mode>.
27602 2017-02-17 Jakub Jelinek <jakub@redhat.com>
27604 PR tree-optimization/79327
27605 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
27606 variable, its initialization and use.
27608 2017-02-17 Julia Koval <julia.koval@intel.com>
27610 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
27611 (OPTION_MASK_ISA_PKU_UNSET): New.
27612 (ix86_handle_option): Handle -mrdpid.
27613 * config/i386/cpuid.h (bit_RDPID): New.
27614 * config/i386/driver-i386.c (host_detect_local_cpu):
27615 Detect RDPID feature.
27616 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
27617 * config/i386/i386-c.c (ix86_target_macros_internal):
27619 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
27620 (ix86_valid_target_attribute_inner_p): Add "rdpid".
27621 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
27622 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
27623 * config/i386/i386.md (define_insn "rdpid"): New.
27624 * config/i386/i386.opt Add -mrdpid.
27625 * config/i386/immintrin.h (_rdpid_u32): New.
27627 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
27629 PR rtl-optimization/79541
27630 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
27631 instead of transforming it into USE.
27633 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
27635 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
27636 If HONOR_SNANS (SFmode) force the input to a register.
27637 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
27638 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
27639 an frsp or similar insn.
27641 2017-02-17 Martin Liska <mliska@suse.cz>
27643 PR rtl-optimization/79577
27644 * params.def (selsched-max-sched-times): Increase minimum to 1.
27646 2017-02-17 Martin Liska <mliska@suse.cz>
27648 PR rtl-optimization/79574
27649 * gcse.c (want_to_gcse_p): Prevent integer overflow.
27651 2017-02-17 Martin Liska <mliska@suse.cz>
27653 PR tree-optimization/79529
27654 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
27655 ssa_defined_default_def_p to handle cases which are implicitly
27657 * tree-ssa.c (ssa_defined_default_def_p): New function.
27658 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
27659 which are implicitly defined.
27660 * tree-ssa.h (ssa_defined_default_def_p): Declare.
27662 2017-02-17 Richard Biener <rguenther@suse.de>
27664 PR middle-end/79576
27665 * params.def (max-ssa-name-query-depth): Limit to 10.
27667 2017-02-17 Richard Biener <rguenther@suse.de>
27669 PR tree-optimization/79552
27670 * tree-ssa-structalias.c (visit_loadstore): Properly verify
27673 2017-02-17 Richard Biener <rguenther@suse.de>
27676 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
27678 2017-02-17 Marek Polacek <polacek@redhat.com>
27680 PR middle-end/79536
27681 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
27682 (fold_negate_expr): New wrapper.
27684 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
27686 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
27687 Correct terminology and de-emphasize pre-standard behavior.
27689 2017-02-16 Alan Modra <amodra@gmail.com>
27691 PR rtl-optimization/79286
27692 * ira.c (def_dominates_uses): New function.
27693 (update_equiv_regs): Don't create an equivalence for insns that
27694 may trap where the register def does not dominate the use.
27696 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
27698 PR rtl-optimization/78127
27699 * lra.c (lra): Call lra_eliminate before finish the loop after
27702 2017-02-16 Richard Biener <rguenther@suse.de>
27704 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
27706 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
27707 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
27708 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
27709 (isl_val_int_from_wi): New function.
27710 (extract_affine_gmp): Rename to ...
27711 (extract_affine_wi): ... this, take a widest_int.
27712 (extract_affine_int): Just wrap extract_affine_wi.
27713 (add_param_constraints): Use isl_val_int_from_wi.
27714 (add_loop_constraints): Likewise, and extract_affine_wi.
27716 2017-02-15 Jeff Law <law@redhat.com>
27718 PR middle-end/79521
27719 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
27720 ira_init_register_move_cost_if_necessary.
27722 2017-02-15 Martin Sebor <msebor@redhat.com>
27724 PR middle-end/32003
27725 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
27726 removed in a prior commit.
27728 2017-02-15 Bin Cheng <bin.cheng@arm.com>
27730 PR tree-optimization/79347
27731 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
27732 counters during peeling.
27734 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
27736 * Makefile.in (site.exp): Remove "set ISLVER".
27738 2017-02-15 Jakub Jelinek <jakub@redhat.com>
27741 * real.c (real_from_integer): Call real_convert even for decimal.
27743 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
27746 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
27748 2017-02-14 Andrew Pinski <apinski@cavium.com>
27750 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
27751 cores and change the partno/implementer to be correct.
27752 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
27753 the 'B" as the implementer.
27754 * config/aarch64/aarch64-tune.md: Regenerate.
27756 2017-02-14 Carl Love <cel@us.ibm.com>
27758 * config/rs6000/rs6000.c: Add case statement entry to make the
27759 xvcvuxdsp built-in argument unsigned.
27760 * config/rs6000/vsx.md: Fix the source and return operand types so they
27761 match the instruction definitions from the ISA document. Fix typo
27762 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
27765 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
27768 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
27769 member early_clobber_alts.
27770 * lra-lives.c (reg_early_clobber_p): New.
27771 (process_bb_lives): Use it.
27772 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
27773 (debug_operand_data): Initialize early_clobber_alts.
27774 (setup_operand_alternative): Set up early_clobber_alts.
27775 (collect_non_operand_hard_regs): Ditto. Pass early clobber
27776 alternatives to new_insn_reg.
27777 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
27779 (lra_update_insn_regno_info): Pass the new arg.
27781 2017-02-14 Jakub Jelinek <jakub@redhat.com>
27783 PR middle-end/79505
27784 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
27785 (new_oacc_loop_raw): Don't clear already cleared fields.
27788 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
27789 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
27790 _mm512_prefetch_i64gather_ps): New inline functions and macros.
27792 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
27795 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
27797 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
27800 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
27801 the extra instruction to the right place to store 128-bit constant
27804 2017-02-14 Martin Sebor <msebor@redhat.com>
27806 PR middle-end/79448
27807 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
27808 warning for strings of unknown length.
27810 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
27812 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
27814 2017-02-14 Jeff Law <law@redhat.com>
27817 * ira-costs.c (scan_one_insn): Initialize register move costs
27818 for pseudos seen in USE/CLOBBER insns.
27820 PR tree-optimization/79095
27821 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
27822 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
27823 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
27824 if the operands are known to be not equal, then the resulting range
27826 (intersect_ranges): If the new range is ~[0,0] and the old range is
27827 wide, then prefer ~[0,0].
27828 * tree-vrp.c (overflow_comparison_p_1): New function.
27829 (overflow_comparison_p): New function.
27830 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
27831 if NAME is used in an overflow test.
27832 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
27833 overflow check that can be expressed as an equality test, then adjust
27834 ops to be that equality test.
27836 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27838 * config/s390/s390-builtin-types.def: Remove flags argument.
27839 * config/s390/s390.c (s390_init_builtins): Likewise.
27841 2017-02-14 Martin Liska <mliska@suse.cz>
27843 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
27844 vector. Fix trailing white spaces.
27846 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
27848 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
27851 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27853 PR rtl-optimization/68664
27854 * config/arm/arm.c (arm_sched_can_speculate_insn):
27855 New function. Declare prototype.
27856 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
27858 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27860 PR rtl-optimization/68664
27861 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
27863 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
27865 2017-02-14 Amit Pawar <amit.pawar@amd.com>
27867 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
27868 max skip bytes for function, loop and jump.
27870 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27872 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
27873 ABS_EXPR for gimple dump.
27875 2017-02-14 Jakub Jelinek <jakub@redhat.com>
27878 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
27880 PR tree-optimization/79408
27881 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
27882 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
27883 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
27884 also if rhs1 is INTEGER_CST.
27886 2017-02-14 Richard Biener <rguenther@suse.de>
27888 PR middle-end/79432
27889 * tree-into-ssa.c (insert_phi_nodes): When the function can
27890 have abnormal edges rewrite SSA names with broken use-def
27891 dominance out of SSA and register them for PHI insertion.
27893 2017-02-13 Martin Sebor <msebor@redhat.com>
27895 PR middle-end/79496
27896 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
27897 clearing info.nowrite flag when snprintf size argument is a range.
27899 2017-02-13 Jakub Jelinek <jakub@redhat.com>
27901 * cprop.c (cprop_jump): Add missing space in string literal.
27902 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
27903 (get_constraint_for_component_ref): Likewise.
27904 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
27905 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
27906 * lra-constraints.c (process_alt_operands): Likewise.
27907 * ipa-inline.c (inline_small_functions): Likewise.
27908 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
27909 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
27910 * trans-mem.c (diagnose_tm_1_op): Likewise.
27911 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
27912 (grid_parallel_clauses_gridifiable): Likewise.
27914 * config/nvptx/mkoffload.c (process): Add space in between
27917 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
27918 "MOD4_SSE_REGS" and "ALL_REGS".
27920 * spellcheck.c (test_data): Add , in between "foo" and "food".
27922 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27925 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
27926 boundary crossing check and subsequent code generation agree.
27928 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27930 * config/aarch64/aarch64.c (has_memory_op): Delete.
27931 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
27934 2017-02-13 Jakub Jelinek <jakub@redhat.com>
27936 PR rtl-optimization/79388
27937 PR rtl-optimization/79450
27938 * combine.c (distribute_notes): When removing TEM_INSN for which
27939 corresponding dest has last value recorded, invalidate that last
27942 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27944 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
27945 of explicit '@'. Add missing assembly comment marker on branch costs
27948 2017-02-13 Nathan Sidwell <nathan@acm.org>
27950 * gengtype-lex.l (<in_struct>): Add '/'.
27952 2017-02-13 Martin Liska <mliska@suse.cz>
27955 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
27957 2017-02-13 Richard Biener <rguenther@suse.de>
27959 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
27961 * configure: Re-generate.
27962 * config.in: Likewise.
27963 * graphite-dependences.c: Simplify as if
27964 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
27965 * graphite-isl-ast-to-gimple.c: Likewise.
27966 * graphite-optimize-isl.c: Likewise.
27967 * graphite-poly.c: Likewise.
27968 * graphite-sese-to-poly.c: Likewise.
27969 * graphite.h: Likewise.
27970 * toplev.c: Include isl/version.h and use isl_version () for
27971 printing the ISL version.
27972 * doc/install.texi: Update ISL requirement.
27974 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
27976 * doc/standards.texi (Standards): Update reference to
27979 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
27981 * doc/extend.texi (Named Address Spaces): sourceware.org now
27983 * doc/install.texi (Binaries): Ditto.
27986 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
27988 * doc/cpp.texi: Replace "stringify"/"stringification" with C
27989 standard terminology "stringize"/"stringizing" throughout.
27990 * doc/cppinternals.texi: Likewise.
27992 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
27994 * doc/extend.texi: Fix some spelling mistakes and typos.
27995 * doc/invoke.texi: Likewise.
27997 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
28000 * params.def (inline-min-speedup) Change from 10 to 8.
28002 2017-02-11 Jakub Jelinek <jakub@redhat.com>
28004 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
28007 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
28010 * ipa-inline-analysis.c (get_minimal_bb): New function.
28011 (record_modified): Use it.
28012 (remap_edge_change_prob): Handle also ancestor functions.
28014 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
28016 * doc/contrib.texi (Contributors): Remove broken link into
28017 the Mauve CVS repository.
28019 2017-02-11 Jakub Jelinek <jakub@redhat.com>
28021 PR middle-end/79454
28022 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
28023 result computation whenever lhs doesn't have vector mode, not
28024 just when it has BLKmode.
28026 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
28028 * doc/makefile.texi (profiledbootstrap): Refer to the
28029 installation instructions only in textual form.
28031 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28034 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
28036 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
28038 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
28039 (Specific): Update mingw-w64 reference.
28041 (Specific): Remove broken link to Renesas RX processor.
28043 2017-02-10 Richard Biener <rguenther@suse.de>
28045 * toplev.c (process_options): Do not mention obsolete graphite
28046 options when printing sorry message about missing graphite support.
28047 Mention -floop-nest-optimize.
28049 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
28051 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
28052 (vtst_p16): Likewise.
28053 (vtstq_p8): Likewise.
28054 (vtstq_p16): Likewise.
28056 (vtstq_p64): Likewise.
28057 * config/arm/arm_neon.h (vgetq_lane_p64): New.
28058 (vset_lane_p64): New.
28059 (vsetq_lane_p64): New.
28061 2017-02-10 Jakub Jelinek <jakub@redhat.com>
28063 PR tree-optimization/79411
28064 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
28065 stmt operands are SSA_NAMEs used in abnormal phis.
28066 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
28069 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
28072 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
28075 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
28077 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
28079 2017-02-09 Jakub Jelinek <jakub@redhat.com>
28081 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
28085 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
28086 not arbitrary TREE_CONSTANT.
28089 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
28090 "omp declare target link" attribute unless is_global_var.
28091 * omp-offload.c (find_link_var_op): Likewise.
28093 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
28094 Chung-Lin Tang <cltang@codesourcery.com>
28096 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
28098 (gimplify_adjust_omp_clauses): Don't delete TILE.
28099 (gimplify_omp_for): Deal with TILE.
28100 * internal-fn.c (expand_GOACC_TILE): New function.
28101 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
28103 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
28104 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
28106 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
28107 avoid DIV for outermost collapse var.
28108 (expand_oacc_for): Insert tile element loop as needed. Adjust.
28109 Remove out of date comments, fix whitespace.
28110 * omp-general.c (omp_extract_for_data): Deal with tiling.
28111 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
28112 adjust OLF_DIM_BASE value.
28113 (struct omp_for_data): Add tiling field.
28114 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
28115 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
28116 for auto loops. Remove default auto determining, moved to
28117 oacc_loop_fixed_partitions.
28118 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
28119 stmts, add e_mask field.
28120 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
28121 (oacc_thread_numbers): Use oacc_dim_call.
28122 (oacc_xform_tile): New.
28123 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
28124 (finish_oacc_loop): Adjust for ifns vector.
28125 (oacc_loop_discover_walk): Append loop abstraction sites to list,
28126 add case for GOACC_TILE fns.
28127 (oacc_loop_xform_loop): Delete.
28128 (oacc_loop_process): Iterate over call list directly, and add
28129 handling for GOACC_TILE fns.
28130 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
28132 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
28133 vector partitioning to outer loops. Assign 2 partitions to loops
28134 when available. Add TILE handling.
28135 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
28136 (execite_oacc_device_lower): Process GOACC_TILE fns,
28137 ignore unknown specs.
28138 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
28139 * tree.c (omp_clause_num_ops): Adjust TILE ops.
28140 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
28142 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
28144 * configure.ac (ACX_BUGURL): Update.
28145 * configure: Regenerate.
28147 2017-02-09 Richard Biener <rguenther@suse.de>
28149 PR tree-optimization/69823
28150 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
28151 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
28153 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
28155 * config/arc/arc-c.def: Add __NPS400__ definition.
28156 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
28157 (TARGET_NPS400): Define.
28159 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
28161 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
28163 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
28164 pointer, arch_info.
28165 (arc_cpu_types): Fill the arch_info field with a pointer into the
28166 arc_arch_types table.
28167 (arc_selected_cpu): Declare.
28168 * config/arc/arc.c (arc_selected_cpu): Make global.
28169 (arc_selected_arch): Delete.
28170 (arc_base_cpu): Delete.
28171 (arc_override_options): Remove references to deleted variables,
28172 update access to arch information.
28173 (ARC_OPT): Update access to arch information.
28174 (ARC_OPTX): Likewise.
28175 * config/arc/arc.h (arc_base_cpu): Remove declaration.
28176 (TARGET_ARC600): Update access to arch information.
28177 (TARGET_ARC601): Likewise.
28178 (TARGET_ARC700): Likewise.
28179 (TARGET_EM): Likewise.
28180 (TARGET_HS): Likewise.
28181 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
28184 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
28187 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
28188 condition/operands for integer GE/LE/GEU/LEU operations.
28190 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
28192 PR translation/79397
28193 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
28196 2017-02-08 Martin Jambor <mjambor@suse.cz>
28199 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
28200 whether allocation happened.
28201 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
28202 nothing was allocated.
28204 2017-02-08 Jakub Jelinek <jakub@redhat.com>
28206 PR tree-optimization/79408
28207 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
28208 constant, but SSA_NAME with a known integer range, use the minimum
28209 of that range instead of op1 to determine if modulo can be replaced
28210 with its first operand.
28212 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28214 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
28216 2017-02-08 Richard Biener <rguenther@suse.de>
28218 PR tree-optimization/71824
28219 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
28220 Check all loops contained in the merged region.
28222 2017-02-07 Andrew Pinski <apinski@cavium.com>
28224 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
28226 2017-02-07 Andrew Pinski <apinski@cavium.com>
28228 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
28229 (thunderxt88): Likewise.
28230 (thunderxt81): Disable LSE and change v8.1 to v8.
28231 (thunderxt83): Likewise.
28233 2017-02-07 Jakub Jelinek <jakub@redhat.com>
28234 Richard Biener <rguenther@suse.de>
28236 PR middle-end/79399
28237 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
28238 type from int to size_t.
28239 * ira-costs.c (struct_costs_size): Change type from int to size_t.
28241 2017-02-07 Jakub Jelinek <jakub@redhat.com>
28243 PR rtl-optimization/79386
28244 * cprop.c (bypass_conditional_jumps): Initialize
28245 bypass_last_basic_block already before splitting bbs after
28246 unconditional traps...
28247 (bypass_conditional_jumps): ... rather than here.
28250 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
28251 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
28252 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
28253 fix -masm=intel patterns.
28255 2017-02-07 Richard Biener <rguenther@suse.de>
28257 PR tree-optimization/79256
28258 PR middle-end/79278
28259 * builtins.c (get_object_alignment_2): Use min_align_of_type
28260 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
28261 and ADJUST_FIELD_ALIGN.
28263 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
28265 * doc/tm.texi: Regenerate.
28266 * stor-layout.c (layout_decl): Adjust.
28267 (update_alignment_for_field): Likewise.
28268 (place_field): Likewise.
28269 (min_align_of_type): Likewise.
28270 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
28271 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
28272 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
28273 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
28274 * config/frv/frv.c (frv_adjust_field_align): Likewise.
28275 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
28276 * config/i386/i386.c (x86_field_alignment): Likewise.
28277 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
28278 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
28279 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
28280 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
28281 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
28282 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
28286 2017-01-30 Richard Biener <rguenther@suse.de>
28288 PR tree-optimization/79256
28289 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
28290 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
28293 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
28295 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
28296 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
28297 builtins to SImode and emit a zero-extend, if necessary.
28299 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
28301 * docs/invoke.texi (RISC-V Options): Alphabetize.
28303 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
28305 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
28308 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
28310 * config/riscv/riscv.c: New file.
28311 * common/config/riscv/riscv-common.c: Likewise.
28312 * config.gcc: Likewise.
28313 * config/riscv/constraints.md: Likewise.
28314 * config/riscv/elf.h: Likewise.
28315 * config/riscv/generic.md: Likewise.
28316 * config/riscv/linux.h: Likewise.
28317 * config/riscv/multilib-generator: Likewise.
28318 * config/riscv/peephole.md: Likewise.
28319 * config/riscv/pic.md: Likewise.
28320 * config/riscv/predicates.md: Likewise.
28321 * config/riscv/riscv-builtins.c: Likewise.
28322 * config/riscv/riscv-c.c: Likewise.
28323 * config/riscv/riscv-ftypes.def: Likewise.
28324 * config/riscv/riscv-modes.def: Likewise.
28325 * config/riscv/riscv-opts.h: Likewise.
28326 * config/riscv/riscv-protos.h: Likewise.
28327 * config/riscv/riscv.h: Likewise.
28328 * config/riscv/riscv.md: Likewise.
28329 * config/riscv/riscv.opt: Likewise.
28330 * config/riscv/sync.md: Likewise.
28331 * config/riscv/t-elf-multilib: Likewise.
28332 * config/riscv/t-linux: Likewise.
28333 * config/riscv/t-linux-multilib: Likewise.
28334 * config/riscv/t-riscv: Likewise.
28335 * configure.ac: Likewise.
28336 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
28337 Waterman as RISC-V maintainers.
28338 * doc/install.texi: Add RISC-V entries.
28339 * doc/invoke.texi: Add RISC-V options section.
28340 * doc/md.texi: Add RISC-V constraints section.
28341 * configure: Regenerated.
28343 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
28346 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
28347 false values to be constant vectors with all 0 or all 1 bits set.
28348 (vcondu<mode><mode>): Likewise.
28349 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
28351 (fpmask_comparison_operator): Update comment.
28352 (vecint_comparison_operator): New predicate.
28353 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
28354 vector conditionals when the true and false values are constant
28355 vectors with all 0 bits or all 1 bits set.
28357 2017-02-06 Martin Sebor <msebor@redhat.com>
28359 PR tree-optimization/79376
28360 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
28362 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
28364 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
28365 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
28366 to simplify split condition.
28368 2017-02-06 Jakub Jelinek <jakub@redhat.com>
28370 * omp-expand.c (oxpand_omp_atomic_fetch_op,
28371 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
28374 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
28376 PR rtl-optimization/68664
28377 * target.def (can_speculate_insn): New hook.
28378 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
28379 * doc/tm.texi: Regenerate.
28380 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
28381 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
28382 (rs6000_sched_can_speculate_insn): New function.
28384 2017-02-06 Jakub Jelinek <jakub@redhat.com>
28386 PR tree-optimization/79284
28387 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
28388 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
28389 vectorizable_mask_load_store, vectorizable_operation,
28390 vect_is_simple_cond, get_same_sized_vectype): Use it instead
28391 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
28392 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
28393 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
28394 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
28395 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
28396 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
28397 is_gimple_assign (stmt). Replace another such test with
28398 is_gimple_assign (stmt).
28400 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
28403 * config/avr/avr.c (rtl-iter.h): Include it.
28404 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
28405 (avr_legitimate_combined_insn): ...and implementation.
28407 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
28409 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
28410 * config/s390/s390.c (s390_const_operand_ok)
28411 (s390_canonicalize_comparison, s390_extract_part)
28412 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
28413 (s390_contiguous_bitmask_p, s390_rtx_costs)
28414 (legitimize_pic_address): Likewise.
28415 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
28416 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
28417 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
28418 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
28419 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
28421 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
28423 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
28424 REGNO($0) == REGNO($1).
28426 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28428 * config/s390/linux.h(SIZE_TYPE): Add comment.
28430 2017-02-06 Julian Brown <julian@codesourcery.com>
28431 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28432 Virendra Pathak <virendra.pathak@broadcom.com>
28434 * config/aarch64/aarch64-cores.def: Change the scheduler
28436 * config/aarch64/aarch64.md: Include thunderx2t99.md.
28437 * config/aarch64/thunderx2t99.md: New file.
28439 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
28441 * doc/standards.texi (Go Language): Update link to language
28444 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
28446 * tree-eh.c (lower_resx): Sanitize profile.
28447 (cleanup_empty_eh_move_lp): Likewise.
28449 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
28452 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
28454 * cfgloopmanip.h (loop_version): Update prototype.
28455 * modulo-sched.c (sms_schedule): Update call of loop_version.
28456 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
28457 * tree-parloops.c (gen_parallel_loop): Likewise.
28458 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
28459 * tree-ssa-loop-split.c (split_loop): Likewise.
28460 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
28461 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
28463 2017-02-05 Martin Liska <mliska@suse.cz>
28466 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
28468 (print_operand_address): Initialize a struct to zero.
28470 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
28472 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
28473 garbage collector only in textual form.
28475 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
28477 * doc/extend.texi (x86 specific memory model extensions for
28478 transactional memory): Simplify a phrase.
28480 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
28483 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
28484 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
28485 (atomic_storedi_1): Likewise.
28487 2017-02-04 Jakub Jelinek <jakub@redhat.com>
28489 PR tree-optimization/79338
28490 * tree-parloops.c (gather_scalar_reductions): Don't call
28491 vect_analyze_loop_form for loop->inner before destroying loop's
28494 2017-02-03 Martin Sebor <msebor@redhat.com>
28496 PR tree-optimization/79327
28497 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
28498 when precision has resulted in leading zeros.
28499 (format_integer): Adjust the likely counter to assume an unknown
28500 argument that may be zero is non-zero.
28502 2017-02-03 Jason Merrill <jason@redhat.com>
28505 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
28506 avoid copying non-taken branch.
28508 2017-02-03 Jakub Jelinek <jakub@redhat.com>
28510 PR tree-optimization/79340
28511 * tree-vect-loop.c (vectorizable_reduction): Release
28512 vec_defs elements after safe_splicing them into other vectors.
28515 PR tree-optimization/79327
28516 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
28517 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
28519 (format_integer): Use wide_int_to_tree instead of build_int_cst
28520 + to_?hwi. If argmin is NULL, just set argmin and argmax to
28521 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
28522 of shortest and longest sequence.
28524 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
28526 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
28527 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
28529 2017-02-03 Walter Lee <walt@tilera.com>
28532 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
28533 after initial stackframe link reg save.
28534 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
28536 2017-02-03 Jakub Jelinek <jakub@redhat.com>
28539 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
28540 wu for stxssp alternative.
28542 2017-02-03 Martin Sebor <msebor@redhat.com>
28544 PR tree-optimization/79352
28545 * gimple-fold.c (get_range_strlen): Add argument.
28546 (get_range_strlen): Change return type to bool.
28547 (get_maxval_strlen): Pass in a dummy argument.
28548 * gimple-fold.h (get_range_strlen): Change return type to bool.
28549 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
28550 * tree.h (array_at_struct_end_p): Add argument.
28551 * tree.c (array_at_struct_end_p): Handle it.
28553 2017-02-03 Martin Liska <mliska@suse.cz>
28556 * multiple_target.c (create_dispatcher_calls): Redirect edge
28557 from a caller of a dispatcher.
28558 (expand_target_clones): Make the clones local.
28559 (ipa_target_clone): Do both target clones and resolvers.
28560 (ipa_dispatcher_calls): Remove the pass.
28561 (pass_dispatcher_calls::gate): Likewise.
28562 (make_pass_dispatcher_calls): Likewise.
28563 * passes.def (pass_target_clone): Put as very first IPA early
28566 2017-02-03 Martin Liska <mliska@suse.cz>
28568 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
28569 in case of a function with ifunc attribute.
28571 2017-02-03 Martin Liska <mliska@suse.cz>
28573 * cgraph.c (cgraph_node::dump): Dump function version info.
28574 * symtab.c (symtab_node::dump_base): Add missing new line.
28576 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
28578 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
28579 (ifcombine_ifandif): Use it.
28581 2017-02-03 Martin Liska <mliska@suse.cz>
28583 * doc/invoke.texi: Document default value for
28584 use-after-scope-direct-emission-threshold.
28586 2017-02-03 Martin Liska <mliska@suse.cz>
28588 PR tree-optimization/79339
28589 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
28590 (format_floating): Likewise.
28592 2017-02-03 Martin Liska <mliska@suse.cz>
28595 * ipa-prop.c (ipa_node_params_t::insert): Remove current
28597 (ipa_node_params_t::remove): Likewise.
28598 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
28599 initialization from removed ipa_node_params_t::insert.
28600 (ipa_node_params::~ipa_node_params): Move from removed
28601 ipa_node_params_t::release.
28602 * symbol-summary.h (symbol_summary::m_released): New member.
28603 Do not release a summary twice. Do not allow to call finalizer
28604 for types of a summary that live in GGC memory.
28606 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28608 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
28611 2017-02-02 Martin Sebor <msebor@redhat.com>
28613 PR middle-end/79275
28614 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
28615 (format_string): Tighten up the range of output for non-constant
28616 strings and correct the expected range for wide non-constant strings.
28618 2017-02-02 Martin Sebor <msebor@redhat.com>
28620 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
28622 PR middle-end/32003
28623 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
28625 (-fdump-tree-@var): Add to index and document how to come up
28626 with pass-specific option and dump file names.
28627 (-fdump-passes): Clarify where to look for output.
28629 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
28631 PR middle-end/77445
28632 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
28633 statistics of the analyzed path; allow threading for speed when
28634 any of BBs along the path are optimized for speed.
28636 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
28638 PR middle-end/78468
28639 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
28640 settings of the virtual registers.
28643 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
28645 * explow.c (get_dynamic_stack_size): Take known alignment of stack
28646 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
28649 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28651 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
28652 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
28654 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28656 * config/s390/s390.md: Add missing comments with the expanded
28658 * config/s390/vector.md: Likewise.
28659 * config/s390/vx-builtins.md: Likewise.
28661 2017-02-02 Jakub Jelinek <jakub@redhat.com>
28664 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
28665 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
28666 conditions on a single line.
28668 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28670 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
28671 __S390_VX__ to __VX__.
28673 2017-02-01 Andrew Pinski <apinski@cavium.com>
28675 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
28676 stmt_info to record_stmt_cost.
28677 (vect_get_known_peeling_cost): Pass stmt_info if known to
28679 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
28680 cpu_vector_cost field into
28681 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
28682 field into vec_int_stmt_cost and vec_fp_stmt_cost.
28683 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
28684 splitting of scalar_stmt_cost and vec_stmt_cost.
28685 (thunderx_vector_cost): Likewise.
28686 (cortexa57_vector_cost): LIkewise.
28687 (exynosm1_vector_cost): Likewise.
28688 (xgene1_vector_cost): Likewise.
28689 (thunderx2t99_vector_cost): Improve after the splitting of the two
28691 (aarch64_builtin_vectorization_cost): Update for the splitting of
28692 scalar_stmt_cost and vec_stmt_cost.
28694 2017-02-01 Torvald Riegel <triegel@redhat.com>
28695 Richard Henderson <rth@redhat.com>
28697 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
28698 conditional on existance of a fast atomic load.
28699 * optabs-query.c (can_atomic_load_p): New function.
28700 * optabs-query.h (can_atomic_load_p): Declare it.
28701 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
28702 no fast atomic load is available for the particular size of access.
28703 (expand_atomic_compare_and_swap): Likewise.
28704 (expand_atomic_load): Likewise.
28705 (expand_atomic_store): Likewise.
28706 (expand_atomic_fetch_op): Likewise.
28707 * testsuite/lib/target-supports.exp
28708 (check_effective_target_sync_int_128): Remove x86 because it provides
28709 no fast atomic load.
28710 (check_effective_target_sync_int_128_runtime): Likewise.
28712 2017-02-01 Richard Biener <rguenther@suse.de>
28714 * graphite.c: Include tree-vectorizer.h for find_loop_location.
28715 (graphite_transform_loops): Provide opt-info for optimized nests.
28716 * tree-parloop.c (parallelize_loops): Provide opt-info for
28717 parallelized loops.
28719 2017-02-01 Richard Biener <rguenther@suse.de>
28721 PR middle-end/79315
28722 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
28723 was not set before.
28725 2017-02-01 Richard Biener <rguenther@suse.de>
28727 PR tree-optimization/71824
28728 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
28729 Verify the loops are valid in the merged SESE region.
28730 (scop_detection::can_represent_loop_1): Check analyzing the
28731 evolution of the number of iterations in the region succeeds.
28733 2017-01-31 Ian Lance Taylor <iant@golang.org>
28735 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
28736 REG_ARGS_SIZE note to 32-bit push insns and call insn.
28738 2017-01-31 David Malcolm <dmalcolm@redhat.com>
28740 PR preprocessor/79210
28741 * input.c (get_substring_ranges_for_loc): Replace line_width
28742 assertion with error-handling.
28744 2017-01-31 Richard Biener <rguenther@suse.de>
28746 PR tree-optimization/77318
28747 * graphite-sese-to-poly.c (extract_affine): Fix assert.
28748 (create_pw_aff_from_tree): Take loop parameter.
28749 (add_condition_to_pbb): Pass loop of the condition to
28750 create_pw_aff_from_tree.
28752 2017-01-31 Jakub Jelinek <jakub@redhat.com>
28754 * config/s390/s390.c (s390_asan_shadow_offset): New function.
28755 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
28757 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
28761 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
28763 (convert_int_to_float128): Likewise.
28764 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
28765 (convert_int_to_float128): Likewise.
28766 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
28767 (UNSPEC_IEEE128_CONVERT): Likewise.
28768 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
28769 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
28770 Use local variables for IBM extended format.
28771 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
28772 (fix_trunc<mode>si2_fprs): Likewise.
28773 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
28774 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
28775 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
28776 to know that we can now have integers of all sizes in vector
28778 (fix<uns>_<mode>di2_hw): Likewise.
28779 (float<uns>_<mode>si2_hw): Likewise.
28780 (fix_<mode>si2_hw): Likewise.
28781 (fixuns_<mode>si2_hw): Likewise.
28782 (float<uns>_<mode>di2_hw): Likewise.
28783 (float_<mode>di2_hw): Likewise.
28784 (float_<mode>si2_hw): Likewise.
28785 (floatuns_<mode>di2_hw): Likewise.
28786 (floatuns_<mode>si2_hw): Likewise.
28787 (xscvqp<su>wz_<mode>): Delete, no longer used.
28788 (xscvqp<su>dz_<mode>): Likewise.
28789 (xscv<su>dqp_<mode>): Likewise.
28790 (ieee128_mfvsrd_64bit): Likewise.
28791 (ieee128_mfvsrd_32bit): Likewise.
28792 (ieee128_mfvsrwz): Likewise.
28793 (ieee128_mtvsrw): Likewise.
28794 (ieee128_mtvsrd_64bit): Likewise.
28795 (ieee128_mtvsrd_32bit): Likewise.
28797 2017-01-31 Martin Liska <mliska@suse.cz>
28800 * ipa-prop.c (ipa_free_all_node_params): Call release method
28801 instead of ~sumbol_summary to not to trigger double times
28804 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
28806 PR tree-optimization/71691
28807 * bitmap.h (class auto_bitmap): New.
28808 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
28809 is_maybe_undefined instead of ssa_undefined_value_p.
28811 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28813 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
28814 __S390_ARCH_LEVEL__ to __ARCH__.
28816 2017-01-31 Jakub Jelinek <jakub@redhat.com>
28818 PR tree-optimization/79267
28819 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
28820 if should_remove_lhs_p is true.
28822 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
28825 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
28826 (add_alignment_attribute): New.
28827 (base_type_die): Add alignment attribute.
28828 (subrange_type_die): Likewise.
28829 (modified_type_die): Likewise.
28830 (gen_array_type_die): Likewise.
28831 (gen_descr_array_type_die: Likewise.
28832 (gen_enumeration_type_die): Likewise.
28833 (gen_subprogram_die): Likewise.
28834 (gen_variable_die): Likewise.
28835 (gen_field_die): Likewise.
28836 (gen_ptr_to_mbr_type_die): Likewise.
28837 (gen_struct_or_union_type_die): Likewise.
28838 (gen_subroutine_type_die): Likewise.
28839 (gen_typedef_die): Likewise.
28840 (base_type_cmp): Compare alignment attribute.
28842 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28845 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
28846 (setb_unsigned) New pattern for setb with CCUNS.
28847 * config/rs6000/rs6000.c (expand_block_compare): Use a different
28848 subfc./subfe sequence to avoid overflow problems. Generate a
28849 shorter sequence with cmpld/setb for power9.
28850 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
28851 for generating subfc. instruction.
28852 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
28853 now uses this instruction.
28855 2017-01-30 Ian Lance Taylor <iant@google.com>
28858 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
28859 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
28861 2017-01-30 Martin Sebor <msebor@redhat.com>
28863 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
28864 Move constant to the right of a relational operator.
28865 (get_mpfr_format_length, format_character, format_string): Ditto.
28866 (should_warn_p, maybe_warn): Same.
28868 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
28870 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
28873 * asan.c (get_translation_unit_decl): Remove function.
28874 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
28876 2017-01-30 Martin Liska <mliska@suse.cz>
28878 PR gcov-profile/79259
28879 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
28880 -fprofile-generate.
28882 2017-01-30 Martin Liska <mliska@suse.cz>
28885 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
28886 Initialize variables with NULL value.
28888 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
28891 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
28893 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
28895 2017-01-30 Richard Biener <rguenther@suse.de>
28897 PR tree-optimization/79276
28898 * tree-vrp.c (process_assert_insertions): Properly adjust common
28899 when removing a duplicate.
28901 * gcc.dg/torture/pr79276.c: New testcase.
28903 2017-01-30 Richard Biener <rguenther@suse.de>
28905 PR tree-optimization/79256
28906 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
28907 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
28909 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
28911 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
28914 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
28915 ("*r<noxa>sbg_<mode>_sll_bitmask")
28916 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
28917 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
28918 Use contiguous_bitmask_nowrap_operand.
28920 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28923 * config/rs6000/altivec.h (vec_xl): Revise #define.
28924 (vec_xst): Likewise.
28926 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
28928 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
28930 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
28932 PR rtl-optimization/79194
28933 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
28934 traps before call to bypass_conditional_jumps.
28936 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
28938 PR tree-optimization/71374
28939 * lra-constraints.c (check_conflict_input_operands): New.
28940 (match_reload): Use it.
28942 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
28945 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
28946 account to calculate conflict_set.
28948 2017-01-27 Bin Cheng <bin.cheng@arm.com>
28950 PR rtl-optimization/78559
28951 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
28952 other_insn in combine.
28954 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
28956 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
28957 uint16_type_node for BT_UINT16.
28959 2017-01-27 David Malcolm <dmalcolm@redhat.com>
28961 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
28962 "RTL Tests" to menu.
28963 (GIMPLE Tests): New node.
28964 (RTL Tests): New node.
28966 2017-01-27 Richard Biener <rguenther@suse.de>
28968 PR tree-optimization/79245
28969 * tree-loop-distribution.c (distribute_loop): Apply cost
28970 modeling also to detected patterns.
28972 2017-01-27 Richard Biener <rguenther@suse.de>
28974 PR tree-optimization/71433
28975 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
28976 (compare_assert_loc): New function.
28977 (process_assert_insertions): Sort and optimize assert locations
28978 to remove duplicates and push down identical assertions on
28979 edges to their destination block.
28981 2017-01-27 Richard Biener <rguenther@suse.de>
28983 PR tree-optimization/79244
28984 * tree-vrp.c (remove_range_assertions): Forcefully propagate
28985 out SSA names even if abnormal.
28987 2017-01-27 Jakub Jelinek <jakub@redhat.com>
28989 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
28990 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
28991 instead of MPFR_RNDN.
28993 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
28996 * arm.c (arm_option_override): Don't call build_target_option_node
28997 until after doing all option overrides.
28998 (arm_valid_target_attribute_tree): Likewise.
29000 2017-01-27 Martin Liska <mliska@suse.cz>
29002 * doc/invoke.texi (-fprofile-arcs): Document profiling support
29003 for {cd}tors and C++ {cd}tors.
29005 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
29007 * config/s390/s390.md ("*setmem_long_and")
29008 ("*setmem_long_and_31z"): Use zero_extend instead of and.
29010 2017-01-26 Martin Sebor <msebor@redhat.com>
29012 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
29015 2017-01-26 Martin Sebor <msebor@redhat.com>
29017 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
29018 HAVE_DFmode before using XFmode or DFmode.
29019 (parse_directive): Avoid using the z length modifier to avoid
29020 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
29022 PR middle-end/78703
29023 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
29024 to accept adjustment as an array.
29025 (get_int_range): New function.
29026 (struct directive): Make width and prec arrays.
29027 (directive::set_width, directive::set_precision): Call get_int_range.
29028 (format_integer, format_floating): Handle width and precision ranges.
29029 (format_string, parse_directive): Same.
29031 2017-01-26 Jakub Jelinek <jakub@redhat.com>
29034 * dwarf2out.c (generate_skeleton_bottom_up): For children with
29035 comdat_type_p set, just clone them, but keep the children in the
29039 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
29040 which have direct callers with -fvar-tracking-assignments enabled
29042 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
29043 inside of type units.
29045 2017-01-26 Martin Sebor <msebor@redhat.com>
29047 PR middle-end/78703
29048 * gimple-ssa-sprintf.c (struct result_range): Add likely and
29050 (struct format_result): Replace number_chars, number_chars_min,
29051 and number_chars_max with a single member of struct result_range.
29053 (format_result::operator+=): Adjust.
29054 (struct fmtresult): Remove bounded. Handle likely and unlikely
29056 (fmtresult::adjust_for_width_or_precision): New function.
29057 (fmtresult:type_max_digits): New function.
29058 (bytes_remaining): Handle likely and unlikely counters.
29059 (min_bytes_remaining): Remove.
29060 (format_percent): Simplify.
29061 (format_integer, format_floating): Set likely and unlikely counters.
29062 (get_string_length, format_character, format_string): Same.
29063 (format_plain, should_warn_p): New function.
29064 (maybe_warn): Call should_warn_p. Update diagnostic messages
29065 and handle those for all directives, including plain strings.
29066 (format_directive): Handle likely and unlikely counters.
29067 Remove unnecessary quoting from diagnostics. Add an informational
29069 (add_bytes): Remove.
29070 (pass_sprintf_length::compute_format_length): Simplify.
29071 (try_substitute_return_value): Handle likely and unlikely counters.
29073 2017-01-26 Carl Love <cel@us.ibm.com>
29075 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
29076 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
29078 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
29081 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
29082 endianess for subregs into account.
29083 * lra-constraints.c (lra_constraints): Do risky transformations
29084 always on the first iteration.
29085 * lra-lives.c (check_pseudos_live_through_calls): Add arg
29086 last_call_used_reg_set.
29087 (process_bb_lives): Define and use last_call_used_reg_set.
29088 * lra.c (lra): Always continue after lra_constraints on the first
29091 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
29093 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
29095 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
29097 2017-01-26 Jakub Jelinek <jakub@redhat.com>
29099 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
29100 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
29101 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
29102 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
29103 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
29104 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
29105 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
29106 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
29107 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
29109 2017-01-26 Marek Polacek <polacek@redhat.com>
29112 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
29113 for the third operand.
29115 2017-01-26 Jakub Jelinek <jakub@redhat.com>
29117 PR middle-end/79236
29118 * omp-low.c (struct omp_context): Add simt_stmt field.
29119 (scan_omp_for): Return omp_context *.
29120 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
29121 context to the _simt_ SIMD stmt.
29122 (lower_omp_for): For combined SIMD with sibling _simt_
29123 SIMD, make sure to use the same decls in _looptemp_
29124 clauses as in the sibling.
29126 2017-01-26 David Sherwood <david.sherwood@arm.com>
29128 PR middle-end/79212
29129 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
29132 2017-01-26 Jakub Jelinek <jakub@redhat.com>
29135 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
29136 emit fld b; fld a; if possible.
29138 * brig-builtins.def: Update copyright years.
29139 * config/arm/arm_acle_builtins.def: Update copyright years.
29141 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
29144 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
29145 constraint instead of o for the stxsd instruction.
29147 2017-01-25 Carl Love <cel@us.ibm.com>
29149 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
29150 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
29152 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
29154 * doc/invoke.texi (C++ Dialect Options): Fix typo.
29156 2017-01-25 Richard Biener <rguenther@suse.de>
29158 PR tree-optimization/69264
29159 * target.def (vector_alignment_reachable): Improve documentation.
29160 * doc/tm.texi: Regenerate.
29161 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
29163 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
29164 earlier changes with respect to TYPE_USER_ALIGN.
29165 (vector_alignment_reachable_p): Likewise. Improve dumping.
29167 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29170 * config/arm/arm.md (xordi3): Force constant operand into a register
29173 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29175 * doc/invoke.texi (-fstore-merging): Correct default optimization
29176 levels at which it is enabled.
29177 (-O): Move -fstore-merging from list to...
29180 2017-01-25 Richard Biener <rguenther@suse.de>
29183 * omp-expand.c: Include debug.h.
29184 (expand_omp_taskreg): Make sure to generate early debug before
29185 outlining anything from a function.
29186 (expand_omp_target): Likewise.
29187 (grid_expand_target_grid_body): Likewise.
29189 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
29192 * asan.c (get_translation_unit_decl): New function.
29193 (asan_add_global): Extract modules file name from globals
29194 TRANSLATION_UNIT_DECL name.
29196 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
29199 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
29200 for long calls with APCS frame and VFP.
29202 2017-01-24 David Malcolm <dmalcolm@redhat.com>
29204 * cfg.c (original_copy_tables_initialized_p): New function.
29205 * cfg.h (original_copy_tables_initialized_p): New decl.
29206 * cfgrtl.c (relink_block_chain): Guard the call to
29207 free_original_copy_tables with a call to
29208 original_copy_tables_initialized_p.
29209 * cgraph.h (symtab_node::native_rtl_p): New decl.
29210 * cgraphunit.c (symtab_node::native_rtl_p): New function.
29211 (symtab_node::needed_p): Don't assert for early assembly output
29212 for __RTL functions.
29213 (cgraph_node::finalize_function): Set "force_output" for __RTL
29215 (cgraph_node::analyze): Bail out early for __RTL functions.
29216 (analyze_functions): Update assertion to support __RTL functions.
29217 (cgraph_node::expand): Bail out early for __RTL functions.
29218 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
29220 * function.h (struct function): Update comment for field
29222 * gimple-expr.c: Include "tree-pass.h".
29223 (gimple_has_body_p): Return false for __RTL functions.
29224 * Makefile.in (OBJS): Add run-rtl-passes.o.
29225 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
29227 (gcc::pass_manager::get_clean_slate): New accessor.
29228 * passes.c: Include "insn-addr.h".
29229 (should_skip_pass_p): Add logging. Update logic for running
29230 "expand" to be compatible with both __GIMPLE and __RTL. Guard
29231 property-provider override so it is only done for gimple passes.
29233 (skip_pass): New function.
29234 (execute_one_pass): Call skip_pass when skipping passes.
29235 * read-md.c (md_reader::read_char): Support filtering
29236 the input to a subset of line numbers.
29237 (md_reader::md_reader): Initialize fields
29238 m_first_line and m_last_line.
29239 (md_reader::read_file_fragment): New function.
29240 * read-md.h (md_reader::read_file_fragment): New decl.
29241 (md_reader::m_first_line): New field.
29242 (md_reader::m_last_line): New field.
29243 * read-rtl-function.c (function_reader::create_function): Only
29244 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
29245 curr_properties. Set DECL_INITIAL to a dummy block.
29246 (read_rtl_function_body_from_file_range): New function.
29247 * read-rtl-function.h (read_rtl_function_body_from_file_range):
29249 * run-rtl-passes.c: New file.
29250 * run-rtl-passes.h: New file.
29252 2017-01-24 Jeff Law <law@redhat.com>
29254 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
29257 2017-01-24 Bin Cheng <bin.cheng@arm.com>
29259 PR tree-optimization/79159
29260 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
29261 (record_nonwrapping_iv): Improve boundary using above function if no
29262 value range information.
29264 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
29265 Martin Jambor <mjambor@suse.cz>
29267 * brig-builtins.def: New file.
29268 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
29269 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
29270 (DEF_HSAIL_SAT_BUILTIN): Likewise.
29271 (DEF_HSAIL_INTR_BUILTIN): Likewise.
29272 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
29273 * builtin-types.def (BT_INT8): New.
29274 (BT_INT16): Likewise.
29275 (BT_UINT8): Likewise.
29276 (BT_UINT16): Likewise.
29277 (BT_FN_ULONG): Likewise.
29278 (BT_FN_UINT_INT): Likewise.
29279 (BT_FN_UINT_ULONG): Likewise.
29280 (BT_FN_UINT_LONG): Likewise.
29281 (BT_FN_UINT_PTR): Likewise.
29282 (BT_FN_ULONG_PTR): Likewise.
29283 (BT_FN_INT8_FLOAT): Likewise.
29284 (BT_FN_INT16_FLOAT): Likewise.
29285 (BT_FN_UINT32_FLOAT): Likewise.
29286 (BT_FN_UINT16_FLOAT): Likewise.
29287 (BT_FN_UINT8_FLOAT): Likewise.
29288 (BT_FN_UINT64_FLOAT): Likewise.
29289 (BT_FN_UINT16_UINT32): Likewise.
29290 (BT_FN_UINT32_UINT16): Likewise.
29291 (BT_FN_UINT16_UINT16_UINT16): Likewise.
29292 (BT_FN_INT_PTR_INT): Likewise.
29293 (BT_FN_UINT_PTR_UINT): Likewise.
29294 (BT_FN_LONG_PTR_LONG): Likewise.
29295 (BT_FN_ULONG_PTR_ULONG): Likewise.
29296 (BT_FN_VOID_UINT64_UINT64): Likewise.
29297 (BT_FN_UINT8_UINT8_UINT8): Likewise.
29298 (BT_FN_INT8_INT8_INT8): Likewise.
29299 (BT_FN_INT16_INT16_INT16): Likewise.
29300 (BT_FN_INT_INT_INT): Likewise.
29301 (BT_FN_UINT_FLOAT_UINT): Likewise.
29302 (BT_FN_FLOAT_UINT_UINT): Likewise.
29303 (BT_FN_ULONG_UINT_UINT): Likewise.
29304 (BT_FN_ULONG_UINT_PTR): Likewise.
29305 (BT_FN_ULONG_ULONG_ULONG): Likewise.
29306 (BT_FN_UINT_UINT_UINT): Likewise.
29307 (BT_FN_VOID_UINT_PTR): Likewise.
29308 (BT_FN_UINT_UINT_PTR: Likewise.
29309 (BT_FN_UINT32_UINT64_PTR): Likewise.
29310 (BT_FN_INT_INT_UINT_UINT): Likewise.
29311 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
29312 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
29313 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
29314 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
29315 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
29316 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
29317 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
29318 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
29319 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
29320 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
29321 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
29322 * doc/frontends.texi: List BRIG FE.
29323 * doc/install.texi (Testing): Add BRIG tesring requirements.
29324 * doc/invoke.texi (Overall Options): Mention BRIG.
29325 * doc/standards.texi (Standards): Doucment BRIG HSA version.
29327 2017-01-24 Richard Biener <rguenther@suse.de>
29329 PR translation/79208
29330 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
29332 2017-01-24 Martin Jambor <mjambor@suse.cz>
29335 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
29336 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
29337 and known_contexts.
29339 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
29341 PR middle-end/79123
29342 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
29343 casts from signed to unsigned really don't have a range.
29345 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
29347 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
29348 GMP_RNDx for compatiblity.
29350 2017-01-24 Martin Liska <mliska@suse.cz>
29353 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
29354 that would prevent us to call alloca with -1 as argument.
29356 2017-01-24 Jakub Jelinek <jakub@redhat.com>
29358 * dwarf2out.c (output_compilation_unit_header, output_file_names):
29359 Avoid -Wformat-security warning.
29361 2017-01-23 Andrew Pinski <apinski@cavium.com>
29363 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
29366 2017-01-23 Martin Sebor <msebor@redhat.com>
29368 PR middle-end/78703
29369 * gimple-ssa-sprintf.c (warn_level): New global.
29370 (format_integer): Use it here and throughout the rest of the file.
29371 Use the same switch to compute sign as base.
29372 (maybe_warn): New function.
29373 (format_directive): Factor out warnings into maybe_warn.
29374 Add debugging output. Use warn_level.
29375 (add_bytes): Use warn_level.
29376 (pass_sprintf_length::compute_format_length): Add debugging output.
29377 (try_substitute_return_value): Same.
29378 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
29380 PR middle-end/78703
29381 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
29382 (struct fmtresult, format_integer, format_floating): Adjust.
29383 (fmtresult::fmtresult): Set max correctly in two argument ctor.
29384 (get_string_length, format_string,format_directive): Same.
29385 (pass_sprintf_length::compute_format_length): Same.
29386 (try_substitute_return_value): Simplify slightly.
29388 PR middle-end/78703
29389 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
29390 (fmtresult::operator+=): Outlined.
29391 (struct fmtresult): Add ctors.
29392 (struct conversion_spec): Rename...
29393 (struct directive): ...to this. Add and remove data members.
29394 (directive::set_width, directive::set_precision): New functions.
29395 (format_percent): Use fmtresult ctor.
29396 (get_width_and_precision): Remove.
29397 (format_integer): Make naming changes. Avoid computing width and
29399 (format_floating): Same. Adjust indentation.
29400 (format_character, format_none): New functions.
29401 (format_string): Moved character handling to format_character.
29402 (format_directive): Remove arguments, change return type.
29403 (parse_directive): New function.
29404 (pass_sprintf_length::compute_format_length): Move directive
29405 parsing to parse_directive.
29407 2017-01-23 Jakub Jelinek <jakub@redhat.com>
29409 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
29410 (assign_assembler_name_if_needed): ... this.
29411 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
29412 (assign_assembler_name_if_needed): ... this.
29413 (free_lang_data_in_cgraph): Adjust callers.
29414 * cgraphunit.c (cgraph_node::analyze): Likewise.
29415 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
29418 2017-01-23 Richard Biener <rguenther@suse.de>
29420 PR tree-optimization/79088
29421 PR tree-optimization/79188
29422 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
29423 resetting loop bounds after last path deletion. Reset loop
29424 bounds of the target loop, make code match the comments.
29425 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
29426 Make sure loops need no fixups.
29428 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
29430 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
29431 exponent support with double type for first argument.
29432 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
29433 type returned by __builtin_vec_extract_sig,
29434 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
29435 functions from "vector int" to "vector unsigned int" or from
29436 "vector long long int" to "vector unsigned long long int".
29437 Changed type returned by __builtin_vec_extract_exp,
29438 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
29439 functions from "vector int" to "vector unsigned int" or from
29440 "vector long long int" to "vector unsigned long long int".
29441 Changed return type of __builtin_vec_test_data_class,
29442 __builtin_vec_test_data_class_sp, and
29443 __builtin_vec_test_data_class_dp from "vector int" to
29444 "vector bool int" or from "vector long long int" to "vector bool
29445 long long int" and changed second argument type from "unsigned
29446 int" to "int". Added new overloaded function forms "vector float
29447 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
29448 "vector float __builtin_vec_insert_exp_sp (vector float, vector
29449 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
29450 double, vector unsigned long long int)" and "vector double
29451 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
29452 long int)". Changed return type of
29453 __builtin_scalar_test_data_class and
29454 __builtin_scalar_test_data_class_sp and
29455 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
29456 int" and changed second argument from "unsigned int" to "int".
29457 Changed type returned by __builtin_scalar_test_neg,
29458 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
29459 from "int" to "bool int". Added new overloaded function form
29460 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
29461 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
29462 exponent double-precision with floating point first argument.
29463 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
29464 documentation of scalar_test_data_class, scalar_test_neg,
29465 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
29466 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
29467 vec_test_data_class built-in functions to reflect refinements in
29468 their type signatures.
29470 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
29472 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
29474 (aarch64_elf_asm_destructor): Likewise.
29476 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
29478 PR rtl-optimization/78634
29479 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
29480 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
29481 * ifcvt.c (noce_try_cmove): Add missing cost check.
29483 PR rtl-optimization/71724
29484 * combine.c (if_then_else_cond): Look for situations where it is
29485 beneficial to undo the work of one of the recursive calls.
29487 2017-01-23 Bin Cheng <bin.cheng@arm.com>
29489 PR tree-optimization/70754
29490 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
29491 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
29492 combined stmt before it if not NULL.
29493 (combine_chains): Process refs reversely and compute dominance point
29496 2017-01-23 Martin Liska <mliska@suse.cz>
29498 PR tree-optimization/79196
29499 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
29500 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
29502 (strlen_optimize_stmt): Call the renamed function.
29504 2017-01-23 Michael Matz <matz@suse.de>
29506 PR tree-optimization/78384
29507 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
29509 2017-01-23 Richard Biener <rguenther@suse.de>
29511 PR tree-optimization/79186
29512 * tree-vrp.c (register_new_assert_for): Make sure we've seen
29513 both incoming edges before moving an assert.
29515 2017-01-23 Martin Jambor <mjambor@suse.cz>
29517 * ipa-prop.c (load_from_param_1): Removed.
29518 (load_from_unmodified_param): Bits from load_from_param_1 put back
29520 (load_from_param): Removed.
29521 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
29522 with stmt. Reverted back to use of load_from_unmodified_param.
29524 2017-01-23 Martin Jambor <mjambor@suse.cz>
29527 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
29528 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
29529 field a pointer to garbage collected vector, mark lattices and
29530 ipcp_orig_node with GTY((skip)).
29531 (ipa_get_param_count): Adjust to descriptors being a pointer.
29532 (ipa_get_param): Likewise.
29533 (ipa_get_type): Likewise.
29534 (ipa_get_param_move_cost): Likewise.
29535 (ipa_set_param_used): Likewise.
29536 (ipa_get_controlled_uses): Likewise.
29537 (ipa_set_controlled_uses): Likewise.
29538 (ipa_is_param_used): Likewise.
29539 (ipa_node_params_t): Move into garbage collector. New methods insert
29541 (ipa_node_params_sum): Annotate wth GTY(()).
29542 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
29544 (ipa_load_from_parm_agg): Adjust declaration.
29545 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
29546 * ipa-profile.c (ipa_profile): Likewise.
29547 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
29548 (ipa_populate_param_decls): Make descriptors parameter garbage
29550 (ipa_dump_param): Adjust to descriptors being a pointer.
29551 (ipa_alloc_node_params): Likewise.
29552 (ipa_initialize_node_params): Likewise.
29553 (load_from_param_1): Make descriptors parameter garbage collected.
29554 (load_from_unmodified_param): Likewise.
29555 (load_from_param): Likewise.
29556 (ipa_load_from_parm_agg): Likewise.
29557 (ipa_node_params::~ipa_node_params): Removed.
29558 (ipa_free_all_node_params): Remove call to delete operator.
29559 (ipa_node_params_t::insert): New.
29560 (ipa_node_params_t::remove): Likewise.
29561 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
29562 copy known_csts and known_contexts vectors.
29563 (ipa_read_node_info): Adjust to descriptors being a pointer.
29564 (ipcp_modif_dom_walker): Make m_descriptors field garbage
29566 (ipcp_transform_function): Make descriptors variable garbage
29569 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
29571 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
29572 * config/i386/avx512dqintrin.h: Ditto.
29573 * config/i386/avx512fintrin.h: Ditto.
29574 * config/i386/i386.c: Handle new builtins.
29575 * config/i386/i386-builtin.def: Add new builtins.
29576 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
29577 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
29579 2017-01-23 Jakub Jelinek <jakub@redhat.com>
29580 Martin Liska <mliska@suse.cz>
29582 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
29583 * asan.c (asan_expand_poison_ifn): Support stores and use
29584 appropriate ASAN report function.
29585 * internal-fn.c (expand_ASAN_POISON_USE): New function.
29586 * internal-fn.def (ASAN_POISON_USE): Declare.
29587 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
29588 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
29589 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
29590 ASAN_POISON calls w/o LHS.
29591 * tree-ssa.c (execute_update_addresses_taken): Create clobber
29592 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
29593 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
29594 * gimplify.c (asan_poison_variables): Add attribute
29595 use_after_scope_memory to variables that really needs to live
29597 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
29598 having the attribute.
29600 2017-01-23 Martin Liska <mliska@suse.cz>
29602 * asan.c (create_asan_shadow_var): New function.
29603 (asan_expand_poison_ifn): Likewise.
29604 * asan.h (asan_expand_poison_ifn): New declaration.
29605 * internal-fn.c (expand_ASAN_POISON): Likewise.
29606 * internal-fn.def (ASAN_POISON): New builtin.
29607 * sanopt.c (pass_sanopt::execute): Expand
29608 asan_expand_poison_ifn.
29609 * tree-inline.c (copy_decl_for_dup_finish): Make function
29611 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
29612 * tree-ssa.c (is_asan_mark_p): New function.
29613 (execute_update_addresses_taken): Rewrite local variables
29614 (identified just by use-after-scope as addressable) into SSA.
29616 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
29618 * doc/install.texi (Specific): opensource.apple.com uses https
29619 now. Remove trailing slash.
29621 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
29623 * README.Portability: Remove note on an Irix compatibility issue.
29625 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
29627 * gcov.c (INCLUDE_ALGORITHM): Define.
29628 (INCLUDE_VECTOR): Define.
29629 No longer include <vector> and <algorithm> directly.
29631 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
29633 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
29635 * doc/invoke.texi (Code Gen Options): Ditto.
29637 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
29640 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
29642 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
29644 rtl-optimization/79125
29645 * cprop.c (local_cprop_pass): Handle cases where we make an
29646 unconditional trap.
29648 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
29652 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
29653 read from, for big endian.
29655 2017-01-20 Jiong Wang <jiong.wang@arm.com>
29657 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
29658 register pauth builtins for LP64 only.
29660 2017-01-20 Marek Polacek <polacek@redhat.com>
29663 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
29666 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
29668 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
29670 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
29671 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
29672 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
29674 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29677 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
29678 in big-endian mode when they are not a single duplicated value.
29680 2017-01-20 Richard Biener <rguenther@suse.de>
29682 * BASE-VER: Bump to 7.0.1.
29684 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
29686 * omp-low.c (omplow_simd_context): New struct. Use it...
29687 (lower_rec_simd_input_clauses): ...here and...
29688 (lower_rec_input_clauses): ...here to hold common data. Adjust all
29689 references to idx, lane, max_vf, is_simt.
29691 2017-01-20 Graham Markall <graham.markall@embecosm.com>
29693 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
29696 2017-01-20 Martin Jambor <mjambor@suse.cz>
29698 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
29699 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
29701 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
29702 (GTFILES): Rename hsa.c to hsa-common.c.
29703 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
29704 * hsa-dump.c: Likewise.
29705 * hsa-gen.c: Likewise.
29706 * hsa-regalloc.c: Likewise.
29707 * ipa-hsa.c: Likewise.
29708 * omp-expand.c: Likewise.
29709 * omp-low.c: Likewise.
29710 * toplev.c: Likewise.
29712 2017-01-20 Marek Polacek <polacek@redhat.com>
29715 * doc/invoke.texi: Document -Wduplicated-branches.
29716 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
29717 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
29718 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
29719 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
29720 return 0 only when not OEP_LEXICOGRAPHIC.
29721 (fold_build_cleanup_point_expr): Use the expression
29722 location when building CLEANUP_POINT_EXPR.
29723 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
29724 * tree.c (add_expr): Handle error_mark_node.
29726 2017-01-20 Martin Liska <mliska@suse.cz>
29729 * tree-profile.c (init_ic_make_global_vars): Do not call
29731 (gimple_init_gcov_profiler): Likewise.
29733 2017-01-20 Martin Liska <mliska@suse.cz>
29736 * cgraph.h (maybe_create_reference): Remove argument and
29738 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
29740 * ipa-cp.c (create_specialized_node): Likewise.
29741 * symtab.c (symtab_node::maybe_create_reference): Handle
29742 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
29744 2017-01-20 Martin Liska <mliska@suse.cz>
29746 * read-rtl-function.c (function_reader::create_function): Use
29747 build_decl instread of build_decl_stat.
29749 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
29751 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
29752 * config/i386/avx512dqintrin.h: Ditto.
29753 * config/i386/avx512fintrin.h: Ditto.
29754 * config/i386/i386-builtin-types.def: Add new types.
29755 * config/i386/i386.c: Handle new types.
29756 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
29757 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
29758 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
29759 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
29760 (__builtin_ia32_kshiftridi): New.
29761 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
29763 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
29767 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
29768 define to rs6000_init_stack_protect_guard.
29769 (rs6000_init_stack_protect_guard): New function.
29771 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
29772 Yunqiang Su <yunqiang.su@imgtec.com>
29774 * config.gcc (supported_defaults): Add madd4.
29775 (with_madd4): Add validation.
29776 (all_defaults): Add madd4.
29777 * config/mips/mips.opt (mmadd4): New option.
29778 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
29780 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
29782 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
29783 (ISA_HAS_FUSED_MADD4): Likewise.
29784 * doc/invoke.texi (-mmadd4): Document the new option.
29785 * doc/install.texi (--with-madd4): Document the new option.
29787 2017-01-19 Jiong Wang <jiong.wang@arm.com>
29789 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
29790 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
29791 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
29792 (aarch64_init_pauth_hint_builtins): New.
29793 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
29794 (aarch64_expand_builtin): Expand new builtins.
29796 2017-01-19 Jiong Wang <jiong.wang@arm.com>
29798 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
29799 * combine-stack-adj.c (no_unhandled_cfa): Handle
29800 REG_CFA_TOGGLE_RA_MANGLE.
29801 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
29802 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
29803 info for return address signing.
29804 (aarch64_expand_epilogue): Likewise.
29806 2017-01-19 Jiong Wang <jiong.wang@arm.com>
29808 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
29809 * config/aarch64/aarch64-protos.h
29810 (aarch64_return_address_signing_enabled): New declaration.
29811 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
29813 (aarch64_expand_prologue): Sign return address before it's pushed onto
29815 (aarch64_expand_epilogue): Authenticate return address fetched from
29817 (aarch64_override_options): Sanity check for ILP32 and ISA level.
29818 (aarch64_attributes): New function attributes for "sign-return-address".
29819 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
29820 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
29821 ("*do_return"): Generate combined instructions according to key index.
29822 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
29823 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
29825 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
29826 * config/aarch64/aarch64.opt (msign-return-address=): New.
29827 * doc/extend.texi (AArch64 Function Attributes): Documents
29828 "sign-return-address=".
29829 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
29831 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
29833 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
29834 overall option summary.
29836 2017-01-19 Jiong Wang <jiong.wang@arm.com>
29838 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
29839 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
29840 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
29841 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
29843 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
29845 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
29846 -mpower9-minmax by default for -mcpu=power9.
29847 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
29848 128-bit floating point.
29850 2017-01-20 Alan Modra <amodra@gmail.com>
29852 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
29853 optimizing for size.
29855 2017-01-20 Alan Modra <amodra@gmail.com>
29858 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
29859 for strcmp and strncmp from corresponding builtin decl.
29861 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
29863 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
29864 instead of i386/rtems-64.h.
29865 * config/i386/rtems-64.h: Remove.
29867 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
29871 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
29873 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
29875 2017-01-19 Tamar Christina <tamar.christina@arm.com>
29877 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
29878 Change int to HOST_WIDE_INT.
29879 * config/aarch64/aarch64-protos.h
29880 (aarch64_simd_gen_const_vector_dup): Likewise.
29881 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
29883 2017-01-19 David Malcolm <dmalcolm@redhat.com>
29885 * langhooks-def.h (lhd_type_for_size): New decl.
29886 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
29887 * langhooks.c (lhd_type_for_size): New function, taken from
29890 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
29892 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
29893 define_bypass for CR latency.
29894 (power9-cracked-alu): Update bypass latency and remove power9-branch.
29895 (power9-alu2): Add define_bypass for CR latency.
29897 (power9-mul): Update insn latency.
29898 (power9-mul-compare): Update insn latency, bypass latency and remove
29901 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29903 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
29905 * config/aarch64/aarch64.md
29906 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
29907 aarch64_nopcrelative_literal_loads.
29908 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
29910 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
29912 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
29913 TARGET_LOONGSON_3A.
29914 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
29916 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
29919 * config.gcc (supported_defaults): Add lxc1-sxc1.
29920 (with_lxc1_sxc1): Add validation.
29921 (all_defaults): Add lxc1-sxc1.
29922 * config/mips/mips.opt (mlxc1-sxc1): New option.
29923 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
29925 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
29926 __mips_no_lxc1_sxc1.
29927 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
29928 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
29929 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
29931 2017-01-19 Richard Biener <rguenther@suse.de>
29933 PR tree-optimization/72488
29934 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
29935 sure to restore SSA info.
29936 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
29938 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
29940 PR rtl-optimization/79121
29941 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
29942 of the inner type when shifting an extended value.
29944 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
29947 * symtab.c (symtab_node::equal_address_to): Fix comparing of
29948 interposable aliases.
29950 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
29953 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
29954 Use the evmergelohi instruction.
29955 (mov_si<mode>_e500_subreg4_2_le): Likewise.
29956 (mov_sitf_e500_subreg8_2_be): Likewise.
29957 (mov_sitf_e500_subreg12_2_le): Likewise.
29958 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
29959 (mov_si<mode>_e500_subreg4_2_be): Likewise.
29960 (mov_sitf_e500_subreg8_2_le): Likewise.
29961 (mov_sitf_e500_subreg12_2_be): Likewise.
29963 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29965 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
29966 attribute from vecsimple to vecperm.
29967 (altivec_vbpermq2): Likewise.
29969 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29972 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
29974 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29975 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
29976 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
29977 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
29978 case where N arg is SIZE_MAX.
29979 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
29980 (cmpstrsi): Add pattern.
29982 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
29984 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
29985 __builtin_vec_revb builtins.
29986 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
29987 built-in functions to support generation of the ISA 3.0 XXBR<x>
29988 vector byte reverse instructions.
29989 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
29990 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
29991 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
29992 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
29993 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
29994 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
29995 (P9V_BUILTIN_VEC_REVB): Likewise.
29996 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
29997 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
29998 (p9_xxbrq_v16qi): Likewise.
29999 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
30000 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
30001 (p9_xxbrh_v8hi): Likewise.
30002 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
30003 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
30004 vec_revb built-in functions.
30006 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
30008 PR rtl-optimization/78952
30009 * config/i386/i386.md (any_extract): New code iterator.
30010 (*insvqi_2): Use any_extract for source operand.
30011 (*insvqi_3): Use any_shiftrt for source operand.
30013 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
30015 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
30017 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
30019 2017-01-18 Matthias Klose <doko@ubuntu.com>
30021 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
30023 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30025 * config/rs6000/altivec.h (vec_bperm): Change #define.
30026 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
30027 (altivec_vbpermq2): New define_insn.
30028 (altivec_vbpermd): Likewise.
30029 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
30030 function interface.
30031 (VBPERMD): Likewise.
30032 (VBPERM): New polymorphic function interface.
30033 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
30034 Add entries for P9V_BUILTIN_VEC_VBPERM.
30035 * doc/extend.texi: Add interfaces for vec_bperm.
30037 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30039 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
30040 first letter of error messages.
30041 (s390_resolve_overloaded_builtin): Likewise.
30042 * config/s390/s390.c (s390_expand_builtin): Likewise.
30043 (s390_invalid_arg_for_unprototyped_fn): Likewise.
30044 (s390_valid_target_attribute_inner_p): Likewise.
30045 * config/s390/s390.md ("tabort"): Likewise.
30047 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
30049 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
30050 (ISA_AVOID_DIV_HILO): New macro.
30051 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
30052 (ISA_HAS_DDIV): Likewise.
30054 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
30056 * doc/invoke.texi (fabi-version): Correct number of occurrences.
30058 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
30060 * doc/invoke.texi (fabi-version): Spelling fix.
30062 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
30065 * doc/invoke.texi (fabi-version): Mention mangling fix for
30068 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
30071 * doc/invoke.texi (fabi-version): Document discriminator mangling.
30073 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
30076 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
30077 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
30079 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
30081 (stack_protect_test): Ditto.
30082 * config/rs6000/rs6000.opt (mstack-protector-guard=,
30083 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
30085 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
30086 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
30087 -mstack-protector-guard-offset=.
30088 (RS/6000 and PowerPC Options): Ditto.
30090 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
30092 * config/i386/i386.h (MASK_CLASS_P): New define.
30093 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
30094 there are no registers from different register sets also when
30095 mask registers are used. Update function comment.
30096 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
30097 to (*k/*r) and (*k/*km) alternatives.
30099 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
30101 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
30102 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
30103 (EH_RETURN_HANDLER_RTX): New define.
30104 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
30105 Force frame pointer in EH return functions.
30106 (aarch64_expand_epilogue): Add barrier for eh_return.
30107 (aarch64_final_eh_return_addr): Remove.
30108 (aarch64_eh_return_handler_rtx): New function.
30109 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
30111 (aarch64_eh_return_handler_rtx): New prototype.
30113 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30115 * config/rs6000/altivec.h (vec_rlmi): New #define.
30116 (vec_vrlnm): Likewise.
30117 (vec_rlnm): Likewise.
30118 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
30119 (UNSPEC_VRLNM): Likewise.
30120 (VIlong): New mode iterator.
30121 (altivec_vrl<VI_char>mi): New define_insn.
30122 (altivec_vrl<VI_char>nm): Likewise.
30123 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
30125 (VRLDNM): Likewise.
30126 (RLNM): New polymorphic function entry.
30127 (VRLWMI): New monomorphic function entry.
30128 (VRLDMI): Likewise.
30129 (RLMI): New polymorphic function entry.
30130 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
30131 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
30132 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
30135 2017-01-17 Jakub Jelinek <jakub@redhat.com>
30138 * dwarf2out.c (field_byte_offset): Restore the
30139 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
30140 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
30141 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
30144 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
30147 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
30149 2017-01-17 Jakub Jelinek <jakub@redhat.com>
30152 * dwarf2out.c (add_data_member_location_attribute): For constant
30153 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
30154 instead of DW_AT_data_member_location, DW_AT_bit_offset and
30155 DW_AT_byte_size attributes.
30157 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
30159 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
30160 after forcing to constant memory when the code model is medium.
30162 2017-01-17 Julia Koval <julia.koval@intel.com>
30165 * config/i386/avx512fintrin.h
30166 (_mm512_i32gather_ps): Change __addr type to void const*.
30167 (_mm512_mask_i32gather_ps): Ditto.
30168 (_mm512_i32gather_pd): Ditto.
30169 (_mm512_mask_i32gather_pd): Ditto.
30170 (_mm512_i64gather_ps): Ditto.
30171 (_mm512_mask_i64gather_ps): Ditto.
30172 (_mm512_i64gather_pd): Ditto.
30173 (_mm512_mask_i64gather_pd): Ditto.
30174 (_mm512_i32gather_epi32): Ditto.
30175 (_mm512_mask_i32gather_epi32): Ditto.
30176 (_mm512_i32gather_epi64): Ditto.
30177 (_mm512_mask_i32gather_epi64): Ditto.
30178 (_mm512_i64gather_epi32): Ditto.
30179 (_mm512_mask_i64gather_epi32): Ditto.
30180 (_mm512_i64gather_epi64): Ditto.
30181 (_mm512_mask_i64gather_epi64): Ditto.
30182 (_mm512_i32scatter_ps): Change __addr type to void*.
30183 (_mm512_mask_i32scatter_ps): Ditto.
30184 (_mm512_i32scatter_pd): Ditto.
30185 (_mm512_mask_i32scatter_pd): Ditto.
30186 (_mm512_i64scatter_ps): Ditto.
30187 (_mm512_mask_i64scatter_ps): Ditto.
30188 (_mm512_i64scatter_pd): Ditto.
30189 (_mm512_mask_i64scatter_pd): Ditto.
30190 (_mm512_i32scatter_epi32): Ditto.
30191 (_mm512_mask_i32scatter_epi32): Ditto.
30192 (_mm512_i32scatter_epi64): Ditto.
30193 (_mm512_mask_i32scatter_epi64): Ditto.
30194 (_mm512_i64scatter_epi32): Ditto.
30195 (_mm512_mask_i64scatter_epi32): Ditto.
30196 (_mm512_i64scatter_epi64): Ditto.
30197 (_mm512_mask_i64scatter_epi64): Ditto.
30198 * config/i386/avx512pfintrin.h
30199 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
30200 (_mm512_mask_prefetch_i32gather_ps): Ditto.
30201 (_mm512_mask_prefetch_i64gather_pd): Ditto.
30202 (_mm512_mask_prefetch_i64gather_ps): Ditto.
30203 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
30204 (_mm512_prefetch_i32scatter_ps): Ditto.
30205 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
30206 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
30207 (_mm512_prefetch_i64scatter_pd): Ditto.
30208 (_mm512_prefetch_i64scatter_ps): Ditto.
30209 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
30210 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
30211 * config/i386/avx512vlintrin.h
30212 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
30213 (_mm_mmask_i32gather_ps): Ditto.
30214 (_mm256_mmask_i32gather_pd): Ditto.
30215 (_mm_mmask_i32gather_pd): Ditto.
30216 (_mm256_mmask_i64gather_ps): Ditto.
30217 (_mm_mmask_i64gather_ps): Ditto.
30218 (_mm256_mmask_i64gather_pd): Ditto.
30219 (_mm_mmask_i64gather_pd): Ditto.
30220 (_mm256_mmask_i32gather_epi32): Ditto.
30221 (_mm_mmask_i32gather_epi32): Ditto.
30222 (_mm256_mmask_i32gather_epi64): Ditto.
30223 (_mm_mmask_i32gather_epi64): Ditto.
30224 (_mm256_mmask_i64gather_epi32): Ditto.
30225 (_mm_mmask_i64gather_epi32): Ditto.
30226 (_mm256_mmask_i64gather_epi64): Ditto.
30227 (_mm_mmask_i64gather_epi64): Ditto.
30228 (_mm256_i32scatter_ps): Change __addr type to void*.
30229 (_mm256_mask_i32scatter_ps): Ditto.
30230 (_mm_i32scatter_ps): Ditto.
30231 (_mm_mask_i32scatter_ps): Ditto.
30232 (_mm256_i32scatter_pd): Ditto.
30233 (_mm256_mask_i32scatter_pd): Ditto.
30234 (_mm_i32scatter_pd): Ditto.
30235 (_mm_mask_i32scatter_pd): Ditto.
30236 (_mm256_i64scatter_ps): Ditto.
30237 (_mm256_mask_i64scatter_ps): Ditto.
30238 (_mm_i64scatter_ps): Ditto.
30239 (_mm_mask_i64scatter_ps): Ditto.
30240 (_mm256_i64scatter_pd): Ditto.
30241 (_mm256_mask_i64scatter_pd): Ditto.
30242 (_mm_i64scatter_pd): Ditto.
30243 (_mm_mask_i64scatter_pd): Ditto.
30244 (_mm256_i32scatter_epi32): Ditto.
30245 (_mm256_mask_i32scatter_epi32): Ditto.
30246 (_mm_i32scatter_epi32): Ditto.
30247 (_mm_mask_i32scatter_epi32): Ditto.
30248 (_mm256_i32scatter_epi64): Ditto.
30249 (_mm256_mask_i32scatter_epi64): Ditto.
30250 (_mm_i32scatter_epi64): Ditto.
30251 (_mm_mask_i32scatter_epi64): Ditto.
30252 (_mm256_i64scatter_epi32): Ditto.
30253 (_mm256_mask_i64scatter_epi32): Ditto.
30254 (_mm_i64scatter_epi32): Ditto.
30255 (_mm_mask_i64scatter_epi32): Ditto.
30256 (_mm256_i64scatter_epi64): Ditto.
30257 (_mm256_mask_i64scatter_epi64): Ditto.
30258 (_mm_i64scatter_epi64): Ditto.
30259 (_mm_mask_i64scatter_epi64): Ditto.
30260 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
30261 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
30262 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
30263 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
30264 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
30265 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
30266 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
30267 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
30268 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
30269 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
30270 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
30271 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
30272 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
30273 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
30274 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
30275 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
30276 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
30277 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
30278 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
30279 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
30280 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
30281 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
30282 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
30283 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
30284 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
30285 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
30286 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
30287 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
30288 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
30289 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
30290 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
30291 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
30292 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
30293 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
30294 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
30295 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
30296 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
30297 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
30298 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
30299 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
30300 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
30301 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
30302 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
30303 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
30304 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
30305 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
30306 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
30307 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
30308 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
30309 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
30310 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
30311 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
30312 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
30313 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
30314 definitions accordingly.
30316 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
30317 Kuan-Lin Chen <kuanlinchentw@gmail.com>
30320 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
30323 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
30326 * ira-conflicts.c (ira_build_conflicts): Update total conflict
30327 hard regs for inner regno.
30329 2017-01-17 Martin Liska <mliska@suse.cz>
30332 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
30333 assumption and add comment.
30335 2017-01-17 Nathan Sidwell <nathan@acm.org>
30337 * ipa-visibility.c (localize_node): New function, broken out of ...
30338 (function_and_variable_visibility): ... here. Call it.
30340 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
30342 PR middle-end/77445
30343 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
30344 correctly set frequency of oudgoing edge.
30345 (duplicate_thread_path): Fix profile updating.
30347 2017-01-17 Jakub Jelinek <jakub@redhat.com>
30350 * configure.ac: Add GCC_BASE_VER.
30351 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
30352 version from BASE-VER file.
30353 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
30354 (gcc.o): Depend on $(BASEVER).
30355 * common.opt (dumpfullversion): New option.
30356 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
30357 * doc/invoke.texi: Document -dumpfullversion.
30358 * doc/install.texi: Document --with-gcc-major-version-only.
30359 * configure: Regenerated.
30361 2017-01-17 Richard Biener <rguenther@suse.de>
30363 PR tree-optimization/71433
30364 * tree-vrp.c (register_new_assert_for): Merge same asserts
30365 on all incoming edges.
30366 (process_assert_insertions_for): Handle insertions at the
30369 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
30371 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
30372 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
30374 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
30377 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
30380 2017-01-17 Alan Modra <amodra@gmail.com>
30383 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
30384 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
30385 symbolic stack limit when pic.
30387 2017-01-16 Martin Sebor <msebor@redhat.com>
30389 PR tree-optimization/78608
30390 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
30392 2017-01-16 Jeff Law <law@redhat.com>
30395 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
30396 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
30397 for several include directories that may be relative to sysroot.
30398 * config/i386/x-mingw32 (gplus_includedir): Define.
30399 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
30400 (native_system_includedir): Likewise.
30401 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
30402 override if TARGET_SYSTEM_ROOT is defined.
30403 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
30405 PR tree-optimization/79090
30406 PR tree-optimization/33562
30407 PR tree-optimization/61912
30408 PR tree-optimization/77485
30409 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
30410 and computed trims into the dump file.
30412 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
30414 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
30416 2017-01-16 Jakub Jelinek <jakub@redhat.com>
30419 * gimplify.c (gimplify_init_constructor): If want_value and
30420 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
30424 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
30425 sequence. Formatting fixes.
30426 (doloop_optimize): Formatting fixes.
30429 * gcc.c (debug_level_greater_than_spec_func): New function.
30430 (static_spec_functions): Add debug-level-gt spec function.
30431 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
30433 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
30434 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
30435 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
30436 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
30437 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
30438 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
30440 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
30442 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
30443 QImode fixups to general and mask registers only.
30445 2017-01-16 Carl Love <cel@us.ibm.com>
30447 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
30448 for built-in functions
30449 vector signed char vec_nabs (vector signed char)
30450 vector signed short vec_nabs (vector signed short)
30451 vector signed int vec_nabs (vector signed int)
30452 vector signed long long vec_nabs (vector signed long long)
30453 vector float vec_nabs (vector float)
30454 vector double vec_nabs (vector double)
30455 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
30457 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
30458 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
30459 * doc/extend.texi: Update the documentation file for the new built-in
30462 2017-01-16 Martin Sebor <msebor@redhat.com>
30464 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
30467 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30469 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
30470 UNSPEC_VSX__XXSPLTD to require special splat handling.
30472 2017-01-16 David Malcolm <dmalcolm@redhat.com>
30475 * system.h: Poison strndup.
30477 2017-01-16 Alan Modra <amodra@gmail.com>
30480 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
30483 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
30485 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
30487 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
30489 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
30490 call recog here. Assert that INSN_CODE (insn) is non-negative.
30492 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
30495 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
30497 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
30498 in the currently scheduled RTL fragment.
30500 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
30502 PR rtl-optimization/78751
30503 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
30506 2017-01-14 Jeff Law <law@redhat.com>
30508 PR tree-optimization/79090
30509 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
30510 variable length stores.
30511 (compute_trims): Delete dead assignment to *trim_tail.
30512 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
30515 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
30517 PR rtl-optimization/78626
30518 PR rtl-optimization/78727
30519 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
30520 of a block, and split such blocks after everything else is finished.
30522 2017-01-14 Alan Modra <amodra@gmail.com>
30525 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
30526 target legitimate_combined_insn.
30527 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
30528 (rs6000_legitimate_combined_insn): New function.
30529 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
30531 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
30532 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
30533 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
30535 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
30537 * doc/frontends.texi (G++ and GCC): Remove references to Java.
30539 2017-01-13 Jeff Law <law@redhat.com>
30541 PR tree-optimization/33562
30542 PR tree-optimization/61912
30543 PR tree-optimization/77485
30544 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
30546 (delete_dead_assignment): Likewise.
30547 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
30548 statement to delete_dead_call and delete_dead_assignment.
30550 2017-01-13 David Malcolm <dmalcolm@redhat.com>
30553 * substring-locations.c (format_warning_va): Strengthen case 1 so
30554 that both endpoints of the substring must be within the format
30555 range for just the substring to be printed.
30557 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
30559 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
30560 * config/i386/i386.c (ix86_target_string): Add missing options
30561 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
30562 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
30563 flags_other and ix86_target_other to flags2_other. Display unknown
30565 (ix86_valid_target_attribute_inner_p): Add missing options and
30566 reorder options by implied ISAs, as in ix86_target_string.
30568 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
30570 * hash-table.h (hash_table::too_empty_p): New function.
30571 (hash_table::expand): Use it.
30572 (hash_table::traverse): Likewise.
30573 (hash_table::empty_slot): Use sizeof (value_type) instead of
30574 sizeof (PTR) to convert bytes to elements. Shrink the table
30575 if the current size is excessive for the current number of
30578 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
30580 * ira-costs.c (record_reg_classes): Break from the inner loop
30581 early once alt_fail is known to be true. Update outer loop
30582 handling accordingly.
30584 2017-01-13 Jeff Law <law@redhat.com>
30586 * tree-ssa-dse.c (decrement_count): New function.
30587 (increment_start_addr, maybe_trim_memstar_call): Likewise.
30588 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
30589 when we know the partially dead statement is a mem* function.
30591 PR tree-optimization/61912
30592 PR tree-optimization/77485
30593 * tree-ssa-dse.c: Include expr.h.
30594 (maybe_trim_constructor_store): New function.
30595 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
30597 PR tree-optimization/33562
30598 PR tree-optimization/61912
30599 PR tree-optimization/77485
30600 * doc/invoke.texi: Document new dse-max-object-size param.
30601 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
30602 * tree-ssa-dse.c: Include params.h.
30603 (dse_store_status): New enum.
30604 (initialize_ao_ref_for_dse): New, partially extracted from
30606 (valid_ao_ref_for_dse, normalize_ref): New.
30607 (setup_live_bytes_from_ref, compute_trims): Likewise.
30608 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
30609 (maybe_trim_partially_dead_store): Likewise.
30610 (maybe_trim_complex_store): Likewise.
30611 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
30612 Track what bytes live from the original store. Return tri-state
30613 for dead, partially dead or live.
30614 (dse_dom_walker): Add constructor, destructor and new private members.
30615 (delete_dead_call, delete_dead_assignment): New extracted from
30617 (dse_optimize_stmt): Make a member of dse_dom_walker.
30618 Use initialize_ao_ref_for_dse.
30620 PR tree-optimization/33562
30621 PR tree-optimization/61912
30622 PR tree-optimization/77485
30623 * sbitmap.h (bitmap_count_bits): Prototype.
30624 (bitmap_clear_range, bitmap_set_range): Likewise.
30625 * sbitmap.c (bitmap_clear_range): New function.
30626 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
30628 2017-01-13 Martin Liska <mliska@suse.cz>
30631 * function.c (set_cfun): Add new argument force.
30632 * function.h (set_cfun): Likewise.
30633 * ipa-inline-transform.c (inline_call): Use the function when
30634 strict alising from is dropped for function we inline to.
30636 2017-01-13 Richard Biener <rguenther@suse.de>
30638 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
30639 for dumping GIMPLE INTEGER_CSTs.
30641 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
30643 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
30644 to 201112L since C++17.
30646 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
30649 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
30650 if -fsanitize=kernel-address is present.
30652 2017-01-13 Richard Biener <rguenther@suse.de>
30654 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
30655 as _Literal ( type ) number in case usual suffixes do not
30656 preserve all information.
30658 2017-01-13 Richard Biener <rguenther@suse.de>
30660 PR tree-optimization/77283
30661 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
30662 and ssa-iterators.h.
30663 (is_feasible_trace): Implement a cost model based on joiner
30666 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
30669 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
30670 char or short to __float128/_Float128 directly.
30672 2017-01-12 Martin Sebor <msebor@redhat.com>
30674 to -Wformat-overflow.
30675 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
30676 (min_bytes_remaining): Same.
30677 (get_string_length): Same.
30678 (format_string): Same.
30679 (format_directive): Same.
30681 (pass_sprintf_length::handle_gimple_call): Same.
30683 2017-01-12 Jakub Jelinek <jakub@redhat.com>
30685 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
30686 info.nowrite calls with no lhs that can't throw. Return bool
30687 whether gsi_remove has been called or not.
30688 (pass_sprintf_length::handle_gimple_call): Return bool whether
30689 try_substitute_return_value called gsi_remove. Formatting fix.
30690 (pass_sprintf_length::execute): Don't use gsi_remove if
30691 handle_gimple_call returned true.
30694 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
30695 be removed due to side-effects, don't remove following barrier nor
30696 turn the successor edge into fallthru edge.
30698 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30701 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
30702 element-reversing loads and stores as not swappable.
30704 2017-01-12 Nathan Sidwell <nathan@acm.org>
30705 Nicolai Stange <nicstange@gmail.com>
30707 * combine.c (try_combine): Don't ignore result of overlap checking
30708 loop. Combine overlap & asm check into single loop.
30710 2017-01-12 Richard Biener <rguenther@suse.de>
30712 * tree-pretty-print.c (dump_generic_node): Provide -gimple
30713 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
30715 2017-01-12 Richard Biener <rguenther@suse.de>
30717 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
30718 and TS_TARGET_OPTION directly derive from TS_BASE.
30719 * tree-core.h (tree_optimization_option): Derive from tree_base.
30720 (tree_target_option): Likewise.
30722 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
30724 * config/i386/i386.c (memory_address_length): Increase len
30725 only when rip_relative_addr_p returns false.
30727 2017-01-11 Julia Koval <julia.koval@intel.com>
30729 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
30730 (OPTION_MASK_ISA_SGX_SET): New.
30731 (ix86_handle_option): Handle OPT_msgx.
30732 * config.gcc: Added sgxintrin.h.
30733 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
30734 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
30735 * config/i386/i386.c (ix86_target_string): Add -msgx.
30737 (ix86_option_override_internal): Handle new options.
30738 (ix86_valid_target_attribute_inner_p): Add sgx.
30739 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
30740 * config/i386/i386.opt: Add msgx.
30741 * config/i386/sgxintrin.h: New file.
30742 * config/i386/x86intrin.h: Add sgxintrin.h.
30744 2017-01-11 Jakub Jelinek <jakub@redhat.com>
30747 * fold-const.c (maybe_nonzero_address): Return 1 for function
30749 (tree_single_nonzero_warnv_p): Don't handle function local objects
30753 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
30756 2017-01-11 David Malcolm <dmalcolm@redhat.com>
30759 * opts.c: Include "spellcheck.h"
30760 (struct string_fragment): New struct.
30761 (struct edit_distance_traits<const string_fragment &>): New
30763 (get_closest_sanitizer_option): New function.
30764 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
30766 2017-01-11 Jakub Jelinek <jakub@redhat.com>
30768 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
30770 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
30771 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
30772 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
30773 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
30774 for initial die_offset if dwarf_split_debug_info.
30775 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
30776 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
30778 (output_skeleton_debug_sections): Formatting fix. Use
30779 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
30780 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
30782 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
30784 * config/arm/cortex-a53.md: Add bypasses for
30785 cortex_a53_r2f_cvt.
30786 (cortex_a53_r2f): Only use for transfers.
30787 (cortex_a53_f2r): Likewise.
30788 (cortex_a53_r2f_cvt): Add reservation for conversions.
30789 (cortex_a53_f2r_cvt): Likewise.
30791 2017-01-11 Tamar Christina <tamar.christina@arm.com>
30793 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
30794 to all inlined functions, change static to extern.
30796 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
30799 * config/arm/arm.c (legitimize_pic_address): Handle reference to
30801 (arm_assemble_integer): Likewise.
30803 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
30805 * config.gcc: Use new awk script to check CPU, FPU and architecture
30806 parameters for --with-... options.
30807 * config/arm/parsecpu.awk: New file
30808 * config/arm/arm-cpus.in: New file.
30809 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
30811 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
30813 * config/arm/t-arm: Update dependency rules.
30814 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
30815 of processing .def files.
30816 * config/arm/genopt.sh: Deleted.
30817 * config/arm/gentune.sh: Deleted.
30818 * config/arm/arm-cores.def: Deleted.
30819 * config/arm/arm-arches.def: Deleted.
30820 * config/arm/arm-fpus.def: Deleted.
30821 * config/arm/arm-tune.md: Regenerated.
30822 * config/arm/arm-tables.opt: Regenerated.
30823 * config/arm/arm-cpu.h: New generated file.
30824 * config/arm/arm-cpu-data.h: New generated file.
30825 * config/arm/arm-cpu-cdata.h: New generated file.
30827 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
30830 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
30832 (input_varpool_node): Unpack dynamically_initialized bit.
30834 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
30836 PR rtl-optimization/79032
30837 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
30838 the alignment of the adjusted memory reference against that of MODE,
30839 instead of the alignment of the original memory reference.
30841 2017-01-11 Martin Jambor <mjambor@suse.cz>
30843 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
30845 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
30846 decorated functions.
30848 2017-01-11 Richard Biener <rguenther@suse.de>
30850 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
30851 set range/nonnull info for PHI results. Do not set it on
30852 stmts marked for removal.
30854 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
30856 * expr.c (store_field): In the bitfield case, fetch the return value
30857 from the registers before applying a single big-endian adjustment.
30858 Always do a final load for a BLKmode value not larger than a word.
30860 2017-01-10 David Malcolm <dmalcolm@redhat.com>
30863 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
30864 that we correctly handle column numbers greater than
30865 LINE_MAP_MAX_COLUMN_NUMBER.
30867 2017-01-10 Martin Sebor <msebor@redhat.com>
30869 PR middle-end/78245
30870 * gimple-ssa-sprintf.c (get_destination_size): Call
30871 {init,fini}object_sizes.
30872 * tree-object-size.c (addr_object_size): Adjust.
30873 (pass_through_call): Adjust.
30874 (pass_object_sizes::execute): Adjust.
30875 * tree-object-size.h (fini_object_sizes): Declare.
30877 2017-01-10 Martin Sebor <msebor@redhat.com>
30879 PR tree-optimization/78775
30880 * builtins.c (get_size_range): Move...
30881 * calls.c: ...to here.
30882 (alloc_max_size): Accept zero argument.
30883 (operand_signed_p): Remove.
30884 (maybe_warn_alloc_args_overflow): Call get_size_range.
30885 * calls.h (get_size_range): Declare.
30887 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
30889 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
30890 from TI's devices.csv file as of September 2016.
30891 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
30893 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
30895 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
30896 * doc/invoke.texi: Likewise.
30897 * doc/md.texi: Likewise.
30898 * doc/objc.texi: Likewise.
30900 2017-01-10 Joshua Conner <joshconner@google.com>
30902 * config/arm/fuchsia-elf.h: New file.
30903 * config/fuchsia.h: New file.
30904 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
30905 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
30907 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
30909 2016-01-10 Richard Biener <rguenther@suse.de>
30911 PR tree-optimization/79034
30912 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
30913 Propagate out degenerate PHIs in the joiner.
30915 2017-01-10 Martin Liska <mliska@suse.cz>
30917 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
30918 (sort_congruence_classes_by_decl_uid): Likewise.
30919 (sort_congruence_class_groups_by_decl_uid): Likewise.
30920 (sem_item_optimizer::merge_classes): Sort class, groups in these
30921 classes and members in the groups by DECL_UID of declarations.
30922 This would make merge operations stable.
30924 2017-01-10 Martin Liska <mliska@suse.cz>
30926 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
30927 usage of m_classes_vec.
30928 (sem_item_optimizer::~sem_item_optimizer): Likewise.
30929 (sem_item_optimizer::get_group_by_hash): Likewise.
30930 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
30931 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
30932 (sem_item_optimizer::verify_classes): Likewise.
30933 (sem_item_optimizer::process_cong_reduction): Likewise.
30934 (sem_item_optimizer::dump_cong_classes): Likewise.
30935 (sem_item_optimizer::merge_classes): Likewise.
30936 * ipa-icf.h (congruence_class_hash): Rename from
30937 congruence_class_group_hash. Remove declaration of m_classes_vec.
30939 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
30941 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
30942 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
30943 * config.gcc: Add avx512vpopcntdqintrin.h.
30944 * config/i386/avx512vpopcntdqintrin.h: New.
30945 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
30946 * config/i386/i386-builtin-types.def: Add new types.
30947 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
30948 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
30949 __builtin_ia32_vpopcountq_v8di_mask): New.
30950 * config/i386/i386-c.c (ix86_target_macros_internal): Define
30951 __AVX512VPOPCNTDQ__.
30952 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
30953 (PTA_AVX512VPOPCNTDQ): Define.
30954 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
30955 TARGET_AVX512VPOPCNTDQ_P): Define.
30956 * config/i386/i386.opt: Add mavx512vpopcntdq.
30957 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
30958 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
30960 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
30962 PR middle-end/77484
30963 * predict.def (PRED_CALL): Set to 67.
30965 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
30967 * expr.c (store_field): In the bitfield case, if the value comes from
30968 a function call and is of an aggregate type returned in registers, do
30969 not modify the field mode; extract the value in all cases if the mode
30970 is BLKmode and the size is not larger than a word.
30972 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
30975 * config/i386/cpuid.h: Fix undefined behavior.
30977 2017-01-04 Jeff Law <law@redhat.com>
30979 PR tree-optimization/79007
30980 PR tree-optimization/67955
30981 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
30982 conservative for pt.null when flag_non_call_exceptions is on.
30984 2017-01-09 Jakub Jelinek <jakub@redhat.com>
30986 PR translation/79019
30987 PR translation/79020
30988 * params.def (PARAM_INLINE_MIN_SPEEDUP,
30989 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
30990 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
30992 * config/avr/avr.opt (maccumulate-args): Likewise.
30993 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
30994 * common.opt (freport-bug): Likewise.
30995 * cif-code.def (CIF_FINAL_ERROR): Likewise.
30996 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
30997 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
30998 translatable string.
30999 * config/i386/i386.c (function_value_32): Likewise.
31000 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
31001 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
31003 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
31004 * common/config/msp430/msp430-common.c (msp430_handle_option):
31006 * symtab.c (symtab_node::verify_base): Likewise.
31007 * opts.c (set_debug_level): Likewise.
31008 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
31009 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
31010 missing whitespace to translatable strings.
31011 * config/avr/avr.md (bswapsi2): Fix typo in comment.
31012 * config/sh/superh.h: Likewise.
31013 * config/i386/xopintrin.h: Likewise.
31014 * config/i386/znver1.md: Likewise.
31015 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
31016 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
31017 * double-int.h (struct double_int): Likewise.
31018 * double-int.c (div_and_round_double): Likewise.
31019 * wide-int.cc: Likewise.
31020 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
31021 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
31022 * cfgcleanup.c (crossjumps_occured): Renamed to ...
31023 (crossjumps_occurred): ... this.
31024 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
31027 PR tree-optimization/78899
31028 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
31029 returning bool return struct loop *, NULL for failure and the new
31031 (versionable_outer_loop_p): Don't version outer loop if it has
31032 dont_vectorized bit set.
31033 (tree_if_conversion): When versioning outer loop, ensure
31034 tree_if_conversion is performed also on the inner loop of the
31035 non-vectorizable outer loop copy.
31036 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
31037 LOOP_VECTORIZED in inner loop of the scalar outer loop and
31038 prevent vectorization of it.
31039 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
31040 the outer loop vectorization of the non-scalar version is attempted
31041 before vectorization of the inner loop in scalar version. If
31042 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
31043 vectorization of its inner loop.
31044 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
31045 has 2 inner loops, rename also on edges from bb whose single pred
31046 is outer_loop->header. Fix typo in function comment.
31048 2017-01-09 Martin Sebor <msebor@redhat.com>
31051 * asan.c (asan_emit_stack_protection): Increase local buffer size
31052 to avoid snprintf truncation warning.
31054 2017-01-09 Andrew Pinski <apinski@cavium.com>
31056 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
31057 to reference thunderx2t99 for the tuning structure
31058 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
31060 (thunderx2t99_extra_costs): This.
31061 * config/aarch64/aarch64-tune.md: Regenerate.
31062 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
31063 (vulcan_addrcost_table): This.
31064 (vulcan_regmove_cost): Rename to ...
31065 (thunderx2t99_regmove_cost): This.
31066 (vulcan_vector_cost): Rename to ...
31067 (thunderx2t99_vector_cost): this.
31068 (vulcan_branch_cost): Rename to ...
31069 (thunderx2t99_branch_cost): This.
31070 (vulcan_tunings): Rename to ...
31071 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
31072 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
31074 2017-01-09 Martin Jambor <mjambor@suse.cz>
31078 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
31079 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
31080 (propagate_vr_accross_jump_function): Use the above function for all
31081 value range computations for pass-through jump functions and type
31082 converasion from explicit value range values.
31083 (ipcp_propagate_stage): Do not attempt to deduce types of formal
31084 parameters from TYPE_ARG_TYPES.
31085 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
31086 (ipa_write_node_info): Stream type of the actual argument.
31087 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
31089 2017-01-09 Martin Liska <mliska@suse.cz>
31092 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
31093 (lookup_compiler): Do not show error message with have_E.
31095 2017-01-09 Jakub Jelinek <jakub@redhat.com>
31097 PR tree-optimization/78938
31098 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
31099 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
31100 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
31101 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
31104 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31106 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
31109 2017-01-09 Richard Biener <rguenther@suse.de>
31111 PR tree-optimization/78997
31112 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
31113 name condition properly.
31115 2017-01-09 Richard Biener <rguenther@suse.de>
31118 * dwarf2out.c (is_cxx): New overload with context.
31119 (is_naming_typedef_decl): Use it.
31121 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
31123 * invoke.texi (Option Summary): Correct spacing in option lists
31124 and add line breaks to fix over-long lines.
31126 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
31128 PR middle-end/17660
31130 * extend.texi (Common Variable Attributes): Add xref to GCC
31131 Internals manual to explain mode attribute keywords.
31133 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
31136 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
31137 and Preprocessor Options.
31138 (Options for Linking): Document -pthread here....
31139 (RS/6000 and PowerPC Options): ...not here.
31140 (Solaris 2 Options): ...or here.
31141 * doc/cppopts.texi: Document -pthread.
31143 2017-01-08 Martin Sebor <msebor@redhat.com>
31145 PR middle-end/77708
31146 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
31147 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
31148 New member functions.
31149 (format_directive): Used them.
31151 (pass_sprintf_length::handle_gimple_call): Same.
31152 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
31153 to avoid truncation for any argument.
31154 (extract_affine_mul): Same.
31155 * tree.c (get_file_function_name): Same.
31157 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
31159 PR middle-end/77484
31160 * predict.def (PRED_INDIR_CALL): Set to 86.
31162 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
31164 PR preprocessor/54124
31165 * doc/cppopts.texi: Reformat -d subtable to list the full name
31166 of the options. Add cross-reference to the docs for the general
31167 compiler -d options.
31168 * doc/invoke.texi (Developer Options): Add cross-reference to the
31169 preprocessor-specific -d option documentation.
31171 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
31173 PR preprocessor/13498
31174 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
31175 redudant material, and reflect new command-line options.
31176 (System Headers): Likewise.
31178 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
31180 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
31181 -isystem, and -idirafter. Copy-edit.
31182 * doc/cppopts.texi: Copy-edit. Remove contradiction about
31183 default for -ftrack-macro-expansion. Delete obsolete and
31184 badly-formatted implementation details about -fdebug-cpp output.
31185 * doc/cppwarnopts.texi: Copy-edit.
31187 2017-01-07 David Malcolm <dmalcolm@redhat.com>
31190 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
31191 that the transition from a max line width >= 1<<10 to narrower
31192 lines works correctly.
31194 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
31196 * doc/options.texi (PerFunction): New.
31197 * opt-functions.awk (switch_flags): Map both Optimization and
31198 PerFunction to CL_OPTIMIZATION.
31199 * opth-gen.awk: Test for PerFunction flag along with
31201 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
31202 it only when the latter is present. Skip those that don't in
31203 the hash function generator.
31204 * common.opt (fvar-tracking): Mark as PerFunction instead of
31206 (fvar-tracking-assignments): Likewise.
31207 (fvar-tracking-assignments-toggle): Likewise.
31208 (fvar-tracking-uninit): Likewise.
31210 2017-01-07 Jakub Jelinek <jakub@redhat.com>
31212 PR translation/79018
31213 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
31216 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
31219 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
31220 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
31221 TARGET_LONG_JUMP_TABLE_OFFSETS.
31222 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
31223 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
31224 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
31225 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
31226 * config/m68k/m68k.md (tablejump expander): Likewise.
31227 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
31228 TARGET_LONG_JUMP_TABLE_OFFSETS.
31229 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
31230 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
31232 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
31233 David Holsgrove <david.holsgrove@xilinx.com>
31235 * common/config/microblaze/microblaze-common.c
31236 (TARGET_EXCEPT_UNWIND_INFO): Remove.
31237 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
31239 * config/microblaze/microblaze.c (microblaze_must_save_register)
31240 (microblaze_expand_epilogue, microblaze_return_addr): Handle
31242 (microblaze_eh_return): New function.
31243 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
31244 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
31245 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
31246 * config/microblaze/microblaze.md (eh_return): New pattern.
31248 2017-01-06 Jakub Jelinek <jakub@redhat.com>
31250 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
31251 GCC_DIAGNOSTIC_STRINGIFY): Define.
31253 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
31255 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31257 * config/arm/arm.md (<mcrr>): New.
31259 * config/arm/arm.c (arm_arch5te): New.
31260 (arm_option_override): Set arm_arch5te.
31261 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
31263 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
31264 (arm_mcrr_qualifiers): ... this. New.
31265 (MRRC_QUALIFIERS): Define to...
31266 (arm_mrrc_qualifiers): ... this. New.
31267 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
31269 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
31270 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
31271 (MRRCI, mrrc, MRRC): New.
31272 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
31273 VUNSPEC_MRRC2): New.
31275 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31277 * config/arm/arm.md (<mcr>): New.
31279 * config/arm/arm.c (arm_coproc_builtin_available): Add
31280 support for mcr, mrc, mcr2 and mrc2.
31281 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
31282 (arm_mcr_qualifiers): ... this. New.
31283 (MRC_QUALIFIERS): Define to ...
31284 (arm_mrc_qualifiers): ... this. New.
31285 (MCR_QUALIFIERS): Define to ...
31286 (arm_mcr_qualifiers): ... this. New.
31287 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
31289 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
31290 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
31291 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
31292 VUNSPEC_MRC2): New.
31294 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31296 * config/arm/arm.md (*ldc): New.
31300 * config/arm/arm.c (arm_coproc_builtin_available): Add
31301 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
31302 (arm_coproc_ldc_stc_legitimate_address): New.
31303 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
31304 'qualifier_const_pointer'.
31305 (LDC_QUALIFIERS): Define to...
31306 (arm_ldc_qualifiers): ... this. New.
31307 (STC_QUALIFIERS): Define to...
31308 (arm_stc_qualifiers): ... this. New.
31309 * config/arm/arm-protos.h
31310 (arm_coproc_ldc_stc_legitimate_address): New.
31311 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
31312 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
31313 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
31314 stc2, stcl, stc2l): New.
31315 * config/arm/constraints.md (Uz): New.
31316 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
31317 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
31318 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
31319 VUNSPEC_STC2L): New.
31321 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31323 * config/arm/arm.md (<cdp>): New.
31324 * config/arm/arm.c (neon_const_bounds): Rename this ...
31325 (arm_const_bounds): ... this.
31326 (arm_coproc_builtin_available): New.
31327 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
31328 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
31329 (CDP_QUALIFIERS): Define to...
31330 (arm_cdp_qualifiers): ... this. New.
31332 (arm_expand_builtin_args): Add case for 6 arguments.
31333 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
31334 (arm_const_bounds): ... this.
31335 (arm_coproc_builtin_available): New.
31336 * config/arm/arm_acle.h (__arm_cdp): New.
31338 * config/arm/arm_acle_builtins.def (cdp): New.
31340 * config/arm/iterators.md (CDPI,CDP,cdp): New.
31341 * config/arm/neon.md: Rename all 'neon_const_bounds' to
31342 'arm_const_bounds'.
31343 * config/arm/types.md (coproc): New.
31344 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
31345 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
31346 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
31347 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
31349 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31351 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
31352 (UBINOP_QUALIFIERS): New.
31354 (acle_builtin_data): New. Change comment.
31355 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
31356 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
31357 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
31358 arm_acle_builtins.def.
31359 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
31360 (arm_init_acle_builtins): New.
31361 (CRC32_BUILTIN): Remove.
31362 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
31363 crc32cb, crc32ch and crc32cw.
31364 (arm_init_crc32_builtins): Remove.
31365 (arm_init_builtins): Use arm_init_acle_builtins rather
31366 than arm_init_crc32_builtins.
31367 (arm_expand_acle_builtin): New.
31368 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
31369 * config/arm/arm_acle_builtins.def: New.
31371 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
31373 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
31374 (arm_builtin_datum): ... this.
31375 (arm_init_neon_builtin): Rename to ...
31376 (arm_init_builtin): ... this. Add a new parameters PREFIX
31377 and USE_SIG_IN_NAME.
31378 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
31379 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
31380 'arm_builtin_datum'.
31381 (arm_init_vfp_builtins): Likewise.
31382 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
31383 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
31384 (arm_expand_neon_args): Rename to ...
31385 (arm_expand_builtin_args): ... this. Rename builtin_arg
31386 enum values and differentiate between ARG_BUILTIN_MEMORY
31387 and ARG_BUILTIN_NEON_MEMORY.
31388 (arm_expand_neon_builtin_1): Rename to ...
31389 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
31390 values, arm_expand_builtin_args and add bool parameter NEON.
31391 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
31392 (arm_expand_vfp_builtin): Likewise.
31393 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
31395 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
31397 PR middle-end/77484
31398 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
31399 * predict.c (tree_estimate_probability_bb): Reverse direction of
31400 polymorphic call predictor.
31402 2017-01-06 David Malcolm <dmalcolm@redhat.com>
31404 * passes.c (execute_one_pass): Split out pass-skipping logic into...
31405 (determine_pass_name_match): ...this new function and...
31406 (should_skip_pass_p): ...this new function.
31408 2017-01-06 Nathan Sidwell <nathan@acm.org>
31410 * ipa-visibility.c (function_and_variable_visibility): Reformat
31411 comments and long lines. Remove extrneous if.
31412 * symtab.c (symtab_node::make_decl_local): Fix code format.
31413 (symtab_node::set_section_for_node): Fix comment typo.
31415 2017-01-06 Martin Liska <mliska@suse.cz>
31418 * lra-constraints.c: Rename invariant to lra_invariant.
31419 * predict.c (set_even_probabilities): Initialize e to NULL.
31421 2017-01-05 Martin Sebor <msebor@redhat.com>
31423 PR tree-optimization/78910
31424 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
31425 (format_integer): Correct off-by-one error in the handling
31426 of precision with negative numbers in signed conversions..
31428 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
31430 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
31432 2017-01-05 Jakub Jelinek <jakub@redhat.com>
31434 PR tree-optimization/71016
31435 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
31436 factor_out_conditional_conversion. Formatting fix.
31437 (factor_out_conditional_conversion): Add cond_stmt argument.
31438 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
31439 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
31442 2017-01-05 David Malcolm <dmalcolm@redhat.com>
31444 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
31445 read-rtl-function.o, and selftest-rtl.o.
31446 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
31447 (selftest::aarch64_test_loading_full_dump): New function.
31448 (selftest::aarch64_run_selftests): New function.
31449 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
31450 selftest::aarch64_run_selftests.
31451 * config/i386/i386.c
31452 (selftest::ix86_test_loading_dump_fragment_1): New function.
31453 (selftest::ix86_test_loading_call_insn): New function.
31454 (selftest::ix86_test_loading_full_dump): New function.
31455 (selftest::ix86_test_loading_unspec): New function.
31456 (selftest::ix86_run_selftests): Call the new functions.
31457 * emit-rtl.c (maybe_set_max_label_num): New function.
31458 * emit-rtl.h (maybe_set_max_label_num): New decl.
31459 * function.c (instantiate_decls): Guard call to
31460 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
31461 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
31463 * gensupport.c (gen_reader::gen_reader): Pass "false"
31464 for new "compact" param of rtx_reader.
31465 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
31466 rather than an empty string for NULL strings.
31467 * read-md.c: Potentially include config.h rather than bconfig.h.
31468 Wrap include of errors.h with #ifdef GENERATOR_FILE.
31469 (have_error): New global, copied from errors.c.
31470 (md_reader::read_name): Rename to...
31471 (md_reader::read_name_1): ...this, adding "out_loc" param,
31472 and converting "missing name or number" to returning false, rather
31474 (md_reader::read_name): Reimplement in terms of read_name_1.
31475 (md_reader::read_name_or_nil): New function.
31476 (md_reader::read_string): Handle "(nil)" by returning NULL.
31477 (md_reader::md_reader): Add new param "compact".
31478 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
31479 (md_reader::read_file): New method.
31480 * read-md.h (md_reader::md_reader): Add new param "compact".
31481 (md_reader::read_file): New method.
31482 (md_reader::is_compact): New accessor.
31483 (md_reader::read_name): Convert return type from void to file_location.
31484 (md_reader::read_name_or_nil): New decl.
31485 (md_reader::read_name_1): New decl.
31486 (md_reader::m_compact): New field.
31487 (noop_reader::noop_reader): Pass "false" for new "compact" param
31489 (rtx_reader::rtx_reader): Add new "compact" param.
31490 (rtx_reader::read_rtx_operand): Make virtual and convert return
31491 type from void to rtx.
31492 (rtx_reader::read_until): New decl.
31493 (rtx_reader::handle_any_trailing_information): New virtual function.
31494 (rtx_reader::postprocess): New virtual function.
31495 (rtx_reader::finalize_string): New virtual function.
31496 (rtx_reader::m_in_call_function_usage): New field.
31497 (rtx_reader::m_reuse_rtx_by_id): New field.
31498 * read-rtl-function.c: New file.
31499 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
31500 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
31501 (selftest::verify_three_block_rtl_cfg): New decl.
31502 * read-rtl-function.h: New file.
31503 * read-rtl.c: Potentially include config.h rather than bconfig.h.
31504 For host, include function.h, memmodel.h, and emit-rtl.h.
31505 (one_time_initialization): New function.
31506 (struct compact_insn_name): New struct.
31507 (compact_insn_names): New array.
31508 (find_code): Handle insn codes in compact dumps.
31509 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
31510 (bind_subst_iter_and_attr): Likewise.
31511 (add_condition_to_string): Likewise.
31512 (add_condition_to_rtx): Likewise.
31513 (apply_attribute_uses): Likewise.
31514 (add_current_iterators): Likewise.
31515 (apply_iterators): Likewise.
31516 (initialize_iterators): Guard usage of apply_subst_iterator with
31517 #ifdef GENERATOR_FILE.
31518 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
31519 (md_reader::read_mapping): Likewise.
31520 (add_define_attr_for_define_subst): Likewise.
31521 (add_define_subst_attr): Likewise.
31522 (read_subst_mapping): Likewise.
31523 (check_code_iterator): Likewise.
31524 (rtx_reader::read_rtx): Likewise. Move one-time initialization
31526 (one_time_initialization): New function.
31527 (rtx_reader::read_until): New method.
31528 (read_flags): New function.
31529 (parse_reg_note_name): New function.
31530 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
31531 Handle reuse_rtx ids.
31532 Wrap iterator lookup within #ifdef GENERATOR_FILE.
31533 Add parsing support for RTL dumps, mirroring the special-cases in
31534 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
31535 values, and calling handle_any_trailing_information.
31536 (rtx_reader::read_rtx_operand): Convert return type from void
31537 to rtx, returning return_rtx. Handle case 'e'. Call
31538 finalize_string on XSTR and XTMPL fields.
31539 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
31540 "(nil)" values were omitted. Call the postprocess vfunc on the
31542 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
31543 class ctor. Initialize m_in_call_function_usage. Call
31544 one_time_initialization.
31545 * rtl-tests.c (selftest::test_uncond_jump): Call
31546 set_new_first_and_last_insn.
31547 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
31548 * selftest-rtl.c: New file.
31549 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
31550 (selftest::get_insn_by_uid): New decl.
31551 * selftest-run-tests.c (selftest::run_tests): Call
31552 read_rtl_function_c_tests.
31553 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
31554 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
31557 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
31559 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
31560 operands in a special way. Assert that pos+len <= mode precision.
31562 2017-01-05 Jakub Jelinek <jakub@redhat.com>
31564 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
31565 3 argument Alias with unlimited for the negative form.
31566 (fno-vect-cost-model): Removed.
31568 2017-01-05 Martin Liska <mliska@suse.cz>
31570 * hsa-gen.c (gen_hsa_divmod): New function.
31571 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
31573 2017-01-05 Martin Liska <mliska@suse.cz>
31576 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
31579 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31581 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
31582 small constant length operands.
31584 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31586 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
31587 between loop iterations.
31589 2017-01-05 Martin Liska <mliska@suse.cz>
31592 * gimplify.c (gimplify_decl_expr): Compare to
31593 asan_poisoned_variables instread of checking flags.
31594 (gimplify_target_expr): Likewise.
31595 (gimplify_expr): Likewise.
31596 (gimplify_function_tree): Conditionally initialize
31597 asan_poisoned_variables.
31599 2017-01-04 Jeff Law <law@redhat.com>
31601 PR tree-optimizatin/78812
31602 * rtl.h (contains_mem_rtx_p): Prototype.
31603 * ifcvt.c (containts_mem_rtx_p): Move from here to...
31604 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
31605 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
31606 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
31607 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
31609 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
31611 * input.c (assert_char_at_range): Default-initialize actual_range.
31613 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
31615 * df-scan.c (df_ref_create_structure): Make regno unsigned,
31616 to match the caller.
31618 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
31620 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
31621 insns after final jump in test to emit dummy move.
31623 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
31625 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
31626 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
31628 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
31630 * multiple_target.c (create_dispatcher_calls): Init e_next.
31631 * tree-ssa-loop-split.c (split_loop): Init border.
31632 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
31635 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
31640 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
31641 (altivec_register_operand): Do not return true if the operand
31642 contains a SUBREG mixing SImode and SFmode.
31643 (vsx_register_operand): Likewise.
31644 (vsx_reg_sfsubreg_ok): New predicate.
31645 (vfloat_operand): Do not return true if the operand contains a
31646 SUBREG mixing SImode and SFmode.
31647 (vint_operand): Likewise.
31648 (vlogical_operand): Likewise.
31649 (gpc_reg_operand): Likewise.
31650 (int_reg_operand): Likewise.
31651 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
31652 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
31653 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
31655 (rs6000_emit_move_si_sf_subreg): New helper function.
31656 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
31657 fixup SUBREGs involving SImode and SFmode.
31658 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
31659 numbers for the new peephole2 optimization.
31660 (peephole2 for SFmode unions): New peephole2 to optimize cases in
31661 the GLIBC math library that do AND/IOR/XOR operations on single
31662 precision floating point.
31663 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
31664 target macros to say whether we need to avoid SUBREGs mixing
31666 (TARGET_ALLOW_SF_SUBREG): Likewise.
31667 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
31668 (UNSPEC_SI_FROM_SF): Likewise.
31669 (iorxor): Change spacing.
31670 (and_ior_xor): New iterator for AND, IOR, and XOR.
31671 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
31672 (movdi_from_sf_zero_ext): Likewise.
31673 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
31674 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
31675 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
31676 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
31677 (fms<mode>4): Likewise.
31678 (fnma<mode>4): Likewise.
31679 (fnms<mode>4): Likewise.
31680 (nfma<mode>4): Likewise.
31681 (nfms<mode>4): Likewise.
31683 2017-01-04 Marek Polacek <polacek@redhat.com>
31686 * doc/invoke.texi: Document -Wpointer-compare.
31688 2017-01-04 Jakub Jelinek <jakub@redhat.com>
31690 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
31693 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
31694 descriptions for -gdwarf-5 and emit them as uleb128 instead of
31697 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
31700 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
31701 documentation of the powerpc_popcntb_ok attribute.
31702 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
31703 code to issue warning messages if a requested CPU configuration is
31704 not supported by the binary (assembler and loader) toolchain.
31705 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
31706 made to define a built-in function that has been disabled.
31707 (paired_init_builtins): Add assertion to prevent ICE if attempt is
31708 made to define a built-in function that has been disabled.
31709 (altivec_init_builtins): Add comment explaining why definition
31710 of the DST built-in functions is not preceded by an assertion
31711 check. Add assertions to prevent ICE if attempts are made to
31712 define an altivec predicate or an abs* built-in function that has
31714 (htm_init_builtins): Add comment explaining why definition of the
31715 htm built-in functions is not preceded by an assertion check.
31717 2017-01-04 Jeff Law <law@redhat.com>
31719 PR tree-optimizatin/67955
31720 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
31721 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
31722 the points-to solution does not include pt_null. Use DECL_PT_UID
31725 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
31727 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
31728 Use gen_int_mode instead of gen_lopwart for const_int operands.
31730 2017-01-04 Jakub Jelinek <jakub@redhat.com>
31732 PR tree-optimization/71563
31733 * match.pd: Simplify X << Y into X if Y is known to be 0 or
31734 out of range value - has low bits known to be zero.
31736 2017-01-04 Alan Modra <amodra@gmail.com>
31738 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
31739 * configure: Regenerate.
31740 * config.in: Regenerate.
31742 2017-01-04 Jakub Jelinek <jakub@redhat.com>
31745 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
31746 a substring of the message, but strcmp with the whole message. Ifdef
31747 ENABLE_NLS, translate the message first using dgettext.
31749 2017-01-03 Jeff Law <law@redhat.com>
31751 PR tree-optimizatin/78856
31752 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
31753 (mark_threaded_blocks): Remove code to truncate thread paths that
31754 cross multiple loop headers. Instead invalidate the cached loop
31755 iteration information and handle case of a thread path walking
31756 into an irreducible region.
31758 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
31761 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
31762 assertions. Add support for doing the signbit if the IEEE 128-bit
31763 floating point value is in a GPR.
31764 * config/rs6000/rs6000.md (Fsignbit): Delete.
31765 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
31766 Update the length attribute if the value is in a GPR.
31767 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
31768 the sign or zero extension instruction, since the value is always 0/1.
31769 (signbit<mode>2_dm2): Delete using <Fsignbit>.
31772 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
31773 extracting SImode to a GPR register so that we can generate a
31774 store, limit the vector to be in a traditional Altivec register
31775 for the vextuwrx instruction.
31777 2017-01-03 Ian Lance Taylor <iant@google.com>
31779 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
31781 2017-01-03 Martin Sebor <msebor@redhat.com>
31783 PR tree-optimization/78696
31784 * gimple-ssa-sprintf.c (format_floating): Correct handling of
31785 precision. Use MPFR for %f for greater fidelity. Correct handling
31787 (pass_sprintf_length::compute_format_length): Set width and precision
31788 specified by asrerisk to void_node for vararg functions.
31789 (try_substitute_return_value): Adjust dump output.
31791 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
31793 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
31795 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
31797 * doc/invoke.texi (SPARC options): Document -mlra as the default.
31798 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
31799 -mlra/-mno-lra was passed to the compiler.
31801 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
31803 PR rtl-optimization/65618
31804 * emit-rtl.c (try_split): Move initialization of "before" and
31805 "after" to just before the call to emit_insn_after_setloc.
31807 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
31809 * doc/md.texi (Standard Names): Remove reference to Java frontend.
31811 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
31813 * dwarf2out.c (gen_enumeration_type_die): When
31814 -gno-strict-dwarf, add a DW_AT_encoding attribute.
31816 2017-01-03 Jakub Jelinek <jakub@redhat.com>
31818 PR tree-optimization/78965
31819 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
31820 Change first argument from const call_info & to call_info &. For %n
31821 set info.nowrite to false.
31823 PR middle-end/78901
31824 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
31825 possibly throwing calls.
31827 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
31828 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
31829 and fns handling, rather than in a separate case SSA_NAME.
31831 2017-01-02 Jeff Law <law@redhat.com>
31833 * config/darwin-driver.c (darwin_driver_init): Const-correctness
31834 fixes for first_period and second_period variables.
31836 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
31839 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
31840 (*insvqi_1): New insn pattern.
31841 (*insvqi_1_mem_rex64): Ditto.
31842 (*insvqi_2): Ditto.
31843 (*insvqi_3): Rename from *insvqi.
31845 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
31847 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
31849 * doc/cfg.texi (Edges): Remove reference to Java.
31850 (Maintaining the CFG): Ditto.
31852 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
31854 PR middle-end/77674
31855 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
31856 transparent aliases.
31858 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
31860 PR middle-end/77484
31861 * predict.def (PRED_CALL): Update hitrate.
31862 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
31863 * predict.c (tree_estimate_probability_bb): Split CALL predictor
31864 into direct/indirect/polymorphic variants.
31866 2017-01-01 Jakub Jelinek <jakub@redhat.com>
31868 Update copyright years.
31870 * gcc.c (process_command): Update copyright notice dates.
31871 * gcov-dump.c (print_version): Ditto.
31872 * gcov.c (print_version): Ditto.
31873 * gcov-tool.c (print_version): Ditto.
31874 * gengtype.c (create_file): Ditto.
31875 * doc/cpp.texi: Bump @copying's copyright year.
31876 * doc/cppinternals.texi: Ditto.
31877 * doc/gcc.texi: Ditto.
31878 * doc/gccint.texi: Ditto.
31879 * doc/gcov.texi: Ditto.
31880 * doc/install.texi: Ditto.
31881 * doc/invoke.texi: Ditto.
31883 Copyright (C) 2017 Free Software Foundation, Inc.
31885 Copying and distribution of this file, with or without modification,
31886 are permitted in any medium without royalty provided the copyright
31887 notice and this notice are preserved.