1 2017-11-09 Jakub Jelinek <jakub@redhat.com>
3 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
4 (split_group): Add total_orig and total_new arguments, estimate the
5 number of statements related to the store group without store merging
6 and with store merging.
7 (imm_store_chain_info::output_merged_store): Adjust split_group
8 callers, punt if estimated number of statements with store merging
9 is not smaller than estimated number of statements without it.
11 (handled_load): Remove has_single_use checks.
12 (pass_store_merging::process_store): Likewise.
14 2017-11-09 Richard Biener <rguenther@suse.de>
16 PR tree-optimization/82902
17 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
19 2017-11-09 Martin Liska <mliska@suse.cz>
22 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
25 2017-11-09 Martin Liska <mliska@suse.cz>
27 PR tree-optimization/82669
28 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
30 2017-11-09 Martin Liska <mliska@suse.cz>
33 * coverage.c (coverage_begin_function): Output also end locus
34 of a function and information whether the function is
36 * gcov-dump.c (tag_function): Parse and print the information.
37 * gcov.c (INCLUDE_MAP): Add include.
38 (INCLUDE_SET): Likewise.
39 (struct line_info): Move earlier in the source file because
40 of vector<line_info> in function_info structure.
41 (line_info::line_info): Likewise.
42 (line_info::has_block): Likewise.
43 (struct source_info): Add new member index.
44 (source_info::get_functions_at_location): New function.
45 (function_info::group_line_p): New function.
46 (output_intermediate_line): New function.
47 (output_intermediate_file): Use the mentioned function.
48 (struct function_start): New.
49 (struct function_start_pair_hash): Likewise.
50 (process_file): Add code that identifies group functions.
51 Assign lines either to global or function scope.
52 (generate_results): Skip artificial functions.
53 (find_source): Assign index for each source file.
54 (read_graph_file): Read new flag artificial and end_line.
55 (add_line_counts): Assign it either to global of function scope.
56 (accumulate_line_counts): Isolate core of the function to
57 accumulate_line_info and call it for both function and global
59 (accumulate_line_info): New function.
60 (output_line_beginning): Fix GNU coding style.
61 (print_source_line): New function.
62 (output_line_details): Likewise.
63 (output_function_details): Likewise.
64 (output_lines): Iterate both source (global) scope and function
66 (struct function_line_start_cmp): New class.
67 * doc/gcov.texi: Reflect changes in documentation.
69 2017-11-09 Jakub Jelinek <jakub@redhat.com>
72 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
73 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
74 and similarly for not instead of neg.
76 2017-11-08 Andi Kleen <ak@linux.intel.com>
78 * config/i386/i386.opt: Add -mforce-indirect-call.
79 * config/i386/predicates.md: Check for flag_force_indirect_call.
80 * doc/invoke.texi: Document -mforce-indirect-call
82 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
84 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
86 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
87 config/riscv/riscv.c (predict.h): New include.
88 (riscv_slow_unaligned_access_p): No longer static.
89 (riscv_block_move_straight): Add require.
90 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
92 2017-11-08 Jakub Jelinek <jakub@redhat.com>
95 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
96 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
97 nonimmediate_operand predicate for operand 1 instead of
100 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
102 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
104 * config/aarch64/constraints.md (Uml): New constraint.
105 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
108 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
110 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
111 of two vec_duplicates into a vec_concat.
113 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
115 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
116 Simplify vec_merge of vec_duplicate and vec_concat.
117 * config/aarch64/constraints.md (Utq): New constraint.
118 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
121 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
123 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
124 Simplify vec_merge of vec_duplicate and const_vector.
125 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
127 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
128 mode iterator. Update predicate on operand 1 to
129 handle non-const_vec constants. Delete constraints.
130 (*aarch64_combinez_be<mode>): Likewise for operand 2.
132 2017-11-08 Jakub Jelinek <jakub@redhat.com>
134 PR tree-optimization/78821
135 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
137 (store_operand_info::store_operand_info): Initialize it to false.
138 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
139 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
140 store in the group, and if chain_info is non-NULL, to ignore altogether
142 (compatible_load_p): Fail if bit_not_p does not match.
143 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
144 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
145 (pass_store_merging::process_store): Adjust
146 terminate_all_aliasing_chains calls to pass NULL in all current spots,
147 call terminate_all_aliasing_chains newly when adding a store into
148 a chain with non-NULL chain_info.
150 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
152 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
154 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
156 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
158 (aarch64_layout_frame): Initialise emit_frame_chain.
159 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
160 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
162 2017-11-08 Martin Liska <mliska@suse.cz>
164 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
165 of gimple_call_internal_p.
167 2017-11-07 Tom de Vries <tom@codesourcery.com>
169 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
171 2017-11-07 Tom de Vries <tom@codesourcery.com>
173 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
176 2017-11-08 Martin Liska <mliska@suse.cz>
179 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
181 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
183 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
185 2017-11-07 Jakub Jelinek <jakub@redhat.com>
188 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
189 (*cmp<mode>_ccz_1): New insn with $k alternative.
192 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
193 RTX_COMM_COMPARE as commutative as well.
194 (ix86_binary_operator_ok): Formatting fix.
195 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
196 *<code><mode>3<mask_name><round_saeonly_name>,
197 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
198 *<s>mul<mode>3_highpart<mask_name>,
199 *vec_widen_umult_even_v16si<mask_name>,
200 *vec_widen_umult_even_v8si<mask_name>,
201 *vec_widen_umult_even_v4si<mask_name>,
202 *vec_widen_smult_even_v16si<mask_name>,
203 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
204 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
205 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
206 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
207 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
208 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
209 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
210 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
211 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
212 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
213 ix86_binary_operator_ok. Formatting fixes.
214 (*<plusminus_insn><mode>3<mask_name><round_name>,
215 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
218 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
220 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
221 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
222 eq and ne if TARGET_ISEL.
223 (cmp): New code_iterator.
224 (UNS, UNSU_, UNSIK): New code_attrs.
225 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
226 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
228 ("eq<mode>3"): ... this.
229 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
231 ("ne<mode>3"): ... this.
233 2017-11-07 Julia Koval <julia.koval@intel.com>
236 * common/config/i386/i386-common.c
237 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
238 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
239 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
240 * config/i386/i386.opt: Ditto.
241 * config/i386/i386.c (ix86_target_string): Ditto.
242 (ix86_option_override_internal): Ditto.
243 (ix86_init_mpx_builtins): Move MPX to args2.
244 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
245 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
246 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
247 __builtin_ia32_vgf2p8affineinvqb_v32qi,
248 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
249 __builtin_ia32_vgf2p8affineinvqb_v16qi,
250 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
252 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
255 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
256 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
257 (zero-extendsidi peephole2): Remove peephole.
259 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
262 * stor-layout.c: Include attribs.h.
263 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
264 explicit lookup of "aligned" attribute.
266 2017-11-07 Andrew Waterman <andrew@sifive.com>
268 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
269 (riscv_expand_block_move): Likewise.
270 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
272 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
273 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
274 * config/riscv/riscv.c (riscv_block_move_straight): New function.
275 (riscv_adjust_block_mem): Likewise.
276 (riscv_block_move_loop): Likewise.
277 (riscv_expand_block_move): Likewise.
278 gcc/config/riscv/riscv.md (movmemsi): New pattern.
280 2017-11-07 Michael Clark <michaeljclark@mac.com>
282 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
283 (MUSL_DYNAMIC_LINKER): Likewise.
285 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
287 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
288 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
291 2017-11-07 Richard Biener <rguenther@suse.de>
293 * match.pd: Fix build.
295 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
296 Jackson Woodruff <jackson.woodruff@arm.com>
298 PR tree-optimization/71026
299 * match.pd: Canonicalize negate in division.
301 2017-11-07 Sudakshina Das <sudi.das@arm.com>
304 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
306 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
308 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
309 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
311 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
313 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
314 non-scalar integral types.
315 * match.pd (negate_expr_p): Handle MINUS_EXPR.
316 (-(A-B), -(~A)): New transformations.
318 2017-11-07 Tom de Vries <tom@codesourcery.com>
320 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
321 semicolon after "do {} while (0)".
322 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
323 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
324 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
325 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
326 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
328 2017-11-07 Tom de Vries <tom@codesourcery.com>
330 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
331 after "do {} while (0)".
332 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
333 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
334 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
335 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
336 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
338 2017-11-07 Tom de Vries <tom@codesourcery.com>
341 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
343 (arm_rtx_costs_internal): Add missing semicolon after
344 HANDLE_NARROW_SHIFT_ARITH call.
346 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
348 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
349 disable isel if it was not set explicitly.
351 2017-11-06 James Bowman <james.bowman@ftdichip.com>
353 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
355 (add_type_attribute) likewise.
357 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
359 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
360 of struct ix86_frame.
361 (ix86_initial_elimination_offset): Likewise.
362 (ix86_expand_split_stack_prologue): Likewise.
364 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
366 * tree-vrp.h (enum value_range_type): Update stale comment.
368 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
369 Alan Hayward <alan.hayward@arm.com>
370 David Sherwood <david.sherwood@arm.com>
372 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
373 (aarch64_expand_vec_perm_const): Take the number of units too.
374 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
375 (aarch64_expand_vec_perm_const): Likewise.
376 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
377 (vec_perm<mode>): Update accordingly.
379 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
380 Alan Hayward <alan.hayward@arm.com>
381 David Sherwood <david.sherwood@arm.com>
383 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
384 Take the number of units too.
385 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
386 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
387 but check for a vector mode before rather than after the call.
388 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
389 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
390 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
391 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
392 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
393 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
394 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
395 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
396 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
397 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
398 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
399 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
400 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
401 (aarch64_sqdmull2_n<mode>): Update accordingly.
403 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
404 Alan Hayward <alan.hayward@arm.com>
405 David Sherwood <david.sherwood@arm.com>
407 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
408 the number of units too.
409 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
410 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
411 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
412 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
413 (vec_store_lanesxi<mode>): Update accordingly.
415 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
416 Alan Hayward <alan.hayward@arm.com>
417 David Sherwood <david.sherwood@arm.com>
419 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
420 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
421 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
422 of units rather than the mode.
423 * config/aarch64/iterators.md (nunits): New mode attribute.
424 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
425 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
426 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
427 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
428 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
429 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
430 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
431 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
432 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
433 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
434 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
435 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
436 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
437 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
438 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
439 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
440 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
441 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
442 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
443 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
444 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
445 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
446 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
447 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
448 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
449 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
450 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
451 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
452 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
453 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
454 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
455 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
456 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
457 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
458 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
459 (aarch64_simd_vec_setv2di): Likewise.
461 2017-11-06 Carl Love <cel@us.ibm.com>
463 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
465 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
466 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
468 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
470 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
471 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
472 Add power 8 macro expansions.
473 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
474 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
475 power 8 instructions. (VSX_XXBR): Add iterator.
477 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
479 * config/arm/arm.md (predicable_short_it): Change default to "no",
480 improve documentation, remove uses that are identical to the default.
481 (enabled_for_depr_it): Rename to enabled_for_short_it.
482 * gcc/config/arm/arm-fixed.md (predicable_short_it): Remove default uses.
483 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
484 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
485 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
486 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
488 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
491 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
492 float128 helper macros, which are no longer used after deleting
493 the old 'q' built-in functions, and moving the round to odd
494 built-in functions to being special built-in functions.
495 (BU_FLOAT128_2): Likewise.
496 (BU_FLOAT128_1_HW): Likewise.
497 (BU_FLOAT128_2_HW): Likewise.
498 (BU_FLOAT128_3_HW): Likewise.
499 (FABSQ): Delete old 'q' built-in functions.
500 (COPYSIGNQ): Likewise.
501 (SQRTF128_ODD): Move round to odd built-in functions to be
502 special built-in functions, so that we can handle
503 -mabi=ieeelongdouble.
504 (TRUNCF128_ODD): Likewise.
505 (ADDF128_ODD): Likewise.
506 (SUBF128_ODD): Likewise.
507 (MULF128_ODD): Likewise.
508 (DIVF128_ODD): Likewise.
509 (FMAF128_ODD): Likewise.
510 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
511 built-in names to 'f128'.
512 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
513 old 'q' built-in functions, as the machine independent code for
514 'f128' built-in functions handles this.
515 (rs6000_expand_builtin): Add expansion for float128 round to odd
516 functions, keying off on -mabi=ieeelongdouble of whether to use
517 the KFmode or TFmode variant.
518 (rs6000_init_builtins): Initialize the _Float128 round to odd
520 * doc/extend.texi (PowerPC Built-in Functions): Document the old
521 _Float128 'q' built-in functions are now mapped into the new
522 'f128' built-in functions.
524 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
526 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
527 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
529 2017-11-06 David Malcolm <dmalcolm@redhat.com>
532 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
533 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
534 * toplev.c: Include "ipa-fnsummary.h".
535 (toplev::finalize): Call ipa_fnsummary_c_finalize.
537 2017-11-06 Jakub Jelinek <jakub@redhat.com>
539 PR tree-optimization/82838
540 * gimple-ssa-store-merging.c
541 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
542 on a separate gimple_seq which is then appended to seq.
544 2017-11-06 Jeff Law <law@redhat.com>
547 * config/i386/i386.c (PROBE_INTERVAL): Remove.
548 (get_probe_interval): New functions.
549 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
550 (ix86_adjust_stack_and_probe): Likewise.
551 (output_adjust_stack_and_probe): Likewise.
552 (ix86_emit_probe_stack_range): Likewise.
553 (ix86_expand_prologue): Likewise.
555 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
557 PR tree-optimization/82816
558 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
559 if the modes of the two types are the same.
560 (convert_plusminus_to_widen): Likewise.
562 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
564 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
566 (usadv16qi): New define_expand.
567 (usadv8hi): New define_expand.
569 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
572 * ipa-inline-transform.c (update_noncloned_frequencies): Always
574 (inline_transform): Likewise.
575 * predict.c (counts_to_freqs): Remove useless conditional.
576 * profile-count.h (profile_count::apply_scale): Move sanity check.
577 * tree-inline.c (copy_bb): Always scale.
578 (copy_cfg_body): Likewise.
580 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
583 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
585 (*cmp_ite0): Add enabled_for_depr_it attribute.
586 (*cmp_ite1): Likewise.
588 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
590 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
593 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
595 * tree-vrp.c (vrp_int_const_binop): Return true on success and
596 return the value by pointer.
597 (extract_range_from_multiplicative_op_1): Update accordingly.
598 Return as soon as an operation fails.
600 2017-11-05 Tom de Vries <tom@codesourcery.com>
603 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
604 (DEF_SANITIZER_BUILTIN): ... here.
605 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
606 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
608 2017-11-05 Tom de Vries <tom@codesourcery.com>
611 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
613 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
614 ASM_OUTPUT_BEFORE_CASE_LABEL call.
615 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
617 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
618 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
619 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
621 2017-11-05 Tom de Vries <tom@codesourcery.com>
624 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
627 2017-11-04 Michael Clark <michaeljclark@mac.com>
629 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
630 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
632 (*addsi3_extended): Likewise.
633 (*addsi3_extended2): Likewise.
634 (<optab>si3): Likewise.
635 (<optab>di3): Likewise.
636 (<optab><mode>3): Likewise.
637 (<*optabe>si3_internal): Likewise.
638 (zero_extendqi<SUPERQI:mode>2): Likewise.
639 (*add<mode>hi3): Likewise.
640 (*xor<mode>hi3): Likewise.
641 (<optab>di3): Likewise.
642 (*<optab>si3_extend): Likewise.
643 (*sge<u>_<X:mode><GPR:mode>): Likewise.
644 (*slt<u>_<X:mode><GPR:mode>): Likewise.
645 (*sle<u>_<X:mode><GPR:mode>): Likewise.
647 2017-11-04 Andrew Waterman <andrew@sifive.com>
649 * config/riscv/riscv.c (riscv_option_override): Conditionally set
650 TARGET_STRICT_ALIGN based upon -mtune argument.
652 2017-11-04 Andrew Waterman <andrew@sifive.com>
654 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
656 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
658 config/i386/i386.c (choose_basereg): Use optional scratch
659 register and add assertion.
660 (x86_emit_outlined_ms2sysv_save): Use scratch register when
661 needed, and don't allocate stack.
662 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
663 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
664 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
666 2017-11-03 Jeff Law <law@redhat.com>
668 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
669 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
670 to probe at the start of a noreturn function.
672 2017-11-03 Jakub Jelinek <jakub@redhat.com>
674 PR tree-optimization/78821
675 * gimple-ssa-store-merging.c: Update the file comment.
676 (MAX_STORE_ALIAS_CHECKS): Define.
677 (struct store_operand_info): New type.
678 (store_operand_info::store_operand_info): New constructor.
679 (struct store_immediate_info): Add rhs_code and ops data members.
680 (store_immediate_info::store_immediate_info): Add rhscode, op0r
681 and op1r arguments to the ctor, initialize corresponding data members.
682 (struct merged_store_group): Add load_align_base and load_align
684 (merged_store_group::merged_store_group): Initialize them.
685 (merged_store_group::do_merge): Update them.
686 (merged_store_group::apply_stores): Pick the constant for
687 encode_tree_to_bitpos from one of the two operands, or skip
688 encode_tree_to_bitpos if neither operand is a constant.
689 (class pass_store_merging): Add process_store method decl. Remove
690 bool argument from terminate_all_aliasing_chains method decl.
691 (pass_store_merging::terminate_all_aliasing_chains): Remove
692 var_offset_p argument and corresponding handling.
693 (stmts_may_clobber_ref_p): New function.
694 (compatible_load_p): New function.
695 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
696 if there is overlap and rhs_code is not INTEGER_CST. For
697 non-overlapping stores terminate group if rhs is not mergeable.
698 (get_alias_type_for_stmts): Change first argument from
699 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
700 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
701 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
703 (get_location_for_stmts): Change first argument from
704 auto_vec<gimple *> & to vec<gimple *> &.
705 (struct split_store): Remove orig_stmts data member, add orig_stores.
706 (split_store::split_store): Create orig_stores rather than orig_stmts.
707 (find_constituent_stmts): Renamed to ...
708 (find_constituent_stores): ... this. Change second argument from
709 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
710 to info structures rather than the statements.
711 (split_group): Rename ALLOW_UNALIGNED argument to
712 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
713 it. Adjust find_constituent_stores caller.
714 (imm_store_chain_info::output_merged_store): Handle rhs_code other
715 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
716 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
717 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
718 (mem_valid_for_store_merging): New function.
719 (handled_load): New function.
720 (pass_store_merging::process_store): New method.
721 (pass_store_merging::execute): Use process_store method. Adjust
722 terminate_all_aliasing_chains caller.
724 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
726 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
727 Return true for more constants, symbols and label references.
728 (aarch64_valid_floating_const): Remove unused function.
730 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
733 * config/aarch64/aarch64.c (aarch64_layout_frame):
734 Undo forcing of LR at bottom of frame.
736 2017-11-03 Jeff Law <law@redhat.com>
739 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
740 for int_registers_saved.
742 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
743 extracted from tree-ssa-dom.c.
744 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
745 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
746 (record_equivalences_from_incoming_edge): Add additional argument
747 to single_pred_edge_ignoring_loop_edges call.
748 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
749 (uncprop_dom_walker::before_dom_children): Add additional argument
750 to single_pred_edge_ignoring_loop_edges call.
751 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
752 single_pred_edge_ignoring_loop_edges rather than open coding.
753 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
755 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
757 * match.pd (-(-A)): Rewrite.
759 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
761 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
762 (rs6000_emit_int_cmove): New declaration.
763 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
764 (rs6000_emit_sISEL): Delete.
765 (rs6000_emit_int_cmove): Make non-static.
766 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
767 instead of rs6000_emit_sISEL.
769 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
771 * asan.c (create_cond_insert_point): Maintain profile.
772 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
774 * basic-block.h (struct basic_block_def): Remove frequency.
775 (EDGE_FREQUENCY): Use to_frequency
776 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
778 (find_traces): Update to use to_frequency.
779 (find_traces_1_round): Likewise; use only IPA counts.
780 (bb_to_key): Likewise.
781 (connect_traces): Use IPA counts only.
782 (copy_bb_p): Update to use to_frequency.
783 (fix_up_crossing_landing_pad): Likewise.
784 (sanitize_hot_paths): Likewise.
785 * bt-load.c (basic_block_freq): Likewise.
786 * cfg.c (init_flow): Set count_max to uninitialized.
787 (check_bb_profile): Remove frequencies; check counts.
788 (dump_bb_info): Do not dump frequencies.
789 (update_bb_profile_for_threading): Update counts only.
790 (scale_bbs_frequencies_int): Likewise.
791 (MAX_SAFE_MULTIPLIER): Remove.
792 (scale_bbs_frequencies_gcov_type): Update counts only.
793 (scale_bbs_frequencies_profile_count): Update counts only.
794 (scale_bbs_frequencies): Update counts only.
795 * cfg.h (struct control_flow_graph): Add count-max.
796 (update_bb_profile_for_threading): Update prototype.
797 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
798 (find_many_sub_basic_blocks): Likewise.
799 * cfgcleanup.c (try_forward_edges): Likewise.
800 (try_crossjump_to_edge): Likewise.
801 * cfgexpand.c (expand_gimple_cond): Likewise.
802 (expand_gimple_tailcall): Likewise.
803 (construct_init_block): Likewise.
804 (construct_exit_block): Likewise.
805 * cfghooks.c (verify_flow_info): Check consistency of counts.
806 (dump_bb_for_graph): Do not dump frequencies.
807 (split_block_1): Do not update frequencies.
808 (split_edge): Do not update frequencies.
809 (make_forwarder_block): Do not update frequencies.
810 (duplicate_block): Do not update frequencies.
811 (account_profile_record): Do not update frequencies.
812 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
813 for global heuristics.
814 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
815 (expected_loop_iterations_unbounded): Use counts only.
816 * cfgloopmanip.c (scale_loop_profile): Simplify.
817 (create_empty_loop_on_edge): Simplify
819 (duplicate_loop_to_header_edge): Simplify
820 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
821 (update_br_prob_note): Take care of removing note when profile
823 (relink_block_chain): Do not dump frequency.
824 (rtl_account_profile_record): Use to_frequency.
825 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
826 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
827 (cgraph_update_edges_for_call_stmt_node): Likewise.
828 (cgraph_edge::verify_count_and_frequency): Update.
829 (cgraph_node::verify_node): Temporarily disable frequency verification.
830 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
832 (cgraph_edge::rebuild_edges): Convert to ipa counts.
833 * cgraphunit.c (init_lowered_empty_function): Do not initialize
835 (cgraph_node::expand_thunk): Update profile.
836 * except.c (dw2_build_landing_pads): Do not update frequency.
837 * final.c (compute_alignments): Use to_frequency.
838 (dump_basic_block_info): Do not dump frequency.
839 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
840 (dump_gimple_bb_header): Do not dump frequency.
841 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
843 * gimple-streamer-in.c (input_bb): Do not stream frequency.
844 * gimple-streamer-out.c (output_bb): Do not stream frequency.
845 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
846 (init_before_recovery): Do not update frequency.
847 (sched_create_recovery_edges): Do not update frequency.
848 * hsa-gen.c (convert_switch_statements): Do not update frequency.
849 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
850 (ipa_cp_c_finalize): Set max_count to uninitialized.
851 * ipa-fnsummary.c (get_minimal_bb): Use counts.
852 (param_change_prob): Use counts.
853 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
855 * ipa-split.c (consider_split): Use to_frequency.
856 (split_function): Use to_frequency.
857 * ira-build.c (loop_compare_func): Likewise.
858 (mark_loops_for_removal): Likewise.
859 (mark_all_loops_for_removal): Likewise.
860 * loop-doloop.c (doloop_modify): Do not update frequency.
861 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
863 * lto-streamer-in.c (input_function): Update count_max.
864 * omp-expand.c (expand_omp_taskreg): Update count_max.
865 * omp-simd-clone.c (simd_clone_adjust): Update profile.
866 * predict.c (maybe_hot_frequency_p): Use to_frequency.
867 (maybe_hot_count_p): Use ipa counts only.
868 (maybe_hot_bb_p): Simplify.
869 (maybe_hot_edge_p): Simplify.
870 (probably_never_executed): Do not take frequency argument.
871 (probably_never_executed_bb_p): Do not pass frequency.
872 (probably_never_executed_edge_p): Likewise.
873 (combine_predictions_for_bb): Check that profile is nonzero.
874 (propagate_freq): Do not set frequency.
875 (drop_profile): Simplify.
876 (counts_to_freqs): Simplify.
877 (expensive_function_p): Use to_frequency.
878 (propagate_unlikely_bbs_forward): Simplify.
879 (determine_unlikely_bbs): Simplify.
880 (estimate_bb_frequencies): Add hack to silence graphite issues.
881 (compute_function_frequency): Use ipa counts.
882 (pass_profile::execute): Update.
883 (rebuild_frequencies): Use counts only.
884 (force_edge_cold): Use counts only.
885 * profile-count.c (profile_count::dump): Dump new count types.
886 (profile_count::differs_from_p): Check compatiblity.
887 (profile_count::to_frequency): New function.
888 (profile_count::to_cgraph_frequency): New function.
889 * profile-count.h (struct function): Declare.
890 (enum profile_quality): Add profile_guessed_local and
891 profile_guessed_global0.
892 (class profile_proability): Decrease number of bits to 29;
893 update from_reg_br_prob_note and to_reg_br_prob_note.
894 (class profile_count: Update comment; decrease number of bits
895 to 61. Check compatibility.
896 (profile_count::compatible_p): New private member function.
897 (profile_count::ipa_p): New member function.
898 (profile_count::operator<): Handle global zero correctly.
899 (profile_count::operator>): Handle global zero correctly.
900 (profile_count::operator<=): Handle global zero correctly.
901 (profile_count::operator>=): Handle global zero correctly.
902 (profile_count::nonzero_p): New member function.
903 (profile_count::force_nonzero): New member function.
904 (profile_count::max): New member function.
905 (profile_count::apply_scale): Handle IPA scalling.
906 (profile_count::guessed_local): New member function.
907 (profile_count::global0): New member function.
908 (profile_count::ipa): New member function.
909 (profile_count::to_frequency): Declare.
910 (profile_count::to_cgraph_frequency): Declare.
911 * profile.c (OVERLAP_BASE): Delete.
912 (compute_frequency_overlap): Delete.
913 (compute_branch_probabilities): Do not use compute_frequency_overlap.
914 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
915 * sched-ebb.c (rank): Use counts only.
916 * shrink-wrap.c (handle_simple_exit): Use counts only.
917 (try_shrink_wrapping): Use counts only.
918 (place_prologue_for_one_component): Use counts only.
919 * tracer.c (find_best_predecessor): Use to_frequency.
920 (find_trace): Use to_frequency.
921 (tail_duplicate): Use to_frequency.
922 * trans-mem.c (expand_transaction): Do not update frequency.
923 * tree-call-cdce.c: Do not update frequency.
924 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
925 (gimple_merge_blocks): Likewise.
926 (gimple_split_edge): Likewise.
927 (gimple_duplicate_sese_region): Likewise.
928 (gimple_duplicate_sese_tail): Likewise.
929 (move_sese_region_to_fn): Likewise.
930 (gimple_account_profile_record): Likewise.
931 (insert_cond_bb): Likewise.
932 * tree-complex.c (expand_complex_div_wide): Likewise.
933 * tree-eh.c (lower_resx): Update profile.
934 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
936 (initialize_cfun): Do not initialize frequencies
937 (freqs_to_counts): Delete.
938 (copy_cfg_body): Ignore count parameter.
940 (expand_call_inline): Update count_max.
941 (optimize_inline_calls): Update count_max.
942 (tree_function_versioning): Update count_max.
943 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
944 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
946 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
947 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
948 (try_peel_loop): Likewise.
949 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
951 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
952 (tree_transform_and_unroll_loop): Do not use frequencies
953 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
954 Use reliable prediction only.
955 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
956 * tree-ssa-sink.c (select_best_block): Use to_frequency.
957 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
959 * tree-ssa-threadupdate.c (create_block_for_threading): Do
961 (any_remaining_duplicated_blocks): Likewise.
962 (update_profile): Likewise.
963 (estimated_freqs_path): Delete.
964 (freqs_to_counts_path): Delete.
965 (clear_counts_path): Delete.
966 (ssa_fix_duplicate_block_edges): Likewise.
967 (duplicate_thread_path): Likewise.
968 * tree-switch-conversion.c (gen_inbound_check): Use counts.
969 * tree-tailcall.c (decrease_profile): Do not update frequency.
970 (eliminate_tail_call): Likewise.
971 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
972 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
973 (optimize_mask_stores): Likewise.
974 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
975 * ubsan.c (ubsan_expand_null_ifn): Update profile.
976 (ubsan_expand_ptr_ifn): Update profile.
977 * value-prof.c (gimple_ic): Simplify.
978 * value-prof.h (gimple_ic): Update prototype.
979 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
980 * ipa-inline.c (compute_uninlined_call_time): Be sure that
982 (want_inline_self_recursive_call_p): Likewise.
983 (resolve_noninline_speculation): Only cummulate defined counts.
984 (inline_small_functions): Use nonzero_p.
985 (ipa_inline): Do not access freed node.
987 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
989 * config/aarch64/aarch64.c (aarch64_override_options_internal):
990 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
992 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
994 * config/riscv/riscv.c (riscv_legitimize_move): Handle
995 non-legitimate address.
997 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
999 * config/rs6000/rs6000.md (*lt0_disi): Delete.
1000 (*lt0_<mode>di, *lt0_<mode>si): New.
1002 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1004 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
1005 TARGET_PAIRED_FLOAT.
1007 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
1008 Jim Wilson <jim.wilson@linaro.org>
1010 * config/aarch64/aarch64-cores.def (saphira): New CPU.
1011 * config/aarch64/aarch64-tune.md: Regenerated.
1012 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
1013 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
1015 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
1017 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
1018 (arc_expand_prologue): Restore blink for millicode.
1019 * config/arc/linux.h (LINK_EH_SPEC): Defined.
1021 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
1024 * config/i386/i386.c (ix86_vector_duplicate_value): Use
1025 gen_vec_duplicate after forcing the scalar into a register.
1027 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
1029 * combine (try_combine): Print the insns input to try_combine to the
1032 2017-11-02 Steve Ellcey <sellcey@cavium.com>
1035 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
1036 Remove second argument from aarch64_process_target_attr call.
1037 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
1039 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
1041 (aarch64_handle_attr_arch): Remove second argument.
1042 (aarch64_handle_attr_cpu): Ditto.
1043 (aarch64_handle_attr_tune): Ditto.
1044 (aarch64_handle_attr_isa_flags): Ditto.
1045 (aarch64_process_one_target_attr): Ditto.
1046 (aarch64_process_target_attr): Ditto.
1047 (aarch64_option_valid_attribute_p): Remove second argument.
1048 on aarch64_process_target_attr call.
1050 2017-11-02 David Malcolm <dmalcolm@redhat.com>
1052 * diagnostic.c: Include "selftest-diagnostic.h".
1053 (selftest::assert_location_text): New function.
1054 (selftest::test_diagnostic_get_location_text): New function.
1055 (selftest::diagnostic_c_tests): Call it.
1057 2017-11-02 David Malcolm <dmalcolm@redhat.com>
1059 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
1060 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
1061 (class selftest::test_diagnostic_context): Move to...
1062 * selftest-diagnostic.c: New file.
1063 * selftest-diagnostic.h: New file.
1065 2017-11-02 James Bowman <james.bowman@ftdichip.com>
1067 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
1068 offset range for FT32B.
1069 * config/ft32/ft32.h: option "mcompress" enables relaxation.
1070 * config/ft32/ft32.md: Add TARGET_NOPM.
1071 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
1072 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
1074 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
1076 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
1078 2017-11-02 Jeff Law <law@redhat.com>
1080 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
1081 virtual keyword on FINAL OVERRIDE members.
1083 * tree-ssa-propagate.h (ssa_propagation_engine): Group
1084 virtuals together. Add virtual destructor.
1085 (substitute_and_fold_engine): Similarly.
1087 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
1089 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
1091 2017-11-02 Richard Biener <rguenther@suse.de>
1093 PR tree-optimization/82795
1094 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
1096 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1098 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
1100 (gcc_GAS_CHECK_FEATURE): Remove.
1101 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
1102 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
1103 * configure: Regenerate.
1105 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
1107 * config/arc/arc.c (hwloop_optimize): Account for empty
1110 2017-11-02 Richard Biener <rguenther@suse.de>
1113 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
1114 Truncate ARRAY_REF index and element size.
1116 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
1118 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
1120 2017-11-01 Jeff Law <law@redhat.com>
1122 * tree-ssa-ccp.c (ccp_folder): New class derived from
1123 substitute_and_fold_engine.
1124 (ccp_folder::get_value): New member function.
1125 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
1126 (ccp_fold_stmt): Remove prototype.
1127 (ccp_finalize): Call substitute_and_fold from the ccp_class.
1128 * tree-ssa-copy.c (copy_folder): New class derived from
1129 substitute_and_fold_engine.
1130 (copy_folder::get_value): Renamed from get_value.
1131 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
1132 * tree-vrp.c (vrp_folder): New class derived from
1133 substitute_and_fold_engine.
1134 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
1135 (vrp_folder::get_value): New member function.
1136 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
1137 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
1138 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
1139 provide a class interface to folder/substitute routines.
1140 (ssa_prop_fold_stmt_fn): Remove typedef.
1141 (ssa_prop_get_value_fn): Likewise.
1142 (subsitute_and_fold): Remove prototype.
1143 (replace_uses_in): Likewise.
1144 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
1145 Renamed from replace_uses_in. Call the virtual member function
1146 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
1147 (substitute_and_fold_dom_walker): Remove initialization of
1148 data member entries for calbacks. Add substitute_and_fold_engine
1149 member and initialize it.
1150 (substitute_and_fold_dom_walker::before_dom_children0: Use the
1151 member functions for get_value, replace_phi_args_in c
1152 replace_uses_in, and fold_stmt calls.
1153 (substitute_and_fold_engine::substitute_and_fold): Renamed from
1154 substitute_and_fold. Remove assert. Update ctor call.
1156 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
1157 (ssa_prop_visit_phi_fn): Likewise.
1158 (class ssa_propagation_engine): New class to provide an interface
1160 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
1162 (ssa_prop_visit_phi): Likewise.
1163 (ssa_propagation_engine::simulate_stmt): Moved into class.
1164 Call visit_phi/visit_stmt from the class rather than via
1165 file scoped static variables.
1166 (ssa_propagation_engine::simulate_block): Moved into class.
1167 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
1168 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
1169 set file scoped statics for the visit_stmt/visit_phi callbacks.
1170 * tree-complex.c (complex_propagate): New class derived from
1171 ssa_propagation_engine.
1172 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
1173 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
1174 (tree_lower_complex): Call ssa_propagate via the complex_propagate
1176 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
1177 ssa_propagation_engine.
1178 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
1179 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
1180 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
1181 * tree-ssa-copy.c (copy_prop): New class derived from
1182 ssa_propagation_engine.
1183 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
1184 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
1185 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
1186 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
1187 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
1188 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
1189 (execute_vrp): Call ssa_propagate from the vrp_prop class.
1191 2017-11-01 Jakub Jelinek <jakub@redhat.com>
1193 PR rtl-optimization/82778
1194 PR rtl-optimization/82597
1195 * compare-elim.c (struct comparison): Add in_a_setter field.
1196 (find_comparison_dom_walker::before_dom_children): Remove killed
1197 bitmap and df_simulate_find_defs call, instead walk the defs.
1198 Compute last_setter and initialize in_a_setter. Merge definitions
1199 with first initialization for a few variables.
1200 (try_validate_parallel): Use insn_invalid_p instead of
1201 recog_memoized. Return insn rather than just the pattern.
1202 (try_merge_compare): Fix up comment. Don't uselessly test if
1203 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
1205 (execute_compare_elim_after_reload): Remove df_chain_add_problem
1208 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1209 Alan Hayward <alan.hayward@arm.com>
1210 David Sherwood <david.sherwood@arm.com>
1212 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
1213 aarch64_hard_regno_nregs to get the number of registers
1216 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1217 Alan Hayward <alan.hayward@arm.com>
1218 David Sherwood <david.sherwood@arm.com>
1220 * config/aarch64/constraints.md (Upl): Rename to...
1222 * config/aarch64/aarch64.md
1223 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
1226 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1227 Alan Hayward <alan.hayward@arm.com>
1228 David Sherwood <david.sherwood@arm.com>
1230 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
1231 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
1234 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1235 Alan Hayward <alan.hayward@arm.com>
1236 David Sherwood <david.sherwood@arm.com>
1238 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
1239 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
1240 (aarch64_evpc_dup): Generate rtl direcly, rather than using
1242 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
1243 of a single element.
1244 * config/aarch64/iterators.md: Add a comment above the permute
1245 unspecs to say that they are generated directly by
1246 aarch64_expand_vec_perm_const.
1247 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
1249 2017-11-01 Nathan Sidwell <nathan@acm.org>
1251 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
1253 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
1255 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
1256 and medany code models, and describe what they do.
1258 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1260 Revert accidental duplicate:
1262 * combine.c (can_change_dest_mode): Reject changes in
1263 REGMODE_NATURAL_SIZE.
1265 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
1267 PR rtl-optimization/64682
1268 PR rtl-optimization/69567
1269 PR rtl-optimization/69737
1270 PR rtl-optimization/82683
1271 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
1272 register mentioned in the note, drop the note, unless it came from I3,
1273 in which case it should go to I3 again.
1275 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1277 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
1278 and return false if not.
1279 (clear_bytes_written_by, live_bytes_read): Update accordingly.
1281 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1283 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
1284 range is known to be empty.
1286 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1287 Alan Hayward <alan.hayward@arm.com>
1288 David Sherwood <david.sherwood@arm.com>
1290 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
1291 and CONST_VECTOR_NUNITS instead of computing the number of units from
1292 the byte sizes of the vector and element.
1293 (simplify_binary_operation_1): Likewise.
1294 (simplify_const_binary_operation): Likewise.
1295 (simplify_ternary_operation): Likewise.
1297 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1298 Alan Hayward <alan.hayward@arm.com>
1299 David Sherwood <david.sherwood@arm.com>
1301 * var-tracking.c (INT_MEM_OFFSET): Replace with...
1302 (int_mem_offset): ...this new function.
1303 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
1304 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
1305 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
1308 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1309 Alan Hayward <alan.hayward@arm.com>
1310 David Sherwood <david.sherwood@arm.com>
1312 * lower-subreg.c (interesting_mode_p): New function.
1313 (compute_costs, find_decomposable_subregs, decompose_register)
1314 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
1315 (resolve_clobber, dump_choices): Use it.
1317 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1318 Alan Hayward <alan.hayward@arm.com>
1319 David Sherwood <david.sherwood@arm.com>
1321 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
1323 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1324 Alan Hayward <alan.hayward@arm.com>
1325 David Sherwood <david.sherwood@arm.com>
1327 * alias.c (find_base_value, find_base_term): Only process integer
1328 truncations. Check the precision rather than the size.
1330 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1331 Alan Hayward <alan.hayward@arm.com>
1332 David Sherwood <david.sherwood@arm.com>
1334 * machmode.h (is_narrower_int_mode): New function
1335 * optabs.c (expand_float, expand_fix): Use it.
1336 * dwarf2out.c (rotate_loc_descriptor): Likewise.
1338 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1339 Alan Hayward <alan.hayward@arm.com>
1340 David Sherwood <david.sherwood@arm.com>
1342 * rtl.h (narrower_subreg_mode): New function.
1343 * ira-color.c (update_costs_from_allocno): Use it.
1345 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1346 Alan Hayward <alan.hayward@arm.com>
1347 David Sherwood <david.sherwood@arm.com>
1349 * optabs-query.h (convert_optab_p): New function, split out from...
1350 (convert_optab_handler): ...here.
1351 (widening_optab_handler): Delete.
1352 (find_widening_optab_handler): Remove permit_non_widening parameter.
1353 (find_widening_optab_handler_and_mode): Likewise. Provide an
1354 override that operates on mode class wrappers.
1355 * optabs-query.c (widening_optab_handler): Delete.
1356 (find_widening_optab_handler_and_mode): Remove permit_non_widening
1357 parameter. Assert that the two modes are the same class and that
1358 the "from" mode is narrower than the "to" mode. Use
1359 convert_optab_handler instead of widening_optab_handler.
1360 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
1361 instead of widening_optab_handler.
1362 * expr.c (expand_expr_real_2): Update calls to
1363 find_widening_optab_handler.
1364 * optabs.c (expand_widen_pattern_expr): Likewise.
1365 (expand_binop_directly): Take the insn_code as a parameter.
1366 (expand_binop): Only call find_widening_optab_handler for
1367 conversion optabs; use optab_handler otherwise. Update calls
1368 to find_widening_optab_handler and expand_binop_directly.
1369 Use convert_optab_handler instead of widening_optab_handler.
1370 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
1371 find_widening_optab_handler and use scalar_mode rather than
1373 (convert_plusminus_to_widen): Likewise.
1375 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1376 Alan Hayward <alan.hayward@arm.com>
1377 David Sherwood <david.sherwood@arm.com>
1379 * machmode.h (fixed_size_mode): New class.
1380 * rtl.h (get_pool_mode): Return fixed_size_mode.
1381 * gengtype.c (main): Add fixed_size_mode.
1382 * target.def (get_raw_result_mode): Return a fixed_size_mode.
1383 (get_raw_arg_mode): Likewise.
1384 * doc/tm.texi: Regenerate.
1385 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
1386 * targhooks.c (default_get_reg_raw_mode): Likewise.
1387 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
1388 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
1389 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
1390 (msp430_get_raw_result_mode): Likewise.
1391 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
1392 * dbxout.c (dbxout_parms): Require fixed-size modes.
1393 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
1394 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
1395 * omp-low.c (lower_oacc_reductions): Likewise.
1396 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
1397 (simplify_subreg): Update accordingly.
1398 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
1399 (force_const_mem): Update accordingly. Return NULL_RTX for modes
1400 that aren't fixed-size.
1401 (get_pool_mode): Return a fixed_size_mode.
1402 (output_constant_pool_2): Take a fixed_size_mode.
1404 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1405 Alan Hayward <alan.hayward@arm.com>
1406 David Sherwood <david.sherwood@arm.com>
1408 * doc/rtl.texi (vec_series): Document.
1409 (const): Say that the operand can be a vec_series.
1410 * rtl.def (VEC_SERIES): New rtx code.
1411 * rtl.h (const_vec_series_p_1): Declare.
1412 (const_vec_series_p): New function.
1413 * emit-rtl.h (gen_const_vec_series): Declare.
1414 (gen_vec_series): Likewise.
1415 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
1416 (gen_vec_series): Likewise.
1417 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
1418 * simplify-rtx.c (simplify_unary_operation): Handle negations
1420 (simplify_binary_operation_series): New function.
1421 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
1422 (test_vector_ops_series): New function.
1423 (test_vector_ops): Call it.
1424 * config/powerpcspe/altivec.md (altivec_lvsl): Use
1425 gen_const_vec_series.
1426 (altivec_lvsr): Likewise.
1427 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
1429 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1430 Alan Hayward <alan.hayward@arm.com>
1431 David Sherwood <david.sherwood@arm.com>
1433 * doc/rtl.texi (const): Update description of address constants.
1434 Say that vector constants are allowed too.
1435 * common.md (E, F): Use CONSTANT_P instead of checking for
1437 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
1438 checking for CONST_VECTOR.
1439 * expmed.c (make_tree): Use build_vector_from_val for a CONST
1441 * expr.c (expand_expr_real_2): Check for vector modes instead
1442 of checking for CONST_VECTOR.
1443 * rtl.h (const_vec_p): New function.
1444 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
1445 (unwrap_const_vec_duplicate): Handle them here too.
1447 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1448 David Malcolm <dmalcolm@redhat.com>
1449 Alan Hayward <alan.hayward@arm.com>
1450 David Sherwood <david.sherwood@arm.com>
1452 * rtl.h (vec_duplicate_p): New function.
1453 * selftest-rtl.c (assert_rtx_eq_at): New function.
1454 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
1455 (assert_rtx_eq_at): Declare.
1456 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
1457 * selftest-run-tests.c (selftest::run_tests): Call it.
1458 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
1459 (simplify_unary_operation_1): Recursively handle vector duplicates.
1460 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
1462 (simplify_subreg): Handle subregs of vector duplicates.
1463 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
1464 (selftest::simplify_rtx_c_tests): New functions.
1466 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1467 Alan Hayward <alan.hayward@arm.com>
1468 David Sherwood <david.sherwood@arm.com>
1470 * emit-rtl.h (gen_const_vec_duplicate): Declare.
1471 (gen_vec_duplicate): Likewise.
1472 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
1474 (gen_const_vector): ...here.
1475 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
1476 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
1477 whose elements are all equal.
1478 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
1479 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
1480 (simplify_relational_operation): Likewise.
1481 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
1483 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
1484 (aarch64_expand_vector_init): Likewise.
1485 * config/arm/arm.c (neon_vdup_constant): Likewise.
1486 (neon_expand_vector_init): Likewise.
1487 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
1488 (arm_block_set_unaligned_vect): Likewise.
1489 (arm_block_set_aligned_vect): Likewise.
1490 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
1491 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
1492 (expand_vec_perm_even_odd_pack): Likewise.
1493 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
1494 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
1495 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
1496 gen_const_vec_duplicate.
1497 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
1498 * config/mips/mips.c (mips_gen_const_int_vector): Use
1499 gen_const_vec_duplicate.
1500 (mips_expand_vector_init): Use CONST0_RTX.
1501 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
1502 (define_split): Use gen_const_vec_duplicate.
1503 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
1504 (define_split): Use gen_const_vec_duplicate.
1505 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
1506 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
1507 * config/spu/spu.c (spu_const): Likewise.
1509 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1510 Alan Hayward <alan.hayward@arm.com>
1511 David Sherwood <david.sherwood@arm.com>
1513 * combine.c (can_change_dest_mode): Reject changes in
1514 REGMODE_NATURAL_SIZE.
1516 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
1518 * configure.ac (--enable-libssp): New.
1519 (gcc_cv_libc_provides_ssp): Check for explicit setting before
1520 trying to determine target-specific default. Adjust indentation.
1521 * configure: Regenerated.
1522 * doc/install.texi (Configuration): Expand --disable-libssp
1525 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
1527 config/i386/i386.c (ix86_expand_epilogue): Correct stack
1530 2017-10-31 Martin Jambor <mjambor@suse.cz>
1533 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
1535 2017-10-31 David Malcolm <dmalcolm@redhat.com>
1537 * auto-profile.c (autofdo_source_profile::read): Use
1538 UNKNOWN_LOCATION rather than 0.
1539 * diagnostic-core.h (warning_at_rich_loc): Rename to...
1540 (warning_at): ...this overload.
1541 (warning_at_rich_loc_n): Rename to...
1542 (warning_n): ...this overload.
1543 (error_at_rich_loc): Rename to...
1544 (error_at): ...this overload.
1545 (pedwarn_at_rich_loc): Rename to...
1546 (pedwarn): ...this overload.
1547 (permerror_at_rich_loc): Rename to...
1548 (permerror): ...this overload.
1549 (inform_at_rich_loc): Rename to...
1550 (inform): ...this overload.
1551 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
1552 (diagnostic_n_impl_richloc): Rename to...
1553 (diagnostic_n_impl): ...this rich_location *-based decl.
1554 (inform_at_rich_loc): Rename to...
1555 (inform): ...this, and add an assertion.
1556 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
1557 (warning_at_rich_loc): Rename to...
1558 (warning_at): ...this, and add an assertion.
1559 (warning_at_rich_loc_n): Rename to...
1560 (warning_n): ...this, and add an assertion.
1561 (warning_n): Update location_t-based implementation for removal of
1562 location_t-based diagnostic_n_impl.
1563 (pedwarn_at_rich_loc): Rename to...
1564 (pedwarn): ...this, and add an assertion.
1565 (permerror_at_rich_loc): Rename to...
1566 (permerror): ...this, and add an assertion.
1567 (error_n): Update for removal of location_t-based diagnostic_n_impl.
1568 (error_at_rich_loc): Rename to...
1569 (error_at): ...this, and add an assertion.
1570 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
1571 (driver::do_spec_on_infiles): Likewise.
1572 * substring-locations.c (format_warning_va): Update for renaming
1573 of inform_at_rich_loc.
1575 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
1577 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
1578 _Float<N>X built-in functions so that the variant without the
1579 "__builtin_" prefix is only enabled for the GNU C and Objective C
1580 languages when they are in non-strict ANSI/ISO mode.
1581 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
1582 * target.def (floatn_builtin_p): Add a target hook to control
1583 whether _Float<N> and _Float<N>X built-in functions without the
1584 "__builtin_" prefix are enabled, and return true for C and
1585 Objective C in the default hook. Include langhooks.h in
1587 * targhooks.h (default_floatn_builtin_p): Likewise.
1588 * targhooks.c (default_floatn_builtin_p): Likewise.
1589 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
1590 floatn_builtin_p target hook.
1591 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
1593 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
1594 Eric Botcazou <ebotcazou@adacore.com>
1596 PR rtl-optimization/81803
1597 * lra-constraints.c (curr_insn_transform): Also reload the whole
1598 register for a strict subreg no wider than a word if this is for
1599 a WORD_REGISTER_OPERATIONS target.
1601 2017-10-31 Jason Merrill <jason@redhat.com>
1603 * gdbinit.in: Skip over inlines from timevar.h.
1605 2017-10-31 Martin Liska <mliska@suse.cz>
1607 * doc/gcov.texi: Document new option.
1608 * gcov.c (print_usage): Likewise print it.
1609 (process_args): Support the argument.
1610 (format_count): New function.
1611 (format_gcov): Use the function.
1613 2017-10-31 Martin Liska <mliska@suse.cz>
1615 * gcov.c (struct name_map): do not use typedef.
1616 Define operator== and operator<.
1617 (name_search): Remove.
1618 (name_sort): Remove.
1619 (main): Do not allocate names.
1620 (process_file): Add vertical space.
1621 (generate_results): Use std::find.
1622 (release_structures): Do not release memory.
1623 (find_source): Use std::find.
1625 2017-10-31 Martin Liska <mliska@suse.cz>
1627 * gcov.c (struct line_info): Remove it's typedef.
1628 (line_info::line_info): Add proper ctor.
1629 (line_info::has_block): Do not use a typedef.
1630 (struct source_info): Do not use typedef.
1631 (circuit): Likewise.
1632 (get_cycles_count): Likewise.
1633 (output_intermediate_file): Iterate via vector iterator.
1634 (add_line_counts): Use std::vector methods.
1635 (accumulate_line_counts): Likewise.
1636 (output_lines): Likewise.
1638 2017-10-31 Martin Liska <mliska@suse.cz>
1640 * gcov.c (struct source_info): Remove typedef.
1641 (source_info::source_info): Add proper ctor.
1642 (accumulate_line_counts): Use struct, not it's typedef.
1643 (output_gcov_file): Likewise.
1644 (output_lines): Likewise.
1645 (main): Do not allocate an array.
1646 (output_intermediate_file): Use size of vector container.
1647 (process_file): Resize the vector.
1648 (generate_results): Do not preallocate, use newly added vector
1650 (release_structures): Do not release sources.
1651 (find_source): Use vector methods.
1652 (add_line_counts): Do not use typedef.
1654 2017-10-31 Martin Liska <mliska@suse.cz>
1656 * doc/gcov.texi: Document that.
1657 * gcov.c (add_line_counts): Mark lines with a non-executed
1659 (output_line_beginning): Handle such lines.
1660 (output_lines): Pass new argument.
1661 (output_intermediate_file): Print it in intermediate format.
1663 2017-10-31 Martin Liska <mliska@suse.cz>
1665 * color-macros.h: New file.
1666 * diagnostic-color.c: Factor out color related to macros to
1668 * doc/gcov.texi: Document -k option.
1669 * gcov.c (INCLUDE_STRING): Include string.h.
1670 (print_usage): Add -k option.
1671 (process_args): Parse it.
1672 (pad_count_string): New function.
1673 (output_line_beginning): Likewise.
1674 (DEFAULT_LINE_START): New macro.
1675 (output_lines): Support color output.
1677 2017-10-31 Martin Liska <mliska@suse.cz>
1679 PR gcov-profile/82633
1680 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
1681 their interaction with GCOV infrastructure.
1682 * configure.ac: Add -fkeep-{inline,static}-functions to
1684 * configure: Regenerate.
1686 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
1689 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
1691 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
1694 * config/rs6000/rs6000.md (allocate_stack): Force update interval
1695 into a register if it does not fit into an immediate offset field.
1697 2017-10-31 Olivier Hainque <hainque@adacore.com>
1699 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
1701 2017-10-31 Julia Koval <julia.koval@intel.com>
1703 * config.gcc: Add gfniintrin.h.
1704 * config/i386/gfniintrin.h: New.
1705 * config/i386/i386-builtin-types.def
1706 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
1707 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
1708 __builtin_ia32_vgf2p8affineinvqb_v32qi,
1709 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
1710 __builtin_ia32_vgf2p8affineinvqb_v16qi,
1711 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
1712 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
1713 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
1714 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
1715 V64QI_FTYPE_V64QI_V64QI_INT): New types.
1716 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1717 * config/i386/immintrin.h: Include gfniintrin.h.
1718 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
1720 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
1722 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
1724 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
1726 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
1727 (arm_ashldi3_1bit): Remove pattern.
1728 (ashrdi3): Remove shift by 1 expansion.
1729 (arm_ashrdi3_1bit): Remove pattern.
1730 (lshrdi3): Remove shift by 1 expansion.
1731 (arm_lshrdi3_1bit): Remove pattern.
1732 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
1733 cost of ashldi3 by 1.
1734 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
1735 (<shift>di3_neon): Likewise.
1737 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
1739 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
1740 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
1741 and (*aarch64_simd_mov<VQ:mode>).
1742 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
1743 pattern alternative.
1744 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
1745 attributes to match pattern alternative.
1747 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
1749 * config.gcc (powerpc*-*-*): Add emmintrin.h.
1750 * config/rs6000/emmintrin.h: New file.
1751 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
1753 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
1755 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
1756 * (movdi_vfp_cortexa8): Remove pattern.
1758 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1760 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
1762 (Specific, alpha*-dec-osf5.1): Remove.
1763 (Specific, mips-sgi-irix5): Remove.
1764 (Specific, mips-sgi-irix6): Remove.
1766 2017-10-30 Jakub Jelinek <jakub@redhat.com>
1769 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
1770 arguments to clear_bit_region_be.
1772 2017-10-30 Jim Wilson <wilson@tuliptree.org>
1774 * gimplify.c: Include memmodel.h.
1776 2017-10-30 Martin Jambor <mjambor@suse.cz>
1778 * omp-grid.c (grid_attempt_target_gridification): Also insert a
1779 condition whether loop should be executed at all.
1781 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
1783 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
1784 gimple folding of vec_madd() intrinsics.
1785 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
1786 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
1787 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
1789 2017-10-30 Richard Biener <rguenther@suse.de>
1791 PR tree-optimization/82762
1793 2017-10-23 Richard Biener <rguenther@suse.de>
1795 PR tree-optimization/82129
1797 2017-08-01 Richard Biener <rguenther@suse.de>
1799 PR tree-optimization/81181
1800 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
1801 (compute_antic): ... end of iteration here.
1803 2017-10-30 Joseph Myers <joseph@codesourcery.com>
1805 * doc/invoke.texi (C Dialect Options): Document -std=c17,
1806 -std=iso9899:2017 and -std=gnu17.
1807 * doc/standards.texi (C Language): Document C17 support.
1808 * doc/cpp.texi (Overview): Mention -std=c17.
1809 (Standard Predefined Macros): Document C11 and C17 values of
1810 __STDC_VERSION__. Do not refer to C99 support as incomplete.
1811 * doc/extend.texi (Inline): Do not list individual options for
1812 standards newer than C99.
1813 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
1815 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
1818 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
1820 * asan.c (asan_finish_file): Align asan globals array by shadow
1823 2017-10-30 Jakub Jelinek <jakub@redhat.com>
1826 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
1827 (struct store_immediate_info): Add bitregion_start and bitregion_end
1829 (store_immediate_info::store_immediate_info): Add brs and bre
1830 arguments and initialize bitregion_{start,end} from those.
1831 (struct merged_store_group): Add bitregion_start, bitregion_end,
1832 align_base and mask fields. Drop unnecessary struct keyword from
1833 struct store_immediate_info. Add do_merge method.
1834 (clear_bit_region_be): Use memset instead of loop storing zeros.
1835 (merged_store_group::do_merge): New method.
1836 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
1837 stores as long as the surrounding bitregions have no gaps.
1838 (merged_store_group::merge_overlapping): Use do_merge.
1839 (merged_store_group::apply_stores): Test that bitregion_{start,end}
1840 is byte aligned, rather than requiring that start and width are
1841 byte aligned. Drop unnecessary struct keyword from
1842 struct store_immediate_info. Allocate and populate also mask array.
1843 Make start of the arrays relative to bitregion_start rather than
1844 start and size them according to bitregion_{end,start} difference.
1845 (struct imm_store_chain_info): Drop unnecessary struct keyword from
1846 struct store_immediate_info.
1847 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
1848 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
1849 struct keyword from struct store_immediate_info.
1850 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
1851 between stores as long as the surrounding bitregions have no gaps.
1853 (struct split_store): Add orig non-static data member.
1854 (split_store::split_store): Initialize orig to false.
1855 (find_constituent_stmts): Return store_immediate_info *, non-NULL
1856 if there is exactly a single original stmt. Change stmts argument
1857 to pointer from reference, if NULL, don't push anything to it. Add
1858 first argument, use it to optimize skipping over orig stmts that
1859 are known to be before bitpos already. Simplify.
1860 (split_group): Return unsigned int count how many stores are or
1861 would be needed rather than a bool. Add allow_unaligned argument.
1862 Change split_stores argument from reference to pointer, if NULL,
1863 only do a dry run computing how many stores would be produced.
1864 Rewritten algorithm to use both alignment and misalign if
1865 !allow_unaligned and handle bitfield stores with gaps.
1866 (imm_store_chain_info::output_merged_store): Set start_byte_pos
1867 from bitregion_start instead of start. Compute allow_unaligned
1868 here, if true, do 2 split_group dry runs to compute which one
1869 produces fewer stores and prefer aligned if equal. Punt if
1870 new count is bigger or equal than original before emitting any
1871 statements, rather than during that. Remove no longer needed
1872 new_ssa_names tracking. Replace num_stmts with
1873 split_stores.length (). Use 32-bit stack allocated entries
1874 in split_stores auto_vec. Try to reuse original store lhs/rhs1
1875 if possible. Handle bitfields with gaps.
1876 (pass_store_merging::execute): Ignore bitsize == 0 stores.
1877 Compute bitregion_{start,end} for the stores and construct
1878 store_immediate_info with that. Formatting fixes.
1880 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
1883 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
1884 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
1886 2017-10-29 Jim Wilson <wilson@tuliptree.org>
1888 * gimplify.c: Include tm_p.h.
1890 * common.opt (gcoff): Re-add as ignored option.
1891 (gcoff1, gcoff2, gcoff3): Likewise.
1893 * Makefile.in (OBJS): Delete sdbout.o.
1894 (GTFILES): Delete $(srcdir)/sdbout.c.
1895 * debug.h: Delete sdb_debug_hooks.
1896 * final.c: Delete sdbout.h include.
1897 (final_scan_insn): Delete SDB_DEBUG check.
1898 (rest_of_clean_state): Likewise.
1899 * output.h: Delete sdb_begin_function_line.
1902 * toplev.c: Delete sdbout.h include.
1903 (process_options): Delete SDB_DEBUG check.
1904 * tree-core.h (tree_type_common): Delete pointer field of
1906 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
1907 TYPE_SYMTAB_POINTER.
1908 * tree.h (TYPE_SYMTAB_POINTER): Delete.
1909 (TYPE_SYMTAB_IS_POINTER): Delete.
1910 (TYPE_SYMTAB_IS_DIE): Renumber.
1911 * xcoffout.c: Refer to former sdbout.c file.
1912 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
1914 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
1915 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
1916 (Debugging Options): Delete -gcoff.
1917 (-gstabs): Delete SDB reference.
1919 (-gcoff@var{level}): Delete.
1920 * doc/passes.texi (Debugging information output): Delete SDB and
1921 sdbout.c references.
1922 * doc/tm.texi: Regenerate.
1923 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
1924 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
1926 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
1927 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
1928 (SDB_DEBUGGING_INFO): Delete.
1929 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
1930 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
1931 * target.def (output_source_filename): Delete COFF reference.
1933 * common.opt (gcoff): Delete.
1934 (gxcoff+): Update Negative chain.
1935 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
1937 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
1938 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
1939 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
1941 (expand_function_start): Change sdb reference to past tense.
1942 (expand_function_end): Change sdb reference to past tense.
1943 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
1944 * opts.c (debug_type_names): Delete coff entry.
1945 (common_handle_option): Delete OPT_gcoff case.
1946 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
1948 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
1949 * config/cris/cris.h: Delete SDB reference in comment.
1950 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
1951 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
1952 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
1953 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
1955 (ix86_expand_prologue): Likewise.
1956 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
1957 * config/ia64/ia64.h: Likewise.
1958 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
1959 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
1961 * config/mmix/mmix.h: Likewise.
1962 * config/nds32/nds32.c: Likewise.
1963 * config/stormy/storym16.h: Likewise.
1964 * config/visium/visium.h: Likewise.
1965 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
1967 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
1969 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
1970 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
1971 FRAME_POINTER_REGNUM point at high end of local var area.
1973 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
1975 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
1976 Move comment around. Do not reset best_edge for a copiable
1977 destination if the copy would cause a partition change.
1978 (better_edge_p): Remove redundant check.
1980 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
1982 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
1984 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
1986 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
1987 for math functions that have _Float<N> and _Float<N>X variants.
1988 (mathfn_built_in_2): Add support for math functions that have
1989 _Float<N> and _Float<N>X variants.
1990 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
1991 (expand_builtin_mathfn_ternary): Add support for fma with
1992 _Float<N> and _Float<N>X variants.
1993 (expand_builtin): Likewise.
1994 (fold_builtin_3): Likewise.
1995 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
1996 create math function _Float<N> and _Float<N>X variants as external
1998 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
1999 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
2000 the __builtin_ prefix and if not strict ansi, without the prefix.
2001 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
2002 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
2003 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
2004 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
2005 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
2006 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
2007 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
2008 function signatures for fma _Float<N> and _Float<N>X variants.
2009 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
2010 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
2011 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
2012 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
2013 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
2014 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
2015 * gencfn-macros.c (print_case_cfn): Add support for math functions
2016 that have _Float<N> and _Float<N>X variants.
2017 (print_define_operator_list): Likewise.
2018 (fltfn_suffixes): Likewise.
2020 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
2021 for math functions that have _Float<N> and _Float<N>X variants.
2022 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
2023 and _Float<N>X variants.
2024 (COPYSIGN): Likewise.
2027 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
2028 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
2030 (integer_valued_read_call_p): Likewise.
2031 * fold-const-call.c (fold_const_call_ss): Likewise.
2032 (fold_const_call_sss): Add support for copysign, fmin, and fmax
2033 _Float<N> and _Float<N>X variants.
2034 (fold_const_call_ssss): Add support for fma _Float<N> and
2035 _Float<N>X variants.
2036 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
2037 support for copysign and fma _Float<N> and _Float<N>X variants.
2038 (backprop::process_builtin_call_use): Likewise.
2039 * tree-call-cdce.c (can_test_argument_range); Add support for
2040 sqrt _Float<N> and _Float<N>X variants.
2041 (edom_only_function): Likewise.
2042 (get_no_error_domain): Likewise.
2043 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
2044 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
2045 copysign _Float<N> and _Float<N>X variants.
2046 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
2047 handled by machine independent code.
2048 (FMAF128): Likewise.
2049 * doc/cpp.texi (Common Predefined Macros): Document defining
2050 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
2051 fma _Float<N> and _Float<N>X variants.
2053 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
2056 * config/i386/i386-modes.def (CCFPU): Remove definition.
2057 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
2058 (ix86_cc_modes_compatible): Ditto.
2059 (ix86_expand_carry_flag_compare): Ditto.
2060 (ix86_expand_int_movcc): Ditto.
2061 (ix86_expand_int_addcc): Ditto.
2062 (ix86_reverse_condition): Ditto.
2063 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
2064 Return true/false for unordered/ordered fp comparisons.
2065 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
2066 (ix86_prepare_fp_compare_args): Update for rename.
2067 (ix86_expand_fp_compare): Update for rename. Generate unordered
2068 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
2069 (ix86_expand_sse_compare_and_jump): Ditto.
2070 * config/i386/predicates.md (fcmov_comparison_operator):
2071 Remove CCFPU mode handling.
2072 (ix86_comparison_operator): Ditto.
2073 (ix86_carry_flag_operator): Ditto.
2074 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
2075 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
2076 (*cmpu<mode>_cc_i387): Ditto.
2077 (FPCMP): Remove mode iterator.
2078 (unord): Remove mode attribute.
2079 (unord_subst): New define_subst transformation
2080 (unord): New define_subst attribute.
2082 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
2083 (*cmpi<unord>xf_i387): Ditto.
2084 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
2085 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
2086 using unord_subst transformation.
2087 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
2088 (round_saeonly): Also handle CCFP mode.
2089 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
2090 Remove UNSPEC_SAHF unspec handling.
2092 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
2094 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
2096 2017-10-27 Jeff Law <law@redhat.com>
2098 * gimple-ssa-sprintf.c: Include domwalk.h.
2099 (class sprintf_dom_walker): New class, derived from dom_walker.
2100 (sprintf_dom_walker::before_dom_children): New function.
2101 (struct call_info): Moved into sprintf_dom_walker class
2102 (compute_formath_length, handle_gimple_call): Likewise.
2103 (sprintf_length::execute): Call the dominator walker rather
2104 than walking the statements.
2106 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
2107 gimple statement locations.
2108 (check_array_bounds): Corresponding changes. Get the statement's
2109 location directly from wi->stmt.
2111 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
2114 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
2116 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
2118 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
2119 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
2121 2017-10-27 Jakub Jelinek <jakub@redhat.com>
2124 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
2125 * config/i386/i386.c (maybe_get_pool_constant): Removed.
2126 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
2127 maybe_get_pool_constant.
2128 * config/i386/predicates.md (zero_extended_scalar_load_operand):
2131 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2133 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
2134 2.26 caveat. Update gas and gld versions.
2135 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
2138 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2140 * cgraph.h (set_malloc_flag): Declare.
2141 * cgraph.c (set_malloc_flag_1): New function.
2142 (set_malloc_flag): Likewise.
2143 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
2144 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
2146 (read_ipa_call_summary): Add support for reading is_return_callee.
2147 (write_ipa_call_summary): Stream is_return_callee.
2148 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
2149 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
2150 ipa-prop.h, ipa-fnsummary.h.
2151 (pure_const_names): Change to static.
2152 (malloc_state_e): Define.
2153 (malloc_state_names): Define.
2154 (funct_state_d): Add field malloc_state.
2155 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
2156 (check_retval_uses): New function.
2157 (malloc_candidate_p): Likewise.
2158 (analyze_function): Add support for malloc attribute.
2159 (pure_const_write_summary): Stream malloc_state.
2160 (pure_const_read_summary): Add support for reading malloc_state.
2161 (dump_malloc_lattice): New function.
2162 (propagate_malloc): New function.
2163 (warn_function_malloc): New function.
2164 (ipa_pure_const::execute): Call propagate_malloc and
2165 ipa_free_fn_summary.
2166 (pass_local_pure_const::execute): Add support for malloc attribute.
2167 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
2168 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
2170 2017-10-27 Martin Liska <mliska@suse.cz>
2172 PR gcov-profile/82457
2173 * doc/invoke.texi: Document that one needs a non-strict ISO mode
2174 for fork-like functions to be properly instrumented.
2176 2017-10-27 Richard Biener <rguenther@suse.de>
2179 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
2180 info when we redirected EH.
2182 2017-10-26 Michael Collison <michael.collison@arm.com>
2184 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
2186 (<optab>_trunchf<GPI:mode>2: New pattern.
2187 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
2188 * config/aarch64/iterators.md (wv): New mode attribute.
2189 (vf, VF): New mode attributes.
2190 (vgp, VGP): New mode attributes.
2191 (s): Update attribute with SImode and DImode prefixes.
2193 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
2195 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
2196 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
2197 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
2198 (nios2_option_overide): Initialize it. Don't allow R0-relative
2199 addressing with PIC.
2200 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
2201 (nios2_symbolic_constant_p): Likewise.
2202 (nios2_legitimate_address_p): Likewise.
2203 (nios2_r0rel_section_name_p): New.
2204 (nios2_symbol_ref_in_r0rel_data_p): New.
2205 (nios2_emit_move_sequence): Handle r0rel_constant_p.
2206 (r0rel_constant_p): New.
2207 (nios2_print_operand_address): Handle r0rel_constant_p.
2208 (nios2_cdx_narrow_form_p): Likewise.
2209 * config/nios2/nios2.opt (mr0rel-sec=): New option.
2210 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
2211 (Nios II Options): Document -mr0rel-sec.
2213 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
2215 * config/nios2/nios2.c: Include xregex.h.
2216 (nios2_gprel_sec_regex): New.
2217 (nios2_option_overide): Initialize it. Don't allow GP-relative
2218 addressing with PIC.
2219 (nios2_small_section_name_p): Check for regex match.
2220 * config/nios2/nios2.opt (mgprel-sec=): New option.
2221 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
2222 (Nios II Options): Document -mgprel-sec.
2224 2017-10-26 Jim Wilson <wilson@tuliptree.org>
2226 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
2228 2017-10-26 Tom de Vries <tom@codesourcery.com>
2230 PR tree-optimization/82707
2231 * gimple.c (gimple_copy): Fix unsharing of
2232 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
2234 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
2236 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
2237 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
2238 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
2239 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
2240 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
2241 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
2242 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
2243 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
2244 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
2245 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
2246 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
2247 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
2248 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
2249 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
2250 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
2251 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
2252 _mm512_mask_cmpunord_ps_mask): New intrinsics.
2254 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
2256 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
2258 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
2259 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
2260 warning to rs6000.c. Remove the Undocumented flag, since it has
2262 (-mabi=ibmlongdouble): Likewise.
2263 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
2264 already set, set the default format for long double.
2265 (rs6000_debug_reg_global): Print whether long double is IBM or
2267 (rs6000_option_override_internal): Rework setting long double
2268 format. Only warn if the user is changing the long double default
2269 and they did not use -Wno-psabi.
2270 * doc/invoke.texi (PowerPC options): Update the documentation for
2271 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
2273 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2274 Alan Hayward <alan.hayward@arm.com>
2275 David Sherwood <david.sherwood@arm.com>
2277 * rtl.h (wider_subreg_mode): New function.
2278 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
2279 rather than an unsigned int *.
2280 * ira-color.c (regno_max_ref_width): Replace with...
2281 (regno_max_ref_mode): ...this new variable.
2282 (coalesced_pseudo_reg_slot_compare): Update accordingly.
2283 Use wider_subreg_mode.
2284 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
2285 rather than an unsigned int *.
2286 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
2287 (process_alt_operands): Likewise.
2288 (invariant_p): Likewise.
2289 * lra-spills.c (assign_mem_slot): Likewise.
2290 (add_pseudo_to_slot): Likewise.
2291 * lra.c (collect_non_operand_hard_regs): Likewise.
2292 (add_regs_to_insn_regno_info): Likewise.
2293 * reload1.c (regno_max_ref_width): Replace with...
2294 (regno_max_ref_mode): ...this new variable.
2295 (reload): Update accordingly. Update call to
2296 ira_sort_regnos_for_alter_reg.
2297 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
2298 (init_eliminable_invariants): Update to use regno_max_ref_mode.
2299 (scan_paradoxical_subregs): Likewise.
2301 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2303 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
2304 (aarch64_frame): Add emit_frame_chain boolean.
2305 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2306 Move eh_return case to aarch64_layout_frame.
2307 (aarch64_layout_frame): Initialize emit_frame_chain.
2308 (aarch64_expand_prologue): Use emit_frame_chain.
2310 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2312 * config/aarch64/aarch64.c (aarch64_layout_frame):
2313 Ensure LR is always stored at the bottom of the callee-saves.
2314 Remove rarely used frame layout which saves callee-saves at top of
2315 frame, so the store of LR can be used as a valid probe in all cases.
2317 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2319 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
2320 Improve unaligned TImode/TFmode base/offset split.
2322 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2323 Alan Hayward <alan.hayward@arm.com>
2324 David Sherwood <david.sherwood@arm.com>
2326 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
2327 * combine.c (find_single_use_1): Likewise.
2328 (expand_field_assignment): Likewise.
2329 (move_deaths): Likewise.
2330 * lra-constraints.c (simplify_operand_subreg): Likewise.
2331 (curr_insn_transform): Likewise.
2332 * lra.c (collect_non_operand_hard_regs): Likewise.
2333 (add_regs_to_insn_regno_info): Likewise.
2334 * rtlanal.c (reg_referenced_p): Likewise.
2335 (covers_regno_no_parallel_p): Likewise.
2337 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2339 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
2340 Don't print any bits outside the precision of the value.
2341 * wide-int.cc (test_printing): Add some new tests.
2343 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2345 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
2346 supports -xbrace_comment option.
2347 * configure: Regenerate.
2348 * config.in: Regenerate.
2349 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
2350 (ASM_CPU_SPEC): Use it.
2352 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2354 * target.def (static_rtx_alignment): New hook.
2355 * targhooks.h (default_static_rtx_alignment): Declare.
2356 * targhooks.c (default_static_rtx_alignment): New function.
2357 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
2358 * doc/tm.texi: Regenerate.
2359 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
2360 instead of targetm.constant_alignment. Remove call to
2362 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2363 (cris_preferred_mininum_alignment): New function, split out from...
2364 (cris_constant_alignment): ...here.
2365 (cris_static_rtx_alignment): New function.
2366 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
2368 (ix86_constant_alignment): ...here.
2369 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2370 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2371 (mmix_static_rtx_alignment): New function.
2372 * config/spu/spu.c (spu_static_rtx_alignment): New function.
2373 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2375 2017-10-26 Tamar Christina <tamar.christina@arm.com>
2378 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
2379 Add flag_trapping_math and flag_fp_int_builtin_inexact.
2381 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
2383 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
2384 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
2385 mark as a sign-extending load.
2386 (local_pic_load_u): Define.
2388 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
2391 * fold-const.c (operand_equal_for_comparison_p): Also return true
2392 if ARG0 is a simple variant of ARG1 with narrower precision.
2393 (fold_ternary_loc): Always pass unstripped operands to the predicate.
2395 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
2397 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
2399 * i386.h (processor_costs): Add gather_static, gather_per_elt,
2400 scatter_static, scatter_per_elt.
2401 * x86-tune-costs.h: Add new cost entries.
2403 2017-10-25 Richard Biener <rguenther@suse.de>
2405 * tree-ssa-sccvn.h (vn_eliminate): Declare.
2406 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
2407 class pass_fre): Move to ...
2408 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
2409 class pass_fre): ... here and adjust for statistics.
2411 2017-10-25 Jakub Jelinek <jakub@redhat.com>
2414 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
2415 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
2416 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
2417 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
2420 2017-10-25 Richard Biener <rguenther@suse.de>
2422 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
2423 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
2424 eliminate_push_avail, eliminate_insert): Move inside...
2425 (class eliminate_dom_walker): ... this class in preparation
2427 (fini_eliminate): Remove by merging with ...
2428 (eliminate): ... this function. Adjust for class changes.
2429 (pass_pre::execute): Remove fini_eliminate call.
2430 (pass_fre::execute): Likewise.
2432 2017-10-24 Jakub Jelinek <jakub@redhat.com>
2435 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
2436 (VPERMI2, VPERMI2I): New mode iterators.
2437 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
2438 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
2440 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
2441 mode iterator. Remove 3 old define_insn patterns.
2442 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
2443 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
2444 VPERMI2 mode iterator, remove the other two expanders.
2445 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
2446 to use VPERMI2 mode iterator, add another alternative for vpermi2*
2447 instructions, remove the other two patterns.
2448 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
2449 mode iterator, remove the other two patterns.
2450 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
2451 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
2452 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
2453 and adjust argument order accordingly.
2454 (ix86_expand_vec_perm): Adjust caller.
2455 (expand_vec_perm_1): Likewise.
2456 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
2457 (expand_vec_perm_vpermt2_vpshub2): ... this.
2458 (ix86_expand_vec_perm_const_1): Adjust caller.
2459 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
2462 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
2463 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
2464 (vec_shl_<mode>): Remove unused expander.
2465 (avx512bw_<shift_insn><mode>3): New define_insn.
2466 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
2467 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
2469 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
2472 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
2475 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
2477 PR rtl-optimization/82396
2478 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
2479 (autopref_multipass_init): Simplify initialization.
2480 (autopref_rank_data): Simplify sort order.
2481 * gcc/sched-int.h (autopref_multipass_data_): Remove
2482 multi_mem_insn_p, min_offset and max_offset.
2484 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
2487 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2488 Check special value of flag_omit_frame_pointer.
2489 (aarch64_can_eliminate): Likewise.
2490 (aarch64_override_options_after_change_1): Simplify handling of
2491 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
2493 2017-10-24 Richard Biener <rguenther@suse.de>
2495 PR tree-optimization/82697
2496 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
2497 zero for conditional load and unconditional store.
2499 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
2501 * doc/install.texi: Document bootstrap-cet.
2503 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
2506 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
2507 ENDBR instruction at function entrance if function is only
2510 2017-10-24 Jakub Jelinek <jakub@redhat.com>
2513 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
2514 patterns to better describe from which operation the CF is computed.
2515 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
2516 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
2517 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
2518 is 0, use _0 suffixed expanders instead of emitting a comparison
2521 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
2523 * config/i386/i386.md(*movsf_internal, *movdf_internal):
2524 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
2526 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
2529 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
2530 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
2531 * loop-iv.c (iv_get_reaching_def): Likewise.
2532 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
2533 variable is promoted and the partition contains undefined values.
2535 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2537 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
2539 (nios2_address_cost): Define.
2540 (nios2_legitimize_address): Recognize (exp + constant) directly.
2541 (TARGET_ADDRESS_COST): Define.
2543 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2545 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
2546 (nios2_symbolic_memory_operand_p): Declare.
2547 (nios2_split_large_constant): Declare.
2548 (nios2_split_symbolic_memory_operand): Declare.
2549 * config/nios2/nios2.c: Adjust includes.
2550 (nios2_symbolic_constant_allowed): New.
2551 (nios2_symbolic_constant_p): New.
2552 (nios2_plus_symbolic_constant_p): New.
2553 (nios2_valid_addr_expr_p): Recognize addresses involving
2555 (nios2_legitimate_address_p): Likewise, also LO_SUM.
2556 (nios2_symbolic_memory_operand_p): New.
2557 (nios2_large_constant_p): New.
2558 (nios2_split_large_constant): New.
2559 (nios2_split_plus_large_constant): New.
2560 (nios2_split_symbolic_memory_operand): New.
2561 (nios2_legitimize_address): Code refactoring. Handle addresses
2562 involving symbolic constants.
2563 (nios2_emit_move_sequence): Likewise.
2564 (nios2_print_operand): Improve error output.
2565 (nios2_print_operand_address): Handle LO_SUM.
2566 (nios2_cdx_narrow_form_p): Likewise.
2567 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
2568 operands involving symbolic constants.
2569 (movhi_internal, movsi_internal): Likewise.
2570 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
2571 (extendhisi2, extendqi<mode>2): Likewise.
2573 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2575 * tree-pass.h (PROP_rtl_split_insns): Define.
2576 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
2578 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2580 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
2582 2017-10-23 Jakub Jelinek <jakub@redhat.com>
2585 * target.def (const_not_ok_for_debug_p): Default to
2586 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
2587 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
2588 * targhooks.c (default_const_not_ok_for_debug_p): New function.
2589 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
2590 which targetm.const_not_ok_for_debug_p returned true.
2591 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
2593 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
2595 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
2596 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
2597 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
2599 (ix86_const_not_ok_for_debug_p): New function.
2600 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
2601 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
2603 2017-10-23 David Malcolm <dmalcolm@redhat.com>
2606 * system.h: Conditionally include "unique-ptr.h" if
2607 INCLUDE_UNIQUE_PTR is defined.
2608 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
2609 of defining INCLUDE_UNIQUE_PTR before including "system.h".
2611 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
2613 * config/rl78/rl78.md: New define_expand "subdi3".
2615 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
2618 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
2619 DF_REF_INSN if DF_REF_INSN_INFO is false.
2621 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
2623 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
2624 xmm_move instead of sse_move.
2625 (sse_store_index): New function.
2626 (ix86_register_move_cost): Be more sensible about mismatch stall;
2627 model AVX moves correctly; make difference between sse->integer and
2629 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
2630 moves; make difference between SSE and AVX.
2631 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
2632 and zmm_move. Increase size of sse load and store tables;
2633 add unaligned load and store tables; add ssemmx_to_integer.
2634 * x86-tune-costs.h: Update all entries according to real
2635 move latencies from Agner Fog's manual and chip documentation.
2637 2017-10-23 Jakub Jelinek <jakub@redhat.com>
2640 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
2641 * config/i386/constraints.md (Wf): New constraint.
2642 * config/i386/i386.md (UNSPEC_SBB): New unspec.
2643 (cmp<dwi>_doubleword): Removed.
2644 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
2645 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
2646 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
2647 expand with cmp<dwi>_doubleword. For LTU and GEU use
2648 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
2650 * common.opt (gcolumn-info): Enable by default.
2651 * doc/invoke.texi (gcolumn-info): Document new default.
2653 2017-10-23 Richard Biener <rguenther@suse.de>
2655 PR tree-optimization/82672
2656 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
2657 Fold the stmt if we propagated into it.
2659 2017-10-23 Richard Biener <rguenther@suse.de>
2661 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
2662 (bitmap_remove_expr_from_set): ... this. All callers call this
2663 for non-constant values.
2664 (bitmap_set_subtract): Rename to...
2665 (bitmap_set_subtract_expressions): ... this. Adjust and
2667 (bitmap_set_contains_value): Remove superfluous check.
2668 (bitmap_set_replace_value): Inline into single caller ...
2669 (bitmap_value_replace_in_set): ... here and simplify.
2670 (dependent_clean): Merge into ...
2671 (clean): ... this using an overload. Adjust.
2672 (prune_clobbered_mems): Adjust.
2673 (compute_antic_aux): Likewise.
2674 (compute_partial_antic_aux): Likewise.
2676 2017-10-23 Richard Biener <rguenther@suse.de>
2678 PR tree-optimization/82129
2680 2017-08-01 Richard Biener <rguenther@suse.de>
2682 PR tree-optimization/81181
2683 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
2684 (compute_antic): ... end of iteration here.
2686 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2688 * target.def (starting_frame_offset): New hook.
2689 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
2690 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
2691 * doc/tm.texi.in: Regenerate.
2692 * hooks.h (hook_hwi_void_0): Declare.
2693 * hooks.c (hook_hwi_void_0): New function.
2694 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
2695 STARTING_FRAME_OFFSET.
2696 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
2697 * reload1.c (reload): Likewise.
2698 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
2699 instead of STARTING_FRAME_OFFSET.
2700 * function.c (try_fit_stack_local): Likewise.
2701 (assign_stack_local_1): Likewise
2702 (instantiate_virtual_regs): Likewise.
2703 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
2704 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
2705 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
2706 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
2707 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
2708 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
2709 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
2710 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
2711 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
2712 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
2713 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
2714 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
2715 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
2716 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
2717 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
2718 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
2719 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
2720 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
2721 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
2722 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
2723 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
2724 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
2725 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
2726 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
2727 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
2728 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
2729 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
2730 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
2731 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
2732 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
2733 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
2734 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
2735 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
2736 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
2737 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
2738 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
2739 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
2740 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
2741 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
2742 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
2743 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
2744 * config/avr/avr.c (avr_starting_frame_offset): Make static and
2745 return a HOST_WIDE_INT.
2746 (avr_builtin_setjmp_frame_value): Use it instead of
2747 STARTING_FRAME_OFFSET.
2748 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2749 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
2750 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
2752 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2753 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
2754 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
2755 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2756 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
2757 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
2758 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2759 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
2760 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
2761 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2762 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
2763 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
2765 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2766 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
2767 * config/mips/mips.c (mips_compute_frame_info): Refer to
2768 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
2769 (mips_starting_frame_offset): New function.
2770 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2771 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
2772 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
2773 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
2774 and return a HOST_WIDE_INT.
2775 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2776 (mmix_initial_elimination_offset): Refer to
2777 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
2778 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
2779 * config/pa/pa.c (pa_starting_frame_offset): New function.
2780 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
2781 (pa_expand_prologue): Likewise.
2782 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2783 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
2784 !FRAME_GROWS_DOWNWARD handling to...
2785 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2786 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
2787 !FRAME_GROWS_DOWNWARD handling to...
2788 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2789 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
2790 !FRAME_GROWS_DOWNWARD handling to...
2791 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2792 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
2794 (rs6000_starting_frame_offset): New function.
2795 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
2796 !FRAME_GROWS_DOWNWARD handling to...
2797 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2798 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
2799 !FRAME_GROWS_DOWNWARD handling to...
2800 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2801 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
2802 !FRAME_GROWS_DOWNWARD handling to...
2803 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2804 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
2805 (rs6000_starting_frame_offset): New function.
2806 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
2807 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
2808 * config/vax/vax.c (vax_starting_frame_offset): New function.
2809 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
2810 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2811 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
2812 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
2813 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2814 * system.h (STARTING_FRAME_OFFSET): Poison.
2816 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2818 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
2819 SCALAR_TYPE_MODE instead of TYPE_MODE.
2821 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2822 Alan Hayward <alan.hayward@arm.com>
2823 David Sherwood <david.sherwood@arm.com>
2825 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
2827 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2828 Alan Hayward <alan.hayward@arm.com>
2829 David Sherwood <david.sherwood@arm.com>
2831 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
2833 2017-10-23 Richard Biener <rguenther@suse.de>
2835 PR tree-optimization/82129
2836 * tree-ssa-pre.c (bitmap_set_and): Remove.
2837 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
2838 canonicalizing expressions in the set to those with lowest
2839 ID rather than taking that from the first edge.
2841 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2843 * combine.c (rtx_equal_for_field_assignment_p): Use
2844 byte_lowpart_offset.
2846 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2847 Alan Hayward <alan.hayward@arm.com>
2848 David Sherwood <david.sherwood@arm.com>
2850 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
2851 to a SUBREG_PROMOTED_VAR.
2853 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2854 Alan Hayward <alan.hayward@arm.com>
2855 David Sherwood <david.sherwood@arm.com>
2857 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
2858 (expand_debug_source_expr): Likewise.
2859 * combine.c (combine_simplify_rtx): Likewise.
2860 * cse.c (fold_rtx): Likewise.
2861 * optabs.c (expand_float): Likewise.
2862 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2863 (simplify_binary_operation_1): Likewise.
2865 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2866 Alan Hayward <alan.hayward@arm.com>
2867 David Sherwood <david.sherwood@arm.com>
2869 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
2870 (record_promoted_value): Likewise.
2871 * expr.c (expand_expr_real_2): Likewise.
2872 * ree.c (update_reg_equal_equiv_notes): Likewise.
2873 (combine_set_extension): Likewise.
2874 * rtlanal.c (low_bitmask_len): Likewise.
2875 * simplify-rtx.c (neg_const_int): Likewise.
2876 (simplify_binary_operation_1): Likewise.
2878 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2879 Alan Hayward <alan.hayward@arm.com>
2880 David Sherwood <david.sherwood@arm.com>
2882 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
2883 * regcprop.c (maybe_mode_change): Likewise.
2884 * reload1.c (alter_reg): Likewise.
2886 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2888 * inchash.h (inchash::hash::add_wide_int): New function.
2889 * lto-streamer-out.c (hash_tree): Use it.
2891 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2893 * inchash.h (inchash::hash::add_wide_int): Rename to...
2894 (inchash::hash::add_hwi): ...this.
2895 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
2896 (polymorphic_call_target_hasher::hash): Likewise.
2897 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
2898 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
2899 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
2900 * lto-streamer-out.c (hash_tree): Likewise.
2901 * optc-save-gen.awk: Likewise.
2902 * tree.c (add_expr): Likewise.
2904 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
2907 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
2908 for ordered inequality comparisons even with TARGET_IEEE_FP.
2910 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
2913 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
2914 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
2915 Expand with cmp<dwi>_doubleword.
2917 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2919 * extend.texi: Add x86 specific to 'nocf_check' attribute.
2920 List CET intrinsics.
2921 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
2922 specific to -fcf-protection option.
2924 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2926 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
2927 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
2928 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
2929 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
2930 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
2931 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
2932 (extra_objs): Add cet.o for Linux/x86 targets.
2933 (tmake_file): Add i386/t-cet for Linux/x86 targets.
2934 * config/i386/cet.c: New file.
2935 * config/i386/cetintrin.h: Likewise.
2936 * config/i386/t-cet: Likewise.
2937 * config/i386/cpuid.h (bit_SHSTK): New.
2938 (bit_IBT): Likewise.
2939 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
2940 pass IBT and SHSTK bits.
2941 * config/i386/i386-builtin-types.def
2942 (VOID_FTYPE_UNSIGNED_PVOID): New.
2943 (VOID_FTYPE_UINT64_PVOID): Likewise.
2944 * config/i386/i386-builtin.def: Add CET intrinsics.
2945 * config/i386/i386-c.c (ix86_target_macros_internal): Add
2946 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
2947 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
2948 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
2950 * config/i386/i386.c (rest_of_insert_endbranch): New.
2951 (pass_data_insert_endbranch): Likewise.
2952 (pass_insert_endbranch): Likewise.
2953 (make_pass_insert_endbranch): Likewise.
2954 (ix86_notrack_prefixed_insn_p): Likewise.
2955 (ix86_target_string): Add -mibt, -mshstk flags.
2956 (ix86_option_override_internal): Add flag_cf_protection
2958 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
2959 (ix86_print_operand): Add 'notrack' prefix output.
2960 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
2961 (ix86_expand_builtin): Expand CET intrinsics.
2962 (x86_output_mi_thunk): Add 'endbranch' instruction.
2963 * config/i386/i386.h (TARGET_IBT): New.
2964 (TARGET_IBT_P): Likewise.
2965 (TARGET_SHSTK): Likewise.
2966 (TARGET_SHSTK_P): Likewise.
2967 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
2968 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
2969 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
2970 (builtin_setjmp_setup): New pattern.
2971 (builtin_longjmp): Likewise.
2972 (rdssp<mode>): Likewise.
2973 (incssp<mode>): Likewise.
2974 (saveprevssp): Likewise.
2975 (rstorssp): Likewise.
2976 (wrss<mode>): Likewise.
2977 (wruss<mode>): Likewise.
2978 (setssbsy): Likewise.
2979 (clrssbsy): Likewise.
2980 (nop_endbr): Likewise.
2981 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
2983 * config/i386/immintrin.h: Include <cetintrin.h>.
2984 * config/i386/linux-common.h
2985 (file_end_indicate_exec_stack_and_cet): New prototype.
2986 (TARGET_ASM_FILE_END): New.
2988 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
2990 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
2991 latencies instead of having separate table; make difference between
2992 integer and float costs.
2993 * i386.h (processor_costs): Remove scalar_stmt_cost,
2994 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
2995 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
2997 * x86-tune-costs.h: Remove entries which has been removed in
2998 procesor_costs from all tables; make cond_taken_branch_cost
2999 and cond_not_taken_branch_cost COST_N_INSNS based.
3001 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
3003 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
3005 2017-10-20 Jakub Jelinek <jakub@redhat.com>
3007 * config/i386/i386.md (isa): Remove fma_avx512f.
3008 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
3009 <avx512>_fmadd_<mode>_mask3<round_name>,
3010 <avx512>_fmsub_<mode>_mask<round_name>,
3011 <avx512>_fmsub_<mode>_mask3<round_name>,
3012 <avx512>_fnmadd_<mode>_mask<round_name>,
3013 <avx512>_fnmadd_<mode>_mask3<round_name>,
3014 <avx512>_fnmsub_<mode>_mask<round_name>,
3015 <avx512>_fnmsub_<mode>_mask3<round_name>,
3016 <avx512>_fmaddsub_<mode>_mask<round_name>,
3017 <avx512>_fmaddsub_<mode>_mask3<round_name>,
3018 <avx512>_fmsubadd_<mode>_mask<round_name>,
3019 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
3020 (*vec_widen_umult_even_v16si<mask_name>,
3021 *vec_widen_smult_even_v16si<mask_name>): Likewise.
3022 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
3024 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3026 * extend.texi: Add 'nocf_check' documentation.
3027 * gimple.texi: Add second parameter to
3028 gimple_build_call_from_tree.
3029 * invoke.texi: Add -fcf-protection documentation.
3030 * rtl.texi: Add REG_CALL_NOTRACK documenation.
3032 2017-10-20 Richard Biener <rguenther@suse.de>
3034 PR tree-optimization/82473
3035 * tree-vect-loop.c (vectorizable_reduction): Properly get at
3036 the largest input type.
3038 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3040 * c-attribs.c (handle_nocf_check_attribute): New function.
3041 (c_common_attribute_table): Add 'nocf_check' handling.
3042 * gimple-parser.c: Add second argument NULL to
3043 gimple_build_call_from_tree.
3044 * attrib.c (comp_type_attributes): Check nocf_check attribute.
3045 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
3047 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
3048 * common.opt: Add fcf-protection flag.
3049 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
3050 * flag-types.h: Add enum cf_protection_level.
3051 * gimple.c (gimple_build_call_from_tree): Add second parameter.
3052 Add 'nocf_check' attribute propagation to gimple call.
3053 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
3054 (gimple_build_call_from_tree): Update prototype.
3055 (gimple_call_nocf_check_p): New function.
3056 (gimple_call_set_nocf_check): Likewise.
3057 * gimplify.c: Add second argument to gimple_build_call_from_tree.
3058 * ipa-icf.c: Add nocf_check attribute in statement hash.
3059 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
3060 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
3061 * toplev.c (process_options): Add flag_cf_protection handling.
3063 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3065 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
3067 2017-10-20 Richard Biener <rguenther@suse.de>
3069 PR tree-optimization/82603
3070 * tree-if-conv.c (predicate_mem_writes): Make sure to only
3071 remove false predicated stores.
3073 2017-10-20 Richard Biener <rguenther@suse.de>
3075 * graphite-isl-ast-to-gimple.c
3076 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
3077 Remove return value and simplify, dump copied stmt after lhs
3079 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
3080 Reduce dump verbosity.
3081 (gsi_insert_earliest): Likewise.
3082 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
3083 * graphite.c (print_global_statistics): Adjust dumping.
3084 (print_graphite_scop_statistics): Likewise.
3085 (print_graphite_statistics): Do not dump loops here.
3086 (graphite_transform_loops): But here.
3088 2017-10-20 Nicolas Roche <roche@adacore.com>
3090 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
3091 * configure: Regenerate.
3093 2017-10-20 Jakub Jelinek <jakub@redhat.com>
3096 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
3097 functions when optimizing replace GIMPLE_RETURN stmts with
3098 calls to __builtin_unreachable ().
3101 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
3102 for -fsanitize=thread link of executables.
3103 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
3104 link of executables.
3107 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
3109 (<shift_insn><mode>3<mask_name>): Change the last of the 3
3110 define_insns for logical vector shifts to use VI248_AVX512BW
3111 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
3112 condition, useless isa and prefix attributes. Change the first
3113 2 of these define_insns to ...
3114 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
3115 define_insn for avx512vl.
3116 (<shift_insn><mode>3): ... and this, new define_insn without
3117 masking for non-avx512vl.
3120 * config/i386/sse.md (*andnot<mode>3,
3121 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
3122 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
3123 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
3124 not applied use empty suffix even for TARGET_AVX512VL.
3125 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
3126 is applied, supply evex,evex or evex,evex,evex instead of just
3129 2017-10-20 Julia Koval <julia.koval@intel.com>
3131 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
3132 (OPTION_MASK_ISA_GFNI_UNSET): New.
3133 (ix86_handle_option): Handle OPT_mgfni.
3134 * config/i386/cpuid.h (bit_GFNI): New.
3135 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
3136 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
3137 * config/i386/i386.c (ix86_target_string): Add -mgfni.
3138 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
3139 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
3140 * config/i386/i386.opt: Add mgfni.
3142 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
3144 * config/msp430/msp430.c (msp430_option_override): Disable
3145 -fdelete-null-pointer-checks.
3146 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
3148 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3150 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
3151 of x87 and SSE instructions.
3153 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3155 * asan.c (create_cond_insert_point): Do not update edge count.
3156 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
3157 (afdo_propagate_circuit): Likewise.
3158 (afdo_calculate_branch_prob): Likewise.
3159 (afdo_annotate_cfg): Likewise.
3160 * basic-block.h (struct edge_def): Remove count.
3161 (edge_def::count): New accessor.
3162 * bb-reorder.c (rotate_loop): Update.
3163 (find_traces_1_round): Update.
3164 (connect_traces): Update.
3165 (sanitize_hot_paths): Update.
3166 * cfg.c (unchecked_make_edge): Update.
3167 (make_single_succ_edge): Update.
3168 (check_bb_profile): Update.
3169 (dump_edge_info): Update.
3170 (update_bb_profile_for_threading): Update.
3171 (scale_bbs_frequencies_int): Update.
3172 (scale_bbs_frequencies_gcov_type): Update.
3173 (scale_bbs_frequencies_profile_count): Update.
3174 (scale_bbs_frequencies): Update.
3175 * cfganal.c (connect_infinite_loops_to_exit): Update.
3176 * cfgbuild.c (compute_outgoing_frequencies): Update.
3177 (find_many_sub_basic_blocks): Update.
3178 * cfgcleanup.c (try_forward_edges): Update.
3179 (try_crossjump_to_edge): Update
3180 * cfgexpand.c (expand_gimple_cond): Update
3181 (expand_gimple_tailcall): Update
3182 (construct_exit_block): Update
3183 * cfghooks.c (verify_flow_info): Update
3184 (redirect_edge_succ_nodup): Update
3185 (split_edge): Update
3186 (make_forwarder_block): Update
3187 (duplicate_block): Update
3188 (account_profile_record): Update
3189 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
3190 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
3191 * cfgloopmanip.c (scale_loop_profile): Update.
3193 (lv_adjust_loop_entry_edge): Update.
3194 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
3195 (force_nonfallthru_and_redirect): Update.
3196 (purge_dead_edges): Update.
3197 (rtl_flow_call_edges_add): Update.
3198 * cgraphunit.c (init_lowered_empty_function): Update.
3199 (cgraph_node::expand_thunk): Update.
3200 * gimple-pretty-print.c (dump_probability): Update.
3201 (dump_edge_probability): Update.
3202 * gimple-ssa-isolate-paths.c (isolate_path): Update.
3203 * haifa-sched.c (sched_create_recovery_edges): Update.
3204 * hsa-gen.c (convert_switch_statements): Update.
3205 * ifcvt.c (dead_or_predicable): Update.
3206 * ipa-inline-transform.c (inline_transform): Update.
3207 * ipa-split.c (split_function): Update.
3208 * ipa-utils.c (ipa_merge_profiles): Update.
3209 * loop-doloop.c (add_test): Update.
3210 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
3211 * lto-streamer-in.c (input_cfg): Update.
3212 (input_function): Update.
3213 * lto-streamer-out.c (output_cfg): Update.
3214 * modulo-sched.c (sms_schedule): Update.
3215 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
3216 * predict.c (maybe_hot_edge_p): Update.
3217 (unlikely_executed_edge_p): Update.
3218 (probably_never_executed_edge_p): Update.
3219 (dump_prediction): Update.
3220 (drop_profile): Update.
3221 (propagate_unlikely_bbs_forward): Update.
3222 (determine_unlikely_bbs): Update.
3223 (force_edge_cold): Update.
3224 * profile.c (compute_branch_probabilities): Update.
3225 * reg-stack.c (better_edge): Update.
3226 * shrink-wrap.c (handle_simple_exit): Update.
3227 * tracer.c (better_p): Update.
3228 * trans-mem.c (expand_transaction): Update.
3229 (split_bb_make_tm_edge): Update.
3230 * tree-call-cdce.c: Update.
3231 * tree-cfg.c (gimple_find_sub_bbs): Update.
3232 (gimple_split_edge): Update.
3233 (gimple_duplicate_sese_region): Update.
3234 (gimple_duplicate_sese_tail): Update.
3235 (gimple_flow_call_edges_add): Update.
3236 (insert_cond_bb): Update.
3237 (execute_fixup_cfg): Update.
3238 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
3239 * tree-complex.c (expand_complex_div_wide): Update.
3240 * tree-eh.c (lower_resx): Update.
3241 (unsplit_eh): Update.
3242 (cleanup_empty_eh_move_lp): Update.
3243 * tree-inline.c (copy_edges_for_bb): Update.
3244 (freqs_to_counts): Update.
3245 (copy_cfg_body): Update.
3246 * tree-ssa-dce.c (remove_dead_stmt): Update.
3247 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
3248 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
3249 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
3250 (unloop_loops): Update.
3251 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
3252 * tree-ssa-loop-split.c (connect_loops): Update.
3253 (split_loop): Update.
3254 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
3255 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
3256 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
3257 * tree-ssa-reassoc.c (branch_fixup): Update.
3258 * tree-ssa-tail-merge.c (replace_block_by): Update.
3259 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
3260 (compute_path_counts): Update.
3261 (update_profile): Update.
3262 (recompute_probabilities): Update.
3263 (update_joiner_offpath_counts): Update.
3264 (estimated_freqs_path): Update.
3265 (freqs_to_counts_path): Update.
3266 (clear_counts_path): Update.
3267 (ssa_fix_duplicate_block_edges): Update.
3268 (duplicate_thread_path): Update.
3269 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
3270 (case_bit_test_cmp): Update.
3271 (collect_switch_conv_info): Update.
3272 (gen_inbound_check): Update.
3273 (do_jump_if_equal): Update.
3274 (emit_cmp_and_jump_insns): Update.
3275 * tree-tailcall.c (decrease_profile): Update.
3276 (eliminate_tail_call): Update.
3277 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
3278 (vect_do_peeling): Update.
3279 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
3280 * ubsan.c (ubsan_expand_null_ifn): Update.
3281 (ubsan_expand_ptr_ifn): Update.
3282 * value-prof.c (gimple_divmod_fixed_value): Update.
3283 (gimple_mod_pow2): Update.
3284 (gimple_mod_subtract): Update.
3285 (gimple_ic): Update.
3286 (gimple_stringop_fixed_value): Update.
3288 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
3291 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
3293 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
3295 PR rtl-optimization/82395
3296 * ira-color.c (allocno_priority_compare_func): Fix comparison step
3297 based on non_spilled_static_chain_regno_p.
3299 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
3301 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
3302 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
3303 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
3305 2017-10-19 Martin Sebor <msebor@redhat.com>
3307 PR tree-optimization/82596
3308 * tree.c (array_at_struct_end_p): Handle STRING_CST.
3310 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
3312 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
3313 (get_mem_refs_of_builtin_call): Likewise.
3314 * builtins.c (expand_builtin_apply): Adjust call to
3315 allocate_dynamic_stack_space.
3316 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
3317 the third argument to allocate_dynamic_stack_space, otherwise -1.
3318 (expand_builtin): Deal with all alloca variants.
3319 (is_inexpensive_builtin): Likewise.
3320 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
3321 * calls.c (special_function_p): Deal with all alloca variants.
3322 (initialize_argument_information): Adjust call to
3323 allocate_dynamic_stack_space.
3324 (expand_call): Likewise.
3325 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
3326 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
3327 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
3328 use it for the stack usage computation.
3329 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
3330 * function.c (gimplify_parameters): Call build_alloca_call_expr.
3331 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
3332 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
3333 (in_loop_p): Remove first argument and useless check.
3334 (pass_walloca::execute): Remove useless test and adjust call to above.
3335 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
3336 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
3337 (gimplify_call_expr): Deal with all alloca variants.
3338 * hsa-gen.c (gen_hsa_alloca): Likewise.
3339 (gen_hsa_insns_for_call): Likewise.
3340 * ipa-pure-const.c (special_builtin_state): Likewise.
3341 * tree-chkp.c (chkp_build_returned_bound): Likewise.
3342 * tree-object-size.c (alloc_object_size): Likewise.
3343 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
3344 (call_may_clobber_ref_p_1): Likewise.
3345 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
3346 (ccp_fold_stmt): Likewise.
3347 (optimize_stack_restore): Likewise.
3348 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
3349 (mark_all_reaching_defs_necessary_1): Likewise.
3350 (propagate_necessity): Likewise.
3351 (eliminate_unnecessary_stmts): Likewise.
3352 * tree.c (build_common_builtin_nodes): Build
3353 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
3354 (build_alloca_call_expr): New function.
3355 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
3356 (CASE_BUILT_IN_ALLOCA): Likewise.
3357 (build_alloca_call_expr): Declare.
3358 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
3360 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
3363 * dwarf2out.c (new_die_raw): New static inline function.
3364 (new_die): Use it to create the DIE.
3365 (add_AT_external_die_ref): Likewise.
3366 (clone_die): Likewise.
3367 (clone_as_declaration): Likewise.
3368 (dwarf2out_vms_debug_main_pointer): Likewise.
3369 (base_type_die): Likewise. Remove early return for corner cases.
3370 Do not call add_pubtype on the DIE here.
3371 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
3372 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
3373 typedefs for base types with DW_AT_endianity. Make sure a DIE with
3374 native order exists for base types, attach the DIE manually and call
3375 add_pubtype on it. Do not equate a reverse order DIE to the type.
3377 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
3379 * config/arm/arm.c (align_ok_ldrd_strd): New function.
3380 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
3382 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
3384 2017-10-19 Jakub Jelinek <jakub@redhat.com>
3386 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
3387 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
3388 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
3389 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
3390 * opts.c (sanitizer_opts): Add builtin.
3391 * ubsan.c (instrument_builtin): New function.
3392 (pass_ubsan::execute): Call it.
3393 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
3394 * doc/invoke.texi: Document -fsanitize=builtin.
3396 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
3397 builtins, store max (log2 (align), 0) into uchar field instead of
3398 align into uptr field.
3399 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
3400 store uchar 0 field instead of uptr 0 field.
3401 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
3402 instead of passing one address of struct with 2 locations pass
3403 two addresses of structs with 1 location each.
3404 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
3405 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
3406 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
3407 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
3408 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
3409 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
3410 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
3411 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
3413 2017-10-19 Martin Liska <mliska@suse.cz>
3416 * file-find.c (remove_prefix): Remove.
3417 * file-find.h (remove_prefix): Likewise.
3418 * gcc-ar.c: Remove smartness of lookup.
3420 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
3422 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
3423 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
3424 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
3426 2017-10-19 Jakub Jelinek <jakub@redhat.com>
3429 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
3430 (setcc + and to xor + setcc): New peephole2.
3432 2017-10-19 Tom de Vries <tom@codesourcery.com>
3434 * doc/sourcebuild.texi (Test Directives, Variants of
3435 dg-require-support): Add dg-require-stack-size.
3437 2017-10-19 Martin Liska <mliska@suse.cz>
3440 * gimplify.c (gimplify_decl_expr): Do not instrument variables
3441 that have a large alignment.
3442 (gimplify_target_expr): Likewise.
3444 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
3446 PR rtl-optimization/82602
3447 * ira.c (rtx_moveable_p): Return false for volatile asm.
3449 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
3452 * config/i386/i386-modes.def (CCGZ): New CC mode.
3453 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
3454 * config/i386/predicates.md (ix86_comparison_operator):
3456 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
3457 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
3458 with double-word subtraction.
3459 (put_condition_code): Handle CCGZmode.
3461 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
3463 * wide-int.cc (debug (const wide_int &)): New.
3464 (debug (const wide_int *)): New.
3465 (debug (const widest_int &)): New.
3466 (debug (const widest_int *)): New.
3468 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
3471 * lra-constraints.c (curr_insn_transform): Use non-input operand
3472 instead of output one for matched reload.
3474 2017-10-18 Bin Cheng <bin.cheng@arm.com>
3476 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
3477 (tree-ssa-loop-ivopts.h): New header file.
3478 (struct builtin_info): New fields.
3479 (classify_builtin_1): Compute and record base and offset parts for
3480 memset builtin partition by calling strip_offset.
3481 (offset_cmp, fuse_memset_builtins): New functions.
3482 (finalize_partitions): Fuse adjacent memset partitions by calling
3484 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
3485 Expose the interface.
3486 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
3488 2017-10-18 Bin Cheng <bin.cheng@arm.com>
3490 PR tree-optimization/82574
3491 * tree-loop-distribution.c (find_single_drs): New parameter. Check
3492 that data reference must be executed exactly once per iteration
3493 against the outermost loop in nest.
3494 (classify_partition): Update call to above function.
3496 2017-10-18 Richard Biener <rguenther@suse.de>
3498 PR tree-optimization/82591
3499 * graphite.c (graphite_transform_loops): Move code gen message
3501 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
3502 Here. Handle scop_to_isl_ast failing.
3503 (scop_to_isl_ast): Limit the number of ISL operations.
3505 2017-10-18 Richard Biener <rguenther@suse.de>
3507 * graphite-isl-ast-to-gimple.c
3508 (translate_isl_ast_to_gimple::set_rename): Simplify.
3509 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
3510 (graphite_copy_stmts_from_block): ... here.
3511 (copy_bb_and_scalar_dependences): Simplify.
3512 (add_parameters_to_ivs_params): Canonicalize.
3513 (generate_entry_out_of_ssa_copies): Simplify.
3514 * graphite-sese-to-poly.c (extract_affine_name): Simplify
3515 by passing in ISL dimension.
3516 (parameter_index_in_region_1): Rename to ...
3517 (parameter_index_in_region): ... this.
3518 (extract_affine): Adjust assert, pass down parameter index.
3519 (add_param_constraints): Use range-info when available.
3520 (build_scop_context): Adjust.
3521 * sese.c (new_sese_info): Adjust.
3522 (free_sese_info): Likewise.
3523 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
3524 Remove unused typedefs.
3525 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
3527 2017-10-18 Martin Liska <mliska@suse.cz>
3529 * combine.c (simplify_compare_const): Add gcc_fallthrough.
3531 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
3533 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
3534 (s390_sched_init): Do not reset s390_sched_state if we entered the
3535 current basic block via a fallthru edge and all others are unlikely.
3537 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
3539 * config/s390/s390.c (NUM_SIDES): New variable.
3540 (LONGRUNNING_THRESHOLD): New variable.
3541 (LATENCY_FACTOR): New variable.
3542 (s390_sched_score): Decrease score for long-running instructions on
3544 (s390_sched_variable_issue): Perform bookkeeping for long-running
3547 2017-10-18 Richard Biener <rguenther@suse.de>
3549 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
3550 Simplify with removal of the parameter rename map.
3551 (set_rename): Likewise.
3552 (should_copy_to_new_region): Likewise.
3553 (graphite_copy_stmts_from_block): Likewise.
3554 (copy_bb_and_scalar_dependences): Remove initialization of
3555 unused copied_bb_map.
3557 (copy_internal_parameters): Likewise.
3558 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
3559 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
3560 Use INTEGRAL_TYPE_P.
3561 (parameter_index_in_region_1): Rename to ...
3562 (assign_parameter_index_in_region): ... this. Assert we have
3563 a parameter we handle.
3564 (scan_tree_for_params): Adjust.
3565 * sese.h (parameter_rename_map_t): Remove.
3566 (struct sese_info_t): Remove unused parameter_rename_map and
3567 copied_bb_map members.
3568 * sese.c (new_sese_info): Adjust.
3569 (free_sese_info): Likewise.
3571 2017-10-18 Martin Liska <mliska@suse.cz>
3574 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
3575 on an abnormal edge.
3577 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
3579 * doc/invoke.texi (ffunction-sections and fdata-sections):
3582 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
3584 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
3585 the use statement can throw internally.
3587 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
3589 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
3590 any RTX present on the RHS of a SET.
3591 * compare-elim.c (try_eliminate_compare): Restore comment.
3593 2017-10-17 Jakub Jelinek <jakub@redhat.com>
3595 * langhooks.h (struct lang_hooks): Document that tree_size langhook
3596 may be also called on tcc_type nodes.
3597 * langhooks.c (lhd_tree_size): Likewise.
3599 2017-10-17 David Malcolm <dmalcolm@redhat.com>
3601 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
3602 format_warning_at_substring.
3603 (maybe_warn): Convert source_range * param to a location_t. Pass
3604 UNKNOWN_LOCATION rather than NULL to fmtwarn.
3605 (format_directive): Remove code to extract source_ranges and
3606 source_range * in favor of just a location_t.
3607 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
3609 * substring-locations.c (format_warning_va): Convert
3610 source_range * param to a location_t.
3611 (format_warning_at_substring): Likewise.
3612 * substring-locations.h (format_warning_va): Likewise.
3613 (format_warning_at_substring): Likewise.
3615 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
3617 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
3619 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
3620 and scatter/gather ops.
3622 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
3623 vec_gather_load and vec_scatter_store.
3624 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
3625 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
3626 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
3627 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
3628 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
3629 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
3631 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
3633 * reg-stack.c (compare_for_stack_reg): Add bool argument.
3634 Detect FTST instruction and handle its register pops. Only pop
3635 second operand if can_pop_second_op is true.
3636 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
3637 set can_pop_second_op to false in the compare_for_stack_reg call.
3639 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
3640 output_fp_compare for stack register operands.
3641 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
3642 instructions here. Do not emit stack register pops here. Assert
3643 that FCOMPP pops next to top stack register. Rewrite function.
3645 2017-10-17 Nathan Sidwell <nathan@acm.org>
3648 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
3649 use DECL_ASSEMBLER_NAME_RAW.
3652 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
3655 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
3656 Wilco Dijkstra <wilco.dijkstra@arm.com>
3658 * builtins.c (expand_builtin_update_setjmp_buf): Add a
3659 converstion to Pmode from the buf_addr.
3661 2017-10-17 Richard Biener <rguenther@suse.de>
3663 * graphite-dependences.c (scop_get_reads_and_writes): Change
3664 output parameters to references.
3666 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
3668 PR 71026/tree-optimization
3669 * fold-const.c (distribute_real_division): Removed.
3670 (fold_binary_loc): Remove calls to distribute_real_divison.
3672 2017-10-17 Richard Biener <rguenther@suse.de>
3674 * graphite-scop-detection.c
3675 (scop_detection::stmt_has_simple_data_refs_p): Always use
3676 the full nest as region.
3677 (try_generate_gimple_bb): Likewise.
3678 * sese.c (scalar_evolution_in_region): Simplify now that
3679 SCEV can handle instantiation in regions.
3680 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
3681 in the non-loop part of a function if requested.
3683 2017-10-17 Richard Biener <rguenther@suse.de>
3685 PR tree-optimization/82563
3686 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
3688 (graphite_regenerate_ast_isl): Call it.
3689 * graphite-scop-detection.c (build_scops): Remove entry edge split.
3691 2017-10-17 Jakub Jelinek <jakub@redhat.com>
3693 PR tree-optimization/82549
3694 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
3695 Formatting fixes. Instead of calling make_bit_field_ref with negative
3698 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
3700 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
3701 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
3702 _mm_maskz_reduce_ss): New.
3703 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
3704 __builtin_ia32_reducess_mask): Ditto..
3705 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
3706 * config/i386/sse.md (reduces<mode>): Renamed to ...
3707 (reduces<mode><mask_scalar_name>): ... this.
3708 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
3710 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
3711 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
3713 2017-10-16 David Malcolm <dmalcolm@redhat.com>
3715 * Makefile.in (OBJS): Add unique-ptr-tests.o.
3716 * selftest-run-tests.c (selftest::run_tests): Call
3717 selftest::unique_ptr_tests_cc_tests.
3718 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
3719 * unique-ptr-tests.cc: New file.
3721 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
3724 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
3726 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
3727 (make_hard_regno_born, make_hard_regno_dead): Update
3728 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
3730 2017-10-16 Jeff Law <law@redhat.com>
3732 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
3734 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
3736 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
3738 2017-10-16 Olivier Hainque <hainque@adacore.com>
3740 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
3741 with_cpu if we were configured for an e500v2 target cpu name.
3743 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
3745 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
3746 * doc/invoke.texi: Document +nodsp as a valid extension for
3749 2017-10-16 Martin Liska <mliska@suse.cz>
3751 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
3752 (test_set_range): Likewise.
3753 (test_range_functions): Rename to ...
3754 (test_bit_in_range): ... this.
3755 (sbitmap_c_tests): Add new test.
3757 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3759 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
3761 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
3762 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
3764 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3766 * config/aarch64/aarch64-builtins.c
3767 (aarch64_types_quadopu_lane_qualifiers): New.
3768 (TYPES_QUADOPU_LANE): New.
3769 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
3770 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
3771 (aarch64_<sur>dot_laneq<vsi2qi>): New.
3772 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
3773 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
3774 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
3775 (Vdottype, DOTPROD): New.
3776 (sur): Add SDOT and UDOT.
3778 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3780 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
3781 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
3782 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
3784 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
3785 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
3786 Enable TARGET_DOTPROD.
3787 (cortex-a75.cortex-a55): Likewise.
3788 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
3790 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3792 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
3793 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
3795 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
3797 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
3798 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
3799 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
3800 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
3801 * config/arm/types.md (neon_dot, neon_dot_q): New.
3802 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
3804 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3806 * config/arm/arm.h (TARGET_DOTPROD): New.
3807 * config/arm/arm.c (arm_arch_dotprod): New.
3808 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
3809 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
3810 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
3811 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
3812 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
3813 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
3814 * doc/invoke.texi (armv8.2-a): Document dotprod
3816 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
3818 * i386.c (ix86_vec_cost): New function.
3819 (ix86_rtx_costs): Handle vector operations better.
3820 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
3821 * x86-tune-costs.h: Add new costs to all tables.
3823 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
3825 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
3827 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
3828 divsd, sqrtss and sqrtsd
3829 * x86-tune-costs.h: Add new entries to all costs.
3830 (znver1_cost): Fix to match real instruction latencies.
3832 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3833 Michael Collison <michael.collison@arm.com>
3835 * compare-elim.c: Include emit-rtl.h.
3836 (can_merge_compare_into_arith): New function.
3837 (try_validate_parallel): Likewise.
3838 (try_merge_compare): Likewise.
3839 (try_eliminate_compare): Call the above when no previous clobber
3841 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
3844 2017-10-14 Jakub Jelinek <jakub@redhat.com>
3848 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
3849 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
3853 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
3854 to be any operand_equal_p operands. For & (B - 1) require
3855 B to be power of 2. Recognize
3856 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
3858 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
3861 * optabs.c (expand_memory_blockage): Fix call of
3862 targetm.have_memory_blockage.
3864 2017-10-14 Jakub Jelinek <jakub@redhat.com>
3867 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
3868 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
3869 objects to extra_objs instead of overwriting it.
3871 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
3873 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
3874 Use any_fp_register_operand as operand[3] predicate. Simplify
3875 equality test for operands[2] and operands[4] memory location.
3876 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
3877 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
3878 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
3879 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
3880 any_fp_register_operand as operand[1] predicate. Simplify
3881 equality test for operands[0] and operands[3] memory location.
3882 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
3883 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
3884 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
3886 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
3888 * target-insns.def: Add memory_blockage.
3889 * optabs.c (expand_memory_blockage): New function.
3890 (expand_asm_memory_barrier): Rename ...
3891 (expand_asm_memory_blockage): ... to this.
3892 (expand_mem_thread_fence): Call expand_memory_blockage
3893 instead of expand_asm_memory_barrier.
3894 (expand_mem_singnal_fence): Ditto.
3895 (expand_atomic_load): Ditto.
3896 (expand_atomic_store): Ditto.
3897 * doc/md.texi (Standard Pattern Names For Generation):
3898 Document memory_blockage instruction pattern.
3900 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
3902 * config/rl78/rl78.c (rl78_emit_libcall): New function.
3903 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
3904 * config/rl78/rl78.md: New define_expand "adddi3".
3906 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
3908 * cfghooks.c (verify_flow_info): Disable check that all probabilities
3911 2017-10-13 Jeff Law <law@redhat.com>
3913 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
3915 2017-10-13 Jakub Jelinek <jakub@redhat.com>
3918 * internal-fn.c (expand_mul_overflow): If both operands have
3919 the same highpart of -1 or 0 and the topmost bit of lowpart
3920 is different, overflow is if res <= 0 rather than res < 0.
3922 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
3924 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
3925 TARGET_P9_VECTOR code for unaligned_load case.
3927 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
3929 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
3931 2017-10-13 Nathan Sidwell <nathan@acm.org>
3933 * tree-core.h (tree_contains_struct): Make bool.
3934 * tree.c (tree_contains_struct): Likewise.
3935 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
3936 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
3937 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
3938 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
3939 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
3941 2017-10-13 Richard Biener <rguenther@suse.de>
3943 * graphite-isl-ast-to-gimple.c
3944 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
3945 parameters and dominance check.
3946 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
3947 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
3948 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
3949 Do not update SSA form here or do intermediate IL verification.
3950 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
3951 (graphite_initialize): Remove check on the number of loops in
3952 the function and inline into graphite_transform_loops.
3953 (graphite_finalize): Inline into graphite_transform_loops.
3954 (graphite_transform_loops): Perform SSA update and IL verification
3956 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
3958 2017-10-13 Richard Biener <rguenther@suse.de>
3960 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
3961 graphite_expression_type_precision): Avoid global constructor
3963 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
3964 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
3965 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
3966 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
3967 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
3968 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
3970 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
3973 * config/i386/i386.h (ix86_red_zone_size): New.
3974 * config/i386/i386.md (push peephole2s): Replace
3975 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
3977 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3978 Alan Hayward <alan.hayward@arm.com>
3979 David Sherwood <david.sherwood@arm.com>
3981 * combine.c (can_change_dest_mode): Reject changes in
3982 REGMODE_NATURAL_SIZE.
3984 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3985 Alan Hayward <alan.hayward@arm.com>
3986 David Sherwood <david.sherwood@arm.com>
3988 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
3989 (expand_debug_source_expr): Likewise.
3990 * combine.c (combine_simplify_rtx): Likewise.
3991 * cse.c (fold_rtx): Likewise.
3992 * fwprop.c (canonicalize_address): Likewise.
3993 * targhooks.c (default_shift_truncation_mask): Likewise.
3995 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3996 Alan Hayward <alan.hayward@arm.com>
3997 David Sherwood <david.sherwood@arm.com>
3999 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
4000 (widened_mode): Likewise.
4001 (expand_unop): Likewise.
4002 * ree.c (transform_ifelse): Likewise.
4003 (merge_def_and_ext): Likewise.
4004 (combine_reaching_defs): Likewise.
4005 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4007 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4008 Alan Hayward <alan.hayward@arm.com>
4009 David Sherwood <david.sherwood@arm.com>
4011 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
4012 * combine.c (gen_lowpart_for_combine): Likewise.
4013 * dwarf2out.c (rtl_for_decl_location): Likewise.
4014 * final.c (alter_subreg): Likewise.
4015 * rtlhooks.c (gen_lowpart_general): Likewise.
4016 (gen_lowpart_if_possible): Likewise.
4018 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4019 Alan Hayward <alan.hayward@arm.com>
4020 David Sherwood <david.sherwood@arm.com>
4022 * calls.c (expand_call): Use subreg_lowpart_offset.
4023 * cse.c (cse_insn): Likewise.
4024 * regcprop.c (copy_value): Likewise.
4025 (copyprop_hardreg_forward_1): Likewise.
4027 2017-10-13 Jakub Jelinek <jakub@redhat.com>
4030 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
4031 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
4032 =Q constraints to +Q and into insn condition add check
4033 that operands[0] and operands[1] are equal.
4034 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
4035 =Q constraints to +Q and into insn condition add check
4036 that operands[0] is equal to either operands[1] or operands[2].
4039 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
4040 instead of handling MINUS_EXPR twice (once for each argument),
4041 canonicalize operand order and handle just once, use rtype where
4042 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
4045 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
4046 any values of __C while still being pattern recognizable as a simple
4049 2017-10-13 Richard Biener <rguenther@suse.de>
4051 PR tree-optimization/82451
4053 2017-10-02 Richard Biener <rguenther@suse.de>
4055 PR tree-optimization/82355
4056 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
4057 a mapping for the enclosing loop but avoid generating one for
4059 (copy_bb_and_scalar_dependences): Remove premature codegen
4060 error on PHIs in blocks duplicated into multiple places.
4061 * graphite-scop-detection.c
4062 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
4063 in the region use it as loop and nest to analyze the DR in.
4064 (try_generate_gimple_bb): Likewise.
4065 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
4066 (add_loop_constraints): For blocks in a loop not in the region
4067 create a dimension with a single iteration.
4068 * sese.h (gbb_loop_at_index): Remove assert.
4070 * cfgloop.c (loop_preheader_edge): For the loop tree root
4071 return the single successor of the entry block.
4072 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
4073 Reset the SCEV hashtable and niters.
4074 * graphite-scop-detection.c
4075 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
4076 assert that we only have POLYNOMIAL_CHREC that vary in loops
4077 contained in the region.
4078 (scop_detection::graphite_can_represent_expr): Adjust.
4079 (scop_detection::stmt_has_simple_data_refs_p): For loops
4080 not in the region set loop to NULL. The nest is now the
4081 entry edge to the region.
4082 (try_generate_gimple_bb): Likewise.
4083 * sese.c (scalar_evolution_in_region): Adjust for
4084 instantiate_scev change.
4085 * tree-data-ref.h (graphite_find_data_references_in_stmt):
4086 Make nest parameter the edge into the region.
4087 (create_data_ref): Likewise.
4088 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
4089 entry edge into a region and adjust instantiate_scev calls.
4090 (create_data_ref): Likewise.
4091 (graphite_find_data_references_in_stmt): Likewise.
4092 (find_data_references_in_stmt): Pass the loop preheader edge
4093 from the nest argument.
4094 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
4095 parameter the edge into the region.
4096 (instantiate_parameters): Use the loop preheader edge as entry.
4097 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
4099 (get_instantiated_value_entry): Make instantiate_below parameter
4100 the edge into the region.
4101 (instantiate_scev_name): Likewise. Adjust dominance checks,
4102 when we cannot use loop-based instantiation instantiate by
4103 walking use-def chains.
4104 (instantiate_scev_poly): Adjust.
4105 (instantiate_scev_binary): Likewise.
4106 (instantiate_scev_convert): Likewise.
4107 (instantiate_scev_not): Likewise.
4108 (instantiate_array_ref): Remove.
4109 (instantiate_scev_3): Likewise.
4110 (instantiate_scev_2): Likewise.
4111 (instantiate_scev_1): Likewise.
4112 (instantiate_scev_r): Do not blindly handle N-operand trees.
4113 Do not instantiate array-refs. Handle all constants and invariants.
4114 (instantiate_scev): Make instantiate_below parameter
4115 the edge into the region.
4116 (resolve_mixers): Use the loop preheader edge for the region
4117 parameter to instantiate_scev_r.
4118 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
4120 2017-10-13 Richard Biener <rguenther@suse.de>
4122 PR tree-optimization/82525
4123 * graphite-isl-ast-to-gimple.c
4124 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
4126 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
4127 Fail code generation when we cannot represent the isl integer.
4128 (binary_op_to_tree): Elide modulo operations that are no-ops
4129 in the type we code generate. Remove now superfluous code
4132 2017-10-13 Richard Biener <rguenther@suse.de>
4134 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
4135 (scop_detection::harmful_loop_in_region): Remove premature
4136 IV type restriction.
4137 (scop_detection::graphite_can_represent_scev): We can handle
4138 pointer IVs just fine.
4140 2017-10-13 Alan Modra <amodra@gmail.com>
4142 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
4143 "Clobbers and Scratch Registers". Add paragraph on
4144 alternative to clobbers for scratch registers and OpenBLAS
4147 2017-10-13 Alan Modra <amodra@gmail.com>
4149 * doc/extend.texi (Clobbers): Correct vax example. Delete old
4150 example of a memory input for a string of known length. Move
4151 commentary out of table. Add a number of new examples
4152 covering array memory inputs.
4154 2017-10-12 Martin Liska <mliska@suse.cz>
4156 PR tree-optimization/82493
4157 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
4158 (test_range_functions): New function.
4159 (sbitmap_c_tests): Likewise.
4160 * selftest-run-tests.c (selftest::run_tests): Run new tests.
4161 * selftest.h (sbitmap_c_tests): New function.
4163 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
4165 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
4167 * config/rs6000/amo.h: Fix spacing issue.
4169 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4172 * config/i386/i386.md (*ashl<mode>3_mask_1,
4173 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
4174 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
4177 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
4179 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
4180 (profile_probability): Set max_probability
4181 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
4183 * profile-count.c (profile_probability::differs_from_p): Do not
4184 rely on max_probaiblity == 10000
4186 2017-10-12 Jeff Law <law@redhat.com>
4188 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
4191 2017-10-12 Martin Sebor <msebor@redhat.com>
4195 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
4196 (handle_alias_pairs): Call it.
4197 * common.opt (-Wattribute-alias): New option.
4198 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
4199 * doc/invoke.texi (-Wattribute-alias): Document.
4201 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
4204 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
4206 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4208 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4209 (make_hard_regno_born, make_hard_regno_dead): Update
4210 bb_killed_pseudos and bb_gen_pseudos.
4212 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
4214 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
4216 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
4218 * config/alpha/alpha.c (alpha_split_conditional_move):
4219 Use std::swap instead of manually swapping.
4220 (alpha_stdarg_optimize_hook): Ditto.
4221 (alpha_canonicalize_comparison): Ditto.
4223 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4225 * tree-loop-distribution.c (struct builtin_info): New struct.
4226 (struct partition): Refactor fields into struct builtin_info.
4227 (partition_free): Free struct builtin_info.
4228 (build_size_arg_loc, build_addr_arg_loc): Delete.
4229 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
4230 information from struct builtin_info.
4231 (find_single_drs): New function refactored from classify_partition.
4232 Also moved builtin validity checks to this function.
4233 (compute_access_range, alloc_builtin): New functions.
4234 (classify_builtin_st, classify_builtin_ldst): New functions.
4235 (classify_partition): Refactor code into functions find_single_drs,
4236 classify_builtin_st and classify_builtin_ldst.
4237 (distribute_loop): Don't do runtime alias check when distributing
4239 (find_seed_stmts_for_distribution): New function.
4240 (pass_loop_distribution::execute): Refactor code finding seed
4241 stmts into above function. Support distribution for the innermost
4242 two-level loop nest. Adjust dump information.
4244 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4246 * tree-loop-distribution.c: Adjust the general comment.
4247 (NUM_PARTITION_THRESHOLD): New macro.
4248 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
4249 (classify_partition): Skip builtin pattern of loop nest's inner loop.
4250 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
4251 in call to build_partition_graph.
4252 (finalize_partitions): New parameter. Make loop distribution more
4253 conservative by fusing more partitions.
4254 (distribute_loop): Don't do runtime alias check in case of loop nest
4256 (find_seed_stmts_for_distribution): New function.
4257 (prepare_perfect_loop_nest): New function.
4258 (pass_loop_distribution::execute): Refactor code finding seed stmts
4259 and loop nest into above functions. Support loop nest distribution.
4260 Adjust dump information accordingly.
4262 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4264 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
4265 and set PTYPE_SEQUENTIAL for merged partition.
4267 2017-10-12 Richard Biener <rguenther@suse.de>
4269 PR tree-optimization/69728
4271 2017-09-19 Richard Biener <rguenther@suse.de>
4273 PR tree-optimization/69728
4274 * graphite-sese-to-poly.c (schedule_error): New global.
4275 (add_loop_schedule): Handle empty domain by failing the
4277 (build_original_schedule): Handle schedule_error.
4279 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
4280 domain by returning an unchanged schedule.
4282 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4284 * genrecog.c (validate_pattern): For VEC_SELECT verify that
4285 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
4287 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
4289 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
4292 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4295 * expr.c (store_field): Don't optimize away bitsize == 0 store
4296 from CALL_EXPR with addressable return type.
4298 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
4300 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
4301 * config/rs6000/rs6000.md (sel): Delete mode attribute.
4302 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
4303 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
4304 TARGET_ISEL instead of TARGET_ISEL<sel>.
4306 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
4308 * config/rs6000/rs6000.c
4309 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
4311 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
4313 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
4315 (reg_or_cint_operand): Fix comment.
4316 (reg_or_zero_operand): New predicate.
4317 * config/rs6000/rs6000-protos.h (output_isel): Delete.
4318 * config/rs6000/rs6000.c (output_isel): Delete.
4319 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
4320 instead of reg_or_cint_operand. Output instruction directly (not via
4322 (isel_unsigned_<mode>): Ditto.
4323 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
4324 gpc_reg_operand. Add an instruction alternative for this. Output
4325 instruction directly.
4326 (*isel_reversed_unsigned_<mode>): Ditto.
4328 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4330 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
4331 (TARGET_CANONICALIZE_COMPARISON): Define.
4333 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
4336 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
4337 Check whether the dest is REG before adding REG_EQUIV note.
4339 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
4342 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4344 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4345 (make_hard_regno_born, make_hard_regno_dead): Update
4346 bb_killed_pseudos and bb_gen_pseudos.
4348 2017-10-11 Nathan Sidwell <nathan@acm.org>
4350 * incpath.h (enum incpath_kind): Name enum, prefix values.
4351 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
4352 * incpath.c (heads, tails): Use INC_MAX.
4353 (add_env_var_paths, add_standard_paths): Use incpath_kind.
4354 (merge_include_chains, split_quote_chain,
4355 register_include_chains): Update incpath_kind names.
4356 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
4357 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
4359 (add_framework_path, darwin_register_objc_includes): Likewise.
4360 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
4362 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4364 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
4365 Do not use float_operator operator predicate.
4366 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
4367 * config/i386/predicates.md (float_operator): Remove predicate.
4369 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4371 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
4372 (*jccxf_i387): Ditto.
4373 (*jcc<mode>_i387): Ditto.
4374 (*jccu<mode>_i387): Ditto.
4375 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
4376 (*jcc_*_i387 splitters): Remove.
4377 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
4378 * config/i386/i386.c (ix86_split_fp_branch): Remove.
4379 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
4382 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
4384 * profile-count.h (slow_safe_scale_64bit): New function.
4385 (safe_scale_64bit): New inline.
4386 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
4387 * profile-count.c: Include wide-int.h
4388 (slow_safe_scale_64bit): New.
4390 2017-10-11 Nathan Sidwell <nathan@acm.org>
4392 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
4393 HAS_DECL_ASSEMBLER_NAME_P.
4394 * gimple-expr.c (gimple_decl_printable_name: Check
4395 HAS_DECL_ASSEMBLER_NAME_P too.
4396 * ipa-utils.h (type_in_anonymous_namespace_p): Check
4397 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
4398 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
4399 * passes.c (rest_of_decl_compilation): Check
4400 HAS_DECL_ASSEMBLER_NAME_P too.
4401 * recog.c (verify_changes): Likewise.
4402 * tree-pretty-print.c (dump_decl_name): Likewise.
4403 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
4405 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
4406 (SET_DECL_ASSEMBLER_NAME): Use it.
4407 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
4408 (COPY_DECL_ASSEMBLER_NAME): Likewise.
4409 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
4411 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
4413 * config.gcc (i386, x86_64): Add extra objects.
4414 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
4415 (ix86_min_insn_size): Declare.
4416 (ix86_issue_rate): Declare.
4417 (ix86_adjust_cost): Declare.
4418 (ia32_multipass_dfa_lookahead): Declare.
4419 (ix86_macro_fusion_p): Declare.
4420 (ix86_macro_fusion_pair_p): Declare.
4421 (ix86_bd_has_dispatch): Declare.
4422 (ix86_bd_do_dispatch): Declare.
4423 (ix86_core2i7_init_hooks): Declare.
4424 (ix86_atom_sched_reorder): Declare.
4425 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
4426 (COSTS_N_BYTES): Move to x86-tune-costs.h.
4427 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
4428 (rip_relative_addr_p): Rename to ...
4429 (ix86_rip_relative_addr_p): ... this one; export.
4430 (memory_address_length): Update.
4431 (ix86_issue_rate): Move to x86-tune-sched.c.
4432 (ix86_flags_dependent): Move to x86-tune-sched.c.
4433 (ix86_agi_dependent): Move to x86-tune-sched.c.
4434 (exact_dependency_1): Move to x86-tune-sched.c.
4435 (exact_store_load_dependency): Move to x86-tune-sched.c.
4436 (ix86_adjust_cost): Move to x86-tune-sched.c.
4437 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
4438 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
4439 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
4440 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
4441 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
4442 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
4443 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
4444 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
4445 (min_insn_size): Rename to ...
4446 (ix86_min_insn_size): ... this one; export.
4447 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
4448 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
4449 (core2i7_first_cycle_multipass_backtrack): Move to
4450 x86-tune-sched-core.c.
4451 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
4452 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
4453 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
4454 (ix86_avoid_jump_mispredicts): Update.
4455 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
4456 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
4457 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
4458 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
4459 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
4460 (MAX_INSN): Move to ix86-tune-sched-bd.c.
4461 (MAX_IMM): Move to ix86-tune-sched-bd.c.
4462 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
4463 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
4464 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
4465 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
4466 (MAX_STORE): Move to ix86-tune-sched-bd.c.
4467 (BIG): Move to ix86-tune-sched-bd.c.
4468 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
4469 (enum insn_path): Move to ix86-tune-sched-bd.c.
4470 (get_mem_group): Move to ix86-tune-sched-bd.c.
4471 (is_cmp): Move to ix86-tune-sched-bd.c.
4472 (dispatch_violation): Move to ix86-tune-sched-bd.c.
4473 (is_branch): Move to ix86-tune-sched-bd.c.
4474 (is_prefetch): Move to ix86-tune-sched-bd.c.
4475 (init_window): Move to ix86-tune-sched-bd.c.
4476 (allocate_window): Move to ix86-tune-sched-bd.c.
4477 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
4478 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
4479 (process_end_window): Move to ix86-tune-sched-bd.c.
4480 (allocate_next_window): Move to ix86-tune-sched-bd.c.
4481 (find_constant): Move to ix86-tune-sched-bd.c.
4482 (get_num_immediates): Move to ix86-tune-sched-bd.c.
4483 (has_immediate): Move to ix86-tune-sched-bd.c.
4484 (get_insn_path): Move to ix86-tune-sched-bd.c.
4485 (get_insn_group): Move to ix86-tune-sched-bd.c.
4486 (count_num_restricted): Move to ix86-tune-sched-bd.c.
4487 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
4488 (add_insn_window): Move to ix86-tune-sched-bd.c.
4489 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
4490 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
4491 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
4492 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
4493 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
4494 (do_dispatch): Move to ix86-tune-sched-bd.c.
4495 (has_dispatch): Move to ix86-tune-sched-bd.c.
4496 * config/i386/t-i386: Add new object files.
4497 * config/i386/x86-tune-costs.h: New file.
4498 * config/i386/x86-tune-sched-atom.c: New file.
4499 * config/i386/x86-tune-sched-bd.c: New file.
4500 * config/i386/x86-tune-sched-core.c: New file.
4501 * config/i386/x86-tune-sched.c: New file.
4503 2017-10-11 Liu Hao <lh_mouse@126.com>
4505 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
4506 the generic version below instead.
4507 (should_colorize): Recognize Windows consoles as terminals
4509 * pretty-print.c [__MINGW32__] (write_all): New function.
4510 [__MINGW32__] (find_esc_head): Likewise.
4511 [__MINGW32__] (find_esc_terminator): Likewise.
4512 [__MINGW32__] (eat_esc_sequence): Likewise.
4513 [__MINGW32__] (mingw_ansi_fputs): New function that handles
4515 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
4518 2017-10-11 Richard Biener <rguenther@suse.de>
4520 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
4521 Properly call analyze_scalar_evolution with the loop of the stmt.
4523 2017-10-11 Richard Biener <rguenther@suse.de>
4525 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
4526 * tree-core.h (tree_base): Add chrec_var union member.
4527 * tree.h (CHREC_VAR): Remove.
4528 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
4529 * tree-chrec.h (build_polynomial_chrec): Adjust.
4530 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
4531 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
4533 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
4535 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
4536 * match.pd: ... here.
4537 ((T) X == (T) Y): Relax condition.
4539 2017-10-11 Bin Cheng <bin.cheng@arm.com>
4541 PR tree-optimization/82472
4542 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
4544 (break_alias_scc_partitions): Update postorder number.
4546 2017-10-11 Martin Liska <mliska@suse.cz>
4549 * opts.c (parse_no_sanitize_attribute): Do not use error_value
4551 * opts.h (parse_no_sanitize_attribute): Remove last argument.
4553 2017-10-11 Martin Liska <mliska@suse.cz>
4555 * print-rtl.c (print_insn): Move declaration of idbuf
4556 to same scope as name.
4558 2017-10-11 Martin Liska <mliska@suse.cz>
4563 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
4566 2017-10-11 Martin Liska <mliska@suse.cz>
4569 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
4572 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
4574 * config.gcc (default_gnu_indirect_function): Default to yes for
4575 arm*-*-linux* with glibc.
4577 2017-10-11 Richard Biener <rguenther@suse.de>
4579 * tree-scalar-evolution.c (get_scalar_evolution): Handle
4580 default-defs and types we do not want to analyze.
4581 (interpret_loop_phi): Replace unreachable code with an assert.
4582 (compute_scalar_evolution_in_loop): Remove and inline ...
4583 (analyze_scalar_evolution_1): ... here, replacing condition with
4584 what makes the intent clearer. Remove handling of cases
4585 get_scalar_evolution now handles.
4587 2017-10-10 Jim Wilson <wilson@tuliptree.org>
4589 PR rtl-optimization/81434
4590 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
4591 comment for main loop. In sched_group_found if, also add checks for
4592 pass and min_cost_group.
4594 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
4596 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
4597 (rs6000_insn_cost): New function.
4598 * config/rs6000/rs6000.md (cost): New attribute.
4600 2017-10-10 Jakub Jelinek <jakub@redhat.com>
4601 H.J. Lu <hongjiu.lu@intel.com>
4605 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
4606 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
4607 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
4608 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
4609 to OPTION_MASK_ISA_AVX512VL - builtins that have both
4610 OPTION_MASK_ISA_MMX and some other bit set require both
4611 mmx and the ISAs without the mmx bit.
4612 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
4613 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
4614 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
4615 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
4616 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
4617 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
4618 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
4619 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
4620 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
4621 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
4622 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
4623 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
4624 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
4625 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
4626 Add OPTION_MASK_ISA_MMX.
4628 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
4630 * config.gcc (armv7*-*-freebsd*): New target.
4631 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
4633 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
4635 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
4636 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
4639 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
4641 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
4642 a template parameter.
4643 (WIDE_INT_REF_FOR): Update accordingly.
4644 * tree.h (wi::int_traits <const_tree>): Delete.
4645 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
4646 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
4647 (wi::tree_to_wide_ref): New typedef.
4648 (wi::to_wide): New function.
4649 * calls.c (get_size_range): Use wi::to_wide when operating on
4651 * cgraph.c (cgraph_node::create_thunk): Likewise.
4652 * config/i386/i386.c (ix86_data_alignment): Likewise.
4653 (ix86_local_alignment): Likewise.
4654 * dbxout.c (stabstr_O): Likewise.
4655 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
4656 * expr.c (const_vector_from_tree): Likewise.
4657 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
4658 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
4659 (fold_negate_expr_1, int_const_binop_1, const_binop)
4660 (fold_convert_const_int_from_real, optimize_bit_field_compare)
4661 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
4662 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
4663 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
4664 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
4665 (fold_not_const, round_up_loc): Likewise.
4666 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
4667 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
4668 (alloca_call_type): Likewise.
4669 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
4670 * godump.c (go_output_typedef): Likewise.
4671 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
4672 * internal-fn.c (get_min_precision): Likewise.
4673 * ipa-cp.c (ipcp_store_vr_results): Likewise.
4674 * ipa-polymorphic-call.c
4675 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
4676 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
4677 (ipa_modify_call_arguments): Likewise.
4678 * match.pd: Likewise.
4679 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
4680 * print-tree.c (print_node_brief, print_node): Likewise.
4681 * stmt.c (expand_case): Likewise.
4682 * stor-layout.c (layout_type): Likewise.
4683 * tree-affine.c (tree_to_aff_combination): Likewise.
4684 * tree-cfg.c (group_case_labels_stmt): Likewise.
4685 * tree-data-ref.c (dr_analyze_indices): Likewise.
4686 (prune_runtime_alias_test_list): Likewise.
4687 * tree-dump.c (dequeue_and_dump): Likewise.
4688 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
4689 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
4690 * tree-pretty-print.c (dump_generic_node): Likewise.
4691 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
4692 (simple_iv_with_niters): Likewise.
4693 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
4694 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
4695 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
4696 * tree-ssa-loop-niter.c (split_to_var_and_offset)
4697 (refine_value_range_using_guard, number_of_iterations_ne_max)
4698 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
4699 (get_cst_init_from_scev, record_nonwrapping_iv)
4700 (scev_var_range_cant_overflow): Likewise.
4701 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
4702 * tree-ssa-pre.c (compute_avail): Likewise.
4703 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
4704 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
4705 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
4706 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
4707 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
4708 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
4709 (dump_case_nodes, try_switch_expansion): Likewise.
4710 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
4711 (vect_do_peeling): Likewise.
4712 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
4713 * tree-vect-stmts.c (vectorizable_load): Likewise.
4714 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
4715 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
4716 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
4717 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
4718 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
4719 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
4720 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
4721 (evrp_dom_walker::before_dom_children): Likewise.
4722 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
4723 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
4724 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
4725 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
4726 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
4727 (get_range_pos_neg): Likewise.
4728 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
4729 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
4730 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
4731 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
4732 * config/avr/avr.c (avr_fold_builtin): Likewise.
4733 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
4734 * config/msp430/msp430.c (msp430_attr): Likewise.
4735 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
4736 * config/powerpcspe/powerpcspe-c.c
4737 (altivec_resolve_overloaded_builtin): Likewise.
4738 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
4739 (rs6000_expand_ternop_builtin): Likewise.
4740 * config/rs6000/rs6000-c.c
4741 (altivec_resolve_overloaded_builtin): Likewise.
4742 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
4743 (rs6000_expand_ternop_builtin): Likewise.
4744 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
4746 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4748 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
4749 when copying loop nest with only one inner loop.
4751 2017-10-10 Richard Biener <rguenther@suse.de>
4753 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
4754 blocks if SCEV is active.
4755 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
4757 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
4758 (scev_initialize): Assert we are not yet initialized.
4760 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4762 * tree-loop-distribution.c (generate_loops_for_partition): Remove
4763 inner loop's exit stmt by making it always exit the loop, otherwise
4764 we would generate an infinite empty loop.
4766 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4768 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
4769 renaming variables in new preheader if it's deleted.
4771 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4773 * tree-loop-distribution.c (struct partition): Remove unused field
4774 loops of the structure.
4775 (partition_alloc, partition_free): Ditto.
4776 (build_rdg_partition_for_vertex): Ditto.
4778 2017-10-09 Jeff Law <law@redhat.com>
4780 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
4781 return type to match prototype and documentation.
4783 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4785 * config/rs6000/rs6000.c (processor_costs): Move to ...
4786 * config/rs6000/rs6000.h: ... here.
4787 (rs6000_cost): Declare.
4789 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
4791 * except.c (setjmp_fn): New global variable.
4792 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
4793 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
4794 if DONT_USE_BUILTIN_SETJMP is defined.
4796 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4798 * target.def (insn_cost): New hook.
4799 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
4800 * doc/tm.texi: Regenerate.
4801 * rtlanal.c (insn_cost): Use the new hook.
4803 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4805 * combine.c (combine_validate_cost): Compute the new insn_cost,
4806 not just pattern_cost.
4807 (try_combine): Adjust comment.
4809 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4811 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
4813 * combine.c (uid_insn_cost): Adjust comment.
4814 (combine_validate_cost): Adjust comment. Use pattern_cost instead
4816 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
4817 * dse.c (find_shift_sequence): Ditto.
4818 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
4819 (bb_valid_for_noce_process_p): Use pattern_cost.
4820 * rtl.h (insn_rtx_cost): Delete.
4821 (pattern_cost): New prototype.
4822 (insn_cost): New prototype.
4823 * rtlanal.c (insn_rtx_cost): Rename to...
4824 (pattern_cost): ... this.
4827 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
4829 * config/i386/i386.md (*jcc_2): Remove insn pattern.
4830 (*jcc<mode>_0_r_i387): Ditto.
4831 (*jccxf_r_i387): Ditto.
4832 (*jcc<mode>_r_i387): Ditto.
4833 (*jccu<mode>_r_i387): Ditto.
4834 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
4835 (*jcc): Rename from *jcc_1.
4837 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4839 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
4840 deferred rescans after the lvx/stvx recombination pre-pass.
4842 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
4844 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
4845 memory operation instruction support.
4846 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
4847 (rs6000-ibm-aix[789]*): Likewise.
4848 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
4849 Document new functions.
4851 2017-10-09 Richard Biener <rguenther@suse.de>
4853 PR tree-optimization/82397
4854 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
4855 equality only for semantically equal trees.
4857 2017-10-09 Richard Biener <rguenther@suse.de>
4859 PR tree-optimization/82449
4860 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
4861 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
4862 allow constant addresses.
4863 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
4866 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4868 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
4871 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4874 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
4877 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4880 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
4882 2017-10-09 Jakub Jelinek <jakub@redhat.com>
4885 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
4886 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
4888 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
4890 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
4891 (WI_BINARY_PREDICATE_RESULT): Likewise.
4892 (wi::binary_traits::operator_result): New type.
4893 (wi::binary_traits::predicate_result): Likewise.
4894 (generic_wide_int::operator~, unary generic_wide_int::operator-)
4895 (generic_wide_int::operator==, generic_wide_int::operator!=)
4896 (generic_wide_int::operator&, generic_wide_int::and_not)
4897 (generic_wide_int::operator|, generic_wide_int::or_not)
4898 (generic_wide_int::operator^, generic_wide_int::operator+
4899 (binary generic_wide_int::operator-, generic_wide_int::operator*):
4901 (operator~, unary operator-, operator==, operator!=, operator&)
4902 (operator|, operator^, operator+, binary operator-, operator*): New
4904 * expr.c (get_inner_reference): Use wi::bit_and_not.
4905 * fold-const.c (fold_binary_loc): Likewise.
4906 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
4907 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
4908 (bit_value_binop): Likewise.
4909 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
4910 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
4911 (extract_range_from_binary_expr_1): Likewise.
4912 (masked_increment): Likewise.
4913 (simplify_bit_ops_using_ranges): Likewise.
4915 2017-10-09 Martin Jambor <mjambor@suse.cz>
4918 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
4919 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
4920 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
4922 (hsa_fixup_mov_insn_type): New function.
4923 (hsa_op_with_type::get_in_type): Use it.
4924 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
4925 immediates in an assert.
4926 (hsa_op_with_type::extend_int_to_32bit): New method.
4927 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
4928 types. Convert to dest type if necessary.
4929 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
4930 (reg_for_gimple_ssa): Pass false as min32int to
4931 hsa_type_for_scalar_tree_type.
4932 (gen_hsa_addr): Fixup type when creating addresable temporary.
4933 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
4934 (gen_hsa_unary_operation): Extend operands and convert to dest type if
4935 necessary. Call hsa_fixup_mov_insn_type.
4936 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
4937 extend operands and convert to dest type if necessary.
4938 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
4939 to dest type if necessary.
4940 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
4941 if conversion nt necessary and size matches.
4942 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
4943 to dest type if necessary.
4944 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
4945 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
4947 (gen_hsa_clrsb): Likewise.
4948 (gen_hsa_ffs): Likewise.
4949 (gen_hsa_divmod): Extend operands and convert to dest type if
4951 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
4953 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
4955 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
4956 Remove empty default arguments. Use a brace block as output
4958 (conditional return): Ditto.
4960 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
4961 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
4963 (group_ending_nop): Ditto.
4964 (doloop_end): Ditto.
4965 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
4966 (splitters for those): Ditto.
4968 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
4970 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
4971 a conditional jump (and the compare for it) so that pc_rtx is the
4973 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
4974 for the deleted and renamed ctr<mode>_internal[234] patterns.
4975 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
4976 Delete second conditional return pattern.
4977 (ctr<mode>_internal2): Delete this second bdnz pattern.
4978 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
4979 (ctr<mode>_internal4): Delete this second bdz pattern.
4981 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
4983 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
4984 (always_initialized_rtx_for_ssa_name_p): New predicate.
4985 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
4986 (finish_out_of_ssa): Free new field of SA.
4987 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
4988 * tree-ssa-coalesce.c: Include tree-ssa.h.
4989 (get_parm_default_def_partitions): Remove extern keyword.
4990 (get_undefined_value_partitions): New function.
4991 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
4992 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
4994 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
4996 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
4998 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
5000 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
5002 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
5003 for targets that preffer 128bit.
5005 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
5007 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
5009 2017-10-08 Olivier Hainque <hainque@adacore.com>
5011 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
5012 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
5014 (thumb_set_return_address): Likewise.
5016 2017-10-08 Olivier Hainque <hainque@adacore.com>
5018 * common/config/arm/arm-common.c (arm_except_unwind_info):
5019 Handle DWARF2_UNWIND_INFO.
5021 2017-10-07 Michael Collison <michael.collison@arm.com>
5023 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
5026 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
5028 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
5029 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
5030 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
5031 defined, force the creation of a new block for a dispatch label.
5033 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
5035 * invoke.texi (Wsuggest-attribute=cold): Document.
5036 * common.opt (Wsuggest-attribute=cold): New
5037 * ipa-pure-const.c (warn_function_cold): New function.
5038 * predict.c (compute_function_frequency): Use it.
5039 * predict.h (warn_function_cold): Declare.
5041 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
5043 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
5046 2017-10-06 Martin Liska <mliska@suse.cz>
5048 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
5049 keyword for member functions.
5050 (struct sanopt_tree_couple): New struct.
5051 (struct sanopt_tree_couple_hash): New function.
5052 (struct sanopt_ctx): Add new hash_map.
5053 (has_dominating_ubsan_ptr_check): New function.
5054 (record_ubsan_ptr_check_stmt): Likewise.
5055 (maybe_optimize_ubsan_ptr_ifn): Likewise.
5056 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
5057 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
5059 2017-10-06 Sudakshina Das <sudi.das@arm.com>
5062 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
5063 aarch64_simd_valid_immediate on CONST_VECTORs.
5064 (aarch64_reg_or_bic_imm): Likewise.
5066 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
5068 PR rtl-optimization/82396
5069 * haifa-sched.c (ready_sort_real): Disable qsort checking.
5071 2017-10-06 Sebastian Pop <sebpop@gmail.com>
5073 * graphite-dependences.c (scop_get_reads): Move code to...
5074 (scop_get_must_writes): Move code to...
5075 (scop_get_may_writes): Move code to...
5076 (scop_get_reads_and_writes): ... here.
5077 (scop_get_dependences): Call scop_get_reads_and_writes.
5079 2017-10-06 Jakub Jelinek <jakub@redhat.com>
5081 PR tree-optimization/82434
5082 * fold-const.h (can_native_encode_type_p,
5083 can_native_encode_string_p): Remove.
5084 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
5085 don't encode anything, just return what would be otherwise returned.
5086 (native_encode_fixed, native_encode_complex, native_encode_vector):
5088 (native_encode_string): Likewise. Inline by hand
5089 can_native_encode_string_p.
5090 (can_native_encode_type_p): Remove.
5091 (can_native_encode_string_p): Remove.
5092 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
5093 STRING_CSTs using can_native_encode_string_p, test all
5094 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
5095 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
5096 argument from native_encode_expr.
5097 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
5098 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
5101 2017-10-06 Richard Biener <rguenther@suse.de>
5103 PR tree-optimization/82397
5104 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
5105 operand_equal_p but rely on data_ref_compare_tree for detecting
5107 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
5108 to match up with dr_group_sort_cmp.
5110 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5113 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
5115 * config/s390/s390-builtin-types.def: Regenerate.
5117 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5120 * config/s390/s390-builtin-types.def: Regenerate.
5121 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
5122 Change flag from B_VXE to B_VX.
5123 (s390_vec_min_dbl): Remove B_VXE flag.
5125 2017-10-06 Richard Biener <rguenther@suse.de>
5127 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
5128 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
5129 translate_isl_ast_to_gimple::is_valid_rename,
5130 translate_isl_ast_to_gimple::get_rename,
5131 translate_isl_ast_to_gimple::get_def_bb_for_const,
5132 translate_isl_ast_to_gimple::get_new_name,
5133 translate_isl_ast_to_gimple::collect_all_ssa_names,
5134 translate_isl_ast_to_gimple::copy_loop_phi_args,
5135 translate_isl_ast_to_gimple::collect_all_ssa_names,
5136 translate_isl_ast_to_gimple::copy_loop_phi_args,
5137 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
5138 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
5139 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
5140 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
5141 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
5142 translate_isl_ast_to_gimple::copy_cond_phi_args,
5143 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
5144 translate_isl_ast_to_gimple::edge_for_new_close_phis,
5145 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
5146 translate_isl_ast_to_gimple::rename_uses,
5147 translate_isl_ast_to_gimple::rename_all_uses): Remove.
5148 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
5149 (set_rename_for_each_def): Likewise.
5150 (graphite_copy_stmts_from_block): Handle debug stmt resetting
5151 here. Handle rewriting SCEV analyzable uses here.
5152 (copy_bb_and_scalar_dependences): Generate code for PHI
5154 (graphite_regenerate_ast_isl): Adjust.
5155 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
5156 (add_write, add_read): New functions.
5157 (build_cross_bb_scalars_def): Use it and simplify.
5158 (build_cross_bb_scalars_use): Likewise.
5159 (graphite_find_cross_bb_scalar_vars): Inline into...
5160 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
5161 simulating out-of-SSA. Compute liveout and add dependencies.
5162 (build_scops): Force an empty entry block.
5163 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
5165 (sese_build_liveouts): Declare.
5166 (sese_trivially_empty_bb_p): Likewise.
5167 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
5168 compute liveout and debug_liveout.
5169 (sese_bad_liveouts_use): Remove.
5170 (sese_reset_debug_liveouts_bb): Likewise.
5171 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
5172 (sese_build_liveouts): Build liveout and debug_liveout and store
5174 (new_sese_info): Adjust.
5175 (free_sese_info): Likewise.
5176 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
5177 do not build liveout here.
5178 (move_sese_in_condition): Adjust region entry.
5179 (scev_analyzable_p): Match up with chrec_apply requirements.
5180 (sese_trivially_empty_bb_p): New.
5181 * tree-into-ssa.c (get_reaching_def): Properly support generating
5182 default-defs for incremental rewrite of anonymous names.
5184 2017-10-06 Richard Biener <rguenther@suse.de>
5186 * graphite-sese-to-poly.c (extract_affine): For casts increasing
5187 precision do not perform modulo reduction.
5189 2017-10-06 Richard Biener <rguenther@suse.de>
5191 PR tree-optimization/82436
5192 * tree-vect-slp.c (vect_supported_load_permutation_p): More
5193 conservatively choose the vectorization factor when checking
5194 whether we can perform the required load permutation.
5195 (vect_transform_slp_perm_load): Assert when we may not fail.
5197 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
5199 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
5200 message for incompatible -msdata=* and -mcall-* options.
5202 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5204 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
5205 rate for post-reload scheduling.
5207 2017-10-05 Tamar Christina <tamar.christina@arm.com>
5209 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
5211 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5213 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
5214 to improve monte carlo in scimark.
5216 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5218 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
5219 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
5220 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
5221 pentium4_cost, nocona_cost): Set reassociation width to 1.
5222 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
5223 width to 2 for fp operations and 1 otherwise.
5224 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
5226 (atom_cost): Set reassociation width to 2.
5227 (slm_cost, generic_cost): Set fp reassociation width
5228 to 2 and 1 otherwise.
5229 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
5230 (core_cost): Set fp reassociation width to 4 and vector to 2.
5231 (ix86_reassociation_width): Rewrite using cost table; special case
5232 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
5233 and TARGET_AVX128_OPTIMAL.
5234 * config/i386/i386.h (processor_costs): Add
5235 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
5236 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
5237 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
5238 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
5239 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
5240 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
5242 2017-10-05 Nathan Sidwell <nathan@acm.org>
5244 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
5246 2017-10-05 Tamar Christina <tamar.christina@arm.com>
5248 * config/arm/arm.c (arm_test_fpu_data): New.
5249 (arm_run_selftests): Call arm_test_fpu_data.
5251 2017-10-04 Nathan Sidwell <nathan@acm.org>
5253 * toplev.c (toplev::main): Remove excess parens on pretty_printer
5255 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
5257 2017-10-04 Sudakshina Das <sudi.das@arm.com>
5259 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
5260 check type for aarch64_simd_valid_immediate.
5261 (aarch64_output_simd_mov_immediate): Update prototype.
5262 (aarch64_simd_valid_immediate): Update prototype.
5263 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
5264 support for ORR-immediate.
5265 (and<mode>3): modified pattern to add support for BIC-immediate.
5266 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
5267 now checks for valid immediate for BIC and ORR based on new enum
5269 (aarch64_output_simd_mov_immediate): Function now used to output
5270 BIC/ORR imm as well based on new enum argument.
5271 * config/aarch64/constraints.md (Do): New vector immediate constraint.
5273 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
5274 (aarch64_reg_or_bic_imm): Likewise.
5276 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5278 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
5279 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
5281 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
5285 PR rtl-optimization/82396
5286 * haifa-sched.c (autopref_multipass_init): Simplify
5288 (autopref_rank_data): Simplify sort order.
5289 * sched-int.h (autopref_multipass_data_): Remove
5290 multi_mem_insn_p, min_offset and max_offset.
5292 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5294 * doc/sourcebuild.texi: Document vect_peeling_profitable.
5296 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5298 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
5301 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5303 * doc/sourcebuild.texi: Document vect_long_mult.
5305 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
5307 PR tree-optimization/82413
5308 * fold-const.c (build_range_check): Use widest_int when comparing
5309 the maximum ETYPE value with HIGH.
5311 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
5313 PR rtl-optimization/82396
5314 * haifa-sched.c (autopref_multipass_init): Simplify
5316 (autopref_rank_data): Simplify sort order.
5317 * sched-int.h (autopref_multipass_data_): Remove
5318 multi_mem_insn_p, min_offset and max_offset.
5320 2017-10-04 Jakub Jelinek <jakub@redhat.com>
5322 PR tree-optimization/82381
5323 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
5324 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
5327 PR tree-optimization/82374
5328 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
5329 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
5330 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
5331 current_function_decl to the new decl.
5333 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
5335 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
5336 helper macro for IEEE float128 hardware built-in functions.
5337 (SQRTF128_ODD): Add built-in functions with the round-to-odd
5339 (TRUNCF128_ODD): Likewise.
5340 (ADDF128_ODD): Likewise.
5341 (SUBF128_ODD): Likewise.
5342 (MULF128_ODD): Likewise.
5343 (DIVF128_ODD): Likewise.
5344 (FMAF128_ODD): Likewise.
5345 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
5346 UNSPEC_TRUNC_ROUND_TO_ODD.
5347 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
5348 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
5349 floating point round to odd instructions.
5350 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
5351 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
5352 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
5353 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
5354 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
5355 (trunc<mode>sf2_hw): Change the truncate with round to odd
5356 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
5357 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
5358 to odd hardware instructions.
5359 (sub<mode>3_odd): Likewise.
5360 (mul<mode>3_odd): Likewise.
5361 (div<mode>3_odd): Likewise.
5362 (sqrt<mode>2_odd): Likewise.
5363 (fma<mode>4_odd): Likewise.
5364 (fms<mode>4_odd): Likewise.
5365 (nfma<mode>4_odd): Likewise.
5366 (nfms<mode>4_odd): Likewise.
5367 (trunc<mode>df2_odd): Change the truncate with round to odd
5368 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
5370 * doc/extend.texi (PowerPC built-in functions): Update documentation
5371 for existing IEEE float128-bit built-in functions. Add built-in
5372 functions that generate the IEEE 128-bit floating point round to
5375 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
5377 PR rtl-optimization/77729
5378 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
5379 to (X&(C1&~C2))|C2 transformations.
5381 2017-10-03 Martin Jambor <mjambor@suse.cz>
5383 PR tree-optimization/82363
5384 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
5385 mismatch, mark lacc written regardless of racc.
5387 2017-10-03 Jakub Jelinek <jakub@redhat.com>
5389 PR tree-optimization/82381
5390 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
5391 stmt_to_insert nor wheather SSA_NAMEs are default defs.
5392 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
5393 fallthrough into reassoc_stmt_dominates_stmt_p.
5396 * combine.c (combine_instructions): Don't combine in unreachable
5399 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
5402 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
5403 function to not use the have_cpu variable. Do not set cpu_index,
5404 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
5406 (rs6000_valid_attribute_p): Remove duplicate initializations of
5407 old_optimize and func_optimize.
5408 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
5409 (rs6000_activate_target_options): Make global.
5410 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
5413 2017-10-02 Jakub Jelinek <jakub@redhat.com>
5415 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
5416 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
5417 Set *poffset to 0 and *psize and *pmax_size to -1 if
5418 *poffset + *psize overflows in HOST_WIDE_INT.
5420 PR tree-optimization/82387
5421 PR tree-optimization/82388
5422 PR tree-optimization/82389
5423 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
5424 instead of live_bytes non-NULL.
5426 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
5429 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
5432 2017-10-02 Richard Biener <rguenther@suse.de>
5434 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
5435 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
5436 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
5438 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
5440 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
5441 requested precision matches the type's.
5442 * calls.c (alloc_max_size): Calculate the new candidate size as
5443 a widest_int and use wi::to_widest when comparing it with the
5444 current candidate size.
5445 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
5446 zero rather than integer_zero_node.
5447 * match.pd: Check for a no-op conversion before using wi::add
5448 rather than after. Use tree_to_uhwi when summing small shift
5449 counts into an unsigned int.
5451 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
5452 Alan Hayward <alan.hayward@arm.com>
5453 David Sherwood <david.sherwood@arm.com>
5456 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
5457 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
5458 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
5459 POINTER_AND_FP_REGS.
5461 2017-10-02 Richard Biener <rguenther@suse.de>
5463 PR tree-optimization/82355
5464 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5465 a mapping for the enclosing loop but avoid generating one for
5467 (copy_bb_and_scalar_dependences): Remove premature codegen
5468 error on PHIs in blocks duplicated into multiple places.
5469 * graphite-scop-detection.c
5470 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5471 in the region use it as loop and nest to analyze the DR in.
5472 (try_generate_gimple_bb): Likewise.
5473 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5474 (add_loop_constraints): For blocks in a loop not in the region
5475 create a dimension with a single iteration.
5476 * sese.h (gbb_loop_at_index): Remove assert.
5478 2017-10-01 Kevin Buettner <kevinb@redhat.com>
5480 * omp-expand.c (adjust_context_scope): New function.
5481 (expand_parallel_call): Call adjust_context_scope.
5483 2017-10-01 Jeff Law <law@redhat.com>
5485 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
5486 dom_opt_dom_walker class with direct access to private members.
5487 Add comments. Call test_for_singularity.
5488 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
5489 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
5490 m_dummy_cond anymore.
5491 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
5493 (pass_dominator:execute): Build the dummy_cond here and pass it
5494 to the dom_opt_dom_walker ctor.
5495 (test_for_singularity): New function.
5497 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
5498 Maya Rashish <coypu@sdf.org>
5500 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
5501 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
5502 (alpha*-*-netbsd*) Use nbsd_tm_file.
5503 (arm*-*-netbsdelf*) Likewise.
5504 (i[34567]86-*-netbsdelf*) Likewise.
5505 (x86_64-*-netbsd*) Likewise.
5506 (mips*-*-netbsd*) Likewise.
5507 (powerpc-*-netbsd*) Likewise.
5508 (sh*-*-netbsd*) Likewise.
5509 (sparc-*-netbsdelf*) Likewise.
5510 (sparc64-*-netbsd*) Likewise.
5511 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
5513 (vax-*-netbsdelf*) Likewise.
5514 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
5515 (UINT_FAST8_TYPE) Likewise.
5516 (INT_FAST16_TYPE) Check CHAR_FAST16.
5517 (UINT_FAST16_TYPE) Likewise.
5519 2017-09-30 Jakub Jelinek <jakub@redhat.com>
5522 * config/i386/i386.md
5523 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
5524 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
5525 *divmodsi4_zext_2): New define_insn_and_split.
5526 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
5527 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
5528 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
5529 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
5530 New define_insn_and_split.
5531 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
5532 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
5533 operands[1] having DImode when mode is SImode.
5535 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
5536 always SImode for DIV and MOD in REG_EQUAL notes.
5538 2017-09-29 Yury Gribov <tetra2005@gmail.com>
5541 * match.pd: Fix handling of NaNs in pattern.
5543 2017-09-29 Jeff Law <law@redhat.com>
5545 * sbitmap.c (bitmap_bit_in_range_p): New function.
5546 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
5547 * tree-ssa-dse.c (live_bytes_read): New function.
5548 (dse_classify_store): Ignore reads of dead bytes.
5550 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
5551 typos and whitespace errors.
5552 * config/i386/predicates.md (address_no_seg_operand): Likewise.
5553 * config/s390/s390.c (s390_emit_prologue): Likewise.
5555 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
5558 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
5559 with a symbol for LRA.
5561 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
5563 PR rtl-optimization/82338
5564 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
5566 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
5568 * genmodes.c (calc_wider_mode): Suppress qsort macro.
5569 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
5570 (qsort_chk): Declare.
5571 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
5572 (qsort_chk): New function.
5574 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5576 PR tree-optimization/82337
5577 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
5578 phi definition if the PHI result appears in an abnormal PHI.
5579 (find_basis_for_base_expr): Don't record a basis if the LHS of the
5580 basis appears in an abnormal PHI.
5582 2017-09-29 Richard Biener <rguenther@suse.de>
5584 * graphite-isl-ast-to-gimple.c
5585 (translate_isl_ast_to_gimple::set_codegen_error): New function.
5586 (binary_op_to_tree): Use it.
5587 (get_rename_from_scev): Likewise.
5588 (copy_loop_phi_nodes): Likewise.
5589 (copy_bb_and_scalar_dependences): Likewise.
5590 (translate_pending_phi_nodes): Likewise.
5592 2017-09-29 Jakub Jelinek <jakub@redhat.com>
5595 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
5596 for movabsq $(i32 << shift), r64.
5598 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
5600 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
5601 index when encoding %esp as %rsp to avoid 0x67 prefix.
5603 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
5605 * config/i386/i386.md (*movsf_internal, *movdf_internal):
5606 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
5608 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
5610 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
5611 Extensions with more than 16 double VFP registers.
5612 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
5613 to_clear_mask and all code related to it. Replace the remaining
5614 entry by a sbitmap and adapt code accordingly.
5616 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
5618 * brig-builtins.def: Change pure attributes to const.
5620 2017-09-28 Joseph Myers <joseph@codesourcery.com>
5622 * config.gcc (default_gnu_indirect_function): Default to yes for
5623 sparc*-*-linux* with glibc.
5625 2017-09-28 Joseph Myers <joseph@codesourcery.com>
5627 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
5628 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
5629 when creating .init_array and .fini_array sections with priority
5632 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
5635 * config/aarch64/aarch64.c
5636 (aarch64_builtin_support_vector_misalignment): Always return false
5637 when misalignment is unknown.
5639 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
5641 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
5642 this function to return false if the definition used by the swap
5643 instruction is artificial, or if the memory address from which the
5644 constant value is loaded is not represented by a base address held
5645 in a register or if the base address register is a frame or stack
5646 pointer. Additionally, return false if the base address of the
5647 loaded constant is a SYMBOL_REF but is not considered to be a
5649 (replace_swapped_load_constant): New function.
5650 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
5651 loaded constant vector with a load of a swapped constant vector.
5653 2017-09-27 Carl Love <cel@us.ibm.com>
5655 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
5656 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
5657 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
5660 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
5662 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
5663 first, always call autopref_rank_data otherwise.
5665 2017-09-27 Richard Biener <rguenther@suse.de>
5667 * graphite-scop-detection.c (find_scop_parameters): Move
5668 loop bound handling ...
5669 (gather_bbs::before_dom_children): ... here, avoiding the need
5670 to build scop_info->loop_nest.
5671 (record_loop_in_sese): Remove.
5672 * sese.h (sese_info_t::loop_nest): Remove.
5673 * sese.c (new_sese_info): Do not allocate loop_nest.
5674 (free_sese_info): Do not free loop_nest.
5676 2017-09-27 Jakub Jelinek <jakub@redhat.com>
5679 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
5680 lhs from calls if the lhs has addressable type.
5682 2017-09-27 Richard Biener <rguenther@suse.de>
5684 * graphite.h (scop::max_alias_set): New member.
5685 * graphite-scop-detection.c: Remove references to non-existing
5686 --param in comments.
5687 (build_alias_sets): Record the maximum alias set used for drs.
5688 (build_scops): Support zero as unlimited for
5689 --param graphite-max-arrays-per-scop.
5690 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
5692 (build_poly_sr_1): ... here. Compute alias set based on the
5693 maximum alias set used for drs rather than
5694 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
5696 2017-09-27 Richard Biener <rguenther@suse.de>
5698 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
5699 --param loop-block-tile-size=0 to disable tiling.
5701 2017-09-27 Richard Biener <rguenther@suse.de>
5703 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
5704 (graphite-max-nb-scop-params): Document special value zero.
5705 * domwalk.h (dom_walker::STOP): New symbolical constant.
5706 (dom_walker::dom_walker): Add optional parameter for bb to
5708 (dom_walker::~dom_walker): Declare.
5709 (dom_walker::before_dom_children): Document STOP return value.
5710 (dom_walker::m_user_bb_to_rpo): New member.
5711 (dom_walker::m_bb_to_rpo): Likewise.
5712 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
5713 mapping here if not provided by the user.
5714 (dom_walker::~dom_walker): Free bb to RPO mapping if not
5715 provided by the user.
5716 (dom_walker::STOP): Define.
5717 (dom_walker::walk): Do not compute bb to RPO mapping here.
5718 Support STOP return value from before_dom_children to stop
5720 * graphite-optimize-isl.c (optimize_isl): If the schedule
5721 is the same still generate code if -fgraphite-identity
5722 or -floop-parallelize-all are given.
5723 * graphite-scop-detection.c: Include cfganal.h.
5724 (gather_bbs::gather_bbs): Get and pass through bb to RPO
5726 (gather_bbs::before_dom_children): Return STOP for BBs
5728 (build_scops): Compute bb to RPO mapping and pass it to
5729 the domwalk. Treat --param graphite-max-nb-scop-params=0
5730 as not limiting the number of params.
5731 * graphite.c (graphite_initialize): Remove limit on the
5732 number of basic-blocks in a function.
5733 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
5734 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
5735 default value of 10.
5737 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5739 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
5740 Adjust code to eliminate needing to do the shift right 32-bits
5741 operation after XSCVDPSPN.
5743 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5745 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
5746 ((X / Y) != 0 -> X >= Y): Likewise.
5748 2017-09-26 Carl Love <cel@us.ibm.com>
5750 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
5751 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
5752 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
5753 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
5754 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
5755 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
5756 definitions and overloading.
5757 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
5758 statement for P9V_BUILTIN_XST_LEN_R.
5759 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
5760 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
5761 define_expand and define_insn for the instructions and builtins.
5762 * doc/extend.texi: Update the built-in documentation file for the new
5764 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
5765 define_insn for the instructions
5767 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
5770 * gcc/config/netbsd-protos.h: New file.
5771 * gcc/config/netbsd.c: New file.
5772 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
5773 * gcc/config/t-netbsd: New file.
5774 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
5775 (tmake_file) Add t-netbsd.
5776 (extra_objs) Add netbsd.o.
5778 2017-09-26 Janus Weil <janus@gcc.gnu.org>
5782 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
5784 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5786 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
5787 sign extension from a vector register to a GPR by doing a 32-bit
5788 direct move and then an EXTSW.
5789 (extendsi<mode>2 splitter): Likewise.
5790 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
5791 right or vector extract after doing XSCVDPSPN. Use
5792 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
5794 (movdi_from_sf_zero_ext): Likewise.
5795 (reload_gpr_from_vsxsf): Likewise.
5796 (p8_mfvsrd_4_disf): Delete, no longer used.
5797 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
5798 then needing to move the SFmode to a GPR to use the XSCVDPSP
5799 instruction instead of FRSP and XSCVDPSPN.
5800 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
5801 it is adjacent to the other XSCVSPDP insns.
5802 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
5803 SFmode to be in traditional Altivec registers.
5804 (vsx_xscvdpspn): Eliminate useless alternative constraint.
5805 (vsx_xscvspdpn): Likewise.
5806 (vsx_xscvspdpn_scalar): Likewise.
5808 2017-09-26 Martin Jambor <mjambor@suse.cz>
5810 * tree-sra.c (compare_access_positions): Put integral types first,
5811 stabilize sorting of integral types, remove conditions putting
5812 non-full-precision integers last.
5813 (sort_and_splice_var_accesses): Disable scalarization if a
5814 non-integert would be represented by a non-full-precision integer.
5816 2017-09-26 Joseph Myers <joseph@codesourcery.com>
5818 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
5819 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
5820 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
5821 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
5822 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
5823 conditionals inside the function instead of around it. Call
5824 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
5825 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
5827 2017-09-26 Richard Biener <rguenther@suse.de>
5829 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
5831 (scop_detection::build_scop_breadth): ... this. Removed.
5832 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
5833 (scop_detection::harmful_stmt_in_bb): Likewise.
5834 (scop_detection::graphite_can_represent_stmt): Likewise.
5835 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
5836 (scop_detection::can_represent_loop): Remove recursion, fold in ...
5837 (scop_detection::can_represent_loop_1): ... this. Removed.
5838 (scop_detection::harmful_loop_in_region): Simplify after inlining
5839 the above and remove more quadraticness.
5840 (build_scops): Adjust.
5841 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
5844 2017-09-26 Jakub Jelinek <jakub@redhat.com>
5847 * config/i386/i386.c (ix86_print_operand_address_as): Only test
5848 REGNO (base) == SP_REG if base is a REG.
5851 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
5852 if it is different SSA_NAME.
5853 (optimize_range_tests_cmp_bitwise): New function.
5854 (optimize_range_tests): Call it.
5856 2017-09-26 Richard Biener <rguenther@suse.de>
5858 PR tree-optimization/82321
5859 * graphite.c (canonicalize_loop_closed_ssa): Properly check
5860 for the def being inside the loop.
5862 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5864 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
5866 * config/s390/s390-builtins.def: Fix constraint on op4.
5868 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5870 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
5871 independent expanders.
5872 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
5873 ("vec_ordered", "vec_unordered"): New expanders.
5875 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5877 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
5880 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5882 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
5883 vec_unpacks_lo_v16qi.
5884 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
5886 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5888 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
5889 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
5890 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
5892 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5894 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
5896 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
5898 ("*vec_slb<mode>"): New insn pattern.
5899 ("vec_shr_<mode>"): New expander.
5900 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
5901 and force the shift count operand to V16QImode.
5902 ("vec_srb<mode>"): Set shift count mode to V16QI.
5904 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5906 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
5907 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
5908 ("vec_widen_smult_hi_<mode>"): New expander definitions.
5910 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
5913 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
5915 2017-09-26 Richard Biener <rguenther@suse.de>
5917 PR tree-optimization/82320
5918 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
5921 2017-09-25 Jeff Law <law@redhat.com>
5923 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
5924 prototype for new argument.
5925 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
5926 mostly extracted from rs6000_emit_allocate_stack.
5927 (rs6000_emit_probe_stack_range_stack_clash): New function.
5928 (rs6000_emit_allocate_stack): Call
5929 rs6000_emit_probe_stack_range_stack_clash as needed.
5930 (rs6000_emit_probe_stack_range): Add additional argument
5931 to call to gen_probe_stack_range{si,di}.
5932 (output_probe_stack_range): New.
5933 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
5934 (output_probe_stack_range_stack_clash): New.
5935 (rs6000_emit_prologue): Emit notes into dump file as requested.
5936 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
5937 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
5938 Add additional operand and pass it to output_probe_stack_range.
5940 2017-09-25 Bin Cheng <bin.cheng@arm.com>
5942 PR tree-optimization/82163
5943 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
5944 (checking_verify_loop_closed_ssa): New parameter.
5945 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
5946 (check_loop_closed_ssa_stmt): Delete.
5947 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
5948 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
5949 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
5952 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
5954 * brig-builtins.def: Treat HSAIL barrier builtins as
5955 setjmp/longjump style functions.
5957 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
5959 * target.def (constant_alignment): New hook.
5960 * defaults.h (CONSTANT_ALIGNMENT): Delete.
5961 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
5962 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
5963 * doc/tm.texi: Regenerate.
5964 * targhooks.h (default_constant_alignment): Declare.
5965 (constant_alignment_word_strings): Likewise.
5966 * targhooks.c (default_constant_alignment): New function.
5967 (constant_alignment_word_strings): Likewise.
5968 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
5969 instead of CONSTANT_ALIGNMENT.
5970 * varasm.c (align_variable, get_variable_align, build_constant_desc)
5971 (force_const_mem): Likewise.
5972 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
5973 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
5974 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
5975 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5976 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
5978 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
5979 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5980 constant_alignment_word_strings.
5981 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
5982 (CONSTANT_ALIGNMENT): Likewise.
5983 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5984 (arm_constant_alignment): New function.
5985 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
5986 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5987 constant_alignment_word_strings.
5988 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
5989 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5990 constant_alignment_word_strings.
5991 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
5992 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5993 (cris_constant_alignment): New function.
5994 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
5995 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5996 (epiphany_constant_alignment): New function.
5997 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
5998 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5999 constant_alignment_word_strings.
6000 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
6001 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6002 constant_alignment_word_strings.
6003 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
6004 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6005 constant_alignment_word_strings.
6006 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
6007 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
6008 * config/i386/i386.c (ix86_constant_alignment): Make static.
6009 Use the same interface as the target hook.
6010 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6011 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
6012 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6013 constant_alignment_word_strings.
6014 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
6015 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
6016 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6017 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
6018 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6019 constant_alignment_word_strings.
6020 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
6021 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6022 constant_alignment_word_strings.
6023 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
6024 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6025 constant_alignment_word_strings.
6026 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
6027 * config/microblaze/microblaze.c (microblaze_constant_alignment):
6029 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6030 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
6031 * config/mips/mips.c (mips_constant_alignment): New function.
6032 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6033 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
6034 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
6035 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6036 (mmix_constant_alignment): Make static. Use the same interface
6038 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
6039 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6040 constant_alignment_word_strings.
6041 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
6042 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6043 constant_alignment_word_strings.
6044 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
6045 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6046 constant_alignment_word_strings.
6047 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
6048 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6049 (rs6000_constant_alignment): New function.
6050 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
6051 * config/riscv/riscv.c (riscv_constant_alignment): New function.
6052 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6053 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
6054 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6055 (rs6000_constant_alignment): New function.
6056 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
6057 * config/s390/s390.c (s390_constant_alignment): New function.
6058 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6059 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
6060 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6061 constant_alignment_word_strings.
6062 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
6063 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6064 (sparc_constant_alignment): New function.
6065 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
6066 * config/spu/spu.c (spu_constant_alignment): New function.
6067 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6068 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
6069 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6070 constant_alignment_word_strings.
6071 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
6072 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6073 constant_alignment_word_strings.
6074 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
6075 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6076 constant_alignment_word_strings.
6077 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
6078 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6079 (visium_constant_alignment): New function.
6080 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
6081 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6082 (xtensa_constant_alignment): New function.
6083 * system.h (CONSTANT_ALIGNMENT): Poison.
6085 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
6087 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6088 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
6089 (rs6000_builtin_valid_without_lhs): New helper function.
6090 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6091 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
6093 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
6095 * target.h (vec_perm_indices): Use unsigned short rather than
6097 (auto_vec_perm_indices): Likewise.
6098 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
6099 Use unsigned int rather than unsigned char.
6100 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
6102 2017-09-25 Richard Biener <rguenther@suse.de>
6104 * cfgloop.h (sort_sibling_loops): Declare.
6105 * cfgloop.c (sort_sibling_loops_cmp): New helper.
6106 (sort_sibling_loops): New function sorting the sibling loop list
6108 * graphite.c (graphite_transform_loops): Sort sibling loops.
6110 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
6112 * target.def (vec_perm_const_ok): Change sel parameter to
6114 * optabs-query.c (can_vec_perm_p): Update accordingly.
6115 * doc/tm.texi: Regenerate.
6116 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
6117 auto_vec_perm_indices and remove separate nelt field.
6118 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
6119 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
6120 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
6121 (aarch64_expand_vec_perm_const): Update accordingly.
6122 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
6123 to vec_perm_indices.
6124 * config/arm/arm.c (expand_vec_perm_d): Change perm to
6125 auto_vec_perm_indices and remove separate nelt field.
6126 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
6127 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
6128 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
6130 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
6131 to vec_perm_indices.
6132 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
6133 sel to vec_perm_indices.
6134 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
6135 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
6136 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
6138 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
6141 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
6144 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
6145 on the FUNCTION_DECL function context if it has a DIE that is a
6148 2017-09-25 Richard Biener <rguenther@suse.de>
6150 PR tree-optimization/82285
6151 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
6154 2017-09-25 Tom de Vries <tom@codesourcery.com>
6158 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
6161 2017-09-25 Richard Biener <rguenther@suse.de>
6163 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
6164 ISL errors other than isl_error_quota happen. Dump if the
6165 schedule is the same.
6166 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
6167 errors instead of aborting inside ISL.
6169 2017-09-25 Iain Sandoe <iain@codesourcery.com>
6172 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
6173 of libgcc_eh for m64.
6174 * config/i386/darwin64.h: Likewise.
6176 2017-09-25 Richard Biener <rguenther@suse.de>
6179 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
6180 attribute for incomplete types nor twice for complete ones.
6182 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
6185 * config/i386/i386.c (ix86_print_operand_address_as): Encode
6186 %esp as %rsp to avoid 0x67 prefix if there is no index or base
6189 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
6192 * config/i386/i386.opt (mprefer-avx256): Use
6193 ix86_target_flags variable.
6194 * config/i386/i386.c (ix86_target_string): Move
6195 -mprefer-avx256 to flag2_opts.
6197 2017-09-22 Jakub Jelinek <jakub@redhat.com>
6200 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
6201 and x != -1 | y != -1 into (x & y) != -1.
6203 2017-09-22 Steve Ellcey <sellcey@cavium.com>
6205 * config.gcc: Add new case statement to set
6206 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
6207 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
6208 s390x-*-linux* case statements. Added aarch64 to the list of
6209 supported architectures.
6211 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6213 PR tree-optimization/82289
6214 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
6215 STMT_VINFO_RELEVANT_P.
6217 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6218 Alan Hayward <alan.hayward@arm.com>
6219 David Sherwood <david.sherwood@arm.com>
6221 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
6222 for VR_RANGE only; don't allow VR_ANTI_RANGE.
6223 (extract_range_from_binary_expr_1): Don't call
6224 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
6226 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6227 Alan Hayward <alan.hayward@arm.com>
6228 David Sherwood <david.sherwood@arm.com>
6230 * target.def (preferred_vector_alignment): New hook.
6231 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
6233 * doc/tm.texi: Regenerate.
6234 * targhooks.h (default_preferred_vector_alignment): Declare.
6235 * targhooks.c (default_preferred_vector_alignment): New function.
6236 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
6238 (DR_TARGET_ALIGNMENT): New macro.
6239 (aligned_access_p): Update commentary.
6240 (vect_known_alignment_in_bytes): New function.
6241 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
6243 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
6244 Calculate the misalignment based on the target alignment rather than
6246 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
6247 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
6248 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
6249 the target alignment, rather than masking the element misalignment
6250 with the number of elements in a vector. Also use the target
6251 alignment when calculating the maximum number of peels.
6252 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
6253 instead of TYPE_ALIGN_UNIT.
6254 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
6255 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
6256 (vect_create_addr_base_for_vector_ref): Update call accordingly.
6257 (vect_create_data_ref_ptr): Likewise.
6258 (vect_setup_realignment): Realign by ANDing with
6259 -DR_TARGET_MISALIGNMENT.
6260 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
6261 the number of peels based on DR_TARGET_ALIGNMENT.
6262 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
6263 with the guaranteed alignment boundary when deciding whether
6265 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
6266 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
6267 (ensure_base_align): Remove stmt_info parameter. Get the
6268 target base alignment from DR_TARGET_ALIGNMENT.
6269 (vectorizable_store): Update call accordingly. Interpret
6270 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
6272 (vectorizable_load): Likewise.
6274 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6275 Alan Hayward <alan.hayward@arm.com>
6276 David Sherwood <david.sherwood@arm.com>
6278 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
6279 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
6280 (vect_enhance_data_refs_alignment): Likewise.
6282 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
6284 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
6285 error. Only quit immediately if parsing is complete.
6286 (BEGIN): Initialize fatal_err and parse_done.
6287 (begin fpu, end fpu): Check number of arguments.
6288 (begin arch, end arch): Likewise.
6289 (begin cpu, end cpu): Likewise.
6290 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
6291 (optalias): Likewise.
6293 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
6295 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
6296 * config/arm/arm-isa.h: Delete. Move definitions to ...
6297 * arm-cpus.in: ... here. Use new feature and fgroup values.
6298 * config/arm/arm.c (arm_option_override): Use lower case for feature
6300 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
6301 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
6302 * config/arm/parsecpu.awk (END): Add new command 'isa'.
6304 (print_isa_bits_for): New function.
6305 (gen_isa): New function.
6306 (gen_comm_data): Use print_isa_bits_for.
6307 (define feature): New keyword.
6308 (define fgroup): New keyword.
6309 * config/arm/t-arm (TM_H): Remove.
6310 (GTM_H): Add arm-isa.h.
6311 (arm-isa.h): Add rule to generate file.
6312 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
6313 case for feature bit names.
6315 2017-09-22 Richard Biener <rguenther@suse.de>
6317 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
6319 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
6320 print of no dependency loops ...
6321 * graphite.c (graphite_transform_loops): ... here.
6322 (canonicalize_loop_closed_ssa_form): Work from inner to outer
6324 (same_close_phi_node, remove_duplicate_close_phi,
6325 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
6326 (canonicalize_loop_closed_ssa): ... here and simplify.
6327 * graphite-optimize-isl.c: Include tree-vectorizer.h.
6328 (optimize_isl): Use dump_printf_loc to tell when we stopped
6329 optimizing because of an ISL timeout.
6331 2017-09-22 Richard Biener <rguenther@suse.de>
6333 PR tree-optimization/82291
6334 * tree-if-conv.c (predicate_mem_writes): Make sure to
6335 remove writes in blocks predicated with false.
6337 2017-09-22 Richard Biener <rguenther@suse.de>
6339 * sese.c: Include cfganal.h.
6340 (if_region_set_false_region): Remove.
6341 (create_if_region_on_edge): Likewise.
6342 (move_sese_in_condition): Re-implement without destroying
6345 2017-09-22 Richard Biener <rguenther@suse.de>
6347 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
6348 Verify both BBs contain loop PHI nodes before dispatching to
6350 (graphite_regenerate_ast_isl): Do not recompute dominators,
6351 do not verify three times. Restructure for clarity.
6352 * graphite-scop-detection.c (same_close_phi_node,
6353 remove_duplicate_close_phi, make_close_phi_nodes_unique,
6354 defined_in_loop_p, canonicalize_loop_closed_ssa,
6355 canonicalize_loop_closed_ssa_form): Simplify, remove excess
6356 checking and SSA rewrite, move to ...
6357 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
6358 (graphite_initialize): Do not pass in ctx, do not reset the
6359 SCEV cache, compute only dominators.
6360 (graphite_transform_loops): Allocate ISL ctx after
6361 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
6362 Maintain post-dominators only around build_scops.
6363 * sese.c (if_region_set_false_region): Make static. Free
6364 and recompute dominators.
6365 (move_sese_in_condition): Assert we don't get called with
6366 post-dominators computed.
6367 * sese.h (if_region_set_false_region): Remove.
6369 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
6371 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
6372 mode attribute for TARGET_AVX512VL.
6374 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
6376 * config/i386/i386.opt (mprefer-avx256): New option.
6377 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
6379 (ix86_preferred_simd_mode): Return 256-bit AVX modes
6380 for TARGET_PREFER_AVX256.
6381 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
6383 2017-09-21 Jeff Law <law@redhat.com>
6385 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
6386 Fix dump output if the only stack space is for pushed registers.
6388 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6390 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
6393 2017-09-21 Martin Sebor <msebor@redhat.com>
6396 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
6397 code (in C++) or code that triggers warnings.
6399 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
6401 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
6403 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
6405 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
6406 * sched-rgn.c: Ditto.
6407 * sel-sched-ir.c: Ditto.
6409 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
6411 * toplev.h (set_random_seed): Adjust return type.
6412 * toplev.c (init_local_tick): Move eager initialization of random_seed
6413 to get_random_seed. Adjust comment.
6414 (init_random_seed): Inline to get_random_seed, delete.
6415 (get_random_seed): Initialize random_seed lazily.
6416 (set_random_seed): Do not return previous value.
6417 (print_switch_value): Do not call get_random_seed.
6419 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
6421 * cgraph.c (delete_function_version): New, broken out from...
6422 (cgraph_node::delete_function_version): ...here. Rename to
6423 cgraph_node::delete_function_version_by_decl. Update all uses.
6424 (cgraph_node::remove): Call delete_function_version.
6426 2017-09-21 Jakub Jelinek <jakub@redhat.com>
6429 * tree-inline.c (expand_call_inline): Emit clobber stmts for
6430 VAR_DECLs to which addressable non-volatile parameters are mapped
6431 and for id->retvar after the return value assignment. Clear
6432 id->retval and id->retbnd after inlining.
6434 2017-09-21 Richard Biener <rguenther@suse.de>
6436 PR tree-optimization/82276
6437 PR tree-optimization/82244
6438 * tree-vrp.c (build_assert_expr_for): Set
6439 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
6441 (remove_range_assertions): Revert earlier change.
6443 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
6446 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
6448 2017-09-21 Richard Biener <rguenther@suse.de>
6450 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
6451 Restore valid IL after code generation errors.
6452 * graphite.c (graphite_transform_loops): Diagnose code
6453 generation issues as MSG_MISSED_OPTIMIZATION and continue
6454 with processing SCOPs.
6456 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6457 Alan Hayward <alan.hayward@arm.com>
6458 David Sherwood <david.sherwood@arm.com>
6460 * calls.c (compute_argument_addresses): Use simplify_gen_binary
6461 rather than choosing between plus_constant and gen_rtx_<CODE>.
6462 * expr.c (emit_push_insn): Likewise.
6463 (expand_expr_real_2): Likewise.
6465 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6466 Alan Hayward <alan.hayward@arm.com>
6467 David Sherwood <david.sherwood@arm.com>
6469 * loop-unroll.c (split_iv): Call copy_rtx on the step.
6471 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6472 Alan Hayward <alan.hayward@arm.com>
6473 David Sherwood <david.sherwood@arm.com>
6475 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
6476 calling tree_to_uhwi.
6478 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6479 Alan Hayward <alan.hayward@arm.com>
6480 David Sherwood <david.sherwood@arm.com>
6482 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
6483 INTEGER_CST rather than a negative test for ADDR_EXPR.
6485 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6486 Alan Hayward <alan.hayward@arm.com>
6487 David Sherwood <david.sherwood@arm.com>
6489 * tree-vrp.c (extract_range_from_binary_expr_1): Check
6490 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
6492 2017-09-21 Richard Biener <rguenther@suse.de>
6494 PR tree-optimization/71351
6495 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
6496 graphite_create_new_loop_guard): Remove, fold remaining parts
6498 (translate_isl_ast_node_for): ... here and simplify.
6500 2017-09-21 Jakub Jelinek <jakub@redhat.com>
6503 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
6504 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
6505 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
6506 alternative always use QI mode, for -Os imov (=R,R) alternative
6507 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
6510 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6511 Jeff Law <law@redhat.com>
6513 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
6514 (allocate_stack_space): New function, partially extracted from
6516 (s390_emit_prologue): Track offset to most recent stack probe.
6517 Code to allocate space moved into allocate_stack_space.
6518 Dump actions when no stack is allocated.
6519 (s390_prologue_plus_offset): New function.
6520 (s390_emit_stack_probe): Likewise.
6522 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
6524 * common.opt (Wa, Wl, Wp, g, gz=): Add
6526 (gno-column-info): Remove.
6527 (gcolumn-info): Drop RejectNegative.
6529 (gno-record-gcc-switches): Remove.
6530 (grecord-gcc-switches): Drop RejectNegative.
6531 (gno-split-dwarf): Remove.
6532 (gsplit-dwarf): Drop RejectNegative.
6533 (gno-strict-dwarf): Remove.
6534 (gstrict-dwarf): Drop RejectNegative.
6535 * config/darwin.opt (gfull, gused): Add RejectNegative.
6536 * dwarf2out.c (gen_producer_string): Drop
6537 gno-record-gcc-switches handler.
6538 * optc-gen.awk: Add g to prefixes with negative forms.
6539 * opts-common.c (remapping_prefix_p): New.
6540 (find_opt): Check it.
6541 (generate_canonical_option): Test g prefix.
6542 (option_map): Add -gno- mapping.
6543 (add_misspelling_candidates): Check remapping_prefix_p.
6545 2017-09-20 Jeff Law <law@redhat.com>
6547 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
6548 thinko in stack clash protection support.
6550 * explow.c (compute_stack_clash_protection_loop_data): Use
6551 CONST_INT_P instead of explicit test. Verify object is a
6552 CONST_INT_P before looking at INTVAL.
6553 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
6554 instead of explicit test.
6556 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
6559 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
6560 address instead of to r1 and r11.
6562 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
6564 * config.gcc: Support "knm".
6565 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
6566 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
6568 * config/i386/i386.c (m_KNM): Define.
6569 (processor_target_table): Add "knm".
6571 (ix86_option_override_internal): Add "knm".
6572 (ix86_issue_rate): Add PROCESSOR_KNM.
6573 (ix86_adjust_cost): Ditto.
6574 (ia32_multipass_dfa_lookahead): Ditto.
6575 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
6576 (fold_builtin_cpu): Add M_INTEL_KNM.
6577 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
6578 (processor_type): Add PROCESSOR_KNM.
6579 * config/i386/x86-tune.def: Add m_KNM.
6580 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
6582 2017-09-20 Richard Biener <rguenther@suse.de>
6584 PR tree-optimization/80213
6585 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
6586 are allowed in empty BBs as well.
6587 (canonicalize_loop_closed_ssa): Also look for other complex
6589 (scop_detection::get_sese): Include the loop-closed PHI block
6591 (scop_detection::merge_sese): Remove code adding extra blocks.
6592 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
6593 (build_scops): Assert the final returned scop is invalid.
6595 2017-09-20 Richard Biener <rguenther@suse.de>
6597 PR tree-optimization/82264
6598 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
6600 (vn_phi_lookup): Likewise.
6601 (vn_phi_insert): Likewise.
6603 2017-09-20 Jakub Jelinek <jakub@redhat.com>
6605 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
6606 that fits into uhwi or shwi, add DW_AT_const_value regardless
6607 of early_dwarf without going through RTL, using add_AT_unsigned
6610 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
6611 (DEBUG_LTO_ABBREV_SECTION): Likewise.
6612 (DEBUG_LTO_MACINFO_SECTION): Likewise.
6613 (DEBUG_MACRO_SECTION): Likewise.
6614 (DEBUG_LTO_MACRO_SECTION): Likewise.
6615 (DEBUG_STR_DWO_SECTION): Likewise.
6616 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
6617 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
6618 (DEBUG_LTO_DWO_LINE_SECTION): Define.
6619 (DEBUG_LTO_LINE_STR_SECTION): Define.
6620 (init_sections_and_labels): Initialize debug_line_str_section
6621 variable. Initialize debug_loc_section for -gdwarf-5 to
6622 DEBUG_LOCLISTS_SECTION. Formatting fixes.
6624 2017-09-20 Richard Biener <rguenther@suse.de>
6626 * graphite-sese-to-poly.c (extract_affine): Properly handle
6627 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
6629 2017-09-20 Richard Biener <rguenther@suse.de>
6631 PR tree-optimization/81373
6632 * graphite-scop-detection.c (build_cross_bb_scalars_def):
6633 Force SESE live-out defs to be handled even if they are
6636 2017-09-19 Jeff Law <law@redhat.com>
6638 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
6639 nothing for stack adjustments with REG_STACK_CHECK.
6640 * sched-deps.c (parse_add_or_inc): Reject insns with
6641 REG_STACK_CHECK from dependency breaking.
6642 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
6643 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
6644 * reg-notes.def (STACK_CHECK): New note.
6646 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
6647 (ix86_expand_prologue): Dump stack clash info as needed.
6648 Call ix86_adjust_stack_and_probe_stack_clash as needed.
6650 * function.c (dump_stack_clash_frame_info): New function.
6651 * function.h (dump_stack_clash_frame_info): Prototype.
6652 (enum stack_clash_probes): New enum.
6654 * config/alpha/alpha.c (alpha_expand_prologue): Also check
6655 flag_stack_clash_protection.
6656 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
6657 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
6658 (arm_frame_pointer_required): Likewise.
6659 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
6660 (ia64_expand_prologue): Likewise.
6661 * config/mips/mips.c (mips_expand_prologue): Likewise.
6662 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
6663 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
6664 (sparc_flat_expand_prologue): Likewise.
6665 * config/spu/spu.c (spu_expand_prologue): Likewise.
6667 * explow.c: Include "params.h".
6668 (anti_adjust_stack_and_probe_stack_clash): New function.
6669 (get_stack_check_protect): Likewise.
6670 (compute_stack_clash_protection_loop_data): Likewise.
6671 (emit_stack_clash_protection_loop_start): Likewise.
6672 (emit_stack_clash_protection_loop_end): Likewise.
6673 (allocate_dynamic_stack_space): Use get_stack_check_protect.
6674 Use anti_adjust_stack_and_probe_stack_clash.
6675 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
6676 (emit_stack_clash_protection_loop_start): Likewise.
6677 (emit_stack_clash_protection_loop_end): Likewise.
6678 * rtl.h (get_stack_check_protect): Prototype.
6679 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
6680 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
6681 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
6683 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
6685 * doc/tm.texi: Rebuilt.
6686 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
6687 get_stack_check_protect.
6688 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
6689 * config/arm/arm.c (arm_expand_prologue): Likewise.
6690 (arm_frame_pointer_required): Likewise.
6691 * config/i386/i386.c (ix86_expand_prologue): Likewise.
6692 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
6693 * config/mips/mips.c (mips_expand_prologue): Likewise.
6694 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
6695 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
6696 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
6697 (sparc_flat_expand_prologue): Likewise.
6699 * common.opt (-fstack-clash-protection): New option.
6700 * flag-types.h (enum stack_check_type): Note difference between
6701 -fstack-check= and -fstack-clash-protection.
6702 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
6703 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
6704 * toplev.c (process_options): Issue warnings/errors for cases
6705 not handled with -fstack-clash-protection.
6706 * doc/invoke.texi (-fstack-clash-protection): Document new option.
6707 (-fstack-check): Note additional problem with -fstack-check=generic.
6708 Note that -fstack-check is primarily for Ada and refer users
6709 to -fstack-clash-protection for stack-clash-protection.
6710 Document new params for stack clash protection.
6712 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
6714 * config/i386/i386.c (ix86_split_long_move): Do not handle
6715 address used for LEA in a special way.
6717 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
6719 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
6721 2017-09-19 Martin Sebor <msebor@redhat.com>
6724 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
6725 of incompatible types.
6727 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
6729 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6730 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
6731 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6732 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
6734 2017-09-19 Richard Biener <rguenther@suse.de>
6736 PR tree-optimization/82244
6737 * tree-vrp.c (remove_range_assertions): Do not propagate
6738 a constant to abnormals but replace the assert with a copy.
6740 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
6742 PR rtl-optimization/57878
6743 PR rtl-optimization/68988
6744 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
6745 avoidance test involving non_reload_pseudos. Move frequency test
6746 below the general fragmentation avoidance test.
6748 2017-09-19 Richard Biener <rguenther@suse.de>
6750 PR tree-optimization/69728
6751 * graphite-sese-to-poly.c (schedule_error): New global.
6752 (add_loop_schedule): Handle empty domain by failing the
6754 (build_original_schedule): Handle schedule_error.
6756 2017-09-19 Richard Biener <rguenther@suse.de>
6758 * graphite-scop-detection.c (scop_detection::can_represent_loop):
6759 Do not iterate to sibling loops but only to siblings of inner
6762 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
6765 * config/m68k/m68k.md (moveq feeding equality comparison): Check
6766 that the registers are different.
6768 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
6770 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
6771 to processor_model and "amdfam17h" to arch_names_table.
6772 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
6774 2017-09-18 Jakub Jelinek <jakub@redhat.com>
6777 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
6779 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6780 Alan Hayward <alan.hayward@arm.com>
6781 David Sherwood <david.sherwood@arm.com>
6783 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
6785 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
6787 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
6788 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
6789 vect_schedule_slp_instance.
6790 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
6791 Update call to vect_slp_analyze_node_operations. Simplify return
6793 (vect_slp_analyze_bb_1): Update call accordingly.
6794 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
6795 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
6796 (vect_schedule_slp): Update call accordingly.
6798 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6799 Alan Hayward <alan.hayward@arm.com>
6800 David Sherwood <david.sherwood@arm.com>
6802 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
6803 with types that aren't in fact scalar.
6805 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6807 * tree-vect-slp.c (vect_record_max_nunits): New function,
6809 (vect_build_slp_tree_1): ...here.
6810 (vect_build_slp_tree_2): Call it for phis too.
6812 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6814 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
6815 to vect_get_vec_def_for_operand when getting the mask operand.
6817 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6818 Alan Hayward <alan.hayward@arm.com>
6819 David Sherwood <david.sherwood@arm.com>
6821 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
6824 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6825 Alan Hayward <alan.hayward@arm.com>
6826 David Sherwood <david.sherwood@arm.com>
6828 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
6829 calculation for vector booleans.
6831 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6832 Alan Hayward <alan.hayward@arm.com>
6833 David Sherwood <david.sherwood@arm.com>
6835 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
6837 (vect_transform_stmt): ...here.
6838 (vect_analyze_stmt): Use it instead of calling
6839 vectorizable_live_operation directly.
6841 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
6843 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
6844 non-SIMT targets in acc vector loops.
6846 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
6848 * configure.ac: Add arc and check if assembler supports gdwarf2.
6849 * configure: Regenerate.
6851 2017-09-18 Richard Biener <rguenther@suse.de>
6853 PR tree-optimization/82220
6854 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
6855 epilogue niters from the min_profitable_iters compute.
6857 2017-09-18 Jakub Jelinek <jakub@redhat.com>
6860 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
6861 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
6862 (ix86_init_pic_reg): Revert 2017-09-01 changes.
6864 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
6867 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
6868 switching to a new text section.
6870 2017-09-18 Richard Biener <rguenther@suse.de>
6872 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
6874 (build_alias_set): Reject aliases with no access function.
6876 2017-09-18 Richard Biener <rguenther@suse.de>
6878 PR tree-optimization/79622
6879 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
6881 (build_cross_bb_scalars_use): Likewise.
6883 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
6885 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
6887 2017-09-18 Alan Modra <amodra@gmail.com>
6890 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
6891 stack_pointer_rtx for count 0. Update comments. Break up
6892 large rtl expression.
6894 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
6896 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
6897 Increase to 20 bytes.
6898 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
6899 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
6900 or avx version of the stub.
6902 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
6905 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
6906 compute the minimum stack alignment. Also update preferred stack
6907 boundary for leaf functions.
6909 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
6911 PR tree-optimization/82228
6912 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
6915 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
6917 * common/config/nds32/nds32-common.c
6918 (nds32_option_optimization_table): Refine formatting.
6919 (nds32_option_optimization_table): Use -fsched-pressure and
6920 -fomit-frame-pointer for specific optimization level.
6922 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
6924 * config/nds32/nds32.c: Refine formatting and comments.
6925 * config/nds32/nds32.h: Likewise.
6926 * config/nds32/nds32.md: Likewise.
6927 * config/nds32/nds32-cost.c: Likewise.
6928 * config/nds32/nds32-isr.c: Likewise.
6929 * config/nds32/nds32-md-auxiliary.c: Likewise.
6930 * config/nds32/nds32-multiple.md: Likewise.
6931 * config/nds32/nds32-predicates.c: Likewise.
6933 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
6934 Jakub Jelinek <jakub@redhat.com>
6936 Add support for -std=c++2a.
6937 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
6939 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
6941 2017-09-15 Steve Ellcey <sellcey@cavium.com>
6944 * doc/extend.texi (Common Function Attributes): Add
6945 references to ARM, AArch64, and S/390 specific attributes.
6946 (Function Specific Option Pragmas): Add AArch64 and S/390
6947 to list of back ends that support the target pragma.
6949 2017-09-15 Nathan Sidwell <nathan@acm.org>
6951 * doc/standards.texi: Fix C++17 description. Update URLs for
6954 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
6956 * common.opt (Wcast-align=strict): New warning option.
6957 * doc/invoke.texi: Document -Wcast-align=strict.
6959 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
6961 * cgraph.h (cgraph_thunk_info): Add comments.
6962 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
6963 assert for VIRTUAL_* arguments stricter.
6965 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
6967 PR tree-optimization/71026
6968 * match.pd: Move RDIV patterns from fold-const.c
6969 * fold-const.c (distribute_real_division): Removed.
6970 (fold_binary_loc): Remove calls to distribute_real_divison.
6972 2017-09-15 Jakub Jelinek <jakub@redhat.com>
6974 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
6975 c++1z and gnu++1z as deprecated. Change other references to
6976 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
6977 Change -Wc++1z-compat to -Wc++17-compat.
6978 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
6979 * dwarf2out.c (highest_c_language): Handle C++17.
6980 (gen_compile_unit_die): Likewise.
6982 2017-09-15 Jakub Jelinek <jakub@redhat.com>
6984 PR rtl-optimization/82192
6985 * combine.c (make_extraction): Don't look through non-paradoxical
6986 SUBREGs or TRUNCATE if pos + len is or might be bigger than
6989 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
6990 Alan Hayward <alan.hayward@arm.com>
6991 David Sherwood <david.sherwood@arm.com>
6993 * target.def (function_arg_offset): New hook.
6994 * targhooks.h (default_function_arg_offset): Declare.
6995 * targhooks.c (default_function_arg_offset): New function.
6996 * function.c (locate_and_pad_parm): Use
6997 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
6998 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
6999 (TARGET_FUNCTION_ARG_OFFSET): ...this.
7000 * doc/tm.texi: Regenerate.
7001 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
7002 * config/spu/spu.c (spu_function_arg_offset): New function.
7003 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
7004 * system.h (FUNCTION_ARG_OFFSET): Poison.
7006 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7007 Alan Hayard <alan.hayward@arm.com>
7008 David Sherwood <david.sherwood@arm.com>
7010 * target.def (truly_noop_truncation): New hook.
7011 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
7012 than TRULY_NOOP_TRUNCATION.
7013 * hooks.h (hook_bool_uint_uint_true): Declare.
7014 * hooks.c (hook_bool_uint_uint_true): New function.
7015 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
7016 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
7017 * doc/tm.texi: Regenerate.
7018 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
7019 rather than TRULY_NOOP_TRUNCATION in comments.
7020 (simplify_comparison): Likewise.
7021 (record_truncated_value): Likewise.
7022 * expmed.c (extract_bit_field_1): Likewise.
7023 (extract_split_bit_field): Likewise.
7024 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
7025 instead of TRULY_NOOP_TRUNCATION.
7026 * function.c (assign_parm_setup_block): Likewise.
7027 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
7028 * rtlhooks.c: Include target.h.
7029 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
7030 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
7031 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
7032 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
7033 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
7034 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
7035 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
7036 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
7037 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
7038 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
7039 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
7040 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
7041 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
7042 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
7043 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
7044 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
7045 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
7046 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
7047 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
7048 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
7049 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
7050 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
7051 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
7052 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
7053 * config/mips/mips.c (mips_truly_noop_truncation): New function.
7054 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7055 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7056 rather than TRULY_NOOP_TRUNCATION in comments.
7057 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
7058 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
7059 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
7060 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
7061 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
7062 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
7063 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
7064 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
7065 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
7066 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
7067 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
7068 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7069 rather than TRULY_NOOP_TRUNCATION in comments.
7070 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
7071 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
7072 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
7073 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
7074 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
7075 TRULY_NOOP_TRUNCATION condition.
7076 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
7077 (TRULY_NOOP_TRUNCATION): Delete.
7078 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
7079 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
7080 * config/spu/spu.c (spu_truly_noop_truncation): New function.
7081 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7082 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
7083 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
7084 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
7085 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7086 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7087 rather than TRULY_NOOP_TRUNCATION in comments.
7088 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
7089 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
7090 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
7091 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
7092 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
7093 * system.h (TRULY_NOOP_TRUNCATION): Poison.
7095 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
7098 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
7099 (*cmp_ior): Likewise.
7100 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
7101 (*ior_scc_scc_cmp): Likewise.
7102 (*and_scc_scc): Likewise.
7103 (*and_scc_scc_cmp): Likewise.
7105 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7106 Alan Hayard <alan.hayward@arm.com>
7107 David Sherwood <david.sherwood@arm.com>
7109 * target.def (can_change_mode_class): New hook.
7110 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
7111 (hard_regno_nregs): Likewise.
7112 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
7113 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
7114 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
7115 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
7116 (LOAD_EXTEND_OP): Update accordingly.
7117 * doc/tm.texi: Regenerate.
7118 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
7119 CANNOT_CHANGE_MODE_CLASS.
7120 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
7121 (REG_CAN_CHANGE_MODE_P): ...this new macro.
7122 * combine.c (simplify_set): Update accordingly.
7123 * emit-rtl.c (validate_subreg): Likewise.
7124 * recog.c (general_operand): Likewise.
7125 * regcprop.c (mode_change_ok): Likewise.
7126 * reload1.c (choose_reload_regs): Likewise.
7127 (inherit_piecemeal_p): Likewise.
7128 * rtlanal.c (simplify_subreg_regno): Likewise.
7129 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
7130 instead of CANNOT_CHANGE_MODE_CLASS.
7131 (reload_cse_simplify_operands): Likewise.
7132 * reload.c (push_reload): Use targetm.can_change_mode_class
7133 instead of CANNOT_CHANGE_MODE_CLASS.
7134 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
7135 REG_CANNOT_CHANGE_MODE_P.
7136 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7137 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
7138 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7139 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7140 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7141 (arm_can_change_mode_class): New function.
7142 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
7143 than CANNOT_CHANGE_MODE_CLASS in comments.
7144 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7145 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
7146 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
7147 (ix86_can_change_mode_class): ...this new function, inverting the
7148 sense of the return value.
7149 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7150 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7151 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7152 (ia64_can_change_mode_class): New function.
7153 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7154 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
7155 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
7156 (m32c_can_change_mode_class): ...this new function, inverting the
7157 sense of the return value.
7158 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7159 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7160 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
7161 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
7162 (mips_can_change_mode_class): ...this new function, inverting the
7163 sense of the return value.
7164 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7165 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7166 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7167 (msp430_can_change_mode_class): New function.
7168 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7169 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
7170 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7171 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7172 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7173 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
7174 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7175 (pa_cannot_change_mode_class): Replace with...
7176 (pa_can_change_mode_class): ...this new function, inverting the
7177 sense of the return value.
7178 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
7179 than CANNOT_CHANGE_MODE_CLASS in comments.
7180 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7181 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
7182 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7183 (pdp11_cannot_change_mode_class): Replace with...
7184 (pdp11_can_change_mode_class): ...this new function, inverting the
7185 sense of the return value.
7186 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7187 * config/powerpcspe/powerpcspe-protos.h
7188 (rs6000_cannot_change_mode_class_ptr): Delete.
7189 * config/powerpcspe/powerpcspe.c
7190 (rs6000_cannot_change_mode_class_ptr): Delete.
7191 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7192 (rs6000_option_override_internal): Assign to
7193 targetm.can_change_mode_class instead of
7194 rs6000_cannot_change_mode_class_ptr.
7195 (rs6000_cannot_change_mode_class): Replace with...
7196 (rs6000_can_change_mode_class): ...this new function, inverting the
7197 sense of the return value.
7198 (rs6000_debug_cannot_change_mode_class): Replace with...
7199 (rs6000_debug_can_change_mode_class): ...this new function.
7200 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7201 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
7202 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7203 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7204 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
7206 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
7207 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7208 (rs6000_option_override_internal): Assign to
7209 targetm.can_change_mode_class instead of
7210 rs6000_cannot_change_mode_class_ptr.
7211 (rs6000_cannot_change_mode_class): Replace with...
7212 (rs6000_can_change_mode_class): ...this new function, inverting the
7213 sense of the return value.
7214 (rs6000_debug_cannot_change_mode_class): Replace with...
7215 (rs6000_debug_can_change_mode_class): ...this new function.
7216 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7217 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
7218 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
7219 (s390_can_change_mode_class): ...this new function, inverting the
7220 sense of the return value.
7221 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7222 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7223 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
7224 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7225 (sh_cannot_change_mode_class): Replace with...
7226 (sh_can_change_mode_class): ...this new function, inverting the
7227 sense of the return value.
7228 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7229 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7230 (sparc_can_change_mode_class): New function.
7231 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7232 * config/spu/spu.c (spu_can_change_mode_class): New function.
7233 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7234 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7235 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7236 (visium_can_change_mode_class): New function.
7237 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
7239 2017-09-15 Richard Biener <rguenther@suse.de>
7241 PR tree-optimization/82217
7242 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
7243 but not undefined case.
7245 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7248 * postreload.c (reload_cse_simplify_operands): Skip
7249 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
7251 2017-09-15 Richard Biener <rguenther@suse.de>
7253 PR tree-optimization/68823
7254 * graphite-scop-detection.c (build_alias_set): If we have a
7255 possible dependence check whether we can handle them by just
7256 looking at the DRs DR_ACCESS_FNs.
7257 (build_scops): If build_alias_set fails, fail the SCOP.
7259 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
7261 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
7262 to support float128 built-in functions that require the ISA 3.0
7264 (BU_FLOAT128_3_HW): Likewise.
7265 (SQRTF128): Add support for the IEEE 128-bit square root and fma
7267 (FMAF128): Likewise.
7269 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
7270 support for built-in functions that need the ISA 3.0 IEEE 128-bit
7271 floating point instructions.
7272 (rs6000_invalid_builtin): Likewise.
7273 (rs6000_builtin_mask_names): Likewise.
7274 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
7275 (RS6000_BTM_FLOAT128_HW): Likewise.
7276 (RS6000_BTM_COMMON): Likewise.
7277 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
7279 * doc/extend.texi (RS/6000 built-in functions): Document the
7280 IEEE 128-bit floating point square root and fused multiply-add
7283 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
7285 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
7286 reg (r2) isn't in the set of registers defined in the prologue.
7288 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7289 Alan Hayward <alan.hayward@arm.com>
7290 David Sherwood <david.sherwood@arm.com>
7292 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
7293 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
7294 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
7295 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
7296 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
7298 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
7299 max_vectorization_factor.
7300 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
7302 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7303 Alan Hayward <alan.hayward@arm.com>
7304 David Sherwood <david.sherwood@arm.com>
7306 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
7307 (vect_worthwhile_without_simd_p): Declare.
7308 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
7309 (vectorizable_reduction): Use it.
7310 * tree-vect-stmts.c (vectorizable_shift): Likewise.
7311 (vectorizable_operation): Likewise.
7313 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7314 Alan Hayward <alan.hayward@arm.com>
7315 David Sherwood <david.sherwood@arm.com>
7317 * tree-vectorizer.h (vect_get_num_copies): New function.
7318 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
7319 * tree-vect-loop.c (vectorizable_reduction): Likewise.
7320 (vectorizable_induction): Likewise.
7321 (vectorizable_live_operation): Likewise.
7322 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
7323 (vectorizable_bswap): Likewise.
7324 (vectorizable_call): Likewise.
7325 (vectorizable_conversion): Likewise.
7326 (vectorizable_assignment): Likewise.
7327 (vectorizable_shift): Likewise.
7328 (vectorizable_operation): Likewise.
7329 (vectorizable_store): Likewise.
7330 (vectorizable_load): Likewise.
7331 (vectorizable_condition): Likewise.
7332 (vectorizable_comparison): Likewise.
7333 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
7335 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7336 Alan Hayward <alan.hayward@arm.com>
7337 David Sherwood <david.sherwood@arm.com>
7339 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
7340 of vect_init_vector.
7342 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7343 Alan Hayward <alan.hayward@arm.com>
7344 David Sherwood <david.sherwood@arm.com>
7346 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
7347 an inline wrapper that provides a location.
7348 (gimple_build_vector): Likewise.
7349 * gimple-fold.c (gimple_build_vector_from_val): New function.
7350 (gimple_build_vector): Likewise.
7351 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
7352 functions to build the initial value. Always return a gimple value.
7353 (get_initial_defs_for_reduction): Likewise. Only compute
7355 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
7356 vect_init_vector on the results from get_initial_def(s)_for_reduction.
7357 (vectorizable_induction): Use gimple_build_vector rather than
7360 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7361 Alan Hayward <alan.hayward@arm.com>
7362 David Sherwood <david.sherwood@arm.com>
7364 * target.h (vec_perm_indices): New typedef.
7365 (auto_vec_perm_indices): Likewise.
7366 * optabs-query.h: Include target.h
7367 (can_vec_perm_p): Take a vec_perm_indices *.
7368 * optabs-query.c (can_vec_perm_p): Likewise.
7369 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
7370 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
7371 * tree-vect-generic.c (lower_vec_perm): Likewise.
7372 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
7373 (vect_grouped_load_supported): Likewise.
7374 (vect_shift_permute_load_chain): Likewise.
7375 (vect_permute_store_chain): Use auto_vec_perm_indices.
7376 (vect_permute_load_chain): Likewise.
7377 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
7378 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
7379 Update uses of can_vec_perm_p.
7380 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
7381 mode with a number of elements. Take a vec_perm_indices *.
7382 (vect_create_epilog_for_reduction): Update accordingly.
7383 Use auto_vec_perm_indices.
7384 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
7385 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
7386 (vect_transform_slp_perm_load): Likewise.
7387 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
7388 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
7389 (vect_gen_perm_mask_checked): Likewise.
7390 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
7391 (vect_gen_perm_mask_checked): Likewise.
7392 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
7393 (vectorizable_store): Likewise.
7394 (vectorizable_load): Likewise.
7395 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
7396 (vectorizable_bswap): Likewise.
7398 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7399 Alan Hayward <alan.hayward@arm.com>
7400 David Sherwood <david.sherwood@arm.com>
7402 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
7403 * tree.c (build_vector): Likewise.
7404 (build_vector_from_ctor): Update accordingly.
7405 (build_vector_from_val): Likewise.
7406 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
7407 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
7408 * tree-vect-generic.c (add_rshift): Likewise.
7409 (expand_vector_divmod): Likewise.
7410 (optimize_vector_constructor): Likewise.
7411 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
7412 (vect_transform_slp_perm_load): Likewise.
7413 (vect_schedule_slp_instance): Likewise.
7414 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
7415 (vectorizable_call): Likewise.
7416 (vect_gen_perm_mask_any): Likewise. Add elements in order.
7417 * expmed.c (make_tree): Likewise.
7418 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
7419 a vector passed to build_vector.
7420 (fold_convert_const): Likewise.
7421 (exact_inverse): Likewise.
7422 (fold_ternary_loc): Likewise.
7423 (fold_relational_const): Likewise.
7424 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
7425 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
7426 (const_unop): Likewise. Store the reduction accumulator in a
7427 variable rather than an array.
7428 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
7429 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
7430 the new vector, rather than constructing it after the input arrays.
7431 (native_interpret_vector): Use auto_vec<tree> when building
7432 a vector passed to build_vector. Add elements in order.
7433 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
7434 auto_vec<tree> when building a vector passed to build_vector.
7435 (vect_create_epilog_for_reduction): Likewise.
7436 (vectorizable_induction): Likewise.
7437 (get_initial_def_for_reduction): Likewise. Fix indentation of
7439 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
7441 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
7442 passed to build_vector.
7444 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7445 Alan Hayward <alan.hayward@arm.com>
7446 David Sherwood <david.sherwood@arm.com>
7448 * tree-core.h (tree_base::u): Add an "nelts" field.
7449 (tree_vector): Use VECTOR_CST_NELTS as the length.
7450 * tree.c (tree_size): Likewise.
7451 (make_vector): Initialize VECTOR_CST_NELTS.
7452 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
7453 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
7454 TYPE_VECTOR_SUBPARTS.
7455 * expr.c (const_vector_mask_from_tree): Consistently use "units"
7456 as the number of units, setting it from VECTOR_CST_NELTS.
7457 (const_vector_from_tree): Likewise.
7458 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
7459 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
7460 (fold_negate_expr_1): Likewise.
7461 (fold_convert_const): Likewise.
7462 (const_binop): Likewise. Differentiate the number of output and
7464 (const_unop): Likewise.
7465 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
7466 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
7467 in cases that did the opposite.
7469 2017-09-14 Richard Biener <rguenther@suse.de>
7471 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
7474 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
7476 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
7478 2017-09-14 Jakub Jelinek <jakub@redhat.com>
7481 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
7482 if and where to split a bb, except for splitting before debug insn
7483 sequences followed by non-label real insn. Delete debug insns
7484 in between basic blocks.
7486 * combine.c (make_compound_operation_int): Formatting fixes.
7488 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
7489 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7490 * config/netbsd.h (LINK_EH_SPEC): Likewise.
7491 * config/sol2.h (LINK_EH_SPEC): Likewise.
7492 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7493 * config/s390/linux.h (LINK_SPEC): Likewise.
7494 * config/freebsd.h (LINK_EH_SPEC): Likewise.
7495 * config/openbsd.h (LINK_EH_SPEC): Likewise.
7496 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7497 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
7498 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
7499 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7500 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
7501 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
7503 2017-09-13 Jakub Jelinek <jakub@redhat.com>
7505 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
7507 (ENDFILE_LINUX_SPEC): Likewise.
7508 (LINK_EH_SPEC): Likewise.
7509 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
7510 (LINK_OS_LINUX_SPEC32): Likewise.
7511 (LINK_OS_LINUX_SPEC64): Likewise.
7512 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
7513 (LINK_OS_LINUX_SPEC): Likewise.
7515 2017-09-13 Martin Liska <mliska@suse.cz>
7518 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
7519 CASE_HIGH is NULL_TREE.
7521 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
7522 Alan Hayward <alan.hayward@arm.com>
7523 David Sherwood <david.sherwood@arm.com>
7525 * target.def (secondary_memory_needed): New hook.
7526 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
7527 instead of SECONDARY_MEMORY_NEEDED.
7528 (secondary_memory_needed_mode): Likewise.
7529 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
7530 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
7531 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
7532 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
7533 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
7534 * doc/tm.texi: Regenerate.
7535 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
7536 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
7537 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7538 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
7539 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
7540 * config/i386/i386.c (inline_secondary_memory_needed): Put the
7541 mode argument first and change the reg_class arguments to reg_class_t.
7542 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
7543 Make static. Update the call to inline_secondary_memory_needed.
7544 (ix86_register_move_cost): Update the call to
7545 inline_secondary_memory_needed.
7546 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7547 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
7549 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
7550 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7552 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
7553 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
7554 * config/mips/mips.c (mips_secondary_memory_needed): Make static
7555 and match hook interface. Add comment from mips.h.
7556 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7557 * config/mmix/mmix.md (truncdfsf2): Refer to
7558 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7560 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
7561 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
7562 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7563 (pa_secondary_memory_needed): New function.
7564 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
7565 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
7566 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7567 (pdp11_secondary_memory_needed): Make static and match hook interface.
7568 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
7569 * config/powerpcspe/powerpcspe-protos.h
7570 (rs6000_secondary_memory_needed_ptr): Delete.
7571 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
7573 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7574 (rs6000_option_override_internal): Assign to
7575 targetm.secondary_memory_needed rather than
7576 rs6000_secondary_memory_needed_ptr.
7577 (rs6000_secondary_memory_needed): Match hook interface.
7578 (rs6000_debug_secondary_memory_needed): Likewise.
7579 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
7580 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
7581 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
7582 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7583 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
7584 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
7586 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
7587 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7588 (rs6000_option_override_internal): Assign to
7589 targetm.secondary_memory_needed rather than
7590 rs6000_secondary_memory_needed_ptr.
7591 (rs6000_secondary_memory_needed): Match hook interface.
7592 (rs6000_debug_secondary_memory_needed): Likewise.
7593 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
7594 * config/s390/s390.c (s390_secondary_memory_needed): New function.
7595 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7596 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
7597 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7598 (sparc_secondary_memory_needed): New function.
7599 * lra-constraints.c (check_and_process_move): Refer to
7600 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7602 (curr_insn_transform): Likewise.
7603 (process_alt_operands): Use targetm.secondary_memory_needed
7604 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7605 (check_secondary_memory_needed_p): Likewise.
7606 (choose_split_class): Likewise.
7607 * reload.c: Unconditionally include code that was previously
7608 conditional on SECONDARY_MEMORY_NEEDED.
7609 (push_secondary_reload): Use targetm.secondary_memory_needed
7610 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7611 (push_reload): Likewise.
7612 * reload1.c: Unconditionally include code that was previously
7613 conditional on SECONDARY_MEMORY_NEEDED.
7614 (choose_reload_regs): Use targetm.secondary_memory_needed
7615 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7616 (gen_reload): Likewise.
7617 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
7619 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
7620 Alan Hayward <alan.hayward@arm.com>
7621 David Sherwood <david.sherwood@arm.com>
7623 * target.def (secondary_memory_needed_mode): New hook:
7624 * targhooks.c (default_secondary_memory_needed_mode): Declare.
7625 * targhooks.h (default_secondary_memory_needed_mode): New function.
7626 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
7627 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
7628 * doc/tm.texi: Regenerate.
7629 * lra-constraints.c (check_and_process_move): Use
7630 targetm.secondary_memory_needed_mode instead of
7631 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
7632 (curr_insn_transform): Likewise.
7633 * reload.c (get_secondary_mem): Likewise.
7634 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7635 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
7637 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7638 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7639 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
7640 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7641 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
7643 * config/powerpcspe/powerpcspe-protos.h
7644 (rs6000_secondary_memory_needed_mode): Delete.
7645 * config/powerpcspe/powerpcspe.c
7646 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7647 (rs6000_secondary_memory_needed_mode): Make static.
7648 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7649 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
7651 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
7653 (rs6000_secondary_memory_needed_mode): Make static.
7654 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7655 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
7656 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7657 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7658 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
7660 (sparc_secondary_memory_needed_mode): New function.
7661 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
7663 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
7665 * config/aarch64/constraints.md (Umq): New constraint.
7666 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
7668 (mov<mode>): Update condition.
7670 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7672 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
7673 when bitposition is the same.
7675 2017-09-13 Richard Biener <rguenther@suse.de>
7677 * dwarf2out.c (output_die_symbol): Remove.
7678 (output_die): Do not output a DIEs symbol.
7680 2017-09-13 Richard Biener <rguenther@suse.de>
7683 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
7684 default-def to avoid breaking iterator update with the weird
7685 interaction with cgraph_update_edges_for_call_stmt_node.
7687 2017-09-13 Richard Biener <rguenther@suse.de>
7689 * tree-cfg.c (verify_gimple_assign_binary): Add verification
7690 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
7691 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
7692 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
7694 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
7696 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
7697 Disable pc relative literal load irrespective of
7698 TARGET_FIX_ERR_A53_84341 for default.
7700 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
7702 * config/sparc/sparc.c (output_return): Output the source location of
7703 the insn in the delay slot, if any.
7704 (output_sibcall): Likewise.
7706 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
7709 * common.opt (-static-pie): New alias.
7710 (shared): Negate static-pie.
7711 (-no-pie): Update help text.
7713 (static-pie): New option.
7714 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
7715 -static-pie support.
7716 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
7717 (LINK_EH_SPEC): Likewise.
7718 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7719 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
7720 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
7721 * gcc.c (LINK_COMMAND_SPEC): Likewise.
7722 (init_gcc_specs): Likewise.
7723 (init_spec): Likewise.
7724 (display_help): Update help message for -pie.
7725 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
7728 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
7730 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
7731 (movdi_aarch64): Likewise.
7732 (movti_aarch64): Likewise.
7734 2017-09-12 Simon Wright <simon@pushface.org>
7737 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
7738 calculation of the minor version, always output as 0.
7740 2017-09-12 Jakub Jelinek <jakub@redhat.com>
7743 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
7744 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
7745 on it early, rather than manual conversion late. For
7746 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
7747 instead of performing manual conversion.
7749 2017-09-12 Carl Love <cel@us.ibm.com>
7751 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
7752 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
7754 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
7755 VMULOSW): Add definitions.
7756 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7757 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
7758 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
7759 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
7760 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
7762 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
7764 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
7766 (movti_aarch64): Likewise.
7767 (movdf_aarch64): Likewise.
7768 (movtf_aarch64): Likewise.
7769 (load_pairdi): Likewise.
7770 (store_pairdi): Likewise.
7771 (load_pairdf): Likewise.
7772 (store_pairdf): Likewise.
7773 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
7774 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
7775 (ldr_got_small_<mode>): Likewise.
7776 (ldr_got_small_28k_<mode>): Likewise.
7777 (ldr_got_tiny): Likewise.
7778 * config/aarch64/iterators.md (ldst_sz): New.
7779 (ldpstp_sz): Likewise.
7780 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
7782 (thunderx_load): Split load_8 to load_16.
7783 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
7785 (thunderx2t99_storepair_basic): Split store_8 to store_16.
7786 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
7787 (xgene1_store_pair): Split store_8 to store_16.
7788 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
7789 (falkor_st_0_st_sd): Split store_8 to store_16.
7791 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
7793 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
7794 and store1/2/3/4 to store_4/8/12/16.
7795 * config/aarch64/aarch64.md: Update for rename.
7796 * config/arm/arm.md: Likewise.: Likewise.
7797 * config/arm/arm.c: Likewise.
7798 * config/arm/thumb1.md: Likewise.
7799 * config/arm/thumb2.md: Likewise.
7800 * config/arm/vfp.md: Likewise.
7801 * config/arm/arm-generic.md: Likewise.
7802 * config/arm/arm1020e.md: Likewise.
7803 * config/arm/arm1026ejs.md: Likewise.
7804 * config/arm/arm1136jfs.md: Likewise.
7805 * config/arm/arm926ejs.md: Likewise.
7806 * config/arm/cortex-a15.md: Likewise.
7807 * config/arm/cortex-a17.md: Likewise.
7808 * config/arm/cortex-a5.md: Likewise.
7809 * config/arm/cortex-a53.md: Likewise.
7810 * config/arm/cortex-a57.md: Likewise.
7811 * config/arm/cortex-a7.md: Likewise.
7812 * config/arm/cortex-a8.md: Likewise.
7813 * config/arm/cortex-a9.md: Likewise.
7814 * config/arm/cortex-m4.md: Likewise.
7815 * config/arm/cortex-m7.md: Likewise.
7816 * config/arm/cortex-r4.md: Likewise.
7817 * config/arm/exynos-m1.md: Likewise.
7818 * config/arm/fa526.md: Likewise.
7819 * config/arm/fa606te.md: Likewise.
7820 * config/arm/fa626te.md: Likewise.
7821 * config/arm/fa726te.md: Likewise.
7822 * config/arm/fmp626.md: Likewise.
7823 * config/arm/iwmmxt.md: Likewise.
7824 * config/arm/ldmstm.md: Likewise.
7825 * config/arm/marvell-pj4.md: Likewise.
7826 * config/arm/xgene1.md: Likewise.
7827 * config/aarch64/thunderx.md: Likewise.
7828 * config/aarch64/thunderx2t99.md: Likewise.
7829 * config/aarch64/falkor.md: Likewise.
7831 2017-09-12 Martin Liska <mliska@suse.cz>
7833 * attribs.c (private_lookup_attribute): New function.
7834 * attribs.h (private_lookup_attribute): Declared here.
7835 (lookup_attribute): Called from this place.
7837 2017-09-12 Richard Biener <rguenther@suse.de>
7839 PR tree-optimization/82157
7840 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
7841 stmts with side-effects.
7843 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7844 Alan Hayward <alan.hayward@arm.com>
7845 David Sherwood <david.sherwood@arm.com>
7847 * target.def (hard_regno_nregs): New hook.
7848 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
7849 * targhooks.h (default_hard_regno_nregs): Declare.
7850 * targhooks.c (default_hard_regno_nregs): New function.
7851 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
7852 (TARGET_HARD_REGNO_NREGS): ...this hook.
7853 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
7854 (CLASS_MAX_NREGS): Likewise.
7855 * doc/tm.texi: Regenerate.
7856 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
7857 instead of HARD_REGNO_NREGS.
7858 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
7859 HARD_REGNO_NREGS in the comment.
7860 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
7861 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
7862 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
7863 Return an unsigned int.
7864 (TARGET_HARD_REGNO_NREGS): Redefine.
7865 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
7866 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
7867 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
7868 (arc_hard_regno_nregs): New function.
7869 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
7870 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
7871 (arm_hard_regno_nregs): New function.
7872 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
7873 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
7874 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
7875 (TARGET_HARD_REGNO_NREGS): Redefine.
7876 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
7877 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
7878 (HARD_REGNO_NREGS): Delete.
7879 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
7880 (cr16_hard_regno_nregs): New function.
7881 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
7882 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
7883 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
7884 (cris_hard_regno_nregs): New function.
7885 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
7886 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
7887 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
7888 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
7889 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
7890 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
7891 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
7892 (frv_hard_regno_nregs): Make static. Take and return an
7894 (frv_class_max_nregs): Remove outdated copy of documentation.
7895 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
7896 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
7897 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
7898 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
7899 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
7900 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
7901 (TARGET_HARD_REGNO_NREGS): Redefine.
7902 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
7903 (CLASS_MAX_NREGS): Update comment.
7904 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
7905 (ia64_hard_regno_nregs): New function.
7906 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
7907 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
7908 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
7909 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
7910 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
7912 (m32c_hard_regno_nregs): Likewise. Make static.
7913 (TARGET_HARD_REGNO_NREGS): Redefine.
7914 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
7915 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
7916 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
7917 (m68k_hard_regno_nregs): New function.
7918 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
7919 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
7920 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
7921 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
7922 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
7923 Take and return an unsigned int.
7924 (TARGET_HARD_REGNO_NREGS): Redefine.
7925 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
7926 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
7927 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
7928 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
7929 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
7930 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
7931 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
7932 (msp430_hard_regno_nregs): Make static. Take and return an
7934 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
7935 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
7936 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
7937 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
7938 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
7939 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
7940 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
7941 (TARGET_HARD_REGNO_NREGS): Redefine.
7942 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
7943 (PA_HARD_REGNO_NREGS): ...this.
7944 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
7945 (PA_HARD_REGNO_NREGS): ...this.
7946 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
7947 (pa_hard_regno_nregs): New function.
7948 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
7949 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
7950 (pdp11_hard_regno_nregs): New function.
7951 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
7952 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
7953 (rs6000_hard_regno_nregs_hook): New function.
7954 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
7955 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
7956 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
7957 Take and return an unsigned int. Move earlier in file.
7958 (TARGET_HARD_REGNO_NREGS): Redefine.
7959 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
7960 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
7961 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
7962 (rl78_hard_regno_nregs): Make static. Take and return an
7964 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
7965 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
7966 (rs6000_hard_regno_nregs_hook): New function.
7967 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
7968 * config/rx/rx.c (rx_hard_regno_nregs): New function.
7969 (TARGET_HARD_REGNO_NREGS): Redefine.
7970 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
7971 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
7972 instead of HARD_REGNO_NREGS.
7973 (s390_hard_regno_nregs): New function.
7974 (s390_hard_regno_mode_ok): Add comment from s390.h.
7975 (TARGET_HARD_REGNO_NREGS): Redefine.
7976 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
7977 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
7978 (sh_hard_regno_nregs): New function.
7979 (sh_pass_in_reg_p): Use it.
7980 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
7981 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
7982 (sparc_hard_regno_nregs): New function.
7983 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
7984 * config/spu/spu.c (spu_hard_regno_nregs): New function.
7985 (spu_function_arg_advance): Use it, supplying a valid register number.
7986 (TARGET_HARD_REGNO_NREGS): Redefine.
7987 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
7988 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
7989 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
7990 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
7991 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
7992 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
7993 (CLASS_MAX_NREGS): Remove copy of old documentation.
7994 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
7995 (visium_hard_regno_nregs): New function.
7996 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
7997 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
7998 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
7999 xtensa_hard_regno_nregs): New function.
8000 * system.h (HARD_REGNO_NREGS): Poison.
8002 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8004 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
8005 hard_regno_nregs instead of HARD_REGNO_NREGS.
8006 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
8007 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
8008 (c6x_expand_epilogue): Likewise.
8009 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
8010 (frv_read_iacc_argument): Likewise.
8011 * config/sh/sh.c: Include regs.h.
8012 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
8013 (regs_used): Likewise.
8014 (output_stack_adjust): Likewise.
8015 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
8016 * expmed.c: Include regs.h.
8017 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
8018 * ree.c: Include regs.h.
8019 (combine_reaching_defs): Use hard_regno_nregs instead of
8021 (add_removable_extension): Likewise.
8023 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8025 * regs.h (hard_regno_nregs): Turn into a function.
8026 (end_hard_regno): Update accordingly.
8027 * caller-save.c (setup_save_areas): Likewise.
8028 (save_call_clobbered_regs): Likewise.
8029 (replace_reg_with_saved_mem): Likewise.
8030 (insert_restore): Likewise.
8031 (insert_save): Likewise.
8032 * combine.c (can_change_dest_mode): Likewise.
8033 (move_deaths): Likewise.
8034 (distribute_notes): Likewise.
8035 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
8036 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
8037 (rs6000_split_multireg_move): Likewise.
8038 (rs6000_register_move_cost): Likewise.
8039 (rs6000_memory_move_cost): Likewise.
8040 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
8041 (rs6000_split_multireg_move): Likewise.
8042 (rs6000_register_move_cost): Likewise.
8043 (rs6000_memory_move_cost): Likewise.
8044 * cselib.c (cselib_reset_table): Likewise.
8045 (cselib_lookup_1): Likewise.
8046 * emit-rtl.c (set_mode_and_regno): Likewise.
8047 * function.c (aggregate_value_p): Likewise.
8048 * ira-color.c (setup_profitable_hard_regs): Likewise.
8049 (check_hard_reg_p): Likewise.
8050 (calculate_saved_nregs): Likewise.
8051 (assign_hard_reg): Likewise.
8052 (improve_allocation): Likewise.
8053 (calculate_spill_cost): Likewise.
8054 * ira-emit.c (modify_move_list): Likewise.
8055 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
8056 (ira_hard_reg_in_set_p): Likewise.
8057 * ira.c (setup_reg_mode_hard_regset): Likewise.
8058 (clarify_prohibited_class_mode_regs): Likewise.
8059 (check_allocation): Likewise.
8060 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8061 (lra_setup_reg_renumber): Likewise.
8062 (setup_try_hard_regno_pseudos): Likewise.
8063 (spill_for): Likewise.
8064 (assign_hard_regno): Likewise.
8065 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
8066 * lra-constraints.c (in_class_p): Likewise.
8067 (lra_constraint_offset): Likewise.
8068 (simplify_operand_subreg): Likewise.
8069 (lra_constraints): Likewise.
8070 (split_reg): Likewise.
8071 (split_if_necessary): Likewise.
8072 (invariant_p): Likewise.
8073 (inherit_in_ebb): Likewise.
8074 * lra-lives.c (process_bb_lives): Likewise.
8075 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
8076 (get_hard_regs): Likewise.
8077 (do_remat): Likewise.
8078 * lra-spills.c (assign_spill_hard_regs): Likewise.
8079 * mode-switching.c (create_pre_exit): Likewise.
8080 * postreload.c (reload_combine_recognize_pattern): Likewise.
8081 * recog.c (peep2_find_free_register): Likewise.
8082 * regcprop.c (kill_value_regno): Likewise.
8083 (set_value_regno): Likewise.
8084 (copy_value): Likewise.
8085 (maybe_mode_change): Likewise.
8086 (find_oldest_value_reg): Likewise.
8087 (copyprop_hardreg_forward_1): Likewise.
8088 * regrename.c (check_new_reg_p): Likewise.
8089 (regrename_do_replace): Likewise.
8090 * reload.c (push_reload): Likewise.
8091 (combine_reloads): Likewise.
8092 (find_dummy_reload): Likewise.
8093 (operands_match_p): Likewise.
8094 (find_reloads): Likewise.
8095 (find_equiv_reg): Likewise.
8096 (reload_adjust_reg_for_mode): Likewise.
8097 * reload1.c (count_pseudo): Likewise.
8098 (count_spilled_pseudo): Likewise.
8099 (find_reg): Likewise.
8100 (clear_reload_reg_in_use): Likewise.
8101 (free_for_value_p): Likewise.
8102 (allocate_reload_reg): Likewise.
8103 (choose_reload_regs): Likewise.
8104 (reload_adjust_reg_for_temp): Likewise.
8105 (emit_reload_insns): Likewise.
8106 (delete_output_reload): Likewise.
8107 * rtlanal.c (subreg_get_info): Likewise.
8108 * sched-deps.c (sched_analyze_reg): Likewise.
8109 * sel-sched.c (init_regs_for_mode): Likewise.
8110 (mark_unavailable_hard_regs): Likewise.
8111 (choose_best_reg_1): Likewise.
8112 (verify_target_availability): Likewise.
8113 * valtrack.c (dead_debug_insert_temp): Likewise.
8114 * var-tracking.c (track_loc_p): Likewise.
8115 (emit_note_insn_var_location): Likewise.
8116 * varasm.c (make_decl_rtl): Likewise.
8117 * reginfo.c (choose_hard_reg_mode): Likewise.
8118 (init_reg_modes_target): Refer directly to
8119 this_target_regs->x_hard_regno_nregs.
8121 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8123 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
8124 instead of hard_regno_nregs.
8126 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8128 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
8129 end_hard_regno instead of hard_regno_nregs.
8130 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
8131 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
8132 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
8133 * ira-color.c (improve_allocation): Likewise.
8134 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8135 * lra-lives.c (mark_regno_live): Likewise.
8136 (mark_regno_dead): Likewise.
8137 * lra-remat.c (operand_to_remat): Likewise.
8138 * lra.c (collect_non_operand_hard_regs): Likewise.
8139 * postreload.c (reload_combine_note_store): Likewise.
8140 (move2add_valid_value_p): Likewise.
8141 * reload.c (regno_clobbered_p): Likewise.
8143 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8145 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
8147 * config/v850/v850.c (v850_reorg): Likewise.
8148 * reload.c (refers_to_regno_for_reload_p): Likewise.
8149 (find_equiv_reg): Likewise.
8150 * reload1.c (reload_reg_reaches_end_p): Likewise.
8152 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8154 * caller-save.c (add_used_regs): Use REG_NREGS instead of
8156 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
8157 * config/arm/arm.c (output_move_neon): Likewise.
8158 (arm_attr_length_move_neon): Likewise.
8159 (neon_split_vcombine): Likewise.
8160 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
8161 (c6x_mark_reg_written): Likewise.
8162 (c6x_dwarf_register_span): Likewise.
8163 * config/i386/i386.c (ix86_save_reg): Likewise.
8164 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
8165 (rws_access_reg): Likewise.
8166 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
8167 * mode-switching.c (create_pre_exit): Likewise.
8168 * ree.c (combine_reaching_defs): Likewise.
8169 (add_removable_extension): Likewise.
8170 * regcprop.c (find_oldest_value_reg): Likewise.
8171 (copyprop_hardreg_forward_1): Likewise.
8172 * reload.c (reload_inner_reg_of_subreg): Likewise.
8173 (push_reload): Likewise.
8174 (combine_reloads): Likewise.
8175 (find_dummy_reload): Likewise.
8176 (reload_adjust_reg_for_mode): Likewise.
8177 * reload1.c (find_reload_regs): Likewise.
8178 (forget_old_reloads_1): Likewise.
8179 (reload_reg_free_for_value_p): Likewise.
8180 (reload_adjust_reg_for_temp): Likewise.
8181 (emit_reload_insns): Likewise.
8182 (delete_output_reload): Likewise.
8183 * sel-sched.c (choose_best_reg_1): Likewise.
8184 (choose_best_pseudo_reg): Likewise.
8186 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8187 Alan Hayward <alan.hayward@arm.com>
8188 David Sherwood <david.sherwood@arm.com>
8190 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
8191 * target.def (slow_unaligned_access): New hook.
8192 * targhooks.h (default_slow_unaligned_access): Declare.
8193 * targhooks.c (default_slow_unaligned_access): New function.
8194 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
8195 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
8196 * doc/tm.texi: Regenerate.
8197 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
8198 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
8199 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
8201 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
8202 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
8204 (rs6000_slow_unaligned_access): New function.
8205 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
8206 (expand_block_compare): Likewise.
8207 (expand_strn_compare): Likewise.
8208 (rs6000_rtx_costs): Likewise.
8209 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
8210 (riscv_slow_unaligned_access): Likewise.
8211 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
8212 (riscv_slow_unaligned_access_p): ...this and make static.
8213 (riscv_option_override): Update accordingly.
8214 (riscv_slow_unaligned_access): New function.
8215 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
8216 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
8217 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
8218 (rs6000_slow_unaligned_access): New function.
8219 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
8220 (rs6000_rtx_costs): Likewise.
8221 * config/rs6000/rs6000-string.c (expand_block_compare)
8222 (expand_strn_compare): Use targetm.slow_unaligned_access instead
8223 of SLOW_UNALIGNED_ACCESS.
8224 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
8225 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
8226 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
8227 of SLOW_UNALIGNED_ACCESS.
8228 * expmed.c (simple_mem_bitfield_p): Likewise.
8229 * expr.c (alignment_for_piecewise_move): Likewise.
8230 (emit_group_load_1): Likewise.
8231 (emit_group_store): Likewise.
8232 (copy_blkmode_from_reg): Likewise.
8233 (emit_push_insn): Likewise.
8234 (expand_assignment): Likewise.
8235 (store_field): Likewise.
8236 (expand_expr_real_1): Likewise.
8237 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
8238 * lra-constraints.c (simplify_operand_subreg): Likewise.
8239 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
8240 * gimple-ssa-store-merging.c: Likewise in block comment at start
8242 * tree-ssa-strlen.c: Include target.h.
8243 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
8244 of SLOW_UNALIGNED_ACCESS.
8245 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
8247 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8249 PR rtl-optimization/82185
8250 * expmed.c (emit_store_flag_int): Only test tem if it has been
8253 2017-09-12 Richard Biener <rguenther@suse.de>
8256 * match.pd ((FTYPE) N CMP CST): Fix typo.
8258 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
8260 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
8262 (mips_near_type_p): Add 'short_call' attribute as a synonym
8264 * doc/extend.texi (short_call): Document new function attribute.
8266 2017-09-12 Jakub Jelinek <jakub@redhat.com>
8269 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
8270 assertion check that in the condition.
8271 (get_atomic_generic_size): Likewise. Before testing if parameter
8272 has pointer type, if it has array type, call for C++
8273 default_conversion to perform array-to-pointer conversion.
8275 2017-09-12 Richard Biener <rguenther@suse.de>
8277 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
8278 for operations we cannot scalarize.
8280 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
8282 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
8283 vectors heap vectors. Clean up comments.
8284 Make visited_bbs a reference.
8285 (profitable_jump_thread_path): Make GC
8286 vectors heap vectors. Clean up comments.
8288 (convert_and_register_jump_thread_path): Make GC vectors heap
8290 (check_subpath_and_update_thread_path): Same. Clean up comments.
8291 Make visited_bbs a reference.
8292 (handle_phi): Abstract common code to to
8293 register_jump_thread_path_if_profitable.
8294 Rename VAR_BB to DEF_BB.
8296 Make GC vectors heap vectors.
8297 Make visited_bbs a reference.
8298 (handle_assignment): Same.
8299 (register_jump_thread_path_if_profitable): New.
8300 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
8302 Make GC vectors heap vectors. Clean up comments.
8303 Make visited_bbs a reference.
8304 (find_jump_threads_backwards): Make visited_bbs live in the stack.
8305 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
8308 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
8311 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
8312 words of E_DImode object are reachable by xtensa_uimm8x4 access.
8314 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
8316 Revert r251800 and r251799.
8318 2017-09-11 Martin Jambor <mjambor@suse.cz>
8321 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
8322 arguments in advance.
8323 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
8324 use it to find predecessor edges.
8325 (naive_outof_ssa): Collect vector of predecessors.
8327 2017-09-08 Jason Merrill <jason@redhat.com>
8329 PR c++/70029 - ICE with ref-qualifier and -flto
8330 * langhooks.h (struct lang_hooks_for_types): Add
8331 copy_lang_qualifiers.
8332 * attribs.c (build_type_attribute_qual_variant): Use it.
8333 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
8335 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
8336 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
8338 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
8341 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
8342 (*mulsi3_sp64): New instruction.
8343 (mulsi3): New expander.
8345 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
8347 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
8349 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8351 * sancov.c: Include memmodel.h.
8353 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
8356 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
8359 2017-09-07 Carl Love <cel@us.ibm.com>
8361 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
8362 the sldi instruction.
8364 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
8366 * sancov.c: Include tm_p.h.
8368 2017-09-07 Jakub Jelinek <jakub@redhat.com>
8371 * output.h (switch_to_other_text_partition): New declaration.
8372 * varasm.c (switch_to_other_text_partition): New function.
8373 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
8374 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
8375 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
8376 to the other text partition before emitting LCL label and switch back
8377 after emitting the word after it.
8379 2017-09-07 Richard Biener <rguenther@suse.de>
8381 * passes.def (pass_split_crit_edges): Remove instance before PRE.
8382 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
8383 critical edges here, after loop init.
8384 (pass_data_pre): Remove PROP_no_crit_edges flags.
8385 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
8386 for valueization of call args to avoid leaking VN_TOP.
8387 (visit_use): Assert we do not visit default defs.
8388 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
8389 Use error_mark_node to more easily detect leaking VN_TOP.
8390 All default-defs are varying, not VN_TOP. Mark them visited.
8391 (run_scc_vn): Make code match comment.
8393 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
8395 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
8396 OPTION_MASK_FLOAT128_KEYWORD.
8397 (POWERPC_MASKS): Likewise.
8398 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
8399 support for the -mfloat128-type option, and make -mfloat128
8400 default on PowerPC Linux systems. Define or undefine
8401 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
8402 Define __float128 to be __ieee128 if IEEE 128-bit support is
8403 enabled, or undefine it.
8404 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
8405 Delete defining __FLOAT128_TYPE__.
8406 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
8407 -mfloat128-type option and make -mfloat128 default on PowerPC
8409 (TARGET_FLOAT128_TYPE): Likewise.
8410 (-mfloat128-type): Likewise.
8411 * config/rs6000/rs6000.c (rs6000_option_override_internal):
8412 Delete the -mfloat128-type option and make -mfloat128 default on
8413 PowerPC Linux systems. Always use __ieee128 to be the keyword for
8414 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
8415 128-bit floating point is enabled. Change tests from using
8416 -mfloat128-type to -mfloat128.
8417 (rs6000_mangle_type): Use the correct mangling for the __float128
8418 type even if normal long double is restricted to 64-bits.
8419 (floatn_mode): Enable the _Float128 type by default on VSX Linux
8421 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
8422 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
8423 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
8425 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
8426 documentation for -mfloat128.
8428 2017-09-06 Olivier Hainque <hainque@adacore.com>
8430 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
8432 2017-09-06 Wish Wu <wishwu007@gmail.com>
8433 Jakub Jelinek <jakub@redhat.com>
8435 * asan.c (initialize_sanitizer_builtins): Add
8436 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
8437 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
8438 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
8439 BT_FN_VOID_UINT64_PTR variables.
8440 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
8441 (BT_FN_VOID_UINT16_UINT16): Likewise.
8442 (BT_FN_VOID_UINT32_UINT32): Likewise.
8443 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
8444 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
8445 (BT_FN_VOID_UINT64_PTR): Likewise.
8446 * common.opt (flag_sanitize_coverage): New variable.
8447 (fsanitize-coverage=trace-pc): Remove.
8448 (fsanitize-coverage=): Add.
8449 * flag-types.h (enum sanitize_coverage_code): New enum.
8450 * fold-const.c (fold_range_test): Disable non-short-circuit
8451 optimization if flag_sanitize_coverage.
8452 (fold_truth_andor): Likewise.
8453 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
8454 * opts.c (COVERAGE_SANITIZER_OPT): Define.
8455 (coverage_sanitizer_opts): New array.
8456 (get_closest_sanitizer_option): Add OPTS argument, handle also
8457 OPT_fsanitize_coverage_.
8458 (parse_sanitizer_options): Adjusted to also handle
8459 OPT_fsanitize_coverage_.
8460 (common_handle_option): Add OPT_fsanitize_coverage_.
8461 * sancov.c (instrument_comparison, instrument_switch): New function.
8462 (sancov_pass): Add trace-cmp support.
8463 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
8464 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
8465 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
8466 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
8467 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
8468 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
8469 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
8470 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
8471 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
8472 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
8474 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
8476 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
8477 error. Only quit immediately if parsing is complete.
8478 (BEGIN): Initialize fatal_err and parse_done.
8479 (begin fpu, end fpu): Check number of arguments.
8480 (begin arch, end arch): Likewise.
8481 (begin cpu, end cpu): Likewise.
8482 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
8483 (optalias): Likewise.
8485 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
8487 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
8488 * config/arm/arm-isa.h: Delete. Move definitions to ...
8489 * arm-cpus.in: ... here. Use new feature and fgroup values.
8490 * config/arm/arm.c (arm_option_override): Use lower case for feature
8492 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
8493 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
8494 * config/arm/parsecpu.awk (END): Add new command 'isa'.
8496 (print_isa_bits_for): New function.
8497 (gen_isa): New function.
8498 (gen_comm_data): Use print_isa_bits_for.
8499 (define feature): New keyword.
8500 (define fgroup): New keyword.
8501 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
8502 (arm-isa.h): Add rule to generate file.
8503 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
8504 case for feature bit names.
8506 2017-09-06 Richard Biener <rguenther@suse.de>
8508 * tree-ssa-pre.c (NECESSARY): Remove.
8509 (create_expression_by_pieces): Do not touch pass-local flags.
8510 (insert_into_preds_of_block): Likewise.
8511 (do_pre_regular_insertion): Likewise.
8512 (eliminate_insert): Likewise.
8513 (eliminate_dom_walker::before_dom_children): Likewise.
8514 (fini_eliminate): Do not look at inserted_exprs.
8515 (mark_operand_necessary): Remove.
8516 (remove_dead_inserted_code): Replace with simple work-list
8517 algorithm based on inserted_exprs and SSA uses.
8518 (pass_pre::execute): Re-order fini_eliminate and
8519 remove_dead_inserted_code.
8521 2017-09-06 Olivier Hainque <hainque@adacore.com>
8523 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
8524 for VxWorks 7. Adjust surrounding comments.
8526 2017-09-06 Richard Biener <rguenther@suse.de>
8528 * gimple-ssa-strength-reduction.c
8529 (find_candidates_dom_walker::before_dom_children): Also allow
8532 2017-09-06 Richard Biener <rguenther@suse.de>
8534 PR tree-optimization/82108
8535 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
8536 for gap in the non-permutation SLP case.
8538 2017-09-06 Martin Jambor <mjambor@suse.cz>
8540 PR tree-optimization/82078
8541 * tree-sra.c (sort_and_splice_var_accesses): Move call to
8542 add_access_to_work_queue...
8543 (build_accesses_from_assign): ...here.
8544 (propagate_all_subaccesses): Make sure racc is the group
8545 representative, if there is one.
8547 2017-09-06 Jakub Jelinek <jakub@redhat.com>
8550 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
8553 2017-09-06 Richard Biener <rguenther@suse.de>
8555 * gimple-ssa-strength-reduction.c
8556 (find_candidates_dom_walker::before_doom_children): Use a
8557 type and not a mode check.
8559 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
8562 * config/arm/predicates.md (arm_general_adddi_operand): Create new
8564 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
8566 2017-09-05 Jeff Law <law@redhat.com>
8568 PR tree-optimization/64910
8569 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
8570 cases where we have 3 or more operands.
8572 2017-09-05 Jakub Jelinek <jakub@redhat.com>
8575 * omp-low.c (lower_omp_for): Recompute tree invariant if
8576 gimple_omp_for_initial/final is ADDR_EXPR.
8579 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
8580 into gimple val before gimplification fo the COND_EXPR.
8582 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
8584 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
8585 REGION_COPY argument.
8586 (thread_through_all_blocks): Remove unused argument to
8587 duplicate_thread_path.
8589 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8590 Alan Hayward <alan.hayward@arm.com>
8591 David Sherwood <david.sherwood@arm.com>
8593 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
8594 Take a scalar_mode rather than a machine_mode.
8595 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8596 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
8597 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8598 (aarch64_gen_adjusted_ldpstp): Likewise.
8599 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
8601 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8602 Alan Hayward <alan.hayward@arm.com>
8603 David Sherwood <david.sherwood@arm.com>
8605 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
8606 Take a scalar_int_mode instead of a machine_mode.
8607 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
8608 (aarch64_output_scalar_simd_mov_immediate): Likewise.
8609 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
8610 (aarch64_simd_attr_length_rglist): Delete.
8611 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
8612 a scalar_int_mode instead of a machine_mode.
8613 (aarch64_add_offset): Likewise.
8614 (aarch64_internal_mov_immediate): Likewise
8615 (aarch64_add_constant_internal): Likewise.
8616 (aarch64_add_constant): Likewise.
8617 (aarch64_movw_imm): Likewise.
8618 (aarch64_rtx_arith_op_extract_p): Likewise.
8619 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
8620 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
8621 Remove assert that the mode isn't a vector.
8622 (aarch64_output_scalar_simd_mov_immediate): Likewise.
8623 (aarch64_expand_mov_immediate): Update calls after above changes.
8624 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
8625 (aarch64_and_bitmask_imm): Check for scalar integer modes.
8626 (aarch64_move_imm): Likewise.
8627 (aarch64_can_const_movi_rtx_p): Likewise.
8628 (aarch64_strip_extend): Likewise.
8629 (aarch64_extr_rtx_p): Likewise.
8630 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
8631 a CONST_INT when the mode parameter is VOIDmode.
8632 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
8634 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8636 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
8637 * stor-layout.c (bitwise_mode_for_mode): Likewise.
8638 (bitwise_type_for_mode): Update accordingly.
8640 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8642 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
8643 * stor-layout.c (mode_for_size_tree): Likewise.
8644 (mode_for_array): Update accordingly.
8645 (layout_decl): Likewise.
8646 (compute_record_mode): Likewise. Only set the mode once.
8648 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8650 * target.def (get_mask_mode): Change return type to opt_mode.
8652 * doc/tm.texi: Regenerate.
8653 * targhooks.h (default_get_mask_mode): Return an opt_mode.
8654 * targhooks.c (default_get_mask_mode): Likewise.
8655 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
8656 * optabs-query.c (can_vec_mask_load_store_p): Update use of
8657 targetm.get_mask_mode.
8658 * tree.c (build_truth_vector_type): Likewise.
8660 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8662 * machmode.h (mode_for_vector): Return an opt_mode.
8663 * stor-layout.c (mode_for_vector): Likewise.
8664 (mode_for_int_vector): Update accordingly.
8665 (layout_type): Likewise.
8666 * config/i386/i386.c (emit_memmov): Likewise.
8667 (ix86_expand_set_or_movmem): Likewise.
8668 (ix86_expand_vector_init): Likewise.
8669 (ix86_get_mask_mode): Likewise.
8670 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
8672 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
8673 * expmed.c (extract_bit_field_1): Likewise.
8674 * expr.c (expand_expr_real_2): Likewise.
8675 * optabs-query.c (can_vec_perm_p): Likewise.
8676 (can_vec_mask_load_store_p): Likewise.
8677 * optabs.c (expand_vec_perm): Likewise.
8678 * targhooks.c (default_get_mask_mode): Likewise.
8679 * tree-vect-stmts.c (vectorizable_store): Likewise.
8680 (vectorizable_load): Likewise.
8681 (get_vectype_for_scalar_type_and_size): Likewise.
8683 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8685 * machmode.h (mode_for_int_vector): New function.
8686 * stor-layout.c (mode_for_int_vector): Likewise.
8687 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
8688 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
8689 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
8690 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
8691 (s390_expand_vcond): Likewise.
8693 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8695 * machmode.h (opt_machine_mode): New type.
8696 (opt_mode<T>): Allow construction from anything that can be
8698 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
8699 (mode_for_size): Return an opt_machine_mode.
8700 * stor-layout.c (mode_for_size): Likewise.
8701 (mode_for_size_tree): Update call accordingly.
8702 (bitwise_mode_for_mode): Likewise.
8703 (make_fract_type): Likewise.
8704 (make_accum_type): Likewise.
8705 * caller-save.c (replace_reg_with_saved_mem): Update call
8707 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8708 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8709 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8710 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8711 * expmed.c (extract_bit_field_1): Likewise.
8712 * reload.c (get_secondary_mem): Likewise.
8713 * varasm.c (assemble_integer): Likewise.
8714 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
8717 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8719 * machmode.h (decimal_float_mode_for_size): New function.
8720 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
8721 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
8722 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
8723 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
8724 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
8725 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
8727 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8729 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
8730 (get_builtin_sync_mode): Likewise.
8731 (expand_ifn_atomic_compare_exchange): Likewise.
8732 (expand_builtin_atomic_clear): Likewise.
8733 (expand_builtin_atomic_test_and_set): Likewise.
8734 (fold_builtin_atomic_always_lock_free): Likewise.
8735 * calls.c (compute_argument_addresses): Likewise.
8736 (emit_library_call_value_1): Likewise.
8737 (store_one_arg): Likewise.
8738 * combine.c (combine_instructions): Likewise.
8739 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
8740 * config/arm/arm.c (arm_function_value): Likewise.
8741 (aapcs_allocate_return_reg): Likewise.
8742 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
8743 * config/i386/i386.c (construct_container): Likewise.
8744 (ix86_gimplify_va_arg): Likewise.
8745 (ix86_expand_sse_cmp): Likewise.
8746 (emit_memmov): Likewise.
8747 (emit_memset): Likewise.
8748 (expand_small_movmem_or_setmem): Likewise.
8749 (ix86_expand_pextr): Likewise.
8750 (ix86_expand_pinsr): Likewise.
8751 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
8752 * config/microblaze/microblaze.c (microblaze_block_move_straight):
8754 * config/mips/mips.c (mips_function_value_1) Likewise.
8755 (mips_block_move_straight): Likewise.
8756 (mips_expand_ins_as_unaligned_store): Likewise.
8757 * config/powerpcspe/powerpcspe.c
8758 (rs6000_darwin64_record_arg_advance_flush): Likewise.
8759 (rs6000_darwin64_record_arg_flush): Likewise.
8760 * config/rs6000/rs6000.c
8761 (rs6000_darwin64_record_arg_advance_flush): Likewise.
8762 (rs6000_darwin64_record_arg_flush): Likewise.
8763 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
8764 (sparc_function_value_1): Likewise.
8765 * config/spu/spu.c (adjust_operand): Likewise.
8766 (spu_emit_branch_or_set): Likewise.
8767 (arith_immediate_p): Likewise.
8768 * emit-rtl.c (gen_lowpart_common): Likewise.
8769 * expr.c (expand_expr_real_1): Likewise.
8770 * function.c (assign_parm_setup_block): Likewise.
8771 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
8772 * reload1.c (alter_reg): Likewise.
8773 * stor-layout.c (mode_for_vector): Likewise.
8774 (layout_type): Likewise.
8776 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8778 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
8779 (spu_convert_move): Likewise.
8780 * lower-subreg.c (resolve_simple_move): Likewise.
8782 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8785 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
8786 define_insn to a define_expand.
8787 (altivec_vsum2sws_direct): New define_insn.
8788 (altivec_vsumsws): Convert from a define_insn to a define_expand.
8790 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
8792 * config/arm/arm.c (arm_option_params_internal): Improve setting of
8795 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
8800 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
8801 realign stack if stack alignment needed is less than incoming
8804 2017-09-05 Marek Polacek <polacek@redhat.com>
8807 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
8810 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
8812 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
8814 2017-09-05 Richard Biener <rguenther@suse.de>
8816 PR tree-optimization/82084
8817 * fold-const.c (can_native_encode_string_p): Handle wide characters.
8819 2017-09-05 Richard Biener <rguenther@suse.de>
8821 PR tree-optimization/82102
8822 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
8824 2017-09-05 Martin Liska <mliska@suse.cz>
8826 PR tree-optimization/82032
8827 * tree-cfg.c (generate_range_test): New function.
8828 * tree-cfg.h (generate_range_test): Declared here.
8829 * tree-cfgcleanup.c (convert_single_case_switch): New function.
8830 (cleanup_control_expr_graph): Use it.
8831 * tree-switch-conversion.c (try_switch_expansion): Remove
8833 (emit_case_nodes): Use generate_range_test.
8835 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
8838 * config/i386/i386.md (*<btsc><mode>_mask): Add
8839 TARGET_USE_BT to insn constraint.
8840 (*btr<mode>_mask): Ditto.
8842 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
8844 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
8845 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
8847 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
8850 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
8851 TARGET_NEON and TARGET_IWMMXT.
8852 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
8853 TARGET_NEON and TARGET_IWMMXT.
8854 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
8856 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
8858 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
8859 (ix86_rewrite_tls_address): Ditto.
8860 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
8861 (ix86_rewrite_tls_address_1): Ditto.
8862 (ix86_rewrite_tls_address): Ditto.
8863 * config/i386/predicates.md (tls_address_pattern): New predicate.
8864 * config/i386/i386.md (TLS address splitter): New splitter.
8866 2017-09-04 Richard Biener <rguenther@suse.de>
8868 PR tree-optimization/82084
8869 * fold-const.h (can_native_encode_string_p): Declare.
8870 * fold-const.c (can_native_encode_string_p): Factor out from ...
8871 (native_encode_string): ... here.
8872 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
8873 vectorizing stores from constants we later cannot handle.
8875 2017-09-04 Marek Polacek <polacek@redhat.com>
8878 * doc/invoke.texi: Update -Wtautological-compare documentation.
8880 2017-09-04 Jeff Law <law@redhat.com>
8882 PR tree-optimization/64910
8883 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
8884 swap the first and last operand if the last is a constant.
8886 2017-09-04 Marek Polacek <polacek@redhat.com>
8889 * convert.c (do_narrow): When sanitizing signed integer overflows,
8890 bail out for signed types.
8891 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
8893 2017-09-04 Richard Biener <rguenther@suse.de>
8895 PR tree-optimization/82060
8896 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
8897 Move devirtualization after stmt folding and before EH/AB/noreturn
8898 cleanup to get the stmt refs canonicalized. Use a bool instead
8899 of gimple_modified_p since that doesn't work for NOPs. Schedule
8900 NOPs generated by folding for removal.
8902 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8903 Alan Hayward <alan.hayward@arm.com>
8904 David Sherwood <david.sherwood@arm.com>
8906 * coretypes.h (pad_direction): New enum.
8907 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
8908 (FUNCTION_ARG_PADDING): Likewise.
8909 * target.def (function_arg_padding): New hook.
8910 * targhooks.h (default_function_arg_padding): Declare.
8911 * targhooks.c (default_function_arg_padding): New function.
8912 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
8913 (TARGET_FUNCTION_ARG_PADDING): ...this.
8914 * doc/tm.texi: Regenerate.
8915 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
8916 instead of direction.
8917 (compute_argument_addresses): Likewise.
8918 (load_register_parameters): Likewise.
8919 (emit_library_call_value_1): Likewise.
8920 (store_one_arg): Use targetm.calls.function_arg_padding instead
8921 of FUNCTION_ARG_PADDING.
8922 (must_pass_in_stack_var_size_or_pad): Likewise.
8923 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
8924 (emit_group_store): Likewise.
8925 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
8926 instead of FUNCTION_ARG_PADDING.
8927 (emit_push_insn): Likewise, and propagate enum change throughout
8929 * function.h (direction): Delete.
8930 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
8932 * function.c (assign_parm_find_stack_rtl): Likewise.
8933 (assign_parm_setup_block_p): Likewise.
8934 (assign_parm_setup_block): Likewise.
8935 (gimplify_parameters): Likewise.
8936 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
8937 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
8939 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
8940 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
8941 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
8942 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
8943 (aarch64_function_arg_padding): ...this new function.
8944 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
8945 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8946 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
8947 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
8948 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
8949 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8950 (arm_pad_arg_upward): Replace with...
8951 (arm_function_arg_padding): ...this new function.
8952 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
8954 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
8955 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
8956 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8957 (ia64_hpux_function_arg_padding): Replace with...
8958 (ia64_function_arg_padding): ...this new function. Use pad_direction
8959 instead of direction. Check for TARGET_HPUX.
8960 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
8961 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8962 (iq2000_function_arg_padding): New function.
8963 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
8964 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
8965 (mips_function_arg_padding): ...this new function.
8966 (mips_pad_reg_upward): Update accordingly.
8967 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8968 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
8969 targetm.calls.function_arg_padding.
8970 (FUNCTION_ARG_PADDING): Delete.
8971 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
8972 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
8973 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
8974 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
8975 (nios2_block_reg_padding): Return pad_direction instead of direction.
8976 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
8977 instead of direction.
8978 (nios2_function_arg_padding): Likewise. Make static.
8979 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8980 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
8981 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
8982 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
8983 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8984 (pa_function_arg_padding): Make static. Return pad_direction instead
8986 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
8987 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
8988 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
8989 instead of direction. Use targetm.calls.function_arg_padding.
8990 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
8991 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
8992 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
8993 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
8994 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
8996 (function_arg_padding): Rename to...
8997 (rs6000_function_arg_padding): ...this. Make static. Return
8998 pad_direction instead of direction.
8999 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
9000 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
9001 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9002 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
9003 instead of direction. Use targetm.calls.function_arg_padding.
9004 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
9005 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
9006 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
9007 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
9008 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9009 (function_arg_padding): Rename to...
9010 (rs6000_function_arg_padding): ...this. Make static. Return
9011 pad_direction instead of direction.
9012 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
9013 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
9014 * config/s390/s390.c (s390_function_arg_padding): New function.
9015 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9016 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
9017 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
9018 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9019 (function_arg_padding): Rename to...
9020 (sparc_function_arg_padding): ...this. Make static. Return
9021 pad_direction instead of direction.
9022 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
9023 * config/spu/spu.c (spu_function_arg_padding): New function.
9024 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9025 * system.h (FUNCTION_ARG_PADDING): Poison.
9027 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9028 Alan Hayward <alan.hayward@arm.com>
9029 David Sherwood <david.sherwood@arm.com>
9031 * target.def (modes_tieable_p): New hook.
9032 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
9033 (TARGET_MODES_TIEABLE_P): ...this.
9034 * doc/tm.texi.in: Regenerate.
9035 * hooks.h (hook_bool_mode_mode_true): Declare.
9036 * hooks.c (hook_bool_mode_mode_true): New function.
9037 * combine.c (subst): Use targetm.modes_tieable_p instead of
9039 * dse.c (find_shift_sequence): Likewise.
9040 * expmed.c (extract_low_bits): Likewise.
9041 * lower-subreg.c: Include target.h.
9042 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
9044 * rtlanal.c (rtx_cost): Likewise.
9045 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
9046 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
9047 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
9048 (TARGET_MODES_TIEABLE_P): Redefine.
9049 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
9050 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
9051 (TARGET_MODES_TIEABLE_P): Redefine.
9052 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
9053 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
9054 (arc_modes_tieable_p): New function.
9055 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
9056 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
9057 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
9058 (arm_modes_tieable_p): Make static.
9059 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
9060 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
9061 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
9062 (TARGET_MODES_TIEABLE_P): Redefine.
9063 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
9064 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
9065 (TARGET_MODES_TIEABLE_P): Redefine.
9066 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
9067 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
9068 (cr16_modes_tieable_p): New function.
9069 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
9070 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
9071 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
9072 (TRULY_NOOP_TRUNCATION): Update comment.
9073 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
9074 (TRULY_NOOP_TRUNCATION): Update comment.
9075 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
9076 (frv_modes_tieable_p): New function.
9077 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
9078 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
9079 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
9080 (TARGET_MODES_TIEABLE_P): Redefine.
9081 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
9082 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
9083 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
9084 (TARGET_MODES_TIEABLE_P): Redefine.
9085 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
9086 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
9087 (ia64_modes_tieable_p): New function.
9088 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
9089 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
9090 (iq2000_modes_tieable_p): New function.
9091 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
9092 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
9093 (lm32_modes_tieable_p): New function.
9094 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
9095 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
9096 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
9097 (TARGET_MODES_TIEABLE_P): Redefine.
9098 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
9099 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
9100 (m32r_modes_tieable_p): New function.
9101 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
9102 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
9103 (m68k_modes_tieable_p): New function.
9104 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
9105 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
9106 (mcore_modes_tieable_p): New function.
9107 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
9108 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
9110 (TARGET_MODES_TIEABLE_P): Redefine.
9111 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
9112 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
9113 * config/mips/mips.c (mips_modes_tieable_p): Make static.
9114 (TARGET_MODES_TIEABLE_P): Redefine.
9115 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
9116 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
9117 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
9118 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
9119 (mn10300_modes_tieable_p): ...this and make static.
9120 (TARGET_MODES_TIEABLE_P): Redefine.
9121 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
9122 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
9123 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
9124 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
9125 (msp430_modes_tieable_p): Make static.
9126 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
9127 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
9128 (TARGET_MODES_TIEABLE_P): Redefine.
9129 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
9130 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
9131 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
9132 (TARGET_MODES_TIEABLE_P): Redefine.
9133 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
9134 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
9135 * config/pa/pa.c (pa_modes_tieable_p): Make static.
9136 (TARGET_MODES_TIEABLE_P): Redefine.
9137 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
9138 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
9139 (pdp11_modes_tieable_p): New function.
9140 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
9141 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
9142 (rs6000_modes_tieable_p): New function.
9143 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
9144 * config/powerpcspe/powerpcspe.md: Update comment.
9145 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
9146 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
9147 (TARGET_MODES_TIEABLE_P): Redefine.
9148 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
9149 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
9150 (rl78_modes_tieable_p): New function.
9151 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
9152 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
9153 (rs6000_modes_tieable_p): New function.
9154 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
9155 * config/rs6000/rs6000.md: Update comment.
9156 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
9157 * config/rx/rx.c (rx_modes_tieable_p): New function.
9158 (TARGET_MODES_TIEABLE_P): Redefine.
9159 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
9160 * config/s390/s390.c (s390_modes_tieable_p): New function.
9161 (TARGET_MODES_TIEABLE_P): Redefine.
9162 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
9163 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
9164 (sh_modes_tieable_p): New function.
9165 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
9166 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
9167 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
9168 (sparc_modes_tieable_p): Make static.
9169 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
9170 * config/spu/spu.c (spu_modes_tieable_p): New function.
9171 (TARGET_MODES_TIEABLE_P): Redefine.
9172 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
9173 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
9174 (TARGET_MODES_TIEABLE_P): Redefine.
9175 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
9176 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
9177 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
9178 * config/v850/v850.c (v850_modes_tieable_p): New function.
9179 (TARGET_MODES_TIEABLE_P): Redefine.
9180 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
9181 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
9182 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
9183 (visium_modes_tieable_p): New function.
9184 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
9185 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
9186 (xtensa_modes_tieable_p): New function.
9187 * system.h (MODES_TIEABLE_P): Poison.
9189 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9190 Alan Hayward <alan.hayward@arm.com>
9191 David Sherwood <david.sherwood@arm.com>
9193 * target.def (hard_regno_mode_ok): New hook.
9194 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
9195 (TARGET_HARD_REGNO_MODE_OK): ...this.
9196 * doc/tm.texi.in: Regenerate.
9197 * hooks.h (hook_bool_uint_mode_true): Declare.
9198 * hooks.c (hook_bool_uint_mode_true): New function.
9199 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
9201 * genpreds.c (write_insn_preds_c): Add an include of target.h.
9202 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
9203 instead of HARD_REGNO_MODE_OK.
9204 * caller-save.c: Include target.h.
9205 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
9207 * combine.c (can_combine_p): Likewise.
9208 (combinable_i3pat): Likewise.
9209 (can_change_dest_mode): Likewise.
9210 * expr.c (init_expr_target): Likewise.
9211 (convert_move): Likewise.
9212 (convert_modes): Likewise.
9213 * ira.c (setup_prohibited_class_mode_regs): Likewise.
9214 (setup_prohibited_mode_move_regs): Likewise.
9215 * ira.h (target_ira): Likewise.
9216 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9217 * lra-constraints.c (process_alt_operands): Likewise.
9218 (split_reg): Likewise.
9219 * recog.c (peep2_find_free_register): Likewise.
9220 * ree.c (combine_reaching_defs): Likewise.
9221 * regcprop.c (maybe_mode_change): Likewise.
9222 * reginfo.c (init_reg_sets_1): Likewise.
9223 (choose_hard_reg_mode): Likewise.
9224 (simplifiable_subregs): Likewise.
9225 * regrename.c (check_new_reg_p): Likewise.
9226 * reload.c (find_valid_class): Likewise.
9227 (find_valid_class_1): Likewise.
9228 (reload_inner_reg_of_subreg): Likewise.
9229 (push_reload): Likewise.
9230 (combine_reloads): Likewise.
9231 (find_dummy_reload): Likewise.
9232 (find_reloads): Likewise.
9233 * reload1.c (find_reg): Likewise.
9234 (set_reload_reg): Likewise.
9235 (allocate_reload_reg): Likewise.
9236 (choose_reload_regs): Likewise.
9237 (reload_adjust_reg_for_temp): Likewise.
9238 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
9239 (simplify_subreg_regno): Likewise.
9240 * sel-sched.c (init_regs_for_mode): Likewise.
9241 * varasm.c (make_decl_rtl): Likewise.
9242 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
9243 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
9245 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
9246 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
9247 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9248 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
9249 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
9250 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9251 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
9252 (arc_mode_class): Delete.
9253 (HARD_REGNO_MODE_OK): Delete.
9254 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9255 (arc_hard_regno_mode_ok): Rename old array to...
9256 (arc_hard_regno_mode_ok_modes): ...this.
9257 (arc_conditional_register_usage): Update accordingly.
9258 (arc_mode_class): Make static.
9259 (arc_hard_regno_mode_ok): New function.
9260 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
9261 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
9262 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9263 (arm_hard_regno_mode_ok): Make static.
9264 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
9266 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
9267 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
9268 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
9270 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9271 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
9272 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
9273 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
9274 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
9275 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9276 * config/bfin/predicates.md (valid_reg_operand): Use
9277 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
9278 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
9279 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
9280 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9281 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
9282 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
9283 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9284 (cr16_hard_regno_mode_ok): Make static and return a bool.
9285 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
9286 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9287 (cris_hard_regno_mode_ok): New function.
9288 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
9289 (epiphany_mode_class): Delete.
9290 (HARD_REGNO_MODE_OK): Delete.
9291 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
9292 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9293 (hard_regno_mode_ok): Rename to...
9294 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
9295 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
9296 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
9298 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
9299 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
9300 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9301 (frv_hard_regno_mode_ok): Make static and return a bool.
9302 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
9304 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
9305 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
9306 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
9307 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
9309 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9310 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
9311 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
9312 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
9314 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9315 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
9316 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9317 (ia64_hard_regno_mode_ok): New function.
9318 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
9319 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9320 (iq2000_hard_regno_mode_ok): New function.
9321 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
9322 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9323 (lm32_hard_regno_mode_ok): New function.
9324 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
9325 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
9326 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
9327 instead of HARD_REGNO_MODE_OK.
9328 (m32c_hard_regno_ok): Rename to...
9329 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
9330 (m32c_cannot_change_mode_class): Update accordingly.
9331 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9332 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
9333 (m32r_mode_class): Delete.
9334 (HARD_REGNO_MODE_OK): Delete.
9335 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9336 (m32r_hard_regno_mode_ok): Rename to...
9337 (m32r_hard_regno_modes): ...this.
9338 (m32r_mode_class): Make static.
9339 (m32r_hard_regno_mode_ok): New function.
9340 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
9341 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
9342 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9343 (m68k_hard_regno_mode_ok): Make static.
9344 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
9345 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9346 (mcore_hard_regno_mode_ok): New function.
9347 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
9348 (HARD_REGNO_MODE_OK): Delete.
9349 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
9351 (microblaze_hard_regno_mode_ok_p): ...this and make static.
9352 (microblaze_hard_regno_mode_ok): New function.
9353 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9354 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
9355 (mips_hard_regno_mode_ok): Delete.
9356 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
9357 (mips_hard_regno_mode_ok_p): ...this and make static.
9358 (mips_hard_regno_mode_ok_p): Rename to...
9359 (mips_hard_regno_mode_ok_uncached): ...this.
9360 (mips_hard_regno_mode_ok): New function.
9361 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
9362 of HARD_REGNO_MODE_OK.
9363 (mips_option_override): Update after above name changes.
9364 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9365 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
9366 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
9367 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
9368 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
9369 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9370 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
9371 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
9372 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
9373 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9374 (msp430_hard_regno_mode_ok): Make static and return a bool.
9375 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
9376 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
9377 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
9379 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9380 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
9381 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
9382 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
9383 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
9384 (PA_HARD_REGNO_MODE_OK): ...this
9385 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
9386 (PA_HARD_REGNO_MODE_OK): ...this.
9387 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9388 (pa_hard_regno_mode_ok): New function.
9389 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
9390 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9391 (pdp11_hard_regno_mode_ok): New function.
9392 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
9393 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
9395 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
9397 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9398 (rs6000_hard_regno_mode_ok): Rename to...
9399 (rs6000_hard_regno_mode_ok_uncached): ...this.
9400 (rs6000_init_hard_regno_mode_ok): Update accordingly.
9401 (rs6000_hard_regno_mode_ok): New function.
9402 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
9403 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
9404 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
9405 (riscv_hard_regno_mode_ok): ...this and make static.
9406 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9407 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
9408 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
9409 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9410 (rl78_hard_regno_mode_ok): Make static and return bool.
9411 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
9412 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
9414 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
9415 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9416 (rs6000_hard_regno_mode_ok): Rename to...
9417 (rs6000_hard_regno_mode_ok_uncached): ...this.
9418 (rs6000_init_hard_regno_mode_ok): Update accordingly.
9419 (rs6000_hard_regno_mode_ok): New function.
9420 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
9421 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
9422 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9423 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
9424 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
9425 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
9426 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9427 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
9428 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
9429 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9430 (sh_hard_regno_mode_ok): Make static.
9431 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
9432 instead of HARD_REGNO_MODE_OK.
9433 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
9434 (sparc_mode_class): Delete.
9435 (HARD_REGNO_MODE_OK): Delete.
9436 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9437 (hard_regno_mode_classes): Make static.
9438 (sparc_mode_class): Likewise.
9439 (sparc_hard_regno_mode_ok): New function.
9440 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
9441 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
9442 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
9444 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9445 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
9446 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
9447 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
9448 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
9449 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9450 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
9451 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
9452 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9453 (visium_hard_regno_mode_ok): New function.
9454 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
9455 instead of HARD_REGNO_MODE_OK.
9456 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
9457 (HARD_REGNO_MODE_OK): Delete.
9458 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
9459 (xtensa_hard_regno_mode_ok_p): ...this and make static.
9460 (xtensa_option_override): Update accordingly.
9461 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9462 (xtensa_hard_regno_mode_ok): New function.
9463 * system.h (HARD_REGNO_MODE_OK): Poison.
9465 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9466 Alan Hayward <alan.hayward@arm.com>
9467 David Sherwood <david.sherwood@arm.com>
9469 * target.def (hard_regno_call_part_clobbered): New hook.
9470 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
9471 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
9472 * doc/tm.texi: Regenerate.
9473 * hooks.h (hook_bool_uint_mode_false): Declare.
9474 * hooks.c (hook_bool_uint_mode_false): New function.
9475 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9476 * cselib.c (cselib_process_insn): Use
9477 targetm.hard_regno_call_part_clobbered instead of
9478 HARD_REGNO_CALL_PART_CLOBBERED.
9479 * ira-conflicts.c (ira_build_conflicts): Likewise.
9480 * ira-costs.c (ira_tune_allocno_costs): Likewise.
9481 * lra-constraints.c (need_for_call_save_p): Likewise.
9482 * lra-lives.c: Include target.h.
9483 (check_pseudos_live_through_calls): Use
9484 targetm.hard_regno_call_part_clobbered instead of
9485 HARD_REGNO_CALL_PART_CLOBBERED.
9486 * regcprop.c: Include target.h.
9487 (copyprop_hardreg_forward_1): Use
9488 targetm.hard_regno_call_part_clobbered instead of
9489 HARD_REGNO_CALL_PART_CLOBBERED.
9490 * reginfo.c (choose_hard_reg_mode): Likewise.
9491 * regrename.c (check_new_reg_p): Likewise.
9492 * reload.c (find_equiv_reg): Likewise.
9493 * reload1.c (emit_reload_insns): Likewise.
9494 * sched-deps.c (deps_analyze_insn): Likewise.
9495 * sel-sched.c (init_regs_for_mode): Likewise.
9496 (mark_unavailable_hard_regs): Likewise.
9497 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
9498 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9499 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
9501 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9502 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9503 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
9505 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
9507 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9508 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9509 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
9511 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9512 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9513 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
9515 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9516 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
9518 * config/powerpcspe/powerpcspe.c
9519 (rs6000_hard_regno_call_part_clobbered): New function.
9520 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9521 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9522 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
9524 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9525 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9526 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
9528 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9529 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9530 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
9532 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9533 Alan Hayward <alan.hayward@arm.com>
9534 David Sherwood <david.sherwood@arm.com>
9536 * rtl.h (subreg_memory_offset): Declare.
9537 * emit-rtl.c (subreg_memory_offset): New function.
9538 * expmed.c (store_bit_field_1): Use it.
9539 * expr.c (undefined_operand_subword_p): Likewise.
9540 * simplify-rtx.c (simplify_subreg): Likewise.
9542 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
9544 PR rtl-optimization/57448
9547 * optabs.c (expand_atomic_load): Place compiler memory barriers if
9548 using atomic_load pattern.
9549 (expand_atomic_store): Likewise.
9551 2017-09-04 Jakub Jelinek <jakub@redhat.com>
9554 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
9555 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
9556 handling. Use replace_call_with_value with NULL instead of
9557 gsi_replace, unlink_stmt_vdef and release_defs.
9559 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
9562 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
9564 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9567 * rtl.h (emit_library_call_value_1): Declare.
9568 (emit_library_call): Replace declaration with a series of overloads.
9569 Remove the parameter count argument.
9570 (emit_library_call_value): Likewise.
9571 * calls.c (emit_library_call_value_1): Make global. Replace varargs
9572 with an "rtx_mode_t *".
9573 (emit_library_call_value): Delete.
9574 (emit_library_call): Likewise.
9575 * asan.c (asan_emit_stack_protection): Update calls accordingly.
9576 (asan_emit_allocas_unpoison): Likewise.
9577 * builtins.c (expand_builtin_powi): Likewise.
9578 (expand_asan_emit_allocas_unpoison): Likewise.
9579 * cfgexpand.c (expand_main_function): Likewise.
9580 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
9581 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
9582 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
9583 * config/arm/arm.c (arm_trampoline_init): Likewise.
9584 (arm_call_tls_get_addr): Likewise.
9585 (arm_expand_divmod_libfunc): Likewise.
9586 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
9587 (smulsi3_highpart): Likewise.
9588 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
9589 (c6x_expand_compare): Likewise.
9590 (c6x_expand_movmem): Likewise.
9591 * config/frv/frv.c (frv_trampoline_init): Likewise.
9592 * config/i386/i386.c (ix86_trampoline_init): Likewise.
9593 (ix86_expand_divmod_libfunc): Likewise.
9594 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
9595 (ia64_expand_compare): Likewise.
9596 (ia64_profile_hook): Likewise.
9597 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
9598 (nonlocal_goto): Likewise.
9599 (restore_stack_nonlocal): Likewise.
9600 * config/m32r/m32r.c (block_move_call): Likewise.
9601 (m32r_trampoline_init): Likewise.
9602 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
9603 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
9604 (m68k_call_m68k_read_tp): Likewise.
9605 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
9606 (microblaze_expand_divide): Likewise.
9607 * config/mips/mips.h (mips_args): Likewise.
9608 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
9609 (MIPS_ICACHE_SYNC): Likewise.
9610 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
9611 (nios2_trampoline_init): Likewise.
9612 * config/pa/pa.c (hppa_tls_call): Likewise.
9613 (pa_trampoline_init): Likewise.
9614 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
9615 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
9616 (expand_strn_compare): Likewise.
9617 (rs6000_generate_compare): Likewise.
9618 (rs6000_expand_float128_convert): Likewise.
9619 (output_profile_hook): Likewise.
9620 (rs6000_trampoline_init): Likewise.
9621 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
9622 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
9623 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
9624 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
9625 (rs6000_generate_compare): Likewise.
9626 (rs6000_expand_float128_convert): Likewise.
9627 (output_profile_hook): Likewise.
9628 (rs6000_trampoline_init): Likewise.
9629 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
9630 * config/sh/sh.c (sh_trampoline_init): Likewise.
9631 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
9632 (sparc_emit_float_lib_cmp): Likewise.
9633 (sparc32_initialize_trampoline): Likewise.
9634 (sparc64_initialize_trampoline): Likewise.
9635 (sparc_profile_hook): Likewise.
9636 * config/spu/spu.c (ea_load_store): Likewise.
9637 * config/spu/spu.md (floatunssidf2): Likewise.
9638 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
9639 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
9640 * config/visium/visium.c (expand_block_move_4): Likewise.
9641 (expand_block_move_2): Likewise.
9642 (expand_block_move_1): Likewise.
9643 (expand_block_set_4): Likewise.
9644 (expand_block_set_2): Likewise.
9645 (expand_block_set_1): Likewise.
9646 (visium_trampoline_init): Likewise.
9647 (visium_profile_hook): Likewise.
9648 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
9649 (xtensa_setup_frame_addresses): Likewise.
9650 (xtensa_trampoline_init): Likewise.
9651 * except.c (sjlj_emit_function_enter): Likewise.
9652 (sjlj_emit_function_exit): Likewise.
9653 * explow.c (allocate_dynamic_stack_space): Likewise.
9654 (probe_stack_range): Likewise.
9655 * expr.c (convert_mode_scalar): Likewise.
9656 * optabs.c (expand_binop): Likewise.
9657 (expand_twoval_binop_libfunc): Likewise.
9658 (expand_unop): Likewise.
9659 (prepare_cmp_insn): Likewise.
9660 (prepare_float_lib_cmp): Likewise.
9661 (expand_float): Likewise.
9662 (expand_fix): Likewise.
9663 (expand_fixed_convert): Likewise.
9664 (maybe_emit_sync_lock_test_and_set): Likewise.
9665 (expand_atomic_compare_and_swap): Likewise.
9666 (expand_mem_thread_fence): Likewise.
9667 (expand_atomic_fetch_op): Likewise.
9669 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
9671 * doc/generic.texi (OpenACC): Adjust URL.
9672 * doc/invoke.texi (C Dialect Options): Ditto.
9674 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
9676 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
9677 predicate for operand 1. Add (m,<S>) constraint.
9678 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
9679 Prevent memory operand 1 with register operand 2.
9681 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
9683 PR rtl-optimization/82024
9684 * combine.c (try_combine): If the combination result is a PARALLEL,
9685 and we only need to retain the SET in there that would be placed
9686 at I2, check that we can place that at I3 instead, before doing so.
9688 2017-09-01 Jakub Jelinek <jakub@redhat.com>
9691 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
9693 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
9694 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
9697 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
9698 Jeff Law <law@redhat.com>
9700 * varasm.c (bss_initializer_p): Do not put constants into .bss
9701 (categorize_decl_for_section): Handle bss_initializer_p returning
9702 false when DECL_INITIAL is NULL.
9704 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9707 * config/s390/s390.c (s390_can_inline_p): New function.
9709 2017-09-01 Jeff Law <law@redhat.com>
9711 PR tree-optimization/82052
9712 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
9713 Always initialize the returned slot after a hash table miss
9714 when INSERT is true.
9716 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
9718 * config/s390/s390.md (mem_signal_fence): Remove.
9719 * doc/md.texi (mem_signal_fence): Remove.
9720 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
9722 * target-insns.def (mem_signal_fence): Remove.
9724 2017-09-01 Jakub Jelinek <jakub@redhat.com>
9727 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
9730 * asan.c (create_odr_indicator): Strip name encoding from assembler
9731 name before appending it after __odr_asan_.
9733 2017-09-01 Martin Liska <mliska@suse.cz>
9735 PR tree-optimization/82059
9736 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
9737 frequency only when an edge is redirected.
9739 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9741 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
9742 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
9743 (arc_conditional_register_usage): Remove ARC600 lp_count
9745 (arc_file_start): Emit Tag_ARC_CPU_variation.
9746 (arc_can_use_doloop_p): New conditions to use ZOLs.
9747 (hwloop_fail): New function.
9748 (hwloop_optimize): Likewise.
9749 (hwloop_pattern_reg): Likewise.
9750 (arc_doloop_hooks): New struct, to be used with reorg_loops.
9751 (arc_reorg_loops): New function, calls reorg_loops.
9752 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
9753 (arc600_corereg_hazard): Remove ZOL checking, case handled by
9755 (arc_loop_hazard): Remove function, functionality moved into
9757 (arc_hazard): Remove arc_loop_hazard call.
9758 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
9759 into hwloop_optimize.
9760 (arc_label_align): Remove ZOL handling.
9761 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
9762 * config/arc/arc.md (doloop_begin): Remove pattern.
9763 (doloop_begin_i): Likewise.
9764 (doloop_end_i): Likewise.
9765 (doloop_fallback): Likewise.
9766 (doloop_fallback_m): Likewise.
9767 (doloop_end): Reimplement expand.
9768 (arc_lp): New pattern for LP instruction.
9769 (loop_end): New pattern.
9770 (loop_fail): Likewise.
9771 (decrement_and_branch_until_zero): Likewise.
9772 * config/arc/arc.opt (mlpc-width): New option.
9773 * doc/invoke.texi (mlpc-width): Document option.
9775 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9777 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
9778 (arc_ccfsm_advance): Fix checking for delay slots.
9779 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
9781 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9783 * config/arc/arc.md (movqi_insn): Add stores to save constant long
9785 (movhi_insn): Update store instruction constraint which are saving
9786 6-bit short immediates.
9787 (movsi_insn): Consider also short scaled load operations.
9788 (zero_extendhisi2_i): Use Usd constraint instead of T.
9789 (extendhisi2_i): Add q constraint.
9790 (arc_clzsi2): Add type and length attributes.
9791 (arc_ctzsi2): Likewise.
9792 * config/arc/constraints.md (Usc): Update constraint, the
9793 assembler can parse two relocations for a single instruction.
9795 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9797 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
9798 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
9800 2017-08-31 Olivier Hainque <hainque@adacore.com>
9802 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
9803 match as powerpc-wrs-vxworks*.
9805 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
9807 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
9808 register constraint for by-element operand.
9809 (aarch64_mls_elt_merge<mode>): Likewise.
9811 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9813 * config/arc/arc.c (arc_can_follow_jump): Check for short
9816 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9818 * config.gcc: Use g.opt for arc.
9819 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
9820 functionality moved to ...
9821 (legitimate_scaled_address_p): New function, ...here.
9822 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
9823 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
9824 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
9826 (arc_override_options): Handle G option.
9827 (arc_output_pic_addr_const): Correct function definition.
9828 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
9829 (arc_decl_anon_ns_mem_p): Delete.
9830 (arc_in_small_data_p): Overhaul this function to take into
9831 consideration the value given via G option.
9832 (arc_rewrite_small_data_1): Renamed and corrected old
9833 arc_rewrite_small_data function.
9834 (arc_rewrite_small_data): New function.
9835 (small_data_pattern): Don't use pic_offset_table_rtx.
9836 * config/arc/arc.h (CC1_SPEC): Recognize G option.
9837 * config/arc/simdext.md (movmisalignv2hi): Use
9838 prepare_move_operands function.
9840 (movmisalign*): Likewise.
9841 * doc/invoke.texi (ARC options): Document -G option.
9843 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9845 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
9847 * config/arc/arc.c (arc_print_operand): Output scalled address for
9848 sdata whenever is possible.
9849 (arc_in_small_data_p): Allow sdata for 64bit datum when double
9850 load/stores are available.
9851 (compact_sda_memory_operand): Check for the alignment required by
9852 code density instructions.
9853 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
9855 * config/arc/constraints.md (Usd): Update constraint.
9856 (Us0): New constraint.
9857 (Usc): Update constraint.
9859 2017-08-31 Richard Biener <rguenther@suse.de>
9862 * dwarf2out.c (dwarf2out_early_global_decl): Process each
9865 2017-08-31 Tamar Christina <tamar.christina@arm.com>
9867 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
9868 Resize type_signature.
9870 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
9871 Alan Hayward <alan.hayward@arm.com>
9872 David Sherwood <david.sherwood@arm.com>
9874 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
9875 subregs whose inner modes can be stored in GPRs.
9876 (aarch64_classify_index): Likewise.
9878 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
9879 Alan Hayward <alan.hayward@arm.com>
9880 David Sherwood <david.sherwood@arm.com>
9882 * config/aarch64/iterators.md (V_cmp_result): Rename to...
9883 (V_INT_EQUIV): ...this.
9884 (v_cmp_result): Rename to...
9885 (v_int_equiv): ...this.
9886 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
9887 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
9888 (copysign<mode>3): Likewise.
9889 (aarch64_simd_bsl<mode>_internal): Likewise.
9890 (aarch64_simd_bsl<mode>): Likewise.
9891 (vec_cmp<mode><mode>): Likewise.
9892 (vcond<mode><mode>): Likewise.
9893 (vcond<v_cmp_mixed><mode>): Likewise.
9894 (vcondu<mode><v_cmp_mixed>): Likewise.
9895 (aarch64_cm<optab><mode>): Likewise.
9896 (aarch64_cmtst<mode>): Likewise.
9897 (aarch64_fac<optab><mode>): Likewise.
9898 (vec_perm_const<mode>): Likewise.
9899 (vcond_mask_<mode><v_cmp_result>): Rename to...
9900 (vcond_mask_<mode><v_int_equiv>): ...this.
9901 (vec_cmp<mode><v_cmp_result>): Rename to...
9902 (vec_cmp<mode><v_int_equiv>): ...this.
9904 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
9905 Alan Hayward <alan.hayward@arm.com>
9906 David Sherwood <david.sherwood@arm.com>
9908 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
9910 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
9911 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
9912 (UNSPEC_LD4_DREG): New unspecs.
9913 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
9914 (aarch64_ld2<mode>_dreg_be): Replace with...
9915 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
9917 (aarch64_ld3<mode>_dreg_le)
9918 (aarch64_ld3<mode>_dreg_be): Replace with...
9919 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
9921 (aarch64_ld4<mode>_dreg_le)
9922 (aarch64_ld4<mode>_dreg_be): Replace with...
9923 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
9926 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9928 PR tree-optimization/81987
9929 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
9930 insert an initializer in a location not dominated by the stride
9933 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
9935 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
9936 on the entire header of the finally block in the fallthru case.
9938 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
9940 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
9942 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
9944 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
9945 rs6000_emit_move_from_cr and call renamed function.
9946 (rs6000_emit_prologue): Call renamed functions.
9947 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
9948 movesi_from_cr, remove volatile CRs.
9950 2017-08-30 Jon Beniston <jon@beniston.com>
9951 Richard Biener <rguenther@suse.de>
9953 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
9954 of VECTOR_MODE_P check.
9955 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
9956 element vector types.
9958 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9960 * df.h (df_read_modify_subreg_p): Remove in favor of...
9961 * rtl.h (read_modify_subreg_p): ...this new function. Take a
9962 const_rtx instead of an rtx.
9963 * cprop.c (local_cprop_find_used_regs): Update accordingly.
9964 * df-problems.c (df_word_lr_mark_ref): Likewise.
9965 * ira-lives.c (mark_pseudo_reg_live): Likewise.
9966 (mark_pseudo_reg_dead): Likewise.
9967 (mark_ref_dead): Likewise.
9968 * reginfo.c (init_subregs_of_mode): Likewise.
9969 * sched-deps.c (sched_analyze_1): Likewise.
9970 * df-scan.c (df_def_record_1): Likewise.
9971 (df_uses_record): Likewise.
9972 (df_read_modify_subreg_p): Remove in favor of...
9973 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
9974 const_rtx instead of an rtx.
9976 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9977 Alan Hayward <alan.hayward@arm.com>
9978 David Sherwood <david.sherwood@arm.com>
9980 * rtl.h (partial_subreg_p): New function.
9981 * caller-save.c (save_call_clobbered_regs): Use it.
9982 * calls.c (expand_call): Likewise.
9983 * combine.c (combinable_i3pat): Likewise.
9984 (simplify_set): Likewise.
9985 (make_extraction): Likewise.
9986 (make_compound_operation_int): Likewise.
9987 (gen_lowpart_or_truncate): Likewise.
9988 (force_to_mode): Likewise.
9989 (make_field_assignment): Likewise.
9990 (reg_truncated_to_mode): Likewise.
9991 (record_truncated_value): Likewise.
9992 (move_deaths): Likewise.
9993 * cse.c (record_jump_cond): Likewise.
9994 (cse_insn): Likewise.
9995 * cselib.c (cselib_lookup_1): Likewise.
9996 * expmed.c (extract_bit_field_using_extv): Likewise.
9997 * function.c (assign_parm_setup_reg): Likewise.
9998 * ifcvt.c (noce_convert_multiple_sets): Likewise.
9999 * ira-build.c (create_insn_allocnos): Likewise.
10000 * lra-coalesce.c (merge_pseudos): Likewise.
10001 * lra-constraints.c (match_reload): Likewise.
10002 (simplify_operand_subreg): Likewise.
10003 (curr_insn_transform): Likewise.
10004 * lra-lives.c (process_bb_lives): Likewise.
10005 * lra.c (new_insn_reg): Likewise.
10006 (lra_substitute_pseudo): Likewise.
10007 * regcprop.c (mode_change_ok): Likewise.
10008 (maybe_mode_change): Likewise.
10009 (copyprop_hardreg_forward_1): Likewise.
10010 * reload.c (push_reload): Likewise.
10011 (find_reloads): Likewise.
10012 (find_reloads_subreg_address): Likewise.
10013 * reload1.c (alter_reg): Likewise.
10014 (eliminate_regs_1): Likewise.
10015 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10017 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
10019 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
10020 back to if statements, including unpack.
10022 2017-08-30 Martin Liska <mliska@suse.cz>
10024 PR inline-asm/82001
10025 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
10027 (func_checker::compare_asm_inputs_outputs): ... this function.
10028 (func_checker::compare_gimple_asm): Use the function to compare
10029 also ASM constrains.
10030 * ipa-icf-gimple.h: Rename the function.
10032 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10033 Alan Hayward <alan.hayward@arm.com>
10034 David Sherwood <david.sherwood@arm.com>
10036 * coretypes.h (complex_mode): New type.
10037 * gdbhooks.py (build_pretty_printer): Handle it.
10038 * machmode.h (complex_mode): New class.
10039 (complex_mode::includes_p): New function.
10040 (is_complex_int_mode): Likewise.
10041 (is_complex_float_mode): Likewise.
10042 * genmodes.c (get_mode_class): Handle complex mode classes.
10043 * function.c (expand_function_end): Use is_complex_int_mode.
10045 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10046 Alan Hayward <alan.hayward@arm.com>
10047 David Sherwood <david.sherwood@arm.com>
10049 * coretypes.h (scalar_mode_pod): New typedef.
10050 * gdbhooks.py (build_pretty_printer): Handle it.
10051 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
10052 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
10053 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
10054 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
10055 as_a <scalar_mode>.
10057 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10058 Alan Hayward <alan.hayward@arm.com>
10059 David Sherwood <david.sherwood@arm.com>
10061 * machmode.h (mode_for_vector): Take a scalar_mode instead
10063 * stor-layout.c (mode_for_vector): Likewise.
10064 * explow.c (promote_mode): Use as_a <scalar_mode>.
10065 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
10067 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10068 Alan Hayward <alan.hayward@arm.com>
10069 David Sherwood <david.sherwood@arm.com>
10071 * target.def (preferred_simd_mode): Take a scalar_mode
10072 instead of a machine_mode.
10073 * targhooks.h (default_preferred_simd_mode): Likewise.
10074 * targhooks.c (default_preferred_simd_mode): Likewise.
10075 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
10076 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
10077 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
10078 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
10079 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
10080 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
10081 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
10082 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
10084 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
10085 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
10086 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
10087 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
10088 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
10089 * doc/tm.texi: Regenerate.
10090 * optabs-query.c (can_vec_mask_load_store_p): Return false for
10093 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10094 Alan Hayward <alan.hayward@arm.com>
10095 David Sherwood <david.sherwood@arm.com>
10097 * target.def (scalar_mode_supported_p): Take a scalar_mode
10098 instead of a machine_mode.
10099 * targhooks.h (default_scalar_mode_supported_p): Likewise.
10100 * targhooks.c (default_scalar_mode_supported_p): Likewise.
10101 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
10102 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
10103 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
10104 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
10105 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
10106 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
10107 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
10108 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
10109 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
10110 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
10111 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
10112 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
10114 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
10115 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
10116 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
10117 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
10118 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
10120 * doc/tm.texi: Regenerate.
10122 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10123 Alan Hayward <alan.hayward@arm.com>
10124 David Sherwood <david.sherwood@arm.com>
10126 * coretypes.h (opt_scalar_mode): New typedef.
10127 * gdbhooks.py (build_pretty_printers): Handle it.
10128 * machmode.h (mode_iterator::get_2xwider): Add overload for
10130 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
10132 * expr.c (convert_mode_scalar): Likewise.
10133 * omp-low.c (omp_clause_aligned_alignment): Likewise.
10134 * optabs.c (expand_float): Likewise.
10135 (expand_fix): Likewise.
10136 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10138 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10139 Alan Hayward <alan.hayward@arm.com>
10140 David Sherwood <david.sherwood@arm.com>
10142 * optabs.c (expand_float): Explicitly check for scalars before
10143 using a branching expansion.
10144 (expand_fix): Likewise.
10146 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10147 Alan Hayward <alan.hayward@arm.com>
10148 David Sherwood <david.sherwood@arm.com>
10150 * expr.c (convert_mode): Split scalar handling out into...
10151 (convert_mode_scalar): ...this new function. Treat the modes
10154 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10155 Alan Hayward <alan.hayward@arm.com>
10156 David Sherwood <david.sherwood@arm.com>
10158 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
10160 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
10162 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10163 Alan Hayward <alan.hayward@arm.com>
10164 David Sherwood <david.sherwood@arm.com>
10166 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
10167 rather than a machine_mode.
10168 (fixed_from_string): Likewise.
10169 (fixed_convert): Likewise.
10170 (fixed_convert_from_int): Likewise.
10171 (fixed_convert_from_real): Likewise.
10172 (real_convert_from_fixed): Likewise.
10173 * fixed-value.c (fixed_from_double_int): Likewise.
10174 (fixed_from_string): Likewise.
10175 (fixed_convert): Likewise.
10176 (fixed_convert_from_int): Likewise.
10177 (fixed_convert_from_real): Likewise.
10178 (real_convert_from_fixed): Likewise.
10179 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
10181 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10182 Alan Hayward <alan.hayward@arm.com>
10183 David Sherwood <david.sherwood@arm.com>
10185 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
10186 of separate mode class checks. Do not allow vector modes here.
10187 (immed_wide_int_const): Use as_a <scalar_mode>.
10188 * explow.c (trunc_int_for_mode): Likewise.
10189 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
10190 (wi::shwi): Likewise.
10191 (wi::min_value): Likewise.
10192 (wi::max_value): Likewise.
10193 * dwarf2out.c (loc_descriptor): Likewise.
10194 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
10195 for CONST_WIDE_INT.
10197 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10198 Alan Hayward <alan.hayward@arm.com>
10199 David Sherwood <david.sherwood@arm.com>
10201 * tree.h (SCALAR_TYPE_MODE): New macro.
10202 * expr.c (expand_expr_addr_expr_1): Use it.
10203 (expand_expr_real_2): Likewise.
10204 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
10205 (fold_convert_const_fixed_from_int): Likewise.
10206 (fold_convert_const_fixed_from_real): Likewise.
10207 (native_encode_fixed): Likewise
10208 (native_encode_complex): Likewise
10209 (native_encode_vector): Likewise.
10210 (native_interpret_fixed): Likewise.
10211 (native_interpret_real): Likewise.
10212 (native_interpret_complex): Likewise.
10213 (native_interpret_vector): Likewise.
10214 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
10215 (simd_clone_adjust_argument_types): Likewise.
10216 (simd_clone_init_simd_arrays): Likewise.
10217 (simd_clone_adjust): Likewise.
10218 * stor-layout.c (layout_type): Likewise.
10219 * tree.c (build_minus_one_cst): Likewise.
10220 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
10221 * tree-inline.c (estimate_move_cost): Likewise.
10222 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
10223 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
10224 (vectorizable_reduction): Likewise.
10225 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
10226 (vect_recog_mixed_size_cond_pattern): Likewise.
10227 (check_bool_pattern): Likewise.
10228 (adjust_bool_pattern): Likewise.
10229 (search_type_for_mask_1): Likewise.
10230 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
10231 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10232 (vectorizable_load): Likewise.
10233 (vectorizable_store): Likewise.
10234 * ubsan.c (ubsan_encode_value): Likewise.
10235 * varasm.c (output_constant): Likewise.
10237 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10238 Alan Hayward <alan.hayward@arm.com>
10239 David Sherwood <david.sherwood@arm.com>
10241 * coretypes.h (scalar_mode): New class.
10242 * machmode.h (scalar_mode): Likewise.
10243 (scalar_mode::includes_p): New function.
10244 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
10245 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
10246 * genmodes.c (get_mode_class): Handle remaining scalar modes.
10247 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
10248 * expmed.c (store_bit_field_1): Likewise.
10249 (extract_bit_field_1): Likewise.
10250 * expr.c (write_complex_part): Likewise.
10251 (read_complex_part): Likewise.
10252 (emit_move_complex_push): Likewise.
10253 (expand_expr_real_2): Likewise.
10254 * function.c (assign_parm_setup_reg): Likewise.
10255 (assign_parms_unsplit_complex): Likewise.
10256 * optabs.c (expand_binop): Likewise.
10257 * rtlanal.c (subreg_get_info): Likewise.
10258 * simplify-rtx.c (simplify_immed_subreg): Likewise.
10259 * varasm.c (output_constant_pool_2): Likewise.
10261 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10262 Alan Hayward <alan.hayward@arm.com>
10263 David Sherwood <david.sherwood@arm.com>
10265 * expmed.c (extract_high_half): Use scalar_int_mode and remove
10267 (expmed_mult_highpart_optab): Likewise.
10268 (expmed_mult_highpart): Likewise.
10270 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10271 Alan Hayward <alan.hayward@arm.com>
10272 David Sherwood <david.sherwood@arm.com>
10274 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
10275 instead of a machine_mode.
10276 (builtin_memset_read_str): Likewise.
10277 * builtins.c (c_readstr): Likewise.
10278 (builtin_memcpy_read_str): Likewise.
10279 (builtin_strncpy_read_str): Likewise.
10280 (builtin_memset_read_str): Likewise.
10281 (builtin_memset_gen_str): Likewise.
10282 (expand_builtin_signbit): Use scalar_int_mode for local variables.
10283 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
10284 instead of a machine_mode.
10285 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
10287 (make_extraction): Likewise.
10288 (try_widen_shift_mode): Take and return scalar_int_modes instead
10290 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
10291 a scalar_int_mode instead of a machine_mode.
10292 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
10293 (avr_addr_space_pointer_mode): Likewise.
10294 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
10295 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
10296 (msp430_unwind_word_mode): Likewise.
10297 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
10298 (spu_addr_space_pointer_mode): Likewise.
10299 (spu_addr_space_address_mode): Likewise.
10300 (spu_libgcc_cmp_return_mode): Likewise.
10301 (spu_libgcc_shift_count_mode): Likewise.
10302 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
10303 (rl78_addr_space_pointer_mode): Likewise.
10304 (fl78_unwind_word_mode): Likewise.
10305 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
10307 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
10308 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
10309 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
10310 (mips_valid_pointer_mode): Likewise.
10311 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
10312 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
10313 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
10315 (ft32_addr_space_address_mode): Likewise.
10316 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
10317 scalar_int_mode instead of a machine_mode.
10318 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
10320 (m32c_addr_space_address_mode): Likewise.
10321 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
10322 (rs6000_eh_return_filter_mode): Likewise.
10323 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
10324 (rs6000_eh_return_filter_mode): Likewise.
10325 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
10326 (s390_libgcc_shift_count_mode): Likewise.
10327 (s390_unwind_word_mode): Likewise.
10328 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
10330 * target.def (mode_rep_extended): Likewise.
10331 (valid_pointer_mode): Likewise.
10332 (addr_space.valid_pointer_mode): Likewise.
10333 (eh_return_filter_mode): Return a scalar_int_mode rather than
10335 (libgcc_cmp_return_mode): Likewise.
10336 (libgcc_shift_count_mode): Likewise.
10337 (unwind_word_mode): Likewise.
10338 (addr_space.pointer_mode): Likewise.
10339 (addr_space.address_mode): Likewise.
10340 * doc/tm.texi: Regenerate.
10341 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
10343 (do_jump): Use scalar_int_mode for local variables.
10344 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
10345 rather than a machine_mode.
10346 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
10347 (scompare_loc_descriptor_wide): Likewise.
10348 (scompare_loc_descriptor_narrow): Likewise.
10349 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
10351 * except.c (sjlj_emit_dispatch_table): Likewise.
10352 (expand_builtin_eh_copy_values): Likewise.
10353 * explow.c (convert_memory_address_addr_space_1): Likewise.
10354 Take a scalar_int_mode rather than a machine_mode.
10355 (convert_memory_address_addr_space): Take a scalar_int_mode rather
10356 than a machine_mode.
10357 (memory_address_addr_space): Use scalar_int_mode for local variables.
10358 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
10359 rather than a machine_mode.
10360 * expmed.c (mask_rtx): Likewise.
10361 (init_expmed_one_conv): Likewise.
10362 (expand_mult_highpart_adjust): Likewise.
10363 (extract_high_half): Likewise.
10364 (expmed_mult_highpart_optab): Likewise.
10365 (expmed_mult_highpart): Likewise.
10366 (expand_smod_pow2): Likewise.
10367 (expand_sdiv_pow2): Likewise.
10368 (emit_store_flag_int): Likewise.
10369 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
10371 (extract_low_bits): Likewise.
10372 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
10374 * expr.c (pieces_addr::adjust): Likewise.
10375 (can_store_by_pieces): Likewise.
10376 (store_by_pieces): Likewise.
10377 (clear_by_pieces_1): Likewise.
10378 (expand_expr_addr_expr_1): Likewise.
10379 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
10380 (expand_expr_real_1): Likewise.
10381 (try_casesi): Likewise.
10382 * final.c (shorten_branches): Likewise.
10383 * fold-const.c (fold_convert_const_int_from_fixed): Change the
10384 type of "mode" to machine_mode.
10385 * internal-fn.c (expand_arith_overflow_result_store): Take a
10386 scalar_int_mode rather than a machine_mode.
10387 (expand_mul_overflow): Use scalar_int_mode for local variables.
10388 * loop-doloop.c (doloop_modify): Likewise.
10389 (doloop_optimize): Likewise.
10390 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
10391 than a machine_mode.
10392 (expand_doubleword_shift_condmove): Likewise.
10393 (expand_doubleword_shift): Likewise.
10394 (expand_doubleword_clz): Likewise.
10395 (expand_doubleword_popcount): Likewise.
10396 (expand_doubleword_parity): Likewise.
10397 (expand_absneg_bit): Use scalar_int_mode for local variables.
10398 (prepare_float_lib_cmp): Likewise.
10399 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
10400 rather than a machine_mode.
10401 (convert_memory_address_addr_space): Likewise.
10402 (get_mode_bounds): Likewise.
10403 (get_address_mode): Return a scalar_int_mode rather than a
10405 * rtlanal.c (get_address_mode): Likewise.
10406 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
10407 than a machine_mode.
10408 * targhooks.c (default_mode_rep_extended): Likewise.
10409 (default_valid_pointer_mode): Likewise.
10410 (default_addr_space_valid_pointer_mode): Likewise.
10411 (default_eh_return_filter_mode): Return a scalar_int_mode rather
10412 than a machine_mode.
10413 (default_libgcc_cmp_return_mode): Likewise.
10414 (default_libgcc_shift_count_mode): Likewise.
10415 (default_unwind_word_mode): Likewise.
10416 (default_addr_space_pointer_mode): Likewise.
10417 (default_addr_space_address_mode): Likewise.
10418 * targhooks.h (default_eh_return_filter_mode): Likewise.
10419 (default_libgcc_cmp_return_mode): Likewise.
10420 (default_libgcc_shift_count_mode): Likewise.
10421 (default_unwind_word_mode): Likewise.
10422 (default_addr_space_pointer_mode): Likewise.
10423 (default_addr_space_address_mode): Likewise.
10424 (default_mode_rep_extended): Take a scalar_int_mode rather than
10426 (default_valid_pointer_mode): Likewise.
10427 (default_addr_space_valid_pointer_mode): Likewise.
10428 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
10430 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
10431 rather than a machine_mode.
10432 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
10433 for local variables.
10434 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10435 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
10436 than a machine_mode.
10438 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10439 Alan Hayward <alan.hayward@arm.com>
10440 David Sherwood <david.sherwood@arm.com>
10442 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
10443 the mode argument to scalar_int_mode.
10444 (do_jump_by_parts_zero_rtx): Likewise.
10445 (do_jump_by_parts_equality_rtx): Likewise.
10446 (do_jump_by_parts_greater): Take a mode argument.
10447 (do_jump_by_parts_equality): Likewise.
10448 (do_jump_1): Update calls accordingly.
10450 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10451 Alan Hayward <alan.hayward@arm.com>
10452 David Sherwood <david.sherwood@arm.com>
10454 * is-a.h (safe_dyn_cast): New function.
10455 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
10456 (jump_table_for_label): Likewise.
10457 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
10458 instead of an rtx_insn *.
10459 (shorten_branches): Use dyn_cast instead of LABEL_P and
10460 JUMP_TABLE_DATA_P. Use jump_table_for_label and
10461 rtx_jump_table_data::get_data_mode.
10462 (final_scan_insn): Likewise.
10464 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10465 Alan Hayward <alan.hayward@arm.com>
10466 David Sherwood <david.sherwood@arm.com>
10468 * combine.c (try_combine): Use is_a <scalar_int_mode> when
10469 trying to combine a full-register integer set with a subreg
10472 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10473 Alan Hayward <alan.hayward@arm.com>
10474 David Sherwood <david.sherwood@arm.com>
10476 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
10477 that is always either address_mode or pointer_mode.
10479 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10480 Alan Hayward <alan.hayward@arm.com>
10481 David Sherwood <david.sherwood@arm.com>
10483 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
10484 when the two are known to be equal.
10486 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10487 Alan Hayward <alan.hayward@arm.com>
10488 David Sherwood <david.sherwood@arm.com>
10490 * simplify-rtx.c (simplify_const_unary_operation): Use
10491 is_a <scalar_int_mode> instead of checking for a nonzero
10492 precision. Forcibly convert op_mode to a scalar_int_mode
10493 in that case. More clearly differentiate the operand and
10494 result modes and use the former when deciding what the value
10495 of a count-bits operation should be. Use is_int_mode instead
10496 of checking for a MODE_INT. Remove redundant check for whether
10497 this mode has a zero precision.
10499 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10500 Alan Hayward <alan.hayward@arm.com>
10501 David Sherwood <david.sherwood@arm.com>
10503 * optabs.c (widen_leading): Change the type of the mode argument
10504 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
10505 (widen_bswap): Likewise.
10506 (expand_parity): Likewise.
10507 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
10508 (expand_ffs): Likewise.
10509 (epand_unop): Check for scalar integer modes before calling the
10512 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10513 Alan Hayward <alan.hayward@arm.com>
10514 David Sherwood <david.sherwood@arm.com>
10516 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
10518 (expand_expr_real_1): Update call accordingly.
10520 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10521 Alan Hayward <alan.hayward@arm.com>
10522 David Sherwood <david.sherwood@arm.com>
10524 * expmed.c (store_bit_field_using_insv): Add op0_mode and
10525 value_mode arguments. Use scalar_int_mode internally.
10526 (store_bit_field_1): Rename the new integer mode from imode
10527 to op0_mode and use it instead of GET_MODE (op0). Update calls
10528 to store_split_bit_field, store_bit_field_using_insv and
10529 store_fixed_bit_field.
10530 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
10531 Use scalar_int_mode internally. Use a bit count rather than a mode
10532 when calculating the largest bit size for get_best_mode.
10533 Update calls to store_split_bit_field and store_fixed_bit_field_1.
10534 (store_fixed_bit_field_1): Add mode and value_mode arguments.
10535 Remove assertion that OP0 has a scalar integer mode.
10536 (store_split_bit_field): Add op0_mode and value_mode arguments.
10537 Update calls to extract_fixed_bit_field.
10538 (extract_bit_field_using_extv): Add an op0_mode argument.
10539 Use scalar_int_mode internally.
10540 (extract_bit_field_1): Rename the new integer mode from imode to
10541 op0_mode and use it instead of GET_MODE (op0). Update calls to
10542 extract_split_bit_field, extract_bit_field_using_extv and
10543 extract_fixed_bit_field.
10544 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
10545 to extract_split_bit_field and extract_fixed_bit_field_1.
10546 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
10547 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
10548 on the target mode.
10549 (extract_split_bit_field): Add an op0_mode argument. Update call
10550 to extract_fixed_bit_field.
10552 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10553 Alan Hayward <alan.hayward@arm.com>
10554 David Sherwood <david.sherwood@arm.com>
10556 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
10557 * explow.c (hard_function_value): Likewise.
10558 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
10559 convert_to_mode call outside the loop.
10560 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
10561 for the mode iterator. Require the mode specified by max_pieces
10563 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
10565 (copy_blkmode_to_reg): Likewise.
10566 (set_storage_via_setmem): Likewise.
10567 * optabs.c (prepare_cmp_insn): Likewise.
10568 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
10569 * stor-layout.c (finish_bitfield_representative): Likewise.
10571 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10572 Alan Hayward <alan.hayward@arm.com>
10573 David Sherwood <david.sherwood@arm.com>
10575 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
10576 * expr.c (convert_move): Use them.
10577 (convert_modes): Likewise.
10578 (store_expr_with_bounds): Likewise.
10580 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10581 Alan Hayward <alan.hayward@arm.com>
10582 David Sherwood <david.sherwood@arm.com>
10584 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
10585 parameter for the mode of "x". Remove the "known_x", "known_mode"
10586 and "known_ret" arguments. Change the type of the mode argument
10587 to scalar_int_mode.
10588 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
10589 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
10590 (reg_num_sign_bit_copies_for_combine): Likewise.
10591 * rtlanal.c (nonzero_bits1): Likewise.
10592 (num_sign_bit_copies1): Likewise.
10593 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
10594 (reg_num_sign_bit_copies_general): Likewise.
10595 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
10596 (reg_nonzero_bits_general): Likewise.
10598 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10599 Alan Hayward <alan.hayward@arm.com>
10600 David Sherwood <david.sherwood@arm.com>
10602 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
10603 than in subroutines. Return 1 for non-integer modes.
10604 (cached_num_sign_bit_copies): Change the type of the mode parameter
10605 to scalar_int_mode.
10606 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
10607 classes. Handle CONST_INT_P first and then check whether X also
10608 has a scalar integer mode. Check the same thing for inner registers
10609 of a SUBREG and for values that are being extended or truncated.
10611 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10612 Alan Hayward <alan.hayward@arm.com>
10613 David Sherwood <david.sherwood@arm.com>
10615 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
10616 in subroutines. Return the mode mask for non-integer modes.
10617 (cached_nonzero_bits): Change the type of the mode parameter
10618 to scalar_int_mode.
10619 (nonzero_bits1): Likewise. Remove early exit for other mode
10620 classes. Handle CONST_INT_P first and then check whether X
10621 also has a scalar integer mode.
10623 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10624 Alan Hayward <alan.hayward@arm.com>
10625 David Sherwood <david.sherwood@arm.com>
10627 * expr.c (widest_int_mode_for_size): Make the comment match the code.
10628 Return a scalar_int_mode and assert that the size is greater than
10630 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
10631 (op_by_pieces_d::op_by_pieces_d): Likewise.
10632 (op_by_pieces_d::run): Likewise.
10633 (can_store_by_pieces): Likewise.
10635 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10636 Alan Hayward <alan.hayward@arm.com>
10637 David Sherwood <david.sherwood@arm.com>
10639 * combine.c (extract_left_shift): Add a mode argument and update
10641 (make_compound_operation_int): Change the type of the mode parameter
10642 to scalar_int_mode and update the call to extract_left_shift.
10644 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10645 Alan Hayward <alan.hayward@arm.com>
10646 David Sherwood <david.sherwood@arm.com>
10648 * combine.c (simplify_and_const_int): Change the type of the mode
10649 parameter to scalar_int_mode.
10650 (simplify_and_const_int_1): Likewise. Update recursive call.
10652 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10653 Alan Hayward <alan.hayward@arm.com>
10654 David Sherwood <david.sherwood@arm.com>
10656 * combine.c (simplify_compare_const): Check that the mode is a
10657 scalar_int_mode (rather than VOIDmode) before testing its
10659 (simplify_comparison): Move COMPARISON_P handling out of the
10660 loop and restrict the latter part of the loop to scalar_int_modes.
10661 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
10662 and when considering SUBREG_REGs. Use is_int_mode instead of
10663 checking GET_MODE_CLASS against MODE_INT.
10665 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10666 Alan Hayward <alan.hayward@arm.com>
10667 David Sherwood <david.sherwood@arm.com>
10669 * combine.c (try_widen_shift_mode): Move check for equal modes to...
10670 (simplify_shift_const_1): ...here. Use scalar_int_mode for
10671 shift_unit_mode and for modes involved in scalar shifts.
10673 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10674 Alan Hayward <alan.hayward@arm.com>
10675 David Sherwood <david.sherwood@arm.com>
10677 * combine.c (force_int_to_mode): New function, split out from...
10678 (force_to_mode): ...here. Keep xmode up-to-date and use it
10679 instead of GET_MODE (x).
10681 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10682 Alan Hayward <alan.hayward@arm.com>
10683 David Sherwood <david.sherwood@arm.com>
10685 * optabs-query.h (extraction_insn::struct_mode): Change type to
10686 opt_scalar_int_mode and update comment.
10687 (extraction_insn::field_mode): Change type to scalar_int_mode.
10688 (extraction_insn::pos_mode): Likewise.
10689 * combine.c (make_extraction): Update accordingly.
10690 * optabs-query.c (get_traditional_extraction_insn): Likewise.
10691 (get_optab_extraction_insn): Likewise.
10692 * recog.c (simplify_while_replacing): Likewise.
10693 * expmed.c (narrow_bit_field_mem): Change the type of the mode
10694 parameter to opt_scalar_int_mode.
10696 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10697 Alan Hayward <alan.hayward@arm.com>
10698 David Sherwood <david.sherwood@arm.com>
10700 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
10701 to a scalar_int_mode instead of a machine_mode.
10702 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
10703 (get_best_mode): Return a boolean and use a pointer argument to store
10704 the selected mode. Replace the limit mode parameter with a bit limit.
10705 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
10706 for the values returned by bit_field_mode_iterator::next_mode.
10707 (store_bit_field): Update call to get_best_mode.
10708 (store_fixed_bit_field): Likewise.
10709 (extract_fixed_bit_field): Likewise.
10710 * expr.c (optimize_bitfield_assignment_op): Likewise.
10711 * fold-const.c (optimize_bit_field_compare): Likewise.
10712 (fold_truth_andor_1): Likewise.
10713 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
10714 Update for new type of m_mode.
10715 (get_best_mode): As above.
10717 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10718 Alan Hayward <alan.hayward@arm.com>
10719 David Sherwood <david.sherwood@arm.com>
10721 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
10722 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
10723 (store_bit_field): Check is_a <scalar_int_mode> before calling
10724 strict_volatile_bitfield_p.
10725 (extract_bit_field): Likewise.
10727 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10728 Alan Hayward <alan.hayward@arm.com>
10729 David Sherwood <david.sherwood@arm.com>
10731 * target.def (cstore_mode): Return a scalar_int_mode.
10732 * doc/tm.texi: Regenerate.
10733 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
10734 * targhooks.h (default_cstore_mode): Likewise.
10735 * targhooks.c (default_cstore_mode): Likewise, using a forced
10737 * expmed.c (emit_cstore): Expect the target of the cstore to be
10740 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10741 Alan Hayward <alan.hayward@arm.com>
10742 David Sherwood <david.sherwood@arm.com>
10744 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
10746 (niter_desc): Likewise mode.
10747 (iv_analyze): Add a mode parameter.
10749 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
10750 and change its type to scalar_int_mode.
10751 * loop-iv.c: Update commentary at head of file.
10752 (iv_constant): Pass the mode paraeter before the rtx it describes
10753 and change its type to scalar_int_mode. Remove VOIDmode handling.
10754 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
10755 (iv_extend): Likewise.
10756 (shorten_into_mode): Likewise.
10757 (iv_add): Use scalar_int_mode.
10758 (iv_mult): Likewise.
10759 (iv_shift): Likewise.
10760 (canonicalize_iv_subregs): Likewise.
10761 (get_biv_step_1): Pass the outer_mode parameter before the rtx
10762 it describes and change its mode to scalar_int_mode. Also change
10763 the type of the returned inner_mode to scalar_int_mode.
10764 (get_biv_step): Likewise, turning outer_mode from a pointer
10765 into a direct parameter. Update call to get_biv_step_1.
10766 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
10767 iv_constant and get_biv_step.
10768 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
10769 and change its type to scalar_int_mode. Don't initialise iv->mode
10770 to VOIDmode and remove later checks for its still being VOIDmode.
10771 Update calls to iv_analyze_op and iv_analyze_expr. Check
10772 is_a <scalar_int_mode> when changing the mode under consideration.
10773 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
10774 Update call to iv_analyze_expr.
10775 (iv_analyze_op): Add a mode parameter. Reject subregs whose
10776 inner register is not also a scalar_int_mode. Update call to
10778 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
10779 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
10780 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
10781 separate mode class checks. Update calls to iv_analyze. Remove
10782 fix-up of VOIDmodes after iv_analyze_biv.
10783 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
10784 don't have a scalar_int_mode. Update call to biv_p.
10786 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10787 Alan Hayward <alan.hayward@arm.com>
10788 David Sherwood <david.sherwood@arm.com>
10790 * cfgexpand.c (convert_debug_memory_address): Use
10791 as_a <scalar_int_mode>.
10792 * combine.c (expand_compound_operation): Likewise.
10793 (make_extraction): Likewise.
10794 (change_zero_ext): Likewise.
10795 (simplify_comparison): Likewise.
10796 * cse.c (cse_insn): Likewise.
10797 * dwarf2out.c (minmax_loc_descriptor): Likewise.
10798 (mem_loc_descriptor): Likewise.
10799 (loc_descriptor): Likewise.
10800 * expmed.c (init_expmed_one_mode): Likewise.
10801 (synth_mult): Likewise.
10802 (emit_store_flag_1): Likewise.
10803 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
10804 of a comparison with size.
10805 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
10806 (reduce_to_bit_field_precision): Likewise.
10807 * function.c (expand_function_end): Likewise.
10808 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
10809 * loop-doloop.c (doloop_modify): Likewise.
10810 * optabs.c (expand_binop): Likewise.
10811 (expand_unop): Likewise.
10812 (expand_copysign_absneg): Likewise.
10813 (prepare_cmp_insn): Likewise.
10814 (maybe_legitimize_operand): Likewise.
10815 * recog.c (const_scalar_int_operand): Likewise.
10816 * rtlanal.c (get_address_mode): Likewise.
10817 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10818 (simplify_cond_clz_ctz): Likewise.
10819 * tree-nested.c (get_nl_goto_field): Likewise.
10820 * tree.c (build_vector_type_for_mode): Likewise.
10821 * var-tracking.c (use_narrower_mode): Likewise.
10823 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10824 Alan Hayward <alan.hayward@arm.com>
10825 David Sherwood <david.sherwood@arm.com>
10827 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
10828 * builtins.c (expand_builtin_signbit): Use it.
10829 * cfgexpand.c (expand_debug_expr): Likewise.
10830 * dojump.c (do_jump): Likewise.
10831 (do_compare_and_jump): Likewise.
10832 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
10833 * expmed.c (make_tree): Likewise.
10834 * expr.c (expand_expr_real_2): Likewise.
10835 (expand_expr_real_1): Likewise.
10836 (try_casesi): Likewise.
10837 * fold-const-call.c (fold_const_call_ss): Likewise.
10838 * fold-const.c (unextend): Likewise.
10839 (extract_muldiv_1): Likewise.
10840 (fold_single_bit_test): Likewise.
10841 (native_encode_int): Likewise.
10842 (native_encode_string): Likewise.
10843 (native_interpret_int): Likewise.
10844 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
10845 * internal-fn.c (expand_addsub_overflow): Likewise.
10846 (expand_neg_overflow): Likewise.
10847 (expand_mul_overflow): Likewise.
10848 (expand_arith_overflow): Likewise.
10849 * match.pd: Likewise.
10850 * stor-layout.c (layout_type): Likewise.
10851 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
10852 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
10853 * tree-ssanames.c (get_range_info): Likewise.
10854 * tree-switch-conversion.c (array_value_type) Likewise.
10855 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
10856 (vect_recog_divmod_pattern): Likewise.
10857 (vect_recog_mixed_size_cond_pattern): Likewise.
10858 * tree-vrp.c (extract_range_basic): Likewise.
10859 (simplify_float_conversion_using_ranges): Likewise.
10860 * tree.c (int_fits_type_p): Likewise.
10861 * ubsan.c (instrument_bool_enum_load): Likewise.
10862 * varasm.c (mergeable_string_section): Likewise.
10863 (narrowing_initializer_constant_valid_p): Likewise.
10864 (output_constant): Likewise.
10866 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10867 Alan Hayward <alan.hayward@arm.com>
10868 David Sherwood <david.sherwood@arm.com>
10870 * machmode.h (NARROWEST_INT_MODE): New macro.
10871 * expr.c (alignment_for_piecewise_move): Use it instead of
10872 GET_CLASS_NARROWEST_MODE (MODE_INT).
10873 (push_block): Likewise.
10874 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
10876 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10878 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10879 Alan Hayward <alan.hayward@arm.com>
10880 David Sherwood <david.sherwood@arm.com>
10882 * postreload.c (move2add_valid_value_p): Change the type of the
10883 mode parameter to scalar_int_mode.
10884 (move2add_use_add2_insn): Add a mode parameter and use it instead
10886 (move2add_use_add3_insn): Likewise.
10887 (reload_cse_move2add): Update accordingly.
10889 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10890 Alan Hayward <alan.hayward@arm.com>
10891 David Sherwood <david.sherwood@arm.com>
10893 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
10895 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
10896 * optabs.c (expand_unop): Likewise.
10898 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10899 Alan Hayward <alan.hayward@arm.com>
10900 David Sherwood <david.sherwood@arm.com>
10902 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
10903 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
10904 (popcount_loc_descriptor): Likewise.
10905 (bswap_loc_descriptor): Likewise.
10906 (rotate_loc_descriptor): Likewise.
10907 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
10908 calling the functions above.
10910 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10911 Alan Hayward <alan.hayward@arm.com>
10912 David Sherwood <david.sherwood@arm.com>
10914 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
10916 (try_combine): Likewise.
10917 (simplify_if_then_else): Likewise.
10918 * cse.c (cse_insn): Likewise.
10919 * dwarf2out.c (mem_loc_descriptor): Likewise.
10920 * emit-rtl.c (gen_lowpart_common): Likewise.
10921 * simplify-rtx.c (simplify_truncation): Likewise.
10922 (simplify_binary_operation_1): Likewise.
10923 (simplify_const_relational_operation): Likewise.
10924 (simplify_ternary_operation): Likewise.
10925 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
10927 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10928 Alan Hayward <alan.hayward@arm.com>
10929 David Sherwood <david.sherwood@arm.com>
10931 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
10932 * reload.c (push_reload): Likewise.
10933 (find_reloads): Likewise.
10935 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10936 Alan Hayward <alan.hayward@arm.com>
10937 David Sherwood <david.sherwood@arm.com>
10939 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
10940 (make_compound_operation_int): Likewise.
10941 (change_zero_ext): Likewise.
10942 * expr.c (convert_move): Likewise.
10943 (convert_modes): Likewise.
10944 * fwprop.c (forward_propagate_subreg): Likewise.
10945 * loop-iv.c (get_biv_step_1): Likewise.
10946 * optabs.c (widen_operand): Likewise.
10947 * postreload.c (move2add_valid_value_p): Likewise.
10948 * recog.c (simplify_while_replacing): Likewise.
10949 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10950 (simplify_binary_operation_1): Likewise. Remove redundant
10951 mode equality check.
10953 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10954 Alan Hayward <alan.hayward@arm.com>
10955 David Sherwood <david.sherwood@arm.com>
10957 * combine.c (combine_simplify_rtx): Add checks for
10958 is_a <scalar_int_mode>.
10959 (simplify_if_then_else): Likewise.
10960 (make_field_assignment): Likewise.
10961 (simplify_comparison): Likewise.
10962 * ifcvt.c (noce_try_bitop): Likewise.
10963 * loop-invariant.c (canonicalize_address_mult): Likewise.
10964 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10966 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10967 Alan Hayward <alan.hayward@arm.com>
10968 David Sherwood <david.sherwood@arm.com>
10970 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
10971 is_a <scalar_int_mode> instead of != BLKmode.
10973 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10974 Alan Hayward <alan.hayward@arm.com>
10975 David Sherwood <david.sherwood@arm.com>
10977 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
10978 instead of != VOIDmode.
10979 * combine.c (if_then_else_cond): Likewise.
10980 (change_zero_ext): Likewise.
10981 * dwarf2out.c (mem_loc_descriptor): Likewise.
10982 (loc_descriptor): Likewise.
10983 * rtlanal.c (canonicalize_condition): Likewise.
10984 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
10986 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10987 Alan Hayward <alan.hayward@arm.com>
10988 David Sherwood <david.sherwood@arm.com>
10990 * simplify-rtx.c (simplify_binary_operation_1): Use
10991 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
10993 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10994 Alan Hayward <alan.hayward@arm.com>
10995 David Sherwood <david.sherwood@arm.com>
10997 * wide-int.h (int_traits<unsigned char>) New class.
10998 (int_traits<unsigned short>) Likewise.
10999 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
11000 Use GET_MODE_UNIT_PRECISION and remove redundant test for
11002 * combine.c (set_nonzero_bits_and_sign_copies): Use
11003 is_a <scalar_int_mode>.
11004 (find_split_point): Likewise.
11005 (combine_simplify_rtx): Likewise.
11006 (simplify_logical): Likewise.
11007 (expand_compound_operation): Likewise.
11008 (expand_field_assignment): Likewise.
11009 (make_compound_operation): Likewise.
11010 (extended_count): Likewise.
11011 (change_zero_ext): Likewise.
11012 (simplify_comparison): Likewise.
11013 * dwarf2out.c (scompare_loc_descriptor): Likewise.
11014 (ucompare_loc_descriptor): Likewise.
11015 (minmax_loc_descriptor): Likewise.
11016 (mem_loc_descriptor): Likewise.
11017 (loc_descriptor): Likewise.
11018 * expmed.c (init_expmed_one_mode): Likewise.
11019 * lra-constraints.c (lra_constraint_offset): Likewise.
11020 * optabs.c (prepare_libcall_arg): Likewise.
11021 * postreload.c (move2add_note_store): Likewise.
11022 * reload.c (operands_match_p): Likewise.
11023 * rtl.h (load_extend_op): Likewise.
11024 * rtlhooks.c (gen_lowpart_general): Likewise.
11025 * simplify-rtx.c (simplify_truncation): Likewise.
11026 (simplify_unary_operation_1): Likewise.
11027 (simplify_binary_operation_1): Likewise.
11028 (simplify_const_binary_operation): Likewise.
11029 (simplify_const_relational_operation): Likewise.
11030 (simplify_subreg): Likewise.
11031 * stor-layout.c (bitwise_mode_for_mode): Likewise.
11032 * var-tracking.c (adjust_mems): Likewise.
11033 (prepare_call_arguments): Likewise.
11035 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11036 Alan Hayward <alan.hayward@arm.com>
11037 David Sherwood <david.sherwood@arm.com>
11039 * machmode.h (is_int_mode): New fuction.
11040 * combine.c (find_split_point): Use it.
11041 (combine_simplify_rtx): Likewise.
11042 (simplify_if_then_else): Likewise.
11043 (simplify_set): Likewise.
11044 (simplify_shift_const_1): Likewise.
11045 (simplify_comparison): Likewise.
11046 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
11047 * cse.c (notreg_cost): Likewise.
11048 (cse_insn): Likewise.
11049 * cselib.c (cselib_lookup_1): Likewise.
11050 * dojump.c (do_jump_1): Likewise.
11051 (do_compare_rtx_and_jump): Likewise.
11052 * dse.c (get_call_args): Likewise.
11053 * dwarf2out.c (rtl_for_decl_init): Likewise.
11054 (native_encode_initializer): Likewise.
11055 * expmed.c (emit_store_flag_1): Likewise.
11056 (emit_store_flag): Likewise.
11057 * expr.c (convert_modes): Likewise.
11058 (store_field): Likewise.
11059 (expand_expr_real_1): Likewise.
11060 * fold-const.c (fold_read_from_constant_string): Likewise.
11061 * gimple-ssa-sprintf.c (get_format_string): Likewise.
11062 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
11063 * optabs.c (expand_binop): Likewise.
11064 (expand_unop): Likewise.
11065 (expand_abs_nojump): Likewise.
11066 (expand_one_cmpl_abs_nojump): Likewise.
11067 * simplify-rtx.c (mode_signbit_p): Likewise.
11068 (val_signbit_p): Likewise.
11069 (val_signbit_known_set_p): Likewise.
11070 (val_signbit_known_clear_p): Likewise.
11071 (simplify_relational_operation_1): Likewise.
11072 * tree.c (vector_type_mode): Likewise.
11074 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11075 Alan Hayward <alan.hayward@arm.com>
11076 David Sherwood <david.sherwood@arm.com>
11078 * machmode.h (smallest_mode_for_size): Fix formatting.
11079 (smallest_int_mode_for_size): New function.
11080 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
11081 instead of smallest_mode_for_size.
11082 * combine.c (make_extraction): Likewise.
11083 * config/arc/arc.c (arc_expand_movmem): Likewise.
11084 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
11085 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
11086 * config/s390/s390.c (s390_expand_insv): Likewise.
11087 * config/sparc/sparc.c (assign_int_registers): Likewise.
11088 * config/spu/spu.c (spu_function_value): Likewise.
11089 (spu_function_arg): Likewise.
11090 * coverage.c (get_gcov_type): Likewise.
11091 (get_gcov_unsigned_t): Likewise.
11092 * dse.c (find_shift_sequence): Likewise.
11093 * expmed.c (store_bit_field_1): Likewise.
11094 * expr.c (convert_move): Likewise.
11095 (store_field): Likewise.
11096 * internal-fn.c (expand_arith_overflow): Likewise.
11097 * optabs-query.c (get_best_extraction_insn): Likewise.
11098 * optabs.c (expand_twoval_binop_libfunc): Likewise.
11099 * stor-layout.c (layout_type): Likewise.
11100 (initialize_sizetypes): Likewise.
11101 * targhooks.c (default_get_mask_mode): Likewise.
11102 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
11104 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11105 Alan Hayward <alan.hayward@arm.com>
11106 David Sherwood <david.sherwood@arm.com>
11108 * machmode.h (opt_mode::else_blk): New function.
11109 (int_mode_for_mode): Declare.
11110 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
11111 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
11113 * cfgexpand.c (expand_debug_expr): Likewise.
11114 * combine.c (gen_lowpart_or_truncate): Likewise.
11115 (gen_lowpart_for_combine): Likewise.
11116 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
11117 * config/avr/avr.c (avr_to_int_mode): Likewise.
11118 (avr_out_plus_1): Likewise.
11119 (avr_out_plus): Likewise.
11120 (avr_out_round): Likewise.
11121 * config/i386/i386.c (ix86_split_to_parts): Likewise.
11122 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
11123 (s390_expand_vcond): Likewise.
11124 * config/spu/spu.c (spu_split_immediate): Likewise.
11125 (spu_expand_mov): Likewise.
11126 * dse.c (get_stored_val): Likewise.
11127 * expmed.c (store_bit_field_1): Likewise.
11128 (convert_extracted_bit_field): Use int_mode_for_mode instead of
11130 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
11131 (extract_low_bits): Likewise.
11132 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
11133 handling rather than repeating the check.
11134 (emit_group_store): Likewise.
11135 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
11136 * optabs.c (expand_absneg_bit): Likewise.
11137 (expand_copysign_absneg): Likewise.
11138 (expand_copysign_bit): Likewise.
11139 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
11140 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
11141 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
11142 * var-tracking.c (prepare_call_arguments): Likewise.
11143 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
11144 int_mode_for_mode instead of mode_for_size.
11145 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
11147 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11148 Alan Hayward <alan.hayward@arm.com>
11149 David Sherwood <david.sherwood@arm.com>
11151 * machmode.h (int_mode_for_size): New function.
11152 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
11153 instead of mode_for_size.
11154 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
11156 * combine.c (expand_field_assignment): Use int_mode_for_size
11157 instead of mode_for_size.
11158 (make_extraction): Likewise.
11159 (simplify_shift_const_1): Likewise.
11160 (simplify_comparison): Likewise.
11161 * dojump.c (do_jump): Likewise.
11162 * dwarf2out.c (mem_loc_descriptor): Likewise.
11163 * emit-rtl.c (init_derived_machine_modes): Likewise.
11164 * expmed.c (flip_storage_order): Likewise.
11165 (convert_extracted_bit_field): Likewise.
11166 * expr.c (copy_blkmode_from_reg): Likewise.
11167 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
11168 * internal-fn.c (expand_mul_overflow): Likewise.
11169 * lower-subreg.c (simple_move): Likewise.
11170 * optabs-libfuncs.c (init_optabs): Likewise.
11171 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11172 * tree.c (vector_type_mode): Likewise.
11173 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
11174 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
11175 * tree-vect-generic.c (expand_vector_parallel): Likewise.
11176 * tree-vect-stmts.c (vectorizable_load): Likewise.
11177 (vectorizable_store): Likewise.
11179 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11180 Alan Hayward <alan.hayward@arm.com>
11181 David Sherwood <david.sherwood@arm.com>
11183 * coretypes.h (pod_mode): New type.
11184 (scalar_int_mode_pod): New typedef.
11185 * machmode.h (pod_mode): New class.
11186 (int_n_data_t::m): Change type to scalar_int_mode_pod.
11187 * genmodes.c (emit_mode_int_n): Update accordingly.
11188 * lower-subreg.h (target_lower_subreg): Change type to
11189 scalar_int_mode_pod.
11190 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
11191 scalar_int_mode_pod.
11193 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11194 Alan Hayward <alan.hayward@arm.com>
11195 David Sherwood <david.sherwood@arm.com>
11197 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
11198 machine_mode to scalar_int_mode.
11199 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
11200 (rs6000_option_override_internal): Remove cast to int.
11201 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
11202 machine_mode to scalar_int_mode.
11203 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
11204 (rs6000_option_override_internal): Remove cast to int.
11205 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
11206 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
11208 * config/s390/s390.c (s390_expand_builtin): Likewise.
11209 * coretypes.h (scalar_int_mode): New type.
11210 (opt_scalar_int_mode): New typedef.
11211 * machmode.h (scalar_int_mode): New class.
11212 (scalar_int_mode::includes_p): New function.
11213 (byte_mode): Change type to scalar_int_mode.
11214 (word_mode): Likewise.
11215 (ptr_mode): Likewise.
11216 * emit-rtl.c (byte_mode): Likewise.
11217 (word_mode): Likewise.
11218 (ptr_mode): Likewise.
11219 (init_derived_machine_modes): Update accordingly.
11220 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
11221 and MODE_PARTIAL_INT.
11222 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
11223 opt_scalar_int_mode.
11225 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11226 Alan Hayward <alan.hayward@arm.com>
11227 David Sherwood <david.sherwood@arm.com>
11229 * target.def (libgcc_floating_mode_supported_p): Take a
11231 * doc/tm.texi: Regenerate.
11232 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
11234 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
11235 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
11238 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11239 Alan Hayward <alan.hayward@arm.com>
11240 David Sherwood <david.sherwood@arm.com>
11242 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
11243 * doc/tm.texi: Regenerate.
11244 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
11245 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
11246 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
11247 * targhooks.h (default_floatn_mode): Likewise.
11248 * targhooks.c (default_floatn_mode): Likewise.
11249 * tree.c (build_common_tree_nodes): Update accordingly.
11251 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11252 Alan Hayward <alan.hayward@arm.com>
11253 David Sherwood <david.sherwood@arm.com>
11255 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
11256 (mode_iterator::iterate_p): Likewise.
11257 (mode_iterator::get_wider): Likewise.
11258 * expr.c (init_expr_target): Use opt_scalar_float_mode.
11260 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11261 Alan Hayward <alan.hayward@arm.com>
11262 David Sherwood <david.sherwood@arm.com>
11264 * coretypes.h (opt_scalar_float_mode): New typedef.
11265 * machmode.h (float_mode_for_size): New function.
11266 * emit-rtl.c (double_mode): Delete.
11267 (init_emit_once): Use float_mode_for_size.
11268 * stor-layout.c (layout_type): Likewise.
11269 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
11271 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11272 Alan Hayward <alan.hayward@arm.com>
11273 David Sherwood <david.sherwood@arm.com>
11275 * output.h (assemble_real): Take a scalar_float_mode.
11276 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
11277 * config/arm/arm.md (consttable_4): Likewise.
11278 (consttable_8): Likewise.
11279 (consttable_16): Likewise.
11280 * config/mips/mips.md (consttable_float): Likewise.
11281 * config/s390/s390.c (s390_output_pool_entry): Likewise.
11282 * varasm.c (assemble_real): Take a scalar_float_mode.
11283 (output_constant_pool_2): Update accordingly.
11284 (output_constant): Likewise.
11286 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11287 Alan Hayward <alan.hayward@arm.com>
11288 David Sherwood <david.sherwood@arm.com>
11290 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
11291 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
11292 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
11293 (native_encode_real): Likewise.
11294 (native_interpret_real): Likewise.
11295 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
11296 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11298 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11299 Alan Hayward <alan.hayward@arm.com>
11300 David Sherwood <david.sherwood@arm.com>
11302 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
11303 <scalar_float_mode>. Simplify.
11304 (gen_extend_conv_libfunc): Likewise.
11306 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11307 Alan Hayward <alan.hayward@arm.com>
11308 David Sherwood <david.sherwood@arm.com>
11310 * coretypes.h (scalar_float_mode): New type.
11311 * machmode.h (mode_traits::from_int): Use machine_mode if
11312 USE_ENUM_MODES is defined.
11313 (is_a): New function.
11315 (dyn_cast): Likewise.
11316 (scalar_float_mode): New class.
11317 (scalar_float_mode::includes_p): New function.
11318 (is_float_mode): Likewise.
11319 * gdbhooks.py (MachineModePrinter): New class.
11320 (build_pretty_printer): Use it for scalar_float_mode.
11321 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
11322 (format_helper::format_helper): Turn into a template.
11323 * genmodes.c (get_mode_class): New function.
11324 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
11325 or machine_mode if none.
11326 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
11327 as_a <scalar_float_mode>.
11328 * dwarf2out.c (mem_loc_descriptor): Likewise.
11329 (insert_float): Likewise.
11330 (add_const_value_attribute): Likewise.
11331 * simplify-rtx.c (simplify_immed_subreg): Likewise.
11332 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
11333 (expand_unop): Update accordingly.
11334 (expand_abs_nojump): Likewise.
11335 (expand_copysign_absneg): Take a scalar_float_mode.
11336 (expand_copysign_bit): Likewise.
11337 (expand_copysign): Update accordingly.
11339 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11340 Alan Hayward <alan.hayward@arm.com>
11341 David Sherwood <david.sherwood@arm.com>
11343 * coretypes.h (opt_mode): New class.
11344 * machmode.h (opt_mode): Likewise.
11345 (opt_mode::else_void): New function.
11346 (opt_mode::require): Likewise.
11347 (opt_mode::exists): Likewise.
11348 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
11349 (GET_MODE_2XWIDER_MODE): Likewise.
11350 (mode_iterator::get_wider): Update accordingly.
11351 (mode_iterator::get_2xwider): Likewise.
11352 (mode_iterator::get_known_wider): Likewise, turning into a template.
11353 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
11354 forcing a wider mode to exist.
11355 * config/cr16/cr16.h (LONG_REG_P): Likewise.
11356 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11357 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
11358 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
11359 * lower-subreg.c (init_lower_subreg): Likewise.
11360 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
11361 on the final iteration.
11362 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
11363 a wider mode exists before asking for a move pattern.
11364 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
11365 forcing a wider mode to exist.
11366 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
11367 returning false if no such mode exists.
11368 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
11369 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
11370 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
11371 Avoid checking for a MODE_INT if we already know the mode is not a
11373 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
11374 forcing a wider mode to exist.
11375 (expmed_mult_highpart_optab): Likewise.
11376 (expmed_mult_highpart): Likewise.
11377 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
11379 * lto-streamer-in.c (lto_input_mode_table): Likewise.
11380 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
11381 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
11382 * internal-fn.c (expand_mul_overflow): Update use of
11383 GET_MODE_2XWIDER_MODE.
11384 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11385 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
11386 GET_MODE_WIDER_MODE.
11387 (convert_plusminus_to_widen): Likewise.
11388 * tree-switch-conversion.c (array_value_type): Likewise.
11389 * var-tracking.c (emit_note_insn_var_location): Likewise.
11390 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11391 Return false inside rather than outside the loop if no wider mode
11393 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
11394 and GET_MODE_2XWIDER_MODE
11395 (can_compare_p): Use else_void.
11396 * gdbhooks.py (OptMachineModePrinter): New class.
11397 (build_pretty_printer): Use it for opt_mode.
11399 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11400 Alan Hayward <alan.hayward@arm.com>
11401 David Sherwood <david.sherwood@arm.com>
11403 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
11404 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
11406 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11407 Alan Hayward <alan.hayward@arm.com>
11408 David Sherwood <david.sherwood@arm.com>
11410 * machmode.h (mode_traits): New structure.
11411 (get_narrowest_mode): New function.
11412 (mode_iterator::start): Likewise.
11413 (mode_iterator::iterate_p): Likewise.
11414 (mode_iterator::get_wider): Likewise.
11415 (mode_iterator::get_known_wider): Likewise.
11416 (mode_iterator::get_2xwider): Likewise.
11417 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
11418 (FOR_EACH_MODE): Likewise.
11419 (FOR_EACH_MODE_FROM): Likewise.
11420 (FOR_EACH_MODE_UNTIL): Likewise.
11421 (FOR_EACH_WIDER_MODE): Likewise.
11422 (FOR_EACH_2XWIDER_MODE): Likewise.
11423 * builtins.c (expand_builtin_strlen): Use new mode iterators.
11424 * combine.c (simplify_comparison): Likewise
11425 * config/i386/i386.c (type_natural_mode): Likewise.
11426 * cse.c (cse_insn): Likewise.
11427 * dse.c (find_shift_sequence): Likewise.
11428 * emit-rtl.c (init_derived_machine_modes): Likewise.
11429 (init_emit_once): Likewise.
11430 * explow.c (hard_function_value): Likewise.
11431 * expmed.c (extract_fixed_bit_field_1): Likewise.
11432 (extract_bit_field_1): Likewise.
11433 (expand_divmod): Likewise.
11434 (emit_store_flag_1): Likewise.
11435 * expr.c (init_expr_target): Likewise.
11436 (convert_move): Likewise.
11437 (alignment_for_piecewise_move): Likewise.
11438 (widest_int_mode_for_size): Likewise.
11439 (emit_block_move_via_movmem): Likewise.
11440 (copy_blkmode_to_reg): Likewise.
11441 (set_storage_via_setmem): Likewise.
11442 (compress_float_constant): Likewise.
11443 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11444 * optabs-query.c (get_best_extraction_insn): Likewise.
11445 * optabs.c (expand_binop): Likewise.
11446 (expand_twoval_unop): Likewise.
11447 (expand_twoval_binop): Likewise.
11448 (widen_leading): Likewise.
11449 (widen_bswap): Likewise.
11450 (expand_parity): Likewise.
11451 (expand_unop): Likewise.
11452 (prepare_cmp_insn): Likewise.
11453 (prepare_float_lib_cmp): Likewise.
11454 (expand_float): Likewise.
11455 (expand_fix): Likewise.
11456 (expand_sfix_optab): Likewise.
11457 * postreload.c (move2add_use_add2_insn): Likewise.
11458 * reg-stack.c (reg_to_stack): Likewise.
11459 * reginfo.c (choose_hard_reg_mode): Likewise.
11460 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11461 * stor-layout.c (mode_for_size): Likewise.
11462 (smallest_mode_for_size): Likewise.
11463 (mode_for_vector): Likewise.
11464 (finish_bitfield_representative): Likewise.
11465 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
11466 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
11467 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11468 * var-tracking.c (prepare_call_arguments): Likewise.
11470 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11471 Alan Hayward <alan.hayward@arm.com>
11472 David Sherwood <david.sherwood@arm.com>
11474 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
11475 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
11476 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
11477 * machmode.h (mode_size): Move earlier in file.
11478 (mode_precision): Likewise.
11479 (mode_inner): Likewise.
11480 (mode_nunits): Likewise.
11481 (mode_unit_size): Likewise.
11482 (unit_unit_precision): Likewise.
11483 (mode_wider): Likewise.
11484 (mode_2xwider): Likewise.
11485 (machine_mode): New class.
11486 (mode_to_bytes): New function.
11487 (mode_to_bits): Likewise.
11488 (mode_to_precision): Likewise.
11489 (mode_to_inner): Likewise.
11490 (mode_to_unit_size): Likewise.
11491 (mode_to_unit_precision): Likewise.
11492 (mode_to_nunits): Likewise.
11493 (GET_MODE_SIZE): Use mode_to_bytes.
11494 (GET_MODE_BITSIZE): Use mode_to_bits.
11495 (GET_MODE_PRECISION): Use mode_to_precision.
11496 (GET_MODE_INNER): Use mode_to_inner.
11497 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
11498 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
11499 (GET_MODE_NUNITS): Use mode_to_nunits.
11500 * system.h (ALWAYS_INLINE): New macro.
11501 * config/powerpcspe/powerpcspe-c.c
11502 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
11503 int for arg1_mode and arg2_mode.
11505 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11506 Alan Hayward <alan.hayward@arm.com>
11507 David Sherwood <david.sherwood@arm.com>
11509 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
11510 Prefix mode names with E_ in case statements.
11511 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11512 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
11513 (aarch64_split_simd_move): Likewise.
11514 (aarch64_gen_storewb_pair): Likewise.
11515 (aarch64_gen_loadwb_pair): Likewise.
11516 (aarch64_gen_store_pair): Likewise.
11517 (aarch64_gen_load_pair): Likewise.
11518 (aarch64_get_condition_code_1): Likewise.
11519 (aarch64_constant_pool_reload_icode): Likewise.
11520 (get_rsqrte_type): Likewise.
11521 (get_rsqrts_type): Likewise.
11522 (get_recpe_type): Likewise.
11523 (get_recps_type): Likewise.
11524 (aarch64_gimplify_va_arg_expr): Likewise.
11525 (aarch64_simd_container_mode): Likewise.
11526 (aarch64_emit_load_exclusive): Likewise.
11527 (aarch64_emit_store_exclusive): Likewise.
11528 (aarch64_expand_compare_and_swap): Likewise.
11529 (aarch64_gen_atomic_cas): Likewise.
11530 (aarch64_emit_bic): Likewise.
11531 (aarch64_emit_atomic_swap): Likewise.
11532 (aarch64_emit_atomic_load_op): Likewise.
11533 (aarch64_evpc_trn): Likewise.
11534 (aarch64_evpc_uzp): Likewise.
11535 (aarch64_evpc_zip): Likewise.
11536 (aarch64_evpc_ext): Likewise.
11537 (aarch64_evpc_rev): Likewise.
11538 (aarch64_evpc_dup): Likewise.
11539 (aarch64_gen_ccmp_first): Likewise.
11540 (aarch64_gen_ccmp_next): Likewise.
11541 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
11542 (alpha_emit_xfloating_libcall): Likewise.
11543 (emit_insxl): Likewise.
11544 (alpha_arg_type): Likewise.
11545 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
11546 (arc_preferred_simd_mode): Likewise.
11547 (arc_secondary_reload): Likewise.
11548 (get_arc_condition_code): Likewise.
11549 (arc_print_operand): Likewise.
11550 (arc_legitimate_constant_p): Likewise.
11551 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11552 * config/arc/arc.md (casesi_load): Likewise.
11553 (casesi_compact_jump): Likewise.
11554 * config/arc/predicates.md (proper_comparison_operator): Likewise.
11555 (cc_use_register): Likewise.
11556 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11557 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
11558 (arm_init_iwmmxt_builtins): Likewise.
11559 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
11560 (neon_expand_vector_init): Likewise.
11561 (arm_attr_length_move_neon): Likewise.
11562 (maybe_get_arm_condition_code): Likewise.
11563 (arm_emit_vector_const): Likewise.
11564 (arm_preferred_simd_mode): Likewise.
11565 (arm_output_iwmmxt_tinsr): Likewise.
11566 (thumb1_output_casesi): Likewise.
11567 (thumb2_output_casesi): Likewise.
11568 (arm_emit_load_exclusive): Likewise.
11569 (arm_emit_store_exclusive): Likewise.
11570 (arm_expand_compare_and_swap): Likewise.
11571 (arm_evpc_neon_vuzp): Likewise.
11572 (arm_evpc_neon_vzip): Likewise.
11573 (arm_evpc_neon_vrev): Likewise.
11574 (arm_evpc_neon_vtrn): Likewise.
11575 (arm_evpc_neon_vext): Likewise.
11576 (arm_validize_comparison): Likewise.
11577 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
11578 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
11579 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
11580 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
11581 (c6x_preferred_simd_mode): Likewise.
11582 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
11583 (epiphany_rtx_costs): Likewise.
11584 * config/epiphany/predicates.md (proper_comparison_operator):
11586 * config/frv/frv.c (condexec_memory_operand): Likewise.
11587 (frv_emit_move): Likewise.
11588 (output_move_single): Likewise.
11589 (output_condmove_single): Likewise.
11590 (frv_hard_regno_mode_ok): Likewise.
11591 (frv_matching_accg_mode): Likewise.
11592 * config/h8300/h8300.c (split_adds_subs): Likewise.
11593 (h8300_rtx_costs): Likewise.
11594 (h8300_print_operand): Likewise.
11595 (compute_mov_length): Likewise.
11596 (output_logical_op): Likewise.
11597 (compute_logical_op_length): Likewise.
11598 (compute_logical_op_cc): Likewise.
11599 (h8300_shift_needs_scratch_p): Likewise.
11600 (output_a_shift): Likewise.
11601 (compute_a_shift_length): Likewise.
11602 (compute_a_shift_cc): Likewise.
11603 (expand_a_rotate): Likewise.
11604 (output_a_rotate): Likewise.
11605 * config/i386/i386.c (classify_argument): Likewise.
11606 (function_arg_advance_32): Likewise.
11607 (function_arg_32): Likewise.
11608 (function_arg_64): Likewise.
11609 (function_value_64): Likewise.
11610 (ix86_gimplify_va_arg): Likewise.
11611 (ix86_legitimate_constant_p): Likewise.
11612 (put_condition_code): Likewise.
11613 (split_double_mode): Likewise.
11614 (ix86_avx256_split_vector_move_misalign): Likewise.
11615 (ix86_expand_vector_logical_operator): Likewise.
11616 (ix86_split_idivmod): Likewise.
11617 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
11618 (ix86_build_const_vector): Likewise.
11619 (ix86_build_signbit_mask): Likewise.
11620 (ix86_match_ccmode): Likewise.
11621 (ix86_cc_modes_compatible): Likewise.
11622 (ix86_expand_branch): Likewise.
11623 (ix86_expand_sse_cmp): Likewise.
11624 (ix86_expand_sse_movcc): Likewise.
11625 (ix86_expand_int_sse_cmp): Likewise.
11626 (ix86_expand_vec_perm_vpermi2): Likewise.
11627 (ix86_expand_vec_perm): Likewise.
11628 (ix86_expand_sse_unpack): Likewise.
11629 (ix86_expand_int_addcc): Likewise.
11630 (ix86_split_to_parts): Likewise.
11631 (ix86_vectorize_builtin_gather): Likewise.
11632 (ix86_vectorize_builtin_scatter): Likewise.
11633 (avx_vpermilp_parallel): Likewise.
11634 (inline_memory_move_cost): Likewise.
11635 (ix86_tieable_integer_mode_p): Likewise.
11636 (x86_maybe_negate_const_int): Likewise.
11637 (ix86_expand_vector_init_duplicate): Likewise.
11638 (ix86_expand_vector_init_one_nonzero): Likewise.
11639 (ix86_expand_vector_init_one_var): Likewise.
11640 (ix86_expand_vector_init_concat): Likewise.
11641 (ix86_expand_vector_init_interleave): Likewise.
11642 (ix86_expand_vector_init_general): Likewise.
11643 (ix86_expand_vector_set): Likewise.
11644 (ix86_expand_vector_extract): Likewise.
11645 (emit_reduc_half): Likewise.
11646 (ix86_emit_i387_round): Likewise.
11647 (ix86_mangle_type): Likewise.
11648 (ix86_expand_round_sse4): Likewise.
11649 (expand_vec_perm_blend): Likewise.
11650 (canonicalize_vector_int_perm): Likewise.
11651 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
11652 (expand_vec_perm_1): Likewise.
11653 (expand_vec_perm_interleave3): Likewise.
11654 (expand_vec_perm_even_odd_pack): Likewise.
11655 (expand_vec_perm_even_odd_1): Likewise.
11656 (expand_vec_perm_broadcast_1): Likewise.
11657 (ix86_vectorize_vec_perm_const_ok): Likewise.
11658 (ix86_expand_vecop_qihi): Likewise.
11659 (ix86_expand_mul_widen_hilo): Likewise.
11660 (ix86_expand_sse2_abs): Likewise.
11661 (ix86_expand_pextr): Likewise.
11662 (ix86_expand_pinsr): Likewise.
11663 (ix86_preferred_simd_mode): Likewise.
11664 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
11665 * config/i386/sse.md (*andnot<mode>3): Likewise.
11666 (<mask_codefor><code><mode>3<mask_name>): Likewise.
11667 (*<code><mode>3): Likewise.
11668 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
11669 (ia64_expand_atomic_op): Likewise.
11670 (ia64_arg_type): Likewise.
11671 (ia64_mode_to_int): Likewise.
11672 (ia64_scalar_mode_supported_p): Likewise.
11673 (ia64_vector_mode_supported_p): Likewise.
11674 (expand_vec_perm_broadcast): Likewise.
11675 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
11676 (iq2000_function_arg_advance): Likewise.
11677 (iq2000_function_arg): Likewise.
11678 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
11679 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
11680 (m68k_libcall_value): Likewise.
11681 (m68k_function_value): Likewise.
11682 (sched_attr_op_type): Likewise.
11683 * config/mcore/mcore.c (mcore_output_move): Likewise.
11684 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
11686 (microblaze_function_arg): Likewise.
11687 * config/mips/mips.c (mips16_build_call_stub): Likewise.
11688 (mips_print_operand): Likewise.
11689 (mips_mode_ok_for_mov_fmt_p): Likewise.
11690 (mips_vector_mode_supported_p): Likewise.
11691 (mips_preferred_simd_mode): Likewise.
11692 (mips_expand_vpc_loongson_even_odd): Likewise.
11693 (mips_expand_vec_unpack): Likewise.
11694 (mips_expand_vi_broadcast): Likewise.
11695 (mips_expand_vector_init): Likewise.
11696 (mips_expand_vec_reduc): Likewise.
11697 (mips_expand_msa_cmp): Likewise.
11698 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
11699 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
11700 (cc_flags_for_mode): Likewise.
11701 * config/msp430/msp430.c (msp430_print_operand): Likewise.
11702 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
11703 (nds32_output_casesi_pc_relative): Likewise.
11704 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11705 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
11706 (nvptx_gen_unpack): Likewise.
11707 (nvptx_gen_pack): Likewise.
11708 (nvptx_gen_shuffle): Likewise.
11709 (nvptx_gen_wcast): Likewise.
11710 (nvptx_preferred_simd_mode): Likewise.
11711 * config/pa/pa.c (pa_secondary_reload): Likewise.
11712 * config/pa/predicates.md (base14_operand): Likewise.
11713 * config/powerpcspe/powerpcspe-c.c
11714 (altivec_resolve_overloaded_builtin): Likewise.
11715 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
11717 (rs6000_preferred_simd_mode): Likewise.
11718 (output_vec_const_move): Likewise.
11719 (rs6000_expand_vector_extract): Likewise.
11720 (rs6000_split_vec_extract_var): Likewise.
11721 (reg_offset_addressing_ok_p): Likewise.
11722 (rs6000_legitimate_offset_address_p): Likewise.
11723 (rs6000_legitimize_address): Likewise.
11724 (rs6000_emit_set_const): Likewise.
11725 (rs6000_const_vec): Likewise.
11726 (rs6000_emit_move): Likewise.
11727 (spe_build_register_parallel): Likewise.
11728 (rs6000_darwin64_record_arg_recurse): Likewise.
11729 (swap_selector_for_mode): Likewise.
11730 (spe_init_builtins): Likewise.
11731 (paired_init_builtins): Likewise.
11732 (altivec_init_builtins): Likewise.
11733 (do_load_for_compare): Likewise.
11734 (rs6000_generate_compare): Likewise.
11735 (rs6000_expand_float128_convert): Likewise.
11736 (emit_load_locked): Likewise.
11737 (emit_store_conditional): Likewise.
11738 (rs6000_output_function_epilogue): Likewise.
11739 (rs6000_handle_altivec_attribute): Likewise.
11740 (rs6000_function_value): Likewise.
11741 (emit_fusion_gpr_load): Likewise.
11742 (emit_fusion_p9_load): Likewise.
11743 (emit_fusion_p9_store): Likewise.
11744 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
11745 (fusion_gpr_mem_load): Likewise.
11746 (fusion_addis_mem_combo_load): Likewise.
11747 (fusion_addis_mem_combo_store): Likewise.
11748 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
11749 (fusion_gpr_mem_load): Likewise.
11750 (fusion_addis_mem_combo_load): Likewise.
11751 (fusion_addis_mem_combo_store): Likewise.
11752 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11754 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
11755 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
11756 (rs6000_preferred_simd_mode): Likewise.
11757 (output_vec_const_move): Likewise.
11758 (rs6000_expand_vector_extract): Likewise.
11759 (rs6000_split_vec_extract_var): Likewise.
11760 (reg_offset_addressing_ok_p): Likewise.
11761 (rs6000_legitimate_offset_address_p): Likewise.
11762 (rs6000_legitimize_address): Likewise.
11763 (rs6000_emit_set_const): Likewise.
11764 (rs6000_const_vec): Likewise.
11765 (rs6000_emit_move): Likewise.
11766 (rs6000_darwin64_record_arg_recurse): Likewise.
11767 (swap_selector_for_mode): Likewise.
11768 (paired_init_builtins): Likewise.
11769 (altivec_init_builtins): Likewise.
11770 (rs6000_expand_float128_convert): Likewise.
11771 (emit_load_locked): Likewise.
11772 (emit_store_conditional): Likewise.
11773 (rs6000_output_function_epilogue): Likewise.
11774 (rs6000_handle_altivec_attribute): Likewise.
11775 (rs6000_function_value): Likewise.
11776 (emit_fusion_gpr_load): Likewise.
11777 (emit_fusion_p9_load): Likewise.
11778 (emit_fusion_p9_store): Likewise.
11779 * config/rx/rx.c (rx_gen_move_template): Likewise.
11780 (flags_from_mode): Likewise.
11781 * config/s390/predicates.md (s390_alc_comparison): Likewise.
11782 (s390_slb_comparison): Likewise.
11783 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
11784 (s390_vector_mode_supported_p): Likewise.
11785 (s390_cc_modes_compatible): Likewise.
11786 (s390_match_ccmode_set): Likewise.
11787 (s390_canonicalize_comparison): Likewise.
11788 (s390_emit_compare_and_swap): Likewise.
11789 (s390_branch_condition_mask): Likewise.
11790 (s390_rtx_costs): Likewise.
11791 (s390_secondary_reload): Likewise.
11792 (__SECONDARY_RELOAD_CASE): Likewise.
11793 (s390_expand_cs): Likewise.
11794 (s390_preferred_simd_mode): Likewise.
11795 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
11796 * config/sh/sh.c (sh_print_operand): Likewise.
11797 (dump_table): Likewise.
11798 (sh_secondary_reload): Likewise.
11799 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11800 * config/sh/sh.md (casesi_worker_1): Likewise.
11801 (casesi_worker_2): Likewise.
11802 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
11803 (fcc_comparison_operator): Likewise.
11804 * config/sparc/sparc.c (sparc_expand_move): Likewise.
11805 (emit_soft_tfmode_cvt): Likewise.
11806 (sparc_preferred_simd_mode): Likewise.
11807 (output_cbranch): Likewise.
11808 (sparc_print_operand): Likewise.
11809 (sparc_expand_vec_perm_bmask): Likewise.
11810 (vector_init_bshuffle): Likewise.
11811 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
11812 (spu_vector_mode_supported_p): Likewise.
11813 (spu_expand_insv): Likewise.
11814 (spu_emit_branch_or_set): Likewise.
11815 (spu_handle_vector_attribute): Likewise.
11816 (spu_builtin_splats): Likewise.
11817 (spu_builtin_extract): Likewise.
11818 (spu_builtin_promote): Likewise.
11819 (spu_expand_sign_extend): Likewise.
11820 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
11821 (tilegx_simd_int): Likewise.
11822 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
11823 (tilepro_simd_int): Likewise.
11824 * config/v850/v850.c (const_double_split): Likewise.
11825 (v850_print_operand): Likewise.
11826 (ep_memory_offset): Likewise.
11827 * config/vax/vax.c (vax_rtx_costs): Likewise.
11828 (vax_output_int_move): Likewise.
11829 (vax_output_int_add): Likewise.
11830 (vax_output_int_subtract): Likewise.
11831 * config/visium/predicates.md (visium_branch_operator): Likewise.
11832 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
11833 (visium_print_operand_address): Likewise.
11834 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11835 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
11836 (xtensa_expand_conditional_branch): Likewise.
11837 (xtensa_copy_incoming_a7): Likewise.
11838 (xtensa_output_literal): Likewise.
11839 * dfp.c (decimal_real_maxval): Likewise.
11840 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
11842 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11843 Alan Hayward <alan.hayward@arm.com>
11844 David Sherwood <david.sherwood@arm.com>
11846 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
11847 (mode_nunits_inline): Likewise.
11848 (mode_inner_inline): Likewise.
11849 (mode_unit_size_inline): Likewise.
11850 (mode_unit_precision_inline): Likewise.
11851 (emit_insn_modes_h): Likewise. Also emit a #define of the
11853 (emit_mode_wider): Add an E_ prefix to mode names.
11854 (emit_mode_complex): Likewise.
11855 (emit_mode_inner): Likewise.
11856 (emit_mode_adjustments): Likewise.
11857 (emit_mode_int_n): Likewise.
11858 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
11859 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
11860 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
11861 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
11862 (CRC32_BUILTIN, ENTRY): Likewise.
11863 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
11864 (aarch64_pop_regs): Likewise.
11865 (aarch64_process_components): Likewise.
11866 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
11867 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
11868 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
11869 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
11870 * config/arm/arm.c (arm_init_libfuncs): Likewise.
11871 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
11873 * config/i386/i386-builtin.def (pcmpestr): Likewise.
11874 (pcmpistr): Likewise.
11875 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
11876 * config/mmix/mmix.c (mmix_output_condition): Likewise.
11877 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
11879 * config/rl78/rl78.c (mduc_regs): Likewise.
11880 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
11881 (htm_expand_builtin): Likewise.
11882 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
11883 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
11884 * config/xtensa/xtensa.c (print_operand): Likewise.
11885 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
11886 (NUM_MODE_VECTOR_INT): Likewise.
11887 * genoutput.c (null_operand): Likewise.
11888 (output_operand_data): Likewise.
11889 * genrecog.c (print_parameter_value): Likewise.
11890 * lra.c (debug_operand_data): Likewise.
11892 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11894 * dwarf2out.c (scompare_loc_descriptor_wide)
11895 (scompare_loc_descriptor_narrow): New functions, split out from...
11896 (scompare_loc_descriptor): ...here.
11897 * expmed.c (emit_store_flag_int): New function, split out from...
11898 (emit_store_flag): ...here.
11900 2017-08-30 Richard Biener <rguenther@suse.de>
11902 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
11903 (dwarf2out_early_finish): Move setting of AT_pubnames from
11904 early debug output to early finish.
11906 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11908 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
11909 and -mdata-region to the assembler.
11911 2017-08-30 Richard Biener <rguenther@suse.de>
11913 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
11915 (gen_subprogram_die): Add DW_AT_object_pointer only early.
11916 (dwarf2out_early_global_decl): Only generate a DIE for the
11917 abstract origin if it doesn't already exist or is a declaration DIE.
11918 (resolve_addr): Do not add the linkage name twice when
11919 generating a stub DIE for the DW_TAG_GNU_call_site target.
11921 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11923 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11924 Use machine_mode rather than int for arg1_mode.
11926 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
11929 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
11930 that the second argument of the built-in functions to unpack
11931 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
11932 switch statement instead a lot of if statements.
11933 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
11934 Allow 64-bit values to be in Altivec registers as well as
11935 traditional floating point registers.
11936 (pack<mode>, FMOVE128_VSX iterator): Likewise.
11938 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
11940 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
11941 MAX_REGS_PER_ADDRESS == 1.
11943 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
11945 * config/i386/i386.opt (flag_fentry): Do not init to -1.
11946 * config/i386/i386.c (ix86_option_override_internal): Simplify
11947 setting of opts->x_flag_entry.
11949 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11950 Jakub Jelinek <jakub@redhat.com>
11951 Richard Biener <rguenther@suse.de>
11953 PR tree-optimization/81503
11954 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
11955 folded constant fits in the target type; reorder tests for clarity.
11957 2017-08-29 Martin Liska <mliska@suse.cz>
11959 * passes.def: Include pass_lower_switch.
11960 * stmt.c (dump_case_nodes): Remove and move to
11961 tree-switch-conversion.
11962 (case_values_threshold): Likewise.
11963 (expand_switch_as_decision_tree_p): Likewise.
11964 (emit_case_decision_tree): Likewise.
11965 (expand_case): Likewise.
11966 (balance_case_nodes): Likewise.
11967 (node_has_low_bound): Likewise.
11968 (node_has_high_bound): Likewise.
11969 (node_is_bounded): Likewise.
11970 (emit_case_nodes): Likewise.
11971 (struct simple_case_node): New struct.
11972 (add_case_node): Remove.
11973 (emit_case_dispatch_table): Use vector instead of case_list.
11974 (reset_out_edges_aux): Remove.
11975 (compute_cases_per_edge): Likewise.
11976 (expand_case): Build list of simple_case_node.
11977 (expand_sjlj_dispatch_table): Use it.
11978 * tree-switch-conversion.c (struct case_node): Moved from
11979 stmt.c and adjusted.
11980 (emit_case_nodes): Likewise.
11981 (node_has_low_bound): Likewise.
11982 (node_has_high_bound): Likewise.
11983 (node_is_bounded): Likewise.
11984 (case_values_threshold): Likewise.
11985 (reset_out_edges_aux): Likewise.
11986 (compute_cases_per_edge): Likewise.
11987 (add_case_node): Likewise.
11988 (dump_case_nodes): Likewise.
11989 (balance_case_nodes): Likewise.
11990 (expand_switch_as_decision_tree_p): Likewise.
11991 (emit_jump): Likewise.
11992 (emit_case_decision_tree): Likewise.
11993 (try_switch_expansion): Likewise.
11994 (do_jump_if_equal): Likewise.
11995 (emit_cmp_and_jump_insns): Likewise.
11996 (fix_phi_operands_for_edge): New function.
11997 (record_phi_operand_mapping): Likewise.
11998 (class pass_lower_switch): New pass.
11999 (pass_lower_switch::execute): New function.
12000 (make_pass_lower_switch): Likewise.
12001 (conditional_probability):
12002 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
12003 * tree-pass.h: Add make_pass_lower_switch.
12005 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12008 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
12011 2017-08-29 Richard Biener <rguenther@suse.de>
12013 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
12014 we do not add a DW_AT_inline attribute twice.
12015 (gen_subprogram_die): Remove code setting DW_AT_inline on
12016 DECL_ABSTRACT_P nodes.
12018 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
12020 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
12021 calls to internal functions.
12022 (gimplify_modify_expr): Likewise.
12023 * tree-call-cdce.c (use_internal_fn): Likewise.
12024 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
12025 (convert_to_divmod): Set the nothrow flag.
12026 * tree-if-conv.c (predicate_mem_writes): Likewise.
12027 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
12028 (vectorizable_call): Likewise.
12029 (vectorizable_store): Likewise.
12030 (vectorizable_load): Likewise.
12031 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
12032 (vect_recog_mask_conversion_pattern): Likewise.
12034 2017-08-29 Martin Liska <mliska@suse.cz>
12037 * gcc.c (driver_handle_option): Add new argument.
12038 * opts-common.c (handle_option): Pass
12039 target_option_override_hook.
12040 * opts-global.c (lang_handle_option): Add new option.
12041 (set_default_handlers): Add new argument.
12042 (decode_options): Likewise.
12043 * opts.c (target_handle_option): Likewise.
12044 (common_handle_option): Call target_option_override_hook.
12045 * opts.h (struct cl_option_handler_func): Add hook for
12046 target option override.
12047 (struct cl_option_handlers): Likewise.
12048 (set_default_handlers): Add new argument.
12049 (decode_options): Likewise.
12050 (common_handle_option): Likewise.
12051 (target_handle_option): Likewise.
12052 * toplev.c (toplev::main): Pass targetm.target_option.override
12055 2017-08-29 Richard Biener <rguenther@suse.de>
12056 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
12058 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
12059 life to the active subtree.
12061 2017-08-28 Jeff Law <law@redhat.com>
12063 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
12064 derive_equivalences.
12065 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
12066 Code moved into....
12067 (edge_info::derive_equivalences): New private member function
12069 * tree-ssa-dom.c (class edge_info): Changed from a struct
12070 to a class. Add ctor/dtor, methods and data members.
12071 (edge_info::edge_info): Renamed from allocate_edge_info.
12072 Initialize additional members.
12073 (edge_info::~edge_info): New.
12074 (free_dom_edge_info): Delete the edge info.
12075 (record_edge_info): Use new class & associated member functions.
12076 Tighten forms for testing for edge equivalences.
12077 (record_temporary_equivalences): Iterate over the simple
12078 equivalences rather than assuming there's only one per edge.
12079 (cprop_into_successor_phis): Iterate over the simple
12080 equivalences rather than assuming there's only one per edge.
12081 (optimize_stmt): Use operand_equal_p rather than pointer
12082 equality for mini-DSE code.
12084 2017-08-28 Nathan Sidwell <nathan@acm.org>
12086 * gcc.c (execute): Fold SIGPIPE handling into switch
12087 statement. Adjust internal error message.
12089 2017-08-28 Richard Biener <rguenther@suse.de>
12092 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
12093 Do nothing for removed DIEs.
12095 2017-08-28 Richard Biener <rguenther@suse.de>
12097 PR tree-optimization/81977
12098 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
12101 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
12104 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
12105 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
12106 using targetm.gen_mem_thread_fence.
12108 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
12111 * config/i386/i386.md (*<btsc><mode>): Change operand 2
12112 predicate to register_operand. Reorder operands.
12113 (*btr<mode>): Ditto.
12114 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
12115 (*btr<mode>_mask): Ditto.
12117 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
12119 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
12120 * config/rs6000/xmmintrin.h: New file.
12121 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
12123 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12126 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
12127 parameter and_insn and return it.
12128 (recombine_lvx_pattern): Insert a copy to ensure availability of
12129 the base register of the copied masking operation at the point of
12130 the instruction replacement.
12131 (recombine_stvx_pattern): Likewise.
12133 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
12135 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
12136 undocumented switches.
12137 (-mpower9-dform-vector): Likewise.
12138 (-mpower9-dform): Likewise.
12139 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
12140 comments to delete references to -mpower9-dform* switches.
12141 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
12142 Delete reference to -mpower9-dform* switches, test for
12143 -mpower9-vector instead.
12144 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
12145 (OTHER_P9_VECTOR_MASKS): Likewise.
12146 (POWERPC_MASKS): Likewise.
12147 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
12148 tests against -mpower9-dform* to -mpower9-vector. Delete code
12149 that checked for -mpower9-dform* consistancy with other options.
12150 Add test for -mpower9-misc to enable other power9 switches.
12151 (rs6000_init_hard_regno_mode_ok): Likewise.
12152 (rs6000_option_override_internal): Likewise.
12153 (rs6000_emit_prologue): Likewise.
12154 (rs6000_emit_epilogue): Likewise.
12155 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
12156 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
12157 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
12159 (emit_fusion_p9_store): Likewise.
12160 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
12161 resetting these macros if the assembler does not support ISA 3.0
12163 (TARGET_P9_DFORM_VECTOR): Likewise.
12164 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
12165 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
12167 2017-08-25 Alan Modra <amodra@gmail.com>
12169 PR rtl-optimization/81747
12170 * cse.c (cse_extended_basic_block): Don't attempt to record
12171 equivalences for degenerate conditional branches that branch
12172 to their fall-through.
12174 2017-08-24 Martin Sebor <msebor@redhat.com>
12176 PR middle-end/81908
12177 * gimple-fold.c (size_must_be_zero_p): New function.
12178 (gimple_fold_builtin_memory_op): Call it.
12180 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
12182 * config/rs6000/mm_malloc.h: New file.
12184 2017-08-24 Bin Cheng <bin.cheng@arm.com>
12186 PR tree-optimization/81913
12187 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
12188 analysis when either IVs in condition can wrap.
12190 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
12192 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
12193 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
12195 2017-08-24 Richard Biener <rguenther@suse.de>
12198 * targhooks.c (default_target_can_inline_p): Properly
12199 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
12200 is present and always compare.
12201 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
12202 infer -mfpmath=sse from TARGET_SSE_P.
12203 (ix86_can_inline_p): Properly use target_option_default_node when
12204 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
12206 2017-08-24 Richard Biener <rguenther@suse.de>
12209 * dwarf2out.c (output_die): Handle flag_generate_offload like
12211 (output_comp_unit): Likewise.
12212 (gen_array_type_die): Likewise.
12213 (dwarf2out_early_finish): Likewise.
12214 (note_variable_value_in_expr): Likewise.
12215 (dwarf2out_finish): Likewise. Adjust assert.
12216 * cgraphunit.c (symbol_table::compile): Move setting of
12217 flag_generate_offload earlier ...
12218 (symbol_table::finalize_compilation_unit): ... here, before
12219 early debug finalization.
12221 2017-08-24 Richard Biener <rguenther@suse.de>
12223 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
12224 and ipa-fnsummary.h.
12225 (ix86_can_inline_p): When ix86_fpmath flags do not match
12226 check whether the callee uses FP math at all.
12228 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
12230 PR middle-end/81931
12231 * tree-ssanames.c (get_nonzero_bits): Use element_precision
12232 instead of TYPE_PRECISION.
12234 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
12235 Alan Hayward <alan.hayward@arm.com>
12236 David Sherwood <david.sherwood@arm.com>
12238 * combine.c (make_extraction): Use subreg_offset_from_lsb.
12240 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
12242 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
12244 (ix86_frame::stack_realign_allocate): New field.
12245 (struct machine_frame_state): Modify comments.
12246 (machine_frame_state::sp_realigned_fp_end): New field.
12247 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
12248 layout calculation.
12249 (sp_valid_at): Add assertion to assure no attempt to access invalid
12250 offset of a realigned stack.
12251 (fp_valid_at): Likewise.
12252 (choose_baseaddr): Modify comments.
12253 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
12254 ix86_expand_prologue.
12255 (ix86_expand_prologue): Modify stack realignment and allocation.
12256 (ix86_expand_epilogue): Modify comments.
12257 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
12258 avx2_runtime, avx512f, and avx512f_runtime.
12260 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
12262 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
12263 (mstackrealign): Do not init to -1.
12264 * config/i386/i386.c (ix86_option_override_internal):
12265 Check opts_set, not opts when setting default value of
12266 opts->x_ix86_force_align_arg_pointer.
12268 2017-08-23 Richard Biener <rguenther@suse.de>
12270 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
12271 lang_hooks.decl_printable_name.
12272 * print-rtl-function.c (print_rtx_function): Likewise.
12273 * tree-pretty-print.c (dump_function_header): Likewise.
12275 2017-08-23 Richard Biener <rguenther@suse.de>
12278 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
12279 -g0 at compile-time.
12281 2017-08-23 Tamar Christina <tamar.christina@arm.com>
12283 PR middle-end/19706
12284 * doc/sourcebuild.texi (Other hardware attributes):
12287 2017-08-23 Tamar Christina <tamar.christina@arm.com>
12289 PR middle-end/19706
12290 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
12291 Fix single-use check.
12293 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
12295 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
12297 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
12299 * doc/install.texi: Modify to add more details on running selected
12302 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
12304 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
12305 is combined with -mabi=ms.
12306 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
12309 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12311 PR tree-optimization/81488
12312 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
12313 and cached_basis fields.
12314 (MAX_SPREAD): New constant.
12315 (alloc_cand_and_find_basis): Initialize new fields.
12316 (clear_visited): New function.
12317 (create_phi_basis_1): Rename from create_phi_basis, set visited
12318 and cached_basis fields.
12319 (create_phi_basis): New wrapper function.
12320 (phi_add_costs_1): Rename from phi_add_costs, add spread
12321 parameter, set visited field, short-circuit when limits reached.
12322 (phi_add_costs): New wrapper function.
12323 (record_phi_increments_1): Rename from record_phi_increments, set
12325 (record_phi_increments): New wrapper function.
12326 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
12327 (phi_incr_cost): New wrapper function.
12328 (all_phi_incrs_profitable_1): Rename from
12329 all_phi_incrs_profitable, set visited field.
12330 (all_phi_incrs_profitable): New wrapper function.
12332 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
12333 Alan Hayward <alan.hayward@arm.com>
12334 David Sherwood <david.sherwood@arm.com>
12336 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
12337 that takes the outer and inner modes.
12338 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
12339 comparison as the canonical test for a paradoxical subreg.
12340 * combine.c (simplify_set): Use paradoxical_subreg_p.
12341 (make_extraction): Likewise.
12342 (force_to_mode): Likewise.
12343 (rtx_equal_for_field_assignment_p): Likewise.
12344 (gen_lowpart_for_combine): Likewise.
12345 (simplify_comparison): Likewise.
12346 * cse.c (equiv_constant): Likewise.
12347 * expmed.c (store_bit_field_1): Likewise.
12348 * final.c (alter_subreg): Likewise.
12349 * fwprop.c (propagate_rtx): Likewise.
12350 (forward_propagate_subreg): Likewise.
12351 * ira-conflicts.c (ira_build_conflicts): Likewise.
12352 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
12353 * lra-constraints.c (curr_insn_transform): Likewise.
12354 (split_reg): Likewise.
12355 * lra-eliminations.c (move_plus_up): Likewise.
12356 (lra_eliminate_regs_1): Likewise.
12357 * recog.c (general_operand): Likewise.
12358 * ree.c (combine_reaching_defs): Likewise.
12359 * reload.c (push_reload): Likewise.
12360 (find_reloads): Likewise.
12361 * reload1.c (elimination_effects): Likewise.
12362 (compute_reload_subreg_offset): Likewise.
12363 (choose_reload_regs): Likewise.
12364 * rtlanal.c (subreg_lsb_1): Likewise.
12365 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12366 (simplify_subreg): Likewise.
12367 * var-tracking.c (track_loc_p): Likewise.
12368 * emit-rtl.c (byte_lowpart_offset): Likewise.
12369 (paradoxical_subreg_p): Delete out-of-line definition.
12371 2017-08-22 Jeff Law <law@redhat.com>
12373 PR tree-optimization/81741
12374 PR tree-optimization/71947
12375 * tree-ssa-dom.c: Include tree-inline.h.
12376 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
12377 equivalences if one is more expensive to compute than the other.
12378 * tree-ssa-scopedtables.h (class const_or_copies): Make
12379 record_const_or_copy_raw method private.
12380 (class avail_exprs_stack): New method simplify_binary_operation.
12381 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
12382 avail_exprs_stack::simplify_binary_operation as needed.
12383 (avail_exprs_stack::simplify_binary_operation): New function.
12385 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
12387 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
12388 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
12389 (DOT_SYMBOLS): Likewise.
12390 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
12391 (RELOCATABLE_NEEDS_FIXUP): Likewise.
12392 (RS6000_ABI_NAME): Likewise.
12393 (TARGET_CMODEL): Likewise.
12394 (TOC_SECTION_ASM_OP): Likewise.
12395 (SET_CMODEL): New macro.
12396 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
12398 2017-08-22 Richard Biener <rguenther@suse.de>
12400 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
12401 to allow for free-lang-data replacements similar to verify_type_variant.
12403 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
12404 Alan Hayward <alan.hayward@arm.com>
12405 David Sherwood <david.sherwood@arm.com>
12407 * config/aarch64/aarch64.md (casesi): Use DImode rather than
12408 VOIDmode for the LABEL_REF.
12410 2017-08-22 Richard Biener <rguenther@suse.de>
12412 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
12414 2017-08-22 Richard Biener <rguenther@suse.de>
12416 * common.opt (feliminate-dwarf2-dups): Ignore.
12417 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
12418 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
12419 same_die_p_wrap, compute_section_prefix,
12420 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
12421 (comdat_symbol_id, comdat_symbol_number): Likewise.
12422 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
12424 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
12425 (output_die): Mark unreachable path unreachable.
12426 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
12427 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
12428 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
12429 (dwarf2out_early_finish): Likewise.
12431 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
12433 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
12435 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
12438 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
12439 not VAR_P. Filter attribute warnings with OPT_Wattributes.
12440 (avr_attribute_table) <io, io_low, address>: Initialize
12441 .decl_required with true.
12443 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
12445 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
12446 undocumented debugging options.
12447 (-mvsx-scalar-double): Likewise.
12448 (-mallow-df-permute): Likewise.
12449 (-mvectorize-builtins): Likewise.
12450 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12451 (rs6000_builtin_vectorized_function): Likewise.
12452 (rs6000_builtin_md_vectorized_function): Likewise.
12453 (rs6000_opt_vars): Likewise.
12455 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
12458 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
12459 (*btrq_imm): Rename from *btrq.
12460 (*btcq_imm): Rename from *btcq.
12461 (btsc): New code attribute.
12462 (*<btsc><mode>): New insn pattern.
12463 (*btr<mode>): Ditto.
12464 (*<btsc><mode>_mask): New insn_and_split pattern.
12465 (*btr<mode>_mask): Ditto.
12467 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12468 Alan Hayward <alan.hayward@arm.com>
12469 David Sherwood <david.sherwood@arm.com>
12471 * function.c (pad_below): Simplify padding calculation.
12473 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12474 Alan Hayward <alan.hayward@arm.com>
12475 David Sherwood <david.sherwood@arm.com>
12477 * target.def (function_prologue): Remove frame size argument.
12478 (function_epilogue): Likewise.
12479 * doc/tm.texi: Regenerate.
12480 * final.c (final_start_function): Update call to function_prologue.
12481 (final_end_function): Update call to function_epilogue.
12482 (default_function_pro_epilogue): Remove frame size argument.
12483 * output.h (default_function_pro_epilogue): Likewise.
12484 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
12485 (arm_output_function_prologue): Likewise.
12486 * config/frv/frv.c (frv_function_prologue): Likewise.
12487 (frv_function_epilogue): Likewise.
12488 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
12489 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
12490 (ia64_output_function_epilogue): Likewise.
12491 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
12492 (m32r_output_function_epilogue): Likewise.
12493 * config/microblaze/microblaze.c (microblaze_function_prologue)
12494 (microblaze_function_epilogue): Likewise.
12495 * config/mips/mips.c (mips_output_function_prologue): Likewise.
12496 (mips_output_function_epilogue): Likewise.
12497 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
12498 (mmix_target_asm_function_epilogue): Likewise.
12499 * config/msp430/msp430.c (msp430_start_function): Likewise.
12500 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
12501 (nds32_asm_function_epilogue): Likewise.
12502 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
12503 * config/pa/pa.c (pa_output_function_prologue): Likewise.
12504 (pa_output_function_epilogue): Likewise.
12505 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
12506 (rs6000_output_function_epilogue): Likewise.
12507 * config/rl78/rl78.c (rl78_start_function): Likewise.
12508 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
12509 (rs6000_output_function_epilogue): Likewise.
12510 * config/rx/rx.c (rx_output_function_prologue): Likewise.
12511 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
12512 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
12513 (sparc_asm_function_epilogue): Likewise.
12515 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12517 * tree.h (type_has_mode_precision_p): New function.
12518 * convert.c (convert_to_integer_1): Use it.
12519 * expr.c (expand_expr_real_2): Likewise.
12520 (expand_expr_real_1): Likewise.
12521 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
12522 * match.pd: Likewise.
12523 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
12524 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
12525 * tree-tailcall.c (process_assignment): Likewise.
12526 * tree-vect-loop.c (vectorizable_reduction): Likewise.
12527 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
12528 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
12529 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12530 (vectorizable_assignment): Likewise.
12531 (vectorizable_shift): Likewise.
12532 (vectorizable_operation): Likewise.
12533 * tree-vrp.c (register_edge_assert_for_2): Likewise.
12535 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
12537 * match.pd: Add pow (C, x) simplification.
12539 2017-08-21 Richard Biener <rguenther@suse.de>
12541 PR tree-optimization/81900
12542 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
12543 for blocks with abnormal predecessors.
12544 (compute_antic): Do not set visited flag prematurely.
12546 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
12549 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
12551 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12553 * stor-layout.h (vector_type_mode): Move to...
12554 * tree.h (vector_type_mode): ...here.
12555 * stor-layout.c (vector_type_mode): Move to...
12556 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
12558 2017-08-21 Richard Biener <rguenther@suse.de>
12560 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
12561 register_external_die hooks.
12562 (debug_false_tree_charstarstar_uhwistar): Declare.
12563 (debug_nothing_tree_charstar_uhwi): Likewise.
12564 * debug.c (do_nothing_debug_hooks): Adjust.
12565 (debug_false_tree_charstarstar_uhwistar): New do nothing.
12566 (debug_nothing_tree_charstar_uhwi): Likewise.
12567 * dbxout.c (dbx_debug_hooks): Adjust.
12568 (xcoff_debug_hooks): Likewise.
12569 * sdbout.c (sdb_debug_hooks): Likewise.
12570 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
12571 * dwarf2out.c (macinfo_label_base): New global.
12572 (dwarf2out_register_external_die): New function for the
12573 register_external_die hook.
12574 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
12575 (dwarf2_debug_hooks): Use them.
12576 (dwarf2_lineno_debug_hooks): Adjust.
12577 (struct die_struct): Add with_offset flag.
12578 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
12579 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
12580 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
12581 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
12582 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
12583 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
12584 defining section names for the early LTO debug variants.
12585 (reset_indirect_string): New helper.
12586 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
12587 (print_dw_val): Add support for offsetted symbol references.
12588 (get_ultimate_context): Split out from is_cxx.
12589 (is_cxx): Use get_ultimate_context.
12590 (is_fortran): Add decl overload.
12591 (compute_comp_unit_symbol): Split out worker from
12592 compute_section_prefix.
12593 (compute_section_prefix): Call compute_comp_unit_symbol and
12594 set comdat_type_p here.
12595 (output_die): Skip DIE symbol output for the LTO added one.
12596 Handle DIE symbol references with offset.
12597 (output_comp_unit): Guard section name mangling properly.
12598 For LTO debug sections emit a symbol at the section beginning
12599 which we use to refer to its DIEs.
12600 (add_abstract_origin_attribute): For DIEs registered via
12601 dwarf2out_register_external_die directly refer to the early
12602 DIE rather than indirectly through the shadow one we created.
12603 Remove obsolete call to dwarf2out_abstract_function for
12604 non-function/block origins.
12605 (gen_array_type_die): When generating early LTO debug do
12606 not emit DW_AT_string_length.
12607 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
12608 late when in LTO. As suggested place a gcc_unreachable for
12609 the DECL_ABSTRACT_P case.
12610 (gen_subprogram_die): Avoid another specification DIE
12611 for early built declarations/definitions for the late LTO case.
12612 (gen_variable_die): Add type references for late duplicated VLA dies
12614 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
12615 we have the abstract instance already.
12616 (process_scope_var): Adjust decl DIE contexts in LTO which
12617 first puts them in limbo.
12618 (gen_decl_die): Do not generate type DIEs late apart from
12619 types for VLAs or for decls we do not yet have a DIE. Do not
12620 call dwarf2out_abstract_function late.
12621 (dwarf2out_early_global_decl): Make sure to create DIEs
12622 for abstract instances of a decl first.
12623 (dwarf2out_late_global_decl): Adjust comment.
12624 (output_macinfo_op): With multiple macro sections use
12625 macinfo_label_base to distinguish labels.
12626 (output_macinfo): Likewise. Update macinfo_label_base.
12627 Pass in the line info label.
12628 (note_variable_value_in_expr): When generating LTO resolve
12629 all variable values here by generating DIEs as needed.
12630 (init_sections_and_labels): Add early LTO debug flag parameter
12631 and generate different sections and names if set. Add generation
12632 counter for the labels so we can have multiple of them.
12633 (reset_dies): Helper to allow DIEs to be output multiple times.
12634 (dwarf2out_finish): When outputting DIEs to the fat part of an
12635 LTO object first reset DIEs.
12636 (dwarf2out_early_finish): Output early DIEs when generating LTO.
12637 (modified_type_die): Check for decl_ultimate_origin being self
12639 (gen_type_die_with_usage): Likewise.
12640 (gen_typedef_die): Allow decl_ultimate_origin being self.
12641 (set_decl_abstract_flags): Remove.
12642 (set_block_abstract_flags): Likewise.
12643 (dwarf2out_abstract_function): Treat the early generated DIEs
12644 as the abstract copy and only add DW_AT_inline and
12645 DW_AT_artificial here and call set_decl_origin_self.
12646 If the DIE has an abstract origin don't do anything.
12647 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
12648 if we have none yet (Go fails to build one, PR78628).
12649 (variably_modified_type_p): Prevent endless recursion for Ada
12650 cyclic pointer types.
12651 * lto-streamer-in.c: Include debug.h.
12652 (dref_queue): New global.
12653 (lto_read_tree_1): Stream in DIE references.
12654 (lto_input_tree): Register DIE references.
12655 (input_function): Stream DECL_DEBUG_ARGS.
12656 * lto-streamer-out.c: Include debug.h.
12657 (lto_write_tree_1): Output DIE references.
12658 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
12659 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
12660 (output_function): Stream DECL_DEBUG_ARGS.
12661 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
12662 Stream DECL_ABSTRACT_ORIGIN.
12663 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
12664 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
12665 DECL_CONTEXT for file-scope decls.
12666 * lto-streamer.h (struct dref_entry): Declare.
12667 (dref_queue): Likewise.
12668 * cfgexpand.c (pass_expand::execute): Do not call the
12669 outlining_inline_function hook here.
12670 * lto-wrapper.c (debug_obj): New global.
12671 (tool_cleanup): Unlink it if required.
12672 (debug_objcopy): New function.
12673 (run_gcc): Handle early debug sections in the IL files by
12674 extracting them to separate files, partially linkin them and
12675 feeding the result back as result to the linker.
12676 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
12677 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
12678 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
12679 sections into a separate segment.
12680 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
12682 (darwin_asm_dwarf_section): Likewise.
12683 (darwin_asm_output_dwarf_offset): Likewise.
12684 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
12686 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12687 Alan Hayward <alan.hayward@arm.com>
12688 David Sherwood <david.sherwood@arm.com>
12690 * read-md.h (md_reader::record_potential_iterator_use): Replace
12691 pointer argument with an rtx and an index.
12692 * read-rtl.c (iterator_group::apply_iterator): Likewise.
12693 (apply_mode_iterator): Likewise.
12694 (apply_code_iterator): Likewise.
12695 (apply_int_iterator): Likewise.
12696 (apply_subst_iterator): Likewise.
12697 (record_iterator_use): Likewise.
12698 (record_attribute_use): Likewise.
12699 (md_reader::record_potential_iterator_use): Likewise. Update calls
12700 to record_iterator_use and apply_iterator.
12701 (iterator_use): Replace ptr with x and index.
12702 (attribute_use): Likewise.
12703 (apply_attribute_uses): Update calls to apply_iterator.
12704 (apply_iterators): Likewise. Update initialization of iterator_use.
12705 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
12706 and record_potential_iterator_use.
12707 (rtx_reader::read_rtx_operand): Likewise.
12709 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12710 Alan Hayward <alan.hayward@arm.com>
12711 David Sherwood <david.sherwood@arm.com>
12713 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
12716 2017-08-21 Richard Biener <rguenther@suse.de>
12718 PR middle-end/81884
12719 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
12720 at struct end conservatively when comparing common bases.
12722 2017-08-21 Richard Biener <rguenther@suse.de>
12724 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
12725 (mem_ref_in_stmt): Remove.
12726 (determine_max_movement): Use ref index to get at the reference.
12727 (invariantness_dom_walker::before_dom_children): Deal with
12728 lim data already initialized.
12729 (gather_mem_refs_stmt): Initialize lim data and record ref index.
12731 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
12733 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
12734 (TARGET_ISA_ROUND): Ditto.
12735 (TARGET_ROUND): Ditto.
12736 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
12737 * config/i386/i386.md: Ditto.
12738 * config/i386/sse.md: Ditto.
12739 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
12740 with OPTION_MASK_ISA_SSE4_1.
12742 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
12745 * doc/extend.texi (x86 Built-in Functions): Correct the name of
12746 __builtin_ia32_lzcnt_u16.
12748 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
12751 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
12752 (rs6000_set_current_function): Rewrite function to use it.
12754 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
12757 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
12758 and TYPE_WARN_IF_NOT_ALIGN.
12759 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
12760 (handle_warn_if_not_align): New.
12761 (place_union_field): Call handle_warn_if_not_align.
12762 (place_field): Call handle_warn_if_not_align.
12763 Copy TYPE_WARN_IF_NOT_ALIGN.
12764 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
12765 (layout_type): Likewise.
12766 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
12768 (tree_decl_common): Add warn_if_not_align.
12769 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
12770 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
12771 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
12772 (DECL_WARN_IF_NOT_ALIGN): Likewise.
12773 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
12774 * doc/extend.texi: Document warn_if_not_aligned attribute.
12775 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
12777 2017-08-17 Martin Liska <mliska@suse.cz>
12780 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
12781 (get_data_dependence): Use it as pointer type.
12782 (distribute_loop): Likewise.
12784 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12786 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
12787 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
12788 (p8_vmrgow_<mode>_direct): New define_insn.
12789 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
12790 handle endianness for vmrgew and vmrgow permute patterns.
12792 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
12794 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
12795 * config/rs6000/rs6000-cpus.def: Remove comment.
12796 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
12797 (POWERPC_MASKS): Likewise.
12798 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
12799 use of TARGET_VSX_TIMODE.
12800 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
12801 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
12802 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
12803 (rs6000_option_override_internal): Remove dead code.
12804 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
12805 (rs6000_legitimize_reload_address): Likewise.
12806 (rs6000_legitimate_address_p): Likewise.
12807 (rs6000_opt_masks): Delete "vsx-timode".
12808 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
12809 from function comment.
12810 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
12811 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
12812 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
12814 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
12815 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
12816 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
12817 (VSX_TI): Likewise.
12819 (define_peephole2): Likewise.
12821 2017-08-17 Martin Sebor <msebor@redhat.com>
12824 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
12825 past the end of an array.
12826 (test_pp_format): Add test cases.
12828 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
12830 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
12831 missing ECF_NOTHROW flags.
12833 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
12836 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
12837 operands residing in integer registers.
12838 (*vsx_le_perm_load_<mode>): Likewise.
12839 (*vsx_le_perm_store_<mode>): Likewise.
12840 (define_peephole2): Add peepholes to optimize the above.
12842 2017-08-17 Marek Polacek <polacek@redhat.com>
12844 PR middle-end/81814
12845 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
12846 to mimic what shorten_compare did. Change the return type to bool.
12847 (fold_cond_expr_with_comparison): Update call to
12848 operand_equal_for_comparison_p.
12849 (fold_ternary_loc): Likewise.
12851 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
12853 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
12855 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
12857 2017-08-17 Richard Biener <rguenther@suse.de>
12859 * tree-ssa-structalias.c (solve_graph): When propagating
12860 to successors update the graphs succ edges and avoid duplicate work.
12862 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
12865 * config/i386/i386.c (ix86_option_override_internal): Save target
12866 specific options after ix86_stack_protector_guard_reg was changed.
12868 2017-08-17 Richard Biener <rguenther@suse.de>
12870 PR tree-optimization/81827
12871 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
12872 (new_var_info): Initialize it conservatively.
12873 (get_call_vi): Mark register vars.
12874 (new_scalar_tmp_constraint_exp): Likewise.
12875 (handle_rhs_call): Likewise.
12876 (handle_const_call): Likewise.
12877 (create_function_info_for): Likewise.
12878 (solve_constraints): Sort varinfos to separate register from
12879 non-register vars to pack points-to solution bitmaps during
12882 2017-08-17 Marek Polacek <polacek@redhat.com>
12884 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
12886 2017-08-17 Richard Biener <rguenther@suse.de>
12888 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
12889 to true when overflow is undefined and we saturated the result.
12891 2017-08-17 Alan Modra <amodra@gmail.com>
12894 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
12895 Don't use store multiple if only one reg needs saving.
12896 (interesting_frame_related_regno): New function.
12897 (rs6000_frame_related): Don't emit frame info for regs that
12899 (rs6000_emit_epilogue): Likewise.
12901 2017-08-16 Nathan Sidwell <nathan@acm.org>
12903 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
12904 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
12905 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
12906 * tree.c (free_lang_data_in_type): Use else-if chain. Always
12907 clear TYPE_LANG_1. Remove obsolete member-function stripping.
12908 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
12909 (verify_type): Adjust for TYPE_BINFO move.
12910 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
12911 process TYPE_BINFO directly.
12912 (hash_tree): Likewise.
12913 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
12915 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
12918 2017-08-16 David Malcolm <dmalcolm@redhat.com>
12920 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
12922 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
12925 * config/i386/i386.md (*anddi_1_btr): Change predicates of
12926 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
12927 Add ix86_binary_operator_ok to insn constraint.
12928 (*iordi_1_bts): Ditto.
12929 (*xordi_1_btc): Ditto.
12930 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
12931 Update corresponding peephole2 pattern.
12935 2017-08-16 Bin Cheng <bin.cheng@arm.com>
12937 PR tree-optimization/81832
12938 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
12939 copy loop header which has IFN_LOOP_DIST_ALIAS call.
12941 2017-08-16 Marek Polacek <polacek@redhat.com>
12944 * fold-const.c (fold_indirect_ref_1): Restore original behavior
12945 regarding size_zero_node.
12947 2017-08-16 Martin Liska <mliska@suse.cz>
12950 * config.gcc: Respect previously set extra_objs in case
12953 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
12955 PR tree-optimization/81835
12956 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
12957 the phi SSA_NAME. Check that the condition in a COND_EXPR does
12958 not depend on the phi.
12960 2017-08-16 Alan Modra <amodra@gmail.com>
12962 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
12965 2017-08-16 Alan Modra <amodra@gmail.com>
12967 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
12968 (save_reg_p): ..into this. Update all callers.
12969 (first_reg_to_save): Simplify.
12971 2017-08-16 Alan Modra <amodra@gmail.com>
12973 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
12976 2017-08-15 Joseph Myers <joseph@codesourcery.com>
12980 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
12981 constant count if that count is less than 32.
12983 2017-08-15 Nathan Sidwell <nathan@acm.org>
12985 * gcc.c (execute): Emit friendlier message if inferior is killed
12986 by an external cause.
12988 2017-08-15 Richard Biener <rguenther@suse.de>
12990 PR tree-optimization/81790
12991 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
12992 CONSTRUCTORs from simplifying and VN.
12994 2017-08-14 Martin Sebor <msebor@redhat.com>
12996 * builtin-attrs.def: Add comments.
12998 2017-08-14 Martin Sebor <msebor@redhat.com>
13001 * doc/extend.texi (attribute nonstring): Document new attribute.
13003 2017-08-14 Martin Sebor <msebor@redhat.com>
13006 * tree-diagnostic.c (default_tree_printer): Handle %G.
13007 * gimple-pretty-print.h (percent_G_format): Declare new function.
13008 * gimple-pretty-print.c (percent_G_format): Define.
13009 * tree-pretty-print.c (percent_K_format): Add argument.
13011 2017-08-14 Martin Sebor <msebor@redhat.com>
13013 PR translation/79998
13014 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
13015 Remove a stray space.
13017 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
13020 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
13021 (*iordi_1_bts): Ditto.
13022 (*xordi_1_btc): Ditto.
13024 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13027 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
13028 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13030 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
13031 quoted strings, and make more translator-friendly.
13032 (darwin_rs6000_override_options): Likewise.
13033 (rs6000_option_override_internal): Likewise.
13034 (rs6000_return_in_memory): Fix overlong line.
13035 (init_cmulative_args): Use quoted strings, and make more
13036 translator-friendly.
13037 (rs6000_pass_by_reference): Fix overlong line.
13038 (def_builtin): Use quoted strings.
13039 (altivec_expand_predicate_builtin): Use quoted strings, and make
13040 more translator-friendly.
13041 (htm_expand_builtin): Use quoted strings.
13042 (cpu_expand_builtin): Use quoted strings, and make more
13043 translator-friendly.
13044 (altivec_expand_builtin): Likewise.
13045 (paired_expand_predicate_builtin): Likewise.
13046 (rs6000_invalid_builtin): Likewise.
13047 (builtin_function_type): Use quoted strings.
13048 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
13049 more translator-friendly.
13050 (rs6000_trampoline_init): Likewise.
13051 (rs6000_handle_altivec_attribute): Likewise.
13052 (rs6000_inner_target_options): Use quoted strings.
13053 (rs6000_disable_incompatible_switches): Likewise.
13054 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
13055 strings, and make more translator-friendly.
13056 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
13058 2017-08-14 Bin Cheng <bin.cheng@arm.com>
13060 PR tree-optimization/81799
13061 * tree-loop-distribution.c (version_loop_by_alias_check): Force
13062 cond_expr to simple gimple operand.
13064 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
13066 PR middle-end/46932
13067 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
13069 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
13073 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
13074 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
13075 TARGET_GASISR_PROLOGUES.
13076 * config/avr/avr.c (avr_option_override): Same.
13077 (avr_pass_pre_proep::execute): Same.
13079 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
13082 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
13083 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
13085 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
13087 * config/i386/i386.md (*load_tp_<mode>): Redefine as
13088 define_insn_and_split. Split to a memory load from 0 in
13089 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
13090 using PTR mode iterator.
13091 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
13092 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
13093 (*add_tp_<mode>): Redefine as define_insn_and_split.
13094 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
13095 address space. Merge with *add_tp_x32 using PTR mode iterator.
13096 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
13097 Split to an add with a memory load from 0 in
13098 DEFAULT_TLS_SEG_REG address space.
13100 2017-08-12 Andrew Pinski <apinski@cavium.com>
13102 * config/aarch64/aarch64-option-extensions.def (rdma):
13103 Fix feature string to what Linux prints out in /proc/cpuinfo.
13105 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
13108 * dwarf2out.c (modified_type_die): For C typedef types that have
13109 an ultimate origin, process the ultimate origin instead of the
13111 (gen_typedef_die): Assert that input DECLs have no ultimate
13113 (gen_type_die_with_usage): For typedef variants that have an
13114 ultimate origin, just call gen_decl_die on the original DECL.
13115 (process_scope_var): Avoid creating DIEs for local typedefs and
13116 concrete static variables.
13118 2017-08-12 Alan Modra <amodra@gmail.com>
13122 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
13123 match gnu-user.h startfile.
13124 (ENDFILE_LINUX_SPEC): Similarly.
13126 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
13129 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
13131 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
13133 2017-08-11 Tamar Christina <tamar.christina@arm.com>
13134 * config/aarch64/aarch64.md (mov<mode>): Change.
13135 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
13136 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
13137 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
13139 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
13141 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
13142 for storage order barriers.
13144 2017-08-11 Martin Liska <mliska@suse.cz>
13147 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
13148 variables of void type.
13150 2017-08-11 Martin Liska <mliska@suse.cz>
13152 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
13153 TARGET_SUPPORTS_ALIASES.
13154 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
13155 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
13156 (optimize_weakref): Likewise.
13157 * symtab.c (symtab_node::noninterposable_alias): Likewise.
13158 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
13159 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
13161 2017-08-11 Martin Liska <mliska@suse.cz>
13164 * config/i386/i386.c (make_resolver_func): Do complete
13165 refactoring of the function.
13167 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
13170 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
13171 * config/i386/i386.c (ix86_stack_protect_guard): Use
13172 ix86_stack_protect_guard_symbol_str to generate varible declaration.
13173 * doc/invoke.texi (x86 Options): Document
13174 -mstack-protector-guard-symbol= option.
13176 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
13178 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
13179 * config/i386/i386.c (ix86_split_stack_guard): New function.
13180 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
13181 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
13182 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
13183 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
13184 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
13185 (split_stack_space_check): Call ix86_split_stack_guard.
13187 2017-08-10 Martin Sebor <msebor@redhat.com>
13189 * print-tree.c (print_node): Print location using the established
13191 Replace spaces with colons.
13192 (debug_raw, debug): Ditto.
13194 2017-08-10 Martin Sebor <msebor@redhat.com>
13197 * pretty-print.c (pp_format): Correct the handling of %s precision.
13199 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
13202 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
13204 (ix86_finalize_stack_frame_flags): This. Also clear
13205 frame_pointer_needed if -fno-omit-frame-pointer is used without
13207 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
13208 with ix86_finalize_stack_frame_flags.
13209 (ix86_expand_epilogue): Likewise.
13210 (ix86_expand_split_stack_prologue): Likewise.
13211 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
13213 2017-08-10 Martin Liska <mliska@suse.cz>
13216 * c-attribs.c (handle_target_attribute):
13217 Report warning for an empty string argument of target attribute.
13219 2017-08-09 Jakub Jelinek <jakub@redhat.com>
13222 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
13224 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
13225 or DECL_NONLOCAL labels.
13226 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
13227 or DECL_NONLOCAL labels here.
13229 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
13231 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
13232 to indicate when early gimple folding has been disabled.
13233 (rs6000_gimple_fold_builtin): Add debug content.
13234 (rs6000_invalid_builtin): Fix whitespace.
13235 (rs6000_expand_builtin): Fix whitespace.
13236 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
13238 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
13241 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
13242 SAVE_MULTIPLE if not all the registers that saves, should be saved.
13244 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
13246 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
13247 (qdf24xx): Likewise.
13248 * config/aarch64/aarch64.md: Include falkor.md.
13249 * config/aarch64/falkor.md: New.
13251 2017-08-09 Marek Polacek <polacek@redhat.com>
13254 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
13255 * diagnostic.c (emit_diagnostic): Add a comment.
13256 (emit_diagnostic_valist): New function.
13258 2017-08-09 Marek Polacek <polacek@redhat.com>
13261 * input.c (make_location): New overload.
13262 * input.h (make_location): Declare.
13264 2017-08-08 Alan Modra <amodra@gmail.com>
13265 H.J. Lu <hongjiu.lu@intel.com>
13268 * gcc.c (NO_PIE_SPEC): Delete.
13269 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
13271 (LINK_PIE_SPEC): ..to here.
13272 (LINK_COMMAND_SPEC): Support -no-pie.
13273 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
13274 chain of crtbegin*.o selection, update for PIE_SPEC changes and
13276 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
13277 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
13278 (ENDFILE_CRTEND_SPEC): Similarly.
13280 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
13283 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
13284 (mstack-protector-guard-offset=): Ditto.
13285 * config/i386/i386.c (ix86_option_override): Handle
13286 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
13288 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
13289 ix86_stack_protect_guard_offset variables.
13290 (TARGET_STACK_PROTECT_GUARD): Always define.
13291 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
13292 and -mstack-protector-guard-offset= options.
13294 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13296 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
13297 boundary case for the last candidate.
13299 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13301 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
13302 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
13304 2017-08-08 Tamar Christina <tamar.christina@arm.com>
13306 PR middle-end/19706
13307 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
13308 * config/aarch64/aarch64-builtins.c
13309 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
13310 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
13311 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
13313 2017-08-08 Tamar Christina <tamar.christina@arm.com>
13314 Andrew Pinski <pinskia@gmail.com>
13316 PR middle-end/19706
13317 * internal-fn.def (XORSIGN): New.
13318 * optabs.def (xorsign_optab): New.
13319 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
13320 (convert_expand_mult_copysign): New.
13321 (pass_optimize_widening_mul::execute): Call
13322 convert_expand_mult_copysign.
13324 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13326 PR tree-optimization/81354
13327 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
13328 Insert on edges rather than explicitly creating landing pads.
13329 (analyze_candidates_and_replace): Commit edge inserts.
13331 2017-08-08 Richard Biener <rguenther@suse.de>
13333 PR middle-end/81719
13334 * tree-ssa-loop-niter.c: Include tree-dfa.h.
13335 (expand_simple_operations): Also look through ADDR_EXPRs with
13336 MEM_REF bases treating them as POINTER_PLUS_EXPR.
13338 2017-08-08 Richard Biener <rguenther@suse.de>
13340 PR tree-optimization/81723
13341 * tree-vect-slp.c (struct bst_traits): New hash traits.
13342 (bst_fail): New global.
13343 (vect_build_slp_tree_2): New worker, split out from ...
13344 (vect_build_slp_tree): ... this now wrapping it with using
13345 bst_fail set to cache SLP tree build fails. Properly handle
13347 (vect_analyze_slp_instance): Allocate and free bst_fail.
13349 2017-08-08 Martin Liska <mliska@suse.cz>
13352 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
13353 LABEL_DECLs that can be from a different function.
13355 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13357 PR tree-optimization/81744
13358 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
13359 loop's number of iterations.
13361 2017-08-08 Martin Liska <mliska@suse.cz>
13363 * asan.c: Include header files.
13364 * attribs.c (build_decl_attribute_variant): New function moved
13366 (build_type_attribute_qual_variant): Likewise.
13367 (cmp_attrib_identifiers): Likewise.
13368 (simple_cst_list_equal): Likewise.
13369 (omp_declare_simd_clauses_equal): Likewise.
13370 (attribute_value_equal): Likewise.
13371 (comp_type_attributes): Likewise.
13372 (build_type_attribute_variant): Likewise.
13373 (lookup_ident_attribute): Likewise.
13374 (remove_attribute): Likewise.
13375 (merge_attributes): Likewise.
13376 (merge_type_attributes): Likewise.
13377 (merge_decl_attributes): Likewise.
13378 (merge_dllimport_decl_attributes): Likewise.
13379 (handle_dll_attribute): Likewise.
13380 (attribute_list_equal): Likewise.
13381 (attribute_list_contained): Likewise.
13382 * attribs.h (lookup_attribute): New function moved from tree.[ch].
13383 (lookup_attribute_by_prefix): Likewise.
13384 * bb-reorder.c: Include header files.
13385 * builtins.c: Likewise.
13386 * calls.c: Likewise.
13387 * cfgexpand.c: Likewise.
13388 * cgraph.c: Likewise.
13389 * cgraphunit.c: Likewise.
13390 * convert.c: Likewise.
13391 * dwarf2out.c: Likewise.
13392 * final.c: Likewise.
13393 * fold-const.c: Likewise.
13394 * function.c: Likewise.
13395 * gimple-expr.c: Likewise.
13396 * gimple-fold.c: Likewise.
13397 * gimple-pretty-print.c: Likewise.
13398 * gimple.c: Likewise.
13399 * gimplify.c: Likewise.
13400 * hsa-common.c: Likewise.
13401 * hsa-gen.c: Likewise.
13402 * internal-fn.c: Likewise.
13403 * ipa-chkp.c: Likewise.
13404 * ipa-cp.c: Likewise.
13405 * ipa-devirt.c: Likewise.
13406 * ipa-fnsummary.c: Likewise.
13407 * ipa-inline.c: Likewise.
13408 * ipa-visibility.c: Likewise.
13410 * lto-cgraph.c: Likewise.
13411 * omp-expand.c: Likewise.
13412 * omp-general.c: Likewise.
13413 * omp-low.c: Likewise.
13414 * omp-offload.c: Likewise.
13415 * omp-simd-clone.c: Likewise.
13416 * opts-global.c: Likewise.
13417 * passes.c: Likewise.
13418 * predict.c: Likewise.
13419 * sancov.c: Likewise.
13420 * sanopt.c: Likewise.
13421 * symtab.c: Likewise.
13422 * toplev.c: Likewise.
13423 * trans-mem.c: Likewise.
13424 * tree-chkp.c: Likewise.
13425 * tree-eh.c: Likewise.
13426 * tree-into-ssa.c: Likewise.
13427 * tree-object-size.c: Likewise.
13428 * tree-parloops.c: Likewise.
13429 * tree-profile.c: Likewise.
13430 * tree-ssa-ccp.c: Likewise.
13431 * tree-ssa-live.c: Likewise.
13432 * tree-ssa-loop.c: Likewise.
13433 * tree-ssa-sccvn.c: Likewise.
13434 * tree-ssa-structalias.c: Likewise.
13435 * tree-ssa.c: Likewise.
13436 * tree-streamer-in.c: Likewise.
13437 * tree-vectorizer.c: Likewise.
13438 * tree-vrp.c: Likewise.
13439 * tsan.c: Likewise.
13440 * ubsan.c: Likewise.
13441 * varasm.c: Likewise.
13442 * varpool.c: Likewise.
13443 * tree.c: Remove functions moved to attribs.[ch].
13444 * tree.h: Likewise.
13445 * config/aarch64/aarch64.c: Add attrs.h header file.
13446 * config/alpha/alpha.c: Likewise.
13447 * config/arc/arc.c: Likewise.
13448 * config/arm/arm.c: Likewise.
13449 * config/avr/avr.c: Likewise.
13450 * config/bfin/bfin.c: Likewise.
13451 * config/c6x/c6x.c: Likewise.
13452 * config/cr16/cr16.c: Likewise.
13453 * config/cris/cris.c: Likewise.
13454 * config/darwin.c: Likewise.
13455 * config/epiphany/epiphany.c: Likewise.
13456 * config/fr30/fr30.c: Likewise.
13457 * config/frv/frv.c: Likewise.
13458 * config/ft32/ft32.c: Likewise.
13459 * config/h8300/h8300.c: Likewise.
13460 * config/i386/winnt.c: Likewise.
13461 * config/ia64/ia64.c: Likewise.
13462 * config/iq2000/iq2000.c: Likewise.
13463 * config/lm32/lm32.c: Likewise.
13464 * config/m32c/m32c.c: Likewise.
13465 * config/m32r/m32r.c: Likewise.
13466 * config/m68k/m68k.c: Likewise.
13467 * config/mcore/mcore.c: Likewise.
13468 * config/microblaze/microblaze.c: Likewise.
13469 * config/mips/mips.c: Likewise.
13470 * config/mmix/mmix.c: Likewise.
13471 * config/mn10300/mn10300.c: Likewise.
13472 * config/moxie/moxie.c: Likewise.
13473 * config/msp430/msp430.c: Likewise.
13474 * config/nds32/nds32-isr.c: Likewise.
13475 * config/nds32/nds32.c: Likewise.
13476 * config/nios2/nios2.c: Likewise.
13477 * config/nvptx/nvptx.c: Likewise.
13478 * config/pa/pa.c: Likewise.
13479 * config/pdp11/pdp11.c: Likewise.
13480 * config/powerpcspe/powerpcspe.c: Likewise.
13481 * config/riscv/riscv.c: Likewise.
13482 * config/rl78/rl78.c: Likewise.
13483 * config/rx/rx.c: Likewise.
13484 * config/s390/s390.c: Likewise.
13485 * config/sh/sh.c: Likewise.
13486 * config/sol2.c: Likewise.
13487 * config/sparc/sparc.c: Likewise.
13488 * config/spu/spu.c: Likewise.
13489 * config/stormy16/stormy16.c: Likewise.
13490 * config/tilegx/tilegx.c: Likewise.
13491 * config/tilepro/tilepro.c: Likewise.
13492 * config/v850/v850.c: Likewise.
13493 * config/vax/vax.c: Likewise.
13494 * config/visium/visium.c: Likewise.
13495 * config/xtensa/xtensa.c: Likewise.
13497 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13500 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
13501 constraints since the -mupper-regs-* switches have been
13503 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
13504 into a vector from a double word element that was extracted from
13505 another vector, and eliminate extra XXPERMDI instructions.
13506 (vsx_concat_<mode>_2): Likewise.
13507 (vsx_concat_<mode>_3): Likewise.
13508 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
13509 concat to allow optimizing inserts from previous extracts.
13511 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
13513 * config/i386/i386.c (ix86_stack_protect_guard): Generate
13514 memory reference to a SSP offset in TLS address space.
13515 (ix86_print_operand) <case '@'>: Remove.
13516 (ix86_print_operand_punct_valid_p): Remove '@' code.
13517 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
13518 UNSPEC_SP_TLS_TEST.
13519 (stack_tls_protect_set_<mode>): Remove.
13520 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
13521 (stack_tls_protect_test_<mode>): Remove.
13522 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
13524 2017-08-07 Olivier Hainque <hainque@adacore.com>
13527 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
13529 2017-08-07 Douglas Rupp <rupp@adacore.com>
13531 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
13532 variable was referenced as multidir in command.
13534 2017-08-07 Jakub Jelinek <jakub@redhat.com>
13537 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
13540 2017-08-07 Martin Liska <mliska@suse.cz>
13542 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
13543 * config/rl78/rl78.c: Add include of attribs.h.
13544 * config/sh/sh.c: Likewise.
13545 * config/v850/v850.c: Likewise.
13547 2017-08-07 Tom de Vries <tom@codesourcery.com>
13549 PR middle-end/78266
13550 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
13552 2017-08-07 Martin Liska <mliska@suse.cz>
13554 * config/mips/mips.c: Include attribs.h.
13556 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
13559 * doc/invoke.texi: Document change in behvaior for -Ofast for
13562 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
13564 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
13566 (aarch64_pop_regs): Likewise.
13567 (aarch64_gen_load_pair): Likewise.
13568 (aarch64_save_callee_saves): Likewise.
13569 (aarch64_restore_callee_saves): Likewise.
13571 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13573 * config/i386/i386.c: Revert the last change.
13575 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13578 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
13580 (ix86_finalize_stack_frame_flags): This. Also clear
13581 frame_pointer_needed if -fno-omit-frame-pointer is used without
13583 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
13584 with ix86_finalize_stack_frame_flags.
13585 (ix86_expand_epilogue): Likewise.
13586 (ix86_expand_split_stack_prologue): Likewise.
13588 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13591 * config/i386/i386.c (get_builtin_code_for_version): Set priority
13592 to P_AES for Westmere.
13594 2017-08-07 Jonathan Yong <10walls@gmail.com>
13596 * config/i386/mingw.opt (fset-stack-executable): Removed.
13597 * config/i386/cygming.opt (fset-stack-executable): Moved
13599 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
13601 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
13603 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
13605 2017-08-07 Marek Polacek <polacek@redhat.com>
13607 PR middle-end/81737
13608 * fold-const.c (fold_indirect_ref_1): Check type_domain.
13610 2017-08-07 Martin Liska <mliska@suse.cz>
13612 * attribs.h (canonicalize_attr_name): New function.
13613 (cmp_attribs): Move from c-format.c and adjusted.
13614 (is_attribute_p): Moved from tree.h.
13615 * tree-inline.c: Add new includes.
13616 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
13617 (private_is_attribute_p): Remove.
13618 (private_lookup_attribute): Likewise.
13619 (private_lookup_attribute_by_prefix): Simplify.
13620 (remove_attribute): Use is_attribute_p.
13621 * tree.h: Remove removed declarations.
13623 2017-08-07 Jakub Jelinek <jakub@redhat.com>
13625 PR middle-end/81698
13626 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
13627 instead of computing it in the function. Formatting fix.
13628 (expand_case): Don't rely on default_edge being the first edge,
13629 clear it if removing it, pass default_edge to
13630 emit_case_dispatch_table.
13631 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
13634 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
13636 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
13637 insn in the function, emit NOP after the insn.
13639 2017-08-06 Tom de Vries <tom@codesourcery.com>
13641 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
13644 2017-08-06 Tom de Vries <tom@codesourcery.com>
13646 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
13649 2017-08-04 Yury Gribov <tetra2005@gmail.com>
13651 PR tree-optimization/57371
13652 * match.pd: New pattern.
13654 2017-08-04 Marek Polacek <polacek@redhat.com>
13656 PR middle-end/81695
13657 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
13658 perform the computation in offset_int.
13660 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13662 PR tree-optimization/81136
13663 * tree-vectorizer.h: Include tree-hash-traits.h.
13664 (vec_base_alignments): New typedef.
13665 (vec_info): Add a base_alignments field.
13666 (vect_record_base_alignments): Declare.
13667 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
13669 (DR_IS_CONDITIONAL_IN_STMT): New macro.
13670 (create_data_ref): Add an is_conditional_in_stmt argument.
13671 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
13672 the is_conditional_in_stmt field.
13673 (data_ref_loc): Add an is_conditional_in_stmt field.
13674 (get_references_in_stmt): Set the is_conditional_in_stmt field.
13675 (find_data_references_in_stmt): Update call to create_data_ref.
13676 (graphite_find_data_references_in_stmt): Likewise.
13677 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
13678 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
13679 (vect_record_base_alignment): New function.
13680 (vect_record_base_alignments): Likewise.
13681 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
13682 for nested statements even if we fail to compute a misalignment.
13683 Use pooled base alignments for unconditional references.
13684 (vect_find_same_alignment_drs): Compare base addresses instead
13686 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
13687 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
13689 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13691 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
13692 Add an explicit name for the enum. Use auto_vec for slp_instances
13693 and grouped_stores.
13694 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
13696 (_bb_vec_info): Add a constructor and destructor.
13697 (vinfo_for_stmt): Return NULL for uids of -1 as well.
13698 (destroy_loop_vec_info): Delete.
13699 (vect_destroy_datarefs): Likewise.
13700 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
13701 (vec_info::vec_info): New function.
13702 (vec_info::~vec_info): Likewise.
13703 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
13704 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
13705 destroy_loop_vec_info.
13706 * tree-vect-loop.c (new_loop_vec_info): Replace with...
13707 (_loop_vec_info::_loop_vec_info): ...this.
13708 (destroy_loop_vec_info): Replace with...
13709 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
13710 the stmt_vec_infos. Leave handling of vec_info information to its
13711 destructor. Remove explicit vector releases.
13712 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
13713 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
13714 * tree-vect-slp.c (new_bb_vec_info): Replace with...
13715 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
13716 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
13717 (destroy_bb_vec_info): Replace with...
13718 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
13719 information to its destructor.
13720 (vect_slp_analyze_bb_1): Use new and delete instead of
13721 new_bb_vec_info and destroy_bb_vec_info.
13722 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
13725 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13727 * tree-data-ref.h (subscript): Add access_fn field.
13728 (data_dependence_relation): Add could_be_independent_p.
13729 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
13730 (same_access_functions): Move to tree-data-ref.c.
13731 * tree-data-ref.c (ref_contains_union_access_p): New function.
13732 (access_fn_component_p): Likewise.
13733 (access_fn_components_comparable_p): Likewise.
13734 (dr_analyze_indices): Add a reference to access_fn_component_p.
13735 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
13737 (constant_access_functions): Likewise.
13738 (add_other_self_distances): Likewise.
13739 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
13740 (initialize_data_dependence_relation): Use XCNEW and remove
13741 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
13742 of access functions that have the same type. Allow the
13743 subsequence to end with different bases in some circumstances.
13744 Record the chosen access functions in SUB_ACCESS_FN.
13745 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
13746 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
13747 (subscript_dependence_tester_1): Likewise dra and drb.
13748 (build_classic_dist_vector): Update calls accordingly.
13749 (subscript_dependence_tester): Likewise.
13750 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
13751 DDR_COULD_BE_INDEPENDENT_P.
13752 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
13753 comp_alias_ddrs instead of may_alias_ddrs.
13754 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
13756 (vect_analyze_data_ref_dependence): Use it if
13757 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
13758 distance vectors if that fails.
13759 (dependence_distance_ge_vf): New function.
13760 (vect_prune_runtime_alias_test_list): Use it. Don't clear
13761 LOOP_VINFO_MAY_ALIAS_DDRS.
13763 2017-08-04 Richard Biener <rguenther@suse.de>
13765 PR middle-end/81705
13766 * fold-const.c (fold_binary_loc): Properly restrict
13767 minus_var0 && minus_var1 case when associating undefined overflow
13770 2017-08-04 Tom de Vries <tom@codesourcery.com>
13772 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
13774 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13776 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13777 Don't start diagnostic messages with a capital letter.
13778 * config/rs6000/rs6000.c (rs6000_option_override_internal):
13780 (rs6000_invalid_builtin): Likewise.
13781 (rs6000_trampoline_init): Likewise.
13783 2017-08-03 Jakub Jelinek <jakub@redhat.com>
13786 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
13787 after setting changeable df flags.
13789 2017-08-03 Richard Biener <rguenther@suse.de>
13791 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
13792 up if the use is in USE - X.
13794 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
13796 * toplev.c (dumpfile.h): New include.
13797 (internal_error_reentered): New static function. Use it...
13798 (internal_error_function): ...here to handle reentered internal_error.
13800 2017-08-03 Richard Biener <rguenther@suse.de>
13802 PR middle-end/81148
13803 * fold-const.c (split_tree): Add minus_var and minus_con
13804 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
13805 here but always use minus_*.
13806 (associate_trees): Assert we never associate with MINUS_EXPR
13807 and NULL first operand. Do not recurse for PLUS_EXPR operands
13808 when associating as MINUS_EXPR either.
13809 (fold_binary_loc): Track minus_var and minus_con.
13811 2017-08-03 Tom de Vries <tom@codesourcery.com>
13814 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
13815 ACCEL_COMPILER, apply finish_options on
13816 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
13818 2017-08-03 Tom de Vries <tom@codesourcery.com>
13821 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
13822 function_entry_patch_area_size > 0.
13824 2017-08-03 Jakub Jelinek <jakub@redhat.com>
13827 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
13828 instead of 10??LU, perform unit multiplication in wide_int,
13829 don't change alloc_object_size_limit if the limit is larger
13832 PR tree-optimization/81655
13833 PR tree-optimization/81588
13834 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
13835 the case when ranges[i].low and high are 1 for unsigned type with
13838 PR middle-end/81052
13839 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
13840 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
13842 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13844 * tree-vrp.h: Add include guard.
13846 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
13849 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
13850 (ud2): New insn pattern.
13851 * config/i386/i386.c (ix86_expand_epilogue):
13852 For naked functions, generate ud2 instead of trap insn.
13854 2017-08-02 Marek Polacek <polacek@redhat.com>
13857 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
13859 2017-08-02 Tom de Vries <tom@codesourcery.com>
13860 Cesar Philippidis <cesar@codesourcery.com>
13862 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
13863 Add missing edge probabilities.
13865 2017-08-02 Tamar Christina <tamar.christina@arm.com>
13867 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
13868 Correct endianness.
13870 2017-08-02 Jakub Jelinek <jakub@redhat.com>
13872 PR middle-end/79499
13873 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
13874 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
13875 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
13876 prologue_seq sequences - if any.
13878 2017-08-02 Richard Biener <rguenther@suse.de>
13880 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
13881 via vectors if supported, integer extracts via punning if supported
13882 or otherwise vector extracts.
13884 2017-08-02 Richard Biener <rguenther@suse.de>
13886 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
13888 (bitmap_insert_into_set): ... this.
13890 2017-08-02 Richard Biener <rguenther@suse.de>
13892 PR tree-optimization/81633
13894 2015-08-17 Alan Hayward <alan.hayward@arm.com>
13896 PR tree-optimization/71752
13897 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
13899 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
13901 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
13902 (machine_function::call_ms2sysv_pad_out): Remove field.
13903 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
13904 (ix86_compute_frame_layout): Likewise.
13906 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
13909 * config/i386/i386.c (ix86_set_func_type): Disallow naked
13910 attribute with interrupt attribute.
13912 2017-08-01 Andrew Pinski <apinski@cavium.com>
13914 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
13915 BIT_INSERT_EXPR's operand 1
13916 to see if the types precision matches.
13918 2017-08-01 Martin Liska <mliska@suse.cz>
13920 PR middle-end/70140
13921 * builtins.c (expand_builtin_memcpy_args): Remove.
13922 (expand_builtin_memcpy): Call newly added function
13923 expand_builtin_memory_copy_args.
13924 (expand_builtin_memcpy_with_bounds): Likewise.
13925 (expand_builtin_mempcpy): Remove last argument.
13926 (expand_builtin_mempcpy_with_bounds): Likewise.
13927 (expand_builtin_memory_copy_args): New function created from
13928 expand_builtin_mempcpy_args with small modifications.
13929 (expand_builtin_mempcpy_args): Remove.
13930 (expand_builtin_stpcpy): Remove unused argument.
13931 (expand_builtin): Likewise.
13932 (expand_builtin_with_bounds): Likewise.
13934 2017-08-01 Martin Liska <mliska@suse.cz>
13937 Make mempcpy more optimal (PR middle-end/70140).
13939 2017-08-01 Jakub Jelinek <jakub@redhat.com>
13942 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
13943 __builtin_vec_cmpne verify both arguments are compatible vectors
13944 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
13945 verify arg1_type is a pointer or array type. For __builtin_vec_st,
13946 move computation of aligned to after checking the argument types.
13950 * config/rs6000/vsx.md (vextract_fp_from_shorth,
13951 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
13954 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
13955 Doug Rupp <rupp@adacore.com>
13956 Olivier Hainque <hainque@adacore.com>
13958 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
13959 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
13961 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
13962 for TARGET_OS_CPP_BUILTIN.
13963 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
13964 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
13966 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
13967 passing required abi options to the assembler for EABI configurations.
13968 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
13969 of .text.hot and .text.unlikely sections for kernel modules when
13970 using ARM style exceptions.
13971 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
13972 options. Add EXTRA_CC1_SPEC.
13973 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
13975 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
13977 (ARM_TARGET2_DWARF_FORMAT): Define.
13978 * config/arm/t-vxworks: Adjust multilib control to removal of the
13979 Diab command line options.
13981 2017-08-01 Martin Liska <mliska@suse.cz>
13983 PR gcov-profile/81561
13984 * gcov.c (unblock): Make unblocking safe as we need to preserve
13985 index correspondence of blocks and block_lists.
13987 2017-08-01 Richard Biener <rguenther@suse.de>
13989 PR tree-optimization/81181
13990 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
13991 (compute_antic): ... end of iteration here.
13993 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
13995 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
13996 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
13997 (ftree-slp-vectorize): Likewise.
13998 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
13999 can no longer be set independent of flag_tree_loop_vectorize.
14000 * omp-general.c (emp_max_vf): Likewise.
14001 * opts.c (enable_fdo_optimizations): Remove references to
14002 flag_tree_vectorize, these are now implicit.
14003 (common_handle_option): Remove handling for OPT_ftree_vectorize,
14004 and leave it for the options machinery.
14006 2017-08-01 Martin Liska <mliska@suse.cz>
14008 PR middle-end/70140
14009 * builtins.c (expand_builtin_memcpy_args): Remove.
14010 (expand_builtin_memcpy): Call newly added function
14011 expand_builtin_memory_copy_args.
14012 (expand_builtin_memcpy_with_bounds): Likewise.
14013 (expand_builtin_mempcpy): Remove last argument.
14014 (expand_builtin_mempcpy_with_bounds): Likewise.
14015 (expand_builtin_memory_copy_args): New function created from
14016 expand_builtin_mempcpy_args with small modifications.
14017 (expand_builtin_mempcpy_args): Remove.
14018 (expand_builtin_stpcpy): Remove unused argument.
14019 (expand_builtin): Likewise.
14020 (expand_builtin_with_bounds): Likewise.
14022 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
14025 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
14026 print "ds:" only for immediates in generic address space.
14028 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
14031 * config/i386/i386.c (ix86_funciton_naked): New prototype.
14032 (ix86_function_ok_for_sibcall): Return false for naked functions.
14034 2017-08-01 Richard Biener <rguenther@suse.de>
14036 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
14037 (compute_antic): Seed worklist with exit block predecessors.
14038 * cfganal.c (dfs_find_deadend): For a cycle return the source
14039 of the edge closing it.
14041 2017-08-01 Tamar Christina <tamar.christina@arm.com>
14043 * config/aarch64/aarch64.c
14044 (aarch64_can_const_movi_rtx_p): Move 0 check.
14046 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14048 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
14049 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
14051 * match.pd: Ditto in address comparison pattern.
14053 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14055 PR tree-optimization/81627
14056 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
14057 closed ssa form for store-store chain.
14059 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14061 PR tree-optimization/81620
14062 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
14063 for store-store chain.
14065 2017-08-01 Jakub Jelinek <jakub@redhat.com>
14067 PR tree-optimization/81588
14068 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
14069 ranges[i].in_p, invert comparison code ccode. For >/>=,
14070 swap rhs1 and rhs2 and comparison code unconditionally,
14071 for </<= don't do that. Don't swap rhs1/rhs2 again if
14072 ranges[i].in_p, instead invert comparison code ccode if
14073 opcode or oe->rank is BIT_IOR_EXPR.
14076 * optabs.def (vec_extract_optab, vec_init_optab): Change from
14077 a direct optab to conversion optab.
14078 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
14079 with GET_MODE_INNER as last argument instead of optab_handler.
14080 * expmed.c (extract_bit_field_1): Likewise. Use vector from
14081 vector extraction if possible and optab is available.
14082 * expr.c (store_constructor): Use convert_optab_handler instead
14083 of optab_handler. Use vector initialization from smaller
14084 vectors if possible and optab is available.
14085 * tree-vect-stmts.c (vectorizable_load): Likewise.
14086 * doc/md.texi (vec_extract, vec_init): Document that the optabs
14087 now have two modes.
14088 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
14089 of vec_init from half-sized vectors with the same element mode.
14090 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
14091 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
14092 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
14093 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
14094 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
14095 after mode in gen_vec_extract* calls.
14096 (vec_extract<mode>): Renamed to ...
14097 (vec_extract<mode><ssescalarmodelower>): ... this.
14098 (vec_extract<mode><ssehalfvecmodelower>): New expander.
14099 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
14100 element mode after mode in gen_vec_init* calls.
14101 (VEC_INIT_HALF_MODE): New mode iterator.
14102 (vec_init<mode>): Renamed to ...
14103 (vec_init<mode><ssescalarmodelower>): ... this.
14104 (vec_init<mode><ssehalfvecmodelower>): New expander.
14105 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
14106 (vec_extractv2sfsf): ... this.
14107 (vec_initv2sf): Renamed to ...
14108 (vec_initv2sfsf): ... this.
14109 (vec_extractv2si): Renamed to ...
14110 (vec_extractv2sisi): ... this.
14111 (vec_initv2si): Renamed to ...
14112 (vec_initv2sisi): ... this.
14113 (vec_extractv4hi): Renamed to ...
14114 (vec_extractv4hihi): ... this.
14115 (vec_initv4hi): Renamed to ...
14116 (vec_initv4hihi): ... this.
14117 (vec_extractv8qi): Renamed to ...
14118 (vec_extractv8qiqi): ... this.
14119 (vec_initv8qi): Renamed to ...
14120 (vec_initv8qiqi): ... this.
14121 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
14122 (vec_init<mode>): Renamed to ...
14123 (vec_init<mode><VEC_base_l>): ... this.
14124 (vec_extract<mode>): Renamed to ...
14125 (vec_extract<mode><VEC_base_l>): ... this.
14126 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
14127 (vec_initv2sfsf): ... this.
14128 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
14129 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
14130 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
14131 element mode after mode in gen_vec_init* calls.
14132 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
14133 (vec_init<mode><Vel>): ... this.
14134 (vec_extract<mode>): Renamed to ...
14135 (vec_extract<mode><Vel>): ... this.
14136 * config/aarch64/iterators.md (Vel): New mode attribute.
14137 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
14138 Add element mode after mode in gen_vec_extract* calls.
14139 * config/s390/vector.md (non_vec_l): New mode attribute.
14140 (vec_extract<mode>): Renamed to ...
14141 (vec_extract<mode><non_vec_l>): ... this.
14142 (vec_init<mode>): Renamed to ...
14143 (vec_init<mode><non_vec_l>): ... this.
14144 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
14145 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
14147 * config/arm/iterators.md (V_elem_l): New mode attribute.
14148 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
14149 (vec_extract<mode><V_elem_l>): ... this.
14150 (vec_extractv2di): Renamed to ...
14151 (vec_extractv2didi): ... this.
14152 (vec_init<mode>): Renamed to ...
14153 (vec_init<mode><V_elem_l>): ... this.
14154 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
14155 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
14156 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
14157 Add element mode after gen_vec_extract* calls.
14158 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
14159 (vec_init<mode><unitmode>): ... this.
14160 (vec_extract<mode>): Renamed to ...
14161 (vec_extract<mode><unitmode>): ... this.
14162 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
14163 (vec_init<mode><unitmode>): ... this.
14164 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
14165 (vec_initv2sfsf): ... this.
14166 (vec_extractv2sf): Renamed to ...
14167 (vec_extractv2sfsf): ... this.
14168 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
14169 Add element mode after gen_vec_extract* calls.
14170 * config/mips/mips.md (unitmode): New mode iterator.
14171 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
14172 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
14173 * config/spu/spu.md (inner_l): New mode attribute.
14174 (vec_init<mode>): Renamed to ...
14175 (vec_init<mode><inner_l>): ... this.
14176 (vec_extract<mode>): Renamed to ...
14177 (vec_extract<mode><inner_l>): ... this.
14178 * config/sparc/sparc.md (veltmode): New mode iterator.
14179 (vec_init<VMALL:mode>): Renamed to ...
14180 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
14181 * config/ia64/vect.md (vec_initv2si): Renamed to ...
14182 (vec_initv2sisi): ... this.
14183 (vec_initv2sf): Renamed to ...
14184 (vec_initv2sfsf): ... this.
14185 (vec_extractv2sf): Renamed to ...
14186 (vec_extractv2sfsf): ... this.
14187 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
14188 (vec_init<mode>): Renamed to ...
14189 (vec_init<mode><VEC_base_l>): ... this.
14190 (vec_extract<mode>): Renamed to ...
14191 (vec_extract<mode><VEC_base_l>): ... this.
14192 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
14193 (vec_initv2sfsf): ... this.
14194 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
14195 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
14196 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
14197 gen_vec_init* calls.
14199 2017-08-01 Richard Biener <rguenther@suse.de>
14201 PR tree-optimization/81297
14202 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
14203 TREE_OVERFLOW from INTEGER_CSTs.
14205 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
14207 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
14209 2017-07-31 Carl Love <cel@us.ibm.com>
14211 * config/rs6000/rs6000-c: Add support for built-in functions
14212 vector signed char vec_xl_be (signed long long, signed char *);
14213 vector unsigned char vec_xl_be (signed long long, unsigned char *);
14214 vector signed int vec_xl_be (signed long long, signed int *);
14215 vector unsigned int vec_xl_be (signed long long, unsigned int *);
14216 vector signed long long vec_xl_be (signed long long, signed long long *);
14217 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
14218 vector signed short vec_xl_be (signed long long, signed short *);
14219 vector unsigned short vec_xl_be (signed long long, unsigned short *);
14220 vector double vec_xl_be (signed long long, double *);
14221 vector float vec_xl_be (signed long long, float *);
14222 * config/rs6000/altivec.h (vec_xl_be): Add #define.
14223 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
14224 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
14226 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
14227 (altivec_expand_builtin): Add switch statement to call
14228 altivec_expand_xl_be for each builtin.
14229 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
14230 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
14231 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
14232 __builtin_vsx_le_be_v16qi.
14233 * doc/extend.texi: Update the built-in documentation file for the
14234 new built-in functions.
14236 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
14239 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
14241 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
14243 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14245 * config.gcc: Add z14.
14246 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
14247 CPU model numbers for z13s and z14.
14248 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
14250 * config/s390/s390-opts.h (enum processor_type): Rename
14251 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
14252 * config/s390/s390.c (processor_table): Add field for CPU name to
14253 be passed to Binutils.
14254 (s390_asm_output_machine_for_arch): Use the new field in
14255 processor_table for Binutils.
14256 (s390_expand_builtin): Replace arch12 with z14.
14257 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
14258 (s390_get_sched_attrmask): Likewise.
14259 (s390_get_unit_mask): Likewise.
14260 * config/s390/s390.opt: Add z14 to processor_type enum.
14262 2017-07-31 Martin Jambor <mjambor@suse.cz>
14265 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
14266 regardless of optimization level.
14268 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
14269 Martin Liska <mliska@suse.cz>
14271 * predict.def: Remove old comment and adjust probability.
14272 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
14273 PREDICT statements.
14275 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
14278 * config/i386/i386.c (ix86_function_naked): New function.
14279 (ix86_can_use_return_insn_p): Return false for naked functions.
14280 (ix86_expand_prologue): Skip prologue for naked functions.
14281 (ix86_expand_epilogue): Skip epilogue for naked functions
14282 and emit trap instruction.
14283 (ix86_warn_func_return): New function.
14284 (ix86_attribute_table): Add "naked" attribute specification.
14285 (TARGET_WARN_FUNC_RETURN): Define.
14286 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
14288 2017-07-31 Martin Liska <mliska@suse.cz>
14290 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
14291 (dump_gimple_bb_header): Always dump BB info.
14292 (pp_cfg_jump): Do not append info about BB when dumping a jump.
14294 2017-07-31 Martin Liska <mliska@suse.cz>
14297 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
14298 also with current_function_decl non-null equality.
14300 2017-07-31 Jakub Jelinek <jakub@redhat.com>
14303 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
14304 change type to the element type, instead add eltype variable and
14305 use it where we are interested in the element type.
14307 PR tree-optimization/81603
14308 * ipa-polymorphic-call.c
14309 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
14310 offset arithmetic in offset_int, bail out if the resulting bit offset
14311 doesn't fit into shwi.
14313 2017-07-31 Martin Liska <mliska@suse.cz>
14315 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
14316 (gimplify_save_expr): Fix comment.
14318 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
14321 * config/i386/i386.c (ix86_function_arg): Update arguments for
14323 (ix86_compute_frame_layout): Set the initial stack offset to
14324 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
14325 INCOMING_FRAME_SP_OFFSET.
14326 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
14327 stack before exception handler returns.
14328 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
14329 the 'ERROR_CODE' for exception handler.
14331 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
14333 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
14334 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
14335 (ASM_OUTPUT_REG_POP): Ditto.
14336 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
14337 instead of asm_fprintf to output pure string.
14339 2017-07-29 Jakub Jelinek <jakub@redhat.com>
14341 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
14342 to imported_module_or_decl hook.
14343 (debug_nothing_tree_tree_tree_bool): Remove.
14344 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
14345 * debug.c (do_nothing_debug_hooks): Use
14346 debug_nothing_tree_tree_tree_bool_bool instead of
14347 debug_nothing_tree_tree_tree_bool.
14348 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
14349 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
14350 * sdbout.c (sdb_debug_hooks): Likewise.
14351 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
14352 (gen_namespace_die): Add DW_AT_export_symbols attribute if
14354 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
14355 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
14356 attribute, don't add anything.
14358 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14360 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
14361 (fold_build2_stat_loc): Likewise.
14362 (fold_build3_stat_loc): Likewise.
14363 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
14364 (fold_build1_loc): Remove macro.
14365 (fold_build2_loc): Likewise.
14366 (fold_build3_loc): Likewise.
14368 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14370 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
14371 (gimple_build_debug_bind_source_stat): Likewise.
14372 * gimple.h (gimple_build_debug_bind): Remove macro.
14373 (gimple_build_debug_bind_source): Likewise.
14375 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14377 * bitmap.c (bitmap_alloc): Adjust.
14378 (bitmap_gc_alloc): Likewise.
14379 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
14381 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14383 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
14384 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
14385 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
14386 (bitmap_gc_alloc_stat): Likewise.
14387 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
14389 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14391 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
14392 * rtl.h (shallow_copy_rtx): Remove macro.
14394 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14396 * emit-rtl.c (gen_raw_REG): Adjust.
14397 * gengenrtl.c (gendef): Likewise.
14398 * rtl.c (rtx_alloc_stat): Remove _stat from name.
14399 * rtl.h (rtx_alloc): Remove macro.
14401 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14403 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
14404 (build_tree_list_stat): Likewise.
14405 * tree.h (build_tree_list): Remove macro.
14406 (build_tree_list_vec): Likewise.
14408 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14410 * tree.c (make_vector_stat): Remove _stat from name.
14411 (build_vector_stat): Likewise.
14412 * tree.h (make_vector_stat): Remove macro.
14413 (build_vector_stat): Likewise.
14415 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14417 * tree.h (build_var_debug_value): Remove prototype.
14419 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14421 * tree.c (tree_cons_stat): Remove _stat from name.
14422 * tree.h (tree_cons): Remove macro.
14424 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14426 * tree.c (build_vl_exp_stat): Remove _stat from name.
14427 * tree.h (build_vl_exp): Remove macro.
14429 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14431 * tree.c (build_decl_stat): Remove _stat from name.
14432 * tree.h (build_decl): Remove macro.
14434 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14436 * gimple.c (gimple_build_with_ops_stat): Adjust.
14437 (gimple_alloc_stat): Remove _stat from name.
14438 * gimple.h (gimple_alloc): Remove macro.
14440 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14442 * tree.c (make_tree_vec_stat): Remove _stat from name.
14443 (grow_tree_vec_stat): Likewise.
14444 * tree.h (make_tree_vec_stat): Adjust prototype.
14445 (grow_tree_vec_stat): Likewise.
14446 (make_tree_vec): Remove macro.
14447 (grow_tree_vec): Likewise.
14449 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14451 * fold-const.c (fold_build1_stat_loc): Adjust.
14452 (fold_build2_stat_loc): Likewise.
14453 (fold_build3_stat_loc): Likewise.
14454 * tree.c (build0_stat): Remove _stat from name.
14455 (build1_stat): Likewise.
14456 (build2_stat): Likewise.
14457 (build3_stat): Likewise.
14458 (build4_stat): Likewise.
14459 (build5_stat): Likewise.
14460 * tree.h (build1_loc): Remove macro, and rename _stat function
14462 (build2_loc): Likewise.
14463 (build3_loc): Likewise.
14464 (build4_loc): Likewise.
14465 (build5_loc): Likewise.
14467 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14469 * tree.c (make_int_cst_stat): Remove _stat from name.
14470 * tree.h (make_int_cst_stat): Adjust prototype.
14471 (make_int_cst): Remove macro.
14473 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14475 * tree.c (make_tre_binfo_stat): Remove _stat from name.
14476 * tree.h (make_tree_binfo_stat): Adjust prototype.
14477 (make_tree_binfo): Remove.
14479 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14481 * tree.c (copy_node_stat): Rename to copy_node.
14482 (build_distinct_type_copy): Adjust.
14483 * tree.h (copy_node_stat): Adjust prototype.
14484 (copy_node): Remove macro.
14486 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14488 * tree.c (make_node_stat): rename to make_node.
14489 (build_tree_list_stat): Adjust.
14490 (build0_stat): Likewise.
14491 (build2_stat): Likewise.
14492 (build3_stat): Likewise.
14493 (build4_stat): Likewise.
14494 (build5_stat): Likewise.
14495 (build_decl_stat): Likewise.
14496 * tree.h (make_node_stat): Adjust prototype.
14497 (make_node): remove macro.
14499 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
14501 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
14502 (PPC_FEATURE2_SCV): Likewise.
14503 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
14505 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14507 * config/aarch64/aarch64.c
14508 (aarch64_internal_mov_immediate): Add new special pattern.
14509 * config/aarch64/aarch64.md (*movdi_aarch64):
14510 Add reg/32bit const mov case.
14512 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14513 Richard Sandiford <richard.sandiford@linaro.org>
14515 * config/aarch64/aarch64.md (mov<mode>): Generalize.
14516 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
14517 Add integer and movi cases.
14518 (movi-split-hf-df-sf split, fp16): New.
14519 (enabled): Added TARGET_FP_F16INST.
14520 * config/aarch64/iterators.md (GPF_HF): New.
14521 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
14523 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14525 * config/aarch64/aarch64.c
14526 (aarch64_simd_container_mode): Add prototype.
14527 (aarch64_expand_mov_immediate): Add HI support.
14528 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
14529 (aarch64_can_const_movi_rtx_p): New.
14530 (aarch64_preferred_reload_class):
14531 Remove restrictions of using FP registers for certain SIMD operations.
14532 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
14533 (aarch64_valid_floating_const): Add integer move validation.
14534 (aarch64_simd_imm_scalar_p): Remove.
14535 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
14536 (aarch64_legitimate_constant_p): Expand list of supported cases.
14537 * config/aarch64/aarch64-protos.h
14538 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
14539 (aarch64_reinterpret_float_as_int): New.
14540 (aarch64_simd_imm_scalar_p): Remove.
14541 * config/aarch64/constraints.md (Uvi): New.
14542 (Dd): Split into Ds and new Dd.
14543 * config/aarch64/aarch64.md (*movsi_aarch64):
14545 (*movdi_aarch64): Add SIMD mov case.
14547 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14549 * tree-predcom.c: (struct chain): Handle store-store chain in which
14550 stores for elimination only store loop invariant values.
14551 (execute_pred_commoning_chain): Ditto.
14552 (prepare_initializers_chain_store_elim): Ditto.
14553 (prepare_finalizers): Ditto.
14554 (is_inv_store_elimination_chain): New function.
14555 (initialize_root_vars_store_elim_1): New function.
14557 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14559 * tree-predcom.c: Revise general description of the pass.
14560 (enum chain_type): New enum type for store elimination.
14561 (struct chain): New field supporting store elimination.
14562 (struct component): Ditto.
14563 (dump_chain): Dump store-stores chain.
14564 (release_chain): Release resources.
14565 (split_data_refs_to_components): Compute and create component
14566 contains only stores for elimination.
14567 (get_chain_last_ref_at): New function.
14568 (make_invariant_chain): Initialization.
14569 (make_rooted_chain): Specify chain type in parameter and record it.
14570 (add_looparound_copies): Skip for store-stores chain.
14571 (determine_roots_comp): Compute type of chain and pass it to
14573 (initialize_root_vars_store_elim_2): New function.
14574 (finalize_eliminated_stores): New function.
14575 (remove_stmt): Handle store for elimination.
14576 (execute_pred_commoning_chain): Execute predictive commoning on
14577 store-store chains.
14578 (determine_unroll_factor): Skip unroll for store-stores chain.
14579 (prepare_initializers_chain_store_elim): New function.
14580 (prepare_initializers_chain): Hanlde store-store chain.
14581 (prepare_finalizers_chain, prepare_finalizers): New function.
14582 (tree_predictive_commoning_loop): Return integer value indicating
14583 if loop is unrolled or lcssa form is corrupted.
14584 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
14586 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14588 * tree-predcom.c (initialize_root): Delete.
14589 (execute_pred_commoning_chain): Initialize root vars and replace
14590 reference of non-combined chain directly, rather than call above
14593 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14595 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
14596 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
14598 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14600 * tree-predcom.c (struct chain): New field init_seq.
14601 (release_chain): Release init_seq.
14602 (prepare_initializers_chain): Record intialization stmts in above
14604 (insert_init_seqs): New function.
14605 (tree_predictive_commoning_loop): Call insert_init_seqs.
14607 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14609 * tree-predcom.c (determine_roots_comp): Skip trivial components.
14611 2017-07-28 Richard Biener <rguenther@suse.de>
14613 * match.pd: Remove superfluous :c.
14614 * genmatch.c (simplify::id): Add member.
14615 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
14617 (current_id): New global.
14618 (dt_node::parent): Move from ...
14619 (dt_operand::parent): ... here. Add for_id member.
14620 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
14621 (decision_tree::find_node): Relax order requirement when
14622 merging DT_TRUE nodes to ones inbetween the current simplify
14623 and the one we try to merge with. Add diagnostic whenever
14624 we need to enforce pattern order by not merging.
14625 (decision_tree::insert): Set current_id.
14626 (decision_tree::print_node): Dump parent node and for_id.
14627 (parser::last_id): Add member.
14628 (parser::push_simplify): Assign unique id.
14629 (parser::parser): Initialize last_id.
14631 2017-07-28 Martin Liska <mliska@suse.cz>
14634 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
14635 gimple_build_debug_bind.
14637 2017-07-28 Richard Biener <rguenther@suse.de>
14639 PR tree-optimization/81502
14640 * match.pd: Add pattern combining BIT_INSERT_EXPR with
14642 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
14644 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
14645 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
14646 for BIT_FIELD_REF args.
14647 * fold-const.c (make_bit_field_ref): Likewise.
14648 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
14650 2017-07-28 Jakub Jelinek <jakub@redhat.com>
14653 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
14654 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
14655 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
14656 Or it into SANITIZER_UNDEFINED.
14657 * ubsan.c: Include gimple-fold.h and varasm.h.
14658 (ubsan_expand_ptr_ifn): New function.
14659 (instrument_pointer_overflow): New function.
14660 (maybe_instrument_pointer_overflow): New function.
14661 (instrument_object_size): Formatting fix.
14662 (pass_ubsan::execute): Call instrument_pointer_overflow
14663 and maybe_instrument_pointer_overflow.
14664 * internal-fn.c (expand_UBSAN_PTR): New function.
14665 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
14666 * sanitizer.def (__ubsan_handle_pointer_overflow,
14667 __ubsan_handle_pointer_overflow_abort): New builtins.
14668 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
14669 * internal-fn.def (UBSAN_PTR): New internal function.
14670 * opts.c (sanitizer_opts): Add pointer-overflow.
14671 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
14672 * fold-const.c (build_range_check): Compute pointer range check in
14673 integral type if pointer arithmetics would be needed. Formatting
14676 2017-07-28 Martin Liska <mliska@suse.cz>
14679 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
14680 parameters that are of a variable-length.
14682 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
14684 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
14686 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
14687 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
14688 (CRT_CALL_STATIC_FUNCTION): Likewise.
14689 (ASM_DEFAULT_SPEC): New define.
14690 (ASM_SPEC32): Likewise.
14691 (ASM_SPEC64): Likewise.
14692 (ASM_SPEC_COMMON): Likewise.
14693 (ASM_SPEC): Likewise.
14694 (INVALID_64BIT): Likewise.
14695 (LINK_OS_DEFAULT_SPEC): Likewise.
14696 (LINK_OS_SPEC32): Likewise.
14697 (LINK_OS_SPEC64): Likewise.
14698 (POWERPC_LINUX): Likewise.
14699 (PTRDIFF_TYPE): Likewise.
14700 (RESTORE_FP_PREFIX): Likewise.
14701 (RESTORE_FP_SUFFIX): Likewise.
14702 (SAVE_FP_PREFIX): Likewise.
14703 (SAVE_FP_SUFFIX): Likewise.
14704 (SIZE_TYPE): Likewise.
14705 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
14706 (TARGET_64BIT): Likewise.
14707 (TARGET_64BIT): Likewise.
14708 (TARGET_AIX): Likewise.
14709 (WCHAR_TYPE_SIZE): Likewise.
14710 (WCHAR_TYPE): Undefine.
14711 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
14712 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
14713 (CPP_OS_RTEMS_SPEC): Delete.
14714 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
14715 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
14717 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
14719 2017-07-28 Jakub Jelinek <jakub@redhat.com>
14721 PR tree-optimization/81578
14722 * tree-parloops.c (build_new_reduction): Bail out if
14723 reduction_code isn't one of the standard OpenMP reductions.
14724 Move the details printing after that decision.
14726 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14728 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
14729 related to reload_in_progress.
14730 (splat_input_operand): Likewise.
14731 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
14733 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
14735 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
14736 (TARGET_INSTANTIATE_DECLS): Likewise.
14737 (legitimate_indexed_address_p): Delete reload_in_progress code.
14738 (rs6000_debug_legitimate_address_p): Likewise.
14739 (rs6000_eliminate_indexed_memrefs): Likewise.
14740 (rs6000_emit_le_vsx_store): Likewise.
14741 (rs6000_emit_move_si_sf_subreg): Likewise.
14742 (rs6000_emit_move): Likewise.
14743 (register_to_reg_type): Likewise.
14744 (rs6000_pre_atomic_barrier): Likewise.
14745 (rs6000_machopic_legitimize_pic_address): Likewise.
14746 (rs6000_allocate_stack_temp): Likewise.
14747 (rs6000_address_for_fpconvert): Likewise.
14748 (rs6000_address_for_altivec): Likewise.
14749 (rs6000_secondary_memory_needed_rtx): Delete function.
14750 (rs6000_check_sdmode): Likewise.
14751 (rs6000_alloc_sdmode_stack_slot): Likewise.
14752 (rs6000_instantiate_decls): Likewise.
14753 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
14754 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
14755 Delete reload_in_progress.
14756 (*vec_reload_and_plus_<mptrsize>): Likewise.
14757 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
14758 (vsx_div_v2di): Likewise.
14759 (vsx_udiv_v2di): Likewise.
14761 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14763 * config/rs6000/rs6000.opt (mlra): Replace with stub.
14764 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
14765 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
14766 (rs6000_debug_reg_global): Delete print of LRA status.
14767 (rs6000_option_override_internal): Delete dead LRA related code.
14768 (rs6000_lra_p): Delete function.
14769 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
14771 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
14773 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
14774 * config/riscv/rtems.h: New file.
14776 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14777 Sudakshina Das <sudi.das@arm.com>
14779 * config/aarch64/aarch64.md
14780 (define_split for and<mode>3nr_compare): Move
14781 non aarch64_logical_operand to a register.
14782 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
14783 register immediate operand to a register.
14784 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
14786 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14788 PR middle-end/81564
14789 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
14791 2017-07-27 Richard Biener <rguenther@suse.de>
14793 PR tree-optimization/81573
14794 PR tree-optimization/81494
14795 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
14796 multi defuse cycle case.
14798 2017-07-27 Richard Biener <rguenther@suse.de>
14800 PR tree-optimization/81571
14801 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
14804 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
14806 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
14807 earlier and only if MASK_FPU is set. Adjust formatting.
14809 2017-07-27 Martin Liska <mliska@suse.cz>
14811 * opt-functions.awk: Add validation of value of Init.
14812 * optc-gen.awk: Pass new argument.
14814 2017-07-27 Martin Liska <mliska@suse.cz>
14816 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
14817 Fix wrong condition.
14819 2017-07-27 Martin Liska <mliska@suse.cz>
14821 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
14822 BBs and edges seen by autoFDO.
14824 2017-07-27 Richard Biener <rguenther@suse.de>
14826 PR tree-optimization/81502
14827 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
14828 with incompatible but same sized type.
14829 (execute_update_addresses_taken): Likewise.
14831 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
14833 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
14834 flag_tree_loop_vectorize rather than flag_tree_vectorize.
14836 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14839 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
14840 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
14841 Change s_operand to memory_operand.
14843 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
14845 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
14846 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
14847 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
14848 Emit instructions rather than returning an expression. Handle TFmode
14849 and KFmode by casting to TImode.
14850 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
14851 (rs6000_emit_le_vsx_store): Likewise.
14852 * config/rs6000/vsx.md (VSX_TI): New iterator.
14853 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
14854 (*vsx_le_undo_permute_<mode>): Likewise.
14855 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
14856 emit the split sequence.
14857 (*vsx_le_perm_store_<mode>): Likewise.
14859 2017-07-27 Jakub Jelinek <jakub@redhat.com>
14861 PR tree-optimization/81555
14862 PR tree-optimization/81556
14863 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
14864 if true, force CHANGED for the recursive invocation.
14865 (reassociate_bb): Remember original length of ops array, pass
14866 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
14868 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
14869 attributes for noipa attribute. For naked attribute use
14870 lookup_attribute first before lookup_attribute_spec.
14871 * final.c (rest_of_handle_final): Disable IPA RA for functions with
14873 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
14874 for functions with noipa attribute.
14875 (cgraph_externally_visible_p): Return true for functions with noipa
14877 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
14878 for functions with noipa attribute.
14879 * doc/extend.texi: Document noipa function attribute.
14880 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
14881 also for functions with noipa attribute.
14882 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
14884 2017-07-26 Andrew Pinski <apinski@cavium.com>
14886 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
14887 vec_unalign_load_cost and vec_unalign_store_cost.
14889 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
14891 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
14892 -mvsx-small-integer option.
14893 (ISA_3_0_MASKS_IEEE): Likewise.
14894 (OTHER_VSX_VECTOR_MASKS): Likewise.
14895 (POWERPC_MASKS): Likewise.
14896 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
14897 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
14898 code, only testing for DImode being allowed in non-VSX floating
14900 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
14901 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
14903 (rs6000_option_override_internal): Delete -mvsx-small-integer.
14904 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
14905 TARGET_P8_VECTOR test.
14906 (rs6000_secondary_reload_simple_move): Likewise.
14907 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
14908 since TARGET_P9_VECTOR was already tested.
14909 (rs6000_opt_masks): Remove -mvsx-small-integer.
14910 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
14911 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
14913 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
14914 test for TARGET_VEXTRACTUB was used, and that uses
14916 (p9 extract splitter): Likewise.
14917 (vsx_extract_<mode>_di_p9): Likewise.
14918 (vsx_extract_<mode>_store_p9): Likewise.
14919 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
14920 for TARGET_P9_VECTOR was used. Delete code that is now dead with
14921 the elimination of TARGET_VSX_SMALL_INTEGER.
14922 (vsx_extract_<mode>_p8): Likewise.
14923 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
14924 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
14925 (vsx_set_<mode>_p9): Likewise.
14926 (vsx_set_v4sf_p9): Likewise.
14927 (vsx_set_v4sf_p9_zero): Likewise.
14928 (vsx_insert_extract_v4sf_p9): Likewise.
14929 (vsx_insert_extract_v4sf_p9_2): Likewise.
14930 * config/rs6000/rs6000.md (sign extend splitter): Change
14931 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
14932 (floatsi<mode>2_lfiwax_mem): Likewise.
14933 (floatunssi<mode>2_lfiwzx_mem): Likewise.
14934 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
14935 since a test for TARGET_P9_VECTOR was used.
14936 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
14937 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
14938 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
14939 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
14940 TARGET_P8_VECTOR test.
14941 (fix_trunc<mode>si2_stfiwx): Likewise.
14942 (fix_trunc<mode>si2_internal): Likewise.
14943 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
14944 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
14946 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
14947 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
14948 TARGET_P8_VECTOR test.
14949 (fixuns_trunc<mode>si2_stfiwx): Likewise.
14950 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
14951 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
14953 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
14954 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
14955 since a test for TARGET_P9_VECTOR was used.
14956 (splitter for loading small constants): Likewise.
14958 2017-07-26 Andrew Pinski <apinski@cavium.com>
14960 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
14963 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
14966 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
14967 (fp_valid_at): Likewise.
14969 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
14971 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
14972 (qdf24xx_addrcost_table): Likewise.
14973 (cortexa57_tunings): Update to use generic_branch_cost.
14974 (cortexa72_tunings): Likewise.
14975 (cortexa73_tunings): Likewise.
14976 (qdf24xx_tunings): Likewise.
14978 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
14980 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
14981 (thunderx2t99_branch_cost): Likewise.
14982 (cortexa35_tunings): Update to use generic_branch_cost.
14983 (cortexa53_tunings): Likewise.
14984 (cortexa57_tunings): Likewise.
14985 (cortexa72_tunings): Likewise.
14986 (cortexa73_tunings): Likewise.
14987 (thunderx2t99_tunings): Likewise.
14989 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
14991 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
14992 (sparc_option_override): Honour MASK_FSMULD.
14993 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
14994 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
14995 * config/sparc/sparc.opt (mfsmuld): New option.
14996 * doc/invoke.texi (mfsmuld): Document option.
14998 2017-07-26 Marek Polacek <polacek@redhat.com>
15000 PR middle-end/70992
15001 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
15003 2017-07-26 Richard Biener <rguenther@suse.de>
15005 * gimple-match-head.c (do_valueize): Return OP if valueize
15007 (get_def): New helper to get at the def stmt of a SSA name
15008 if valueize allows.
15009 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
15010 do_valueize to get at the def stmt.
15011 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
15013 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
15015 PR middle-end/46932
15016 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
15018 2017-07-26 Martin Liska <mliska@suse.cz>
15021 * function.c (expand_function_start): Make expansion of
15022 nonlocal_goto_save_area after parm_birth_insn.
15024 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
15026 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
15027 from all CPU target flags enable members.
15029 2017-07-26 Richard Biener <rguenther@suse.de>
15031 * genmatch.c (dt_simplify::gen): Make iterator vars const.
15032 (decision_tree::gen): Make 'type' const.
15033 (write_predicate): Likewise.
15035 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
15037 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
15038 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
15039 (rs6000_option_override_internal): Likewise.
15040 (rs6000_expand_vector_set): Likewise.
15041 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
15042 (TARGET_UPPER_REGS_SF): Likewise.
15043 (TARGET_UPPER_REGS_DI): Likewise.
15044 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
15045 (TARGET_DIRECT_MOVE_64BIT): Likewise.
15046 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
15047 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15048 (Splitters for DI constants in Altivec registers): Likewise.
15049 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
15050 (vsx_set_v4sf_p9): Likewise.
15051 (vsx_set_v4sf_p9_zero): Likewise.
15052 (vsx_insert_extract_v4sf_p9): Likewise.
15053 (vsx_insert_extract_v4sf_p9_2): Likewise.
15055 2017-07-25 Carl Love <cel@us.ibm.com>
15057 * doc/extend.texi: Update the built-in documentation file for the
15058 existing built-in functions
15059 vector signed char vec_cnttz (vector signed char);
15060 vector unsigned char vec_cnttz (vector unsigned char);
15061 vector signed short vec_cnttz (vector signed short);
15062 vector unsigned short vec_cnttz (vector unsigned short);
15063 vector signed int vec_cnttz (vector signed int);
15064 vector unsigned int vec_cnttz (vector unsigned int);
15065 vector signed long long vec_cnttz (vector signed long long);
15066 vector unsigned long long vec_cnttz (vector unsigned long long);
15068 2017-07-25 Andrew Pinski <apinski@cavium.com>
15070 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
15071 accesses where the use is for the first operand of a BIT_INSERT.
15073 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
15076 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
15077 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
15079 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
15081 * config/i386/gstabs.h: Delete.
15082 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
15084 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
15086 * config/i386/i386.c (ix86_decompose_address): Do not check for
15087 register RTX when looking at index_reg or base_reg.
15088 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
15090 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
15092 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
15093 to update EH info here.
15095 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15097 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
15099 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15101 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
15103 2017-07-25 Torsten Duwe <duwe@suse.de>
15105 * common.opt: Introduce -fpatchable-function-entry
15106 command line option, and its variables function_entry_patch_area_size
15107 and function_entry_patch_area_start.
15108 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
15109 including a two-value parser.
15110 * target.def (print_patchable_function_entry): New target hook.
15111 * targhooks.h (default_print_patchable_function_entry): New function.
15112 * targhooks.c (default_print_patchable_function_entry): Likewise.
15113 * toplev.c (process_options): Switch off IPA-RA if
15114 patchable function entries are being generated.
15115 * varasm.c (assemble_start_function): Look at the
15116 patchable-function-entry command line switch and current
15117 function attributes and maybe generate NOP instructions by
15118 calling the print_patchable_function_entry hook.
15119 * doc/extend.texi: Document patchable_function_entry attribute.
15120 * doc/invoke.texi: Document -fpatchable_function_entry
15121 command line option.
15122 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
15124 * doc/tm.texi: Re-generate.
15126 2017-07-25 Jakub Jelinek <jakub@redhat.com>
15129 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
15130 TARGET_AVX512DQ rather than TARGET_AVX512BW.
15132 2017-07-25 Tamar Christina <tamar.christina@arm.com>
15134 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
15136 2017-07-25 Richard Biener <rguenther@suse.de>
15138 PR tree-optimization/81455
15139 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
15140 not walk in cycles when looking for guards.
15142 2017-07-25 Richard Biener <rguenther@suse.de>
15144 PR tree-optimization/81529
15145 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
15146 when optimizing backedge uses.
15148 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
15150 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
15152 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
15153 to dl_section_ref. On AIX, append an expression to subtract
15154 the size of the section length to dl_section_ref.
15156 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
15158 * configure.ac: If any of the config.* scripts fail, exit 1.
15159 * configure: Regenerate.
15161 2017-07-25 Richard Biener <rguenther@suse.de>
15163 PR middle-end/81546
15164 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
15165 of immediate uses, be more verbose on errors.
15167 2017-07-25 Richard Biener <rguenther@suse.de>
15169 PR tree-optimization/81510
15170 * tree-vect-loop.c (vect_is_simple_reduction): When the
15171 reduction stmt is not inside the loop bail out.
15173 2017-07-25 Richard Biener <rguenther@suse.de>
15175 PR tree-optimization/81303
15176 * tree-vect-loop-manip.c (vect_loop_versioning): Build
15177 profitability check against LOOP_VINFO_NITERSM1.
15179 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15181 * domwalk.c (cmp_bb_postorder): Simplify.
15182 (sort_bbs_postorder): New function. Use it...
15183 (dom_walker::walk): ...here to optimize common cases.
15185 2017-07-25 Martin Liska <mliska@suse.cz>
15188 * ipa-visibility.c (function_and_variable_visibility): Make the
15189 redirection just on target that supports aliasing.
15190 Fix GNU coding style.
15192 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
15195 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
15197 * config/arm/rtems.h: Likewise.
15198 * config/bfin/rtems.h: Likewise.
15199 * config/i386/rtemself.h: Likewise.
15200 * config/lm32/rtems.h: Likewise.
15201 * config/m32c/rtems.h: Likewise.
15202 * config/m68k/rtemself.h: Likewise.
15203 * config/microblaze/rtems.h: Likewise.
15204 * config/mips/rtems.h: Likewise.
15205 * config/moxie/rtems.h: Likewise.
15206 * config/nios2/rtems.h: Likewise.
15207 * config/powerpcspe/rtems.h: Likewise.
15208 * config/rs6000/rtems.h: Likewise.
15209 * config/rtems.h: Likewise.
15210 * config/sh/rtems.h: Likewise.
15211 * config/sh/rtemself.h: Likewise.
15212 * config/sparc/rtemself.h: Likewise.
15214 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
15217 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
15218 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
15219 * tree-ssa-structalias.c (alias_get_name): Same.
15221 2017-07-25 Bin Cheng <bin.cheng@arm.com>
15224 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
15225 instructions if no du chain is found.
15227 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
15229 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
15231 2017-07-25 Richard Biener <rguenther@suse.de>
15233 PR middle-end/81505
15234 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
15237 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
15239 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
15240 upper-regs options.
15241 (ISA_2_7_MASKS_SERVER): Likewise.
15242 (ISA_3_0_MASKS_IEEE): Likewise.
15243 (OTHER_P8_VECTOR_MASKS): Likewise.
15244 (OTHER_VSX_VECTOR_MASKS): Likewise.
15245 (POWERPC_MASKS): Likewise.
15246 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
15247 duplicate list of options.
15248 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
15249 explicit -mupper-regs options.
15250 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
15251 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
15252 alias for -mupper-regs-df.
15253 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
15254 (rs6000_init_hard_regno_mode_ok): Likewise.
15255 (rs6000_option_override_internal): Likewise.
15256 (rs6000_opt_masks): Likewise.
15257 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
15258 options in terms of whether -mvsx or -mpower8-vector was used.
15259 (TARGET_UPPER_REGS_DI): Likewise.
15260 (TARGET_UPPER_REGS_SF): Likewise.
15261 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
15262 -mupper-regs-* options.
15264 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
15266 * passes.c (emergency_dump_function): Print some empty lines and a
15267 header before the RTL dump.
15269 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
15271 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
15273 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
15276 * config/aarch64/aarch64.c (aarch64_classify_symbol):
15277 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
15279 2017-07-24 Carl Love <cel@us.ibm.com>
15281 * config/rs6000/rs6000-c.c: Add support for built-in functions
15282 vector float vec_extract_fp32_from_shorth (vector unsigned short);
15283 vector float vec_extract_fp32_from_shortl (vector unsigned short);
15284 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
15285 vec_extract_fp_from_shortl): Add defines for the two builtins.
15286 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
15287 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
15289 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
15290 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
15291 * doc/extend.texi: Update the built-in documentation file for the
15292 new built-in function.
15294 2017-07-24 Jakub Jelinek <jakub@redhat.com>
15297 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
15299 * doc/generic.texi: Likewise.
15300 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
15301 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
15303 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
15305 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
15306 (aarch64_mls_elt_merge<mode>): Likewise.
15308 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
15310 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
15311 having __cxa_atexit.
15313 2017-07-23 Michael Collison <michael.collison@arm.com>
15315 * config/arm/arm.c (arm_option_override): Deprecate
15316 use of -mstructure-size-boundary.
15317 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
15318 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
15320 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15323 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
15324 Reduce cost estimate for direct moves.
15326 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
15329 * config/i386/i386.c (ix86_option_override_internal): Disable
15330 BMI, BMI2 and TBM instructions for -m16.
15332 2017-07-21 Carl Love <cel@us.ibm.com>
15334 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15335 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15336 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15337 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15338 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15339 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
15340 VMULOSW): New enum "unspec" values.
15341 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
15342 altivec_vmulosw): New patterns.
15343 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15344 VMULOSW): Add definitions.
15346 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
15348 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
15349 (qdf24xx): Likewise.
15350 * config/aarch64/aarch64-options-extensions.def (rdma); New.
15351 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
15352 (AARCH64_FL_V8_1): Renumber.
15353 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
15354 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
15355 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
15356 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
15357 rdma to feature modifiers list.
15359 2017-07-21 Yury Gribov <tetra2005@gmail.com>
15361 PR middle-end/56727
15362 * ipa-visibility (function_and_variable_visibility): Convert
15363 recursive PLT call to direct call if appropriate.
15365 2017-07-21 Andrew Pinski <apinski@cavium.com>
15367 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
15368 operand 1 to see if the types precision matches.
15369 * fold-const.c (operand_equal_p): Likewise.
15371 2017-07-21 Richard Biener <rguenther@suse.de>
15373 PR tree-optimization/81303
15374 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
15375 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
15376 (vect_peeling_hash_get_lowest_cost): Adjust.
15377 (vect_enhance_data_refs_alignment): Likewise. Use
15378 vect_get_peeling_costs_all_drs to compute the penalty for no
15379 peeling to match up costs.
15381 2017-07-21 Richard Biener <rguenther@suse.de>
15383 PR tree-optimization/81500
15384 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
15385 we didn't identify a reduction path.
15387 2017-07-21 Tom de Vries <tom@codesourcery.com>
15388 Cesar Philippidis <cesar@codesourcery.com>
15390 PR gcov-profile/81442
15391 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
15394 2017-07-21 Tom de Vries <tom@codesourcery.com>
15397 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
15399 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
15400 nvptx_override_options_after_change.
15402 2017-07-21 Ulrich Drepper <drepper@redhat.com>
15404 * dwarf2out.c (output_file_names): Avoid double testing for
15405 dwarf_version >= 5.
15407 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
15409 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
15411 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
15413 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
15415 (try_crossjump_to_edge): Do not punt on partitioned functions.
15417 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
15419 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
15420 Put all BBs reachable only via paths crossing cold region to cold
15422 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
15424 2016-07-21 Richard Biener <rguenther@suse.de>
15426 PR tree-optimization/81303
15427 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
15428 into account prologue and epilogue iterations when raising
15429 min_profitable_iters to sth at least covering one vector iteration.
15431 2017-07-21 Tamar Christina <tamar.christina@arm.com>
15433 * config/arm/arm.c (arm_test_cpu_arch_dat):
15436 2017-07-20 Nathan Sidwell <nathan@acm.org>
15438 Remove TYPE_METHODS.
15439 * tree.h (TYPE_METHODS): Delete.
15440 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
15441 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
15442 (dbxout_type_methods): Scan TYPE_FIELDS.
15443 (dbxout_type): Don't check TYPE_METHODS here.
15444 * function.c (use_register_for_decl): Always ignore register for
15445 class types when not optimizing.
15446 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
15447 * tree.c (free_lang_data_in_type): Stitch out member functions and
15448 templates from TYPE_FIELDS.
15449 (build_distinct_type_copy, verify_type_variant,
15450 verify_type): Member fns are on TYPE_FIELDS.
15451 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
15452 * tree-pretty-print.c (dump_generic_node): Likewise.
15454 2017-07-20 Jakub Jelinek <jakub@redhat.com>
15457 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
15458 V2TImode and V4TImode.
15459 (ix86_expand_vector_extract): Likewise.
15460 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
15461 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
15462 (ssescalarmode): Handle V4TImode and V2TImode.
15463 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
15464 (*vec_extractv2ti, *vec_extractv4ti): New insns.
15465 (VEXTRACTI128_MODE): New mode iterator.
15466 (splitter for *vec_extractv?ti first element): New.
15467 (VEC_INIT_MODE): New mode iterator.
15468 (vec_init<mode>): Consolidate 3 expanders into one using
15469 VEC_INIT_MODE mode iterator.
15471 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
15473 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
15474 non_spilled_static_chain_regno_p.
15476 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
15478 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
15480 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
15482 * bb-reorder.c (connect_traces): Allow copying of blocks within
15485 2017-07-20 Richard Biener <rguenther@suse.de>
15487 * gimple.h (gimple_phi_result): Add gphi * overload.
15488 (gimple_phi_result_ptr): Likewise.
15489 (gimple_phi_arg): Likewise. Adjust index assert to only
15490 allow actual argument accesses rather than all slots available
15492 (gimple_phi_arg_def): Add gphi * overload.
15493 * tree-phinodes.c (make_phi_node): Initialize only actual
15495 (resize_phi_node): Clear memory not covered by old node,
15496 do not initialize excess argument slots.
15497 (reserve_phi_args_for_new_edge): Initialize new argument slot
15500 2017-07-20 Bin Cheng <bin.cheng@arm.com>
15502 PR tree-optimization/81388
15504 2016-07-21 Bin Cheng <bin.cheng@arm.com>
15506 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
15507 by removing computation of may_be_zero.
15509 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15510 Tom de Vries <tom@codesourcery.com>
15512 PR middle-end/81030
15513 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
15514 when gimple level profile disagrees with what RTL expander did.
15516 2017-07-20 Richard Biener <rguenther@suse.de>
15518 PR tree-optimization/61171
15519 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
15520 (vect_analyze_stmt): Add slp instance parameter.
15521 (vectorizable_reduction): Likewise.
15522 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
15523 (vect_is_simple_reduction): Deal with chains not detected
15524 as SLP reduction chain, specifically not properly associated
15525 chains containing a mix of plus/minus.
15526 (get_reduction_op): Remove.
15527 (get_initial_defs_for_reduction): Simplify, pass in whether
15528 this is a reduction chain, pass in the SLP node for the PHIs.
15529 (vect_create_epilog_for_reduction): Get the SLP instance as
15531 (vectorizable_reduction): Get the SLP instance as arg.
15532 During analysis remember the SLP node with the PHIs in the
15533 instance. Simplify getting at the vectorized reduction PHIs.
15534 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
15535 through SLP instance.
15536 (vect_slp_analyze_operations): Likewise.
15537 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
15538 (vect_transform_stmt): Likewise.
15540 2017-07-20 Tom de Vries <tom@codesourcery.com>
15542 PR tree-optimization/81489
15543 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
15544 read of phi arg location to before loop that modifies phi.
15546 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15548 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
15551 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15553 PR middle-end/81331
15554 * except.c (execute): Fix ordering issue.
15556 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
15558 PR rtl-optimization/81423
15559 * combine.c (make_compound_operation_int): Don't try to optimize
15560 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
15562 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
15564 PR rtl-optimization/81423
15565 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
15566 with a constant that is -1 in the truncated to mode.
15568 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15570 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
15571 (determine_unlikely_bbs): ... here.
15572 * predict.h (propagate_unlikely_bbs_forward): Declare.
15573 * cfgexpand.c (pass_expand::execute): Use it.
15574 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
15576 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
15577 propagate_unlikely_bbs_forward.
15579 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15581 PR middle-end/81331
15582 * except.c (maybe_add_nop_after_section_switch): New function.
15585 2017-07-19 Tom de Vries <tom@codesourcery.com>
15587 * gimple.h (gimple_phi_set_arg): Make assert more strict.
15589 2017-07-19 Tom de Vries <tom@codesourcery.com>
15591 * gimple.h (gimple_phi_arg): Make assert more strict.
15593 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
15595 * config.gcc (powerpc*-*-*): Add mmintrin.h.
15596 * config/rs6000/mmintrin.h: New file.
15597 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
15599 2017-07-19 Jakub Jelinek <jakub@redhat.com>
15601 PR tree-optimization/81346
15602 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
15604 2017-07-19 Tom de Vries <tom@codesourcery.com>
15606 * config/nvptx/nvptx.md (VECIM): Add V2DI.
15608 2017-07-19 Tom de Vries <tom@codesourcery.com>
15610 * config/nvptx/nvptx-modes.def: Add V2DImode.
15611 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
15612 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
15613 (nvptx_output_mov_insn): Handle lack of mov.b128.
15614 (nvptx_print_operand): Handle 'H' and 'L' codes.
15615 (nvptx_vector_mode_supported): Allow V2DImode.
15616 (nvptx_preferred_simd_mode): New function.
15617 (nvptx_data_alignment): New function.
15618 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
15619 nvptx_preferred_simd_mode.
15620 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
15622 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
15624 2017-07-19 Tom de Vries <tom@codesourcery.com>
15626 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
15627 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
15628 (nvptx_vector_mode_supported): New function. Allow V2SImode.
15629 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
15630 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
15631 (mov<VECIM>_insn): New define_insn.
15632 (define_expand "mov<VECIM>): New define_expand.
15634 2017-07-19 Tom de Vries <tom@codesourcery.com>
15636 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
15638 2017-07-19 Jakub Jelinek <jakub@redhat.com>
15640 PR tree-optimization/81346
15641 * fold-const.h (fold_div_compare, range_check_type): Declare.
15642 * fold-const.c (range_check_type): New function.
15643 (build_range_check): Use range_check_type.
15644 (fold_div_compare): No longer static, rewritten into
15645 a match.pd helper function.
15646 (fold_comparison): Don't call fold_div_compare here.
15647 * match.pd (X / C1 op C2): New optimization using fold_div_compare
15648 as helper function.
15650 2017-07-19 Nathan Sidwell <nathan@acm.org>
15652 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
15653 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
15654 * tree.c (find_decls_types_r, verify_type): Use
15655 TYPE_{MIN,MAX}_VALUE_RAW.
15656 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
15657 (hash_tree): Likewise.
15658 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
15660 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
15663 2017-07-18 Tom de Vries <tom@codesourcery.com>
15665 PR middle-end/81464
15666 * omp-expand.c (expand_omp_for_static_chunk): Handle
15667 equal-argument loop exit phi.
15669 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
15672 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
15673 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
15674 operand 2 predicate.
15675 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
15676 operand 2 predicate.
15677 (ror,rol -> rorx splitters): Use const_int_operand as
15678 operand 2 predicate.
15680 2017-06-18 Richard Biener <rguenther@suse.de>
15682 PR tree-optimization/81410
15683 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
15684 the gap in the ! slp_perm SLP case after each group.
15686 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15688 PR middle-end/81463
15689 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
15692 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15694 PR middle-end/81462
15695 * predict.c (set_even_probabilities): Cleanup; do not affect
15696 probabilities that are already known.
15697 (combine_predictions_for_bb): Call even when count is set.
15699 2017-07-18 Nathan Sidwell <nathan@acm.org>
15701 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
15704 2017-07-18 Bin Cheng <bin.cheng@arm.com>
15707 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
15708 optimization for loop niter analysis.
15710 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
15713 * config/avr/avr.c (avr_optimize_casesi): Don't use
15714 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
15716 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
15718 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
15719 body_cost_vec from _vect_peel_extended_info.
15720 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
15721 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
15724 2017-07-18 Bin Cheng <bin.cheng@arm.com>
15726 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
15728 2017-07-18 Richard Biener <rguenther@suse.de>
15730 PR tree-optimization/80620
15731 PR tree-optimization/81403
15732 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
15733 info when re-using a VN table entry.
15735 2017-07-18 Richard Biener <rguenther@suse.de>
15737 PR tree-optimization/81418
15738 * tree-vect-loop.c (vectorizable_reduction): Properly compute
15739 vectype_in. Verify that with lane-reducing reduction operations
15740 we have a single def-use cycle.
15742 2017-07-17 Carl Love <cel@us.ibm.com>
15744 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
15746 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15747 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15748 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15749 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15750 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15751 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
15752 VMULOSW): New enum "unspec" values.
15753 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
15754 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
15755 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
15756 altivec_vmulosw): New patterns.
15757 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15758 VMULOSW): Add definitions.
15759 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
15761 * config/alpha/alpha.c: Include predict.h.
15763 2017-07-17 Yury Gribov <tetra2005@gmail.com>
15765 * tree-vrp.c (compare_assert_loc): Fix comparison function
15766 to return predictable results.
15768 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15770 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
15772 (subdi3): Likewise.
15773 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
15774 * doc/invoke.texi (mexpand-adddi): Update text.
15776 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15778 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
15779 that also clobbers the CC register. The old expand code is moved
15781 (*arc_clzsi2): ... here.
15782 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
15783 the CC register. The old expand code is moved to ...
15784 (arc_ctzsi2): ... here.
15786 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15788 * config/arc/arc.opt (mindexed-loads): Use initial value
15789 TARGET_INDEXED_LOADS_DEFAULT.
15790 (mauto-modify-reg): Use initial value
15791 TARGET_AUTO_MODIFY_REG_DEFAULT.
15792 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
15793 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
15794 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
15795 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
15797 2017-07-17 Martin Liska <mliska@suse.cz>
15800 * opts.c (finish_options): Do not allow -fgnu-tm
15801 w/ -fsanitize={kernel-,}address. Say sorry.
15803 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15806 * tree-loop-distribution.c (classify_partition): Only assert on
15807 numer of iterations.
15808 (merge_dep_scc_partitions): Delete prameter. Update function call.
15809 (distribute_loop): Remove code handling loop with unknown niters.
15810 (pass_loop_distribution::execute): Skip loop with unknown niters.
15812 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15815 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
15816 function sort_partitions_by_post_order.
15818 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15820 PR tree-optimization/81374
15821 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
15822 the max index of basic blocks, rather than number of basic blocks.
15824 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15826 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
15828 (arc_legitimate_pic_operand_p): Likewise.
15829 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
15831 (arc_needs_pcl_p): Likewise.
15832 (arc_legitimate_pc_offset_p): Likewise.
15833 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
15834 function is also used in constrains.md.
15835 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
15836 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
15837 PLUS. Only return true/false in known cases, otherwise assert.
15838 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
15839 is already called in arc_legitimate_constant_p.
15840 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
15842 (LEGITIMATE_PIC_OPERAND_P): Use
15843 arc_raw_symbolic_reference_mentioned_p function.
15844 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
15849 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15850 Andrew Burgess <andrew.burgess@embecosm.com>
15852 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
15853 (arc_return_address_register): New function.
15854 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
15855 (arc_handle_fndecl_attribute): Add naked attribute.
15856 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
15857 (TARGET_WARN_FUNC_RETURN): Likewise.
15858 (arc_allocate_stack_slots_for_args): New function.
15859 (arc_warn_func_return): Likewise.
15860 (machine_function): Change type fn_type.
15861 (arc_compute_function_type): Consider new naked function type,
15862 change function return type.
15863 (arc_must_save_register): Adapt to handle new
15864 arc_compute_function_type's return type.
15865 (arc_expand_prologue): Likewise.
15866 (arc_expand_epilogue): Likewise.
15867 (arc_return_address_regs): Delete.
15868 (arc_return_address_register): New function.
15869 (arc_epilogue_uses): Use above function.
15870 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
15871 (arc_function_type): Change encoding, add naked type.
15872 (ARC_INTERRUPT_P): Change to handle the new encoding.
15873 (ARC_FAST_INTERRUPT_P): Likewise.
15874 (ARC_NORMAL_P): Define.
15875 (ARC_NAKED_P): Likewise.
15876 (arc_compute_function_type): Delete prototype.
15877 * config/arc/arc.md (in_ret_delay_slot): Use
15878 arc_return_address_register function.
15879 (simple_return): Likewise.
15880 (p_return_i): Likewise.
15882 2017-07-17 Jakub Jelinek <jakub@redhat.com>
15884 PR tree-optimization/81428
15885 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
15886 can't be built for those types.
15888 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
15890 Remove stuff dead since r239246.
15892 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
15893 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
15894 (avr_inform_devices): Remove dead stuff.
15896 2017-07-17 Tamar Christina <tamar.christina@arm.com>
15898 * config/arm/arm_neon.h: Fix softp typo.
15900 2017-07-17 Jakub Jelinek <jakub@redhat.com>
15902 PR tree-optimization/81365
15903 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
15904 aggregate moves onto bb predecessor edges, make sure there are no
15905 loads that could alias the lhs in between the start of bb and the
15908 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
15911 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
15912 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
15913 [LSHIFTRT, outer_code = TRUNCATE]: Same.
15915 2017-07-17 Jakub Jelinek <jakub@redhat.com>
15917 PR tree-optimization/81396
15918 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
15919 (init_symbolic_number): Initialize it to 1.
15920 (perform_symbolic_merge): Add n_ops from both operands into the new
15922 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
15923 without base_addr as useless if they need more than one operation.
15924 (bswap_replace): Handle !bswap case for NULL base_addr.
15926 2017-07-17 Tom de Vries <tom@codesourcery.com>
15929 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
15932 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
15934 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
15935 conditional builtin define __FIX_LEON3FT_B2BST.
15937 2017-07-17 Daniel Cederman <cederman@gaisler.com>
15939 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
15940 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
15943 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
15945 PR rtl-optimization/81424
15946 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
15947 to remove potential trapping from operands if -fnon-call-exceptions.
15949 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15951 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
15952 profile_proability for scalling.
15953 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
15955 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15957 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
15959 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15961 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
15962 fixpoint arithmetics.
15964 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15966 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
15967 fixpoint arithmetics.
15969 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15971 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
15972 fixpoint arithmetics.
15974 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15976 * profile-count.h (profile_probability::from_reg_br_prob_note,
15977 profile_probability::to_reg_br_prob_note): New functions.
15978 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
15979 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
15980 * predict.c (probability_reliable_p): Update.
15981 (edge_probability_reliable_p): Update.
15982 (br_prob_note_reliable_p): Update.
15983 (invert_br_probabilities): Update.
15984 (add_reg_br_prob_note): New function.
15985 (combine_predictions_for_insn): Update.
15986 * asan.c (asan_clear_shadow): Update.
15987 * cfgbuild.c (compute_outgoing_frequencies): Update.
15988 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
15989 (update_br_prob_note): Update.
15990 (rtl_verify_edges): Update.
15991 (purge_dead_edges): Update.
15992 (fixup_reorder_chain): Update.
15993 * emit-rtl.c (try_split): Update.
15994 * ifcvt.c (cond_exec_process_insns): Update.
15995 (cond_exec_process_if_block): Update.
15996 (dead_or_predicable): Update.
15997 * internal-fn.c (expand_addsub_overflow): Update.
15998 (expand_neg_overflow): Update.
15999 (expand_mul_overflow): Update.
16000 * loop-doloop.c (doloop_modify): Update.
16001 * loop-unroll.c (compare_and_jump_seq): Update.
16002 * optabs.c (emit_cmp_and_jump_insn_1): Update.
16003 * predict.h: Update.
16004 * reorg.c (mostly_true_jump): Update.
16006 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
16007 * config/alpha/alpha.c (emit_unlikely_jump): Update.
16008 * config/arc/arc.c: (emit_unlikely_jump): Update.
16009 * config/arm/arm.c: (emit_unlikely_jump): Update.
16010 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
16011 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
16012 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
16013 (ix86_print_operand): Update.
16014 (ix86_split_fp_branch): Update.
16015 (predict_jump): Update.
16016 * config/ia64/ia64.c (ia64_print_operand): Update.
16017 * config/mmix/mmix.c (mmix_print_operand): Update.
16018 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
16019 (rs6000_expand_split_stack_prologue): Update.
16020 * config/rs6000/rs6000.c: Update.
16021 * config/s390/s390.c (s390_expand_vec_strlen): Update.
16022 (s390_expand_vec_movstr): Update.
16023 (s390_expand_cs_tdsi): Update.
16024 (s390_expand_split_stack_prologue): Update.
16025 * config/sh/sh.c (sh_print_operand): Update.
16026 (expand_cbranchsi4): Update.
16027 (expand_cbranchdi4): Update.
16028 * config/sparc/sparc.c (output_v9branch): Update.
16029 * config/spu/spu.c (get_branch_target): Update.
16030 (ea_load_store_inline): Update.
16031 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
16032 * config/tilepro/tilepro.c: Update.
16034 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
16036 * gimplify.c (mostly_copy_tree_r): Revert latest change.
16037 (gimplify_save_expr): Likewise.
16039 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16041 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
16043 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16045 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
16047 * timevar.def (TV_IPA_FNSUMMARY): Define.
16049 2017-07-16 Daniel Cederman <cederman@gaisler.com>
16051 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
16052 to back store errata sensitive sequence from being generated.
16053 (sqrtdf2_fix): Likewise.
16055 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16057 * tree-ssa-threadupdate.c (compute_path_counts,
16058 update_joiner_offpath_counts): Use profile_probability.
16060 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
16063 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16065 * config/arm/arm-c.c (arm_cpu_builtins): Define
16066 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
16068 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
16070 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16071 array entries to represent __ieee128 versions of the
16072 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
16073 scalar_extract_sig, and scalar_insert_exp built-in functions.
16074 (altivec_resolve_overloaded_builtin): Add special case handling
16075 for the __builtin_scalar_insert_exp function, as represented by
16076 the P9V_BUILTIN_VEC_VSIEDP constant.
16077 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
16078 exponent support for __ieee128 argument.
16079 (VSESQP): Add scalar extract signature support for __ieee128
16081 (VSTDCNQP): Add scalar test negative support for __ieee128
16083 (VSIEQP): Add scalar insert exponent support for __int128 argument
16084 with __ieee128 result.
16085 (VSIEQPF): Add scalar insert exponent support for __ieee128
16086 argument with __ieee128 result.
16087 (VSTDCQP): Add scalar test data class support for __ieee128
16089 (VSTDCNQP): Add overload support for scalar test negative with
16090 __ieee128 argument.
16091 (VSTDCQP): Add overload support for scalar test data class
16092 __ieee128 argument.
16093 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
16094 UNSPEC_VSX_SXSIGDP.
16095 (UNSPEC_VSX_SIEXPQP): New constant.
16096 (xsxexpqp): New insn for VSX scalar extract exponent quad
16098 (xsxsigqp): New insn for VSX scalar extract significand quad
16100 (xsiexpqpf): New insn for VSX scalar insert exponent quad
16101 precision with floating point argument.
16102 (xststdcqp): New expand for VSX scalar test data class quad
16104 (xststdcnegqp): New expand for VSX scalar test negative quad
16106 (xststdcqp): New insn to match expansions for VSX scalar test data
16107 class quad precision and VSX scalar test negative quad precision.
16108 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
16109 special case operand checking to enforce that second operand of
16110 VSX scalar test data class with quad precision argument is a 7-bit
16112 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
16113 prototypes and descriptions of __ieee128 versions of
16114 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
16115 scalar_test_data_class, and scalar_test_neg built-in functions.
16117 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16119 PR tree-optimization/81162
16120 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
16121 replace a negate with an add.
16123 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
16125 * doc/invoke.texi (arm/-mcpu): Document +crypto.
16127 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16129 * config/arm/arm-c.c (arm_cpu_builtins): Define
16130 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
16132 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16134 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
16135 (armv8-r): Set ARM Cortex-R52 as default CPU.
16136 * config/arm/arm-tables.opt: Regenerate.
16137 * config/arm/arm-tune.md: Regenerate.
16138 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
16140 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
16141 extension for -mcpu=cortex-r52.
16143 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16145 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
16146 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
16147 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
16148 (fp-armv8): Define it as FP_ARMv8 only.
16149 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
16150 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
16152 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
16153 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
16154 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
16155 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
16156 than TARGET_FPU_ARMV8.
16157 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
16158 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
16159 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
16161 * config/arm/neon.md (neon_vrint): Likewise.
16162 (neon_vcvt): Likewise.
16163 (neon_<fmaxmin_op><mode>): Likewise.
16164 (<fmaxmin><mode>3): Likewise.
16165 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
16166 * config/arm/predicates.md (arm_cond_move_operator): Check against
16167 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
16169 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
16171 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
16172 to top of function.
16174 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16176 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
16177 loop in comment with memset.
16179 2017-07-14 Martin Liska <mliska@suse.cz>
16181 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
16182 * dwarf2out.c (is_java): Remove the function.
16183 (output_pubname): Remove usage of the function.
16184 (lower_bound_default): Remove usage of DW_LANG_Java.
16185 (gen_compile_unit_die): Likewise.
16186 * gcc.c: Remove compiler defaults for .java and .zip files.
16187 * gimple-expr.c (remove_suffix): Change as there's no longer
16188 extension than 4-letter one.
16189 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
16190 (gimplify_save_expr): Likewise.
16191 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
16192 as it's possible even for other languages than Java.
16193 * langhooks.h (struct lang_hooks): Remove Java from a comment.
16194 * lto-opts.c (lto_write_options): Remove reference to Java.
16195 * opts.c (strip_off_ending): Update file extension handling.
16196 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
16197 * tree-eh.c (lower_resx): Likewise.
16198 * tree.c (free_lang_data_in_type): Remove dead code.
16199 (find_decls_types_r): Likewise.
16200 (build_common_builtin_nodes): Remove Java from a comment.
16201 (verify_type): Remove dead code.
16202 * varasm.c (assemble_external): Remove Java from a comment.
16204 2017-07-14 Martin Liska <mliska@suse.cz>
16206 * opts.c (finish_options): Add quotes.
16207 (common_handle_option): Likewise.
16209 2017-07-14 Martin Liska <mliska@suse.cz>
16211 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
16212 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
16213 Remove N_SO_PASCAL.
16214 * dwarf2out.c (lower_bound_default): Do not handle
16216 (gen_compile_unit_die): Likewise.
16217 * gcc.c: Remove default extension binding for GNU Pascal.
16218 * stmt.c: Remove Pascal language from a comment.
16219 * xcoffout.c: Likewise.
16221 2017-07-13 David Malcolm <dmalcolm@redhat.com>
16224 * diagnostic-show-locus.c (fixit_cmp): New function.
16225 (layout::layout): Sort m_fixit_hints.
16226 (column_range::column_range): Assert that the values are valid.
16227 (struct char_span): New struct.
16228 (correction::overwrite): New method.
16229 (struct source_line): New struct.
16230 (line_corrections::add_hint): Add assertions. Reimplement memcpy
16231 calls in terms of classes source_line and char_span, and
16232 correction::overwrite.
16233 (selftest::test_overlapped_fixit_printing_2): New function.
16234 (selftest::diagnostic_show_locus_c_tests): Call it.
16236 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
16238 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
16239 early if there is no lhs.
16241 2017-07-13 Martin Liska <mliska@suse.cz>
16243 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
16244 (gen_reference_type_die): Likewise.
16245 * stor-layout.c: Remove Pascal-related comment.
16247 2017-07-13 Martin Liska <mliska@suse.cz>
16249 * opts.c (finish_options): Add quotes to error messages.
16250 (parse_sanitizer_options): Likewise.
16252 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
16254 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
16256 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
16258 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
16260 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
16262 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
16264 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
16266 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
16269 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
16270 provides the hardware capability bits, define the macro
16271 __BUILTIN_CPU_SUPPORTS__.
16272 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
16273 if GLIBC does not provide the hardware capability bits. Add a
16274 gcc_unreachable call if the built-in cpu function is neither
16275 __builtin_cpu_is nor __builtin_cpu_supports.
16276 (rs6000_get_function_versions_dispatcher): Change the warning
16277 that an old GLIBC is used which does not export the capability
16278 bits to be an error.
16279 * doc/extend.texi (target_clones attribute): Document the
16280 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
16281 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
16282 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
16283 the macros defined by GCC if the newer GLIBC is available.
16285 2017-07-12 Jeff Law <law@redhat.com>
16287 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
16288 remaining includes slightly.
16289 * config/riscv/riscv-builtins.c: Include profile-count.h.
16291 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
16294 * config/avr/avr.c (avr_set_current_function): In diagnostic
16295 messages: Quote keywords and (parts of) identifiers.
16296 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
16299 2017-07-12 Carl Love <cel@us.ibm.com>
16301 * config/rs6000/rs6000-c.c: Add support for built-in functions
16302 vector bool char vec_revb (vector bool char);
16303 vector bool short vec_revb (vector short char);
16304 vector bool int vec_revb (vector bool int);
16305 vector bool long long vec_revb (vector bool long long);
16306 * doc/extend.texi: Update the built-in documentation file for the
16307 new built-in functions.
16309 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16311 * config/s390/s390.md: Remove movcc splitter.
16313 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16315 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
16316 load/store on condition.
16318 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
16321 * config/avr/avr.c (avr_encode_section_info)
16322 [progmem && !TREE_READONLY]: Error if progmem object needs
16325 2017-07-11 Michael Collison <michael.collison@arm.com>
16327 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
16330 2017-07-11 Carl Love <cel@us.ibm.com>
16332 * config/rs6000/rs6000-c.c: Add support for builtins
16333 vector unsigned int vec_parity_lsbb (vector signed int);
16334 vector unsigned int vec_parity_lsbb (vector unsigned int);
16335 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
16336 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
16337 vector unsigned long long vec_parity_lsbb (vector signed long long);
16338 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
16339 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
16340 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
16341 * doc/extend.texi: Update the built-in documentation file for the
16342 new built-in functions.
16344 2017-07-11 David Malcolm <dmalcolm@redhat.com>
16346 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
16347 (layout::m_primary_loc): New field.
16348 (layout::layout): Initialize new field. Move location filtering
16349 logic from here to...
16350 (layout::maybe_add_location_range): ...this new method. Add
16351 support for filtering to just the lines already specified by other
16353 (layout::will_show_line_p): New method.
16354 (gcc_rich_location::add_location_if_nearby): New method.
16355 (selftest::test_add_location_if_nearby): New test function.
16356 (selftest::diagnostic_show_locus_c_tests): Call it.
16357 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
16360 2017-07-11 Tom de Vries <tom@codesourcery.com>
16362 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
16363 (bb_first_real_insn): New function.
16364 (nvptx_single): Add extra initialization of broadcasted condition
16367 2017-07-11 Nathan Sidwell <nathan@acm.org>
16369 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
16371 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
16373 * doc/extend.texi (AVR Function Attributes): Remove weblink to
16374 Binutils doc as TEXI will mess them up.
16375 * doc/invoke.texi (AVR Options): Same here.
16377 2017-07-11 Daniel Cederman <cederman@gaisler.com>
16379 * config/sparc/sparc.opt (mfix-ut700): New option.
16380 (mfix-gr712rc): Likewise.
16381 (sparc_fix_b2bst): New variable.
16382 * doc/invoke.texi (SPARC options): Document them.
16383 (ARM options): Fix warnings.
16384 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
16385 instructions to prevent sequences that can trigger the store-store
16386 errata for certain LEON3FT processors.
16387 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
16388 (sparc_option_override): Set sparc_fix_b2bst appropriately.
16389 * config/sparc/sparc.md (fix_b2bst): New attribute.
16390 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
16392 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
16395 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
16397 (*rsqrtsf2_sse): Ditto.
16399 (div<mode>3): Macroize insn from divdf3 and divsf3
16400 using MODEF mode iterator.
16402 2017-07-10 Martin Sebor <msebor@redhat.com>
16404 PR tree-optimization/80397
16405 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
16406 instead of testing for equality to INTEGER_TYPE.
16408 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
16410 * config.gcc: Remove uclibc from arc target spec.
16412 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
16414 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
16416 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16419 * lto-wrapper.c (remove_option): New function.
16420 (merge_and_complain): Merge PIC/PIE options more realistically.
16422 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
16424 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
16428 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
16429 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
16430 * config.in: Regenerate.
16431 * configure: Regenerate.
16432 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
16433 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
16434 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
16435 (TARGET_GASISR_PROLOGUES): ...target mask.
16436 * common/config/avr/avr-common.c
16437 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
16438 Set -mgas-isr-prologues.
16439 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
16440 INSERT_PASS_BEFORE for it.
16441 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
16442 * config/avr/avr.c (avr_option_override)
16443 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
16444 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
16445 (avr_attribute_table) <no_gccisr>: Add new function attribute.
16446 (avr_set_current_function) <is_no_gccisr>: Init machine field.
16447 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
16449 (make_avr_pass_pre_proep): New function.
16450 (emit_push_sfr) <treg>: Add argument to function and use it
16451 instead of TMP_REG.
16452 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
16453 and set machine->gasisr.yes.
16454 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
16455 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
16456 __gcc_isr.n_pushed to .L__stack_usage.
16457 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
16458 (avr_asm_final_postscan_insn): ...this new static function.
16459 * config/avr/avr.h (machine_function)
16460 <is_no_gccisr, use_L__stack_usage>: New fields.
16461 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
16462 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
16463 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
16464 (gasisr, *gasisr): New expander and insn.
16465 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
16466 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
16467 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
16469 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
16471 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
16474 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
16476 Move jump-tables out of .text again.
16479 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
16480 (ASM_OUTPUT_ADDR_VEC): New function.
16481 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
16482 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
16483 INSN_ADDRESSes as asm comment.
16484 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
16485 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
16486 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
16487 * config/avr/avr.md (*tablejump): Adjust comment.
16488 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
16489 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
16491 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
16492 (avr_output_addr_vec): New proto.
16493 (avr_log_t) <insn_addresses>: New field.
16495 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
16498 * config/i386/i386.c (ix86_function_arg_advance): Set
16499 outgoing_args_on_stack to true if there are outgoing arguments
16501 (ix86_function_arg): Likewise.
16502 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
16503 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
16504 * config/i386/i386.h (machine_function): Add
16505 outgoing_args_on_stack.
16507 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
16509 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
16510 supporting pthreds.
16511 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
16513 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
16515 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
16516 (REAL_H): Remove $(MACHMODE_H).
16517 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
16519 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
16520 $(MACHMODE_H) and double-int.h.
16521 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
16523 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
16526 2017-07-07 Andrew Pinski <apinski@cavium.com>
16528 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
16529 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
16531 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16533 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
16534 Add warning if GCC was not configured to link against a GLIBC that
16535 exports the hardware capability bits.
16536 (make_resolver_func): Make resolver function private and not a
16537 COMDAT function. Create the name with clone_function_name instead
16538 of make_unique_name.
16541 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
16542 correct operand in doing the split.
16544 2017-07-07 Carl Love <cel@us.ibm.com>
16546 * config/rs6000/rs6000-c: Add support for built-in function
16547 vector unsigned short vec_pack_to_short_fp32 (vector float,
16549 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
16550 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
16551 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
16552 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
16553 (convert_4f32_8i16): Add define_expand.
16554 * doc/extend.texi: Update the built-in documentation file for the
16555 new built-in function.
16557 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16559 * config/sparc/m8.md: New file.
16560 * config/sparc/sparc.md: Include m8.md.
16562 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16564 * config/sparc/sparc.opt: New option -mvis4b.
16565 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
16566 (sparc_option_override): Handle VIS4B.
16567 (enum sparc_builtins): Define
16568 SPARC_BUILTIN_DICTUNPACK{8,16,32},
16569 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
16570 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
16571 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
16572 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
16573 (check_constant_argument): New function.
16574 (sparc_vis_init_builtins): Define builtins
16575 __builtin_vis_dictunpack{8,16,32},
16576 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
16577 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
16578 __builtin_vis_fpcmpde{8,16,32}shl and
16579 __builtin_vis_fpcmpur{8,16,32}shl.
16580 (sparc_expand_builtin): Check that the constant operands to
16581 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
16582 constant and in range.
16583 * config/sparc/sparc-c.c (sparc_target_macros): Handle
16585 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
16586 (SPARC_IMM5_P): Likewise.
16587 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
16588 (enabled): Handle vis4b.
16589 (UNSPEC_DICTUNPACK): New unspec.
16590 (UNSPEC_FPCMPSHL): Likewise.
16591 (UNSPEC_FPUCMPSHL): Likewise.
16592 (UNSPEC_FPCMPDESHL): Likewise.
16593 (UNSPEC_FPCMPURSHL): Likewise.
16594 (cpu_feature): New CPU feature `vis4b'.
16595 (dictunpack{8,16,32}): New insns.
16596 (FPCSMODE): New mode iterator.
16597 (fpcscond): New code iterator.
16598 (fpcsucond): Likewise.
16599 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
16600 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
16601 (fpcmpde{8,16,32}{si,di}shl): Likewise.
16602 (fpcmpur{8,16,32}{si,di}shl): Likewise.
16603 * config/sparc/constraints.md: Define constraints `q' for unsigned
16604 2-bit integer constants and `t' for unsigned 5-bit integer
16606 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
16608 (imm5_operand_dictunpack16): Likewise.
16609 (imm5_operand_dictunpack32): Likewise.
16610 (imm2_operand): Likewise.
16611 * doc/invoke.texi (SPARC Options): Document -mvis4b.
16612 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
16613 ditunpack* and fpcmp*shl builtins.
16615 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16617 * config.gcc: Handle m8 in --with-{cpu,tune} options.
16618 * config.in: Add HAVE_AS_SPARC6 define.
16619 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
16621 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
16623 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
16624 (CPP_CPU_SPEC): Handle m8.
16625 (ASM_CPU_SPEC): Likewise.
16626 * config/sparc/sparc-opts.h (enum processor_type): Add
16628 * config/sparc/sparc.c (m8_costs): New struct.
16629 (sparc_option_override): Handle TARGET_CPU_m8.
16630 (sparc32_initialize_trampoline): Likewise.
16631 (sparc64_initialize_trampoline): Likewise.
16632 (sparc_issue_rate): Likewise.
16633 (sparc_register_move_cost): Likewise.
16634 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
16635 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
16636 (ASM_CPU64_DEFAULT_SPEC): Likewise.
16637 (CPP_CPU_SPEC): Handle M8.
16638 (ASM_CPU_SPEC): Likewise.
16639 (AS_M8_FLAG): Define.
16640 * config/sparc/sparc.md: Add m8 to the cpu attribute.
16641 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
16642 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
16644 * configure: Regenerate.
16645 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
16648 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16650 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
16652 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
16653 ("*movdi_insn_sp32"): Do not set v3pipe.
16654 ("*movsi_insn"): Likewise.
16655 ("*movdi_insn_sp64"): Likewise.
16656 ("*movsf_insn"): Likewise.
16657 ("*movdf_insn_sp32"): Likewise.
16658 ("*movdf_insn_sp64"): Likewise.
16659 ("*zero_extendsidi2_insn_sp64"): Likewise.
16660 ("*sign_extendsidi2_insn"): Likewise.
16661 ("*mov<VM32:mode>_insn"): Likewise.
16662 ("*mov<VM64:mode>_insn_sp64"): Likewise.
16663 ("*mov<VM64:mode>_insn_sp32"): Likewise.
16664 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
16665 ("<vlop:code><VL:mode>3"): Likewise.
16666 ("*not_<vlop:code><VL:mode>3"): Likewise.
16667 ("*nand<VL:mode>_vis"): Likewise.
16668 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
16669 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
16670 ("one_cmpl<VL:mode>2"): Likewise.
16671 ("faligndata<VM64:mode>_vis"): Likewise.
16672 ("alignaddrsi_vis"): Likewise.
16673 ("alignaddrdi_vis"): Likweise.
16674 ("alignaddrlsi_vis"): Likewise.
16675 ("alignaddrldi_vis"): Likewise.
16676 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
16677 ("bmaskdi_vis"): Likewise.
16678 ("bmasksi_vis"): Likewise.
16679 ("bshuffle<VM64:mode>_vis"): Likewise.
16680 ("cmask8<P:mode>_vis"): Likewise.
16681 ("cmask16<P:mode>_vis"): Likewise.
16682 ("cmask32<P:mode>_vis"): Likewise.
16683 ("pdistn<P:mode>_vis"): Likewise.
16684 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
16686 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16688 * config/sparc/sparc.md ("subtype"): New insn attribute.
16689 ("*wrgsr_sp64"): Set insn subtype.
16690 ("*rdgsr_sp64"): Likewise.
16691 ("alignaddrsi_vis"): Likewise.
16692 ("alignaddrdi_vis"): Likewise.
16693 ("alignaddrlsi_vis"): Likewise.
16694 ("alignaddrldi_vis"): Likewise.
16695 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
16696 ("fexpand_vis"): Likewise.
16697 ("fpmerge_vis"): Likewise.
16698 ("faligndata<VM64:mode>_vis"): Likewise.
16699 ("bshuffle<VM64:mode>_vis"): Likewise.
16700 ("cmask8<P:mode>_vis"): Likewise.
16701 ("cmask16<P:mode>_vis"): Likewise.
16702 ("cmask32<P:mode>_vis"): Likewise.
16703 ("fchksm16_vis"): Likewise.
16704 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
16705 ("fmean16_vis"): Likewise.
16706 ("fp<plusminus_insn>64_vis"): Likewise.
16707 ("<plusminus_insn>v8qi3"): Likewise.
16708 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
16709 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
16710 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
16711 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
16712 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
16713 ("*movqi_insn"): Likewise.
16714 ("*movhi_insn"): Likewise.
16715 ("*movsi_insn"): Likewise.
16716 ("movsi_pic_gotdata_op"): Likewise.
16717 ("*movdi_insn_sp32"): Likewise.
16718 ("*movdi_insn_sp64"): Likewise.
16719 ("movdi_pic_gotdata_op"): Likewise.
16720 ("*movsf_insn"): Likewise.
16721 ("*movdf_insn_sp32"): Likewise.
16722 ("*movdf_insn_sp64"): Likewise.
16723 ("*zero_extendhisi2_insn"): Likewise.
16724 ("*zero_extendqihi2_insn"): Likewise.
16725 ("*zero_extendqisi2_insn"): Likewise.
16726 ("*zero_extendqidi2_insn"): Likewise.
16727 ("*zero_extendhidi2_insn"): Likewise.
16728 ("*zero_extendsidi2_insn_sp64"): Likewise.
16729 ("ldfsr"): Likewise.
16730 ("prefetch_64"): Likewise.
16731 ("prefetch_32"): Likewise.
16732 ("tie_ld32"): Likewise.
16733 ("tie_ld64"): Likewise.
16734 ("*tldo_ldub_sp32"): Likewise.
16735 ("*tldo_ldub1_sp32"): Likewise.
16736 ("*tldo_ldub2_sp32"): Likewise.
16737 ("*tldo_ldub_sp64"): Likewise.
16738 ("*tldo_ldub1_sp64"): Likewise.
16739 ("*tldo_ldub2_sp64"): Likewise.
16740 ("*tldo_ldub3_sp64"): Likewise.
16741 ("*tldo_lduh_sp32"): Likewise.
16742 ("*tldo_lduh1_sp32"): Likewise.
16743 ("*tldo_lduh_sp64"): Likewise.
16744 ("*tldo_lduh1_sp64"): Likewise.
16745 ("*tldo_lduh2_sp64"): Likewise.
16746 ("*tldo_lduw_sp32"): Likewise.
16747 ("*tldo_lduw_sp64"): Likewise.
16748 ("*tldo_lduw1_sp64"): Likewise.
16749 ("*tldo_ldx_sp64"): Likewise.
16750 ("*mov<VM32:mode>_insn"): Likewise.
16751 ("*mov<VM64:mode>_insn_sp64"): Likewise.
16752 ("*mov<VM64:mode>_insn_sp32"): Likewise.
16754 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16756 * config/sparc/sparc.md ("type"): New insn type viscmp.
16757 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
16759 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
16760 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
16761 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
16762 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
16764 ("n7_vis_logical_11cycle"): Likewise.
16765 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
16766 * config/sparc/niagara2.md ("niag3_vis": Likewise.
16767 * config/sparc/niagara.md ("niag_vis"): Likewise.
16768 * config/sparc/ultra3.md ("us3_fga"): Likewise.
16769 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
16771 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16773 * config/sparc/sparc.md: New instruction type `bmask'.
16774 (bmaskdi_vis): Use the `bmask' type.
16775 (bmasksi_vis): Likewise.
16776 * config/sparc/ultra3.md (us3_array): Likewise.
16777 * config/sparc/niagara7.md (n7_array): Likewise.
16778 * config/sparc/niagara4.md (n4_array): Likewise.
16779 * config/sparc/niagara2.md (niag2_vis): Likewise.
16780 (niag3_vis): Likewise.
16781 * config/sparc/niagara.md (niag_vis): Likewise.
16783 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16785 * ipa-comdats.c: Remove optimize check from gate.
16786 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
16787 for functions not optimized.
16788 (ipa_fn_summary_read): Skip optimize check.
16789 (ipa_fn_summary_write): Likewise.
16790 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
16792 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
16794 (can_inline_edge_p): Check flag_pcc_struct_return for match.
16795 (check_callers): Give up on caller which is not optimized.
16796 (inline_small_functions): Likewise.
16797 (ipa_inline): Do not give up when not optimizing.
16798 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
16799 away unoptimizes cdtors.
16800 (whole_program_function_and_variable_visibility): Do
16801 ipa_discover_readonly_nonaddressable_vars in LTO mode.
16802 * ipa.c (process_references): Do not check optimize.
16803 (symbol_table::remove_unreachable_nodes): Update optimize check.
16804 (set_writeonly_bit): Update optimize check.
16805 (pass_ipa_cdtor_merge::gate): Do not check optimize.
16806 (pass_ipa_single_use::gate): Remove.
16808 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16810 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
16811 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
16812 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
16813 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
16814 permute_load, permute_store, adjust_extract, adjust_splat,
16815 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
16816 replace_swap_with_copy, dump_swap_insn_table,
16817 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
16818 recombine_lvx_pattern, recombine_stvx_pattern,
16819 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
16820 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
16821 to file rs6000-p8swap.c.
16822 * config/rs6000/rs6000-p8swap.c: New file.
16823 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
16824 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
16825 and rs6000*-*-* targets.
16827 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16829 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
16831 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16833 * lto-wrapper.c (merge_and_complain): Do not merge
16834 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
16835 fsigned_zeros, ftrapping_math, fwrapv.
16836 (append_compiler_options): Do not track these options.
16837 (append_linker_options): Likewie
16839 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16841 * cgraphunit.c (cgraph_node::finalize_function): When
16842 !flag_toplevel_reorde set no_reorder flag.
16843 (varpool_node::finalize_decl): Likewise.
16844 (symbol_table::compile): Drop no toplevel reorder path.
16846 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16848 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
16849 edges; zero probability is not better than uninitialized.
16851 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
16853 * asan.h (asan_sanitize_allocas_p): Declare.
16854 * asan.c (asan_sanitize_allocas_p): New function.
16855 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
16856 (handle_builtin_alloca): Likewise.
16857 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
16858 if !asan_sanitize_allocas_p.
16859 * params.def (asan-instrument-allocas): Add new option.
16860 * params.h (ASAN_PROTECT_ALLOCAS): Define.
16861 * opts.c (common_handle_option): Disable allocas sanitization for
16864 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
16866 * asan.c: Include gimple-fold.h.
16867 (get_last_alloca_addr): New function.
16868 (handle_builtin_stackrestore): Likewise.
16869 (handle_builtin_alloca): Likewise.
16870 (asan_emit_allocas_unpoison): Likewise.
16871 (get_mem_refs_of_builtin_call): Add new parameter, remove const
16872 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
16873 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
16874 (instrument_builtin_call): Pass gimple iterator to
16875 get_mem_refs_of_builtin_call.
16876 (last_alloca_addr): New global.
16877 * asan.h (asan_emit_allocas_unpoison): Declare.
16878 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
16879 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
16880 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
16881 if function calls alloca.
16882 * gimple-fold.c (replace_call_with_value): Remove static keyword.
16883 * gimple-fold.h (replace_call_with_value): Declare.
16884 * internal-fn.c: Include asan.h.
16885 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
16886 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
16888 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16890 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
16891 (C_SELFTEST_FLAGS): New.
16892 (CPP_SELFTEST_FLAGS): New.
16893 (SELFTEST_DEPS): New, from deps of s-selftest.
16894 (C_SELFTEST_DEPS): New, from deps of s-selftest.
16895 (CPP_SELFTEST_DEPS): New.
16896 (selftest): Add dependency on s-selftest-c++.
16897 (s-selftest): Rename to...
16898 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
16899 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
16900 than SELFTEST_FLAGS.
16901 (selftest-gdb): Rename to...
16902 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
16904 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
16905 (selftest-valgrind): Rename to...
16906 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
16908 (selftest-valgrind): Reintroduce as an alias for
16909 selftest-c-valgrind.
16910 (s-selftest-c++): New.
16911 (selftest-c++-gdb): New.
16912 (selftest-c++-valgrind): New.
16914 2017-07-06 Olivier Hainque <hainque@adacore.com>
16916 * gcc.c (process_command): When deciding if undefined variables
16917 should be ignored when processing specs, accept "gcc -v" as well.
16919 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16921 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
16922 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
16924 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
16926 * config/arm/arm-cpus.in (armv8-r): Add new entry.
16927 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
16928 * config/arm/arm-tables.opt: Regenerate.
16929 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
16931 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
16933 2017-07-06 Carl Love <cel@us.ibm.com>
16935 * ChangeLog: Clean up from mid air collision
16937 2017-07-06 Carl Love <cel@us.ibm.com>
16939 * config/rs6000/rs6000-c.c: Add support for built-in functions
16940 vector signed int vec_subc (vector signed int, vector signed int);
16941 vector signed __int128 vec_subc (vector signed __int128,
16942 vector signed __int128);
16943 vector unsigned __int128 vec_subc (vector unsigned __int128,
16944 vector unsigned __int128);
16945 vector signed int vec_sube (vector signed int, vector signed int,
16946 vector signed int);
16947 vector unsigned int vec_sube (vector unsigned int,
16948 vector unsigned int,
16949 vector unsigned int);
16950 vector signed __int128 vec_sube (vector signed __int128,
16951 vector signed __int128,
16952 vector signed__int128);
16953 vector unsigned __int128 vec_sube (vector unsigned __int128,
16954 vector unsigned __int128,
16955 vector unsigned __int128);
16956 vector signed int vec_subec (vector signed int, vector signed int,
16957 vector signed int);
16958 vector unsigned int vec_subec (vector unsigned int,
16959 vector unsigned int,
16960 vector unsigned int);
16961 vector signed __int128 vec_subec (vector signed __int128,
16962 vector signed __int128,
16963 vector signed__int128);
16964 vector unsigned __int128 vec_subec (vector unsigned __int128,
16965 vector unsigned __int128,
16966 vector unsigned __int128);
16967 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
16968 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
16969 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
16970 BU_ALTIVEC_OVERLOAD_X definitions.
16971 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
16972 * doc/extend.texi: Update the built-in documentation file for the new
16973 built-in functions.
16975 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16978 * diagnostic-show-locus.c (layout::layout): Use start and finish
16979 spelling location for the start and finish of each range.
16980 * genmatch.c (linemap_client_expand_location_to_spelling_point):
16981 Add unused aspect param.
16982 * input.c (expand_location_1): Add "aspect" param, and use it
16983 to access the correct part of the location.
16984 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
16986 (expand_location_to_spelling_point): Likewise.
16987 (linemap_client_expand_location_to_spelling_point): Add "aspect"
16988 param, and pass it to expand_location_1.
16990 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
16992 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
16993 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
16994 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
16995 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
16996 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
16997 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
16998 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
16999 _mm_maskz_getmant_ss): New intrinsics.
17000 (__builtin_ia32_getexpss128_mask): Changed to ...
17001 __builtin_ia32_getexpss128_round ... this.
17002 (__builtin_ia32_getexpsd128_mask): Changed to ...
17003 __builtin_ia32_getexpsd128_round ... this.
17004 * config/i386/i386-builtin-types.def
17005 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
17006 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
17007 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
17008 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
17009 __builtin_ia32_getmantss_mask_round): New builtins.
17010 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
17011 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
17012 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
17013 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
17014 * config/i386/sse.md
17015 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
17016 avx512f_sgetexp<mode><mask_scalar_name>
17017 <round_saeonly_scalar_name> ... this.
17018 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
17019 %0, %1, %2<round_saeonly_op3>}): Changed to ...
17020 vgetexp<ssescalarmodesuffix>
17021 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17022 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
17023 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
17024 avx512f_vgetmant<mode><mask_scalar_name>
17025 <round_saeonly_scalar_name> ... this.
17026 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
17027 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
17028 vgetmant<ssescalarmodesuffix>
17029 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
17030 %0<mask_scalar_operand4>, %1, %2
17031 <round_saeonly_scalar_mask_op4>, %3} ... this.
17032 * config/i386/subst.md (mask_scalar_operand4,
17033 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
17034 round_saeonly_scalar_nimm_predicate): New subst attributes.
17036 2017-07-06 Julia Koval <julia.koval@intel.com>
17038 * config/i386/i386.c (ix86_erase_embedded_rounding):
17039 Remove code for old rounding pattern.
17041 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
17043 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
17045 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
17047 * doc/sourcebuild.texi (Test Directives, Variants of
17048 dg-require-support): Add documentation for dg-require-stack-check.
17050 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
17052 * config/i386/subst.md (mask_scalar, round_scalar,
17053 round_saeonly_scalar): New meta-templates.
17054 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
17055 round_scalar_mask_operand3, round_scalar_mask_op3,
17056 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
17057 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
17058 round_saeonly_scalar_constraint,
17059 round_saeonly_scalar_prefix): New subst attribute.
17060 * config/i386/sse.md
17061 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
17062 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
17063 <round_scalar_name> ... this.
17064 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
17065 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
17066 <round_scalar_name> ... this.
17067 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
17068 <sse>_vm<code><mode>3<mask_scalar_name>
17069 <round_saeonly_scalar_name> ... this.
17070 (v<plusminus_mnemonic><ssescalarmodesuffix>
17071 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
17072 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
17073 v<plusminus_mnemonic><ssescalarmodesuffix>
17074 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17075 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
17076 (v<multdiv_mnemonic><ssescalarmodesuffix>
17077 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
17078 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
17079 v<multdiv_mnemonic><ssescalarmodesuffix>
17080 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17081 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
17082 (v<maxmin_float><ssescalarmodesuffix>
17083 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
17084 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
17085 v<maxmin_float><ssescalarmodesuffix>
17086 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17087 %0<mask_scalar_operand3>, %1, %<iptr>2
17088 <round_saeonly_scalar_mask_op3>} ... this.
17090 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
17092 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
17093 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
17095 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
17096 Alan Hayward <alan.hayward@arm.com>
17097 David Sherwood <david.sherwood@arm.com>
17099 * combine.c (simplify_if_then_else): Remove "enum" before
17101 * compare-elim.c (can_eliminate_compare): Likewise.
17102 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
17104 (aarch64_lookup_simd_builtin_type): Likewise.
17105 (aarch64_simd_builtin_type): Likewise.
17106 (aarch64_init_simd_builtin_types): Likewise.
17107 (aarch64_simd_expand_args): Likewise.
17108 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
17110 (aarch64_reverse_mask): Likewise.
17111 (aarch64_simd_emit_reg_reg_move): Likewise.
17112 (aarch64_gen_adjusted_ldpstp): Likewise.
17113 (aarch64_ccmp_mode_to_code): Likewise.
17114 (aarch64_operands_ok_for_ldpstp): Likewise.
17115 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17116 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
17118 (aarch64_min_divisions_for_recip_mul): Likewise.
17119 (aarch64_reassociation_width): Likewise.
17120 (aarch64_get_condition_code_1): Likewise.
17121 (aarch64_simd_emit_reg_reg_move): Likewise.
17122 (aarch64_simd_attr_length_rglist): Likewise.
17123 (aarch64_reverse_mask): Likewise.
17124 (aarch64_operands_ok_for_ldpstp): Likewise.
17125 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17126 (aarch64_gen_adjusted_ldpstp): Likewise.
17127 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
17129 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
17130 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
17131 (arm_lookup_simd_builtin_type): Likewise.
17132 (arm_simd_builtin_type): Likewise.
17133 (arm_init_simd_builtin_types): Likewise.
17134 (arm_expand_builtin_args): Likewise.
17135 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
17136 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
17137 (ft32_setup_incoming_varargs): Likewise.
17138 (ft32_function_arg): Likewise.
17139 (ft32_function_arg_advance): Likewise.
17140 (ft32_pass_by_reference): Likewise.
17141 (ft32_arg_partial_bytes): Likewise.
17142 (ft32_valid_pointer_mode): Likewise.
17143 (ft32_addr_space_pointer_mode): Likewise.
17144 (ft32_addr_space_legitimate_address_p): Likewise.
17145 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
17147 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
17148 (ix86_emit_outlined_ms2sysv_restore): Likewise.
17149 (iamcu_alignment): Likewise.
17150 (canonicalize_vector_int_perm): Likewise.
17151 (ix86_noce_conversion_profitable_p): Likewise.
17152 (ix86_mpx_bound_mode): Likewise.
17153 (ix86_operands_ok_for_move_multiple): Likewise.
17154 * config/microblaze/microblaze-protos.h
17155 (microblaze_expand_conditional_branch_reg): Likewise.
17156 * config/microblaze/microblaze.c
17157 (microblaze_expand_conditional_branch_reg): Likewise.
17158 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
17160 (rs6000_reassociation_width): Likewise.
17161 (rs6000_invalid_binary_op): Likewise.
17162 (fusion_p9_p): Likewise.
17163 (emit_fusion_p9_load): Likewise.
17164 (emit_fusion_p9_store): Likewise.
17165 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
17167 (riscv_hard_regno_mode_ok_p): Likewise.
17168 (riscv_address_insns): Likewise.
17169 (riscv_split_symbol): Likewise.
17170 (riscv_legitimize_move): Likewise.
17171 (riscv_function_value): Likewise.
17172 (riscv_hard_regno_nregs): Likewise.
17173 (riscv_expand_builtin): Likewise.
17174 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
17175 (riscv_build_integer): Likewise.
17176 (riscv_split_integer): Likewise.
17177 (riscv_legitimate_constant_p): Likewise.
17178 (riscv_cannot_force_const_mem): Likewise.
17179 (riscv_regno_mode_ok_for_base_p): Likewise.
17180 (riscv_valid_base_register_p): Likewise.
17181 (riscv_valid_offset_p): Likewise.
17182 (riscv_valid_lo_sum_p): Likewise.
17183 (riscv_classify_address): Likewise.
17184 (riscv_legitimate_address_p): Likewise.
17185 (riscv_address_insns): Likewise.
17186 (riscv_load_store_insns): Likewise.
17187 (riscv_force_binary): Likewise.
17188 (riscv_split_symbol): Likewise.
17189 (riscv_force_address): Likewise.
17190 (riscv_legitimize_address): Likewise.
17191 (riscv_move_integer): Likewise.
17192 (riscv_legitimize_const_move): Likewise.
17193 (riscv_legitimize_move): Likewise.
17194 (riscv_address_cost): Likewise.
17195 (riscv_subword): Likewise.
17196 (riscv_output_move): Likewise.
17197 (riscv_canonicalize_int_order_test): Likewise.
17198 (riscv_emit_int_order_test): Likewise.
17199 (riscv_function_arg_boundary): Likewise.
17200 (riscv_pass_mode_in_fpr_p): Likewise.
17201 (riscv_pass_fpr_single): Likewise.
17202 (riscv_pass_fpr_pair): Likewise.
17203 (riscv_get_arg_info): Likewise.
17204 (riscv_function_arg): Likewise.
17205 (riscv_function_arg_advance): Likewise.
17206 (riscv_arg_partial_bytes): Likewise.
17207 (riscv_function_value): Likewise.
17208 (riscv_pass_by_reference): Likewise.
17209 (riscv_setup_incoming_varargs): Likewise.
17210 (riscv_print_operand): Likewise.
17211 (riscv_elf_select_rtx_section): Likewise.
17212 (riscv_save_restore_reg): Likewise.
17213 (riscv_for_each_saved_reg): Likewise.
17214 (riscv_register_move_cost): Likewise.
17215 (riscv_hard_regno_mode_ok_p): Likewise.
17216 (riscv_hard_regno_nregs): Likewise.
17217 (riscv_class_max_nregs): Likewise.
17218 (riscv_memory_move_cost): Likewise.
17219 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
17220 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
17221 (rl78_addr_space_address_mode): Likewise.
17222 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17224 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17225 (rs6000_reassociation_width): Likewise.
17226 (rs6000_invalid_binary_op): Likewise.
17227 (fusion_p9_p): Likewise.
17228 (emit_fusion_p9_load): Likewise.
17229 (emit_fusion_p9_store): Likewise.
17230 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
17231 (ok_for_simple_move_operands): Likewise.
17232 (ok_for_simple_move_strict_operands): Likewise.
17233 (ok_for_simple_arith_logic_operands): Likewise.
17234 (visium_legitimize_reload_address): Likewise.
17235 (visium_select_cc_mode): Likewise.
17236 (output_cbranch): Likewise.
17237 (visium_split_double_move): Likewise.
17238 (visium_expand_copysign): Likewise.
17239 (visium_expand_int_cstore): Likewise.
17240 (visium_expand_fp_cstore): Likewise.
17241 * config/visium/visium.c (visium_pass_by_reference): Likewise.
17242 (visium_function_arg): Likewise.
17243 (visium_function_arg_advance): Likewise.
17244 (visium_libcall_value): Likewise.
17245 (visium_setup_incoming_varargs): Likewise.
17246 (visium_legitimate_constant_p): Likewise.
17247 (visium_legitimate_address_p): Likewise.
17248 (visium_legitimize_address): Likewise.
17249 (visium_secondary_reload): Likewise.
17250 (visium_register_move_cost): Likewise.
17251 (visium_memory_move_cost): Likewise.
17252 (prepare_move_operands): Likewise.
17253 (ok_for_simple_move_operands): Likewise.
17254 (ok_for_simple_move_strict_operands): Likewise.
17255 (ok_for_simple_arith_logic_operands): Likewise.
17256 (visium_function_value_1): Likewise.
17257 (rtx_ok_for_offset_p): Likewise.
17258 (visium_legitimize_reload_address): Likewise.
17259 (visium_split_double_move): Likewise.
17260 (visium_expand_copysign): Likewise.
17261 (visium_expand_int_cstore): Likewise.
17262 (visium_expand_fp_cstore): Likewise.
17263 (visium_split_cstore): Likewise.
17264 (visium_select_cc_mode): Likewise.
17265 (visium_split_cbranch): Likewise.
17266 (output_cbranch): Likewise.
17267 (visium_print_operand_address): Likewise.
17268 * expmed.c (flip_storage_order): Likewise.
17269 * expmed.h (emit_cstore): Likewise.
17270 (flip_storage_order): Likewise.
17271 * genrecog.c (validate_pattern): Likewise.
17272 * hsa-gen.c (gen_hsa_addr): Likewise.
17273 * internal-fn.c (expand_arith_overflow): Likewise.
17274 * ira-color.c (allocno_copy_cost_saving): Likewise.
17275 * lra-assigns.c (find_hard_regno_for_1): Likewise.
17276 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
17277 (process_invariant_for_inheritance): Likewise.
17278 * lra-eliminations.c (move_plus_up): Likewise.
17279 * omp-low.c (lower_oacc_reductions): Likewise.
17280 * simplify-rtx.c (simplify_subreg): Likewise.
17281 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
17282 (TARGET_CHKP_BOUND_MODE): Likewise..
17283 * targhooks.c (default_chkp_bound_mode): Likewise.
17284 (default_setup_incoming_vararg_bounds): Likewise.
17285 * targhooks.h (default_chkp_bound_mode): Likewise.
17286 (default_setup_incoming_vararg_bounds): Likewise.
17287 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
17288 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
17289 (have_whole_vector_shift): Likewise.
17290 * tree-vect-stmts.c (vectorizable_load): Likewise.
17291 * doc/tm.texi: Regenerate.
17293 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
17295 Graceful degrade if Binutils PR21472 is not available.
17298 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
17299 .rodata in flash test fails.
17300 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
17301 * confgure: Regenerate.
17302 * config.in: Regenerate.
17303 * config/avr/avr.c (avr_asm_named_section)
17304 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
17305 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
17306 (avr_asm_init_sections): Same.
17308 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17310 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
17311 (fma<VH:mode>4_intrinsic): Likewise.
17312 (*fmsub<VCVTF:mode>4): Likewise.
17313 (*fmsub<VH:mode>4_intrinsic): Likewise.
17315 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
17318 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
17319 Don't depend on "optimize > 0".
17320 (out_movhi_r_mr, out_movqi_mr_r): Same.
17321 (out_movhi_mr_r, out_movqi_r_mr): Same.
17322 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
17323 io_address_operand on "optimize > 0".
17325 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17327 * tree-loop-distribution.c: Add general explanantion on the pass.
17328 (generate_loops_for_partition): Mark distributed loop.
17329 (pg_add_dependence_edges): New parameter. Handle alias data
17330 dependence specially and record it in the parameter if asked.
17331 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
17332 (init_partition_graph_vertices, add_partition_graph_edge): New.
17333 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
17334 (free_partition_graph_vdata, build_partition_graph): New.
17335 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
17336 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
17337 (data_ref_segment_size, latch_dominated_by_data_ref): New.
17338 (compute_alias_check_pairs, version_loop_by_alias_check): New.
17339 (version_for_distribution_p, finalize_partitions): New.
17340 (distribute_loop): Handle alias data dependence specially. Factor
17341 out loop fusion code as functions and call these functions.
17343 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17345 * tree-loop-distribution.c (classify_partition): New parameter and
17346 better handle reduction statement.
17347 (rdg_build_partitions): Revise comment.
17348 (distribute_loop): Compute statements in all partitions and pass it
17349 to classify_partition.
17351 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17353 * tree-loop-distribution.c (enum partition_type): New.
17354 (struct partition): New field type.
17355 (partition_merge_into): Add parameter. Update partition type.
17356 (data_dep_in_cycle_p, update_type_for_merge): New functions.
17357 (build_rdg_partition_for_vertex): Compute partition type.
17358 (rdg_build_partitions): Dump partition type.
17359 (distribute_loop): Update calls to partition_merge_into.
17361 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17363 * tree-loop-distribution.c (struct ddr_hasher): New.
17364 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
17366 (classify_partition): Call get_data_dependence.
17367 (pg_add_dependence_edges): Ditto.
17368 (distribute_loop): Release data dependence hash table.
17370 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17372 * tree-loop-distribution.c (ref_base_address): Delete.
17373 (similar_memory_accesses): Rename ...
17374 (share_memory_accesses): ... to this. Check if partitions access
17375 the same memory reference.
17376 (distribute_loop): Call share_memory_accesses.
17378 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17380 * tree-loop-distribution.c (struct partition): New field recording
17381 its data reference.
17382 (partition_alloc, partition_free): Init and release data refs.
17383 (partition_merge_into): Merge data refs.
17384 (build_rdg_partition_for_vertex): Collect data refs for partition.
17385 (pg_add_dependence_edges): Change parameters from vector to bitmap.
17387 (distribute_loop): Remve data refs from vertice data of partition
17390 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17392 * tree-loop-distribution.c (params.h): Include header file.
17393 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
17394 (datarefs_vec): New global var.
17395 (create_rdg_vertices): Use datarefs_vec directly.
17396 (free_rdg): Don't free data references.
17397 (build_rdg): Update use. Don't free data references.
17398 (distribute_loop): Compute global variable for data references.
17399 Bail out if there are too many data references.
17401 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17403 * tree-loop-distribution.c (loop_nest): New global var.
17404 (build_rdg): Use loop directly, rather than loop nest.
17405 (pg_add_dependence_edges): Remove loop nest parameter. Use global
17407 (distribute_loop): Compute global variable loop nest. Update use.
17409 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17411 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
17412 (partition_merge_into): New parameter. Dump reason for fusion.
17413 (distribute_loop): Update use of partition_merge_into.
17415 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17417 * tree-loop-distribution.c (bb_top_order_index): New.
17418 (bb_top_order_index_size, bb_top_order_cmp): New.
17419 (stmts_from_loop): Use topological order.
17420 (pass_loop_distribution::execute): Compute and release topological
17421 order for basic blocks.
17423 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17425 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
17428 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17430 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
17431 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
17432 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
17433 * internal-fn.def (LOOP_DIST_ALIAS): New.
17434 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
17435 (fold_loop_internal_call): ... this.
17436 (vect_loop_dist_alias_call): New function.
17437 (set_uid_loop_bbs): Call fold_loop_internal_call.
17438 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
17441 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
17444 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
17445 Require dead FLAGS_REG at the beginning of a peephole.
17447 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
17450 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
17451 arguments in the call to __builtin_ia32_sbb_u32.
17452 (_subborrow_u64): Swap _X and _Y arguments in the call to
17453 __builtin_ia32_sbb_u64.
17455 2017-07-04 Jakub Jelinek <jakub@redhat.com>
17458 * tree-vrp.c (compare_assert_loc): Turn into a function template
17459 with stable template parameter. Only test if a->e is NULL,
17460 !a->e == !b->e has been verified already. Use e == NULL or
17461 e != NULL instead of e or ! e tests. If stable is true, don't use
17462 iterative_hash_expr, on the other side allow a or b or both NULL
17463 and sort the NULLs last.
17464 (process_assert_insertions): Sort using compare_assert_loc<false>
17465 instead of compare_assert_loc, later sort using
17466 compare_assert_loc<true> before calling process_assert_insertions_for
17467 in a loop. Use break instead of continue once seen NULL pointer.
17469 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
17471 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
17472 Cortex-R7 and Cortex-R8 processors.
17474 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17476 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
17477 uninitialized while src is not.
17479 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
17481 * common/config/arm/arm-common.c: Adjust include path for
17483 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
17484 (arm-cpu.h): Create in build directory. Adjust dependency rules.
17485 (arm-cpu-data.h): Likewise.
17486 (arm-cpu-cdata.h): Likewise.
17487 * config/arm/arm-cpu.h: Delete.
17488 * config/arm/arm-cpu-cdata.h: Delete.
17489 * config/arm/arm-cpu-data.h: Delete.
17491 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
17493 * config/arm/arm-cpus.in (cortex-a55): New.
17494 (cortex-a75): Likewise.
17495 (cortex-a75.cortex-a55): Likewise.
17496 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
17498 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
17499 * config/arm/arm-cpu-cdata.h: Regenerate.
17500 * config/arm/arm-cpu-data.h: Regenerate.
17501 * config/arm/arm-cpu.h: Regenerate.
17502 * config/arm/arm-tables.opt: Regenerate.
17503 * config/arm/arm-tune.md: Regenerate.
17505 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17507 * haifa-sched.c (sched_create_recovery_edges): Update profile.
17509 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17511 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
17514 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
17516 PR tree-optimization/81292
17517 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
17518 full_string_p, also call adjust_related_strinfos if the adjustment
17519 is simple, otherwise invalidate related strinfos.
17521 2017-07-04 Martin Liska <mliska@suse.cz>
17524 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
17525 newly created variable as DECL_IGNORED_P.
17527 2017-07-04 Martin Liska <mliska@suse.cz>
17530 * ipa-inline.c (inline_small_functions):
17531 Use xstrdup_for_dump.
17533 2017-07-04 Tom de Vries <tom@codesourcery.com>
17535 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
17537 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
17540 * config/darwin.c (darwin_function_switched_text_sections):
17543 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
17545 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
17547 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
17549 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
17551 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17553 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
17554 min_profitable_iters, and th as inclusive lower bounds.
17555 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
17556 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
17557 for min_profitable_iters and min_profitable_estimate.
17558 (vect_transform_loop): Treat th as an inclusive lower bound.
17559 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
17561 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
17564 * config/darwin.c (darwin_function_switched_text_sections):
17565 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
17566 in two pieces, and suppress the use of buf.
17568 2017-07-03 Nathan Sidwell <nathan@acm.org>
17570 * hash-table.h (hash_table_mod1): Fix indentation.
17572 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17574 PR middle-end/81290
17575 * predict.c (force_edge_cold): Be more careful about propagation
17577 * profile-count.h (profile_probability::guessed,
17578 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
17580 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
17582 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
17584 * doc/invoke.texi (rcpc architecture extension): Document it.
17586 2017-07-03 Richard Biener <rguenther@suse.de>
17588 PR tree-optimization/60510
17589 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
17590 the scalar reduction PHI and use it.
17591 (vectorizable_reduction): Properly guard the single_defuse_cycle
17592 path for non-SLP reduction chains where we cannot use it.
17593 Rework reduc_def/index and vector type deduction. Rework
17594 vector operand gathering during reduction op code-gen.
17595 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
17596 chains dissolve the chain and leave it to non-SLP reduction
17599 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17601 * tree-data-ref.h (dr_alignment): Declare.
17602 * tree-data-ref.c (dr_alignment): New function.
17603 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
17604 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
17606 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
17608 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17610 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
17611 and base_misalignment fields.
17612 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
17613 * tree-data-ref.c: Include builtins.h.
17614 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
17615 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
17616 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
17617 * tree-vect-data-refs.c: Include tree-cfg.h.
17618 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
17619 fields instead of calculating an alignment here.
17620 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
17621 innermost_loop_behavior fields.
17623 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17625 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
17627 (DR_STEP_ALIGNMENT): New macro.
17628 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
17629 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
17630 (create_data_ref): Print it.
17631 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
17632 to tell whether the step preserves vector (mis)alignment.
17633 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
17634 Move the check for an integer step and generalise to all INTEGER_CST.
17635 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
17636 Print the outer step alignment.
17638 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17640 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
17641 with offset_alignment.
17642 (DR_ALIGNED_TO): Delete.
17643 (DR_OFFSET_ALIGNMENT): New macro.
17644 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
17645 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
17646 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
17647 (create_data_ref): Likewise.
17648 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
17649 (vect_analyze_data_refs): Likewise.
17650 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
17651 creating dummy innermost behavior.
17653 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17655 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
17656 with a "innermost_loop_behavior *" and refeence tree.
17657 * tree-data-ref.c (dr_analyze_innermost): Likewise.
17658 (create_data_ref): Update call accordingly.
17659 * tree-predcom.c (find_looparound_phi): Likewise.
17661 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17663 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
17664 fields with dr_wrt_vec_loop.
17665 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
17666 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
17667 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
17668 (vect_dr_behavior): New function.
17669 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
17670 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
17671 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
17672 track whether the step preserves the misalignment.
17673 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
17674 Use vect_dr_behavior.
17675 (vect_setup_realignment): Update call accordingly.
17676 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
17677 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
17678 call to vect_create_addr_base_for_vector_ref.
17679 (vect_create_cond_for_align_checks): Likewise.
17680 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
17681 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
17682 (vect_recog_mask_conversion_pattern): Likewise.
17683 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
17684 (new_stmt_vec_info): Remove redundant zeroing.
17686 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
17688 * common/config/arm/arm-common.c (arm_be8_option): New function.
17689 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
17690 (ISA_ARMv6): Add isa_bit_be8.
17691 * config/arm/arm.h (arm_be8_option): Add prototype.
17692 (BE8_SPEC_FUNCTION): New define.
17693 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
17694 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
17695 (mlittle-endian): Similarly.
17696 (mbe8, mbe32): New options.
17697 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
17698 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
17700 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17702 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
17704 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17706 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
17707 (cleanup_tree_cfg_bb): Use it.
17708 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
17710 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
17712 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17715 * loop-doloop.c (add_test): Update profile.
17717 2017-07-03 Martin Liska <mliska@suse.cz>
17720 * sanopt.c (rewrite_usage_of_param): New function.
17721 (sanitize_rewrite_addressable_params): Likewise.
17722 (pass_sanopt::execute): Call rewrite_usage_of_param.
17724 2017-07-03 Richard Biener <rguenther@suse.de>
17726 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
17727 back to using VIEW_CONVERT_EXPR.
17729 2017-07-03 Martin Liska <mliska@suse.cz>
17732 * doc/extend.texi: Document when a resolver function is
17733 generated for target_clones.
17735 2017-07-03 Martin Liska <mliska@suse.cz>
17737 * asan.c (asan_emit_stack_protection): Unpoison just red zones
17738 and shadow memory of auto variables which are subject of
17739 use-after-scope sanitization.
17740 (asan_expand_mark_ifn): Add do set only when is_poison.
17742 2016-07-03 Richard Biener <rguenther@suse.de>
17744 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
17746 (vect_force_simple_reduction): Record reduction def -> phi mapping.
17747 (vectorizable_reduction): Perform reduction PHI creation when
17748 visiting a reduction PHI and adjust and simplify code generation
17749 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
17750 (vect_transform_loop): Visit reduction PHIs.
17751 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
17752 defs into the SLP tree.
17753 (vect_build_slp_tree): Reduction defs terminate the recursion.
17754 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
17756 (vect_get_vec_defs_for_stmt_copy): Export.
17757 (vect_get_vec_defs): Likewise.
17758 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
17760 (vect_get_vec_defs_for_stmt_copy): Declare.
17761 (vect_get_vec_defs): Likewise.
17763 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17765 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
17766 parameter with a "loop" parameter and use it instead of the
17767 loop containing DR_STMT. Don't check simple_iv when doing
17768 BB analysis. Describe the two analysis modes in the comment.
17770 2017-07-03 Tom de Vries <tom@codesourcery.com>
17772 PR tree-optimization/69468
17773 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
17774 (find_same_succ_bb): Handle ignore_edge_flags.
17776 2017-07-03 Tom de Vries <tom@codesourcery.com>
17778 PR tree-optimization/81192
17779 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
17781 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
17783 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
17785 2017-07-03 Tom de Vries <tom@codesourcery.com>
17787 PR tree-optimization/81192
17788 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
17789 BB_SAME_SUCC (bb) == NULL.
17791 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17793 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
17796 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17798 * dumpfile.c: Include profile-count.h
17799 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
17801 (insert_cond_bb): Update profile.
17802 * tree-cfg.h (insert_cond_bb): Update prototype.
17803 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
17804 * tree-dump.c: Do not include tree-cfg.
17806 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17808 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
17810 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17812 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
17815 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17817 * tree-complex.c (expand_complex_div_wide): update profile.
17819 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17820 Alan Hayward <alan.hayward@arm.com>
17821 David Sherwood <david.sherwood@arm.com>
17823 * Makefile.in (MACHMODE_H): Remove insn-modes.h
17824 (CORETYPES_H): New define.
17825 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
17826 (insn-modes-inline.h, s-modes-inline-h): New rules.
17827 (generated_files): Add insn-modes-inline.h.
17828 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
17829 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
17830 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
17831 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
17832 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
17833 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
17834 (build/gencodes.o, build/genconditions.o): Likewise.
17835 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
17836 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
17837 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
17838 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
17839 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
17840 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
17841 * coretypes.h: Include everything up to real.h for generators.
17842 Include insn-modes.h first. Include wide-int-print.h after
17843 wide-int.h. Include insn-modes-inline.h and then machmode.h.
17844 * machmode.h: Don't include insn-modes.h here.
17845 * function-tests.c: Remove includes of signop.h, machmode.h,
17846 double-int.h and wide-int.h.
17848 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
17850 * optc-save-gen.awk: Likewise.
17851 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
17852 * godump.c: Remove include of wide-int-print.h.
17853 * pretty-print.h: Likewise.
17854 * wide-int-print.cc: Likewise.
17855 * wide-int.cc: Likewise.
17856 * hash-map-tests.c: Remove include of signop.h.
17857 * hash-set-tests.c: Likewise.
17858 * rtl-tests.c: Likewise.
17859 * mkconfig.sh: Remove include of machmode.h.
17860 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
17862 (emit_insn_modes_inline_h): ...this new function. Emit the code
17863 into an insn-modes-inline.h header file, adding appropriate
17864 include guards and end comments.
17865 (emit_insn_modes_c_header): Remove include of machmode.h.
17866 (emit_min_insn_modes_c_header): Include coretypes.h rather than
17868 (main): Handle -i flag and call emit_insn_modes_inline_h when
17871 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17873 * tree-ssa-strlen.c (strinfo): Rename the length field to
17874 nonzero_chars. Add a full_string_p field.
17875 (compare_nonzero_chars, zero_length_string_p): New functions.
17876 (get_addr_stridx): Add an offset_out parameter.
17877 Use compare_nonzero_chars.
17878 (get_stridx): Update accordingly. Use compare_nonzero_chars.
17879 (new_strinfo): Update after above changes to strinfo.
17880 (set_endptr_and_length): Set full_string_p.
17881 (get_string_length): Update after above changes to strinfo.
17882 (unshare_strinfo): Update call to new_strinfo.
17883 (maybe_invalidate): Likewise.
17884 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
17885 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
17886 as a uhwi instead of an shwi. Update after above changes to
17887 strinfo and new_strinfo.
17888 (zero_length_string): Assert that chainsi contains full strings.
17889 Use zero_length_string_p. Update call to new_strinfo.
17890 (adjust_related_strinfos): Update after above changes to strinfo.
17891 Copy full_string_p from origsi.
17892 (adjust_last_stmt): Use zero_length_string_p.
17893 (handle_builtin_strlen): Update after above changes to strinfo and
17894 new_strinfo. Install the lhs as the string length if the previous
17895 entry didn't describe a full string.
17896 (handle_builtin_strchr): Update after above changes to strinfo
17898 (handle_builtin_strcpy): Likewise.
17899 (handle_builtin_strcat): Likewise.
17900 (handle_builtin_malloc): Likewise.
17901 (handle_pointer_plus): Likewise.
17902 (handle_builtin_memcpy): Likewise. Track nonzero characters
17903 that aren't necessarily followed by a nul terminator.
17904 (handle_char_store): Likewise.
17906 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17908 PR tree-optimization/80769
17909 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
17910 for malloc and calloc. Document the new invariant that all related
17911 strinfos have delayed lengths or none do.
17912 (verify_related_strinfos): Move earlier in file.
17913 (set_endptr_and_length): New function, split out from...
17914 (get_string_length): ...here. Also set the lengths of related
17916 (zero_length_string): Assert that chainsi has known (rather than
17918 (adjust_related_strinfos): Likewise.
17920 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17922 PR tree-optimization/81136
17923 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
17924 assert that two references with the same misalignment have the same
17925 compile-time misalignment if those compile-time misalignments
17928 2017-07-01 Andi Kleen <ak@linux.intel.com>
17930 * print-tree.c (print_node): Print all attributes.
17932 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17934 * cfg.c (scale_bbs_frequencies): New function.
17935 * cfg.h (scale_bbs_frequencies): Declare it.
17936 * cfgloopanal.c (single_likely_exit): Cleanup.
17937 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
17939 (scale_loop_profile): Likewise.
17940 (loop_version): Likewise.
17941 (create_empty_loop_on_edge): Update.
17942 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
17943 scale_loop_frequencies, scale_loop_profile, loopify,
17944 loop_version): Update prototypes.
17945 * modulo-sched.c (sms_schedule): Update.
17946 * predict.c (unlikely_executed_edge_p): Also check probability.
17947 (probably_never_executed_edge_p): Fix typo.
17948 * tree-if-conv.c (version_loop_for_if_conversion): Update.
17949 * tree-parloops.c (gen_parallel_loop): Update.
17950 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
17951 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
17952 * tree-ssa-loop-split.c (split_loop): Update.
17953 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
17954 * tree-vect-loop-manip.c (vect_do_peeling): Update.
17955 (vect_loop_versioning): Update.
17956 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
17958 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17960 * trans-mem.c (split_bb_make_tm_edge): Update profile.
17962 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17964 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
17965 to keep profile consistent.
17967 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17969 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
17970 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
17971 * profile-count.h (max_safe_multiplier): Make unsigned.
17972 (profile_count::guessed_zero): New.
17974 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17976 * bb-reorder.c (fix_up_crossing_landing_pad,
17977 fix_crossing_conditional_branches): Use make_single_succ_edge
17978 to keep profile consistent.
17980 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17982 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
17985 2017-07-01 Jakub Jelinek <jakub@redhat.com>
17988 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
17989 the right scopes, make sure cond_jump isn't preserved between multiple
17990 iterations. Search for fallthru edge whenever there are 3+ edges and
17991 use find_fallthru_edge for it.
17993 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17995 Patch by Alexander Monakov <amonakov@ispras.ru>
17996 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
17997 probabilities consistently.
17999 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18001 * pa.c (pa_expand_compare_and_swap_loop): Update call of
18002 emit_cmp_and_jump_insns.
18004 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18007 * tree-inline.c (expand_call_inline): Combine profile statuses.
18009 2017-06-30 Andrew Pinski <apinski@cavium.com>
18011 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
18012 fold_stmt returned true.
18014 2017-06-30 Nathan Sidwell <nathan@acm.org>
18016 * ggc.h (empty_string): Delete.
18017 * cfgexpand.c (expand_asm_stmt): Use plain "".
18018 * optabs.c (expand_asm_memory_barrier): Likewise.
18019 * stringpool.c (empty_string): Delete.
18020 (digit_vector, digit_string): Delete.
18021 (ggc_alloc_string): Use plain "", don't optimize single digit
18022 strings. Use ggc_alloc_atomic.
18024 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
18026 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
18027 comparison set and one other set, use the cost of the non-comparison
18030 2017-06-30 Nathan Sidwell <nathan@acm.org>
18032 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
18035 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
18037 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
18038 loops. Remove now unneeded calls to gimple_switch_set_label() that
18039 just set removed labels to NULL_TREE.
18041 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
18043 * tree-ssanames.c (set_range_info_raw): Abstract from ...
18044 (set_range_info): ...here. Only call set_range_info_raw if domain
18046 (set_nonzero_bits): Call set_range_info_raw.
18047 * tree-ssanames.h (set_range_info_raw): New.
18049 2017-06-30 Jakub Jelinek <jakub@redhat.com>
18052 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
18053 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
18054 of nonimmediate_operand and <store_mask_constraint> instead of m
18055 for the input operand. For V8FI iterator, always split if input
18056 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
18057 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
18058 <store_mask_predicate> instead of register_operand and
18059 <store_mask_constraint> instead of v for the input operand. Make
18060 sure both operands aren't MEMs for if not <mask_applied>.
18062 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
18064 * lto-wrapper.c (copy_file) Close both file descriptors before
18067 2017-06-30 Martin Liska <mliska@suse.cz>
18070 * multiple_target.c (create_dispatcher_calls): Make ifunc
18071 also for function that don't have calls or are not referenced.
18073 2017-06-30 Richard Biener <rguenther@suse.de>
18075 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
18076 analyze the first scalar stmt. Move vector type computation
18077 for the BB case here from ...
18078 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
18079 live operation processing in the SLP case properly.
18081 2017-06-30 Richard Biener <rguenther@suse.de>
18083 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
18085 2017-06-30 Martin Liska <mliska@suse.cz>
18088 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
18089 before BUILT_IN_UNWIND_RESUME when ASAN is used.
18091 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
18093 * doc/invoke.texi (AArch64): Add missing options and remove redundant
18096 2017-06-30 Richard Biener <rguenther@suse.de>
18098 PR tree-optimization/81249
18099 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
18100 condition reduction result to original scalar type.
18102 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18104 * profile-count.h (enum profile_quality): Fix typos and whitespace
18107 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18109 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
18110 type for branch probabilities.
18112 2017-06-29 Julian Brown <julian@codesourcery.com>
18113 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18115 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
18116 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
18117 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
18118 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
18120 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18122 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
18123 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
18124 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
18125 CC usage from generic code to here.
18126 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
18127 CC usage into the target macros.
18129 2017-06-29 Maya Rashish <coypu@sdf.org>
18131 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
18134 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18136 * arm/arm-builtins.c: Include profile-count.h
18137 * except.c (sjlj_emit_function_enter): Use
18138 profile_probability::unlikely.
18140 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18142 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
18143 and tocrel_offset be pointer args rather than implicitly using
18145 (legitimate_constant_pool_address_p, rs6000_emit_move,
18146 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
18147 tocrel_offset and use in toc_relative_expr_p call.
18148 (print_operand, print_operand_address): Use static tocrel_base_oac
18149 and tocrel_offset_oac.
18150 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
18153 2017-06-29 Maya Rashish <coypu@sdf.org>
18155 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
18157 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
18159 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
18160 objects, take into account only the alignment of 'op0' and 'mode1' if
18163 2017-06-29 Steve Ellcey <sellcey@cavium.com>
18165 * ccmp.c (ccmp_tree_comparison_p): New function.
18166 (ccmp_candidate_p): Update to use above function.
18167 (get_compare_parts): New function.
18168 (expand_ccmp_next): Update to use new functions.
18169 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
18171 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
18172 take mode as argument.
18173 * ccmp.h (expand_ccmp_expr): Add mode as argument.
18174 * expr.c (expand_expr_real_1): Pass mode as argument.
18176 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
18178 * combine.c (combine_instructions): Print insns to dump_file, together
18181 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18183 * asan.c (asan_emit_stack_protection): Update.
18184 (create_cond_insert_point): Update.
18185 * auto-profile.c (afdo_propagate_circuit): Update.
18186 * basic-block.h (struct edge_def): Turn probability to
18187 profile_probability.
18188 (EDGE_FREQUENCY): Update.
18189 * bb-reorder.c (find_traces_1_round): Update.
18190 (better_edge_p): Update.
18191 (sanitize_hot_paths): Update.
18192 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
18193 (make_single_succ_edge): Update.
18194 (check_bb_profile): Update.
18195 (dump_edge_info): Update.
18196 (update_bb_profile_for_threading): Update.
18197 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
18198 probabilitycount to 0.
18199 * cfgbuild.c (compute_outgoing_frequencies): Update.
18200 * cfgcleanup.c (try_forward_edges): Update.
18201 (outgoing_edges_match): Update.
18202 (try_crossjump_to_edge): Update.
18203 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
18204 (expand_gimple_tailcall): Update.
18205 (construct_init_block): Use make_single_succ_edge.
18206 (construct_exit_block): Use make_single_succ_edge.
18207 * cfghooks.c (verify_flow_info): Update.
18208 (redirect_edge_succ_nodup): Update.
18209 (split_edge): Update.
18210 (account_profile_record): Update.
18211 * cfgloopanal.c (single_likely_exit): Update.
18212 * cfgloopmanip.c (scale_loop_profile): Update.
18213 (set_zero_probability): Remove.
18214 (duplicate_loop_to_header_edge): Update.
18215 * cfgloopmanip.h (loop_version): Update prototype.
18216 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
18217 (force_nonfallthru_and_redirect): Update.
18218 (update_br_prob_note): Update.
18219 (rtl_verify_edges): Update.
18220 (purge_dead_edges): Update.
18221 (rtl_lv_add_condition_to_bb): Update.
18222 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
18223 * cgraphunit.c (init_lowered_empty_function): Update.
18224 (cgraph_node::expand_thunk): Update.
18225 * cilk-common.c: Include profile-count.h
18226 * dojump.c (inv): Remove.
18227 (jumpifnot): Update.
18228 (jumpifnot_1): Update.
18229 (do_jump_1): Update.
18231 (do_jump_by_parts_greater_rtx): Update.
18232 (do_compare_rtx_and_jump): Update.
18233 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
18234 do_jump_1. do_compare_rtx_and_jump): Update prototype.
18235 * dwarf2cfi.c: Include profile-count.h
18236 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
18237 (sjlj_emit_dispatch_table): Likewise.
18238 * explow.c: Include profile-count.h
18239 * expmed.c (emit_store_flag_force): Update.
18240 (do_cmp_and_jump): Update.
18241 * expr.c (compare_by_pieces_d::generate): Update.
18242 (compare_by_pieces_d::finish_mode): Update.
18243 (emit_block_move_via_loop): Update.
18244 (store_expr_with_bounds): Update.
18245 (store_constructor): Update.
18246 (expand_expr_real_2): Update.
18247 (expand_expr_real_1): Update.
18248 * expr.h (try_casesi, try_tablejump): Update prototypes.
18249 * gimple-pretty-print.c (dump_probability): Update.
18250 (dump_profile): New.
18251 (dump_gimple_label): Update.
18252 (dump_gimple_bb_header): Update.
18253 * graph.c (draw_cfg_node_succ_edges): Update.
18254 * hsa-gen.c (convert_switch_statements): Update.
18255 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
18256 (find_if_case_1): Update.
18257 (find_if_case_2): Update.
18258 * internal-fn.c (expand_arith_overflow_result_store): Update.
18259 (expand_addsub_overflow): Update.
18260 (expand_neg_overflow): Update.
18261 (expand_mul_overflow): Update.
18262 (expand_vector_ubsan_overflow): Update.
18263 * ipa-cp.c (good_cloning_opportunity_p): Update.
18264 * ipa-split.c (split_function): Use make_single_succ_edge.
18265 * ipa-utils.c (ipa_merge_profiles): Update.
18266 * loop-doloop.c (add_test): Update.
18267 (doloop_modify): Update.
18268 * loop-unroll.c (compare_and_jump_seq): Update.
18269 (unroll_loop_runtime_iterations): Update.
18270 * lra-constraints.c (lra_inheritance): Update.
18271 * lto-streamer-in.c (input_cfg): Update.
18272 * lto-streamer-out.c (output_cfg): Update.
18273 * mcf.c (adjust_cfg_counts): Update.
18274 * modulo-sched.c (sms_schedule): Update.
18275 * omp-expand.c (expand_omp_for_init_counts): Update.
18276 (extract_omp_for_update_vars): Update.
18277 (expand_omp_ordered_sink): Update.
18278 (expand_omp_for_ordered_loops): Update.
18279 (expand_omp_for_generic): Update.
18280 (expand_omp_for_static_nochunk): Update.
18281 (expand_omp_for_static_chunk): Update.
18282 (expand_cilk_for): Update.
18283 (expand_omp_simd): Update.
18284 (expand_omp_taskloop_for_outer): Update.
18285 (expand_omp_taskloop_for_inner): Update.
18286 * omp-simd-clone.c (simd_clone_adjust): Update.
18287 * optabs.c (expand_doubleword_shift): Update.
18288 (expand_abs): Update.
18289 (emit_cmp_and_jump_insn_1): Update.
18290 (expand_compare_and_swap_loop): Update.
18291 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
18292 * predict.c (predictable_edge_p): Update.
18293 (edge_probability_reliable_p): Update.
18294 (set_even_probabilities): Update.
18295 (combine_predictions_for_insn): Update.
18296 (combine_predictions_for_bb): Update.
18297 (propagate_freq): Update.
18298 (estimate_bb_frequencies): Update.
18299 (force_edge_cold): Update.
18300 * profile-count.c (profile_count::dump): Add missing space into dump.
18301 (profile_count::debug): Add newline.
18302 (profile_count::differs_from_p): Explicitly convert to unsigned.
18303 (profile_count::stream_in): Update.
18304 (profile_probability::dump): New member function.
18305 (profile_probability::debug): New member function.
18306 (profile_probability::differs_from_p): New member function.
18307 (profile_probability::differs_lot_from_p): New member function.
18308 (profile_probability::stream_in): New member function.
18309 (profile_probability::stream_out): New member function.
18310 * profile-count.h (profile_count_quality): Rename to ...
18311 (profile_quality): ... this one.
18312 (profile_probability): New.
18313 (profile_count): Update.
18314 * profile.c (compute_branch_probabilities): Update.
18315 * recog.c (peep2_attempt): Update.
18316 * sched-ebb.c (schedule_ebbs): Update.
18317 * sched-rgn.c (find_single_block_region): Update.
18318 (compute_dom_prob_ps): Update.
18319 (schedule_region): Update.
18320 * sel-sched-ir.c (compute_succs_info): Update.
18321 * stmt.c (struct case_node): Update.
18322 (do_jump_if_equal): Update.
18323 (get_outgoing_edge_probs): Update.
18324 (conditional_probability): Update.
18325 (emit_case_dispatch_table): Update.
18326 (expand_case): Update.
18327 (expand_sjlj_dispatch_table): Update.
18328 (emit_case_nodes): Update.
18329 * targhooks.c: Update.
18330 * tracer.c (better_p): Update.
18331 (find_best_successor): Update.
18332 * trans-mem.c (expand_transaction): Update.
18333 * tree-call-cdce.c: Update.
18334 * tree-cfg.c (gimple_split_edge): Upate.
18335 (move_sese_region_to_fn): Upate.
18336 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
18337 * tree-eh.c (lower_resx): Upate.
18338 (cleanup_empty_eh_move_lp): Upate.
18339 * tree-if-conv.c (version_loop_for_if_conversion): Update.
18340 * tree-inline.c (copy_edges_for_bb): Update.
18341 (copy_cfg_body): Update.
18342 * tree-parloops.c (gen_parallel_loop): Update.
18343 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
18344 (gimple_gen_time_profiler): Update.
18345 * tree-ssa-dce.c (remove_dead_stmt): Update.
18346 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
18347 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
18348 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
18349 (unloop_loops): Update.
18350 (try_peel_loop): Update.
18351 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18352 * tree-ssa-loop-split.c (connect_loops): Update.
18353 (split_loop): Update.
18354 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
18355 (hoist_guard): Update.
18356 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
18357 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
18358 (value_replacement): Update.
18359 * tree-ssa-reassoc.c (branch_fixup): Update.
18360 * tree-ssa-tail-merge.c (replace_block_by): Update.
18361 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
18362 (create_edge_and_update_destination_phis): Update.
18363 (compute_path_counts): Update.
18364 (recompute_probabilities): Update.
18365 (update_joiner_offpath_counts): Update.
18366 (freqs_to_counts_path): Update.
18367 (duplicate_thread_path): Update.
18368 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
18369 (struct switch_conv_info): Update.
18370 (gen_inbound_check): Update.
18371 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
18372 (vect_do_peeling): Update.
18373 (vect_loop_versioning): Update.
18374 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18375 (optimize_mask_stores): Update.
18376 * ubsan.c (ubsan_expand_null_ifn): Update.
18377 * value-prof.c (gimple_divmod_fixed_value): Update.
18378 (gimple_divmod_fixed_value_transform): Update.
18379 (gimple_mod_pow2): Update.
18380 (gimple_mod_pow2_value_transform): Update.
18381 (gimple_mod_subtract): Update.
18382 (gimple_mod_subtract_transform): Update.
18383 (gimple_ic): Update.
18384 (gimple_stringop_fixed_value): Update.
18385 (gimple_stringops_transform): Update.
18386 * value-prof.h: Update.
18388 2017-06-29 Carl Love <cel@us.ibm.com>
18390 * config/rs6000/rs6000-c.c: Add support for built-in functions
18391 vector signed int vec_signed (vector float);
18392 vector signed long long vec_signed (vector double);
18393 vector signed int vec_signed2 (vector double, vector double);
18394 vector signed int vec_signede (vector double);
18395 vector signed int vec_signedo (vector double);
18396 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
18397 instruction generator.
18398 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
18399 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
18400 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
18402 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
18403 vunsignede_v2df): Add define_expands.
18404 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
18405 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
18406 VEC_UNSIGNEDO): Add definitions.
18407 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
18408 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
18409 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
18410 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
18411 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
18412 * config/rs6000/altivec.h (vec_signed, vec_signed2,
18413 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
18414 vec_unsignede, vec_unsignedo): Add builtin defines.
18415 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
18417 * doc/extend.texi: Update the built-in documentation file for the
18418 new built-in functions.
18420 2017-06-29 Richard Biener <rguenther@suse.de>
18422 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
18423 reduction chains to LOOP_VINFO_REDUCTIONS.
18424 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
18425 SLP reductions after processing reduction chains.
18427 2017-06-29 Nathan Sidwell <nathan@acm.org>
18429 * builtins.c (fold_builtin_FUNCTION): Use
18430 lang_hooks.decl_printable_name.
18432 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
18434 PR middle-end/81194
18435 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
18436 with only one label.
18437 * stmt.c (expand_case): Assert NCASES is greater than one.
18439 2017-06-29 Richard Biener <rguenther@suse.de>
18441 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
18443 (group_case_labels): Likewise.
18444 (find_taken_edge): Push sanity checking on val to workers...
18445 (find_taken_edge_cond_expr): ... here
18446 (find_taken_edge_switch_expr): ... and here, handle cases
18447 with just a default label.
18448 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
18449 (group_case_labels): Likewise.
18450 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
18451 group_case_labels does anything cleanup the CFG again.
18453 2017-06-29 Bin Cheng <bin.cheng@arm.com>
18455 PR tree-optimization/81196
18456 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
18457 exit condition comparing two IVs.
18459 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
18461 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
18462 profile to the dummy entry at the end of the list of architectures.
18463 * config/arm/arm-cpu-cdata.h: Regenerated.
18465 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18466 Michael Collison <michael.collison@arm.com>
18469 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
18471 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
18472 (*aarch64_reg_<mode>3_minus_mask): New pattern.
18473 (*aarch64_<optab>_reg_di3_mask2): New pattern.
18474 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
18475 of shift when the shift amount is masked with constant equal to
18476 the size of the mode.
18477 * config/aarch64/predicates.md (subreg_lowpart_operator): New
18480 2017-06-29 Martin Liska <mliska@suse.cz>
18482 * config/i386/i386.opt: Change range from [1,5] to [0,5].
18484 2017-06-29 Yury Gribov <tetra2005@gmail.com>
18487 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
18490 (edge_growth_cache_entry::edge_growth_cache_entry): New
18492 (reset_edge_growth_cache): Update to use constructor.
18494 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18496 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
18497 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
18498 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
18500 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
18502 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
18503 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
18505 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
18507 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
18508 (*-linux-uclibc*): Add t-uclibc tmake_file.
18509 * config/t-musl: New.
18510 * config/t-uclibc: New.
18512 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
18514 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
18516 (gen_comm_data): Emit architectural setting of arch_prof.
18517 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
18519 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
18520 (armv8-m.base, armv8-m.main): Likewise.
18521 * arm-protos.h (arm_build_target): Add profile field.
18522 (arch_option): Likewise.
18523 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
18525 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
18526 arm_active_target.profile.
18528 2017-06-28 Richard Biener <rguenther@suse.de>
18530 PR middle-end/81227
18531 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
18532 TYPE_OVERFLOW_WRAPS.
18533 * match.pd (negate_expr_p): Likewise.
18534 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
18535 fold_build2, not fold_binary.
18537 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
18539 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
18540 Convert memory address to Pmode.
18541 (aarch64_print_operand): Assert MEM operands are always Pmode.
18543 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
18546 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
18547 Remove redundant if.
18548 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
18549 * config/arm/aarch-common-protos.h
18550 (aarch_forward_to_shift_is_not_shifted_re): Remove.
18551 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
18553 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
18556 * multiple_target.c (create_dispatcher_calls): Set the default
18557 clone to be static, not public.
18559 2017-06-28 Richard Biener <rguenther@suse.de>
18561 * tree-vect-loop.c (vectorizable_reduction): Move special
18562 cond reduction IV var creation ...
18563 (vect_create_epilog_for_reduction): ... here. Remove induction_index
18564 parameter. Use STMT_VINFO_VECTYPE.
18565 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
18568 2017-06-28 Martin Liska <mliska@suse.cz>
18571 * ipa-visibility.c (non_local_p): Handle visibility.
18573 2017-06-28 Martin Liska <mliska@suse.cz>
18576 * common.opt: Add IntegerRange to various options.
18577 * opt-functions.awk (integer_range_info): New function.
18578 * optc-gen.awk: Add integer_range_info to cl_options struct.
18579 * opts-common.c (decode_cmdline_option): Handle
18580 CL_ERR_INT_RANGE_ARG.
18581 (cmdline_handle_error): Likewise.
18582 * opts.c (print_filtered_help): Show valid interval in
18583 when --help is provided.
18584 * opts.h (struct cl_option): Add range_min and range_max fields.
18585 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
18587 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
18589 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
18590 (x * C EQ/NE y * C): New transformation.
18592 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
18594 * genmultilib (combination_space): Accept '+' in option names.
18596 2017-06-28 Martin Liska <mliska@suse.cz>
18599 * asan.c (instrument_derefs): Bail out inner references
18600 that are hard register variables.
18602 2017-06-28 Jakub Jelinek <jakub@redhat.com>
18605 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
18606 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
18608 2017-06-28 Richard Biener <rguenther@suse.de>
18610 * tree-vectorizer.h (vect_get_vec_defs): Remove.
18611 (vect_get_slp_defs): Adjust.
18612 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
18614 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
18616 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
18617 get_initial_defs_for_reduction instead of vect_get_vec_defs.
18618 (vectorizable_reduction): Adjust.
18619 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
18621 (vect_get_slp_defs): Likewise.
18622 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
18623 (vectorizable_bswap): Adjust.
18624 (vectorizable_call): Likewise.
18625 (vectorizable_conversion): Likewise.
18626 (vectorizable_assignment): Likewise.
18627 (vectorizable_shift): Likewise.
18628 (vectorizable_operation): Likewise.
18629 (vectorizable_store): Likewise.
18630 (vectorizable_condition): Likewise.
18631 (vectorizable_comparison): Likewise.
18633 2017-06-28 Michael Collison <michael.collison@arm.com>
18636 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
18638 (arm_gen_movmemqi): Removed unused variable 'i'.
18639 Convert 'for' loop into 'while' loop.
18640 (arm_expand_prologue): Remove last unnecessary set of insn.
18641 (thumb_pop): Remove unused variable 'pushed_words'.
18642 (thumb_exit): Remove last unnecessary set of regs_to_pop.
18644 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18646 * config/s390/predicates.md: Use s390_rel_address_ok_p.
18647 * config/s390/s390-protos.h: Add prototype of
18648 s390_rel_address_ok_p.
18649 * config/s390/s390.c (s390_got_symbol): New function.
18650 (s390_rel_address_ok_p): New function.
18651 (legitimize_pic_address): Use s390_rel_address_ok_p.
18652 (s390_load_got): Use s390_got_symbol.
18653 (s390_option_override): Issue error if
18654 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
18655 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
18657 * config/s390/s390.opt: New option mpic-data-is-text-relative.
18659 2017-06-27 Andrew Pinski <apinski@cavium.com>
18661 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
18662 (X * copysign (1.0, X)): New pattern.
18663 (X * copysign (1.0, -X)): New pattern.
18664 (copysign (-1.0, CST)): New pattern.
18666 2017-06-27 Joseph Myers <joseph@codesourcery.com>
18668 * genmultilib (combination_space): Remove variable.
18669 Validate reuse rules against regular expression for any sequence
18670 of multilib options in any order.
18672 2017-06-27 Michael Collison <michael.collison@arm.com>
18674 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
18675 call aarch64_split_simd_combine.
18676 * (aarch64_combine_internal<mode>): Delete pattern.
18677 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
18678 Allow register and subreg operands.
18680 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18682 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
18683 specific need, just fallback on defaults.
18684 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
18686 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18687 Olivier Hainque <hainque@adacore.com>
18689 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
18691 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
18692 targets. Pick a default if no particular attempt applied.
18693 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
18696 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18698 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
18699 (x86_64-wrs-vxworks7): Likewise.
18701 2017-06-27 Marek Polacek <polacek@redhat.com>
18704 * ubsan.c (instrument_null): Check get_base_address's result for null.
18706 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
18708 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
18710 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
18712 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
18713 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
18714 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
18715 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
18716 New function types.
18717 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
18718 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
18719 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
18720 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
18721 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
18722 BUILT_IN_FEUPDATEENV): New builtins.
18723 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
18724 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
18725 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
18726 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
18728 (builtin_structptr_types): Adjust size.
18729 * tree.c (builtin_structptr_types): Add four entries.
18731 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18732 Olivier Hainque <hainque@adacore.com>
18734 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
18735 (TLS_SYM): New local macro, forcing reference to __tls__ on
18736 link command lines for VxWorks 7 RTPs, triggering initialization
18738 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
18739 OS features TLS support, true for RTPs on VxWorks 7.
18740 * config/vxworks.c (vxworks_override_options): Setup emutls
18743 2017-06-27 Jakub Jelinek <jakub@redhat.com>
18745 * predict.c (test_prediction_value_range): Use -1U instead of -1
18746 to avoid narrowing conversion warning.
18747 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
18748 to avoid narrowing conversion warning.
18749 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
18751 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
18753 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18755 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
18756 64bit configurations.
18757 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
18758 (SIZE_TYPE): Likewise.
18759 * config/vxworks.c (vxworks_emutls_var_fields): Use
18760 long_unsigned_type_node instead of unsigned_type_node as the offset
18761 field type, which is "pointer" mode in emutls.c.
18763 2017-06-27 Jakub Jelinek <jakub@redhat.com>
18766 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
18768 PR middle-end/81207
18769 * gimple-fold.c (replace_call_with_call_and_fold): Handle
18770 gimple_vuse copying separately from gimple_vdef copying.
18772 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18774 * value-prof.c (free_hist): Remove call to memset and the enclosing if
18777 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
18778 Olivier Hainque <hainque@adacore.com>
18780 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
18781 for all vxworks7 targets.
18782 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
18783 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
18784 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
18785 variations for VX6/VX7 and 32/64bits later on in ...
18786 (VXWORKS_LIB_SPEC): Leverage new macros.
18787 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
18788 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
18790 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
18792 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
18793 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
18795 2017-06-26 Carl Love <cel@us.ibm.com>
18797 * config/rs6000/rs6000-c.c: Add support for built-in functions
18798 vector bool char vec_reve (vector bool char);
18799 vector signed char vec_reve (vector signed char);
18800 vector unsigned char vec_reve (vector unsigned char);
18801 vector bool int vec_reve (vector bool int);
18802 vector signed int vec_reve (vector signed int);
18803 vector unsigned int vec_reve (vector unsigned int);
18804 vector bool long long vec_reve (vector bool long long);
18805 vector signed long long vec_reve (vector signed long long);
18806 vector unsigned long long vec_reve (vector unsigned long long);
18807 vector bool short vec_reve (vector bool short);
18808 vector signed short vec_reve (vector signed short);
18809 vector double vec_reve (vector double);
18810 vector float vec_reve (vector float);
18811 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
18812 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
18813 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
18814 (altivec_vreve): New pattern.
18815 * config/rs6000/altivec.h (vec_reve): New define.
18816 * doc/extend.texi (vec_rev): Update the built-in documentation file
18817 for the new built-in functions.
18819 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18821 PR tree-optimization/71815
18822 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
18824 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
18826 (slsr_process_phi): Likewise.
18827 (replace_uncond_cands_and_profitable_phis): Don't replace a
18828 multiply candidate with a stride of 1 (copy or cast).
18829 (phi_incr_cost): Call uses_consumed_by_stmt rather than
18831 (lowest_cost_path): Likewise.
18832 (total_savings): Likewise.
18834 2017-06-26 Richard Biener <rguenther@suse.de>
18837 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
18838 Use def_builtin_pure for all gather builtins.
18840 2017-06-26 Richard Biener <rguenther@suse.de>
18842 PR tree-optimization/81203
18843 * tree-tailcall.c (find_tail_calls): Do not move stmts into
18844 non-dominating BBs.
18846 2017-06-26 Marek Polacek <polacek@redhat.com>
18849 * doc/invoke.texi: Document -Wmultistatement-macros.
18851 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
18853 * doc/sourcebuild.texi (ARM-specific attributes): Document new
18854 arm_neon_ok_no_float_abi effective target.
18856 2017-06-26 Richard Biener <rguenther@suse.de>
18858 PR tree-optimization/80928
18859 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
18860 (copy_bbs): Set BB_DUPLICATED flag early.
18861 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
18863 (execute_on_shrinking_pred): Likewise.
18864 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
18865 BB_DUPLICATED blocks.
18866 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
18867 iterate over all PHIs considering removal of *gsi.
18869 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
18871 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
18874 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18876 * config/rs6000/rs6000-string.c: (expand_block_clear,
18877 do_load_for_compare, select_block_compare_mode,
18878 compute_current_alignment, expand_block_compare,
18879 expand_strncmp_align_check, expand_strn_compare,
18880 expand_block_move, rs6000_output_load_multiple)
18881 Move functions related to string/block move/compare
18882 to a separate file.
18883 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
18884 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
18885 for this function which is now used in two files.
18886 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
18887 * config.gcc: Add rs6000-string.o to extra_objs for
18888 targets powerpc*-*-* and rs6000*-*-*.
18890 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
18893 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
18894 32-bit, since indexed is not valid for DImode.
18895 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
18896 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
18897 (define_peephole2 for Altivec d-form load): Add 32-bit support.
18898 (define_peephole2 for Altivec d-form store): Likewise.
18901 * multiple_target.c (create_dispatcher_calls): Only create the
18902 dispatcher call if the function is the default clone of a
18903 versioned function.
18905 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
18907 PR middle-end/80902
18908 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
18909 a call, force the call to not be a tail call.
18911 2017-06-23 Jeff Law <law@redhat.com>
18913 * doc/contrib.texi: Add entry for Steven Pemberton's work on
18916 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
18918 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
18919 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
18920 handling for early expansion of vector shifts (sl,sr,sra,rl).
18921 (builtin_function_type): Add vector shift right instructions
18922 to the unsigned argument list.
18924 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
18926 rtl-optimizatoin/79286
18927 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
18928 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
18929 trap. PIC register plus a const unspec without offset can never trap.
18931 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
18933 * tree.h (builtin_structptr_type): New type.
18934 (builtin_structptr_types): Declare new array.
18935 * tree.c (builtin_structptr_types): New array.
18936 (free_lang_data, build_common_tree_nodes): Use it.
18938 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
18941 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
18944 2017-06-22 Matt Turner <mattst88@gmail.com>
18946 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
18947 Lake models to skylake case. Assume skylake for unknown
18948 models with clflushopt.
18950 2017-06-22 Jeff Law <law@redhat.com>
18952 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
18953 frame sizes that do not satisfy aarch64_uimm12_shift.
18955 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
18957 * profile-count.h (apply_probability,
18958 apply_scale, probability_in): Fix checks for zero.
18960 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18962 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
18963 * doc/cppdiropts.texi (-I @var{dir}): Document it.
18965 2016-06-22 Richard Biener <rguenther@suse.de>
18967 * tree-vect-loop.c (vect_model_reduction_cost): Handle
18968 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
18969 REDUC_MAX_EXPR support.
18970 (vectorizable_reduction): Likewise.
18971 (vect_create_epilog_for_reduction): Likewise.
18973 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
18975 * match.pd (A / (1 << B) -> A >> B): New.
18976 * generic-match-head.c: Include optabs-tree.h.
18977 * gimple-match-head.c: Likewise.
18978 * optabs-tree.h (target_supports_op_p): New.
18979 * optabs-tree.c (target_supports_op_p): New.
18981 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18983 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
18984 $gcc_cv_ld --help output.
18985 (gcc_cv_ld_demangle): Likewise.
18986 (gcc_cv_ld_eh_frame_hdr): Likewise.
18987 (gcc_cv_ld_pie): Likewise.
18988 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
18989 (gcc_cv_ld_buildid): Likewise.
18990 (gcc_cv_ld_sysroot): Likewise.
18991 (ld_bndplt_support): Likewise.
18992 (ld_pushpopstate_support): Likewise.
18993 * configure: Regenerate.
18994 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
18996 2017-06-21 Jakub Jelinek <jakub@redhat.com>
18999 * config/i386/sse.md (round<mode>2): Renumber match_dup and
19000 operands indexes to avoid gap between operands and match_dups.
19002 2017-06-21 Andrew Pinski <apinski@cavium.com>
19004 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
19005 Increment Arith_shift and Arith_shift_reg by 1.
19006 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
19008 * config/aarch64/aarch64.c (thunderx_tunings): Enable
19009 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
19010 (aarch64_strip_extend): Add new argument and test for it.
19011 (aarch64_cheap_mult_shift_p): New function.
19012 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
19013 add a cost if it is true.
19014 Update calls to aarch64_strip_extend.
19015 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
19017 2017-06-21 Andrew Pinski <apinski@cavium.com>
19019 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
19021 (thunderxt88): Likewise.
19022 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
19023 (thunderx_prefetch_tune): New variable.
19024 (thunderx2t99_prefetch_tune): Update for the correct values.
19025 (thunderxt88_tunings): New variable.
19026 (thunderx_tunings): Use thunderx_prefetch_tune instead of
19027 generic_prefetch_tune.
19028 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
19030 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19032 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
19033 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
19034 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
19035 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
19036 (aarch64_atomic_cas<mode>, GPI): Likewise.
19038 2017-06-21 Martin Liska <mliska@suse.cz>
19040 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
19041 statements on cold and hot labels.
19042 * predict.c (tree_estimate_probability_bb): Remove the
19043 prediction from this place.
19045 2017-06-21 Martin Liska <mliska@suse.cz>
19047 PR tree-optimization/79489
19048 * gimplify.c (maybe_add_early_return_predict_stmt): New
19050 (gimplify_return_expr): Call the function.
19051 * predict.c (tree_estimate_probability_bb): Remove handling
19053 * predict.def: Update comment about early return predictor.
19054 * gimple-predict.h (is_gimple_predict): New function.
19055 * predict.def: Change default value of early return to 66.
19056 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
19058 * passes.def: Put pass_strip_predict_hints to the beginning of
19061 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
19063 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
19064 FUNCTION_DECL declarations.
19065 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
19067 (dwaf2out_decl): Likewise.
19068 * godump.c (go_early_global_decl): Skip call to the real debug hook
19069 for FUNCTION_DECL declarations.
19070 * passes.c (rest_of_decl_compilation): Skip call to the
19071 early_global_decl debug hook for FUNCTION_DECL declarations, unless
19072 -fdump-go-spec is passed.
19074 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
19076 * config/i386/i386.c (struct builtin_isa): New field pure_p.
19077 Reorder for compactness.
19078 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
19079 (def_builtin_pure, def_builtin_pure2): New functions.
19080 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
19082 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
19084 * match.pd (nop_convert): New predicate.
19085 ((A +- CST1) +- CST2): Allow some NOP conversions.
19087 2017-06-21 Jakub Jelinek <jakub@redhat.com>
19090 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
19091 with ctors/dtors if GOVD_SHARED is set.
19093 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19095 * config/aarch64/aarch64.md (movti_aarch64):
19096 Emit mov rather than orr.
19097 (movtf_aarch64): Likewise.
19098 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
19099 Emit mov rather than orr.
19101 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19103 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
19104 Swap alternatives, make integer dup more expensive.
19106 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19108 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
19109 Return true for non-tls symbols.
19111 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
19113 * config/aarch64/aarch64-cores.def (cortex-a55): New.
19114 (cortex-a75): Likewise.
19115 (cortex-a75.cortex-a55): Likewise.
19116 * config/aarch64/aarch64-tune.md: Regenerate.
19117 * doc/invoke.texi (-mtune): Document new values for -mtune.
19119 2017-06-21 Tom de Vries <tom@codesourcery.com>
19121 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
19122 stack_size feature.
19123 (Effective-Target Keywords, Other attributes): Suggest using
19124 dg-add-options stack_size feature to get stack limit in stack_size
19125 effective target documentation.
19127 2017-06-21 Julian Brown <julian@codesourcery.com>
19128 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19130 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
19131 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
19132 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
19134 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
19135 attribute type list for neon_multiply.
19136 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
19137 attribute type list for neon_multiply.
19138 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
19139 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
19140 attribute type list for neon_multiply.
19141 * config/arm/types.md (crypto_pmull): Add.
19142 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
19143 attribute type list.
19145 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
19147 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
19150 2017-06-20 Jakub Jelinek <jakub@redhat.com>
19152 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
19153 to make sure not to dereference a NULL cost_classes_ptr pointer.
19155 2017-06-20 Carl Love <cel@us.ibm.com>
19157 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19158 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
19159 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
19160 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
19161 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
19162 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
19163 VMULOSW): New enum "unspec" values.
19164 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
19165 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
19166 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
19167 altivec_vmulosw): New patterns.
19168 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
19169 VMULOSW): Add definitions.
19171 2017-06-20 Julia Koval <julia.koval@intel.com>
19173 * config/i386/i386.c: Fix rounding expand for new pattern.
19174 * config/i386/subst.md: Fix pattern (parallel -> unspec).
19176 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19178 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
19179 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
19181 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19183 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
19186 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19188 * config/aarch64/aarch64-cores.def: Rearrange to sort by
19189 architecture, then by implementer ID.
19190 * config/aarch64/aarch64-tune.md: Regenerate.
19192 2017-06-20 Richard Biener <rguenther@suse.de>
19194 PR middle-end/81097
19195 * fold-const.c (split_tree): Fold to type before negating.
19197 2017-06-20 David Malcolm <dmalcolm@redhat.com>
19199 * diagnostic-show-locus.c
19200 (selftest::test_fixit_deletion_affecting_newline): New function.
19201 (selftest::diagnostic_show_locus_c_tests): Call it.
19203 2017-06-20 Andreas Schwab <schwab@suse.de>
19206 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
19209 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19211 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
19212 __ARM_FEATURE_COPROC according to support.
19214 2017-06-20 Jakub Jelinek <jakub@redhat.com>
19216 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
19217 Rewritten to avoid overflow for > 32-bit pointers.
19220 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
19221 by removing enum keyword.
19222 (ubsan_type_descriptor): Likewise. Formatting fix.
19225 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
19226 splitter): Require TARGET_SSE2 in the condition.
19228 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
19231 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
19232 for doing vector set of SFmode on ISA 3.0.
19233 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
19234 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
19236 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
19237 SFmode value into a V4SF variable that was extracted from another
19238 V4SF variable without converting the element to double precision
19239 and back to single precision vector format.
19240 (vsx_insert_extract_v4sf_p9_2): Likewise.
19242 2017-06-19 Jakub Jelinek <jakub@redhat.com>
19244 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
19245 in UWHI to avoid undefined overflow.
19248 * ubsan.h (enum ubsan_encode_value_phase): New.
19249 (ubsan_encode_value): Change second argument to
19250 enum ubsan_encode_value_phase with default value of
19251 UBSAN_ENCODE_VALUE_GENERIC.
19252 * ubsan.c (ubsan_encode_value): Change second argument to
19253 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
19254 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
19255 create_tmp_var_raw instead of create_tmp_var and use a
19257 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
19258 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
19259 ubsan_encode_value callers.
19262 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
19263 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
19264 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
19266 2017-06-19 Richard Biener <rguenther@suse.de>
19268 PR middle-end/81118
19269 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
19270 estimates if we changed anything.
19272 2017-06-19 Richard Biener <rguenther@suse.de>
19274 PR tree-optimization/80887
19275 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
19276 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
19277 simplified lookups, then reset mprts_hook.
19278 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
19280 (try_to_simplify): Likewise.
19282 2017-06-19 Martin Liska <mliska@suse.cz>
19285 * gimplify.c (gimplify_switch_expr):
19286 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
19288 2017-06-19 Martin Liska <mliska@suse.cz>
19290 * doc/install.texi: Document that PGO runs in 4 stages.
19292 2017-06-19 Martin Liska <mliska@suse.cz>
19295 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
19296 to dispatcher function name.
19297 * multiple_target.c (replace_function_decl): New function.
19298 (create_dispatcher_calls): Redirect both edges and references.
19300 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
19302 * profile-count.c (profile_count::dump): Dump quality.
19303 (profile_count::differs_from_p): Update for unsigned val.
19304 * profile-count.h (profile_count_quality): New enum.
19305 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
19307 2017-06-19 Richard Biener <rguenther@suse.de>
19309 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
19310 struct function as arg.
19311 (estimate_numbers_of_iterations): Export overload with loop arg.
19312 (free_numbers_of_iterations_estimates_loop): Use an overload of
19313 free_numbers_of_iterations_estimates instead.
19314 * tree-cfg.c (remove_bb): Adjust.
19315 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
19316 * tree-parloops.c (gen_parallel_loop): Likewise.
19317 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
19319 (tree_unroll_loops_completely): Likewise.
19320 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
19321 Use an overload instead and export.
19322 (estimated_loop_iterations): Adjust.
19323 (max_loop_iterations): Likewise.
19324 (likely_max_loop_iterations): Likewise.
19325 (estimate_numbers_of_iterations): Take struct function as arg
19327 (loop_exits_before_overflow): Adjust.
19328 (free_numbers_of_iterations_estimates_loop): Use an overload.
19329 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
19330 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
19332 2017-06-19 Richard Biener <rguenther@suse.de>
19335 * ipa-prop.c (find_constructor_constant_at_offset): Handle
19336 RANGE_EXPR conservatively.
19338 2017-06-16 Carl Love <cel@us.ibm.com>
19340 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19341 definitions for vec_float, vec_float2, vec_floato,
19342 vec_floate built-ins.
19343 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
19344 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
19346 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
19347 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
19348 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
19349 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
19350 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
19351 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
19352 vec_floato): Add builtin defines.
19353 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
19354 Update the built-in documentation file for the new built-in
19357 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19359 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
19360 (mthumb): Mark as the negative of -marm.
19362 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19364 * doc/invoke.texi (ARM Options, -mcpu): Document supported
19366 (ARM Options, -mtune): Document that this accepts the same
19367 extension options as -mcpu.
19368 (ARM Options, -mfpu): Document addition of -mfpu=auto.
19370 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19372 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
19373 permitted extensions.
19375 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19377 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
19378 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
19379 (armv8-m.main): Add option +nodsp.
19380 * config/arm/arm-cpu-cdata.h: Regenerated.
19382 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19384 * config/arm/t-fuchsia: New file.
19385 * config.gcc (arm*-*-fuchsia*): Use it.
19387 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19389 * config/arm/t-symbian: Rewrite for new option infrastructure.
19391 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19393 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
19394 (MULTILIB_REQUIRED): Likewise.
19396 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19398 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
19399 (MULTILIB_RESUE): Likewise.
19400 (MULTILIB_MATCHES): Likewise.
19401 (MULTLIB_REQUIRED): Likewise.
19403 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19405 * config/arm/t-rtems: Rewrite for new option framework.
19407 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19409 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
19410 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
19411 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
19412 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
19413 * config/arm/t-multilib: ... here.
19414 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
19415 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
19416 armv7-a and armv8*-a when A-profile libraries have not been built.
19417 * config/arm/t-rmprofile: Rewrite.
19419 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19421 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
19422 with a backslash. Remove the backslash after substituting unescaped
19424 * doc/fragments.texi (MULTILIB_REUSE): Document it.
19426 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19428 * config.gcc: (arm*-*-*): When building a-profile libraries, force
19429 the driver to pass through the default setting of -mfloat-abi.
19430 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
19432 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
19433 (all_feat_combs): New rule.
19434 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
19436 * config/arm/t-aprofile: Rewrite.
19438 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19440 * config/arm/arm.h (FPUTYPE_AUTO): Define.
19441 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
19442 fpu is not specified by the user/command-line.
19443 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
19444 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
19445 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
19446 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
19447 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
19448 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
19450 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19452 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
19453 * config/arm/t-arm-elf: Rewritten.
19455 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19457 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
19458 have some floating-point instructions.
19459 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
19460 (TARGET_MAYBE_HARD_FLOAT): New macro.
19461 * config/arm/arm-builtins.c (arm_init_builtins): Use
19462 TARGET_MAYBE_HARD_FLOAT.
19463 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
19465 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19467 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
19468 (configargs.h): Include it.
19469 (arm_print_hint_for_fpu_option): New function.
19470 (arm_parse_fpu_option): New function.
19471 (candidate_extension): New class.
19472 (arm_canon_for_multilib): New function.
19473 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
19474 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
19475 (ARCH_CANONICAL_SPECS): New macro.
19476 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
19478 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19480 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
19481 are set after handling multilib fragments. Set target_cpu_default2
19484 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19486 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
19488 (arm*-*-*): Set target_cpu_default2 to a quoted string.
19489 * config/arm/parsecpu.awk (check_cpu): Validate any extension
19491 (check_arch): Likewise.
19492 * config/arm/arm.c (arm_configure_build_target): Handle
19493 TARGET_CPU_DEFAULT being a string constant. Scan any feature
19494 options in the default.
19496 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19498 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
19499 when an option is an alias of another.
19500 * config/arm/parsecpu.awk (optalias): New parser token.
19501 (gen_comm_data): Mark non-alias options as such. Emit entries
19502 for extension aliases.
19503 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
19504 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
19505 (armv6kz, armv6zk, armv6t2): Likewise.
19506 (armv7): Make vfpv3-d16 an alias.
19507 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
19509 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
19510 Sort in canonical order.
19511 (armv8-a): Sort in canonical order.
19512 (armv8.1-a, armv8.2-a): Likewise.
19513 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
19515 (cortex-a9): Sort in canonical order.
19516 * config/arm/arm.c (selftests.h): Include it.
19517 (arm_test_cpu_arch_data): New function.
19518 (arm_run_self_tests): New function.
19519 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
19520 (targetm): Move declaration to the end of the file.
19521 * arm-cpu-cdata.h: Regenerated.
19523 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19525 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
19526 call to target_mode_check describing the type of option passed.
19527 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
19528 (arm_target_thumb_only): Use arm_parse_arch_option_name or
19529 arm_parse_cpu_option_name to match parameters against list of
19531 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
19532 arm_arch_core_flags data structure.
19533 * config/arm/arm-cpu_cdata.h: Regenerated.
19535 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19537 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
19539 (arm_print_hint_for_cpu_option): Likewise.
19540 (arm_print_hint_for_arch_option): Likewise.
19541 (arm_parse_cpu_option_name): Likewise.
19542 (arm_parse_arch_option_name): Likewise.
19543 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
19544 of entries in the all_fpus list.
19545 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
19546 (arm_parse_cpu_option_name): Declare.
19547 (arm_parse_arch_option_name): Declare.
19548 (arm_parse_option_features): Declare.
19549 (arm_intialize_isa): Declare.
19550 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
19552 (gen_comm_data): ... here. Make definitions non-static.
19553 * config/arm/arm-cpu-data.h: Regenerated.
19554 * config/arm/arm-cpu-cdata.h: Regenerated.
19556 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19558 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
19559 (cpu_arch_extension): New structure.
19560 (cpu_arch_option, arch_option, cpu_option): New structures.
19561 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
19562 architecture types.
19563 (gen_data): Generate new format data tables.
19564 * config/arm/arm.c (cpu_tune): New structure.
19565 (cpu_option, processors): Delete.
19566 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
19567 (arm_print_hint_for_cpu_option): ... this and ...
19568 (arm_print_hint_for_arch_option): ... this.
19569 (arm_parse_arch_cpu_name): Delete. Replace with ...
19570 (arm_parse_cpu_option_name): ... this and ...
19571 (arm_parse_arch_option_name): ... this.
19572 (arm_unrecognized_feature): Change type of target parameter to
19574 (arm_parse_arch_cpu_features): Delete. Replace with ...
19575 (arm_parse_option_features): ... this.
19576 (arm_configure_build_target): Rework to use new configuration data
19578 (arm_print_tune_info): Rework for new configuration data tables.
19579 * config/arm/arm-cpu-data.h: Regenerated.
19580 * config/arm/arm-cpu.h: Regenerated.
19582 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19584 * Makefile.in (OBJS): Move sbitmap.o from here ...
19585 (OBJS-libcommon): ... to here.
19587 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19589 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
19590 (ISA_ALL_CRYPTO): New macro.
19591 (ISA_ALL_SIMD): New macro
19592 (ISA_ALL_FP): New macro.
19593 * config/arm/arm.c (fpu_bitlist): Update initializer.
19594 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
19596 (arm9e): Add fpu. Add option for nofp
19597 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
19598 (arm926ej-s, arm1026ej-s): Likewise.
19599 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
19600 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
19601 neon-fp16, neon-vfpv4, nofp and nosimd.
19602 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
19603 (cortex-a8): Add fpu. Add option for nofp.
19604 (cortex-a9): Add fpu. Add options for nosimd and nofp.
19605 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
19606 (cortex-r4f): Add fpu.
19607 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
19608 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
19610 (cortex-r8): Likewise.
19611 (cortex-m4): Add fpu. Add option for nofp.
19612 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
19613 (cortex-a17.cortex-a7): Likewise.
19614 (cortex-a32): Add fpu. Add options for crypto and nofp.
19615 (cortex-a35, cortex-a53): Likewise.
19616 (cortex-a57): Add fpu. Add option for crypto.
19617 (cortex-a72, cortex-a73): Likewise.
19618 (exynos-m1): Likewise.
19619 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
19620 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
19621 (cortex-m33): Add fpu. Add option for nofp.
19622 * config/arm/arm-cpu-cdata.h: Regenerated
19623 * config/arm/arm-cpu-data.h: Regenerated.
19625 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19627 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
19628 (armv5te, armv5tej): Likewise.
19629 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
19630 (armv7): Add options fp and vfpv3-d16.
19631 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
19632 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
19634 (armv7ve): Likewise.
19635 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
19636 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
19637 (armv8-a): Add nocrypto option.
19638 (armv8.1-a, armv8.2-a): Likewise.
19639 (armv8-m.main): add options fp, fp.dp and nofp.
19641 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19643 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
19645 (armv8-a+crc): Delete.
19646 (armv8.1-a): Add options simd, crypto and nofp.
19647 (armv8.2-a): Add options fp16, simd, crypto and nofp.
19648 (armv8.2-a+fp16): Delete.
19649 (armv8-m.main): Add option dsp.
19650 (armv8-m.main+dsp): Delete.
19651 (cortex-a8): Add fpu. Add nofp option.
19652 (cortex-a9): Add fpu. Add nofp and nosimd options.
19653 * config/arm/parsecpu.awk (gen_data): Generate option tables and
19654 link to main cpu and architecture data structures.
19655 (gen_comm_data): Only put isa attributes from the main architecture
19657 (option): New statement for architecture and CPU entries.
19658 * arm.c (struct cpu_option): New structure.
19659 (struct processors): Add entry for options.
19660 (arm_unrecognized_feature): New function.
19661 (arm_parse_arch_cpu_name): Ignore any characters after the first
19663 (arm_parse_arch_cpu_feature): New function.
19664 (arm_configure_build_target): Separate out any CPU and architecture
19665 features and parse separately. Don't error out if -mfpu=auto is
19666 used with only an architecture string.
19667 (arm_print_asm_arch_directives): New function.
19668 (arm_file_start): Call it.
19669 * config/arm/arm-cpu-cdata.h: Regenerated.
19670 * config/arm/arm-cpu-data.h: Likewise.
19671 * config/arm/arm-tables.opt: Likewise.
19673 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19675 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
19676 assembler when it is not -mfpu=auto.
19678 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19680 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
19681 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
19682 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
19683 (ASM_CPU_SPEC): Rewrite.
19684 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
19685 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
19686 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
19687 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
19688 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
19689 copied string is NUL-terminated. Also strip any characters prefixed
19691 (arm_rewrite_selected_arch): New function.
19692 (arm_rewrite_march): New function.
19694 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19696 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
19697 (x_arm_cpu_string, x_arm_tune_string): Likewise.
19698 (march, mcpu, mtune): Convert to string-based options.
19699 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
19700 (arm_parse_arch_cpu_name): New function.
19701 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
19702 identify selected architecture or CPU.
19703 (arm_option_save): New function.
19704 (TARGET_OPTION_SAVE): Redefine.
19705 (arm_option_restore): Restore string options.
19706 (arm_option_print): Print string options.
19708 2017-06-16 Martin Sebor <msebor@redhat.com>
19710 PR tree-optimization/80933
19711 PR tree-optimization/80934
19712 * builtins.c (fold_builtin_3): Do not handle bcmp here.
19713 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
19714 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
19715 (gimple_fold_builtin): Call them.
19717 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19719 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
19720 as unlikely; update profile.
19722 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19724 * predict.c (force_edge_cold): Handle declaring edges impossible
19727 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19729 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
19731 (try_unroll_loop_completely): Fix reporting.
19733 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19735 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
19737 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
19740 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
19741 if given a non-constant argument for an intrinsic which requires a
19744 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19746 * profile.c (compare_freqs): New function.
19747 (branch_prob): Sort edge list.
19748 (find_spanning_tree): Assume that the list is priority sorted.
19750 2017-06-16 Richard Biener <rguenther@suse.de>
19752 PR tree-optimization/81090
19753 * passes.def (pass_record_bounds): Remove.
19754 * tree-pass.h (make_pass_record_bounds): Likewise.
19755 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
19756 make_pass_record_bounds): Likewise.
19757 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
19758 not free niter estimates at the beginning but at the end.
19759 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
19761 2017-06-16 Richard Biener <rguenther@suse.de>
19763 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
19764 initializer to workaround ICE in host GCC 4.8.
19766 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19768 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
19770 (clone_inlined_nodes): Update.
19772 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19774 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
19775 prefetch settings, and enable prefetching by default at -O3.
19777 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19779 * config/aarch64/aarch64.c (aarch64_override_options_internal):
19780 Set flag_prefetch_loop_arrays according to tuning data.
19782 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19784 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
19785 New tune structure.
19786 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
19787 [Unrelated to main purpose of the patch] Place the pointer field last
19788 to enable type checking errors when tune structure are wrongly merged.
19789 * config/aarch64/aarch64.c (generic_prefetch_tune,)
19790 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
19791 (thunderx2t99_prefetch_tune): New tune constants.
19792 (tune_params *_tunings): Update all tunings (no functional change).
19793 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
19794 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
19795 from tunings structures.
19797 2017-06-16 Jakub Jelinek <jakub@redhat.com>
19800 * ubsan.c (instrument_null): Add T argument, use it instead
19801 of computing it based on IS_LHS.
19802 (instrument_object_size): Likewise.
19803 (pass_ubsan::execute): Adjust instrument_null and
19804 instrument_object_size callers to pass gimple_get_lhs or
19805 gimple_assign_rhs1 result to it. Use instrument_null instead of
19806 calling get_base_address and instrument_mem_ref. Handle
19807 aggregate call arguments for object-size sanitization.
19809 2017-06-16 Yury Gribov <tetra2005@gmail.com>
19811 PR tree-optimization/81089
19812 * tree-vrp.c (is_masked_range_test): Validate operands of
19815 2017-06-15 Martin Sebor <msebor@redhat.com>
19818 * dumpfile.c (dump_register): Avoid calling memset to initialize
19819 a class with a default ctor.
19820 * gcc.c (struct compiler): Remove const qualification.
19821 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
19822 * hash-table.h: Ditto.
19823 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
19825 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
19826 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
19828 * params.h (struct param_info): Make struct members non-const.
19829 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
19830 with default initialization.
19831 * vec.h (vec_copy_construct, vec_default_construct): New helper
19833 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
19834 with vec_copy_construct.
19835 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
19836 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
19837 * doc/invoke.texi (-Wclass-memaccess): Document.
19839 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19841 * emit-rtl.h (is_leaf): Update comment about local
19842 register allocator.
19844 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19847 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
19848 for a variable to have a section before checking if the section has a
19850 Set section to.persistent if persistent attribute is set.
19851 Warn if .persistent attribute is used on an automatic variable.
19853 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
19855 PR rtl-optimization/80474
19856 * reorg.c (update_block): Do not ignore instructions in a delay slot.
19858 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
19860 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
19863 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
19865 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
19866 (casesi): Emit bounds checking as RTL.
19867 (casesi_internal_mips16_<mode>): Remove bounds checking.
19869 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
19871 * config/xtensa/xtensa.c (xtensa_option_override): Append
19872 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
19873 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
19874 xtensa_doloop_hooks): Define unconditionally.
19875 (xtensa_reorg_loops): Only call reorg_loops in the presence of
19877 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
19878 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
19879 for it in xtensa_option_override.
19880 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
19881 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
19883 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
19885 * doc/cppopts.texi: Document '-' special value to -MF.
19887 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
19889 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
19890 (cortex_a53_fconst): Likewise.
19891 (cortex_a53_fpmul): Likewise.
19892 (cortex_a53_f_load_64): Likewise.
19893 (cortex_a53_f_load_many): Likewise.
19894 (cortex_a53_advsimd_alu): Likewise.
19895 (cortex_a53_advsimd_alu_q): Likewise.
19896 (cortex_a53_advsimd_mul): Likewise.
19897 (cortex_a53_advsimd_mul_q): Likewise.
19898 (fpmac bypass): Add new bypass for fpmac-fpmac case.
19899 Add missing fmul, r2f_cvt and fconst cases.
19901 2017-06-14 Richard Biener <rguenther@suse.de>
19903 PR middle-end/81088
19904 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
19906 (fold_binary_loc): When associating do not treat pre-existing
19907 TREE_OVERFLOW on literal constants as a reason to allow
19908 TREE_OVERFLOW on associated literal constants.
19910 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
19912 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
19913 (MASK_FEATURES): New macro.
19914 * config/sparc/sparc.c (sparc_option_override): Remove the special
19915 handling of -mfpu and generalize it to all MASK_FEATURES switches.
19917 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
19919 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
19920 a division of 0 if non-call exceptions are enabled.
19922 2017-06-14 Andrew Pinski <apinski@cavium.com>
19923 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19926 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
19927 Improve vector initialization code gen for only variable case.
19929 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
19931 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
19933 2017-06-14 Richard Biener <rguenther@suse.de>
19935 PR tree-optimization/81083
19936 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
19939 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19941 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
19942 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
19943 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
19944 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
19945 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
19946 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
19948 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19950 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
19951 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
19953 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19955 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
19957 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19959 * config/rs6000/t-rtems: Don't handle SPE.
19961 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19963 * config/rs6000/t-linux: Don't handle SPE.
19965 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19967 * config/rs6000/eabispe.h: Delete file.
19969 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19971 * config/rs6000/t-spe: Delete file.
19973 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19975 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
19976 (rs6000_legitimate_offset_address_p): Return false for anything in
19977 V2SImode or V2SFmode.
19979 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19981 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
19982 except V2SF and V2SI. Rearrange the vector modes, and add comments.
19983 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
19985 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
19986 (rs6000_legitimate_offset_address_p): Ditto.
19987 (rs6000_emit_move): Ditto.
19988 (rs6000_init_builtins): Remove V4HI_type_node.
19990 2017-06-13 Martin Liska <mliska@suse.cz>
19993 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
19994 (gate_asan): Likewise.
19995 * asan.h (asan_no_sanitize_address_p): Remove the function.
19996 (sanitize_flags_p): New function.
19997 * builtins.def: Fix coding style.
19998 * common.opt: Use renamed enum value.
19999 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
20000 * doc/extend.texi: Document no_sanitize attribute.
20001 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
20002 to SANITIZE_UNDEFINED_NONDEFAULT.
20003 * gcc.c (sanitize_spec_function): Use the renamed enum value.
20004 * gimple-fold.c (optimize_atomic_compare_exchange_p):
20005 Use sanitize_flags_p.
20006 * gimplify.c (gimplify_function_tree): Likewise.
20007 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
20008 * opts.c (parse_no_sanitize_attribute): New function.
20009 (common_handle_option): Use renamed enum value.
20010 * opts.h (parse_no_sanitize_attribute): Declare.
20011 * tree.c (sanitize_flags_p): New function.
20012 * tree.h: Declared here.
20013 * tsan.c: Use sanitize_flags_p.
20014 * ubsan.c (ubsan_expand_null_ifn): Likewise.
20015 (instrument_mem_ref): Likewise.
20016 (instrument_bool_enum_load): Likewise.
20017 (do_ubsan_in_current_function): Remove the function.
20018 (pass_ubsan::execute): Use sanitize_flags_p.
20019 * ubsan.h: Remove do_ubsan_in_current_function
20020 * tree-cfg.c (print_no_sanitize_attr_value): New function.
20021 (dump_function_to_file): Use it here.
20023 2017-06-13 Martin Jambor <mjambor@suse.cz>
20025 PR tree-optimization/80803
20026 PR tree-optimization/81063
20027 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
20028 (propagate_subaccesses_across_link): Enqueue subtree whenever
20029 necessary instead of relying on the caller.
20031 2017-06-13 Martin Jambor <mjambor@suse.cz>
20033 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
20034 that have a first_link.
20035 (sort_and_splice_var_accesses): Do not check first_link before
20037 (subtree_mark_written_and_enqueue): Likewise.
20038 (propagate_all_subaccesses): Likewise and do not stop at first
20039 parent with a first_link.
20041 2017-06-13 Martin Jambor <mjambor@suse.cz>
20043 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
20046 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20048 * match.pd: New pattern.
20050 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20052 * tree-vrp.c (is_masked_range_test): New function.
20053 (register_edge_assert_for): Determine ranges for
20056 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20058 PR tree-optimization/67328
20059 * fold-const.c (maskable_range_p): New function.
20060 (build_range_check): Generate bittests if possible.
20062 2017-06-13 Martin Liska <mliska@suse.cz>
20064 * gimple-pretty-print.c (dump_probability): Add new argument.
20065 (dump_edge_probability): Dump both probability and count.
20066 (dump_gimple_label): Likewise.
20067 (dump_gimple_bb_header): Likewise.
20069 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
20072 * config/avr/avr-devices.c: Fix indentation.
20073 * config/avr/gen-avr-mmcu-specs.c: Dito.
20075 2017-06-13 Richard Biener <rguenther@suse.de>
20077 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
20078 instead get vector type from stmt_info.
20079 (vectorizable_reduction): Adjust. Remove dead code.
20081 2017-06-13 Richard Biener <rguenther@suse.de>
20083 PR middle-end/81065
20084 * fold-const.c (extract_muldiv_1): Remove bogus distribution
20085 case of C * (x * C2 + C3).
20086 (fold_addr_of_array_ref_difference): Properly fold index difference.
20088 2017-06-12 David S. Miller <davem@davemloft.net>
20091 * config/sparc/sparc.md (return expander): Emit frame blockage if
20092 function uses alloca.
20094 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
20096 * combine.c (make_field_assignment): Check len rather than the mode
20097 precision when calling force_to_mode.
20099 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
20101 Support multilibs and devices that see flash in RAM address range.
20104 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
20105 (avr_mcu_t) <flash_pm_offset>: New field.
20106 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
20107 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
20108 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
20109 (AVR_TINY_PM_OFFSET): Remove macro.
20110 * config/avr/avr.opt (-mshort-calls): New option.
20111 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
20112 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
20113 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
20114 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
20115 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
20116 instead of avr_arch->have_jmp_call.
20117 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
20118 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
20119 avr_arch->flash_pm_offset to define.
20120 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
20121 new field flash_pm_offset. Add entry for avrxmega3.
20122 (avr_texinfo): Add entry for avrxmega3.
20123 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
20124 attiny212, attiny214,
20125 attiny412, attiny414, attiny416, attiny417,
20126 attiny814, attiny816, attiny817,
20127 attiny1614, attiny1616, attiny1617,
20128 attiny3214, attiny3216, attiny3217.
20129 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
20130 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
20131 (avr_print_operand_address) [AVR_TINY]: Same.
20132 (avr_asm_init_sections) <readonly_data_section>: Only patch
20133 callback if avr_arch->flash_pm_offset = 0.
20134 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
20135 for rodata if avr_arch->flash_pm_offset != 0.
20136 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
20137 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
20138 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
20139 (m_options): Append opt_rcall.
20140 (m_dirnames): Append dir_rcall.
20141 * config/avr/t-multilib: Regenerate.
20143 * configure.ac [target=avr]: Check whether avrxmega3 default
20144 linker description file works as needed.
20145 * configure: Regenerate.
20146 * doc/avr-mmcu.texi: Regenerate.
20147 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
20148 <__AVR_ARCH__>: Document avrxmega3 and 103.
20149 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
20150 <__AVR_SHORT_CALLS__>: Document it.
20151 <__AVR_PM_BASE_ADDRESS__>: Document it.
20152 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
20153 (AVR Variable Attributes) <progmem>: Document this is
20154 not needed for avrxmega3.
20155 (AVR Named Address Spaces) <__flash>: Dito.
20157 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
20159 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
20161 2017-06-12 Doug Rupp <rupp@adacore.com>
20163 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
20164 Append vxworks-stdint.h to the tm_file list.
20165 * config/vxworks-stdint.h: New file.
20167 2017-06-12 Martin Liska <mliska@suse.cz>
20169 PR tree-optimization/81041
20170 * tree-profile.c (gimple_gen_ic_func_profiler):
20171 Create an extra BB in profile-generate
20172 (gimple_gen_time_profiler): Likewise.
20174 2017-06-12 Jakub Jelinek <jakub@redhat.com>
20176 PR tree-optimization/81003
20177 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
20178 (update_range_test): Use it instead of force_gimple_operand_gsi.
20180 2017-06-12 Richard Biener <rguenther@suse.de>
20182 PR tree-optimization/81053
20183 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
20184 with backedge value not defined in loop. Simplify def stmt
20187 2017-06-11 Tom de Vries <tom@codesourcery.com>
20190 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
20192 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
20193 nvptx_cannot_force_const_mem.
20195 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20197 * opts.c (finish_options): Move test for flag_split_stack after
20198 it has been initialized.
20200 2017-06-11 Jason Merrill <jason@redhat.com>
20202 * tree.h (id_equal): New.
20203 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
20204 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
20205 instead of strcmp of IDENTIFIER_POINTER.
20207 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20209 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
20210 (mark_all_inlined_calls_cdtor): Fix formating.
20211 (inline_transform): Rescale profile before inlining.
20213 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20215 * cgraph.h (cgraph_edge::clone): Update prototype.
20216 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
20217 (cgraph_node::create_clone): Update.
20218 (cgraph_node::create_version_clone): Update.
20219 * tree-inline.c (copy_bb): Update.
20220 (expand_call_inline): Update.
20222 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
20224 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
20225 factored out from ...
20226 (rs6000_emit_prologue): ... here.
20228 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
20230 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
20231 factored out from ...
20232 (rs6000_emit_prologue): ... here.
20234 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20236 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
20238 (handle_missing_profiles): Fix computation of tp_first_run.
20239 (counts_to_freqs): Do not touch freqs when count is 0.
20241 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20243 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
20246 2017-06-10 Tom de Vries <tom@codesourcery.com>
20248 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
20249 attributes): Document signal effective target.
20251 2017-06-10 Tom de Vries <tom@codesourcery.com>
20253 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
20254 Document effective target stack_size.
20256 2017-06-09 David Malcolm <dmalcolm@redhat.com>
20258 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
20259 to the edit_context if they can be auto-applied.
20261 2017-06-9 Ian Lance Taylor <iant@golang.org>
20263 * opts.c (finish_options): If -fsplit-stack, disable implicit
20264 -forder-blocks-and-partition.
20265 * doc/invoke.texi (Optimize Options): Document that when using
20266 -fsplit-stack -forder-blocks-and-partition is not implicitly
20269 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20271 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
20272 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
20273 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
20274 * builtins.def (abort, trap, unreachable): Declare cold.
20275 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
20276 * tree-core.h (ECF_COLD): New.
20277 * tree.c (set_call_expr_flags): Handle ECF_COLD.
20278 (build_common_builtin_nodes): Mark unreachable and abort as cold.
20280 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20282 * predict.c (unlikely_executed_stmt_p): Cleanup.
20284 2017-06-09 Richard Biener <rguenther@suse.de>
20286 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
20287 model if the ref is always written to.
20289 2017-06-09 Tamar Christina <tamar.christina@arm.com>
20291 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
20293 2017-06-09 Tamar Christina <tamar.christina@arm.com>
20295 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
20298 2017-06-09 Tom de Vries <tom@codesourcery.com>
20301 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
20302 "target cannot support label values" when encountering LABEL_REF.
20304 2017-06-09 Martin Liska <mliska@suse.cz>
20306 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
20307 (gimple_gen_ic_func_profiler): Emit direct comparison
20308 of __gcov_indirect_call_callee with NULL.
20309 (gimple_gen_time_profiler): Change probability from
20310 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
20312 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20314 * profile.c (edge_gcov_counts): Turn to pointer.
20315 (compute_branch_probabilities, compute_branch_probabilities): Update.
20316 (branch_prob): Do not clear edge_gcov_count.
20317 * profile.h (edge_gcov_counts): Turn to pointer.
20318 (edge_gcov_count): Update.
20320 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20322 * gimple.h (gimple_check_failed): Mark cold.
20324 2017-06-09 Richard Biener <rguenther@suse.de>
20326 PR tree-optimization/66623
20327 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
20328 refactor check_reduction into two parts, properly computing
20329 whether we have to check reduction validity for outer loop
20332 2017-06-09 Richard Biener <rguenther@suse.de>
20334 PR tree-optimization/79483
20335 * graphite-scop-detection.c (order): New global.
20336 (get_order): Compute bb to order mapping that satisfies code
20337 generation constraints.
20338 (cmp_pbbs): New helper.
20339 (build_scops): Start domwalk at entry block, sort generated
20342 2017-06-09 Richard Biener <rguenther@suse.de>
20344 PR middle-end/81007
20345 * ipa-polymorphic-call.c
20346 (ipa_polymorphic_call_context::restrict_to_inner_class):
20347 Skip FIELD_DECLs with error_mark_node type.
20348 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
20351 2017-06-09 Martin Liska <mliska@suse.cz>
20353 * predict.c (struct branch_predictor): New struct.
20354 (test_prediction_value_range): New test.
20355 (predict_c_tests): New function.
20356 * selftest-run-tests.c (selftest::run_tests): Run the function.
20357 * selftest.h: Declare new tests.
20359 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
20362 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
20363 gen_add3_insn did not fail.
20364 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
20365 r0, construct that number in a temporary reg and add that reg to r0.
20366 If asked to put the result in r0 as well, fail.
20368 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
20370 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
20371 for early expansion of vec_eqv.
20373 2017-06-08 Jakub Jelinek <jakub@redhat.com>
20375 PR middle-end/81005
20376 * ubsan.c (instrument_null): Avoid pointless code temporary.
20377 (pass_ubsan::execute): Instrument aggregate arguments of calls.
20379 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
20383 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
20386 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
20389 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20391 * predict.c (unlikely_executed_edge_p): Move ahead.
20392 (probably_never_executed_edge_p): Use it.
20394 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
20396 PR middle-end/79988
20397 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
20398 gimple_call_builtin_p call.
20400 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20402 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
20403 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
20404 rtl_check_failed_type2, rtl_check_failed_code1,
20405 rtl_check_failed_code2, rtl_check_failed_code_mode,
20406 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
20407 rtvec_check_failed_bounds, rtl_check_failed_flag,
20408 _fatal_insn_not_found, _fatal_insn): Likewise.
20409 * tree.h (tree_contains_struct_check_failed,
20410 tree_check_failed, tree_not_check_failed,
20411 tree_class_check_failed, tree_range_check_failed,
20412 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
20413 tree_vec_elt_check_failed, phi_node_elt_check_failed,
20414 tree_operand_check_failed, omp_clause_check_failed,
20415 omp_clause_operand_check_failed, omp_clause_range_check_failed):
20418 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20420 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
20421 flag_branch_probabilities.
20422 * ipa-inline.c (edge_badness): Likewise.
20423 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
20424 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
20425 * predict.c (maybe_hot_frequency_p): Likewise.
20426 (probably_never_executed): Likewise.
20427 * sched-ebb.c (schedule_ebbs): Likewise.
20428 * sched-rgn.c (find_single_block_region): Likewise.
20429 * tracer.c (tail_duplicate): Likewise.
20431 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20433 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
20434 longer requires x_flag_profile_use.
20436 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20438 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
20439 instead of flag_reorder_blocks_and_partition.
20440 * dbxout.c (dbxout_function_end): Likewise.
20441 * dwarf2out.c (gen_subprogram_die): Likewise.
20442 * haifa-sched.c (sched_create_recovery_edges): Likewise.
20443 * hw-doloop.c (reorg_loops): Likewise.
20444 * varasm.c (assemble_start_function,
20445 assemble_end_function): Likewise.
20446 (decide_function_section): Do not check for
20447 flag_reorder_blocks_and_partition.
20449 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
20451 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
20453 (chkp_get_hard_register_fake_addr_expr): Ditto.
20454 (chkp_build_addr_expr): Add check for hard reg case.
20455 (chkp_parse_array_and_component_ref): Ditto.
20456 (chkp_find_bounds_1): Ditto.
20457 (chkp_process_stmt): Don't generate bounds store for
20460 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20462 * predict.c (maybe_hot_bb_p): Do not check profile status.
20463 (maybe_hot_edge_p): Likewise.
20464 (probably_never_executed): Check for zero counts even if profile
20466 (unlikely_executed_edge_p): New function.
20467 (unlikely_executed_stmt_p): New function.
20468 (unlikely_executed_bb_p): New function.
20469 (set_even_probabilities): Use unlikely predicates.
20470 (combine_predictions_for_bb): Likewise.
20471 (predict_paths_for_bb): Likewise.
20472 (predict_paths_leading_to_edge): Likewise.
20473 (determine_unlikely_bbs): New function.
20474 (estimate_bb_frequencies): Use it.
20475 (compute_function_frequency): Use zero counts even if profile is
20477 * profile-count.h: Fix typo.
20479 2017-08-08 Julia Koval <julia.koval@intel.com>
20481 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
20482 _mm512_mask_cvtsepi16_storeu_epi8,
20483 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
20484 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
20485 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
20486 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
20487 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
20488 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
20489 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
20490 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
20491 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
20492 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
20493 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
20494 __builtin_ia32_pmovuswb256mem_mask,
20495 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
20496 __builtin_ia32_pmovwb512mem_mask): New builtins.
20498 2017-08-08 Julia Koval <julia.koval@intel.com>
20500 PR target/73350,80862
20501 * config/i386/subst.md (round): Fix round pattern.
20502 * config/i386/i386.c (ix86_erase_embedded_rounding):
20503 Fix erasing rounding for the fixed pattern.
20505 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20507 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
20509 2017-06-08 Martin Liska <mliska@suse.cz>
20511 PR gcov-profile/80911
20512 * gcov.c (block_info::block_info): New constructor.
20514 2017-06-07 Carl Love <cel@us.ibm.com>
20516 * config/rs6000/rs6000-c: The return type of the following
20517 built-in functions was implemented as int not long long. Fix sign
20518 of return value for the unsigned version of vec_mulo and vec_mule.
20519 vector unsigned long long vec_bperm (vector unsigned long long,
20520 vector unsigned char)
20521 vector signed long long vec_mule (vector signed int,
20523 vector unsigned long long vec_mule (vector unsigned int,
20524 vector unsigned int)
20525 vector signed long long vec_mulo (vector signed int,
20527 vector unsigned long long vec_mulo (vector unsigned int,
20528 vector unsigned int)
20529 * doc/extend.texi: Fix the documentation for the built-in
20532 2017-06-07 Carl Love <cel@us.ibm.com>
20535 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
20538 2017-06-07 Carl Love <cel@us.ibm.com>
20540 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
20541 support, Generate doublehv for signed int/float for BE case only.
20543 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
20545 * doc/invoke.texi (mcx16): Rewrite.
20547 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20549 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
20550 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
20551 *mov<mode>_softfloat, and an anonymous splitter): Use
20552 nonimmediate_operand instead of rs6000_nonimmediate_operand.
20554 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20556 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
20558 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
20559 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
20560 (rs6000_debug_reg_global): Adjust.
20561 (rs6000_init_hard_regno_mode_ok): Adjust.
20562 (rs6000_dbx_register_number): Adjust.
20563 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
20564 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
20565 Remove SPE_ACC and SPEFSCR.
20566 (REG_ALLOC_ORDER): Ditto.
20567 (FRAME_POINTER_REGNUM): Change to 111.
20568 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
20569 (REG_CLASS_NAMES): Ditto.
20570 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
20571 (REGISTER_NAMES): Ditto.
20572 (ADDITIONAL_REG_NAMES): Ditto.
20573 (rs6000_reg_names): Ditto.
20574 * config/rs6000/rs6000.md: Renumber some register number
20577 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20579 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
20581 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
20582 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
20584 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
20585 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
20586 Delete the SPE high registers.
20587 (REG_ALLOC_ORDER): Ditto.
20588 (enum reg_class): Remove SPE_HIGH_REGS.
20589 (REG_CLASS_NAMES): Ditto.
20590 (REG_CLASS_CONTENTS): Delete the SPE high registers.
20591 (REGISTER_NAMES): Ditto.
20592 (rs6000_reg_names): Ditto.
20593 * doc/tm.texi.in: Remove SPE as example.
20594 * doc/tm.texi: Regenerate.
20596 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20598 * config/rs6000/8540.md (ppc8540_brinc): Delete.
20599 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
20600 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
20601 * config/rs6000/rs6000.md (type): Remove "brinc".
20603 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20605 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
20606 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
20607 * config/rs6000/linuxspe.h: Delete file.
20608 * config/rs6000/rs6000.md: Don't include spe.md.
20609 * config/rs6000/spe.h: Delete file.
20610 * config/rs6000/spe.md: Delete file.
20611 * config/rs6000/t-rs6000: Remove spe.md.
20613 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20615 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
20616 (reg_or_none500mem_operand): Delete.
20617 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
20618 instead of reg_or_none500mem_operand.
20620 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20622 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
20623 handling of SPE flags.
20624 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
20626 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20628 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
20630 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
20631 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
20632 paired_divv2sf3): Similar.
20633 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
20634 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
20635 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
20637 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
20638 Rename the paired_* instruction patterns.
20639 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
20641 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
20642 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
20643 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
20644 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
20645 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
20646 PAIRED_VECTOR_MODE.
20647 (struct machine_function): Delete field spe_insn_chain_scanned_p.
20648 (spe_func_has_64bit_regs_p): Delete.
20649 (spe_expand_predicate_builtin): Delete.
20650 (spe_expand_evsel_builtin): Delete.
20651 (TARGET_DWARF_REGISTER_SPAN): Do not define.
20652 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
20653 (invalid_e500_subreg): Delete.
20654 (rs6000_legitimize_address): Always force_reg op2 as well, for
20655 paired single memory accesses.
20656 (rs6000_member_type_forces_blk): Delete.
20657 (rs6000_spe_function_arg): Delete.
20658 (rs6000_expand_unop_builtin): Delete SPE handling.
20659 (rs6000_expand_binop_builtin): Ditto.
20660 (spe_expand_stv_builtin): Delete.
20661 (bdesc_2arg_spe): Delete.
20662 (spe_expand_builtin): Delete.
20663 (spe_expand_predicate_builtin): Delete.
20664 (spe_expand_evsel_builtin): Delete.
20665 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
20666 (spe_init_builtins): Delete.
20667 (spe_func_has_64bit_regs_p): Delete.
20668 (savres_routine_name): Delete "info" parameter. Adjust callers.
20669 (rs6000_emit_stack_reset): Ditto.
20670 (rs6000_dwarf_register_span): Delete.
20671 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
20672 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
20673 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
20675 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
20677 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
20678 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
20679 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
20680 mulv2sf3, divv2sf3): Delete expanders.
20682 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20684 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
20686 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20688 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
20689 * config/rs6000/rs6000.c: Ditto.
20691 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20693 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
20694 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
20695 comparison_operator.
20697 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20699 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
20700 * config/rs6000/rs6000.opt: Ditto.
20701 * config/rs6000/t-rtems: Ditto.
20703 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20705 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
20706 TARGET_E500_SINGLE by 0, simplify.
20707 * config/rs6000/rs6000.c: Ditto.
20708 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
20709 (spe_build_register_parallel): Delete.
20710 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
20711 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
20712 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
20713 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
20714 (E500_CONVERT): Delete.
20715 * config/rs6000/spe.md: Remove many patterns and all define_constants.
20717 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20719 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
20720 * config/rs6000/dfp.md: Ditto.
20721 (negdd2, *negdd2_fpr): Merge.
20722 (absdd2, *absdd2_fpr): Merge.
20723 (negtd2, *negtd2_fpr): Merge.
20724 (abstd2, *abstd2_fpr): Merge.
20725 * config/rs6000/e500.h: Delete file.
20726 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
20727 TARGET_FPRS by 1 and simplify.
20728 * config/rs6000/rs6000-c.c: Ditto.
20729 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
20730 TARGET_DF_SPE by 0.
20731 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
20733 * config/rs6000/rs6000.md: Ditto.
20734 (floatdidf2, *floatdidf2_fpr): Merge.
20735 (move_from_CR_gt_bit): Delete.
20736 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
20737 (E500_CR_IOR_COMPARE): Delete.
20738 (All patterns that require !TARGET_FPRS): Delete.
20739 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
20741 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20743 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
20745 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20747 * graphds.c (add_edge): Intitialize edge's attached data.
20748 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
20749 pointer parameter. Call pointed function on each edge during
20750 graph traversing. Skip traversing the edge when the function
20752 (graphds_dfs, graphds_scc): Ditto.
20753 (for_each_edge): New parameter. Pass the new parameter to callback
20755 * graphds.h (skip_edge_callback): New function pointer type.
20756 (graphds_dfs, graphds_scc): New function pointer parameter.
20757 (graphds_edge_callback, for_each_edge): New parameter.
20759 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20761 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
20762 out code checking if runtime alias check is possible to below ...
20763 Call the new function.
20764 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
20765 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
20767 2017-06-07 Marek Polacek <polacek@redhat.com>
20770 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
20771 TYPE_OVERFLOW_WRAPS check.
20773 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20775 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
20776 if versioning is required.
20777 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
20778 peeling with the check for versioning.
20780 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20782 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
20783 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
20784 Set true to new parameter if new ssa variable is defined.
20785 (vect_gen_vector_loop_niters): Refactor. Set range information
20786 for the new vector loop bound variable.
20787 (vect_do_peeling): Ditto.
20789 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20791 * tree-affine.c (ssa.h): Include header file.
20792 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
20793 has wrapping overflow behavior.
20795 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20797 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
20799 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20801 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
20802 (tree_to_aff_combination): ... here.
20804 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20806 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
20807 reg_pressure model function.
20808 (ivopts_global_cost_for_size): Delete.
20809 (determine_set_costs, iv_ca_recount_cost): Call new model function
20810 ivopts_estimate_reg_pressure.
20812 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20814 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
20815 expensive than udiv. Remove floating point cases from mod.
20817 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20819 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
20820 Increase idiv cost.
20822 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20824 * config/aarch64/aarch64.md
20825 (copysignsf3): Fix mask generation.
20827 2017-06-07 Jakub Jelinek <jakub@redhat.com>
20829 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
20831 (class dump_manager): Add register_dumps method.
20832 * dumpfile.c: Include langhooks.h.
20833 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
20834 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
20835 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
20836 (dump_manager::dump_register): Start with 512 entries instead of 32.
20837 (dump_manager::register_dumps): New method.
20838 * toplev.c (general_init): Instead of invoking register_dumps
20839 langhook, invoke register_dumps method on the dump manager.
20840 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
20843 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
20845 * doc/md.texi: Clarify the restrictions on a define_insn condition.
20846 Say that # requires an associated define_split to exist, and that
20847 the define_split must be suitable for use after register allocation.
20849 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20851 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
20852 (compute_outgoing_frequencies): Also initialize zero counts.
20853 (find_many_sub_basic_blocks): Do not produce uninitialized profile
20854 around loops; preserve more of profile when nothing changes.
20856 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
20858 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
20860 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
20861 * config/arm/arm-cpu-cdata.h: Regenerate.
20862 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
20863 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
20864 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
20865 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
20866 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
20868 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
20870 * config/arm/t-rmprofile: Likewise.
20871 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
20873 2017-06-06 David S. Miller <davem@davemloft.net>
20876 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
20877 blockage if function uses alloca.
20879 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20881 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
20882 New "uid" fields to hold pretty-print IDs of group and ref.
20883 Memory references are now identified as <group_id>:<ref_id>
20884 instead of using [random] addresses.
20885 (dump_mem_details): Simplify, no functional change.
20886 (dump_mem_ref): Simplify and make output more concise.
20887 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
20888 (find_or_create_group): Initialize group uid.
20889 (record_ref): Initialize ref uid. Improve debug output.
20890 (prune_group_by_reuse, should_issue_prefetch_p,)
20891 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
20892 (mark_nontemporal_store, determine_loop_nest_reuse):
20893 Improve debug output.
20895 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20897 * dbgcnt.def (prefetch): New debug counter.
20898 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
20899 (schedule_prefetches): Stop issueing prefetches if debug counter
20902 2017-06-06 Tom de Vries <tom@codesourcery.com>
20904 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
20905 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
20907 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20909 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
20910 Use aarch64_reg_or_zero predicate for operand 4.
20911 (aarch64_compare_and_swap<mode> define_insn_and_split):
20912 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
20913 (aarch64_store_exclusive<mode>): Likewise for operand 2.
20915 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
20917 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
20918 (arm_compute_save_core_reg_mask): This.
20919 (thumb1_compute_save_reg_mask): Rename into ...
20920 (thumb1_compute_save_core_reg_mask): This.
20921 (arm_compute_save_reg0_reg12_mask): Adapt comment.
20922 (arm_compute_frame_layout): Likewise.
20924 2017-06-06 Richard Biener <rguenther@suse.de>
20926 PR tree-optimization/80974
20927 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
20928 keep or clear leaders SSA info.
20930 2017-06-06 Tom de Vries <tom@codesourcery.com>
20932 * config/nvptx/nvptx.c (split_mode_p): New function.
20933 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
20935 2017-06-06 Tom de Vries <tom@codesourcery.com>
20937 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
20939 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20942 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
20945 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20947 * shrink-wrap.c (handle_simple_exit): Update profile.
20948 (try_shrink_wrapping): Upate profile.
20950 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20952 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
20953 (tree_guess_outgoing_edge_probabilities): New.
20954 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
20955 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
20957 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20959 * ipa-split.c (split_function): Initialize return bb profile.
20961 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20963 * profile.c (compute_branch_probabilities): Also initialize
20964 EXIT_BLOCK profile.
20966 2017-06-06 Richard Biener <rguenther@suse.de>
20968 PR tree-optimization/80928
20969 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
20970 (vect_analyze_loop_operations): Properly guard analysis for
20972 (vect_transform_loop): Likewise.
20973 (vect_analyze_loop_2): Also reset SLP type on PHIs.
20974 (vect_model_induction_cost): Do not cost for pure SLP.
20975 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
20976 of induction in inner loop vectorization.
20977 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
20978 (vect_get_and_check_slp_defs): Handle vect_induction_def.
20979 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
20981 (vect_analyze_slp_cost_1): Cost induction.
20982 (vect_detect_hybrid_slp_stmts): Handle PHIs.
20983 (vect_get_slp_vect_defs): Likewise.
20984 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
20985 (vect_transform_stmt): Handle SLP reductions.
20986 * tree-vectorizer.h (vectorizable_induction): Adjust.
20988 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
20990 * config/rs6000/rs6000.c (make_resolver_func): Update
20991 init_lowered_empty_function call.
20993 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
20995 * doc/invoke.texi: Document the -fprofile-abs-path option.
20996 * common.opt (fprofile-abs-path): New option.
20997 * gcov-io.h (gcov_write_filename): Declare.
20998 * gcov-io.c (gcov_write_filename): New function.
20999 * coverage.c (coverage_begin_function): Use gcov_write_filename.
21000 * profile.c (output_location): Likewise.
21002 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21004 * shring-wrap.c: Revert accidental commit.
21006 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
21008 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
21010 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21012 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
21014 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
21015 profile in callgraph edge.
21016 * profile-count.h (apply_probability): If THIS is 0, then result is 0
21017 (apply_scale): Likewise.
21018 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
21019 Also scale profile when inlining function with zero profile.
21020 (initialize_cfun): Update exit block profile even when it is zero.
21021 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
21022 when profile is read.
21024 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21026 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
21027 (CLONE_*): New constants to define the processors we can generate
21028 code for with the target_clone attribute.
21029 (rs6000_clone_map): New array to identify which clone processors
21030 the current program is running on.
21031 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
21032 target_clone attribute.
21033 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
21034 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
21035 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
21036 (cpu_expand_builtin): Add support for target_clone attribute.
21037 (rs6000_valid_attribute_p): Allow "default" attribute.
21038 (get_decl_name): New debug function to simplify printing the
21039 current function name in debugging statements.
21040 (rs6000_clone_priority): New functions to support the target_clone
21041 attribute, and be able to generate code to switch between ISA 2.05
21042 through ISA 3.0 (power6 through power9).
21043 (rs6000_compare_version_priority): Likewise.
21044 (rs6000_get_function_versions_dispatcher): Likewise.
21045 (make_resolver_func): Likewise.
21046 (add_condition_to_bb): Likewise.
21047 (dispatch_function_versions): Likewise.
21048 (rs6000_generate_version_dispatcher_body): Likewise.
21049 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
21050 (fusion_gpr_load_p): Fix a spacing issue.
21051 * doc/extend.texi (Common Function Attributes): Document that the
21052 PowerPC supports the target_clone attribute.
21054 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
21056 * config/arm/arm.h: explain F symbol found in description of ARM
21057 register allocation in its legend.
21059 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21061 * config/mips/frame-header-opt.c: Include profile-count.h.
21062 * config/riscv/riscv.c: Include profile-count.h
21064 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21066 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
21068 (sm_set_flag_if_changed): Add bbs field.
21069 (execute_sm_if_changed_flag_set): Pass BBS.
21070 (execute_sm): Update.
21072 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21074 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
21077 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21079 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
21080 (peephole2): New peephole2 to emit the above.
21081 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
21083 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21085 * config/aarch64/aarch64.c (define_peephole2 above
21086 *sub_<shift>_<mode>): New peephole.
21088 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21090 * config/i386/i386.c (make_resolver_func): Update.
21091 * Makefile.in: Add profile-count.h and profile-count.o
21092 * auto-profile.c (afdo_indirect_call): Update to new API.
21093 (afdo_set_bb_count): Update.
21094 (afdo_propagate_edge): Update.
21095 (afdo_propagate_circuit): Update.
21096 (afdo_calculate_branch_prob): Update.
21097 (afdo_annotate_cfg): Update.
21098 * basic-block.h: Include profile-count.h
21099 (struct edge_def): Turn count to profile_count.
21100 (struct basic_block_def): Likewie.
21101 (REG_BR_PROB_BASE): Move to profile-count.h
21102 (RDIV): Move to profile-count.h
21103 * bb-reorder.c (max_entry_count): Turn to profile_count.
21104 (find_traces): Update.
21105 (rotate_loop):Update.
21106 (connect_traces):Update.
21107 (sanitize_hot_paths):Update.
21108 * bt-load.c (migrate_btr_defs): Update.
21109 * cfg.c (RDIV): Remove.
21110 (init_flow): Use alloc_block.
21111 (alloc_block): Uninitialize count.
21112 (unchecked_make_edge): Uninitialize count.
21113 (check_bb_profile): Update.
21114 (dump_edge_info): Update.
21115 (dump_bb_info): Update.
21116 (update_bb_profile_for_threading): Update.
21117 (scale_bbs_frequencies_int): Update.
21118 (scale_bbs_frequencies_gcov_type): Update.
21119 (scale_bbs_frequencies_profile_count): New.
21120 * cfg.h (update_bb_profile_for_threading): Update.
21121 (scale_bbs_frequencies_profile_count): Declare.
21122 * cfgbuild.c (compute_outgoing_frequencies): Update.
21123 (find_many_sub_basic_blocks): Update.
21124 * cfgcleanup.c (try_forward_edges): Update.
21125 (try_crossjump_to_edge): Update.
21126 * cfgexpand.c (expand_gimple_tailcall): Update.
21127 (construct_exit_block): Update.
21128 * cfghooks.c (verify_flow_info): Update.
21129 (dump_bb_for_graph): Update.
21130 (split_edge): Update.
21131 (make_forwarder_block): Update.
21132 (duplicate_block): Update.
21133 (account_profile_record): Update.
21134 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
21135 (get_estimated_loop_iterations): Update.
21136 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
21137 (single_likely_exit): Update.
21138 * cfgloopmanip.c (scale_loop_profile): Update.
21140 (set_zero_probability): Update.
21141 (lv_adjust_loop_entry_edge): Update.
21142 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
21143 (purge_dead_edges): Update.
21144 (rtl_account_profile_record): Update.
21145 * cgraph.c (cgraph_node::create): Uninitialize count.
21146 (symbol_table::create_edge): Uninitialize count.
21147 (cgraph_update_edges_for_call_stmt_node): Update.
21148 (cgraph_edge::dump_edge_flags): Update.
21149 (cgraph_node::dump): Update.
21150 (cgraph_edge::maybe_hot_p): Update.
21151 * cgraph.h: Include profile-count.h
21152 (create_clone), create_edge, create_indirect_edge): Update.
21153 (cgraph_node): Turn count to profile_count.
21154 (cgraph_edge0: Likewise.
21155 (make_speculative, clone): Update.
21156 (create_edge): Update.
21157 (init_lowered_empty_function): Update.
21158 * cgraphclones.c (cgraph_edge::clone): Update.
21159 (duplicate_thunk_for_node): Update.
21160 (cgraph_node::create_clone): Update.
21161 * cgraphunit.c (cgraph_node::analyze): Update.
21162 (cgraph_node::expand_thunk): Update.
21163 * final.c (dump_basic_block_info): Update.
21164 * gimple-streamer-in.c (input_bb): Update.
21165 * gimple-streamer-out.c (output_bb): Update.
21166 * graphite.c (print_global_statistics): Update.
21167 (print_graphite_scop_statistics): Update.
21168 * hsa-brig.c: Include basic-block.h.
21169 * hsa-dump.c: Include basic-block.h.
21170 * hsa-gen.c (T sum_slice): Update.
21171 (convert_switch_statements):Update.
21172 * hsa-regalloc.c: Include basic-block.h.
21173 * ipa-chkp.c (chkp_produce_thunks): Update.
21174 * ipa-cp.c (struct caller_statistics): Update.
21175 (init_caller_stats): Update.
21176 (gather_caller_stats): Update.
21177 (ipcp_cloning_candidate_p): Update.
21178 (good_cloning_opportunity_p): Update.
21179 (get_info_about_necessary_edges): Update.
21180 (dump_profile_updates): Update.
21181 (update_profiling_info): Update.
21182 (update_specialized_profile): Update.
21183 (perhaps_add_new_callers): Update.
21184 (decide_about_value): Update.
21185 (ipa_cp_c_finalize): Update.
21186 * ipa-devirt.c (struct odr_type_warn_count): Update.
21187 (struct decl_warn_count): Update.
21188 (struct final_warning_record): Update.
21189 (possible_polymorphic_call_targets): Update.
21190 (ipa_devirt): Update.
21191 * ipa-fnsummary.c (redirect_to_unreachable): Update.
21192 * ipa-icf.c (sem_function::merge): Update.
21193 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
21194 * ipa-inline.c (compute_uninlined_call_time): Update.
21195 (compute_inlined_call_time): Update.
21196 (want_inline_small_function_p): Update.
21197 (want_inline_self_recursive_call_p): Update.
21198 (edge_badness): Update.
21199 (lookup_recursive_calls): Update.
21200 (recursive_inlining): Update.
21201 (inline_small_functions): Update.
21202 (dump_overall_stats): Update.
21203 (dump_inline_stats): Update.
21204 * ipa-profile.c (ipa_profile_generate_summary): Update.
21205 (ipa_propagate_frequency): Update.
21206 (ipa_profile): Update.
21207 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
21208 * ipa-utils.c (ipa_merge_profiles): Update.
21209 * loop-doloop.c (doloop_modify): Update.
21210 * loop-unroll.c (report_unroll): Update.
21211 (unroll_loop_runtime_iterations): Update.
21212 * lto-cgraph.c (lto_output_edge): Update.
21213 (lto_output_node): Update.
21214 (input_node): Update.
21215 (input_edge): Update.
21216 (merge_profile_summaries): Update.
21217 * lto-streamer-in.c (input_cfg): Update.
21218 * lto-streamer-out.c (output_cfg): Update.
21219 * mcf.c (create_fixup_graph): Update.
21220 (adjust_cfg_counts): Update.
21221 (sum_edge_counts): Update.
21222 * modulo-sched.c (sms_schedule): Update.
21223 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
21224 * predict.c (maybe_hot_count_p): Update.
21225 (probably_never_executed): Update.
21226 (dump_prediction): Update.
21227 (combine_predictions_for_bb): Update.
21228 (propagate_freq): Update.
21229 (handle_missing_profiles): Update.
21230 (counts_to_freqs): Update.
21231 (rebuild_frequencies): Update.
21232 (force_edge_cold): Update.
21233 * predict.h: Include profile-count.h
21234 (maybe_hot_count_p, counts_to_freqs): UPdate.
21235 * print-rtl-function.c: Do not include cfg.h
21236 * print-rtl.c: Include basic-block.h
21237 * profile-count.c: New file.
21238 * profile-count.h: New file.
21239 * profile.c (is_edge_inconsistent): Update.
21240 (correct_negative_edge_counts): Update.
21241 (is_inconsistent): Update.
21242 (set_bb_counts): Update.
21243 (read_profile_edge_counts): Update.
21244 (compute_frequency_overlap): Update.
21245 (compute_branch_probabilities): Update; Initialize and deinitialize
21247 (branch_prob): Update.
21248 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
21249 (edge_gcov_count): New.
21250 (bb_gcov_count): New.
21251 * shrink-wrap.c (try_shrink_wrapping): Update.
21252 * tracer.c (better_p): Update.
21253 * trans-mem.c (expand_transaction): Update.
21254 (ipa_tm_insert_irr_call): Update.
21255 (ipa_tm_insert_gettmclone_call): Update.
21256 * tree-call-cdce.c: Update.
21257 * tree-cfg.c (gimple_duplicate_sese_region): Update.
21258 (gimple_duplicate_sese_tail): Update.
21259 (gimple_account_profile_record): Update.
21260 (execute_fixup_cfg): Update.
21261 * tree-inline.c (copy_bb): Update.
21262 (copy_edges_for_bb): Update.
21263 (initialize_cfun): Update.
21264 (freqs_to_counts): Update.
21265 (copy_cfg_body): Update.
21266 (expand_call_inline): Update.
21267 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
21268 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
21269 (try_unroll_loop_completely): Update.
21270 (try_peel_loop): Update.
21271 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
21272 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
21273 * tree-ssa-loop-split.c (connect_loops): Update.
21274 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
21275 * tree-ssa-reassoc.c (branch_fixup): Update.
21276 * tree-ssa-tail-merge.c (replace_block_by): Update.
21277 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
21278 (compute_path_counts): Update.
21279 (update_profile): Update.
21280 (recompute_probabilities): Update.
21281 (update_joiner_offpath_counts): Update.
21282 (estimated_freqs_path): Update.
21283 (freqs_to_counts_path): Update.
21284 (clear_counts_path): Update.
21285 (ssa_fix_duplicate_block_edges): Update.
21286 (duplicate_thread_path): Update.
21287 * tree-switch-conversion.c (case_bit_test_cmp): Update.
21288 (struct switch_conv_info): Update.
21289 * tree-tailcall.c (decrease_profile): Update.
21290 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
21291 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
21292 * value-prof.c (check_counter): Update.
21293 (gimple_divmod_fixed_value): Update.
21294 (gimple_mod_pow2): Update.
21295 (gimple_mod_subtract): Update.
21296 (gimple_ic_transform): Update.
21297 (gimple_stringop_fixed_value): Update.
21298 * value-prof.h (gimple_ic): Update.
21300 2017-06-02 Carl Love <cel@us.ibm.com>
21302 * config/rs6000/rs6000-c: Add support for built-in functions
21303 vector double vec_doublee (vector signed int);
21304 vector double vec_doublee (vector unsigned int);
21305 vector double vec_doublee (vector float);
21306 vector double vec_doubleh (vector signed int);
21307 vector double vec_doubleh (vector unsigned int);
21308 vector double vec_doubleh (vector float);
21309 vector double vec_doublel (vector signed int);
21310 vector double vec_doublel (vector unsigned int);
21311 vector double vec_doublel (vector float);
21312 vector double vec_doubleo (vector signed int);
21313 vector double vec_doubleo (vector unsigned int);
21314 vector double vec_doubleo (vector float);.
21315 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
21316 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
21318 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
21319 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
21320 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
21322 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
21323 vec_doublel, vec_doubleh.
21324 * doc/extend.texi: Update the built-in documentation file for the
21325 new built-in functions.
21327 2017-06-02 David Malcolm <dmalcolm@redhat.com>
21330 * ipa-inline-analysis.c (free_growth_caches): Set
21331 edge_removal_hook_holder to NULL after removing it.
21333 2017-06-02 Sudakshina Das <sudi.das@arm.com>
21335 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
21336 comparision with zero.
21338 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
21339 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21340 for early expansion of vec_min and vec_max builtins.
21341 (builtin_function_type): Add min/max unsigned variants to those
21342 identified as having unsigned arguments.
21344 2017-06-02 Olivier Hainque <hainque@adacore.com>
21346 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
21348 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21350 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
21351 Use VALL_F16 iterator rather than VALL.
21353 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21355 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
21356 Emit CBNZ inside loop when doing a strong exchange and comparing
21357 against zero. Generate the CC flags after the loop.
21359 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
21361 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
21362 (dl_section_ref): New.
21363 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
21364 On AIX, append an expression to subtract the size of the
21365 section length to dl_section_ref.
21367 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
21369 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21370 for early expansion of vector absolute builtins.
21372 2017-06-02 Richard Biener <rguenther@suse.de>
21374 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
21375 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
21377 2017-06-02 Richard Biener <rguenther@suse.de>
21379 PR tree-optimization/80948
21380 * tree-tailcall.c (find_tail_calls): Track stmts to move in
21381 stmt order as well.
21383 2017-06-02 Richard Biener <rguenther@suse.de>
21385 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
21387 * tree-vect-stmts.c (process_use): Do not mark backedge defs
21388 for inductions as relevant.
21390 2017-06-02 Richard Biener <rguenther@suse.de>
21392 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
21393 (vectorizable_induction): ... this. Remove dead code.
21395 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
21397 * builtins. (expand_builtin_alloca): Remove second parameter and
21398 infer its value from the first parameter instead.
21399 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
21401 2017-06-02 Jakub Jelinek <jakub@redhat.com>
21403 PR rtl-optimization/80903
21404 * loop-doloop.c (add_test): Unshare sequence.
21406 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
21408 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
21410 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
21412 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
21414 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
21415 xlogue_layout::get_instance, logue_layout::xlogue_layout,
21416 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
21417 (xlogue_layout::get_stub_rtx): Make static.
21418 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
21419 (xlogue_layout::compute_stub_managed_regs): Rename to...
21420 (xlogue_layout::count_stub_managed_regs): ...this.
21421 (xlogue_layout::is_stub_managed_reg): New function.
21422 (xlogue_layout::m_stub_names): Rename to...
21423 (xlogue_layout::s_stub_names): ...this, make static.
21424 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
21425 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
21426 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
21427 xlogue_layout::s_stub_names): Instantiate statics.
21428 (stub_managed_regs): Remove.
21429 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
21430 (disable_call_ms2sysv_xlogues): Rename to...
21431 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
21432 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
21434 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
21435 change after reload_completed.
21436 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
21438 (ix86_expand_prologue): Likewise.
21439 (ix86_expand_epilogue): Likewise.
21440 (ix86_expand_split_stack_prologue): Likewise.
21441 (ix86_compute_frame_layout): Remove frame parameter ...
21442 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
21443 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
21445 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
21446 (ix86_frame): Move from here ...
21447 * config/i386/i386.h (ix86_frame): ... to here.
21448 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
21449 complete ix86_frame data structure instead. Remove some_ld_name.
21451 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
21453 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
21454 symbols that hold a DECL_VALUE_EXPR.
21456 2017-06-01 Martin Jambor <mjambor@suse.cz>
21458 PR tree-optimization/80898
21459 * tree-sra.c (process_subtree_disqualification): Removed.
21460 (disqualify_candidate): Do not acll
21461 process_subtree_disqualification.
21462 (subtree_mark_written_and_enqueue): New function.
21463 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
21464 RHS has been disqualified and re-queue LHS if necessary. Apart
21465 from that, ignore disqualified RHS.
21467 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21469 * config/s390/s390.c (s390_emit_epilogue): Disable early return
21470 address fetch for z10 or later.
21472 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21474 * config/arc/arc.md (tst_movb): Add guard when splitting.
21476 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21478 * config/arc/arc.c (arc_can_eliminate): Test against
21479 arc_frame_pointer_needed.
21481 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21483 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
21484 to prevent store reordering.
21485 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
21486 (type): Add block type.
21487 (stack_tie): Define special instruction to be used in
21490 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21492 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
21493 constraint. It is not valid for the pattern.
21494 (noncommutative_binary_comparison): Likewise.
21496 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21498 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
21501 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21503 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
21504 be used by the reg-alloc.
21506 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21508 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
21509 reg-alloc when having mul64 or mul32x16 instructions.
21510 (mulsidi3): Likewise.
21511 (umulsidi3): Likewise.
21512 (mulsi32x16): New pattern.
21513 (mulsi64): Likewise.
21514 (mulsidi64): Likewise.
21515 (umulsidi64): Likewise.
21516 (MUL32x16_REG): Define.
21517 (mul64_600): Use MUL32x16_REG.
21518 (mac64_600): Likewise.
21519 (umul64_600): Likewise.
21520 (umac64_600): Likewise.
21522 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21524 * config/arc/arc.md (mulsi3_700): Make it commutative.
21526 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
21528 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
21530 (*sign_extendsidi2_insn): Likewise for movstosw.
21532 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
21534 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
21535 the type of the input discriminant value. Convert the
21536 discriminant value of signedness vary.
21538 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
21540 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
21541 Add to -Wall section.
21543 2017-06-01 Richard Biener <rguenther@suse.de>
21545 PR middle-end/66313
21546 * fold-const.c (fold_plusminus_mult_expr): If the factored
21547 factor may be zero use a wrapping type for the inner operation.
21548 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
21549 and handle moved defs.
21550 (process_assignment): Properly guard the unary op case. Return a
21551 tri-state indicating that moving the stmt before the call may allow
21552 to continue. Pass through to_move.
21553 (find_tail_calls): Handle moving unrelated defs before
21556 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
21559 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
21560 splitter result in the canonical way.
21562 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
21564 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
21565 also for 32bit target. Update insn attributes.
21566 (zero-extendsidi2 splitter): Allow all registers for operand 1.
21568 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
21570 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
21571 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
21572 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
21573 (_mm_maskz_min_ss): New intrinsics.
21575 2017-05-31 Martin Liska <mliska@suse.cz>
21577 * tree-vect-loop.c (vect_create_epilog_for_reduction):
21578 Change comment style to one we normally use.
21579 (vectorizable_reduction): Likewise.
21580 (vectorizable_induction): Likewise.
21581 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
21582 (vectorizable_call): Likewise.
21583 (vectorizable_simd_clone_call): Likewise.
21584 (vectorizable_conversion): Likewise.
21585 (vectorizable_assignment): Likewise.
21586 (vectorizable_shift): Likewise.
21587 (vectorizable_operation): Likewise.
21588 (vectorizable_store): Likewise.
21589 (vectorizable_load): Likewise.
21590 * tree-vectorizer.h: Likewise.
21592 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
21594 * passes.c (emergency_dump_function): New.
21595 * tree-pass.h (emergency_dump_function): Declare.
21596 * plugin.c (plugins_internal_error_function): Remove.
21597 * plugin.h (plugins_internal_error_function): Remove declaration.
21598 * toplev.c (internal_error_function): New static function. Use it...
21599 (general_init): ...here.
21601 2017-05-31 Graham Markall <graham.markall@embecosm.com>
21603 * config/arc/arc.c (arc_print_operand): Handle constant operands.
21604 (arc_rtx_costs): Add costs for new patterns.
21605 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
21606 * config/arc/predicates.md: Add _1_2_3_operand predicate.
21608 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
21610 * tree-ssa-strlen.c (get_next_strinfo): New function.
21611 (get_stridx_plus_constant): Use it.
21612 (zero_length_string): Likewise.
21613 (adjust_related_strinfos): Likewise.
21614 (adjust_last_stmt): Likewise.
21616 2017-05-31 Richard Biener <rguenther@suse.de>
21619 * config/i386/i386.c (ix86_expand_builtin): Remove assert
21620 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
21622 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
21624 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
21625 loop_vinfo argument and use of dependence distance vectors.
21626 Check instead whether the two references differ only in their
21627 initial value and assume that they have the same alignment if the
21628 difference is a multiple of the vector alignment.
21629 (vect_analyze_data_refs_alignment): Update call accordingly.
21631 2017-05-31 Martin Liska <mliska@suse.cz>
21634 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
21636 2017-05-31 Bin Cheng <bin.cheng@arm.com>
21638 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
21639 (create_intersect_range_checks): Move from ...
21640 * tree-data-ref.c (create_intersect_range_checks_index)
21641 (create_intersect_range_checks): ... to here.
21642 (create_runtime_alias_checks): New function factored from ...
21643 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
21644 here. Call above function.
21645 * tree-data-ref.h (create_runtime_alias_checks): New function.
21647 2017-05-31 Bin Cheng <bin.cheng@arm.com>
21649 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
21650 segment length for dr_b and compute it in wide_int.
21652 2017-05-31 Richard Biener <rguenther@suse.de>
21654 PR tree-optimization/80906
21655 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
21656 and pass through iv_map.
21657 (copy_bb_and_scalar_dependences): Adjust.
21658 (translate_pending_phi_nodes): Likewise.
21659 (copy_loop_close_phi_args): Handle code-generating IVs instead
21662 2017-05-30 David Malcolm <dmalcolm@redhat.com>
21664 * diagnostic-color.c (color_dict): Add "type-diff".
21665 (parse_gcc_colors): Update comment.
21666 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
21667 -fdiagnostics-show-template-tree and -fno-elide-type.
21668 (GCC_COLORS): Add type-diff to example.
21670 (-fdiagnostics-show-template-tree): New.
21671 (-fno-elide-type): New.
21672 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
21673 the pp_format_decoder callback. Call any m_format_postprocessor's
21675 (pretty_printer::pretty_printer): Initialize
21676 m_format_postprocessor.
21677 (pretty_printer::~pretty_printer): Delete any
21678 m_format_postprocessor.
21679 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
21680 (class format_postprocessor): New class.
21681 (struct pretty_printer::format_decoder): Document the new parameters.
21682 (struct pretty_printer::m_format_postprocessor): New field.
21683 * tree-diagnostic.c (default_tree_printer): Update for new
21684 bool and const char ** params.
21685 * tree-diagnostic.h (default_tree_printer): Likewise.
21687 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
21689 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
21690 (lwa_operand): Delete rs6000_gen_cell_microcode test.
21691 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
21692 rs6000_gen_cell_microcode code.
21693 (rs6000_final_prescan_insn): Delete.
21694 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
21695 "warn-cell-microcode" entries.
21696 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
21697 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
21698 throughout. Change cc_reg_not_micro_cr0_operand to
21699 cc_reg_not_cr0_operand throughout.
21700 (*extendhi<mode>2_noload): Delete.
21701 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
21702 (mwarn-cell-microcode): Delete.
21703 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
21704 -mgen-cell-microcode and -mwarn-cell-microcode.
21706 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
21709 * config/i386/constraints.md (Yd): New constraint.
21711 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
21712 and (?Yd, r) alternatives. Update insn attributes.
21713 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
21714 and (?*Yd, r) alternatives. Update insn attributes.
21715 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
21717 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
21719 * gimplify.c (gimplify_modify_expr): Don't create a
21720 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
21723 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
21725 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
21727 2017-05-30 Richard Biener <rguenther@suse.de>
21729 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
21730 and reduc_def fields.
21731 (STMT_VINFO_REDUC_TYPE): New define.
21732 (STMT_VINFO_REDUC_DEF): Likewise.
21733 (vect_force_simple_reduction): Adjust prototype.
21734 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
21735 (vect_is_simple_reduction): Remove check_reduction argument.
21736 (vect_force_simple_reduction): Adjust and set
21737 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
21738 (vectorizable_reduction): Do not re-do reduction analysis
21739 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
21740 * tree-parloops.c (gather_scalar_reductions): Adjust.
21742 2017-05-30 Richard Biener <rguenther@suse.de>
21744 PR middle-end/80901
21745 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
21748 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
21750 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
21751 Introduce unknown_misalignment parameter and remove vf.
21752 (vect_peeling_hash_get_lowest_cost):
21753 Pass unknown_misalignment parameter.
21754 (vect_enhance_data_refs_alignment):
21755 Fix unsupportable data ref treatment.
21757 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21759 * tree-vect-data-refs.c (vect_get_data_access_cost):
21760 Workaround for SLP handling.
21761 (vect_enhance_data_refs_alignment):
21762 Compute costs for doing no peeling at all, compare to the best
21763 peeling costs so far and avoid peeling if cheaper.
21765 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21767 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
21768 Return peeling info and set costs to zero for unlimited cost
21770 (vect_enhance_data_refs_alignment): Also inspect all datarefs
21771 with unknown misalignment. Compute and costs for unknown
21772 misalignment, compare them to the costs for known misalignment
21773 and choose the cheapest for peeling.
21775 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21777 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
21778 (vect_get_peeling_costs_all_drs): Create function.
21779 (vect_peeling_hash_get_lowest_cost):
21780 Use vect_get_peeling_costs_all_drs.
21781 (vect_peeling_supportable): Create function.
21782 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
21784 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21786 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
21787 DR_HAS_NEGATIVE_STEP.
21788 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
21789 (vect_enhance_data_refs_alignment): Use.
21790 (vect_duplicate_ssa_name_ptr_info): Use.
21791 * tree-vectorizer.h (dr_misalignment): Use.
21792 (known_alignment_for_access_p): Use.
21794 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
21797 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
21799 (has_section_name): New function.
21801 2017-05-30 Martin Liska <mliska@suse.cz>
21804 * auto-profile.c (get_function_decl_from_block): Fix
21807 2017-05-30 Richard Biener <rguenther@suse.de>
21809 PR middle-end/80876
21810 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
21812 2017-05-30 Martin Liska <mliska@suse.cz>
21814 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
21815 * dumpfile.h (struct dump_file_info): Remove ctors.
21817 2017-05-30 Martin Liska <mliska@suse.cz>
21819 * predict.def: Fix GNU coding style.
21821 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
21823 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
21824 Mark 'to' argument with ATTRIBUTE_UNUSED.
21826 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
21828 * config/xtensa/xtensa.c (xtensa_emit_call): Use
21829 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
21830 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
21833 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
21835 * doc/install.texi (Options specification): Restore entry of
21836 --enable-sjlj-exceptions.
21838 2017-05-27 Michael Eager <eager@eagercon.com>
21841 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
21843 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
21845 * config/microblaze/microblaze.h
21846 (FIXED_REGISTERS): Update in macro.
21847 (CALL_USED_REGISTERS): Update in macro.
21849 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
21851 * doc/install.texi: Add links to macOS binary distributions.
21853 2017-05-27 Jakub Jelinek <jakub@redhat.com>
21857 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21859 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
21861 2017-05-26 Martin Liska <mliska@suse.cz>
21863 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
21865 2017-05-26 Martin Liska <mliska@suse.cz>
21867 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
21868 always leading ';; '.
21869 (dump_bb_info): Likewise.
21870 (brief_dump_cfg): Likewise.
21871 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
21872 * dumpfile.c: Remove usage of TDF_VERBOSE.
21873 * dumpfile.h (enum dump_kind): Likewise.
21874 (dump_gimple_bb_header): Do not use TDF_COMMENT.
21875 * print-tree.c (debug_verbose): Remove.
21876 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
21877 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
21878 * tree-diagnostic.c (default_tree_printer): Replace
21879 TDF_DIAGNOSTIC with TDF_SLIM.
21881 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21883 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
21884 in parameter loop, rather than loop_vinfo.
21885 (create_intersect_range_checks): Ditto.
21886 (vect_create_cond_for_alias_checks): Update call to above functions.
21888 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21890 PR tree-optimization/80815
21891 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
21892 for merging runtime alias checks. Handle negative DR_STEPs.
21894 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21896 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
21898 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
21899 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
21900 out code pruning runtime alias checks.
21901 * tree-data-ref.c (prune_runtime_alias_test_list): New function
21902 factored out from above.
21903 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
21905 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
21907 (prune_runtime_alias_test_list): New decalaration.
21909 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21911 * tree-vect-data-refs.c (compare_tree): Rename and move ...
21912 * tree-data-ref.c (data_ref_compare_tree): ... to here.
21913 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
21914 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
21915 (operator==, comp_dr_with_seg_len_pair): Ditto.
21916 (vect_prune_runtime_alias_test_list): Ditto.
21918 2017-05-26 Martin Liska <mliska@suse.cz>
21921 * params.def: Bound partial-inlining-entry-probability param.
21923 2017-05-26 Marek Polacek <polacek@redhat.com>
21926 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
21929 2017-05-26 Richard Biener <rguenther@suse.de>
21931 PR tree-optimization/80842
21932 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
21935 2017-05-26 Richard Biener <rguenther@suse.de>
21937 PR tree-optimization/80844
21938 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
21940 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
21942 * doc/md.texi (Machine Constraints): Update x86 family
21943 machine constraints section to match 'config/i386/constraints.md'.
21945 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
21947 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
21949 2017-05-25 Nathan Sidwell <nathan@acm.org>
21951 * doc/invoke.texi (--enable-languages): Update documentation.
21953 2017-05-25 Martin Liska <mliska@suse.cz>
21955 * dumpfile.c: Add TDF_FOLDING.
21956 * dumpfile.h (enum dump_kind): Likewise.
21957 * genmatch.c (dt_simplify::gen_1): Use it.
21959 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21961 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
21963 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21965 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
21966 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
21968 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21970 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
21971 * match.pd (X == C): Rewrite it here.
21972 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
21973 with_certain_nonzero_bits2): New predicates.
21974 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
21976 2017-05-24 Nathan Sidwell <nathan@acm.org>
21978 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
21981 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
21984 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
21986 * config/powerpcspe: New port. Files are copied from the rs6000
21987 port, with "rs6000" in filenames replaced by "powerpcspe".
21989 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
21991 PR rtl-optimization/80754
21992 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
21994 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
21996 * config/sparc/sparc.md (length): Return the correct value for -mflat
21997 sibcalls to match output_sibcall.
21999 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
22003 * config/rs6000/rs6000.c (struct machine_function): Add new field
22005 (rs6000_get_separate_components): Init that field, use it.
22006 (rs6000_components_for_bb): Use the field.
22008 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
22010 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
22012 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
22014 PR middle-end/80823
22015 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
22017 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22020 * config/s390/s390.c (s390_check_qrst_address): Check incoming
22021 address against address_operand predicate.
22022 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
22024 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
22026 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
22027 parameters passed indirectly.
22029 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
22031 * config/i386/i386.md (*movdi_internal): Remove SSE4
22032 alternative 18 (?r, *v). Update insn attributes.
22033 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
22034 Update insn attributes.
22035 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
22036 Update insn attributes.
22037 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
22038 alternative 1 (r, v). Remove isa attribute.
22039 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
22040 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
22041 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
22043 2017-05-23 Tom de Vries <tom@codesourcery.com>
22045 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
22048 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22050 * cgraphunit.c (symbol_table::process_new_functions): Update.
22051 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
22052 (inline_generate_summary): Rename to ...
22053 (ipa_fn_summary_generate): ... this one.
22054 (inline_read_summary): Rename to ...
22055 (ipa_fn_summary_read): ... this one.
22056 (inline_write_summary): Rename to ...
22057 (ipa_fn_summary_write): ... this one.
22058 (inline_free_summary): Rename to ...
22059 (ipa_free_fn_summary): ... this one.
22060 (pass_data_local_fn_summary, pass_local_fn_summary,
22061 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
22062 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
22063 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
22064 make_pass_ipa_fn_summary): New.
22065 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
22066 inline_write_summary, inline_free_summary): Remove.
22067 (ipa_free_fn_summary) : New.
22068 * ipa-inline.c (ipa_inline): Update.
22069 (pass_ipa_inline): Do not generate summaries.
22070 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
22072 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
22073 and add pass_ipa_fn_summary.
22074 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
22076 (make_pass_inline_parameters): Remove.
22078 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
22080 * omp-low.c (struct omp_context): Remove "default_kind" member.
22083 * omp-offload.c (execute_oacc_device_lower): Remove the
22084 parallelism dimensions function attributes for unparallelized
22085 OpenACC kernels constructs.
22087 2017-05-23 Martin Liska <mliska@suse.cz>
22089 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
22091 (cgraph_edge::make_speculative): Likewise.
22092 (cgraph_edge::resolve_speculation): Likewise.
22093 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
22094 (cgraph_node::dump): Likewise.
22095 * cgraph.h: Likewise.
22096 * cgraphunit.c (analyze_functions): Likewise.
22097 (symbol_table::compile): Likewise.
22098 * ipa-cp.c (print_all_lattices): Likewise.
22099 (determine_versionability): Likewise.
22100 (initialize_node_lattices): Likewise.
22101 (ipcp_verify_propagated_values): Likewise.
22102 (estimate_local_effects): Likewise.
22103 (update_profiling_info): Likewise.
22104 (create_specialized_node): Likewise.
22105 (perhaps_add_new_callers): Likewise.
22106 (decide_about_value): Likewise.
22107 (decide_whether_version_node): Likewise.
22108 (identify_dead_nodes): Likewise.
22109 (ipcp_store_bits_results): Likewise.
22110 * ipa-devirt.c (dump_targets): Likewise.
22111 (ipa_devirt): Likewise.
22112 * ipa-icf.c (sem_item::dump): Likewise.
22113 (sem_function::equals): Likewise.
22114 (sem_variable::equals): Likewise.
22115 (sem_item_optimizer::read_section): Likewise.
22116 (sem_item_optimizer::execute): Likewise.
22117 (congruence_class::dump): Likewise.
22118 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
22119 (dump_inline_summary): Likewise.
22120 (estimate_node_size_and_time): Likewise.
22121 (inline_analyze_function): Likewise.
22122 * ipa-inline-transform.c (inline_call): Likewise.
22123 * ipa-inline.c (report_inline_failed_reason): Likewise.
22124 (want_early_inline_function_p): Likewise.
22125 (edge_badness): Likewise.
22126 (update_edge_key): Likewise.
22127 (inline_small_functions): Likewise.
22128 * ipa-profile.c (ipa_profile): Likewise.
22129 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
22130 (ipa_make_edge_direct_to_target): Likewise.
22131 (remove_described_reference): Likewise.
22132 (ipa_impossible_devirt_target): Likewise.
22133 (propagate_controlled_uses): Likewise.
22134 (ipa_print_node_params): Likewise.
22135 (ipcp_transform_function): Likewise.
22136 * ipa-pure-const.c (pure_const_read_summary): Likewise.
22137 (propagate_pure_const): Likewise.
22138 * ipa-reference.c (generate_summary): Likewise.
22139 (read_write_all_from_decl): Likewise.
22140 (propagate): Likewise.
22141 (ipa_reference_read_optimization_summary): Likewise.
22142 * ipa-utils.c (ipa_merge_profiles): Likewise.
22143 * ipa.c (walk_polymorphic_call_targets): Likewise.
22144 (symbol_table::remove_unreachable_nodes): Likewise.
22145 (ipa_single_use): Likewise.
22146 * passes.c (execute_todo): Likewise.
22147 * predict.c (drop_profile): Likewise.
22148 * symtab.c (symtab_node::get_dump_name): New function.
22149 (symtab_node::dump_name): Likewise.
22150 (symtab_node::dump_asm_name): Likewise.
22151 (symtab_node::dump_references): Likewise.
22152 (symtab_node::dump_referring): Likewise.
22153 (symtab_node::dump_base): Likewise.
22154 (symtab_node::debug_symtab): Likewise.
22155 * tree-sra.c (convert_callers_for_node): Likewise.
22156 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
22157 * value-prof.c (init_node_map): Likewise.
22159 2017-05-23 Martin Liska <mliska@suse.cz>
22161 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
22162 and symtab_node::debug_symtab to symbol_table::debug.
22163 * cgraphunit.c (analyze_functions): Use the renamed function.
22164 (symbol_table::compile): Likewise.
22165 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
22166 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
22167 * passes.c (execute_todo): Likewise.
22168 * symtab.c (symbol_table::dump): New function.
22169 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
22171 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22173 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
22174 that nonconst implies exec.
22176 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22178 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
22179 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
22180 (inline_edge_summary_vec): Turn into ...
22181 (ipa_call_summaries): ... this one.
22182 (redirect_to_unreachable, edge_set_predicate,
22183 evaluate_properties_for_edge, inline_summary_alloc,
22184 reset_ipa_call_summary, reset_inline_summary,
22185 inline_summary_t::duplicate): Update.
22186 (inline_edge_duplication_hook): Turn to ...
22187 (ipa_call_summary_t::duplicate): ... this one.
22188 (inline_edge_removal_hook): Turn to ...
22189 (ipa_call_summary_t::remove): ... this one.
22190 (dump_inline_edge_summary): Turn to ...
22191 (dump_ipa_call_summary): ... this one.
22192 (estimate_function_body_sizes): Update.
22193 (inline_update_callee_summaries): Update.
22194 (remap_edge_change_prob): Update.
22195 (remap_edge_summaries): Update.
22196 (inline_merge_summary): Update.
22197 (do_estimate_edge_time): Update.
22198 (inline_generate_summary): Update.
22199 (inline_read_section): Update.
22200 (inline_read_summary): Update.
22201 (inline_free_summary): Update.
22202 * ipa-inline.c (can_inline_edge_p): Update.
22203 (compute_inlined_call_time): Update.
22204 (want_inline_small_function_p): Update.
22205 (edge_badness): Update.
22206 (early_inliner): Update.
22207 * ipa-inline.h (inline_edge_summary): Turn to ...
22208 (ipa_call_summary): ... this one.
22209 (ipa_call_summary_t): New class.
22210 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
22211 (ipa_call_summaries): New.
22212 (inline_edge_summary): Remove.
22213 (estimate_edge_growth): Update.
22214 * ipa-profile.c (ipa_propagate_frequency_1): Update.
22215 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22216 * ipa-split.c (execute_split_functions): Update.
22217 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
22219 2017-05-23 Tom de Vries <tom@codesourcery.com>
22221 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
22222 attributes): Document rdrand effective target.
22224 2017-05-23 Tom de Vries <tom@codesourcery.com>
22226 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
22227 attributes): Sort alphabetically.
22229 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
22231 * config/avr/genmultilib.awk: Use gsub instead of gensub.
22233 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
22236 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
22237 V2DF/V2DI splat into two separate patterns, one that handles
22238 registers, and the other that only handles memory. Drop support
22239 for splatting from a GPR on ISA 2.07 and then splitting the
22240 splat into direct move and splat.
22241 (vsx_splat_<mode>_reg): Likewise.
22242 (vsx_splat_<mode>_mem): Likewise.
22244 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
22246 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
22248 2017-05-22 Jakub Jelinek <jakub@redhat.com>
22250 PR middle-end/80809
22251 * omp-low.c (finish_taskreg_remap): New function.
22252 (finish_taskreg_scan): If unit size of ctx->record_type
22253 is non-constant, unshare the size expression and replace
22254 decls in it with possible outer var refs.
22256 PR middle-end/80809
22257 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
22258 GOVD_SHARED rather than GOVD_PRIVATE with it.
22259 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
22260 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
22262 PR middle-end/80853
22263 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
22264 as last argument to build_outer_var_ref for pointer bases of array
22265 section reductions.
22267 2017-05-19 Martin Sebor <msebor@redhat.com>
22269 * print-tree.c (print_node): Print DECL_READ_P flag.
22271 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22273 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
22274 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
22275 * cgraph.c: Likewise.
22276 * cgraphunit.c: Likewise.
22277 * gengtype.c: Likewise.
22278 * ipa-cp.c: Likewise.
22279 * ipa-devirt.c: Likewise.
22280 * ipa-icf.c: Likewise.
22281 * ipa-predicate.c: Likewise.
22282 * ipa-profile.c: Likewise.
22283 * ipa-prop.c: Likewise.
22284 * ipa-split.c: Likewise.
22286 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
22287 edge_predicate_pool, dump_inline_hints,
22288 inline_summary::account_size_time, redirect_to_unreachable,
22289 edge_set_predicate, set_hint_predicate,
22290 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
22291 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
22292 inline_summary_t::remove, remap_hint_predicate_after_duplication,
22293 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
22294 ipa_call_summary_t::remove, initialize_growth_caches,
22295 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
22296 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
22297 mark_modified, unmodified_parm_1, unmodified_parm,
22298 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
22299 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
22300 compute_bb_predicates, will_be_nonconstant_expr_predicate,
22301 will_be_nonconstant_predicate, record_modified_bb_info,
22302 get_minimal_bb, record_modified, param_change_prob,
22303 phi_result_unknown_predicate, predicate_for_phi_result,
22304 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
22305 estimate_function_body_sizes, compute_inline_parameters,
22306 compute_inline_parameters_for_curren, pass_data_inline_parameters,
22307 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
22308 inline_update_callee_summaries, remap_edge_change_prob,
22309 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
22310 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
22311 inline_analyze_function, inline_summary_t::insert,
22312 inline_generate_summary, read_ipa_call_summary, inline_read_section,
22313 inline_read_summary, write_ipa_call_summary, inline_write_summary,
22314 inline_free_summary): Move to ipa-fnsummary.h
22315 (predicate_t): Remove.
22316 * ipa-fnsummary.c: New file.
22317 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
22318 (enum inline_hints_vals, inline_hints, agg_position_info,
22319 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
22320 inline_summaries, ipa_call_summary, ipa_call_summary_t,
22321 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
22322 dump_inline_summary, dump_inline_hints, inline_generate_summary,
22323 inline_read_summary, inline_write_summary, inline_free_summary,
22324 inline_analyze_function, initialize_inline_failed,
22325 inline_merge_summary, inline_update_overall_summary,
22326 compute_inline_parameters): Move to ipa-fnsummary.h
22327 * ipa-fnsummary.h: New file.
22328 * ipa-inline-transform.h: Include ipa-inline.h.
22329 * ipa-inline.c: LIkewise.
22331 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22333 * ipa-inline.c (edge_badness): Use inlined_time instead of
22334 inline_summaries->get.
22336 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22338 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
22340 2017-05-22 Nathan Sidwell <nathan@acm.org>
22342 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
22343 (fdump-lang): Document 'raw' option.
22344 * dumpfile.h (TDI_tu): Delete.
22345 * dumpfile.c (dump_files): Remove translation-unit.
22346 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
22348 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
22350 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
22351 command option from $(AWK) call.
22352 * config/avr/genmultilib.awk: Simplify and rewrite so that it
22353 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
22354 [FORMAT]: Remove handling of variable.
22355 * config/avr/t-multilib: Regenerate.
22357 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22359 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
22361 (dump_inline_summary): Do not print self_time.
22362 (estimate_function_body_sizes): Do not set self_time.
22363 (compute_inline_parameters): Likewise.
22364 (inline_read_section, inline_write_summary): Do not stream self_time.
22365 * ipa-inline.h (inline_summary): Drop self_time.
22367 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22369 * ipa-inline-analysis.c (account_size_time): Rename to ...
22370 (inline_summary::account_size_time): ... this one.
22371 (reset_ipa_call_summary): Turn to ...
22372 (ipa_call_summary::reset): ... this one.
22373 (reset_inline_summary): Turn to ...
22374 (inline_summary::reset): ... this one.
22375 (inline_summary_t::remove): Update.
22376 (inline_summary_t::duplicate): Update.
22377 (ipa_call_summary_t::remove): Update.
22378 (dump_inline_summary): Update.
22379 (estimate_function_body_sizes): Update.
22380 (compute_inline_parameters): Update.
22381 (estimate_node_size_and_time): Update.
22382 (inline_merge_summary): Update.
22383 (inline_update_overall_summary): Update.
22384 (inline_read_section): Update.
22385 (inline_write_summary): Update.
22386 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
22387 add account_size_time and reset member functions.
22388 (ipa_call_summary): Add reset function.
22389 * ipa-predicate.h (predicate::operator &): Constify.
22391 2017-05-22 Richard Biener <rguenther@suse.de>
22393 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
22395 2017-05-19 Jason Merrill <jason@redhat.com>
22397 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
22399 2017-05-19 Marek Polacek <polacek@redhat.com>
22402 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
22403 TYPE_OVERFLOW_WRAPS checks.
22405 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
22407 * tree-core.h (enum omp_clause_default_kind): Add
22408 "OMP_CLAUSE_DEFAULT_PRESENT".
22409 * tree-pretty-print.c (dump_omp_clause): Handle it.
22410 * gimplify.c (enum gimplify_omp_var_data): Add
22411 "GOVD_MAP_FORCE_PRESENT".
22412 (gimplify_adjust_omp_clauses_1): Map it to
22413 "GOMP_MAP_FORCE_PRESENT".
22414 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
22416 * gimplify.c (oacc_default_clause): Clarify.
22418 2017-05-19 Nathan Sidwell <nathan@acm.org>
22420 LANG_HOOK_REGISTER_DUMPS
22421 * toplev.c (general_init): Call register dump lang hook.
22422 * doc/invoke.texi: Document -fdump-lang option family.
22423 * dumpfile.c (dump_files): Remove class dump here.
22424 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
22425 * dumpfile.h (tree_dump_index): Remove TDI_class.
22426 * langhooks-def.h (lhd_register_dumps): Declare.
22427 (LANG_HOOKS_REGISTER_DUMPS): Define.
22428 (LANG_HOOKS_INITIALIZER): Add it.
22429 * langhooks.c (lhd_register_dumps): Define.
22430 * langhooks.h (struct lang_hooks): Add register_dumps.
22432 2017-05-19 Nathan Sidwell <nathan@acm.org>
22434 * context.h (context::set_passes): New.
22435 * context.c (context::context): Do not create pass manager.
22436 * toplev.c (general_init): Create pass manager here.
22438 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
22440 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
22441 use this splitter if two add or or instructions would also work for
22442 the constant we want to generate.
22444 2017-05-19 Richard Biener <rguenther@suse.de>
22447 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
22448 predicate evaluation.
22450 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
22452 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
22454 * ipa-inline.c (want_inline_small_function_p): Do not cast to
22457 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
22459 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
22460 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
22461 (inline_edge_summary_vec): Turn into ...
22462 (ipa_call_summaries): ... this one.
22463 (redirect_to_unreachable, edge_set_predicate,
22464 evaluate_properties_for_edge, inline_summary_alloc,
22465 reset_ipa_call_summary, reset_inline_summary,
22466 inline_summary_t::duplicate): Update.
22467 (inline_edge_duplication_hook): Turn to ...
22468 (ipa_call_summary_t::duplicate): ... this one.
22469 (inline_edge_removal_hook): Turn to ...
22470 (ipa_call_summary_t::remove): ... this one.
22471 (dump_inline_edge_summary): Turn to ...
22472 (dump_ipa_call_summary): ... this one.
22473 (estimate_function_body_sizes): Update.
22474 (inline_update_callee_summaries): Update.
22475 (remap_edge_change_prob): Update.
22476 (remap_edge_summaries): Update.
22477 (inline_merge_summary): Update.
22478 (do_estimate_edge_time): Update.
22479 (inline_generate_summary): Update.
22480 (inline_read_section): Update.
22481 (inline_read_summary): Update.
22482 (inline_free_summary): Update.
22483 * ipa-inline.c (can_inline_edge_p): Update.
22484 (compute_inlined_call_time): Update.
22485 (want_inline_small_function_p): Update.
22486 (edge_badness): Update.
22487 (early_inliner): Update.
22488 * ipa-inline.h (inline_edge_summary): Turn to ...
22489 (ipa_call_summary): ... this one.
22490 (ipa_call_summary_t): New class.
22491 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
22492 (ipa_call_summaries): New.
22493 (inline_edge_summary): Remove.
22494 (estimate_edge_growth): Update.
22495 * ipa-profile.c (ipa_propagate_frequency_1): Update.
22496 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22497 * ipa-split.c (execute_split_functions): Update.
22498 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
22500 2017-05-19 Richard Biener <rguenther@suse.de>
22502 PR middle-end/80764
22503 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
22505 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
22507 * config/rs6000/rs6000.c (struct machine_function): Add field
22508 fpr_is_wrapped_separately.
22509 (rs6000_get_separate_components): Use 64 components. Handle the
22510 new FPR components.
22511 (rs6000_components_for_bb): Handle the FPR components.
22512 (rs6000_emit_prologue_components): Handle the FPR components.
22513 (rs6000_emit_epilogue_components): Handle the FPR components.
22514 (rs6000_set_handled_components): Handle the FPR components.
22515 (rs6000_emit_prologue): Don't output prologue code for those FPRs
22516 that are already separately shrink-wrapped.
22517 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
22518 that are already separately shrink-wrapped.
22520 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
22523 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
22526 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
22527 (define_peephole2 for Altivec d-form load): Add peepholes to catch
22528 cases where the register allocator uses a move and an offsettable
22529 memory operation to/from a FPR register on ISA 2.06/2.07.
22530 (define_peephole2 for Altivec d-form store): Likewise.
22532 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
22535 * config/i386/mmx.md (*mov<mode>_internal): Enable
22536 alternatives 11, 12, 13 and 14 also for 32bit targets.
22537 Remove alternatives 15, 16, 17 and 18.
22538 * config/i386/sse.md (vec_concatv2di): Change
22539 alternative (!x, *y) to (x, ?!*Yn).
22541 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
22543 * dumpfile.h (enum dump_kind): Remove stray comma.
22545 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
22547 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
22548 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
22549 predicate::num_conditions
22550 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
22551 (CHANGED): turn into predicate::changed.
22552 (agg_position_info): Move to ipa-predicate.h
22553 (add_condition, predicate::add_clause, predicate::operator &=,
22554 predicate::or_with, predicate::evaluate, predicate::probability,
22555 dump_condition, dump_clause, predicate::dump,
22556 predicate::remap_after_duplication, predicate::remap_after_inlining,
22557 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
22558 (evaluate_conditions_for_known_args): Update.
22559 (set_cond_stmt_execution_predicate): Update.
22560 * ipa-inline.h: Include ipa-predicate.h
22561 (condition, inline_param_summary, conditions, agg_position_info,
22562 predicate): Move to ipa-predicate.h
22563 * ipa-predicate.c: New file.
22564 * ipa-predicate.h: New file.
22566 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
22568 * final.c (leaf_function_p): Check we are not in a sequence.
22570 2017-05-18 Martin Liska <mliska@suse.cz>
22572 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
22573 * dumpfile.c (dump_register): Use new enum dump_kind.
22574 (get_dump_file_name): Likewise.
22575 (dump_enable_all): Likewise.
22576 (dump_switch_p_1): Likewise.
22577 (enable_rtl_dump_file): Remove usage of TDF_RTL.
22578 * dumpfile.h (enum dump_kind): New enum type.
22579 (struct dump_file_info): Create constructor and
22580 format fields and comments.
22581 * passes.c (pass_manager::register_one_dump_file):
22583 * statistics.c (statistics_early_init): Likewise.
22584 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
22585 TDF_TREE with TDF_SLIM.
22586 (gather_memory_references_ref): Likewise.
22588 2017-05-18 Martin Liska <mliska@suse.cz>
22590 * vec.h (struct vnull): Use it.
22592 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
22594 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
22595 (true_predicate, false_predicate, true_predicate_p,
22596 false_predicate_p): Remove.
22597 (single_cond_predicate, not_inlined_predicate): Turn to member function
22599 (add_condition): Update.
22600 (add_clause): Turn to...
22601 (predicate::add_clause): ... this one; update; allow passing NULL
22603 (and_predicates): Turn to ...
22604 (predicate::operator &=): ... this one.
22605 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
22606 (or_predicates): Turn to ...
22607 (predicate::or_with): ... this one.
22608 (evaluate_predicate): Turn to ...
22609 (predicate::evaluate): ... this one.
22610 (predicate_probability): Turn to ...
22611 (predicate::probability): ... this one.
22612 (dump_condition): Update.
22613 (dump_predicate): Turn to ...
22614 (predicate::dump): ... this one.
22615 (account_size_time): Update.
22616 (edge_set_predicate): Update.
22617 (set_hint_predicate): UPdate.
22618 (evaluate_conditions_for_known_args): Update.
22619 (evaluate_properties_for_edge): Update.
22620 (remap_predicate_after_duplication): Turn to...
22621 (predicate::remap_after_duplication): ... this one.
22622 (remap_hint_predicate_after_duplication): Update.
22623 (inline_summary_t::duplicate): UPdate.
22624 (dump_inline_edge_summary): Update.
22625 (dump_inline_summary): Update.
22626 (set_cond_stmt_execution_predicate): Update.
22627 (set_switch_stmt_execution_predicate): Update.
22628 (compute_bb_predicates): Update.
22629 (will_be_nonconstant_expr_predicate): Update.
22630 (will_be_nonconstant_predicate): Update.
22631 (phi_result_unknown_predicate): Update.
22632 (predicate_for_phi_result): Update.
22633 (array_index_predicate): Update.
22634 (estimate_function_body_sizes): Update.
22635 (estimate_node_size_and_time): Update.
22636 (estimate_ipcp_clone_size_and_time): Update.
22637 (remap_predicate): Rename to ...
22638 (predicate::remap_after_inlining): ... this one.
22639 (remap_hint_predicate): Update.
22640 (inline_merge_summary): Update.
22641 (inline_update_overall_summary): Update.
22642 (estimate_size_after_inlining): Update.
22643 (read_predicate): Rename to ...
22644 (predicate::stream_in): ... this one.
22645 (read_inline_edge_summary): Update.
22646 (write_predicate): Rename to ...
22647 (predicate::stream_out): ... this one.
22648 (write_inline_edge_summary): Update.
22649 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
22650 (clause_t): Turn to uint32_t
22651 (predicate): Turn to class; implement constructor and operators
22653 (size_time_entry): Update.
22654 (inline_summary): Update.
22655 (inline_edge_summary): Update.
22657 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
22659 * fold-const.c (fold_binary_loc): Move transformation...
22660 * match.pd (C - X CMP X): ... here.
22662 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
22664 * config/sparc/sparc.c (sparc_option_override): Set function
22665 alignment for -mcpu=niagara7 to 64 to match the I$ line.
22666 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
22668 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
22670 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
22672 2017-05-18 Marek Polacek <polacek@redhat.com>
22675 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
22676 (pass_ubsan::execute): Call gimple_assign_single_p instead of
22677 gimple_assign_load_p.
22679 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
22681 PR middle-end/80692
22682 * real.c (do_compare): Give decimal_do_compare preference over
22683 comparing just the signs.
22685 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
22687 * doc/md.texi (Canonicalization of Instructions): Describe the
22688 canonical form of instructions that inherently set a condition
22691 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
22693 PR middle-end/80775
22694 * tree-cfg.c: Move deletion of unreachable case statements to after
22695 the merging of consecutive case labels.
22697 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
22699 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
22700 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
22701 restoring of callee-saved registers.
22703 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
22705 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
22706 * config/visium/visium.c (single_set_and_flags): Likewise.
22707 * config/visium/visium.md (Substitutions): Likewise.
22709 2017-05-17 Martin Liska <mliska@suse.cz>
22711 * cfg.c: Introduce dump_flags_t type and
22712 use it instead of int type.
22714 * cfghooks.c: Likewise.
22715 * cfghooks.h (struct cfg_hooks): Likewise.
22716 * cfgrtl.c: Likewise.
22717 * cfgrtl.h: Likewise.
22718 * cgraph.c (cgraph_node::get_body): Likewise.
22719 * coretypes.h: Likewise.
22720 * domwalk.c: Likewise.
22721 * domwalk.h: Likewise.
22722 * dumpfile.c (struct dump_option_value_info): Likewise.
22723 (dump_enable_all): Likewise.
22724 (dump_switch_p_1): Likewise.
22725 (opt_info_switch_p): Likewise.
22726 * dumpfile.h (enum tree_dump_index): Likewise.
22727 (struct dump_file_info): Likewise.
22728 * genemit.c: Likewise.
22729 * generic-match-head.c: Likewise.
22730 * gengtype.c (open_base_files): Likewise.
22731 * gimple-pretty-print.c: Likewise.
22732 * gimple-pretty-print.h: Likewise.
22733 * graph.c (print_graph_cfg): Likewise.
22734 * graphite-scop-detection.c (dot_all_sese): Likewise.
22735 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
22736 * loop-unroll.c (report_unroll): Likewise.
22737 * passes.c (pass_manager::register_one_dump_file): Likewise.
22738 * print-tree.c: Likewise.
22739 * statistics.c: Likewise.
22740 * tree-cfg.c: Likewise.
22741 * tree-cfg.h: Likewise.
22742 * tree-dfa.c: Likewise.
22743 * tree-dfa.h: Likewise.
22744 * tree-dump.c (dump_function): Likewise.
22745 * tree-dump.h (struct dump_info): Likewise.
22746 * tree-pretty-print.c: Likewise.
22747 * tree-pretty-print.h: Likewise.
22748 * tree-ssa-live.c: Likewise.
22749 * tree-ssa-live.h: Likewise.
22750 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
22751 * tree-vect-loop.c: Likewise.
22752 * tree-vect-slp.c: Likewise.
22754 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
22755 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22757 PR tree-optimization/80457
22758 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
22759 of all arguments to a statement as scalar_to_vec operations.
22760 (vectorizable_call): Adjust call to vect_model_simple_cost for
22762 (vectorizable_conversion): Likewise.
22763 (vectorizable_assignment): Likewise.
22764 (vectorizable_shift): Likewise.
22765 (vectorizable_operation): Likewise.
22766 (vectorizable_comparison): Likewise.
22767 (vect_is_simple_cond): Record the def types for operands.
22768 (vectorizable_condition): Likewise, call vect_model_simple_cost.
22769 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
22770 for statement argument count.
22772 2017-05-16 Carl Love <cel@us.ibm.com>
22774 * config/rs6000/rs6000-c: Add support for built-in functions
22775 vector unsigned long long vec_bperm (vector unsigned long long,
22776 vector unsigned char)
22777 vector signed long long vec_mule (vector signed int,
22779 vector unsigned long long vec_mule (vector unsigned int,
22780 vector unsigned int)
22781 vector signed long long vec_mulo (vector signed int,
22783 vector unsigned long long vec_mulo (vector unsigned int,
22784 vector unsigned int)
22785 vector signed char vec_sldw (vector signed char,
22786 vector signed char,
22788 vector unsigned char vec_sldw (vector unsigned char,
22789 vector unsigned char,
22791 vector signed short vec_sldw (vector signed short,
22792 vector signed short,
22794 vector unsigned short vec_sldw (vector unsigned short,
22795 vector unsigned short,
22797 vector signed int vec_sldw (vector signed int,
22800 vector unsigned int vec_sldw (vector unsigned int,
22801 vector unsigned int,
22803 vector signed long long vec_sldw (vector signed long long,
22804 vector signed long long,
22806 vector unsigned long long vec_sldw (vector unsigned long long,
22807 vector unsigned long long,
22809 * config/rs6000/rs6000-c: Add support for built-in functions
22810 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
22811 * config/rs6000/altivec.h: Add defintion for vec_sldw.
22812 * doc/extend.texi: Update the built-in documentation for the
22813 new built-in functions.
22815 2017-05-16 Marek Polacek <polacek@redhat.com>
22819 * tree.c (save_expr): Don't fold the expression.
22821 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
22823 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
22824 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
22825 and (?*y,m). Update insn attributes.
22827 2017-05-16 Martin Liska <mliska@suse.cz>
22829 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
22830 flags argument of print_gimple_stmt, print_gimple_expr,
22831 print_generic_stmt and print_generic_expr.
22832 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
22833 * coretypes.h: Likewise.
22834 * except.c (dump_eh_tree): Likewise.
22835 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
22836 * gimple-pretty-print.h: Likewise.
22837 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
22838 (backprop::push_to_worklist): Likewise.
22839 (backprop::pop_from_worklist): Likewise.
22840 (backprop::process_use): Likewise.
22841 (backprop::intersect_uses): Likewise.
22842 (note_replacement): Likewise.
22843 * gimple-ssa-store-merging.c
22844 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
22845 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
22846 (pass_store_merging::execute): Likewise.
22847 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
22848 (ssa_base_cand_dump_callback): Likewise.
22849 (dump_incr_vec): Likewise.
22850 (replace_refs): Likewise.
22851 (replace_mult_candidate): Likewise.
22852 (create_add_on_incoming_edge): Likewise.
22853 (create_phi_basis): Likewise.
22854 (insert_initializers): Likewise.
22855 (all_phi_incrs_profitable): Likewise.
22856 (introduce_cast_before_cand): Likewise.
22857 (replace_one_candidate): Likewise.
22858 * gimplify.c (gimplify_expr): Likewise.
22859 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
22860 (set_rename): Likewise.
22861 (rename_uses): Likewise.
22862 (copy_loop_phi_nodes): Likewise.
22863 (add_close_phis_to_merge_points): Likewise.
22864 (copy_loop_close_phi_args): Likewise.
22865 (copy_cond_phi_args): Likewise.
22866 (graphite_copy_stmts_from_block): Likewise.
22867 (translate_pending_phi_nodes): Likewise.
22868 * graphite-poly.c (print_pdr): Likewise.
22869 (dump_gbb_cases): Likewise.
22870 (dump_gbb_conditions): Likewise.
22871 (print_scop_params): Likewise.
22872 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
22873 (build_cross_bb_scalars_use): Likewise.
22874 (gather_bbs::before_dom_children): Likewise.
22875 * hsa-dump.c (dump_hsa_immed): Likewise.
22876 * ipa-cp.c (print_ipcp_constant_value): Likewise.
22877 (get_replacement_map): Likewise.
22878 * ipa-inline-analysis.c (dump_condition): Likewise.
22879 (estimate_function_body_sizes): Likewise.
22880 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
22881 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
22882 * ipa-prop.c (ipa_dump_param): Likewise.
22883 (ipa_print_node_jump_functions_for_edge): Likewise.
22884 (ipa_modify_call_arguments): Likewise.
22885 (ipa_modify_expr): Likewise.
22886 (ipa_dump_param_adjustments): Likewise.
22887 (ipa_dump_agg_replacement_values): Likewise.
22888 (ipcp_modif_dom_walker::before_dom_children): Likewise.
22889 * ipa-pure-const.c (check_stmt): Likewise.
22890 (pass_nothrow::execute): Likewise.
22891 * ipa-split.c (execute_split_functions): Likewise.
22892 * omp-offload.c (dump_oacc_loop_part): Likewise.
22893 (dump_oacc_loop): Likewise.
22894 * trans-mem.c (tm_log_emit): Likewise.
22895 (tm_memopt_accumulate_memops): Likewise.
22896 (dump_tm_memopt_set): Likewise.
22897 (dump_tm_memopt_transform): Likewise.
22898 * tree-cfg.c (gimple_verify_flow_info): Likewise.
22899 (print_loop): Likewise.
22900 * tree-chkp-opt.c (chkp_print_addr): Likewise.
22901 (chkp_gather_checks_info): Likewise.
22902 (chkp_get_check_result): Likewise.
22903 (chkp_remove_check_if_pass): Likewise.
22904 (chkp_use_outer_bounds_if_possible): Likewise.
22905 (chkp_reduce_bounds_lifetime): Likewise.
22906 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
22907 (chkp_mark_completed_bounds): Likewise.
22908 (chkp_register_incomplete_bounds): Likewise.
22909 (chkp_mark_invalid_bounds): Likewise.
22910 (chkp_maybe_copy_and_register_bounds): Likewise.
22911 (chkp_build_returned_bound): Likewise.
22912 (chkp_get_bound_for_parm): Likewise.
22913 (chkp_build_bndldx): Likewise.
22914 (chkp_get_bounds_by_definition): Likewise.
22915 (chkp_generate_extern_var_bounds): Likewise.
22916 (chkp_get_bounds_for_decl_addr): Likewise.
22917 * tree-chrec.c (chrec_apply): Likewise.
22918 * tree-data-ref.c (dump_data_reference): Likewise.
22919 (dump_subscript): Likewise.
22920 (dump_data_dependence_relation): Likewise.
22921 (analyze_overlapping_iterations): Likewise.
22922 * tree-inline.c (expand_call_inline): Likewise.
22923 (tree_function_versioning): Likewise.
22924 * tree-into-ssa.c (dump_defs_stack): Likewise.
22925 (dump_currdefs): Likewise.
22926 (dump_names_replaced_by): Likewise.
22927 (dump_update_ssa): Likewise.
22928 (update_ssa): Likewise.
22929 * tree-object-size.c (pass_object_sizes::execute): Likewise.
22930 * tree-parloops.c (build_new_reduction): Likewise.
22931 (try_create_reduction_list): Likewise.
22932 (ref_conflicts_with_region): Likewise.
22933 (oacc_entry_exit_ok_1): Likewise.
22934 (oacc_entry_exit_single_gang): Likewise.
22935 * tree-pretty-print.h: Likewise.
22936 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
22937 (get_scalar_evolution): Likewise.
22938 (add_to_evolution): Likewise.
22939 (get_loop_exit_condition): Likewise.
22940 (analyze_evolution_in_loop): Likewise.
22941 (analyze_initial_condition): Likewise.
22942 (analyze_scalar_evolution): Likewise.
22943 (instantiate_scev): Likewise.
22944 (number_of_latch_executions): Likewise.
22945 (gather_chrec_stats): Likewise.
22946 (final_value_replacement_loop): Likewise.
22947 (scev_const_prop): Likewise.
22948 * tree-sra.c (dump_access): Likewise.
22949 (disqualify_candidate): Likewise.
22950 (create_access): Likewise.
22951 (reject): Likewise.
22952 (maybe_add_sra_candidate): Likewise.
22953 (create_access_replacement): Likewise.
22954 (analyze_access_subtree): Likewise.
22955 (analyze_all_variable_accesses): Likewise.
22956 (sra_modify_assign): Likewise.
22957 (initialize_constant_pool_replacements): Likewise.
22958 (find_param_candidates): Likewise.
22959 (decide_one_param_reduction): Likewise.
22960 (replace_removed_params_ssa_names): Likewise.
22961 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
22962 * tree-ssa-copy.c (dump_copy_of): Likewise.
22963 (copy_prop_visit_cond_stmt): Likewise.
22964 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
22965 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
22966 (record_equivalences_from_stmt): Likewise.
22967 * tree-ssa-dse.c (compute_trims): Likewise.
22968 (delete_dead_call): Likewise.
22969 (delete_dead_assignment): Likewise.
22970 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
22971 (forward_propagate_into_cond): Likewise.
22972 (pass_forwprop::execute): Likewise.
22973 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
22974 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
22976 (move_computations_worker): Likewise.
22977 (execute_sm): Likewise.
22978 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
22979 (remove_exits_and_undefined_stmts): Likewise.
22980 (remove_redundant_iv_tests): Likewise.
22981 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
22982 (adjust_iv_update_pos): Likewise.
22983 * tree-ssa-math-opts.c (bswap_replace): Likewise.
22984 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
22985 (value_replacement): Likewise.
22986 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
22987 * tree-ssa-pre.c (print_pre_expr): Likewise.
22988 (get_representative_for): Likewise.
22989 (create_expression_by_pieces): Likewise.
22990 (insert_into_preds_of_block): Likewise.
22991 (eliminate_insert): Likewise.
22992 (eliminate_dom_walker::before_dom_children): Likewise.
22993 (eliminate): Likewise.
22994 (remove_dead_inserted_code): Likewise.
22995 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
22996 * tree-ssa-reassoc.c (get_rank): Likewise.
22997 (eliminate_duplicate_pair): Likewise.
22998 (eliminate_plus_minus_pair): Likewise.
22999 (eliminate_not_pairs): Likewise.
23000 (undistribute_ops_list): Likewise.
23001 (eliminate_redundant_comparison): Likewise.
23002 (update_range_test): Likewise.
23003 (optimize_range_tests_var_bound): Likewise.
23004 (optimize_vec_cond_expr): Likewise.
23005 (rewrite_expr_tree): Likewise.
23006 (rewrite_expr_tree_parallel): Likewise.
23007 (linearize_expr): Likewise.
23008 (break_up_subtract): Likewise.
23009 (linearize_expr_tree): Likewise.
23010 (attempt_builtin_powi): Likewise.
23011 (attempt_builtin_copysign): Likewise.
23012 (transform_stmt_to_copy): Likewise.
23013 (transform_stmt_to_multiply): Likewise.
23014 (dump_ops_vector): Likewise.
23015 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
23016 (print_scc): Likewise.
23017 (set_ssa_val_to): Likewise.
23018 (visit_reference_op_store): Likewise.
23019 (visit_use): Likewise.
23020 (sccvn_dom_walker::before_dom_children): Likewise.
23021 (run_scc_vn): Likewise.
23022 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
23024 (expr_hash_elt::print): Likewise.
23025 (const_and_copies::pop_to_marker): Likewise.
23026 (const_and_copies::record_const_or_copy_raw): Likewise.
23027 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
23028 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
23029 (dump_predicates): Likewise.
23030 (find_uninit_use): Likewise.
23031 (warn_uninitialized_phi): Likewise.
23032 (pass_late_warn_uninitialized::execute): Likewise.
23033 * tree-ssa.c (verify_vssa): Likewise.
23034 (verify_ssa): Likewise.
23035 (maybe_optimize_var): Likewise.
23036 * tree-vrp.c (dump_value_range): Likewise.
23037 (dump_all_value_ranges): Likewise.
23038 (dump_asserts_for): Likewise.
23039 (register_edge_assert_for_2): Likewise.
23040 (vrp_visit_cond_stmt): Likewise.
23041 (vrp_visit_switch_stmt): Likewise.
23042 (vrp_visit_stmt): Likewise.
23043 (vrp_visit_phi_node): Likewise.
23044 (simplify_cond_using_ranges_1): Likewise.
23045 (fold_predicate_in): Likewise.
23046 (evrp_dom_walker::before_dom_children): Likewise.
23047 (evrp_dom_walker::push_value_range): Likewise.
23048 (evrp_dom_walker::pop_value_range): Likewise.
23049 (execute_early_vrp): Likewise.
23051 2017-05-16 Richard Biener <rguenther@suse.de>
23053 * dwarf2out.c (loc_list_from_tree_1): Do not create
23054 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
23056 2017-05-16 Richard Biener <rguenther@suse.de>
23058 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
23060 (note_variable_value_in_expr): If we resolved the decl ref
23061 do not push to the stack.
23063 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
23065 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
23066 operations in fast-math mode.
23067 (vaddq_f16): Likewise.
23068 (vmul_f16): Likewise.
23069 (vmulq_f16): Likewise.
23070 (vsub_f16): Likewise.
23071 (vsubq_f16): Likewise.
23072 * config/arm/neon.md (add<mode>3): New.
23074 (fma:<VH:mode>3): New. Also remove outdated comment.
23077 2017-05-16 Martin Liska <mliska@suse.cz>
23081 * ipa-devirt.c (warn_types_mismatch): Fix typo.
23082 (odr_types_equivalent_p): Likewise.
23084 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
23086 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
23088 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
23091 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
23092 non-interunit SSE move alternatives with '?'.
23093 (zero-extendsidi peephole2): New peephole to skip intermediate
23094 general register in SSE zero-extend sequence.
23096 2017-05-15 Jeff Law <law@redhat.com>
23098 * reorg.c (relax_delay_slots): Create a new variable to hold
23099 the temporary target rather than clobbering TARGET_LABEL.
23101 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
23102 missing argument to extract_bit_field call.
23103 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
23105 2017-05-15 Martin Liska <mliska@suse.cz>
23108 * gcc.c (process_command): Do not allow empty argument of -o option.
23110 2017-05-15 Renlin Li <renlin.li@arm.com>
23112 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
23113 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
23114 * config/aarch64/constraints.md (Usf): Add long call check.
23115 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
23116 (call_value): Likewise.
23117 (sibcall): Likewise.
23118 (sibcall_value): Likewise.
23120 (call_value_insn): New.
23121 (sibcall_insn): Update rtx pattern.
23122 (sibcall_value_insn): Likewise.
23123 (call_internal): Remove.
23124 (call_value_internal): Likewise.
23125 (sibcall_internal): Likewise.
23126 (sibcall_value_internal): Likewise.
23127 (call_reg): Likewise.
23128 (call_symbol): Likewise.
23129 (call_value_reg): Likewise.
23130 (call_value_symbol): Likewise.
23132 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
23135 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
23137 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
23139 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
23140 compatible with CCGOCmode and with CCZmode.
23142 2017-05-14 Martin Sebor <msebor@redhat.com>
23144 PR middle-end/77671
23145 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
23146 (gimple_fold_builtin_snprintf): Same.
23147 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
23148 (gimple_fold_builtin_snprintf): Same.
23149 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
23150 of character types.
23151 (is_call_safe): New function.
23152 (try_substitute_return_value): Call it.
23153 (try_simplify_call): New function.
23154 (pass_sprintf_length::handle_gimple_call): Call it.
23156 2017-05-14 Martin Sebor <msebor@redhat.com>
23158 PR middle-end/80669
23159 * builtins.c (expand_builtin_stpncpy): Simplify.
23161 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
23163 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
23164 * config/i386/i386.h
23165 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
23166 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
23167 (struct machine_function): Add new members call_ms2sysv,
23168 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
23169 (struct machine_frame_state): New fields sp_realigned and
23170 sp_realigned_offset.
23171 * config/i386/i386.c
23172 (enum xlogue_stub): New enum.
23173 (enum xlogue_stub_sets): New enum.
23174 (class xlogue_layout): New class.
23175 (struct ix86_frame): New fields stack_realign_allocate_offset,
23176 stack_realign_offset and outlined_save_offset. Modify comments to
23177 detail stack layout when using out-of-line stubs.
23178 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
23179 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
23180 -mcall-ms2sysv-xlogues.
23181 (stub_managed_regs): New static variable.
23182 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
23183 registers managed by out-of-line stub.
23184 (disable_call_ms2sysv_xlogues): New function.
23185 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
23186 m->call_ms2sysv when appropriate and compute frame layout for
23188 (sp_valid_at, fp_valid_at): New inline functions.
23189 (choose_basereg): New function.
23190 (choose_baseaddr): Add align parameter, use choose_basereg and modify
23192 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
23193 Use align parameter of choose_baseaddr to generated aligned SSE movs
23195 (pro_epilogue_adjust_stack): Modify to track
23196 machine_frame_state::sp_realigned.
23197 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
23198 (ix86_nsaved_sseregs): Likewise.
23199 (ix86_emit_save_regs): Likewise.
23200 (ix86_emit_save_regs_using_mov): Likewise.
23201 (ix86_emit_save_sse_regs_using_mov): Likewise.
23202 (get_scratch_register_on_entry): Likewise.
23203 (gen_frame_set): New function.
23204 (gen_frame_load): Likewise.
23205 (gen_frame_store): Likewise.
23206 (emit_outlined_ms2sysv_save): Likewise.
23207 (emit_outlined_ms2sysv_restore): Likewise.
23208 (ix86_expand_prologue): Modify stack re-alignment code and call
23209 emit_outlined_ms2sysv_save when appropriate.
23210 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
23211 parameter rtx_insn *insn, which allows the function to be used to only
23212 generate the notes.
23213 (ix86_expand_epilogue): Modify validity checks of frame and stack
23214 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
23215 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
23216 * config/i386/predicates.md
23217 (save_multiple): New predicate.
23218 (restore_multiple): Likewise.
23219 * config/i386/sse.md
23220 (save_multiple<mode>): New pattern.
23221 (save_multiple_realign<mode>): Likewise.
23222 (restore_multiple<mode>): Likewise.
23223 (restore_multiple_and_return<mode>): Likewise.
23224 (restore_multiple_leave_return<mode>): Likewise.
23225 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
23227 2017-05-14 Julia Koval <julia.koval@intel.com>
23229 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
23230 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
23231 (__builtin_ia32_xsetbv): New builtins.
23232 * config/i386/i386.c (ix86_expand_special_args_builtin):
23234 (ix86_expand_builtin): Special expand for new intrinsics.
23235 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
23236 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
23237 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
23239 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23241 * cfganal.c (inverted_post_order_compute): Change argument type
23243 * cfganal.h (inverted_post_order_compute): Adjust prototype.
23244 * df-core.c (rest_of_handle_df_initialize): Adjust.
23245 (rest_of_handle_df_finish): Likewise.
23246 (df_analyze_1): Likewise.
23247 (df_analyze): Likewise.
23248 (loop_inverted_post_order_compute): Change argument to be a vec *.
23249 (df_analyze_loop): Adjust.
23250 (df_get_n_blocks): Likewise.
23251 (df_get_postorder): Likewise.
23252 * df.h (struct df_d): Change field to be a vec.
23253 * lcm.c (compute_laterin): Adjust.
23254 (compute_available): Likewise.
23255 * lra-lives.c (lra_create_live_ranges_1): Likewise.
23256 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
23257 * tree-ssa-pre.c (compute_antic): Likewise.
23259 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23261 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
23262 (depth_first_search::depth_first_search): Change structure init
23263 function to this constructor.
23264 (depth_first_search::add_bb): Rename function to this member.
23265 (depth_first_search::execute): Likewise.
23266 (flow_dfs_compute_reverse_finish): Adjust.
23268 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23270 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
23271 (longest_simple_path): Likewise.
23272 * shrink-wrap.c (spread_components): Likewise.
23273 (disqualify_problematic_components): Likewise.
23274 (emit_common_heads_for_components): Likewise.
23275 (emit_common_tails_for_components): Likewise.
23276 (insert_prologue_epilogue_for_components): Likewise.
23278 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23280 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
23283 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23285 * df-core.c (df_set_blocks): Start using auto_bitmap.
23286 (df_compact_blocks): Likewise.
23287 * df-problems.c (df_rd_confluence_n): Likewise.
23288 * df-scan.c (df_insn_rescan_all): Likewise.
23289 (df_process_deferred_rescans): Likewise.
23290 (df_update_entry_block_defs): Likewise.
23291 (df_update_exit_block_uses): Likewise.
23292 (df_entry_block_bitmap_verify): Likewise.
23293 (df_exit_block_bitmap_verify): Likewise.
23294 (df_scan_verify): Likewise.
23295 * lra-constraints.c (lra_constraints): Likewise.
23296 (undo_optional_reloads): Likewise.
23297 (lra_undo_inheritance): Likewise.
23298 * lra-remat.c (calculate_gen_cands): Likewise.
23299 (do_remat): Likewise.
23300 * lra-spills.c (assign_spill_hard_regs): Likewise.
23301 (spill_pseudos): Likewise.
23302 * tree-ssa-pre.c (bitmap_set_and): Likewise.
23303 (bitmap_set_subtract_values): Likewise.
23305 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23307 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
23308 management with auto_bitmap.
23309 (fix_inter_tick): Likewise.
23310 (fix_recovery_deps): Likewise.
23311 * ira.c (add_store_equivs): Likewise.
23312 (find_moveable_pseudos): Likewise.
23313 (split_live_ranges_for_shrink_wrap): Likewise.
23314 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
23315 (rtx_reuse_manager::seen_def_p): Likewise.
23316 (rtx_reuse_manager::set_seen_def): Likewise.
23317 * print-rtl.h (class rtx_reuse_manager): Likewise.
23319 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23321 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
23323 (migrate_btr_def): Likewise.
23324 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
23325 * df-core.c (loop_post_order_compute): Likewise.
23326 (loop_inverted_post_order_compute): Likewise.
23327 * hsa-common.h: Likewise.
23328 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
23329 * init-regs.c (initialize_uninitialized_regs): Likewise.
23330 * ipa-inline.c (resolve_noninline_speculation): Likewise.
23331 (inline_small_functions): Likewise.
23332 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
23333 * ira.c (combine_and_move_insns): Likewise.
23334 (build_insn_chain): Likewise.
23335 * loop-invariant.c (find_invariants): Likewise.
23336 * lower-subreg.c (propagate_pseudo_copies): Likewise.
23337 * predict.c (tree_predict_by_opcode): Likewise.
23338 (predict_paths_leading_to): Likewise.
23339 (predict_paths_leading_to_edge): Likewise.
23340 (estimate_loops_at_level): Likewise.
23341 (estimate_loops): Likewise.
23342 * shrink-wrap.c (try_shrink_wrapping): Likewise.
23343 (spread_components): Likewise.
23344 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
23345 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
23346 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
23347 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
23348 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
23349 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
23350 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
23351 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
23352 (mark_threaded_blocks): Likewise.
23353 (thread_through_all_blocks): Likewise.
23354 * tree-ssa.c (verify_ssa): Likewise.
23355 (execute_update_addresses_taken): Likewise.
23356 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
23358 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23360 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
23362 (post_order_compute): Likewise.
23363 (inverted_post_order_compute): Likewise.
23364 (pre_and_rev_post_order_compute_fn): Likewise.
23366 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23368 * genrecog.c (int_set::int_set): Explicitly construct our
23369 auto_vec base class.
23370 * vec.h (auto_vec::auto_vec): New constructor.
23372 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23374 * bitmap.h (class auto_bitmap): New constructor taking
23375 bitmap_obstack * argument.
23377 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23379 * bitmap.h (class auto_bitmap): Change type of m_bits to
23380 bitmap_head, and adjust ctor / dtor and member operators.
23382 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
23384 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
23385 when returned register mode doesn't match original mode.
23387 2017-05-12 Jeff Law <law@redhat.com>
23388 Jakub Jelinek <jakub@redhat.com>
23390 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
23391 we look for cc setter after the compare-elim changes.
23392 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
23393 within the vector to match what compare-elim now expects.
23394 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
23395 (xorsi3_flags, one_cmplsi2_flags): Likewise.
23397 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
23398 after the compare-elim changes.
23399 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
23400 the vector to match what compare-elim now expects.
23401 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
23402 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
23403 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
23404 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
23405 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
23407 * config/visium/visium.c (single_set_and_flags): Fix where
23408 we look for cc setter after the compare-elim changes.
23409 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
23410 with the vector to match what compare-elim now expects.
23411 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
23412 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
23413 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
23414 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
23415 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
23416 (neg<mode>2_insn_set_overflow): Likewise.
23418 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
23420 PR middle-end/79794
23421 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
23422 maybe_expand_insn call, set ops[0].target. If still set after call,
23423 set alt_rtl. Add extra arg to recursive calls.
23424 (extract_bit_field): Add alt_rtl argument. Pass to
23426 * expmed.h (extract_bit_field): Fix prototype.
23427 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
23428 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
23429 to extract_bit_field_calls.
23430 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
23431 Pass alt_rtl to extract_bit_field calls.
23432 * calls.c (store_unaligned_arguments_into_psuedos)
23433 load_register_parameters): Pass extra NULL to extract_bit_field calls.
23434 * optabs.c (maybe_legitimize_operand): Clear op->target when call
23436 * optabs.h (struct expand_operand): Add target bitfield.
23438 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
23440 * compare-elim.c (try_eliminate_compare): Canonicalize
23441 operation with embedded compare to
23442 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
23443 (set (reg) (operation)].
23445 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
23447 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
23450 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
23451 cost of adding a carry flag for ADC instruction.
23452 [case MINUS]: Ignore the cost of subtracting a carry flag
23453 for SBB instruction.
23455 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
23457 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
23459 * config/rs6000/bmiintrin.h: New file.
23460 * config/rs6000/bmi2intrin.h: New file.
23461 * config/rs6000/x86intrin.h: New file.
23463 2017-05-12 Jeff Law <law@redhat.com>
23465 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
23468 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
23470 PR middle-end/80707
23471 * tree-cfg.c: Remove cfg edges of unreachable case statements.
23473 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23475 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
23476 early expansion of vector divide builtins.
23477 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
23478 builtins identified as having unsigned arguments.
23480 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23482 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
23483 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
23484 expansion of vector logical operations (and, andc, or, xor,
23487 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23489 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
23490 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
23492 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23494 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
23495 early GIMPLE expansion of vector multiplies.
23497 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23499 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
23500 TARGET_HAVE_MOVT conditional.
23501 (movt splitter): Likewise.
23503 2017-05-12 Richard Biener <rguenther@suse.de>
23505 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
23506 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
23507 Fold all stmts not inplace.
23509 2017-05-12 Richard Biener <rguenther@suse.de>
23511 PR tree-optimization/80713
23512 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
23513 inserted_exprs bit for not removed stmts.
23515 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
23517 PR middle-end/69921
23518 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
23519 parallelized" attribute for parallelized OpenACC kernels.
23520 * omp-offload.c (execute_oacc_device_lower): Use it.
23522 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
23523 Set "oacc kernels" attribute.
23524 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
23525 parameter. Adjust all users.
23526 (oacc_fn_attrib_kernels_p): Remove function.
23527 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
23528 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
23529 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
23530 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
23531 assert "oacc kernels" attribute is set.
23533 2017-05-11 Carl Love <cel@us.ibm.com>
23535 * config/rs6000/rs6000-c: Add support for built-in functions
23536 vector unsigned char vec_popcnt (vector signed char)
23537 vector unsigned char vec_popcnt (vector unsigned char)
23538 vector unsigned short vec_popcnt (vector signed short)
23539 vector unsigned short vec_popcnt (vector unsigned short)
23540 vector unsigned int vec_popcnt (vector signed int)
23541 vector unsigned int vec_popcnt (vector unsigned int)
23542 vector unsigned long long vec_popcnt (vector signed long long)
23543 vector unsigned long long vec_popcnt (vector unsigned long long)
23544 vector signed long long vec_slo (vector signed long long,
23545 vector signed char)
23546 vector signed long long vec_slo (vector signed long long,
23547 vector unsigned char)
23548 vector unsigned long long vec_slo (vector unsigned long long,
23549 vector signed char)
23550 vector unsigned long long vec_slo (vector unsigned long long,
23551 vector unsigned char)
23552 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
23553 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
23554 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
23555 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
23556 * doc/extend.texi: Update the built-in documentation file for the
23557 new built-in functions.
23559 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
23561 * attribs.h (sorted_attr_string): Move machine independent
23562 functions for target clone support from the i386 port to common
23563 code. Rename ix86_function_versions to common_function_versions.
23564 Rename make_name to make_unique_name.
23565 (common_function_versions): Likewise.
23566 (make_unique_name): Likewise.
23567 (make_dispatcher_decl): Likewise.
23568 (is_function_default_version): Likewise.
23569 * attribs.c (attr_strcmp): Likewise.
23570 (sorted_attr_string): Likewise.
23571 (common_function_versions): Likewise.
23572 (make_unique_name): Likewise.
23573 (make_dispatcher_decl): Likewise.
23574 (is_function_default_version): Likewise.
23575 * config/i386/i386.c (attr_strcmp): Likewise.
23576 (sorted_attr_string): Likewise.
23577 (ix86_function_versions): Likewise.
23578 (make_name): Likewise.
23579 (make_dispatcher_decl): Likewise.
23580 (is_function_default_version): Likewise.
23581 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
23583 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23586 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
23587 Account for direct move costs for vec_construct of integer
23590 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
23593 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
23594 (UNSPEC_STX_ATOMIC): Ditto.
23595 (loaddi_via_sse): New insn.
23596 (storedi_via_sse): Ditto.
23597 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
23598 Update corresponding peephole2 patterns.
23599 (atomic_storedi_fpu): Ditto.
23601 2017-05-11 Julia Koval <julia.koval@intel.com>
23603 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
23604 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
23606 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
23607 (__builtin_ia32_rsqrt14ss_mask): New builtins.
23608 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
23610 2017-05-11 Nathan Sidwell <nathan@acm.org>
23612 * graphite-poly.c: Include dumpfile.h.
23614 * dumpfle.h (dump_function): Declare here ...
23615 * tree-dump.h (dump_function): ... not here.
23616 * dumpfile.c: #include tree-cfg.h.
23617 (dump_function): Move here from ...
23618 * tree-dump.c (dump_function): ... here.
23619 * gimplify.c: #include splay-tree.h, not tree-dump.h.
23620 * graphite-poly.c: Don't include tree-dump.h.
23621 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
23622 * print-tree.c: Likewise.
23623 * stor-layout.c: Likewise.
23624 * tree-nested.c: Likewise.
23626 * dumpfile.c (dump_start): Use TDF_FLAGS.
23627 (dump_enable_all): Fix TDF_KIND check thinko.
23629 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
23631 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23632 array entries to represent two legal parameterizations of the
23633 overloaded __builtin_cmpb function, as represented by the
23634 P6_OV_BUILTIN_CMPB constant.
23635 (altivec_resolve_overloaded_builtin): Add special case handling
23636 for the __builtin_cmpb function, as represented by the
23637 P6_OV_BUILTIN_CMPB constant.
23638 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
23639 (BU_P6_64BIT_2): New macro.
23640 (BU_P6_OVERLOAD_2): New macro
23641 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
23642 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
23643 (CMPB): Add overload support to represent both 32-bit and 64-bit
23644 compare-bytes function.
23645 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
23646 support for TARGET_CMPB.
23647 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
23648 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23649 documentation of the __builtin_cmpb overloaded built-in function.
23651 2017-05-11 Richard Biener <rguenther@suse.de>
23653 PR tree-optimization/80705
23654 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
23655 bases are not vectorizable.
23657 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23659 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
23660 when counting register pressure.
23662 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23664 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
23665 (iv_ca_more_deps): Renamed to ...
23666 (iv_ca_compare_deps): ... this.
23667 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
23669 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23671 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
23673 (determine_group_iv_costs): ... here.
23674 (find_inv_vars_cb): Record inv var if it's not recorded before.
23676 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23678 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
23679 (get_shiftadd_cost): Ditto.
23681 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23683 * tree-ssa-address.c: Include header file.
23684 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
23686 (add_to_parts): Refactor.
23687 (addr_to_parts): New parameter. Update use of move_hint_to_base.
23688 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
23691 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23693 PR tree-optimization/53090
23694 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
23696 (extract_cond_operands): Detect condition with IV on both sides
23697 and return COMP_IV_EXPR_2.
23698 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
23699 (rewrite_use_compare): Simplify by removing call to function
23700 extract_cond_operands.
23702 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23704 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
23705 (extract_cond_operands): Detect condition comparing against non-
23706 invariant bound and return appropriate enum value.
23707 (find_interesting_uses_cond): Update use of extract_cond_operands.
23708 Handle its return value accordingly.
23709 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
23711 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23713 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
23714 nonlinear iv_use computation in loop invariant sensitive way.
23716 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23718 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
23719 (find_iv_candidates): Call relate_compare_use_with_all_cands.
23721 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23723 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
23724 (dump_cand): Support iv_cand.inv_exprs.
23725 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
23727 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
23730 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23732 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
23734 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
23735 as local function. Include necessary header files.
23736 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
23738 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23740 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
23742 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23744 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
23745 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
23746 RSHIFT_EXPR and BIT_NOT_EXPR.
23748 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23750 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
23751 (adjust_setup_cost): New parameter supporting round up adjustment.
23752 (struct address_cost_data): Delete.
23753 (force_expr_to_var_cost): Don't bound cost with spill_cost.
23754 (split_address_cost, ptr_difference_cost): Delete.
23755 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
23756 (struct ainc_cost_data): New struct.
23757 (get_address_cost_ainc): New function.
23758 (get_address_cost, get_computation_cost): Reimplement.
23759 (determine_group_iv_cost_address): Record inv_expr for all uses of
23761 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
23762 (iv_ca_has_deps): Reimplemented to ...
23763 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
23765 (iv_ca_extend): Call iv_ca_more_deps.
23767 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23769 * tree-ssa-address.c (struct mem_address): Move to header file.
23770 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
23771 * tree-ssa-address.h (struct mem_address): Move from C file.
23772 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
23774 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23776 * tree-affine.h (aff_combination_type): New interface.
23777 (aff_combination_zero_p): Remove static.
23778 (aff_combination_const_p): New interface.
23779 (aff_combination_singleton_var_p): New interfaces.
23781 2017-05-11 Richard Biener <rguenther@suse.de>
23783 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
23784 Skip unreachable blocks and destinations.
23785 (eliminate): Move stmt removal and fixup ...
23786 (fini_eliminate): ... here. Skip inserted exprs.
23787 (pass_pre::execute): Move fini_pre after fini_eliminate.
23788 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
23789 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
23790 PRE to get rid of dead code that has invalid SSA form and
23791 split critical edges again.
23793 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23795 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
23797 2017-05-11 Richard Biener <rguenther@suse.de>
23799 * passes.c (execute_function_todo): Verify loops if they are
23800 said to be up-to-date.
23801 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
23802 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
23804 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
23807 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
23808 handle calling assemble_external ourself.
23811 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
23812 modes with zero size. Enhance comment.
23814 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23816 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
23817 built-ins for vec_xl and vec_xst with short and char pointer
23820 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
23822 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
23823 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
23824 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
23825 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
23826 (_mm_maskz_min_round_ss): New intrinsics.
23827 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
23828 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
23829 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
23830 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
23831 (__builtin_ia32_minss_mask_round): New builtins.
23832 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23833 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
23834 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
23836 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
23837 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
23839 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
23842 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
23844 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
23845 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
23846 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
23847 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
23848 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
23849 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
23850 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
23851 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23852 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
23853 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
23854 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
23855 (__builtin_ia32_mulss_mask_round): New builtins.
23856 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23857 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
23858 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
23860 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
23861 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
23863 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
23866 2017-05-10 Julia Koval <julia.koval@intel.com>
23868 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
23869 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
23870 (_mm256_setr_m128i): New intrinsics.
23872 2017-05-10 Julia Koval <julia.koval@intel.com>
23874 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
23875 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
23876 (_mm_maskz_rcp14_ss): New intrinsics.
23877 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
23878 (__builtin_ia32_rcp14ss_mask): New builtins.
23879 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
23881 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
23883 PR tree-optimization/51513
23884 * tree-cfg.c (gimple_seq_unreachable_p): New function.
23885 (assert_unreachable_fallthru_edge_p): Use it.
23886 (group_case_labels_stmt): Likewise.
23887 * tree-cfg.h: Prototype it.
23888 * stmt.c: Include cfghooks.h and tree-cfg.h.
23889 (emit_case_dispatch_table) <gap_label>: New local variable.
23890 Use it to fill dispatch table gaps.
23891 Test for default_label before updating probabilities.
23892 (expand_case) <default_label>: Remove unneeded initialization.
23893 Test for unreachable default case statement and remove its edge.
23894 Set default_label accordingly.
23895 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
23897 2017-05-10 Carl Love <cel@us.ibm.com>
23899 * config/rs6000/rs6000-c: Add support for built-in functions
23900 vector signed char vec_neg (vector signed char)
23901 vector signed short int vec_neg (vector short int)
23902 vector signed int vec_neg (vector signed int)
23903 vector signed long long vec_neg (vector signed long long)
23904 vector float vec_neg (vector float)
23905 vector double vec_neg (vector double)
23906 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
23908 * config/rs6000/altivec.h: Add define for vec_neg
23909 * doc/extend.texi: Update the built-in documentation for the
23910 new built-in functions.
23912 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23914 PR tree-optimization/77644
23915 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
23917 2017-05-10 Nathan Sidwell <nathan@acm.org>
23919 * dumpfile.h (TDI_lang_all): New.
23920 (TDF_KIND): New. Renumber others
23921 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
23923 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
23925 (get_dump_file_name): Adjust suffix generation.
23926 (dump_enable_all): Use TDF_KIND.
23927 * doc/invoke.texi (-fdump-lang-all): Document.
23929 * dumpfile.h: Tabify.
23931 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
23934 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
23935 Move member access before delete.
23937 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
23939 * tree-inline.c (expand_call_inline): Split block at stmt
23942 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
23945 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
23946 are now unused after splitting mov{sf,sd}_hardfloat.
23947 (f32_lr2): Likewise.
23948 (f32_lm): Likewise.
23949 (f32_lm2): Likewise.
23950 (f32_li): Likewise.
23951 (f32_li2): Likewise.
23952 (f32_lv): Likewise.
23953 (f32_sr): Likewise.
23954 (f32_sr2): Likewise.
23955 (f32_sm): Likewise.
23956 (f32_sm2): Likewise.
23957 (f32_si): Likewise.
23958 (f32_si2): Likewise.
23959 (f32_sv): Likewise.
23960 (f32_dm): Likewise.
23961 (f32_vsx): Likewise.
23962 (f32_av): Likewise.
23963 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
23964 For movsf, order stores so the VSX stores occur before the GPR
23965 store which encourages the register allocator to use a traditional
23966 FPR instead of a GPR. For movsd, order the stores so that the GPR
23967 store comes before the VSX stores to allow the power6 to work.
23968 This is due to the power6 not having a 32-bit integer store
23969 instruction from a FPR.
23970 (movsf_hardfloat): Likewise.
23971 (movsd_hardfloat): Likewise.
23973 2017-05-09 Martin Sebor <msebor@redhat.com>
23975 PR translation/80280
23976 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
23979 PR translation/80280
23980 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
23981 data member added in r247778.
23982 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
23984 2017-05-09 Nathan Sidwell <nathan@acm.org>
23986 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
23988 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
23991 2017-05-09 Marek Polacek <polacek@redhat.com>
23993 * doc/invoke.texi: Fix typo.
23995 2017-05-09 Richard Biener <rguenther@suse.de>
23997 * tree-vrp.c (vrp_val_is_max): Adjust comment.
23998 (vrp_val_is_min): Likewise.
23999 (set_value_range_to_value): Likewise.
24000 (set_value_range_to_nonnegative): Likewise.
24001 (gimple_assign_nonzero_p): Likewise.
24002 (gimple_stmt_nonzero_p): Likewise.
24003 (vrp_int_const_binop): Likewise. Remove unreachable case.
24004 (adjust_range_with_scev): Adjust comments.
24005 (compare_range_with_value): Likewise.
24006 (extract_range_from_phi_node): Likewise.
24007 (test_for_singularity): Likewise.
24009 2017-05-09 Richard Biener <rguenther@suse.de>
24011 * tree-vrp.c (get_single_symbol): Add assert that we don't
24012 get overflowed constants as invariant part.
24013 (compare_values_warnv): Add comment before the TREE_NO_WARNING
24014 checks. Use wi::cmp instead of recursing for integer constants.
24015 (compare_values): Just ignore whether we assumed undefined
24016 overflow instead of failing the compare.
24017 (extract_range_for_var_from_comparison_expr): Add comment before the
24018 TREE_NO_WARNING sets.
24019 (test_for_singularity): Likewise.
24020 (extract_range_from_comparison): Do not disable optimization
24021 when we assumed undefined overflow.
24022 (extract_range_basic): Remove init of unused var.
24024 2017-05-09 Richard Biener <rguenther@suse.de>
24026 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
24027 (extract_range_from_multiplicative_op_1): Adjust.
24028 (extract_range_from_binary_expr_1): Use int_const_binop.
24030 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
24033 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
24034 rs6000_store_data_bypass_p in seven define_bypass directives and
24035 in several comments.
24036 * config/rs6000/rs6000-protos.h: Add prototype for
24037 rs6000_store_data_bypass_p function.
24038 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
24039 function implements slightly different (rs6000-specific) semantics
24040 than store_data_bypass_p, returning false rather than aborting
24041 with assertion error when arguments do not satisfy the
24042 requirements of store data bypass.
24043 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
24044 rs6000_store_data_bypass_p.
24046 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
24048 * config/xtensa/xtensa-protos.h
24049 (xtensa_initial_elimination_offset): New declaration.
24050 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
24051 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
24052 macro definition, add case for FRAME_POINTER_REGNUM when
24053 FRAME_GROWS_DOWNWARD.
24054 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
24055 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
24056 xtensa_initial_elimination_offset.
24058 2017-05-08 Nathan Sidwell <nathan@acm.org>
24060 * doc/invoke.texi: Alphabetize -fdump options.
24062 2017-05-08 Martin Sebor <msebor@redhat.com>
24064 PR translation/80280
24065 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
24067 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
24069 * target.def (compute_frame_layout): New optional target hook.
24070 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
24071 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
24072 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
24074 * reload1.c (verify_initial_elim_offsets): Likewise.
24075 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
24076 (use_simple_return_p): Call arm_compute_frame_layout if needed.
24077 (arm_get_frame_offsets): Split up into this ...
24078 (arm_compute_frame_layout): ... and this function.
24080 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
24082 * config/aarch64/constraints.md (Usa): New constraint.
24083 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
24085 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
24087 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
24088 with_multilib_list after it has been checked.
24090 2017-05-08 Richard Biener <rguenther@suse.de>
24092 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
24093 (bitmap_set_subtract_values): Likewise.
24095 2017-05-08 Richard Biener <rguenther@suse.de>
24097 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
24098 (gimple_assign_nonzero): ... this and remove strict_overflow_p
24100 (gimple_stmt_nonzero_warnv_p): Rename to ...
24101 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
24103 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
24104 (extract_range_basic): Adjust, do not disable propagation on
24105 strict overflow sensitive simplification.
24106 (vrp_visit_cond_stmt): Likewise.
24108 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
24110 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
24111 body size unconditionally.
24113 2017-05-07 Jeff Law <law@redhat.com>
24116 2017-05-06 Jeff Law <law@redhat.com>
24117 PR tree-optimization/78496
24118 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
24121 PR tree-optimization/78496
24122 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
24123 (simplify_stmt_using_ranges): Call it.
24124 (vrp_dom_walker::before_dom_children): Extract equivalences
24125 from an ASSERT_EXPR with an equality comparison against a
24128 2017-05-06 Jeff Law <law@redhat.com>
24130 PR tree-optimization/78496
24131 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
24134 PR tree-optimization/78496
24135 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
24136 (simplify_stmt_using_ranges): Call it.
24137 (vrp_dom_walker::before_dom_children): Extract equivalences
24138 from an ASSERT_EXPR with an equality comparison against a
24141 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
24143 * lra-constraints.c (lra_copy_reg_equiv): New function.
24144 (split_reg): Use it to copy equivalence information from the
24145 original register to the spill register.
24147 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
24149 PR rtl-optimization/75964
24150 * simplify-rtx.c (simplify_const_relational_operation): Remove
24151 invalid handling of comparisons of integer ABS.
24153 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
24155 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
24156 initialize to zero.
24157 (init_regs): Remove declaration.
24158 (function_arg_advance_32): Initialize error_p as boolean variable.
24160 2017-05-05 Nathan Sidwell <nathan@acm.org>
24162 * store-motion.c (remove_reachable_equiv_notes): Reformat long
24163 lines. Use for (;;).
24165 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24167 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
24168 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
24169 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
24170 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
24171 VF=2 that require versioning.
24173 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24175 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
24178 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24180 * diagnostic.h (diagnostic_override_option_index): Convert from
24181 macro to inline function.
24183 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24185 * diagnostic.c (last_module_changed_p): New function.
24186 (set_last_module): New function.
24187 (diagnostic_report_current_module): Convert macro usage to
24188 the above functions.
24189 * diagnostic.h (diagnostic_context::last_module): Strengthen
24190 from const line_map * to const line_map_ordinary *.
24191 (diagnostic_last_module_changed): Delete macro.
24192 (diagnostic_set_last_module): Delete macro.
24194 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24196 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
24197 with diagnostic_report_diagnostic.
24198 (diagnostic_n_impl_richloc): Likewise.
24199 * diagnostic.h (report_diagnostic): Delete macro.
24200 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
24201 with diagnostic_report_diagnostic.
24202 * substring-locations.c (format_warning_va): Likewise.
24204 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24206 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
24207 save/restor of format_spec. Move option-printing code to...
24208 (print_option_information): ...this new function, and
24209 reimplement by simply printing to the pretty_printer,
24210 rather than appending to the format string.
24212 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24214 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
24215 handling logic into...
24216 (update_effective_level_from_pragmas): ...this new function.
24218 2017-05-04 Andrew Waterman <andrew@sifive.com>
24220 * config/riscv/riscv.opt (mstrict-align): New option.
24221 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
24222 (SLOW_UNALIGNED_ACCESS): Define.
24223 (riscv_slow_unaligned_access): Declare.
24224 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
24226 (riscv_slow_unaligned_access): New variable.
24227 (rocket_tune_info): Set slow_unaligned_access to true.
24228 (optimize_size_tune_info): Set slow_unaligned_access to false.
24229 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
24230 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
24231 (riscv_option_override): Set riscv_slow_unaligned_access.
24232 * doc/invoke.texi: Add -mstrict-align to RISC-V.
24234 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
24236 * config/riscv/riscv.md: Unify indentation.
24238 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
24243 * config/rs6000/rs6000.md (u code attribute): Add FIX and
24245 (extendsi<mode>2): Add support for doing sign extension via
24246 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
24247 don't have ISA 3.0 instructions.
24248 (extendsi<mode>2 splitter): Likewise.
24249 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
24250 generate the normal insns since SImode can now go in vector
24251 registers. Disallow the special UNSPECs needed for previous
24252 machines to hide SImode being used. Add new insns
24253 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
24254 (fix_trunc<mode>si2_stfiwx): Likewise.
24255 (fix_trunc<mode>si2_internal): Likewise.
24256 (fixuns_trunc<mode>si2): Likewise.
24257 (fixuns_trunc<mode>si2_stfiwx): Likewise.
24258 (fctiw<u>z_<mode>_smallint): Likewise.
24259 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
24260 of floating point to 32-bit integer from doing a direct move to
24261 the GPR registers to do a store.
24262 (fctiwz_<mode>): Break long line.
24264 2017-05-05 Bin Cheng <bin.cheng@arm.com>
24266 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
24267 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
24268 (addr_list, addr_offset_valid_p): New.
24269 (split_address_groups): Check offset validity with above function.
24270 (gt-tree-ssa-loop-ivopts.h): Include header file.
24272 2017-05-05 Nathan Sidwell <nathan@acm.org>
24274 * config.gcc (arm*-*-*): Add missing 'fi'.
24276 2017-05-05 Steve Ellcey <sellcey@cavium.com>
24278 * doc/invoke.texi (-fopt-info): Explicitly say order of options
24279 included in -fopt-info does not matter.
24280 * doc/optinfo.texi (-fopt-info): Fix description of default
24281 behavour. Explicitly say order of options included in -fopt-info
24284 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
24286 * config.gcc: Allow combinations of aprofile and rmprofile values for
24287 --with-multilib-list.
24288 * config/arm/t-multilib: New file.
24289 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
24290 variables. Remove setting of ISA and floating-point ABI in
24291 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
24292 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
24293 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
24294 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
24296 * config/arm/t-rmprofile: Likewise except for the matches changes.
24297 * doc/install.texi (--with-multilib-list): Document the combination of
24298 aprofile and rmprofile values and warn about pitfalls in doing that.
24300 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
24302 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
24303 (movdi_aarch64): Likewise.
24305 2017-05-05 Jakub Jelinek <jakub@redhat.com>
24307 PR tree-optimization/80632
24308 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
24310 (build_arrays): Initialize it for virtual phis.
24311 (fix_phi_nodes): Use it for virtual phis.
24313 PR tree-optimization/80558
24314 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
24315 [x, y] op z into [x op, y op z] for op & or | if conditions
24318 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
24319 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24322 * config/arm/arm.md (use_literal_pool): Remove.
24323 (64-bit immediate split): No longer takes cost into consideration
24324 if arm_disable_literal_pool is enabled.
24325 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
24326 used when arm_disable_literal_pool is enabled.
24327 (arm_max_const_double_inline_cost): Remove use of
24328 arm_disable_literal_pool.
24329 (push_minipool_fix): Add assert.
24330 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
24331 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
24332 (no_literal_pool_sf_immediate): New.
24334 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24336 PR tree-optimization/80613
24337 * tree-ssa-dce.c (propagate_necessity): Remove cases for
24338 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
24340 2017-05-05 Richard Biener <rguenther@suse.de>
24342 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
24344 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
24346 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
24347 of this flag from insn conditions due to removal from r247495.
24349 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
24351 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
24353 (arm_early_store_addr_dep_ptr): Likewise.
24354 * config/arm/aarch-common-protos.h
24355 (arm_early_load_addr_dep_ptr): Add prototype.
24356 (arm_early_store_addr_dep_ptr): Likewise.
24357 * config/arm/cortex-a53.md: Add new bypasses.
24359 2017-05-05 Jakub Jelinek <jakub@redhat.com>
24361 * tree.c (next_type_uid): Change type to unsigned.
24362 (type_hash_canon): Decrement back next_type_uid if
24363 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
24364 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
24367 2017-05-04 Martin Sebor <msebor@redhat.com>
24369 * builtins.c: Fix a trivial typo in a comment.
24371 PR middle-end/79234
24372 * builtins.c (check_sizes): Adjust to handle reading past the end.
24373 Avoid printing excessive upper bound of ranges. Use %E to print
24374 tree nodes instead of converting them to %wu.
24375 (expand_builtin_memchr): New function.
24376 (compute_dest_size): Rename...
24377 (compute_objsize): ...to this.
24378 (expand_builtin_memcpy): Adjust.
24379 (expand_builtin_mempcpy): Adjust.
24380 (expand_builtin_strcat): Adjust.
24381 (expand_builtin_strcpy): Adjust.
24382 (check_strncat_sizes): Adjust.
24383 (expand_builtin_strncat): Adjust.
24384 (expand_builtin_strncpy): Adjust and simplify.
24385 (expand_builtin_memset): Adjust.
24386 (expand_builtin_bzero): Adjust.
24387 (expand_builtin_memcmp): Adjust.
24388 (expand_builtin): Handle memcmp.
24389 (maybe_emit_chk_warning): Check strncat just once.
24391 2017-05-04 Martin Sebor <msebor@redhat.com>
24393 PR preprocessor/79214
24394 PR middle-end/79222
24395 PR middle-end/79223
24396 * builtins.c (check_sizes): Add inlining context and issue
24397 warnings even when -Wno-system-headers is set.
24398 (check_strncat_sizes): Same.
24399 (expand_builtin_strncat): Same.
24400 (expand_builtin_memmove): New function.
24401 (expand_builtin_stpncpy): Same.
24402 (expand_builtin): Handle memmove and stpncpy.
24404 2017-05-04 Bin Cheng <bin.cheng@arm.com>
24406 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
24407 which is not used any more.
24409 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24411 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
24413 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24415 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
24416 (cortexa53_tunings): Likewise.
24417 (cortexa57_tunings): Likewise.
24418 (cortexa72_tunings): Likewise.
24419 (cortexa73_tunings): Likewise.
24421 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24423 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
24424 Set loop alignment to 8.
24426 2017-05-04 Martin Sebor <msebor@redhat.com>
24428 PR translation/80280
24429 * builtins.c (expand_builtin_object_size): Add missing quoting to
24430 %D and like directives.
24431 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
24432 (hsa_type_for_tree_type): Same.
24433 (verify_function_arguments): Same.
24434 * symtab.c (symbol_table::change_decl_assembler_name): Same.
24435 * varasm.c (get_section): Same.
24438 2017-05-04 Martin Sebor <msebor@redhat.com>
24440 PR translation/80280
24441 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
24443 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24445 * config/aarch64/aarch64.c (generic_addrcost_table):
24446 Change HI/TI mode setting.
24448 2017-05-04 Martin Jambor <mjambor@suse.cz>
24450 PR tree-optimization/80622
24451 * tree-sra.c (comes_initialized_p): New function.
24452 (build_accesses_from_assign): Only set write lazily when
24453 comes_initialized_p is false.
24454 (analyze_access_subtree): Use comes_initialized_p.
24455 (propagate_subaccesses_across_link): Assert !comes_initialized_p
24456 instead of testing for PARM_DECL.
24458 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24460 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
24461 constraint on operand 0 to allow more general addressing modes.
24462 Adjust output template.
24463 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
24465 * config/aarch64/aarch64-protos.h
24466 (aarch64_address_valid_for_prefetch_p): Declare prototype.
24467 * config/aarch64/constraints.md (Dp): New address constraint.
24468 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
24471 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
24473 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
24474 update use of estimate_ipcp_clone_size_and_time.
24475 (estimate_local_effects): Update use of
24476 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
24477 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
24478 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
24479 Return nonspecialized time.
24481 2017-05-04 Richard Biener <rguenther@suse.de>
24483 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
24484 for the last VUSE which def dominates the PHI. Directly call
24486 (get_continuation_for_phi_1): Remove.
24488 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
24490 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
24491 to explain the use of truncating division. Cap the number of
24492 iterations to the maximum given by nb_iterations_upper_bound,
24495 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24497 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
24498 * configure: Regenerate.
24499 * config.in: Regenerate.
24500 * config/i386/driver-mingw32.c: new file.
24501 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
24502 * config.host: Link driver-mingw32.o on MinGW host.
24503 * doc/install.texi: Document new --enable-mingw-wildcard configure
24506 2017-05-04 Marek Polacek <polacek@redhat.com>
24508 PR tree-optimization/80612
24509 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
24511 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24512 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
24514 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
24515 (movt splitter): Likewise.
24516 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
24517 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
24518 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
24519 block for Thumb-1 with MOVT.
24520 (thumb2_legitimate_address_p): Move code block ...
24521 (can_avoid_literal_pool_for_label_p): ... into this new function.
24522 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
24524 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
24525 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
24526 "M-profile targets with the MOVT instruction".
24528 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24530 * config/arm/arm-builtins.c (arm_init_builtins): Rename
24531 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
24532 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
24534 2017-05-04 Martin Liska <mliska@suse.cz>
24536 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
24537 variable cond_code.
24539 2017-05-04 Richard Biener <rguenther@suse.de>
24541 * tree.c (array_at_struct_end_p): Handle arrays at struct
24542 end with flexarrays more conservatively. Refactor and treat
24543 arrays of arrays or aggregates more strict. Fix
24544 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
24545 * tree.c (array_at_struct_end_p): Adjust prototype.
24546 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
24547 * gimple-fold.c (get_range_strlen): Likewise.
24548 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
24550 2017-05-04 Richard Biener <rguenther@suse.de>
24552 PR tree-optimization/31130
24553 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
24555 (supports_overflow_infinity): Likewise.
24556 (is_negative_overflow_infinity): Likewise.
24557 (is_positive_overflow_infinity): Likewise.
24558 (is_overflow_infinity): Likewise.
24559 (stmt_overflow_infinity): Likewise.
24560 (overflow_infinity_range_p): Likewise.
24561 (usable_range_p): Remove as always returning true.
24562 (make_overflow_infinity): Remove.
24563 (negative_overflow_infinity): Likewise.
24564 (positive_overflow_infinity): Likewise.
24565 (avoid_overflow_infinity): Likewise.
24566 (set_value_range): Adjust accordingly.
24567 (set_value_range_to_nonnegative): Likewise, remove now unused
24568 overflow_infinity arg.
24569 (vrp_operand_equal_p): Adjust.
24570 (update_value_range): Likewise.
24571 (range_int_cst_singleton_p): Likewise.
24572 (operand_less_p): Likewise.
24573 (compare_values_warnv): Likewise.
24574 (extract_range_for_var_from_comparison_expr): Likewise.
24575 (vrp_int_const_binop): Likewise.
24576 (zero_nonzero_bits_from_vr): Likewise.
24577 (extract_range_from_multiplicative_op_1): Likewise.
24578 (extract_range_from_binary_expr_1): Likewise.
24579 (extract_range_from_unary_expr): Likewise.
24580 (extract_range_from_comparison): Likewise.
24581 (extract_range_basic): Likewise.
24582 (adjust_range_with_scev): Likewise.
24583 (compare_ranges): Likewise.
24584 (compare_range_with_value): Likewise.
24585 (dump_value_range): Likewise.
24586 (test_for_singularity): Likewise, remove strict_overflow_p parameter
24588 (simplify_cond_using_ranges): Adjust.
24590 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
24592 * brig-builtins.def: Added a builtin for class_f64.
24593 * builtin-types.def: Added a builtin type needed by class_f64.
24595 2017-05-03 Jason Merrill <jason@redhat.com>
24597 * timevar.def: Add TV_CONSTEXPR.
24599 2017-05-03 David Malcolm <dmalcolm@redhat.com>
24601 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
24603 2017-05-03 Martin Jambor <mjambor@suse.cz>
24605 * ipa-prop.c (ipa_update_after_lto_read): Removed.
24606 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
24607 * ipa-cp.c (ipcp_propagate_stage): Do not call
24608 ipa_update_after_lto_read.
24609 * ipa-inline.c (ipa_inline): Likewise.
24611 2017-05-03 Martin Jambor <mjambor@suse.cz>
24613 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
24614 tag. Added a default constructor and a destructor.
24615 (ipa_edge_args_sum_t): New class;
24616 (ipa_edge_args_sum): Declare.
24617 (ipa_edge_args_vector): Remove declaration.
24618 (IPA_EDGE_REF): Use ipa_edge_args_sum.
24619 (ipa_free_edge_args_substructures): Remove declaration.
24620 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
24621 (ipa_edge_args_info_available_for_edge_p): Likewise.
24622 * ipa-prop.c (ipa_edge_args_vector): Removed.
24623 (edge_removal_hook_holder): Likewise.
24624 (edge_duplication_hook_holder): Likewise.
24625 (ipa_edge_args_sum): New variable.
24626 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
24627 ipa_edge_args_vector.
24628 (ipa_free_edge_args_substructures): Likewise.
24629 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
24630 ipa_edge_args_vector.
24631 (ipa_edge_removal_hook): Turned into method
24632 ipa_edge_args_sum_t::remove.
24633 (ipa_edge_duplication_hook): Turned into method
24634 ipa_edge_args_sum_t::duplicate.
24635 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
24636 registering edge hooks.
24637 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
24638 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
24639 ipa_edge_args_sum instead of ipa_edge_args_vector.
24640 * ipa-profile.c (ipa_profile): Likewise.
24642 2017-05-03 Martin Jambor <mjambor@suse.cz>
24644 * symbol-summary.h (function_summary): New method exists.
24645 (function_summary::symtab_removal): Deallocate through release.
24646 (call_summary): New class.
24647 (gt_ggc_mx): New overload.
24648 (gt_pch_nx): Likewise.
24649 (gt_pch_nx): Likewise.
24651 2017-05-03 Jeff Law <law@redhat.com>
24653 PR tree-optimization/78496
24654 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
24655 from simplify_cond_using_ranges. Split off code to walk
24656 backwards through casts into ...
24657 (simplify_cond_using_ranges_2): New function.
24658 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
24659 (execute_vrp): After identifying jump threads, call
24660 simplify_cond_using_ranges_2.
24662 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
24665 * ipa-inline.h (inline_summary): Add ctor.
24666 (create_ggc): Do not use ggc_cleared_alloc.
24668 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
24669 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24671 * gcc.c (handle_braces): Support escaping in switch matching
24673 * doc/invoke.texi (Spec Files): Document it.
24674 Remove superfluous @code markup in items.
24676 2017-05-03 David Malcolm <dmalcolm@redhat.com>
24678 * diagnostic-show-locus.c (struct column_range): New struct.
24679 (get_affected_columns): New function.
24680 (get_printed_columns): New function.
24681 (struct correction): New struct.
24682 (correction::ensure_capacity): New function.
24683 (correction::ensure_terminated): New function.
24684 (struct line_corrections): New struct.
24685 (line_corrections::~line_corrections): New dtor.
24686 (line_corrections::add_hint): New function.
24687 (layout::print_trailing_fixits): Reimplement in terms of the new
24689 (selftest::test_overlapped_fixit_printing): New function.
24690 (selftest::diagnostic_show_locus_c_tests): Call it.
24692 2017-05-03 Nathan Sidwell <nathan@acm.org>
24694 Canonicalize canonical type hashing
24695 * tree.h (type_hash_canon_hash): Declare.
24696 * tree.c (type_hash_list, attribute_hash_list): Move into
24697 type_hash_canon_hash.
24698 (build_type_attribute_qual_variant): Break out hash code calc into
24699 type_hash_canon_hash.
24700 (type_hash_canon_hash): New. Generic type hash computation.
24701 (build_range_type_1, build_array_type_1, build_function_type,
24702 build_method_type_directly, build_offset_type, build_complex_type,
24703 make_vector_type): Call it.
24705 2017-05-03 Richard Biener <rguenther@suse.de>
24707 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
24708 When all DRs have unknown misaligned do not always peel
24709 when there is a store but apply the same costing model as if
24710 there were only loads.
24712 2017-05-03 Richard Biener <rguenther@suse.de>
24715 PR tree-optimization/80492
24716 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
24717 compare_base_decls returning dont-know properly.
24719 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
24721 * config/arm/iterators.md (CCSI): New mode iterator.
24722 (arch): New mode attribute.
24723 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
24724 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
24725 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
24726 code iterator for success result mode.
24727 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
24728 the corresponding new insn generators.
24730 2017-05-03 Bin Cheng <bin.cheng@arm.com>
24733 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24734 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24736 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
24738 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
24739 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
24740 (DDR_A): Wrap DDR argument in brackets.
24741 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
24742 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
24743 (DDR_REVERSED_P): Likewise.
24745 2017-05-03 Jakub Jelinek <jakub@redhat.com>
24747 PR tree-optimization/79472
24748 * tree-switch-conversion.c (struct switch_conv_info): Add
24749 contiguous_range and default_case_nonstandard fields.
24750 (collect_switch_conv_info): Compute contiguous_range and
24751 default_case_nonstandard fields, don't clear final_bb if
24752 contiguous_range and only the default case doesn't have the required
24754 (check_all_empty_except_final): Set default_case_nonstandard instead
24755 of failing if contiguous_range and the default case doesn't have empty
24757 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
24758 and only the default case doesn't have the required constants. Skip
24760 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
24761 if default_case_nonstandard.
24762 (build_constructors): Build constant 1 just once. Assert that default
24763 values aren't inserted in between cases if contiguous_range. Skip
24765 (build_arrays): Skip virtual phis.
24766 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
24767 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
24768 Handle virtual phis.
24769 (gen_inbound_check): Handle default_case_nonstandard case.
24770 (process_switch): Adjust check_final_bb caller. Call
24771 gather_default_values with the first non-default case instead of
24772 default case if default_case_nonstandard.
24774 2017-05-02 Nathan Sidwell <nathan@acm.org>
24776 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
24777 check. Fix formatting.
24779 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
24781 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
24782 errors when comparing specialized and unspecialized times.
24784 2017-05-02 David Malcolm <dmalcolm@redhat.com>
24786 * diagnostic-show-locus.c
24787 (layout::should_print_annotation_line_p): Make private.
24788 (layout::print_annotation_line): Make private.
24789 (layout::annotation_line_showed_range_p): Make private.
24790 (layout::show_ruler): Make private.
24791 (layout::print_source_line): Make private. Pass in line and
24792 line_width, rather than calling location_get_source_line. Drop
24794 (layout::print_leading_fixits): New method.
24795 (layout::print_any_fixits): Rename to...
24796 (layout::print_trailing_fixits): ...this, and make private.
24797 Don't print newline fixits.
24798 (diagnostic_show_locus): Move logic for printing one row into...
24799 (layout::print_line): ...this new function. Move the
24800 location_get_source_line call and error-handling from
24801 print_source_line to here. Call print_leading_fixits, and rename
24802 print_any_fixits to print_trailing_fixits.
24803 (selftest::test_fixit_insert_containing_newline): Update now that
24804 newlines are partially supported.
24805 (selftest::test_fixit_insert_containing_newline_2): New test.
24806 (selftest::test_fixit_replace_containing_newline): Update comments.
24807 (selftest::diagnostic_show_locus_c_tests): Call the new test.
24808 * edit-context.c (class added_line): New class.
24809 (class edited_line): Describe newline handling in comment.
24810 (edited_line::actually_edited_p): New method.
24811 (edited_line::print_content): Delete redundant decl.
24812 (edited_line::m_predecessors): New field.
24813 (edited_file::print_content): Call edited_line::print_content.
24814 (edited_file::print_diff): Update to support newlines.
24815 (edited_file::print_diff_hunk): Likewise.
24816 (edited_file::print_run_of_changed_lines): New function.
24817 (edited_file::print_diff_line): Convert to...
24818 (print_diff_line): ...this.
24819 (edited_file::get_effective_line_count): New function.
24820 (edited_line::edited_line): Initialize new field m_predecessors.
24821 (edited_line::~edited_line): Clean up m_predecessors.
24822 (edited_line::apply_fixit): Handle newlines.
24823 (edited_line::get_effective_line_count): New function.
24824 (edited_line::print_content): New function.
24825 (edited_line::print_diff_lines): New function.
24826 (selftest::test_applying_fixits_insert_containing_newline): New
24828 (selftest::test_applying_fixits_replace_containing_newline): New
24830 (selftest::insert_line): New function.
24831 (selftest::test_applying_fixits_multiple_lines): Add example of
24833 (selftest::edit_context_c_tests): Call the new tests.
24835 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24837 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
24838 parameter cand. Update dump information.
24839 (get_computation_cost): Update uses.
24841 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24843 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
24844 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
24845 (get_computation_at, rewrite_use_address): Update use of
24846 get_computation_aff.
24848 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24850 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
24851 (get_computation): Delete.
24852 (get_computation_cost): Implement like get_computation_cost_at.
24853 Use get_computation_at.
24854 (get_computation_cost_at): Delete.
24855 (rewrite_use_nonlinear_expr): Use get_computation_at.
24856 (rewrite_use_compare, remove_unused_ivs): Ditto.
24858 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24860 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
24862 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24864 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
24865 (ivopts_global_cost_for_size): Rename parameter and update uses.
24866 (iv_ca_recount_cost): Update uses.
24867 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
24868 candidates seperately in n_invs and n_cands.
24869 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
24871 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24873 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
24874 (find_inv_vars_cb): New.
24875 (find_depends): Renamed to ...
24876 (find_inv_vars): ... this.
24877 (add_candidate_1, force_var_cost): Call find_inv_vars.
24878 (split_address_cost, determine_group_iv_cost_cond): Ditto.
24880 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24882 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
24883 inv_vars. Add inv_exprs.
24884 (struct iv_cand): Rename depends_on to inv_vars.
24885 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
24886 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
24887 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
24888 (dump_cand): Dump inv_vars.
24889 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
24890 (record_invariant, find_depends, add_candidate_1): Ditto.
24891 (set_group_iv_cost, force_var_cost): Ditto.
24892 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
24893 (get_computation_cost_at, get_computation_cost): Ditto.
24894 (determine_group_iv_cost_generic): Ditto.
24895 (determine_group_iv_cost_address): Ditto.
24896 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
24897 (determine_group_iv_costs): Ditto.
24898 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
24899 (iv_ca_set_remove_invariants): Renamed to ...
24900 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
24901 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
24902 (iv_ca_set_add_invariants): Renamed to ...
24903 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
24904 (iv_ca_set_cp): Use iv_ca_set_add_invs.
24905 (iv_ca_has_deps): Support inv_vars and inv_exprs.
24906 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
24907 (create_new_ivs): Remove useless dump.
24909 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24911 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
24913 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
24914 (iv_ca_set_no_cp, create_new_iv): Ditto.
24916 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24918 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24920 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
24922 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
24923 function tree_check2.
24925 2017-05-02 Martin Liska <mliska@suse.cz>
24927 * doc/gcov.texi: Add missing preposition.
24928 * gcov.c (function_info::function_info): Properly fill up
24929 all member variables.
24931 2017-05-02 Tamar Christina <tamar.christina@arm.com>
24933 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
24935 2017-05-02 Tamar Christina <tamar.christina@arm.com>
24937 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
24939 2017-05-02 Martin Liska <mliska@suse.cz>
24942 * lto-streamer-in.c (lto_read_tree_1): Remove
24943 LTO_STREAMER_DEBUG.
24944 * lto-streamer.c (struct tree_hash_entry): Likewise.
24945 (struct tree_entry_hasher): Likewise.
24946 (tree_entry_hasher::hash): Likewise.
24947 (tree_entry_hasher::equal): Likewise.
24948 (lto_streamer_init): Likewise.
24949 (lto_orig_address_map): Likewise.
24950 (lto_orig_address_get): Likewise.
24951 (lto_orig_address_remove): Likewise.
24952 * lto-streamer.h: Likewise.
24953 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
24954 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
24956 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
24958 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
24959 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
24960 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
24961 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
24962 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
24963 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
24964 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
24965 (mm_maskz_sub_ss): New intrinsics.
24966 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24967 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
24968 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
24969 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
24970 (__builtin_ia32_subss_mask_round): New builtins.
24971 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24972 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24973 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
24975 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
24976 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
24978 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
24981 2017-05-02 Martin Jambor <mjambor@suse.cz>
24983 PR tree-optimization/78687
24984 * tree-sra.c (access): New field parent.
24985 (process_subtree_disqualification): New function.
24986 (disqualify_candidate): Call it.
24987 (build_accesses_from_assign): Reset write flag if creating an
24989 (build_access_subtree): Fill in parent field and also prpagate
24990 down grp_write flag.
24991 (create_artificial_child_access): New parameter set_grp_write, set
24992 grp_write to its value.
24993 (propagate_subaccesses_across_link): Also propagate grp_write flag
24995 (propagate_all_subaccesses): Push the closest parent back to work
24996 queue if add_access_to_work_queue returned true.
24998 2017-05-02 Richard Biener <rguenther@suse.de>
25000 * common.opt (fstrict-overflow): Alias negative to fwrapv.
25001 * doc/invoke.texi (fstrict-overflow): Remove all traces of
25002 -fstrict-overflow documentation.
25003 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
25004 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
25005 flag_strict_overflow.
25006 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
25007 * lto-opts.c (lto_write_options): Do not stream it.
25008 * lto-wrapper.c (merge_and_complain): Do not handle it.
25009 * opts.c (default_options_table): Do not set -fstrict-overflow.
25010 (finish_options): Likewise do not clear it when sanitizing.
25011 * simplify-rtx.c (simplify_const_relational_operation): Do not
25012 test flag_strict_overflow.
25014 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
25016 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
25017 using enabled attribute.
25018 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
25019 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
25020 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
25021 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
25022 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
25023 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
25024 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
25025 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
25026 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
25027 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
25029 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
25031 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
25033 2017-05-02 Richard Biener <rguenther@suse.de>
25035 PR tree-optimization/80591
25037 2017-04-10 Richard Biener <rguenther@suse.de>
25039 * tree-ssa-structalias.c (find_func_aliases): Properly handle
25042 2017-05-02 Richard Biener <rguenther@suse.de>
25044 PR tree-optimization/80549
25045 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
25046 (cleanup_tree_cfg_noloop): Create forwarders to known loop
25047 headers if they do not have a preheader.
25049 2017-05-02 Martin Liska <mliska@suse.cz>
25052 * common.opt: Fix typo.
25053 * doc/invoke.texi: Likewise.
25055 2017-05-01 Jan Beulich <jbeulich@suse.com>
25057 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
25058 swapping, add (x,x,m,x,n) alternative.
25060 2017-05-01 Nathan Sidwell <nathan@acm.org>
25062 * calls.c (combine_pending_stack_adjustment_and_call): Remove
25063 unnecessary unadjusted_alignment check.
25065 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
25068 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
25070 * gimplify.c (gimplify_cilk_detach): New function.
25071 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
25072 * tree-core.h: Document EXPR_CILK_SPAWN.
25073 * tree.h (EXPR_CILK_SPAWN): Define.
25075 2017-05-01 David Malcolm <dmalcolm@redhat.com>
25077 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
25078 to use new fixit_hint representation, using the "replace" logic.
25079 (get_line_span_for_fixit_hint): Likewise.
25080 (layout::print_any_fixits): Likewise.
25081 (selftest::test_one_liner_many_fixits): Rename to...
25082 (selftest::test_one_liner_many_fixits_1): ...this, and update
25083 comment and expected output to reflect that the multiple fix-it
25084 hints are now consolidated into one insertion.
25085 (selftest::test_one_liner_many_fixits_2): New test.
25086 (selftest::test_diagnostic_show_locus_one_liner): Update for
25088 (selftest::test_fixit_consolidation): Update for fix-it API
25090 * diagnostic.c (print_parseable_fixits): Likewise.
25091 * edit-context.c (edited_line::m_line_events): Convert from
25092 auto_vec <line_event *> to auto_vec <line_event>.
25093 (class line_event): Convert from abstract base class to a concrete
25094 class, taking over the role of replace_event.
25095 (class insert_event): Delete.
25096 (class replace_event): Rename to class line_event. Convert to
25098 (edit_context::add_fixits): Reimplement.
25099 (edit_context::apply_insert): Delete.
25100 (edit_context::apply_replace): Rename to...
25101 (edit_context::apply_fixit): ...this. Convert to half-open range.
25102 (edited_file::apply_insert): Delete.
25103 (edited_file::apply_replace): Rename to...
25104 (edited_file::apply_fixit): ...this.
25105 (edited_line::~edited_line): Drop deletion of events.
25106 (edited_line::apply_insert): Delete.
25107 (edited_line::apply_replace): Rename to...
25108 (edited_line::apply_fixit): ...this. Convert to half-open range.
25109 Update for change to type of m_line_events.
25110 * edit-context.h (edit_context::apply_insert): Delete.
25111 (edit_context::apply_replace): Rename to...
25112 (edit_context::apply_fixit): ...this.
25114 2017-05-01 Martin Sebor <msebor@redhat.com>
25116 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
25119 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
25122 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
25123 __get_cpuid_max returns 0.
25124 (__get_cpuid_count): Ditto.
25126 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
25128 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
25129 replacement expression is another instance of one of its arguments.
25131 2017-05-01 Jakub Jelinek <jakub@redhat.com>
25134 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
25135 check for stack push/pop autoinc.
25136 * config/i386/i386.c (ix86_agi_dependent): Return false
25137 if the only reason why modified_in_p returned true is that
25138 addr is SP based and set_insn is a push or pop.
25140 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
25142 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
25145 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
25148 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
25149 (account_size_time): Use two predicates - exec_pred and
25151 (evaluate_conditions_for_known_args): Compute both clause and
25153 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
25154 (inline_summary_t::duplicate): Update.
25155 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
25157 (compute_inline_parameters): Likewise.
25158 (estimate_edge_size_and_time): Update caluclation of time.
25159 (estimate_node_size_and_time): Compute both time and nonspecialized
25161 (estimate_ipcp_clone_size_and_time): Update.
25162 (inline_merge_summary): Update.
25163 (do_estimate_edge_time): Update.
25164 (do_estimate_edge_size): Update.
25165 (do_estimate_edge_hints): Update.
25166 (inline_read_section, inline_write_summary): Stream both new predicates.
25167 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
25169 (compute_inlined_call_time): Cleanup.
25170 (big_speedup_p): Update.
25171 (edge_badness): Update.
25172 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
25173 (size_time_entry): Replace predicate by exec_predicate and
25174 nonconst_predicate.
25175 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
25176 (estimate_edge_time): Return also nonspec_time.
25177 (reset_edge_growth_cache): Update.
25179 2017-04-29 Jakub Jelinek <jakub@redhat.com>
25181 PR rtl-optimization/80491
25182 * ifcvt.c (noce_process_if_block): When looking for x setter
25183 with missing else_bb, don't check only the insn right before
25184 cond_earliest, but look for the last insn that x is modified in
25185 within the same bb.
25187 PR rtl-optimization/80491
25188 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
25190 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
25192 PR tree-optimization/80487
25193 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
25195 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25197 PR tree-optimization/79697
25198 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
25199 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
25200 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
25202 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
25203 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
25205 2017-04-28 Martin Sebor <msebor@redhat.com>
25207 PR tree-optimization/80523
25208 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
25209 (init_target_to_host_charmap, target_to_host, target_strtol10): New
25211 (maybe_warn, format_directive, parse_directive): Use new functions.
25212 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
25214 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
25216 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
25218 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
25220 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
25221 target_header_dir): Set correctly.
25222 * configure: Regenerated.
25223 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
25224 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
25225 instead of SYSTEM_HEADER_DIR.
25227 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
25229 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
25230 (estimate_local_effects): Likewise.
25231 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
25232 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
25233 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
25234 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
25235 do_estimate_edge_time, estimate_edge_time): Likewise.
25236 * ipa-inline-analysis.c (estimate_node_size_and_time,
25237 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
25238 (estimate_time_after_inlining): Remove.
25240 2017-04-28 Martin Liska <mliska@suse.cz>
25242 * doc/gcov.texi: Enhance documentation of gcov.
25244 2017-04-28 Martin Liska <mliska@suse.cz>
25246 * doc/gcov.texi: Sort options in alphabetic order.
25247 * doc/gcov-dump.texi: Likewise.
25248 * doc/gcov-tool.texi: Likewise.
25249 * gcov.c (print_usage): Likewise.
25250 * gcov-dump.c (print_usage): Likewise.
25251 * gcov-tool.c (print_merge_usage_message): Likewise.
25252 (print_rewrite_usage_message): Likewise.
25253 (print_overlap_usage_message): Likewise.
25255 2017-04-28 Martin Liska <mliska@suse.cz>
25257 PR gcov-profile/53915
25258 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
25260 2017-04-28 Martin Liska <mliska@suse.cz>
25262 PR gcov-profile/79891
25263 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
25264 is marked by compiler as living on a line.
25265 (get_cycles_count): Remove usage of the union.
25266 (output_intermediate_file): Likewise.
25267 (find_source): Fix GNU coding style.
25268 (accumulate_line_counts): Remove old non-all block mode.
25269 (output_lines): Remove usage of the union.
25270 * profile.c (output_location): Include all BBs, even if
25271 belonging to a same line (and file) as a previous BB.
25273 2017-04-28 Martin Liska <mliska@suse.cz>
25275 * gcov.c (process_args): Handle new argument 'w'.
25276 (read_graph_file): Assign ID to BBs.
25277 (output_branch_count): Display BB # if verbose flag is set.
25278 (output_lines): Likewise for arcs.
25279 (print_usage): Add '--verbose' option help.
25280 * doc/gcov.texi: Document --verbose (-w) option.
25282 2017-04-28 Martin Liska <mliska@suse.cz>
25284 * gcov.c (struct block_location_info): New struct.
25285 (process_file): Fill up the new structure.
25286 (read_graph_file): Replace usage of encoding by the newly added
25288 (add_line_counts): Likewise.
25289 (accumulate_line_counts): Remove usage of the union.
25290 (function_info::function_info): New function.
25291 (function_info::~function_info): Likewise.
25292 (process_file): Call delete instead of release_function.
25293 (release_function): Release the function.
25294 (release_structures): Call delete instead of release_function.
25295 (solve_flow_graph): Replace usage of num_blocks.
25296 (find_exception_blocks): Likewise.
25297 (output_lines): Fix GNU coding style.
25299 2017-04-28 Martin Liska <mliska@suse.cz>
25302 * coverage.c (coverage_remove_note_file): New function.
25303 * coverage.h: Declare the function.
25304 * toplev.c (finalize): Clean if an error has been seen.
25306 2017-04-28 Martin Liska <mliska@suse.cz>
25308 PR gcov-profile/80031
25309 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
25310 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
25311 * gcov.c (read_graph_file): Read just number of blocks.
25312 * profile.c (branch_prob): Do not stream 0 flags per a basic
25315 2017-04-28 Martin Liska <mliska@suse.cz>
25317 * gcov-dump.c (tag_*): Add new argument to declarations.
25318 (dump_gcov_file): Likewise.
25319 (tag_blocks): Add and use new argument depth.
25320 (tag_arcs): Likewise.
25321 (tag_lines): Likewise.
25322 (tag_counters): Likewise.
25323 (tag_summary): Likewise.
25324 (dump_working_sets): Use depth to do a proper indentation.
25326 2017-04-28 Jakub Jelinek <jakub@redhat.com>
25329 * cgraph.h (symtab_node::debug_symtab): No longer inline.
25330 * symtab.c (symtab_node::debug_symtab): Move definition here.
25332 2017-04-28 Richard Biener <rguenther@suse.de>
25334 * lto-streamer.h (LTO_major_version): Bump to 7.
25336 2017-04-28 Richard Biener <rguenther@suse.de>
25338 * tree-vrp.c (assert_info): New struct.
25339 (add_assert_info): New helper.
25340 (register_edge_assert_for_2): Refactor to add asserts to a vector
25342 (register_edge_assert_for_1): Likewise.
25343 (register_edge_assert_for): Likewise.
25344 (finish_register_edge_assert_for): New helper actually registering
25345 asserts where live on edge.
25346 (find_conditional_asserts): Adjust.
25347 (find_switch_asserts): Likewise.
25348 (evrp_dom_walker::try_find_new_range): Generalize.
25349 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
25351 2017-04-27 Marek Polacek <polacek@redhat.com>
25354 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
25355 arg10 and arg11 to itype.
25357 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
25359 * doc/extend.texi (Object Size Checking): Improve grammar.
25361 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
25364 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
25365 that the logic for permitting reciprocal estimates matches that
25368 2017-04-27 Jakub Jelinek <jakub@redhat.com>
25371 * tree.c (type_cache_hasher::equal): Only compare
25372 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
25373 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
25374 non-aggregate element types.
25375 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
25376 about the flag on ARRAY_TYPEs in the comment, formatting fix.
25378 2017-04-27 Richard Biener <rguenther@suse.de>
25380 PR middle-end/80533
25381 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
25382 stripping ARRAY_REFs from MEM_EXPR make sure we're not
25383 keeping a reference to a trailing array.
25385 2017-04-27 Richard Biener <rguenther@suse.de>
25387 PR middle-end/80539
25388 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
25389 being in loop-closed SSA form conservatively.
25390 (chrec_fold_multiply_poly_poly): Likewise.
25392 2017-04-27 Tamar Christina <tamar.christina@arm.com>
25394 PR middle-end/79665
25395 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
25396 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
25398 2017-04-27 Jakub Jelinek <jakub@redhat.com>
25401 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
25402 (aarch64_function_arg_alignment): Return unsigned int again, but still
25403 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
25404 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
25405 Don't emit -Wpsabi note.
25406 (aarch64_function_arg_boundary): Likewise.
25407 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
25410 2017-04-26 Nathan Sidwell <nathan@acm.org>
25412 * tree.h (crc32_unsigned_n): Declare.
25413 (crc32_unsigned, crc32_unsigned): Make inline.
25414 * tree.c (crc32_unsigned_bits): Replace with ...
25415 (crc32_unsigned_n): ... this.
25416 (crc32_unsigned, crc32_byte): Remove.
25417 (crc32_string): Remove unnecessary braces.
25419 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
25421 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
25422 * ipa-inline-analysis.c (MAX_TIME): Remove.
25423 (account_size_time): Use sreal for time.
25424 (dump_inline_summary): Update.
25425 (estimate_function_body_sizes): Update.
25426 (estimate_edge_size_and_time): Update.
25427 (estimate_calls_size_and_time): Update.
25428 (estimate_node_size_and_time): Update.
25429 (inline_merge_summary): Update.
25430 (inline_update_overall_summary): Update.
25431 (estimate_time_after_inlining): Update.
25432 (inline_read_section): Update.
25433 (inline_write_summary): Update.
25434 * ipa-inline.c (compute_uninlined_call_time): Update.
25435 (compute_inlined_call_time): Update.
25436 (recursive_inlining): Update.
25437 (inline_small_functions): Update.
25438 (dump_overall_stats): Update.
25439 * ipa-inline.h: Include sreal.h.
25440 (size_time_entry): Turn time to sreal.
25441 (inline_summary): Turn self_time nad time to sreal.
25443 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
25445 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
25447 (sreal::stream_out, sreal::stream_in): New.
25448 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
25450 2017-04-25 Jakub Jelinek <jakub@redhat.com>
25452 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
25455 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
25458 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
25460 (dimode_scalar_chain::compute_convert_gain): Ditto.
25461 (dimode_scalar_chain::make_vector_copies): Ditto.
25462 (dimode_scalar_chain::convert_reg): Ditto.
25463 (dimode_scalar_chain::convert_insn): Ditto.
25464 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
25465 (VI248_AVX512BW_1): New mode iterator.
25466 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
25467 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
25470 2017-04-25 Martin Sebor <msebor@redhat.com>
25472 PR tree-optimization/80497
25473 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
25474 constants are representable in HOST_WIDE_INT.
25475 (parse_directive): Ditto.
25477 2017-04-25 Martin Sebor <msebor@redhat.com>
25480 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
25481 (new_zero_array): Adjust signature.
25482 (dom_info::dom_init): Used unsigned rather that size_t.
25483 (dom_info::dom_info): Same.
25485 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25486 Jakub Jelinek <jakub@redhat.com>
25489 * config/arm/arm.c: Include gimple.h.
25490 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
25491 returns negative, increment ncrn only if it returned positive.
25492 (arm_needs_doubleword_align): Return int instead of bool,
25493 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
25494 members, but if there is any such non-FIELD_DECL
25495 > PARM_BOUNDARY aligned decl, return -1 instead of false.
25496 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
25497 returns negative, increment nregs only if it returned positive.
25498 (arm_setup_incoming_varargs): Likewise.
25499 (arm_function_arg_boundary): Emit -Wpsabi note if
25500 arm_needs_doubleword_align returns negative, return
25501 DOUBLEWORD_ALIGNMENT only if it returned positive.
25503 2017-04-25 Marek Polacek <polacek@redhat.com>
25506 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
25507 first argument to type.
25509 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
25512 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
25513 type checks to test for compatibility instead of equality.
25515 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25516 Jakub Jelinek <jakub@redhat.com>
25519 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
25521 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
25522 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
25523 the alignment computation, but return their maximum in warn_alignment.
25524 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
25525 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
25527 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
25528 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
25531 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25533 * config/arc/simdext.md (dmpyh): Fix typo.
25535 2017-04-25 Richard Biener <rguenther@suse.de>
25537 PR tree-optimization/80492
25538 * alias.c (compare_base_decls): Handle registers with asm
25539 specification conservatively.
25540 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
25541 compare_base_decls returning dont-know properly.
25543 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25545 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
25546 (legitimate_offset_address_p): New function.
25547 (arc_legitimate_address_p): Use above function.
25549 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25551 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
25553 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25555 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
25556 ACCH registers whenever they are available.
25558 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25560 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
25561 double regs fix when not used.
25563 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25565 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
25567 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
25568 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
25570 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25572 * config/arc/arc.c (arc_output_addsi): Check for h-register class
25573 when emitting short ADD instructions.
25575 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25577 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
25579 (cmpsi_cc_c_insn): Likewise.
25580 (cbranchsi4_scratch): Compute proper instruction length using
25581 compact_hreg_operand.
25582 * config/arc/predicates.md (compact_hreg_operand): New predicate.
25584 2017-04-25 Richard Biener <rguenther@suse.de>
25586 PR middle-end/80509
25587 * passes.c (pass_manager::pass_manager): Initialize
25588 m_name_to_pass_map.
25590 2017-04-25 Richard Biener <rguenther@suse.de>
25592 PR tree-optimization/79201
25593 * tree-ssa-sink.c (statement_sink_location): Handle calls.
25595 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25598 * config/s390/vector.md: Split MEM->GPR vector moves for
25599 non-s_operand addresses.
25601 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25604 * config/s390/predicates.md (reload_const_wide_int_operand): New
25606 * config/s390/s390.md ("movti"): Remove d/P alternative.
25607 ("movti_bigconst"): New pattern definition.
25609 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25612 * s390-protos.h (s390_expand_cs_hqi): Removed.
25613 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
25614 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
25615 modes as well as CCZ1mode and CCZmode.
25616 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
25617 signature of s390_emit_compare_and_swap.
25618 (s390_expand_cs_hqi): Likewise, make static.
25619 (s390_expand_cs_tdsi): Generate an explicit compare before trying
25620 compare-and-swap, in some cases.
25621 (s390_expand_cs): Wrapper function.
25622 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
25624 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
25625 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
25626 patterns for small and large integers. Forbid symref memory operands.
25627 Move expander to s390.c. Require cc register.
25628 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
25629 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
25630 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
25631 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
25632 symref memory operands. Remove CC mode and call s390_match_ccmode
25634 ("atomic_exchange<mode>"): Allow and implement all integer modes.
25636 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25638 * config/s390/s390.md (define_peephole2): New peephole to help
25639 combining the load-and-test pattern with volatile memory.
25641 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25643 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
25644 with CCZmode for TARGET_Z196.
25646 2017-04-25 Jakub Jelinek <jakub@redhat.com>
25648 PR rtl-optimization/80501
25649 * combine.c (make_compound_operation_int): Set subreg_code to SET
25650 even for AND with mask of the sign bit of mode.
25652 PR rtl-optimization/80500
25653 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
25654 sum's initial value.
25656 2017-04-25 Julian Brown <julian@codesourcery.com>
25657 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25659 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
25661 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
25663 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
25665 2017-04-25 Julian Brown <julian@codesourcery.com>
25666 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25668 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
25669 (thunderx2t99_sha): New Reservation.
25671 2017-04-25 Julian Brown <julian@codesourcery.com>
25672 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25674 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
25675 type for 1-element load.
25677 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
25679 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
25681 2017-04-24 Martin Jambor <mjambor@suse.cz>
25683 PR tree-optimization/80293
25684 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
25685 char arrays not totally scalarizable if it is false.
25686 (analyze_all_variable_accesses): Pass correct value in the new
25687 parameter. Add a statistics counter.
25689 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
25691 PR middle-end/79931
25692 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
25694 2017-04-24 Richard Biener <rguenther@suse.de>
25696 PR tree-optimization/80494
25697 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
25698 out for complex types.
25700 2017-04-24 Richard Biener <rguenther@suse.de>
25702 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
25703 * tree-ssa-sccvn.c (print_scc): Print SCC size.
25704 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
25705 (DFS): Adjust and never fail.
25706 (sccvn_dom_walker::fail): Remove.
25707 (sccvn_dom_walker::before_dom_children): Adjust.
25708 (run_scc_vn): Likewise and never fail.
25709 * tree-ssa-pre.c (pass_pre::execute): Adjust.
25710 (pass_fre::execute): Likewise.
25712 2017-04-24 Richard Biener <rguenther@suse.de>
25714 PR tree-optimization/79725
25715 * tree-ssa-sink.c (statement_sink_location): Return whether
25716 failure reason was zero uses. Move that check later.
25717 (sink_code_in_bb): Deal with zero uses by removing the stmt
25720 2017-04-24 Richard Biener <rguenther@suse.de>
25723 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
25724 pointer-based references.
25726 2017-04-24 Richard Biener <rguenther@suse.de>
25729 * pass_manager.h (pass_manager::operator new): Remove.
25730 (pass_manager::operator delete): Likewise.
25731 * passes.c (pass_manager::operator new): Remove.
25732 (pass_manager::operator delete): Likewise.
25733 (pass_manager::pass_manager): Zero individual pass members.
25735 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
25738 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
25739 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
25740 Check "XEXP (src, 1)" operand here.
25741 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
25742 Check "XEXP (src, 1)" operand here.
25743 (dimode_scalar_chain::make_vector_copies): Detect count register
25744 of a shift instruction. Zero extend count register from QImode
25745 to DImode to satisfy vector shift pattern count operand predicate.
25746 Substitute vector shift count operand with a DImode copy.
25747 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
25750 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
25752 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
25753 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
25754 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
25755 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
25756 (UNSPEC_NOREX_MEM): Remove definition.
25758 2017-04-21 Richard Biener <rguenther@suse.de>
25760 PR tree-optimization/79547
25761 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
25762 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
25763 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
25764 without any constraints.
25766 2017-04-21 Richard Biener <rguenther@suse.de>
25768 PR tree-optimization/78847
25769 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
25771 2017-04-21 Richard Biener <rguenther@suse.de>
25773 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
25774 (build_distinct_type_copy): Likewise.
25775 (build_variant_type_copy): Likewise.
25776 * tree.c (build_qualified_type): Pass down mem-stat info.
25777 (build_distinct_type_copy): Likewise.
25778 (build_variant_type_copy): Likewise.
25780 2017-04-21 Richard Biener <rguenther@suse.de>
25782 PR tree-optimization/80237
25783 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
25785 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
25786 for a simplified result.
25788 2016-04-21 Richard Biener <rguenther@suse.de>
25790 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
25791 sth as strict as a simple_iv but a chrec without symbols and an
25792 operand defined in the loop we are peeling (and not some subloop).
25793 (propagate_constants_for_unrolling): Propagate all constants.
25795 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
25798 * config/i386/i386.c (print_reg): Remove assert for disalowed
25799 regno values, call output_operand_lossage instead.
25801 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
25804 * config/i386/constraints.md (Yc): New register constraint.
25805 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
25806 Use Yc constraint for alternative 2 of operand 0. Remove
25807 preferred_for_speed attribute.
25809 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
25811 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
25812 lastprivate clauses in SIMT case.
25814 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
25816 * doc/invoke.texi (-Wextra-semi): Document new warning option.
25818 2017-04-20 Richard Biener <rguenther@suse.de>
25820 PR tree-optimization/57796
25821 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
25822 as N scalar stores.
25823 (vect_model_load_cost): Cost gathers as N scalar loads.
25825 2017-04-20 Richard Biener <rguenther@suse.de>
25827 * ggc-page.c (ggc_allocated_p): Rename to ...
25828 (safe_lookup_page_table_entry): ... this and return the lookup
25830 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
25832 2017-04-20 Richard Biener <rguenther@suse.de>
25834 PR tree-optimization/80453
25835 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
25836 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
25837 from the conditions.
25838 (vn_phi_eq): Pass them down.
25839 (vn_phi_lookup): Record them.
25840 (vn_phi_insert): Likewise.
25842 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
25844 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
25845 uninitialized variable warning to avoid buffer overrun.
25847 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
25850 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
25851 is suppressed for '{ 0 }' in C.
25853 2017-04-20 Jakub Jelinek <jakub@redhat.com>
25855 * BASE-VER: Set to 8.0.0.
25857 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
25859 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
25860 priority .init_array and .fini_array section with SECTION_NOTYPE
25863 2017-04-20 Jakub Jelinek <jakub@redhat.com>
25865 PR middle-end/80423
25866 * tree.h (build_array_type): Add typeless_storage default argument.
25867 * tree.c (type_cache_hasher::equal): Also compare
25868 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
25869 (build_array_type): Add typeless_storage argument, set
25870 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
25872 (build_nonshared_array_type): Adjust build_array_type_1 caller.
25873 (build_array_type): Likewise. Add typeless_storage argument.
25875 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
25876 Jakub Jelinek <jakub@redhat.com>
25878 PR tree-optimization/80426
25879 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
25880 operation on symbolic operands, also compute the overflow for the
25881 invariant part when the operation degenerates into a negation.
25883 2017-04-19 Jakub Jelinek <jakub@redhat.com>
25886 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
25887 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
25890 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
25892 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
25895 * config/avr/avr.c (tree.h): Include it.
25896 (cgraph.h): Include it.
25897 (avr_encode_section_info): Don't warn for uninitialized progmem
25898 variable if it's just an alias.
25900 2017-04-19 Richard Biener <rguenther@suse.de>
25903 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
25904 when needed by AutoPGO.
25906 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
25909 * doc/lto.texi: Remove an extra 'that'.
25911 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
25913 PR rtl-optimization/80429
25914 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
25915 are only used in debug insns.
25917 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
25918 Vladimir Makarov <vmakarov@redhat.com>
25920 * config/sparc/predicates.md (input_operand): Add comment. Return
25921 true for any memory operand when LRA is in progress.
25922 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
25924 2017-04-18 Jeff Law <law@redhat.com>
25927 * mips.md ({return,simple_return}_internal): Do not overwrite
25930 2017-04-18 Jakub Jelinek <jakub@redhat.com>
25932 PR tree-optimization/80443
25933 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
25934 instead of adding 1, subtract -1 and similarly instead of subtracting
25937 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
25939 PR rtl-optimization/80357
25940 * haifa-sched.c (tmp_bitmap): New variable.
25941 (model_recompute): Handle duplicate use records.
25942 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
25943 (free_global_sched_pressure_data): Free it.
25945 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
25948 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
25949 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
25950 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
25951 instead of SYSTEM_HEADER_DIR.
25953 2017-04-18 Jeff Law <law@redhat.com>
25955 PR middle-end/80422
25956 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
25957 predecessors after walking up the insn chain.
25959 2017-04-18 Jakub Jelinek <jakub@redhat.com>
25962 * dwarf2out.c (modified_type_die): Try harder not to emit internal
25963 sizetype type into debug info.
25965 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
25968 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
25969 unneeded test for TARGET_UPPER_REGS_SF.
25970 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
25972 2017-04-18 Jakub Jelinek <jakub@redhat.com>
25975 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
25976 instead of gsi_after_labels.
25978 2017-04-18 Jeff Law <law@redhat.com>
25980 * regcprop.c (maybe_mode_change): Avoid creating copies of the
25984 2017-04-13 Jeff Law <law@redhat.com>
25985 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
25986 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
25988 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
25991 * config/avr/avr.c (intl.h): Include it.
25992 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
25994 2017-04-18 Martin Liska <mliska@suse.cz>
25996 PR gcov-profile/78783
25997 * gcov-tool.c (gcov_output_files): Validate that destination
25998 file is either removed by the tool or by a user.
26000 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
26001 Guy Benyei <guybe@mellanox.com>
26003 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
26004 block, and do not negate it, the stored id is already negative.
26006 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
26008 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
26010 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
26013 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
26014 masks of options that should be turned off if the VSX vector
26015 options are turned off.
26016 (OTHER_P8_VECTOR_MASKS): Likewise.
26017 (OTHER_VSX_VECTOR_MASKS): Likewise.
26018 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
26019 rs6000_disable_incompatible_switches to validate no type switches
26021 (rs6000_incompatible_switch): New function to disallow turning on
26022 other vector options if -mno-vsx, -mno-power8-vector, or
26023 -mno-power9-vector are specified.
26025 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26027 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
26029 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26031 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
26032 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
26033 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
26034 (ARG_POINTER_CFA_OFFSET): Likewise.
26036 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26038 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
26039 conditions to take advantage of various optimizations.
26041 2017-04-13 Jeff Law <law@redhat.com>
26043 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
26044 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
26045 (zero_extendsidi2_dext): Likewise.
26047 2017-04-13 Jakub Jelinek <jakub@redhat.com>
26050 * fold-const.c (fold_ternary_loc): Revert
26051 use op0 instead of fold_convert_loc (loc, type, arg0) part of
26054 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
26056 PR rtl-optimization/80343
26057 * lra-remat.c (update_scratch_ops): Assign original hard reg to
26058 new scratch pseudo.
26060 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
26063 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
26064 to ubsan_encode_value.
26066 2017-04-13 Jeff Law <law@redhat.com>
26068 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
26069 appearing in DEBUG_INSNs.
26071 2017-04-13 Martin Liska <mliska@suse.cz>
26073 PR gcov-profile/80413
26074 * gcov-io.c (gcov_write_string): Copy to buffer just when
26075 allocated size is greater than zero.
26077 2017-04-13 Jakub Jelinek <jakub@redhat.com>
26080 * dwarf2out.c (decls_for_scope): Ignore declarations of
26081 current_function_decl in BLOCK_NONLOCALIZED_VARS.
26083 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
26086 * ipa-visibility.c (non_local_p): Fix typos.
26087 (localize_node): When localizing symbol in same comdat group,
26088 dissolve the group only when we know external symbols are going
26090 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
26092 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26094 PR tree-optimization/79390
26095 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
26096 order does not result in usable sequence, retry with reversed operand
26102 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
26103 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
26104 op0 instead of fold_convert_loc (loc, type, arg0).
26106 2017-04-12 Jeff Law <law@redhat.com>
26108 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
26109 has a delay slot in the generated code.
26111 * config/cris/cris.md (cris_preferred_reload_class): Return
26112 GENNONACR_REGS rather than GENERAL_REGS.
26114 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26117 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
26118 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
26119 signedness of the result type.
26121 2017-04-12 Richard Biener <rguenther@suse.de>
26122 Jeff Law <law@redhat.com>
26124 PR tree-optimization/80359
26125 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
26126 trim stores to TARGET_MEM_REFs.
26128 2017-04-12 Richard Biener <rguenther@suse.de>
26130 PR tree-optimization/79390
26131 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
26132 threading case even more.
26134 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
26137 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
26138 for quad_address_p for TImode, instead of just not indexed_address.
26140 2017-04-12 Richard Biener <rguenther@suse.de>
26141 Bernd Edlinger <bernd.edlinger@hotmail.de>
26143 PR middle-end/79671
26144 * alias.c (component_uses_parent_alias_set_from): Handle
26145 TYPE_TYPELESS_STORAGE.
26146 (get_alias_set): Likewise.
26147 * tree-core.h (tree_type_common): Add typeless_storage flag.
26148 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
26149 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
26150 for types containing members with TYPE_TYPELESS_STORAGE.
26151 (place_field): Likewise.
26152 (layout_type): Likewise for ARRAY_TYPE.
26153 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
26154 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
26155 TYPE_TYPELESS_STORAGE.
26156 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
26158 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26161 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
26162 first argument to type.
26164 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26168 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
26169 CONST0_RTX (mode) rather than const0_rtx where appropriate.
26170 (rs6000_expand_binop_builtin): Likewise.
26171 (rs6000_expand_ternop_builtin): Likewise; also add missing
26172 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
26173 vshasigma built-ins.
26174 * doc/extend.texi: Document that vec_xxpermdi's third argument
26175 must be a constant.
26177 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
26179 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
26180 Use shift_const cost parameter when calculating gain of STV shifts.
26182 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
26184 PR rtl-optimization/70478
26185 * lra-constraints.c (process_alt_operands): Check memory for
26186 disfavoring memory insn operand.
26188 2017-04-11 Jakub Jelinek <jakub@redhat.com>
26190 PR middle-end/80100
26191 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
26192 left shift in unsigned HOST_WIDE_INT type.
26194 PR rtl-optimization/80385
26195 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
26196 (not (neg X)) into (plus X -1) for complex or non-integral modes.
26199 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
26200 if they have any depend clauses.
26202 2017-04-11 Martin Liska <mliska@suse.cz>
26205 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
26206 * ipa-split.c (split_function): Create a local comdat symbol
26207 if caller is in a comdat group.
26209 2017-04-11 Martin Liska <mliska@suse.cz>
26212 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
26215 2017-04-11 Martin Sebor <msebor@redhat.com>
26217 PR middle-end/80364
26218 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
26219 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
26221 (directive::set_width, directive::set_precision, format_character):
26223 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
26226 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
26229 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
26230 conflict, set target->arch_name instead of target->cpu_name.
26232 2017-04-11 Richard Biener <rguenther@suse.de>
26234 PR tree-optimization/80374
26235 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
26236 build_zero_cst, remove fold_convertible_p check again.
26238 2017-04-11 Martin Liska <mliska@suse.cz>
26241 * ubsan.c (instrument_object_size): Do not instrument register
26244 2017-04-11 Jakub Jelinek <jakub@redhat.com>
26247 * config/i386/i386-builtin-types.def
26248 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
26249 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
26250 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
26251 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
26252 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
26253 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
26254 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
26255 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
26256 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
26257 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
26258 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
26259 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
26260 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
26261 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
26262 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
26263 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
26264 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
26265 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
26266 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
26267 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
26268 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
26269 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
26270 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
26271 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
26272 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
26273 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
26274 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
26275 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
26276 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
26277 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
26278 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
26279 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
26280 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
26281 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
26282 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
26283 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
26284 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
26285 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
26286 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
26287 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
26288 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
26289 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
26290 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
26291 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
26292 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
26293 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
26295 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
26296 flag to second_arg_count, handle 4 argument function type _COUNT
26297 aliases, handle second_arg_count on second argument rather than last.
26299 2017-04-10 Jeff Law <law@redhat.com>
26301 PR tree-optimization/80374
26302 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
26303 record anything if we can not convert integer_zero_node to the
26306 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
26309 * config/rs6000/rs6000.c (rs6000_option_override_internal):
26310 Enhance special handling given to the TARGET_P9_MINMAX option in
26311 relation to certain other options.
26313 2017-04-10 Bin Cheng <bin.cheng@arm.com>
26315 PR tree-optimization/80153
26316 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
26317 remove POINTER_PLUS_EXPR's base part directly, rather than through
26320 2017-04-10 Richard Biener <rguenther@suse.de>
26321 Bin Cheng <bin.cheng@arm.com>
26323 PR tree-optimization/80153
26324 * tree-affine.c (aff_combination_to_tree): Get base pointer from
26325 the first element of pointer type aff_tree. Build result expr in
26327 (add_elt_to_tree): Convert to type unconditionally. Remove other
26328 fold_convert calls.
26329 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
26330 (rewrite_use_nonlinear_expr): Check invariant using iv information.
26332 2017-04-10 Richard Biener <rguenther@suse.de>
26334 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26337 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
26339 PR rtl-optimization/70478
26340 * lra-constraints.c (curr_small_class_check): New.
26341 (update_and_check_small_class_inputs): New.
26342 (process_alt_operands): Update curr_small_class_check. Disfavor
26343 alternative insn memory operands. Check available regs for small
26346 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
26349 * config/mips/mips.opt (-mvirt): Update description.
26350 * doc/invoke.texi (-mvirt): Likewise.
26352 2017-04-10 Richard Biener <rguenther@suse.de>
26354 PR middle-end/80362
26355 * fold-const.c (fold_binary_loc): Look at unstripped ops when
26356 looking for NEGATE_EXPR in -A / -B to A / B folding.
26358 2017-04-10 Martin Liska <mliska@suse.cz>
26360 PR gcov-profile/80224
26361 * gcov.c (print_usage): Fix usage string.
26362 (get_gcov_intermediate_filename): Remove.
26363 (output_gcov_file): Use both for normal and intermediate format.
26364 (generate_results): Do not initialize special file for
26365 intermediate format.
26367 2017-04-10 Richard Biener <rguenther@suse.de>
26369 PR tree-optimization/80304
26370 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
26373 2017-04-10 Nathan Sidwell <nathan@acm.org>
26376 * config/rs6000/rs6000.c (rs6000_vector_type): New.
26377 (rs6000_init_builtins): Use it.
26379 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26381 * config/arm/arm.md (<mrc>): Add mode to SET source.
26382 (<mrrc>): Likewise.
26384 2017-04-10 Richard Biener <rguenther@suse.de>
26386 PR middle-end/80344
26387 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
26389 2017-04-10 Jakub Jelinek <jakub@redhat.com>
26392 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
26393 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
26394 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
26395 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
26396 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
26397 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
26398 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
26399 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
26400 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
26401 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
26402 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
26403 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
26404 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
26405 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
26406 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
26407 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
26408 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
26409 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
26410 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
26411 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
26412 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
26413 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
26414 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
26416 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
26418 PR rtl-optimization/70478
26419 * lra-constraints.c: Reverse the last patch.
26421 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
26423 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
26424 Add comment for WCHAR_T.
26426 2017-04-08 Martin Liska <mliska@suse.cz>
26429 2017-04-07 Martin Liska <mliska@suse.cz>
26432 * ipa-split.c (split_function): Add function part to a same comdat
26435 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26438 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
26440 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
26442 * rs6000/rs6000.c (vec_load_pendulum): Rename...
26443 (vec_pairing): ...to this.
26444 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
26445 (rs6000_sched_init): Adjust for name change.
26446 (struct rs6000_sched_context): Likewise.
26447 (rs6000_init_sched_context): Likewise.
26448 (rs6000_set_sched_context): Likewise.
26450 2017-04-07 Jakub Jelinek <jakub@redhat.com>
26456 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
26458 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
26459 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
26460 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
26462 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
26464 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
26466 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
26468 PR rtl-optimization/70703
26469 * ira-color.c (update_conflict_hard_regno_costs): Use
26470 int64_t instead of HOST_WIDE_INT.
26472 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
26474 PR rtl-optimization/70478
26475 * lra-constraints.c (process_alt_operands): Disfavor alternative
26476 insn memory operands.
26478 2017-04-07 Jeff Law <law@redhat.com>
26480 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
26481 CALL and NOTE_INSN_CALL_ARG_LOCATION.
26483 2017-04-07 Martin Liska <mliska@suse.cz>
26486 * config/aarch64/aarch64.c (aarch64_process_target_attr):
26487 Show error message instead of an ICE.
26489 2017-04-07 Martin Liska <mliska@suse.cz>
26492 * ipa-split.c (split_function): Add function part to a same comdat
26495 2017-04-07 Richard Biener <rguenther@suse.de>
26497 PR middle-end/80341
26498 * tree.c (get_unwidened): Also handle ! for_type case for
26500 * convert.c (do_narrow): Split out from ...
26501 (convert_to_integer_1): ... here. Do not pass final truncation
26502 type to get_unwidened for TRUNC_DIV_EXPR.
26504 2017-04-07 Richard Biener <rguenther@suse.de>
26506 * tree-affine.c (wide_int_ext_for_comb): Take type rather
26508 (aff_combination_const): Adjust.
26509 (aff_combination_scale): Likewise.
26510 (aff_combination_add_elt): Likewise.
26511 (aff_combination_add_cst): Likewise.
26512 (aff_combination_convert): Likewise.
26513 (add_elt_to_tree): Likewise. Remove unused argument.
26514 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
26516 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
26518 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
26520 * config/arm/arm.c (arm_default_short_enums): Use
26521 ARM_DEFAULT_SHORT_ENUMS.
26522 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
26524 2017-04-06 Jakub Jelinek <jakub@redhat.com>
26527 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
26528 members with redundant out-of-class redeclaration.
26530 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
26533 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
26534 * config/i386/i386.md (*zero_extendsidi2):
26535 Add (?*x,*x) and (?*v,*v) alternatives.
26537 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
26540 * config/i386/i386.c (ix86_expand_builtin)
26541 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
26542 mode from insn data. Convert operands to insn operand mode.
26543 Copy operands that don't satisfy insn predicate to a register.
26545 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
26547 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
26550 2017-04-06 Richard Biener <rguenther@suse.de>
26552 PR tree-optimization/80334
26553 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
26554 preserve alignment of accesses.
26556 2017-04-06 Richard Biener <rguenther@suse.de>
26558 PR tree-optimization/80262
26559 * tree-sra.c (build_ref_for_offset): Preserve address-space
26561 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
26562 Drop useless address-space information on MEM_REF offsets.
26564 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
26566 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
26568 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
26570 PR rtl-optimization/70703
26571 * ira-color.c (update_conflict_hard_regno_costs): Use
26572 HOST_WIDE_INT instead of long.
26574 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
26577 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
26578 not defined for x86_64 target. Add -mmmx target option when __SSE2__
26580 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
26581 for x86_64 target. Handle -m3dnowa option.
26583 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
26585 PR rtl-optimization/70703
26586 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
26587 (update_conflict_hard_regno_costs): Use long instead of unsigned
26588 arithmetic for cost calculation.
26590 2017-04-05 Jakub Jelinek <jakub@redhat.com>
26591 Bernd Edlinger <bernd.edlinger@hotmail.de>
26594 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
26597 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
26600 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
26601 ptr_mode with Pmode throughout.
26602 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
26603 into probe_stack_range and use DImode.
26605 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
26608 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
26609 call_eh_return is true.
26611 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26613 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
26614 Initialize last_match_fntype_index.
26616 2017-04-05 Jakub Jelinek <jakub@redhat.com>
26619 * tree-nvr.c: Include internal-fn.h.
26620 (pass_return_slot::execute): Ignore internal calls without
26623 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26624 Richard Biener <rguenther@suse.de>
26627 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
26628 captures used multiple times, except for the last use.
26629 * generic-match-head.c: Include gimplify.h.
26631 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26633 PR tree-optimization/79390
26634 * target.h (struct noce_if_info): Declare.
26635 * targhooks.h (default_noce_conversion_profitable_p): Declare.
26636 * target.def (noce_conversion_profitable_p): New target hook.
26637 * ifcvt.h (struct noce_if_info): New type, moved from ...
26638 * ifcvt.c (struct noce_if_info): ... here.
26639 (noce_conversion_profitable_p): Renamed to ...
26640 (default_noce_conversion_profitable_p): ... this. No longer
26642 (noce_try_store_flag_constants, noce_try_addcc,
26643 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
26644 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
26645 instead of noce_conversion_profitable_p.
26646 * config/i386/i386.c: Include ifcvt.h.
26647 (ix86_option_override_internal): Don't override
26648 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
26649 (ix86_noce_conversion_profitable_p): New function.
26650 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
26651 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
26652 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
26653 * doc/tm.texi: Regenerated.
26655 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26657 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
26660 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
26663 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
26664 instructions for small multiply cores.
26666 2017-04-04 Jeff Law <law@redhat.com>
26668 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
26670 (mips_expand_vec_perm_const): Initialize elements in orig_perm
26671 that are not set by the loop over the elements.
26673 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26676 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
26677 int mode, convert_modes it to mode as unsigned, otherwise use
26678 lowpart_subreg to mode rather than SImode.
26679 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
26680 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
26681 Use DImode instead of SImode for the shift count operand.
26682 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
26685 2017-04-04 Richard Biener <rguenther@suse.de>
26687 PR middle-end/80281
26688 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
26689 arithmetic done for the negate or the plus. Simplify.
26690 (A - (-B) -> A + B): Likewise.
26691 * fold-const.c (split_tree): Make sure to not negate pointers.
26693 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
26695 PR rtl-optimization/60818
26696 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
26697 a compare of comparisons with the thing compared if this results
26698 in a different machine mode.
26700 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
26702 * alias.c (base_alias_check): Fix typo in comment.
26703 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
26704 * cgraphunit.c (symbol_table::compile): Likewise.
26705 * collect2.c (maybe_run_lto_and_relink): Likewise.
26706 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
26707 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
26708 * config/avr/avr.c (avr_map_op_t): Likewise.
26709 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
26710 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
26711 * config/epiphany/epiphany.md (movcc): Likewise.
26712 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
26713 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
26715 * config/mips/mips.c (mips_save_restore_reg): Likewise.
26716 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
26717 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
26718 * config/sh/sh.c (sh_rtx_costs): Likewise.
26719 * fold-const.c (fold_truth_andor): Likewise.
26720 * genautomata.c (collapse_flag): Likewise.
26721 * gengtype.h (struct type::u::s): Likewise.
26722 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
26723 * input.c (FORMAT_AMOUNT): Likewise.
26724 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
26725 (known_aggs_to_agg_replacement_list): Likewise.
26726 * ipa-inline-analysis.c: Likewise.
26727 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
26728 * ipa-polymorphic-call.c
26729 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
26730 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
26731 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
26733 * modulo-sched.c (apply_reg_moves): Likewise.
26734 * omp-expand.c (build_omp_regions_1): Likewise.
26735 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
26736 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
26737 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
26738 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
26739 * value-prof.c: Likewise.
26740 * var-tracking.c (val_reset): Likewise.
26742 2017-04-03 Richard Biener <rguenther@suse.de>
26744 PR tree-optimization/80275
26745 * fold-const.c (split_address_to_core_and_offset): Handle
26748 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
26750 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
26751 descriptors is at least equal to that of functions.
26753 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
26755 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
26757 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
26760 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
26761 (mov<IMOD4:mode>): New expander.
26762 (*mov<IMOD4:mode>_internal): New insn and split pattern.
26764 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
26766 PR rtl-optimization/79405
26767 * fwprop.c (propagations_left): New variable.
26768 (forward_propagate_into): Decrement it.
26769 (fwprop_init): Initialize it.
26770 (fw_prop): If the variable has reached zero, stop propagating.
26771 (fwprop_addr): Ditto.
26773 2017-03-31 Jakub Jelinek <jakub@redhat.com>
26776 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
26777 a FUNCTION_DECL, pass it as decl instead of origin to
26780 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
26782 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
26785 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
26788 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
26789 TARGET_VSX_SMALL_INTEGER.
26791 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26793 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
26794 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
26796 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
26798 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
26799 extraction from odd-numbered MSA register.
26801 2017-03-31 Jakub Jelinek <jakub@redhat.com>
26803 PR middle-end/80173
26804 * expmed.c (store_bit_field_1): Don't attempt to create
26805 a word subreg out of hard registers wider than word if they
26806 have HARD_REGNO_NREGS of 1 for their mode.
26808 PR middle-end/80163
26809 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
26810 conversions to integer types wider than word and pointer.
26813 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
26814 (rtx_equal_for_cselib_p): Pass 0 to it.
26815 * cselib.c (cselib_hasher::equal): Likewise.
26816 (rtx_equal_for_cselib_1): Add depth argument. If depth
26817 is 128, don't look up VALUE locs and punt. Increment
26818 depth in recursive calls when walking VALUE locs.
26820 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
26822 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
26823 (make_gcov_file_name): Use the canonical path name for generating
26825 (read_line): Fix handling of files with ascii null bytes.
26827 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
26829 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
26830 to initialise a vector register instead
26831 of using a const_int.
26833 2017-03-30 Jakub Jelinek <jakub@redhat.com>
26835 PR translation/80189
26836 * gimplify.c (omp_default_clause): Use %qs instead of %s in
26837 diagnostic messages.
26839 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
26842 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
26843 (dfp_diex_<mode>): Update mode of operand 1.
26844 * doc/extend.texi (dxex, dxexq): Document change to return type.
26845 (diex, diexq): Document change to argument type.
26847 2017-03-30 Martin Jambor <mjambor@suse.cz>
26850 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
26851 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
26852 it reflects the signature changes performed at the callee side.
26853 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
26854 to cgraph_build_function_type_skip_args.
26855 (build_function_decl_skip_args): Adjust call to the above function.
26857 2017-03-30 Jakub Jelinek <jakub@redhat.com>
26860 * config/i386/sse.md
26861 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
26862 register as dest whenever it is a MEM not rtx_equal_p to the
26863 corresponding dup operand, and when forcing into reg move the
26864 reg into the memory afterwards.
26865 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
26866 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
26867 for the force_reg mode.
26868 (avx512vl_vextractf128<mode>): Use register as dest either
26869 always when a MEM, or when it is a MEM not rtx_equal_p to the
26870 corresponding dup operand, or even not when it is a CONST_VECTOR
26871 depending on the mode and lo vs. hi.
26872 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
26874 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
26875 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
26876 Likewise. Require that operands[2] is even.
26877 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
26878 Remove extraneous parens. Require that operands[2] is a multiple
26880 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
26881 operands[0] is a MEM if <mask_applied>, the predicates/constraints
26882 disallow memory then.
26884 2017-03-30 Richard Biener <rguenther@suse.de>
26886 PR tree-optimization/77498
26887 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
26888 to non-constants over backedges.
26890 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
26892 PR rtl-optimization/80233
26893 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
26894 as last_combined_insn. Do not test for BARRIER_P separately.
26896 2017-03-29 Andreas Schwab <schwab@suse.de>
26899 * calls.c (prepare_call_address): Convert funexp to Pmode before
26900 copying to temp reg.
26902 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26904 PR tree-optimization/80158
26905 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
26906 Handle possible future case of more than one alternate
26908 (replace_rhs_if_not_dup): Likewise.
26909 (replace_one_candidate): Likewise.
26911 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
26913 PR rtl-optimization/80193
26914 * ira.c (ira): Do not check allocation for LRA.
26916 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
26918 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
26919 (nvptx_output_simt_exit): Declare.
26920 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
26921 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
26922 (init_softstack_frame): Move initialization of crtl->is_leaf to...
26923 (nvptx_declare_function_name): ...here. Emit declaration of local
26924 memory space buffer for omp_simt_enter insn.
26925 (nvptx_output_unisimt_switch): New.
26926 (nvptx_output_softstack_switch): New.
26927 (nvptx_output_simt_enter): New.
26928 (nvptx_output_simt_exit): New.
26929 * config/nvptx/nvptx.h (struct machine_function): New fields
26930 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
26931 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
26932 (UNSPECV_SIMT_EXIT): Ditto.
26933 (omp_simt_enter_insn): New insn.
26934 (omp_simt_enter): New expansion.
26935 (omp_simt_exit): New insn.
26936 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
26938 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
26939 (expand_GOMP_SIMT_ENTER_ALLOC): New.
26940 (expand_GOMP_SIMT_EXIT): New.
26941 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
26942 (GOMP_SIMT_ENTER_ALLOC): Ditto.
26943 (GOMP_SIMT_EXIT): Ditto.
26944 * target-insns.def (omp_simt_enter): New insn.
26945 (omp_simt_exit): Ditto.
26946 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
26948 (lower_rec_simd_input_clauses): Implement SIMT privatization.
26949 (lower_rec_input_clauses): Likewise.
26950 (lower_lastprivate_clauses): Handle SIMT privatization.
26952 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
26953 (ompdevlow_adjust_simt_enter): New.
26954 (find_simtpriv_var_op): New.
26955 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
26956 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
26958 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
26959 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
26960 (copy_decl_for_dup_finish): Ditto.
26962 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
26964 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
26967 * config/i386/i386.c (ix86_option_override_internal): Always
26968 allow -mpreferred-stack-boundary=3 for 64-bit targets.
26970 2017-03-28 Bin Cheng <bin.cheng@arm.com>
26972 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
26974 2017-03-28 Bin Cheng <bin.cheng@arm.com>
26976 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
26977 mark new edge's irreducible flag accordign to it.
26978 (vect_do_peeling): Check loop preheader edge's irreducible flag
26979 and pass it to function slpeel_add_loop_guard.
26981 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
26983 PR tree-optimization/80218
26984 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
26985 Update block frequencies and counts.
26987 2017-03-28 Richard Biener <rguenther@suse.de>
26989 PR tree-optimization/78644
26990 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
26991 of a simplification result we may not use it at all.
26993 2017-03-28 Richard Biener <rguenther@suse.de>
26996 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
26997 without arguments, generate default definition of a SSA name.
26999 2017-03-28 Richard Biener <rguenther@suse.de>
27001 PR middle-end/80222
27002 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
27003 TYPE_REF_CAN_ALIAS_ALL references.
27004 * fold-const.c (fold_indirect_ref_1): Likewise.
27006 2017-03-28 Martin Liska <mliska@suse.cz>
27009 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
27010 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
27012 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
27013 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
27015 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
27016 (EXTRA_SPECS): Define.
27017 (SUBTARGET_EXTRA_SPECS): Likewise.
27018 (SUBTARGET_CPP_SPEC): Likewise.
27019 * config/arc/elf.h (EXTRA_SPECS): Renamed to
27020 SUBTARGET_EXTRA_SPECS.
27021 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
27023 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
27025 * config/arc/simdext.md (vst64_insn): Update pattern.
27026 (vld32wh_insn): Likewise.
27027 (vld32wl_insn): Likewise.
27028 (vld64_insn): Likewise.
27029 (vld32_insn): Likewise.
27031 2017-03-28 Marek Polacek <polacek@redhat.com>
27034 * fold-const.c (fold_comparison): Use protected_set_expr_location
27035 instead of SET_EXPR_LOCATION.
27037 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
27039 * tree.c (add_expr): Avoid name lookup warning.
27041 2017-03-27 Jeff Law <law@redhat.com>
27043 PR tree-optimization/80216
27044 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
27045 function name. Limit recursion depth.
27046 (record_temporary_equivalences): Corresponding changes.
27048 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
27050 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
27053 2017-03-27 Jakub Jelinek <jakub@redhat.com>
27056 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
27058 * cfgcleanup.c (reg_note_cfa_p): New array.
27059 (insns_have_identical_cfa_notes): New function.
27060 (old_insns_match_p): Don't cross-jump in between /f
27061 and non-/f instructions. If both i1 and i2 are frame related,
27062 verify all CFA notes, their order and content.
27064 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
27067 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
27068 HImode and SImode with zero extend to DImode to one insn.
27069 (bswap<mode>2_extenddi): Likewise.
27070 (bswapsi2_extenddi): Likewise.
27071 (bswaphi2_extendsi): Likewise.
27072 (bswaphi2): Combine bswap HImode and SImode into one insn.
27073 Separate memory insns from swapping register.
27074 (bswapsi2): Likewise.
27075 (bswap<mode>2): Likewise.
27076 (bswaphi2_internal): Delete, no longer used.
27077 (bswapsi2_internal): Likewise.
27078 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
27079 store, and gpr<-gpr swap insns.
27080 (bswap<mode>2_store): Likewise.
27081 (bswaphi2_reg): Register only splitter, combine with the splitter.
27082 (bswaphi2 splitter): Likewise.
27083 (bswapsi2_reg): Likewise.
27084 (bswapsi2 splitter): Likewise.
27085 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
27086 the insns into load, store, and register/register insns.
27087 (bswapdi2_ldbrx): Likewise.
27088 (bswapdi2_load): Likewise.
27089 (bswapdi2_store): Likewise.
27090 (bswapdi2_reg): Likewise.
27092 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
27094 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
27095 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
27097 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
27100 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
27102 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27103 special handling for target option conflicts between dform
27104 options (-mpower9-dform, -mpower9-dform-vector,
27105 -mpower9-dform-scalar) and -mno-direct-move.
27107 2017-03-27 Richard Biener <rguenther@suse.de>
27109 PR tree-optimization/80181
27110 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
27112 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27114 * config/arc/predicates.md (move_double_src_operand): Replace the
27115 call to move_double_src_operand with a call to address_operand.
27117 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27119 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
27120 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
27121 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
27123 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27125 * config/arc/predicates.md (long_immediate_loadstore_operand):
27126 Consider scaled addresses cases.
27128 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27130 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
27131 restored when in interrupt.
27132 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
27133 doesn't have delay slot.
27135 2017-03-27 Richard Biener <rguenther@suse.de>
27138 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
27139 inlined thunk clones.
27141 2017-03-27 Jakub Jelinek <jakub@redhat.com>
27144 * asan.c (instrument_derefs): Copy over last operand from
27145 original COMPONENT_REF to the new COMPONENT_REF with
27146 DECL_BIT_FIELD_REPRESENTATIVE.
27147 * ubsan.c (instrument_object_size): Likewise.
27149 2017-03-27 Richard Biener <rguenther@suse.de>
27151 PR tree-optimization/80170
27152 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
27153 sure DR/SCEV didnt fold in constants we do not see when looking
27154 at the reference base alignment.
27156 2017-03-27 Richard Biener <rguenther@suse.de>
27158 PR middle-end/80171
27159 * gimple-fold.c (fold_ctor_reference): Properly guard against
27160 NULL return value from canonicalize_constructor_val.
27162 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
27165 * config/i386/i386.c (ix86_expand_builtin)
27166 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
27167 flags reg setting and flags reg using instructions.
27168 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
27169 clobbering instructions to zero extend op2.
27171 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
27173 * doc/install.texi (Configuration) <--with-aix-soname>:
27174 Update link to AIX ld.
27176 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
27178 PR rtl-optimization/80160
27179 PR rtl-optimization/80159
27180 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
27181 reg_alternate_class into account.
27183 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
27186 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
27187 to consider in curr_insn_transform.
27189 2017-03-24 Jakub Jelinek <jakub@redhat.com>
27191 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
27192 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
27193 and emit_mode_inner.
27195 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27197 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
27198 argument to the overloaded builtin variants. Use the new flag to
27199 deprecate certain builtin variants.
27200 * config/s390/s390-builtin-types.def: Add new builtin types.
27201 * config/s390/s390-builtins.h: Support new flags field for
27202 overloaded builtins.
27203 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
27204 (s390_macro_to_expand): Enable vector float data type.
27205 (s390_cpu_cpp_builtins_internal): Indicate support of the new
27206 builtins by incrementing the __VEC__ version number.
27207 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
27209 (s390_resolve_overloaded_builtin): Emit error messages depending
27210 on the builtin flags.
27211 * config/s390/s390.c (s390_expand_builtin): Support additional
27212 flags argument. Change error message to match the messages
27213 emitted in s390-c.c.
27214 * config/s390/s390.md: New UNSPEC_* constants.
27215 (op_type): Add new instruction types.
27216 * config/s390/vecintrin.h: Add new builtins and test data class
27218 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
27219 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
27220 (VEC_INEXACT, VEC_NOINEXACT): New constants.
27221 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
27222 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
27223 ("vec_mergel<mode>"): V_HW -> VEC_HW.
27225 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
27226 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
27227 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
27228 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
27230 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
27231 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
27232 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
27233 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
27235 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
27236 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
27237 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
27238 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
27239 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
27240 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
27241 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
27243 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
27244 ("vec_scatter_element<V_HW_4:mode>_DI")
27245 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
27246 ("vec_fpint<mode>", "vflls")
27247 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
27248 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
27249 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
27250 ("*vec_cmphe<mode>_cc"): ... these.
27252 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
27253 mode constant instead of magic value.
27255 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27257 * config/s390/s390.c (s390_expand_vec_compare): Support other
27258 vector floating point modes than just V2DF.
27259 (s390_expand_vcond): Likewise.
27260 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
27261 (s390_cannot_change_mode_class): Prevent mode changes between TF
27262 and V1TF in vector registers.
27263 * config/s390/s390.md (DF, SF): New mode attributes.
27264 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
27265 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
27266 SFmode support for VRs.
27267 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
27269 (VFT, VF_HW): New mode iterators.
27270 (vw, sdx): New mode attributes.
27271 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
27272 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
27273 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
27274 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
27275 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
27276 also the new vector floating point modes. Renaming to ...
27278 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
27279 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
27280 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
27281 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
27282 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
27283 ("vec_unordered<mode>"): ... these.
27285 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
27286 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
27287 ("*vec_extendv2df"): New insn definitions.
27289 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27291 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
27292 ("mulditi3_2", "*muldi3_sign"): New patterns.
27293 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
27294 rename the pattern definition.
27296 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27298 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
27300 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
27302 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27304 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
27305 instruction if possible.
27306 * config/s390/vector.md (vec_halfnumelts): New mode
27308 ("*vec_vllezlf<mode>"): New pattern.
27310 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27312 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
27313 ("popcountv4si2", "popcountv2di2"): Rename to ...
27314 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
27315 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
27317 ("popcount<mode>2_vxe"): New pattern.
27319 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27321 * common/config/s390/s390-common.c (processor_flags_table): Add
27323 * config.gcc: Add arch12.
27324 * config/s390/driver-native.c (s390_host_detect_local_cpu):
27325 Default to arch12 for unknown CPU model numbers.
27326 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
27327 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
27328 PROCESSOR_max sanity check.
27329 * config/s390/s390-opts.h (enum processor_type): Add
27331 * config/s390/s390.c (processor_table): Add arch12.
27332 (s390_expand_builtin): Add check for B_VXE flag.
27333 (s390_issue_rate): Add PROCESSOR_ARCH12.
27334 (s390_get_sched_attrmask): Likewise.
27335 (s390_get_unit_mask): Likewise.
27336 (s390_sched_score): Enable z13 scheduling for arch12.
27337 (s390_sched_reorder): Likewise.
27338 (s390_sched_variable_issue): Likewise.
27339 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
27341 (s390_tune_attr): Use z13 scheduling also for arch12.
27342 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
27343 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
27344 (TARGET_VXE_P): New macros.
27345 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
27346 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
27347 * config/s390/s390.opt: Add arch12 as processor_type.
27349 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27351 * config/s390/s390.md
27352 ("fixuns_truncdddi2", "fixuns_trunctddi2")
27353 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
27354 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
27356 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
27357 Rename expanders to ...
27359 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
27360 ("fixuns_truncdddi2_emu"): ... these.
27362 ("fixuns_trunc<mode>si2_emu"): New expander.
27364 ("*fixuns_truncdfdi2_z13"): Rename to ...
27365 ("*fixuns_truncdfdi2_vx"): ... this.
27367 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27369 * config/s390/2964.md: Remove the single element vector compare
27370 instructions which are no longer used.
27371 * config/s390/s390.c (s390_select_ccmode): Remove handling of
27373 (s390_canonicalize_comparison): Remove handling of DFmode
27375 (s390_expand_vec_compare_scalar): Remove function.
27376 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
27377 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
27379 ("*cmp<mode>_ccs"): Add wfcdb instruction.
27381 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27383 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
27385 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
27386 will anyway by matched by mov<mode>_64dfp.
27388 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27390 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
27391 vlef/vstef. Add missing operand to vleif.
27393 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27395 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
27396 pair for all vector types with 64 bit elements.
27397 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
27398 * config/s390/vector.md (V_HW_64): ... here.
27399 (V_128_NOSINGLE): New mode iterator.
27400 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
27401 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
27402 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
27403 ("*vec_load_pairv2di"): Change to ...
27404 ("*vec_load_pair<mode>"): ... this one.
27406 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27408 * config/s390/constraints.md: Add comments.
27409 (jKK): Reject element sizes > 8 bytes.
27410 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
27412 * config/s390/s390.md: Add the s_operand checks formerly in
27413 s390_split_ok_p to various splitters where they are still
27415 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
27416 for 128 bit vectors. Plus two splitters.
27418 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27420 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
27423 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27426 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
27427 error if the boundary argument is not constant.
27429 2017-03-24 Jakub Jelinek <jakub@redhat.com>
27431 PR rtl-optimization/80112
27432 * loop-doloop.c (doloop_condition_get): Don't check condition
27433 if cmp isn't SET with IF_THEN_ELSE src.
27435 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27437 PR tree-optimization/80158
27438 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
27439 replacing a candidate statement, also replace it for the
27440 candidate's alternate interpretation.
27441 (replace_rhs_if_not_dup): Likewise.
27442 (replace_one_candidate): Likewise.
27444 2017-03-24 Richard Biener <rguenther@suse.de>
27446 PR tree-optimization/80167
27447 * graphite-isl-ast-to-gimple.c
27448 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
27450 (translate_isl_ast_to_gimple::get_rename): Likewise.
27452 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
27454 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
27455 handling of certain combinations of target options, including the
27456 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
27457 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
27459 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27462 * config/arm/arm.md (*load_multiple): Add reload_completed to
27463 matching condition.
27465 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27466 Richard Biener <rguenth@suse.de>
27468 PR tree-optimization/79908
27469 PR tree-optimization/80136
27470 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
27471 been cast away, gimplify_and_add suffices.
27473 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
27475 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
27477 2017-03-23 Richard Biener <rguenther@suse.de>
27479 PR tree-optimization/80032
27480 * gimplify.c (gimple_push_cleanup): Forced unconditional
27481 cleanups still have to go to the conditional_cleanups
27484 2017-03-22 Jakub Jelinek <jakub@redhat.com>
27486 PR tree-optimization/80072
27487 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
27489 (next_operand_entry_id): Change type to unsigned int.
27490 (sort_by_operand_rank): Make sure to return the right return value
27491 even if unsigned fields are bigger than INT_MAX.
27492 (struct oecount): Change cnt and id type to unsigned int.
27493 (oecount_hasher::equal): Formatting fix.
27494 (oecount_cmp): Make sure to return the right return value
27495 even if unsigned fields are bigger than INT_MAX.
27496 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
27499 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
27500 TREE_READONLY on result if writing it more than once.
27503 * doc/invoke.texi (-fsanitize=thread): Document that with
27504 -fnon-call-exceptions atomics are not able to throw
27508 * tsan.c: Include tree-eh.h.
27509 (instrument_builtin_call): Call maybe_clean_eh_stmt or
27510 maybe_clean_or_replace_eh_stmt where needed.
27511 (instrument_memory_accesses): Add cfg_changed argument.
27512 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
27513 if it returned true.
27514 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
27516 PR rtl-optimization/63191
27517 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
27518 wrapper function, moved the whole old content into ...
27519 (ix86_delegitimize_address_1): ... this. New inline function.
27520 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
27521 true as last argument instead of ix86_delegitimize_address.
27523 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
27525 * config/aarch64/aarch64.c (generic_branch_cost): Copy
27526 cortexa57_branch_cost.
27528 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
27530 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
27532 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27535 * doc/md.texi (Constraints): Document wA constraint.
27536 * config/rs6000/constraints.md (wA): New.
27537 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
27538 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
27539 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
27540 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
27542 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
27545 * gimplify.c (is_oacc_declared): New function.
27546 (oacc_default_clause): Use it to set default flags for acc declared
27547 variables inside parallel regions.
27548 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
27549 declared variables.
27550 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
27551 declare attribute to any decl as necessary.
27553 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
27556 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
27557 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
27558 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
27559 (arm_arch_lpae): This.
27560 * config/arm/arm.c (arm_arch7ve): Rename into ...
27561 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
27562 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
27565 2017-03-22 Martin Liska <mliska@suse.cz>
27568 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
27569 error message instead of an ICE.
27571 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27573 * doc/extend.texi (6.11 Additional Floating Types): Revise.
27575 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
27577 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
27579 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27582 2017-03-21 Martin Sebor <msebor@redhat.com>
27584 * doc/extend.texi: Use "cannot" instead of "can't."
27585 * doc/hostconfig.texi: Same.
27586 * doc/install.texi: Same.
27587 * doc/invoke.texi: Same.
27588 * doc/loop.texi: Same.
27589 * doc/md.texi: Same.
27590 * doc/objc.texi: Same.
27591 * doc/rtl.texi: Same.
27592 * doc/tm.texi: Same.
27593 * doc/tm.texi.in: Same.
27594 * doc/trouble.texi: Same.
27596 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
27599 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
27600 (collect_checksum_attributes): Set it.
27601 (die_checksum_ordered): Use it.
27603 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27605 PR tree-optimization/79908
27606 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
27607 change: For a VA_ARG whose LHS has been cast away, use
27608 force_gimple_operand to construct the side effects.
27610 2017-03-21 David Malcolm <dmalcolm@redhat.com>
27612 PR translation/80001
27613 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
27614 more amenable to translation.
27615 (oacc_loop_auto_partitions): Likewise.
27617 2017-03-21 Marek Polacek <polacek@redhat.com>
27618 Martin Sebor <msebor@redhat.com>
27620 PR tree-optimization/80109
27621 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
27622 on INTEGRAL_TYPE_P.
27624 2017-03-21 Jakub Jelinek <jakub@redhat.com>
27625 Segher Boessenkool <segher@kernel.crashing.org>
27628 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
27629 check reg_used_between_p between insn and one of succ or succ2
27630 depending on if succ is artificial insn not inserted into insn
27633 2017-03-21 Martin Liska <mliska@suse.cz>
27635 PR gcov-profile/80081
27636 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
27637 * doc/gcc.texi: Include gcov-dump stuff.
27638 * doc/gcov-dump.texi: New file.
27640 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
27642 PR rtl-optimization/79150
27643 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
27644 conditional jump, if the jump is the last insn of the loop.
27646 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27647 Richard Biener <rguenth@suse.de>
27649 PR tree-optimization/79908
27650 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
27651 been cast away, use force_gimple_operand to construct the side
27654 2017-03-21 Martin Liska <mliska@suse.cz>
27656 PR libfortran/79956
27657 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
27660 2017-03-21 Brad Spengler <spender@grsecurity.net>
27663 * plugin.c (htab_hash_plugin): New function.
27664 (add_new_plugin): Use it and adjust.
27665 (parse_plugin_arg_opt): Adjust.
27666 (init_one_plugin): Likewise.
27668 2017-03-21 Richard Biener <rguenther@suse.de>
27670 PR tree-optimization/80032
27671 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
27672 if set force the cleanup to happen unconditionally.
27673 (gimplify_target_expr): Push inserted clobbers with force_uncond
27674 to avoid them being removed by control-dependent DCE.
27676 2017-03-21 Richard Biener <rguenther@suse.de>
27678 PR tree-optimization/80122
27679 * tree-inline.c (copy_bb): Do not expans va-arg packs or
27680 va_arg_pack_len when the inlined call stmt requires pack
27682 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
27684 2017-03-21 Jakub Jelinek <jakub@redhat.com>
27687 * tsan.c (instrument_builtin_call): If the memory model argument
27688 is not a constant, assume it is valid.
27691 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
27694 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
27696 PR rtl-optimization/79910
27697 * combine.c (can_combine_p): Do not allow combining an I0 or I1
27698 if its dest is used by an insn before I2 (other than the combined
27699 insns themselves, which are properly handled already).
27701 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
27704 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
27706 * combine.c (record_used_regs): New static function.
27707 (try_combine): Handle situations where there is an additional
27708 instruction between I2 and I3 which needs to have a LOG_LINK
27712 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
27714 * combine.c (try_combine): Delete redundant i1 test. Call
27715 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
27717 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27720 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
27721 alternatives 13/14.
27723 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27725 PR tree-optimization/80054
27726 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
27727 the optimization if a PHI or any of its arguments is not dominated
27728 by the candidate's basis. Use gphi* rather than gimple* as
27730 (replace_profitable_candidates): Clean up a gimple* variable that
27731 should be a gphi* variable.
27733 2017-03-20 Martin Sebor <msebor@redhat.com>
27736 * doc/extend.texi (attribute constructor): Document present limitation.
27738 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
27741 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
27742 __POWER9_VECTOR__ #ifdef control, change template definition to
27743 use Power9-specific built-in function.
27744 (vec_any_eq): Likewise.
27745 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
27746 to control outcomes from this test.
27747 (vector_ae_<mode>p): For VEC_F modes, likewise.
27749 2017-03-20 Ian Lance Taylor <iant@google.com>
27751 * config/i386/i386.c (ix86_function_regparm): Save an extra
27752 register for -fsplit-stack with DECL_STATIC_CHAIN.
27754 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
27757 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
27758 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
27760 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
27762 * config/riscv/riscv.c (riscv_print_operand): Use "fence
27764 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
27767 2017-03-20 Marek Polacek <polacek@redhat.com>
27770 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
27772 2017-03-20 Richard Biener <rguenther@suse.de>
27774 PR tree-optimization/80113
27775 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
27776 allocate extra SSA name for PHI def.
27777 (add_close_phis_to_outer_loops): Likewise.
27778 (add_close_phis_to_merge_points): Likewise.
27779 (copy_loop_close_phi_args): Likewise.
27780 (copy_cond_phi_nodes): Likewise.
27782 2017-03-20 Martin Liska <mliska@suse.cz>
27784 PR middle-end/79753
27785 * tree-chkp.c (chkp_build_returned_bound): Do not build
27786 returned bounds for a LHS that's not a BOUNDED_P type.
27788 2017-03-20 Martin Liska <mliska@suse.cz>
27792 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
27793 COMPLEX_CST and VECTOR_CST.
27795 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27798 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
27799 target operand. A new splitter adds the clobber statement in case
27800 the target operand is dead anyway.
27802 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
27804 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
27805 to age-old versions of binutils and glibc.
27807 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
27809 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
27811 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
27813 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
27815 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
27817 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
27818 requirement for binutils 2.13.
27820 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
27822 * combine.c (try_combine): Delete redundant i1 test. Call
27823 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
27825 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
27827 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
27828 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
27830 <riscv64-*-elf>: Re-arrange section
27831 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
27832 <riscv32-*-linux>: Likewise.
27833 <riscv64-*-elf>: Likewise
27834 <riscv64-*-linux>: Likewise.
27836 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
27839 * aarch64.opt(verbose-cost-dump): Fix typo.
27841 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
27844 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
27845 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
27847 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
27849 * reload.c (find_reloads): When reloading a nonoffsettable address,
27850 use RELOAD_OTHER for it and its address reloads.
27852 PR rtl-optimization/79910
27853 * combine.c (record_used_regs): New static function.
27854 (try_combine): Handle situations where there is an additional
27855 instruction between I2 and I3 which needs to have a LOG_LINK
27858 2017-03-17 Jeff Law <law@redhat.com>
27860 PR tree-optimization/71437
27861 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
27862 conditional in the hash table first.
27863 (vrp_dom_walker::before_dom_children): Extract condition from
27864 ASSERT_EXPR. Record condition, its inverion and any implied
27865 conditions as well.
27867 2017-03-17 Marek Polacek <polacek@redhat.com>
27868 Markus Trippelsdorf <markus@trippelsdorf.de>
27870 PR tree-optimization/80079
27871 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
27874 2017-03-17 Richard Biener <rguenther@suse.de>
27876 PR middle-end/80075
27877 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
27878 Properly verify the LHS before the RHS possibly claims to be
27880 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
27883 2017-03-17 Martin Jambor <mjambor@suse.cz>
27885 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
27886 (List of -O2 options): Likewise.
27887 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
27890 2017-03-17 Tom de Vries <tom@codesourcery.com>
27892 * gcov-dump.c (print_usage): Print bug_report_url.
27894 2017-03-17 Richard Biener <rguenther@suse.de>
27896 PR middle-end/80050
27897 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
27898 (parser::peek): Likewise.
27900 2017-03-17 Richard Biener <rguenther@suse.de>
27902 PR tree-optimization/80048
27903 * sese.c (free_sese_info): Properly release rename_map and
27904 copied_bb_map elements.
27906 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
27908 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
27909 Add linked-list forward and backlinks. Insert on
27910 construction, remove on destruction.
27911 (class pass_store_merging): Add m_stores_head field.
27912 (pass_store_merging::terminate_and_process_all_chains):
27913 Iterate over m_stores_head list.
27914 (pass_store_merging::terminate_all_aliasing_chains):
27916 (pass_store_merging::execute): Check for debug stmts first.
27917 Push new chains onto the m_stores_head stack.
27919 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
27922 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
27923 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
27924 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
27926 2017-03-16 Jeff Law <law@redhat.com>
27928 PR tree-optimization/71437
27929 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
27930 member function. Implementation moved into after_dom_children
27931 member function and into the threader's thread_outgoing_edges
27933 (dom_opt_dom_walker::after_dom_children): Simplify by moving
27934 some code into new thread_outgoing_edges.
27935 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
27936 definition. Simplify marker handling (do it here). Assume we always
27937 have the available expression and the const/copies tables.
27938 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
27940 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
27941 * tree-vrp.c (equiv_stack): No longer file scoped.
27942 (vrp_dom_walker): New class.
27943 (vrp_dom_walker::before_dom_children): New member function.
27944 (vrp_dom_walker::after_dom_children): Likewise.
27945 (identify_jump_threads): Setup domwalker. Use it rather than
27946 walking edges in a random order by hand. Simplify setup/finalization.
27947 (finalize_jump_threads): Remove.
27948 (vrp_finalize): Do not call identify_jump_threads here.
27949 (execute_vrp): Do it here instead and call thread_through_all_blocks
27952 PR tree-optimization/71437
27953 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
27955 (simplify_stmt_for_jump_threading): Add basic_block argument. All
27957 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
27958 (dom_opt_dom_walker::thread_across_edge): Remove
27959 handle_dominating_asserts argument. All callers changed.
27960 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
27961 changes. Remove calls to lhs_of_dominating_assert. Other
27962 uses of handle_dominating_asserts turn into unconditional code
27963 (simplify_control_stmt_condition_1): Likewise.
27964 (simplify_control_stmt_condition): Likewise.
27965 (thread_through_normal_block, thread_across_edge): Likewise.
27966 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
27967 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
27968 object if it is not an SSA_NAME.
27969 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
27970 before calling into the VRP specific simplifiers.
27971 (identify_jump_threads): Remove handle_dominating_asserts
27974 2017-03-16 Jakub Jelinek <jakub@redhat.com>
27977 * tree-diagnostic.c (default_tree_printer): No longer static.
27978 * tree-diagnostic.h (default_tree_printer): New prototype.
27980 2017-03-16 Tamar Christina <tamar.christina@arm.com>
27982 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
27983 Change ins into fmov.
27985 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27987 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
27988 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
27989 Use h_con constraint for operand 1.
27990 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
27991 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
27993 2017-03-15 Jeff Law <law@redhat.com>
27995 PR tree-optimization/71437
27996 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
27997 (record_temporary_equivalences): Use it.
27999 PR tree-optimization/71437
28000 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
28001 tree-ssa-scopedtables.
28002 (lookup_avail_expr, build_and_record_new_cond): Likewise.
28003 (record_conditions, record_cond, vuse_eq): Likewise.
28004 (record_edge_info): Adjust to API tweak of record_conditions.
28005 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
28006 (record_temporary_equivalences, optimize_stmt): Likewise.
28007 (eliminate_redundant_computations): Likewise.
28008 (record_equivalences_from_stmt): Likewise.
28009 * tree-ssa-scopedtables.c: Include options.h and params.h.
28010 (vuse_eq): New function, moved from tree-ssa-dom.c
28011 (build_and_record_new_cond): Likewise.
28012 (record_conditions): Likewise. Accept vector of conditions rather
28013 than edge_equivalence structure for first argument.
28014 for the first argument.
28015 (avail_exprs_stack::lookup_avail_expr): New member function, moved
28016 from tree-ssa-dom.c.
28017 (avail_exprs_stack::record_cond): Likewise.
28018 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
28019 from tree-ssa-dom.c.
28020 (avail_exprs_stack): Add new member functions lookup_avail_expr
28022 (record_conditions): Declare.
28024 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
28027 * lra-constraints.c (process_alt_operands): Increase reject for
28028 reloading an input/output operand.
28030 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
28033 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
28034 insns to convert from signed/unsigned char/short to IEEE 128-bit
28036 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
28038 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
28041 * config/i386/i386.c (ix86_vector_duplicate_value): Create
28042 subreg of inner mode for values already in registers.
28044 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
28046 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
28047 iteration reg is used after the loop.
28049 2017-03-14 Martin Sebor <msebor@redhat.com>
28051 PR tree-optimization/79800
28052 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
28053 precision in negative-positive range.
28054 (format_floating): Call non-const overload with adjusted precision.
28056 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
28059 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
28062 2017-03-14 Martin Sebor <msebor@redhat.com>
28064 PR middle-end/80020
28065 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
28066 * builtins.def (aligned_alloc): Use it.
28069 * Makefile.in (GTFILES): Add calls.c.
28070 * calls.c: Include "gt-calls.h".
28072 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
28074 PR rtl-optimization/79728
28075 * regs.h (struct target_regs): New field
28076 x_contains_allocatable_regs_of_mode.
28077 (contains_allocatable_regs_of_mode): New macro.
28078 * reginfo.c (init_reg_sets_1): Initialize it, and change
28079 contains_reg_of_mode so it includes global regs as well.
28080 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
28081 rather than contains_regs_of_mode.
28083 2017-03-14 Martin Liska <mliska@suse.cz>
28085 * doc/invoke.texi: Document options that can't be combined with
28086 -fcheck-pointer-bounds.
28088 2017-03-14 Martin Liska <mliska@suse.cz>
28090 PR middle-end/79831
28091 * doc/invoke.texi (-Wchkp): Document the option.
28093 2017-03-14 Martin Liska <mliska@suse.cz>
28095 * Makefile.in: Install gcov-dump.
28097 2017-03-14 Martin Liska <mliska@suse.cz>
28099 * multiple_target.c (expand_target_clones): Bail out for
28100 an invalid attribute.
28102 2017-03-14 Richard Biener <rguenther@suse.de>
28104 * alias.c (struct alias_set_entry): Pack properly.
28105 * cfgloop.h (struct loop): Likewise.
28106 * cse.c (struct set): Likewise.
28107 * ipa-utils.c (struct searchc_env): Likewise.
28108 * loop-invariant.c (struct invariant): Likewise.
28109 * lra-remat.c (struct cand): Likewise.
28110 * recog.c (struct change_t): Likewise.
28111 * rtl.h (struct address_info): Likewise.
28112 * symbol-summary.h (function_summary): Likewise.
28113 * tree-loop-distribution.c (struct partition): Likewise.
28114 * tree-object-size.c (struct object_size_info): Likewise.
28115 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
28116 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
28117 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
28118 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
28119 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
28120 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
28121 (struct _stmt_vec_info): Likewise.
28123 2017-03-14 Martin Liska <mliska@suse.cz>
28126 * multiple_target.c (create_dispatcher_calls): Check that
28127 a target can create a function dispatcher.
28129 2017-03-14 Martin Liska <mliska@suse.cz>
28132 * multiple_target.c (expand_target_clones): Drop local.local
28133 flag for default implementation.
28135 2017-03-14 Richard Biener <rguenther@suse.de>
28137 PR tree-optimization/80030
28138 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
28140 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
28142 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
28143 gcc_fallthrough() instead of __attribute__((fallthrough));
28145 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
28147 * doc/gcc.texi: Remove "up" link to (DIR).
28148 * doc/gccint.texi: Ditto.
28150 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
28152 * doc/install.texi (Specific) <avr>: Remove reference to
28155 2017-03-13 Jeff Law <law@redhat.com>
28157 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
28158 attribute rather than comments.
28160 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
28161 match_scratch operand is highest.
28163 2017-03-13 Martin Liska <mliska@suse.cz>
28165 PR middle-end/78339
28166 * ipa-pure-const.c (warn_function_noreturn): If the declarations
28167 is a CHKP clone, use original declaration.
28169 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28171 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
28172 (arc_conditional_register_usage): Use a different allocation order
28173 when optimizing for size.
28174 * common/config/arc/arc-common.c (arc_option_optimization_table):
28175 Section anchors default on when optimizing for size.
28177 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28179 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
28181 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28183 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
28184 * config/arc/arc.md (cpu_facility): Add cd variant.
28185 (*movqi_insn): Add code density variant.
28186 (*movhi_insn): Likewise.
28187 (*movqi_insn): Likewise.
28188 (*addsi3_mixed): Likewise.
28189 (subsi3_insn): Likewise.
28191 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28193 * config/arc/arc.md (movsi_cond_exec): Update constraint.
28195 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28197 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
28198 expressions with MINUS and UNARY ops.
28200 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28203 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
28205 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
28206 between vec_select and vector argument.
28207 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
28208 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
28209 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
28210 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
28211 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
28212 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
28214 2017-03-13 Richard Biener <rguenther@suse.de>
28217 * params.def (vect-max-peeling-for-alignment): Fix typo.
28219 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28221 * doc/install.texi (Specific) <mips-*-*>: Remove description of
28222 issue that only occurred with binutils below 2.18.
28224 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28226 * doc/install.texi (Specific) <cris-axis-elf>: No longer
28227 refer to binutils 2.11/2.12 minimum.
28229 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28231 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
28232 ftp.kernel.org and simplify binutils requirement.
28234 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
28236 * doc/invoke.texi (Warning Options): Fix spelling of link-time
28238 (Optimize Options): Ditto. Also remove redundancy.
28240 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28242 PR translation/79848
28243 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
28245 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
28246 to G_ to avoid double translation.
28248 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28250 PR translation/79923
28251 * auto-profile.c (get_combined_location): Convert leading
28252 character of diagnostics to lower case and remove trailing period.
28253 (read_profile): Likewise for various diagnostics.
28254 * config/arm/arm.c (arm_option_override): Remove trailing period
28255 from various diagnostics.
28256 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
28257 (msp430_expand_delay_cycles): Likewise.
28259 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28262 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
28263 full command-line argument, rather than just "str".
28264 (aarch64_validate_march): Likewise.
28265 (aarch64_validate_mtune): Likewise.
28267 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
28269 PR rtl-optimization/78911
28270 * lra-assigns.c (must_not_spill_p): New function.
28271 (spill_for): Use it.
28273 2017-03-10 Jakub Jelinek <jakub@redhat.com>
28275 PR tree-optimization/79981
28276 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
28277 ATOMIC_COMPARE_EXCHANGE ifn result.
28278 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
28279 IFN_ATOMIC_COMPARE_EXCHANGE.
28281 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28284 * opts.c (parse_sanitizer_options): Add missing question mark to
28285 "did you mean" message.
28287 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28289 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
28291 (VMULEUH_UNS): Likewise.
28292 (VMULOUB_UNS): Likewise.
28293 (VMULOUH_UNS): Likewise.
28294 * config/rs6000/rs6000.c (builtin_function_type): Remove
28295 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
28297 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28300 * read-rtl-function.c (function_reader::read_rtx_operand): Update
28301 x with result of extra_parsing_for_operand_code_0.
28302 (function_reader::extra_parsing_for_operand_code_0): Convert
28303 return type from void to rtx, returning x. When reading
28304 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
28305 larger size containing struct block_symbol.
28307 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
28309 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
28310 -mfloat128-hardware without -m64.
28312 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
28315 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
28316 entries to the case statement that marks unsigned arguments to
28317 overloaded functions.
28319 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
28321 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
28322 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
28324 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
28327 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
28328 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
28330 2017-03-10 Martin Liska <mliska@suse.cz>
28334 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
28335 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
28338 2017-03-10 Olivier Hainque <hainque@adacore.com>
28340 * tree-switch-conversion (array_value_type): Start by resetting
28341 candidate type to it's main variant.
28343 2017-03-10 Jakub Jelinek <jakub@redhat.com>
28345 PR rtl-optimization/79909
28346 * combine.c (try_combine): Use simplify_replace_rtx on individual
28347 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
28348 of the whole CALL_INSN_FUNCTION_USAGE.
28350 PR tree-optimization/79972
28351 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
28352 get_range_info on SSA_NAMEs. Formatting fixes.
28354 2017-03-10 Richard Biener <rguenther@suse.de>
28355 Jakub Jelinek <jakub@redhat.com>
28357 PR tree-optimization/77975
28358 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
28359 edge to be constant.
28360 (get_val_for): For constant x return it. Formatting fix.
28361 (loop_niter_by_eval): Avoid pointless looping if the next iteration
28362 would use the same bases as the current one.
28364 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28366 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
28367 instead of vec_select for V1TImode.
28368 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
28370 (VSX_LE_128): Add V1TI to this mode iterator.
28371 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
28372 (*vsx_le_perm_store_<mode>): Likewise.
28373 (pre-reload splitter for VSX stores): Likewise.
28374 (post-reload splitter for VSX stores): Likewise.
28375 (*vsx_xxpermdi2_le_<mode>): Likewise.
28376 (*vsx_lxvd2x2_le_<mode>): Likewise.
28377 (*vsx_stxvd2x2_le_<mode>): Likewise.
28379 2017-03-09 Michael Eager <eager@eagercon.com>
28381 Correct failures with --enable-checking=yes,rtl.
28383 * config/microblaze/microblaze.c (microblaze_expand_shift):
28384 Replace GET_CODE test with CONST_INT_P and INTVAL test with
28385 test for const0_rtx.
28386 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
28387 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
28389 2017-03-09 Richard Biener <rguenther@suse.de>
28391 PR tree-optimization/79977
28392 * graphite-scop-detection.c (scop_detection::merge_sese):
28393 Handle the case of extra exits to blocks dominating the entry.
28395 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
28397 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
28400 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
28402 PR rtl-optimization/79949
28403 * lra-constraints.c (process_alt_operands): Check memory when
28404 trying to predict a cycle. Print about the overall increase.
28406 2017-03-09 Richard Biener <rguenther@suse.de>
28408 PR middle-end/79971
28409 * gimple-expr.c (useless_type_conversion_p): Preserve
28410 TYPE_SATURATING for fixed-point types.
28412 2017-03-09 Richard Biener <rguenther@suse.de>
28415 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
28416 alignment of BLKmode params.
28418 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28421 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
28422 (VALL_NO_V2Q): Likewise.
28424 * config/aarch64/aarch64-simd.md
28425 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
28427 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
28428 VALL_NO_V2Q mode iterator.
28429 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
28431 2017-03-09 Martin Liska <mliska@suse.cz>
28433 PR tree-optimization/79631
28434 * tree-chkp-opt.c (chkp_is_constant_addr): Call
28435 tree_int_cst_sign_bit just for INTEGER constants.
28437 2017-03-09 Martin Liska <mliska@suse.cz>
28441 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
28444 2017-03-09 Marek Polacek <polacek@redhat.com>
28447 * tree.c (inchash::add_expr): Handle TREE_VEC.
28449 2017-03-09 Martin Liska <mliska@suse.cz>
28452 (chkp_narrow_size_and_offset): New function.
28453 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
28454 (void chkp_parse_bit_field_ref): New function.
28455 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
28456 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
28458 2017-03-09 Martin Liska <mliska@suse.cz>
28461 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
28462 (chkp_find_bounds_1): Remove gcc_unreachable.
28464 2017-03-09 Jakub Jelinek <jakub@redhat.com>
28467 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
28468 BUILT_IN_SYNC*, determine the access type from the size suffix and
28469 always build a MEM_REF with that type. Handle forgotten
28470 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
28473 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
28474 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
28475 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
28476 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
28477 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
28478 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
28479 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
28480 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
28481 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
28482 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
28483 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
28484 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
28485 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
28486 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
28487 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
28488 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
28489 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
28490 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
28491 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
28492 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
28493 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
28494 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
28495 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
28496 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
28497 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
28498 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
28499 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
28500 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
28501 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
28502 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
28503 definitions outside of __OPTIMIZE__ guarded section.
28506 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
28507 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
28508 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
28509 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
28512 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28514 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
28515 ("vfenez<mode>"): Add missing constraints.
28517 2017-03-08 Martin Sebor <msebor@redhat.com>
28520 * config/nds32/nds32.c (nds32_option_override):
28521 Fix misspelled diagnostic.
28523 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28526 * gimplify.c (gimplify_omp_for): Replace index var in outer
28527 taskloop statement with an artificial variable and add
28528 OMP_CLAUSE_PRIVATE clause for it.
28530 2017-03-08 Richard Biener <rguenther@suse.de>
28532 PR tree-optimization/79955
28533 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
28534 for accesses that are completely outside of the variable.
28536 2017-03-08 Andrew Haley <aph@redhat.com>
28538 PR tree-optimization/79943
28539 * tree-ssa-loop-split.c (compute_new_first_bound): When
28540 calculating the new upper bound, (END-BEG) should be added, not
28543 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28545 * config/avr/avr.md (setmemhi): Make sure match_dup
28546 operand number comes before match_scratch.
28548 2017-03-08 Richard Biener <rguenther@suse.de>
28550 PR tree-optimization/79920
28551 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
28552 with ncopies == 1 to ...
28553 (vect_transform_slp_perm_load): ... here. Properly compute
28554 all element loads by iterating VF times over the group. Do
28555 not handle ncopies (computed in a broken way) in
28556 vect_create_mask_and_perm.
28558 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28561 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
28562 is a uniform vector, use uniform_vector_p return value instead of
28563 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
28565 2017-03-07 Marek Polacek <polacek@redhat.com>
28567 PR middle-end/79809
28568 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
28569 (alloca_call_type): Likewise.
28571 2017-03-07 Martin Liska <mliska@suse.cz>
28573 * gcov.c (process_args): Put comment to correct location.
28575 2017-03-07 Martin Liska <mliska@suse.cz>
28577 PR middle-end/68270
28578 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
28579 Use array_at_struct_end_p instead of DECL_CHAIN (field).
28580 (chkp_narrow_bounds_for_field): Likewise.
28581 (chkp_parse_array_and_component_ref): Pass one more argument to
28584 2017-03-07 Richard Biener <rguenther@suse.de>
28586 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
28589 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
28591 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
28592 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
28594 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28597 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
28598 to end of description.
28599 (PARAM_MAX_STORES_TO_MERGE): Likewise.
28601 2017-03-07 Jakub Jelinek <jakub@redhat.com>
28603 PR rtl-optimization/79901
28604 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
28606 (*avx512f_<code><mode>3<mask_name>): ... this.
28607 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
28608 iterator instead of VI8_AVX2_AVX512BW.
28610 PR rtl-optimization/79901
28611 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
28612 min/max expander, expand it using expand_vec_cond_expr.
28615 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
28618 2017-03-06 Jakub Jelinek <jakub@redhat.com>
28621 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
28622 to void * for PCH reasons.
28623 * dwarf2out.c (output_loc_operands, output_die): Cast
28624 v.val_vec.array to unsigned char *.
28626 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
28629 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
28632 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
28634 PR rtl-optimization/79571
28635 * lra-constraints.c (process_alt_operands): Calculate static
28636 reject and subtract it from overall when only addresses will be
28639 2017-03-06 Julia Koval <julia.koval@intel.com>
28642 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
28643 incoming stack boundary to 128 for 64-bit targets.
28645 2017-03-06 Richard Biener <rguenther@suse.de>
28647 PR tree-optimization/79894
28648 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
28649 to NULL after folding it.
28651 2017-03-06 Richard Biener <rguenther@suse.de>
28653 PR tree-optimization/79824
28654 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
28655 check disabling peeling for gaps.
28657 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
28659 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
28660 attributes): Document gettimeofday.
28662 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
28664 * config/s390/s390.c (s390_option_override_internal): Set
28665 PARAM_MIN_VECT_LOOP_BOUND
28667 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
28669 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
28670 * config/s390/s390.md: Likewise.
28672 2017-03-06 Jakub Jelinek <jakub@redhat.com>
28675 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
28676 (<avx2_avx512>_perm<mode>): Rename to ...
28677 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
28679 (<avx512>_perm<mode>_mask): Rename to ...
28680 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
28682 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
28683 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
28684 instead of VI8F_256_512.
28685 (avx512f_perm<mode>): New define_expand.
28686 (avx512f_perm<mode>_mask): Likewise.
28687 (avx512f_perm<mode>_1<mask_name>): New define_insn.
28688 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
28690 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28692 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
28693 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
28695 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
28697 2017-03-06 Martin Liska <mliska@suse.cz>
28700 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
28701 when having a SSA NAME w/o VAR_DECL assigned to it.
28703 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28705 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
28706 msa_dpsub_<su>_d): Fix MODE for vec_select.
28708 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28710 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
28712 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
28714 2017-03-06 Richard Biener <rguenther@suse.de>
28716 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
28717 * plugin.c (register_plugin_info): Likewise.
28718 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
28720 2017-03-05 Jakub Jelinek <jakub@redhat.com>
28722 * config/i386/sse.md (sse_storehps, sse_storelps,
28723 avx_<castmode><avxsizesuffix>_<castmode>,
28724 avx512f_<castmode><avxsizesuffix>_<castmode>,
28725 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
28726 in condition that at least one operand is not a MEM.
28728 2017-03-03 Jakub Jelinek <jakub@redhat.com>
28730 PR middle-end/79805
28731 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
28732 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
28734 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
28735 gimple_call_nothrow_p flag based on whether original builtin can throw.
28736 If it can, emit following stmts on the fallthrough edge.
28737 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
28738 don't create new bb if inserting just debug stmts on the edge, try to
28739 insert them on the fallthru bb or just reset debug stmts.
28741 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
28744 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
28745 restore recog_data (including the operand rtxes inside it) around
28746 the call to get_insn_template.
28748 2017-03-03 Martin Sebor <msebor@redhat.com>
28750 PR tree-optimization/79699
28751 * context.c (context::~context): Free MPFR caches to avoid
28752 a memory leak on program exit.
28754 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28756 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
28757 Use wide_int::ulow () instead of .elt (0).
28759 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
28761 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
28762 (*pushxf): Limit oF constraint to 32bit targets and add oC
28763 constraint for 64bit targets.
28764 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
28765 (*pushdf): Change rmF constraint to rmC.
28767 2017-03-03 Martin Liska <mliska@suse.cz>
28769 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
28770 Remove unused variable.
28772 2017-03-03 Jakub Jelinek <jakub@redhat.com>
28775 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
28776 is a memory operand, increase num_memory.
28777 (ix86_expand_args_builtin): Likewise.
28779 2017-03-03 Jan Hubicka <jh@suse.cz>
28782 * ipa-devirt.c (maybe_record_node): Properly handle
28783 __cxa_pure_virtual visibility.
28785 2017-03-03 Martin Liska <mliska@suse.cz>
28787 PR tree-optimization/79803
28788 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
28790 (pass_loop_prefetch::execute): Disabled optimization if an
28791 assumption about L1 cache size is not met.
28793 2017-03-03 Martin Liska <mliska@suse.cz>
28795 PR rtl-optimization/79574
28796 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
28797 (hash_scan_set): Likewise.
28798 (dump_hash_table): Likewise.
28799 (hoist_code): Likewise.
28801 2017-03-03 Richard Biener <rguenther@suse.de>
28803 * fixed-value.c (fixed_from_string): Restore use of elt (1)
28804 in place of uhigh ().
28805 (fixed_convert_from_real): Likewise.
28807 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
28810 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
28812 2017-03-03 Richard Biener <rguenther@suse.de>
28814 PR middle-end/79818
28815 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
28816 TYPE_OVERFLOW_UNDEFINED check.
28818 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28820 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
28822 (vector_ae_<mode>_p): Likewise.
28823 (vector_nez_<mode>_p): Likewise.
28824 (vector_ne_v2di_p): Likewise.
28825 (vector_ae_v2di_p): Likewise.
28826 (vector_ne_<mode>_p): Likewise.
28827 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
28829 (vsx_tsqrt<mode>2_fe): Likewise.
28831 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
28834 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
28836 2017-03-02 Jakub Jelinek <jakub@redhat.com>
28838 PR rtl-optimization/79780
28839 * cprop.c (one_cprop_pass): When second and further conditional trap
28840 in a single basic block is turned into an unconditional trap, turn it
28841 into a deleted note to avoid RTL verification failures.
28843 2017-03-02 Richard Biener <rguenther@suse.de>
28845 * fold-const.c (const_binop): Use ulow () instead of elt (0).
28847 2017-03-02 Richard Biener <rguenther@suse.de>
28849 PR tree-optimization/79345
28851 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
28852 param and abort the walk, returning -1 if it is hit.
28853 (walk_aliased_vdefs): Take a limit param and pass it on.
28854 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
28855 defaulting to 0 and return a signed int.
28856 * tree-ssa-uninit.c (struct check_defs_data): New struct.
28857 (check_defs): New helper.
28858 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
28859 about uninitialized memory.
28860 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
28861 bogus uninitialized warning.
28862 (fixed_convert_from_real): Likewise.
28864 2017-03-02 Bin Cheng <bin.cheng@arm.com>
28866 PR tree-optimization/66768
28867 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
28868 iv_use if base object can't be determined.
28870 2017-03-02 Jakub Jelinek <jakub@redhat.com>
28872 PR tree-optimization/79345
28873 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
28874 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
28875 (get_pattern_stats): Initialize it.
28876 * genemit.c (gen_expand): Verify match_scratch numbers come after
28877 match_operand/match_dup numbers.
28878 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
28879 match_scratch numbers.
28880 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
28882 * config/s390/s390.md (trunctdsd2): Likewise.
28884 2017-03-02 Richard Biener <rguenther@suse.de>
28886 * wide-int.h (wide_int_storage::operator=): Implement in terms
28889 2017-03-02 Richard Biener <rguenther@suse.de>
28891 PR tree-optimization/79777
28892 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
28893 the to insert expression to sth existing.
28895 2017-03-01 Martin Sebor <msebor@redhat.com>
28897 PR middle-end/79692
28898 * gimple-ssa-sprintf.c
28899 (directive::known_width_and_precision): New function.
28900 (format_integer): Use it.
28901 (get_mpfr_format_length): Consider the full range of precision
28902 when computing %g output with the # flag. Set the likely byte
28903 count to 3 rather than 1 when precision is indeterminate.
28904 (format_floating): Correct the lower bound of precision.
28906 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28908 * doc/invoke.texi: Document default code model for 64-bit Linux.
28910 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28913 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
28914 udiv rather than div since input pattern is unsigned.
28916 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
28918 * config/i386/i386.c (print_reg): Warn for values of
28919 unsupported size in integer register.
28921 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
28924 * config/rs6000/predicates.md (current_file_function_operand): Do
28925 not allow self calls to be local if the function is replaceable.
28927 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
28930 * config/rs6000/altivec.h (vec_ctz and others): Change the
28931 preprocessor macro that controls conditional compilation from
28932 _ARCH_PWR9 to __POWER9_VECTOR__.
28933 (vec_all_ne): Change parameterization of __altivec_scalar_pred
28934 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
28935 control (instead of _ARCH_PWR9 control) so that template
28936 definition uses power9-specific function.
28937 (vec_any_eq): Likewise.
28938 (vec_all_ne): Change macro definition to use a power9-specific
28939 expansion under #ifdef __POWER9_VECTOR__ control (instead of
28940 _ARCH_PWR9 control).
28941 (vec_any_eq) Likewise.
28942 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
28943 expansion for CMPNEF to remove support for xvcmpnesp instruction.
28944 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
28945 support for xvcmpnedp instruction.
28946 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
28947 macro expansion so that Power9 implementation of vec_all_ne does
28948 not use the AltiVec predicate framework.
28949 (VCMPNEH_P): Likewise.
28950 (VCMPNEW_P): Likewise.
28951 (VCMPNED_P): Likewise.
28952 (VCMPNEFP_P): Likewise.
28953 (VCMPNEDP_P): Likewise.
28954 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
28955 implementation of vec_any_eq to not use AltiVec predicate
28957 (VCMPAEH_P): Likewise.
28958 (VCMPAEW_P): Likewise.
28959 (VCMPAED_P): Likewise.
28960 (VCMPAEFP_P): Likewise.
28961 (VCMPAEDP_P): Likewise.
28962 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
28963 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
28964 not use the AltiVec predicate framework.
28965 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
28966 of vec_any_eq to not use AltiVec predicate framework.
28967 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
28968 support for predefined __POWER9_VECTOR__ macro to indicate that
28969 Power9 instruction selection is enabled.
28970 (altivec_overloaded_builtins): Remove extraneous
28971 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
28972 function argument types RS6000_BTI_bool_V16QI and
28973 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
28974 entry for overloaded function argument types RS6000_BTI_bool_V4SI
28975 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
28976 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
28977 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
28978 Power9 for implementations of vec_cmpne. Change the signature for
28979 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
28980 (representing vec_all_ne) to remove the previously described first
28981 argument of type RS6000_BTI_INTSI, as this was an artifact of
28982 reliance on the AltiVec predicate framework, which is no longer
28983 used in the implementation of these functions. Add
28984 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
28985 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
28986 since, unlike the AltiVec predicate framework implementation, we
28987 do not share function descriptors between vec_alle and vec_anyeq.
28988 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
28989 set of modes that receive special treatment even when
28990 TARGET_P9_VECTOR is true. The special treatment emits code that
28991 does not depend on Power9 instructions.
28992 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
28993 define_expand to not rely on AltiVec predicate framework.
28994 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
28996 (vector_ne_v2di_p): Change this define_expand to not rely on
28997 AltiVec predicate framework.
28998 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
29000 (vector_ne_<mode>_p): Change this define_expand to not rely on
29001 AltiVec predicate framework.
29002 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
29004 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
29005 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
29006 define_insn pattern.
29007 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
29008 define_insn pattern because the xvcmpne<VSs>. instruction is not
29010 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
29011 instruction is not supported.
29013 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29015 * config/nvptx/nvptx.c: Include intl.h.
29017 2017-03-01 Martin Jambor <mjambor@suse.cz>
29020 * ipa-prop.h (ipa_bits): Removed field known.
29021 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
29022 to pointers. Adjusted their comments to warn about their sharing.
29023 (ipcp_transformation_summary): Change bits to a vector of pointers.
29024 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
29025 (ipa_get_ipa_bits_for_value): Declare.
29026 * tree-vrp.h (value_range): Mark as GTY((for_user)).
29027 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
29028 (ipa_bits_hash_table): Likewise.
29029 (ipa_vr_ggc_hash_traits): Likewise.
29030 (ipa_vr_hash_table): Likewise.
29031 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
29032 being pointers and vr_known being removed.
29033 (ipa_set_jf_unknown): Likewise.
29034 (ipa_get_ipa_bits_for_value): New function.
29035 (ipa_set_jfunc_bits): Likewise.
29036 (ipa_get_value_range): New overloaded functions.
29037 (ipa_set_jfunc_vr): Likewise.
29038 (ipa_compute_jump_functions_for_edge): Use the above functions to
29039 construct bits and vr parts of jump functions.
29040 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
29041 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
29043 (ipcp_grow_transformations_if_necessary): Also allocate
29044 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
29046 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
29047 them. Fix too long lines.
29048 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
29049 vr_known being removed.
29050 (ipa_read_jump_function): Use new setter functions to construct bits
29051 and vr parts of jump functions or set them to NULL.
29052 (write_ipcp_transformation_info): Adjust for bits being pointers.
29053 (read_ipcp_transformation_info): Likewise.
29054 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
29056 Include gt-ipa-prop.h.
29057 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
29059 (ipcp_store_bits_results): Likewise.
29060 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
29061 Do not write to existing jump functions but use a temporary instead.
29063 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29066 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
29067 attempt to use its first operand as BIT_FIELD_REF base.
29069 2017-03-01 Richard Biener <rguenther@suse.de>
29071 PR middle-end/79721
29072 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
29073 interpolating formula in wrapping arithmetic.
29074 (chrec_apply): Convert chrec_evaluate return value to wanted type.
29076 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29078 PR tree-optimization/79734
29079 * tree-vect-generic.c (expand_vector_condition): Optimize
29080 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
29081 Handle VEC_COND_EXPR where comparison has different inner width from
29082 type's inner width.
29084 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
29086 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
29087 markup, and similar issues. Remove @opindex entries for things
29088 that aren't options. Add missing -mmpy-option entries.
29090 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29092 PR tree-optimization/79737
29093 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
29094 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
29095 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
29096 instead of byte_size. Formatting fix.
29097 (shift_bytes_in_array_right): Formatting fix.
29099 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
29102 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
29103 condition on optimize for the leaf function test.
29105 2017-02-28 Martin Liska <mliska@suse.cz>
29108 * read-rtl-function.c (function_reader::handle_unknown_directive):
29109 Bail out when one uses -flto.
29111 2017-02-28 Martin Liska <mliska@suse.cz>
29113 * common.opt: Replace space with tabular for options of <number>
29115 * config/i386/i386.opt: Show <number> value for
29116 -mlarge-data-threshold.
29117 * opts.c (print_filtered_help): Do not display number in hexadecimal
29120 2017-02-28 Martin Liska <mliska@suse.cz>
29122 * common.opt: Fix --help=option -Q for options which are of
29125 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
29127 * config/i386/i386.c (print_reg): Error out for values
29128 of 8-bit size in invalid integer register.
29130 2017-02-28 Martin Sebor <msebor@redhat.com>
29132 PR tree-optimization/79691
29133 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
29135 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29138 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
29139 gcc_unreachable with output_operand_lossage.
29141 2017-02-28 Richard Biener <rguenther@suse.de>
29143 PR tree-optimization/79740
29144 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
29146 (visit_nary_op): Insert the nary into the hashtable if we
29147 pattern-matched sth.
29148 * tree-ssa-pre.c (eliminate_insert): Robustify.
29150 2017-02-28 Richard Biener <rguenther@suse.de>
29152 PR middle-end/79731
29153 * fold-const.c (decode_field_reference): Reject out-of-bound
29156 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29158 * config/i386/i386.c: Include intl.h.
29159 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
29160 instead of just cond ? "..." : "...".
29161 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
29162 * coverage.c (read_counts_file): Likewise.
29163 * omp-offload.c: Include intl.h.
29164 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
29165 of just cond ? "..." : "...".
29166 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
29167 of just cond ? "..." : "...".
29169 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
29172 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
29174 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
29175 'tune for' CPU name.
29176 * config/arm/arm-cpu-data.h: Regenerated.
29178 2017-02-28 Richard Biener <rguenther@suse.de>
29180 PR tree-optimization/79732
29181 * tree-inline.c (expand_call_inline): Do not shadow var.
29183 2017-02-28 Richard Biener <rguenther@suse.de>
29185 PR tree-optimization/79723
29186 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
29187 address-space properly.
29189 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
29191 * doc/optinfo.texi (Optimization groups): Fix option used for
29193 * doc/invoke.texi (-fopt-info): Document "omp".
29194 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
29195 (OPTGROUP_ALL): Add OPTGROUP_OMP.
29196 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
29197 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
29198 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
29200 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
29202 * dumpfile.c (optgroup_options): Instead of "openmp", associate
29203 OPTGROUP_OMP with "omp".
29205 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
29208 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
29209 for arithmetic shift of unsigned V2DI.
29211 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
29213 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
29214 arc/linux.h headers.
29215 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
29216 (LINK_SPEC): Likewise.
29217 (ARC_TLS_EXTRA_START_SPEC): Likewise.
29218 (EXTRA_SPECS): Likewise.
29219 (STARTFILE_SPEC): Likewise.
29220 (ENDFILE_SPEC): Likewise.
29221 (LIB_SPEC): Likewise.
29222 (TARGET_SDATA_DEFAULT): Likewise.
29223 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
29224 (MULTILIB_DEFAULTS): Likewise.
29225 (DWARF2_UNWIND_INFO): Likewise.
29226 * config/arc/big.h: New file.
29227 * config/arc/elf.h: Likewise.
29228 * config/arc/linux.h: Likewise.
29229 * config/arc/t-uClibc: Remove.
29231 2017-02-27 Bin Cheng <bin.cheng@arm.com>
29233 PR tree-optimization/77536
29234 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
29235 (tree_transform_and_unroll_loop): Use above function to compute the
29236 estimated niter of unrolled loop and use it when scaling profile.
29237 Also use count info rather than frequency if it's non-zero.
29238 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
29239 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
29240 (vect_transform_loop): Call above function.
29242 2017-02-27 Richard Biener <rguenther@suse.de>
29244 PR tree-optimization/45397
29245 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
29246 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
29247 (visit_nary_op): Add pattern matching for CSEing sign-changed
29248 or truncated operations with wider ones.
29250 2017-02-27 Richard Biener <rguenther@suse.de>
29252 PR tree-optimization/79690
29253 * tree-vect-stmts.c (vectorizable_store): Use vector type
29254 built from the DR with address-space.
29256 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
29258 * doc/invoke.texi (Optimize Options): Refine the description
29259 of asan-use-after-return.
29261 2017-02-25 Alan Modra <amodra@gmail.com>
29263 PR rtl-optimization/79584
29264 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
29265 base, not ad->base_term, the reg within base. Remove assertion
29266 that ad->base == ad->base_term. Replace gen_int_mode using
29267 bogus mode with const0_rtx.
29269 2017-02-25 Jakub Jelinek <jakub@redhat.com>
29271 PR middle-end/79396
29272 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
29273 FMA_EXPR like tcc_binary or tcc_unary.
29275 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
29278 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
29280 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
29281 (output_loc_operands): Handle DW_OP_call_ref and
29282 DW_OP_GNU_variable_value.
29283 (struct variable_value_struct): New type.
29284 (struct variable_value_hasher): Likewise.
29285 (variable_value_hash): New variable.
29286 (string_types): Remove.
29287 (copy_loc_descr): New function.
29288 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
29289 (prepend_loc_descr_to_each): New function.
29290 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
29291 instead of add_loc_descr_to_each if the first argument is single
29292 location list and the second has multiple.
29293 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
29294 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
29295 when looking for variable value which doesn't have other location info.
29296 (loc_list_from_tree): Formatting fix.
29297 (gen_array_type_die): Simplify DW_AT_string_length handling.
29298 (adjust_string_types): Remove.
29299 (gen_subprogram_die): Don't call adjust_string_types nor test/set
29300 string_types. Call resolve_variable_values.
29301 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
29302 (resolve_addr_in_expr): Likewise. Add A argument.
29303 (copy_deref_exprloc): Remove deref argument. Adjust for the
29304 original expression being DW_OP_GNU_variable_value with optionally
29305 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
29306 optionally after it.
29307 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
29308 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
29309 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
29310 (variable_value_hasher::hash, variable_value_hasher::equal): New
29312 (resolve_variable_value_in_expr, resolve_variable_value,
29313 resolve_variable_values, note_variable_value_in_expr,
29314 note_variable_value): New functions.
29315 (dwarf2out_early_finish): Call note_variable_value on all toplevel
29318 2017-02-24 Jakub Jelinek <jakub@redhat.com>
29321 * opts.h (handle_generated_option): Add GENERATED_P argument.
29322 * opts-common.c (handle_option): Adjust function comment.
29323 (handle_generated_option): Add GENERATED_P argument, pass it to
29325 (control_warning_option): Pass false to handle_generated_option
29327 * opts.c (maybe_default_option): Pass true to handle_generated_option
29329 * optc-gen.awk: Likewise.
29331 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29333 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
29334 a REG, look at the REG it is a SUBREG of.
29335 (splitter for cmpeqsi_t): Ditto.
29337 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29339 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
29340 the special USEs with the pattern of the insn, not the insn itself.
29342 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
29345 * doc/invoke.texi: Document -mload-store-pairs.
29347 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29348 Sandra Loosemore <sandra@codesourcery.com>
29350 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
29351 argument isn't a CONST_INT.
29352 (nios2_alternate_compare_const): Assert op is a CONST_INT.
29353 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
29354 (nios2_validate_compare): Bypass alternate compare logic if *op2
29355 is not a CONST_INT.
29356 (ldstwm_operation_p): Return false if first_base is not a REG or
29357 if first_offset is not a CONST_INT.
29359 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29361 * config/cris/cris.md: Use correct operand in a define_peephole2.
29363 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29365 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
29367 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29369 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
29370 this_insn if it is an INSN or JUMP_INSN.
29371 (force_offsettable): Look at base, not at addr.
29372 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
29373 on things that aren't necessarily CONST_INTs.
29375 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
29377 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
29378 -mfpmath=sse is the default also for x86-32 targets with SSE2
29379 instruction set when @option{-ffast-math} is enabled
29381 2017-02-24 Jeff Law <law@redhat.com>
29383 PR rtl-optimizatoin/79286
29384 * ira.c (update_equiv_regs): Drop may_trap_p exception to
29387 2017-02-24 Richard Biener <rguenther@suse.de>
29389 PR tree-optimization/79389
29390 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
29393 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
29395 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
29396 function comment to reflect reality.
29397 (loop_exits_before_overflow): Fix typo in function description.
29399 2017-02-24 Richard Biener <rguenther@suse.de>
29401 PR tree-optimization/79389
29402 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
29403 properly that a threading opportunity exists. Detect conditional
29404 copy/constant propagation opportunities.
29406 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
29408 * config/visium/visium.md (type): Add trap.
29409 (b): New mode attribute.
29410 (*btst): Rename into...
29411 (*btst<mode>): ...this and adjust.
29412 (*cbranchsi4_btst_insn): Rename into...
29413 (*cbranch<mode>4_btst_insn): ...this and adjust.
29414 (trap): New define_insn.
29416 2017-02-23 Jakub Jelinek <jakub@redhat.com>
29418 PR tree-optimization/79389
29419 * ifcvt.c (struct noce_if_info): Add rev_cond field.
29420 (noce_reversed_cond_code): New function.
29421 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
29422 reversed_comparison_code. Formatting fix.
29423 (noce_try_store_flag): Test rev_cond != NULL in addition to
29424 reversed_comparison_code.
29425 (noce_try_store_flag_constants): Likewise.
29426 (noce_try_store_flag_mask): Likewise.
29427 (noce_try_addcc): Use rev_cond if non-NULL instead of
29428 reversed_comparison_code.
29429 (noce_try_cmove_arith): Likewise. Formatting fixes.
29430 (noce_try_minmax, noce_try_abs): Clear rev_cond.
29431 (noce_find_if_block): Initialize rev_cond.
29432 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
29433 instead of false as last argument never attempt to reverse it
29436 2017-02-23 Bin Cheng <bin.cheng@arm.com>
29438 PR tree-optimization/79663
29439 * tree-predcom.c (combine_chains): Process refs in reverse order
29440 only for ZERO length chains, and add explaining comment.
29442 2017-02-23 Jeff Law <law@redhat.com>
29444 PR tree-optimization/79578
29445 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
29446 in call to operand_equal_p.
29448 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
29451 * config/i386/cpuid.h: Fix another undefined behavior.
29453 2017-02-23 Richard Biener <rguenther@suse.de>
29455 PR tree-optimization/79683
29456 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
29457 vector types for data-refs.
29459 2017-02-23 Martin Liska <mliska@suse.cz>
29461 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
29463 2017-02-23 Jakub Jelinek <jakub@redhat.com>
29465 PR middle-end/79665
29466 * internal-fn.c (get_range_pos_neg): Moved to ...
29467 * tree.c (get_range_pos_neg): ... here. No longer static.
29468 * tree.h (get_range_pos_neg): New prototype.
29469 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
29470 are known to be in between 0 and signed maximum inclusive, try to
29471 expand both unsigned and signed divmod and use the cheaper one from
29474 2017-02-22 Jeff Law <law@redhat.com>
29476 PR tree-optimization/79578
29477 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
29478 to compare base operands.
29480 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
29483 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
29484 gpc_reg_operand instead of fpr_reg_operand.
29486 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
29488 * config/mips/mips.c (mips_return_in_memory): Force FP
29489 vector types to be returned in memory for o32 ABI.
29491 2017-02-22 Jakub Jelinek <jakub@redhat.com>
29493 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
29494 instead of DW_TAG_member for static data member declarations and don't
29495 set no_linkage_name for static inline data members.
29496 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
29499 2017-02-22 Martin Liska <mliska@suse.cz>
29501 * doc/invoke.texi: Replace inequality signs with square brackets
29504 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
29507 * lra-constraints.c (simplify_operand_subreg): Handle
29508 WORD_REGISTER_OPERATIONS targets.
29510 2017-02-22 Jakub Jelinek <jakub@redhat.com>
29513 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
29514 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
29515 elimination by swapping fld*.
29517 2017-02-22 Richard Biener <rguenther@suse.de>
29519 PR tree-optimization/79673
29520 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
29521 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
29522 irrelevant address-space qualifiers and avoiding a
29523 ADDR_SPACE_CONVERT_EXPR from fold_convert.
29525 2017-02-22 Richard Biener <rguenther@suse.de>
29527 PR tree-optimization/79666
29528 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
29529 to not symbolically negate if that may introduce undefined
29532 2017-02-22 Martin Liska <mliska@suse.cz>
29535 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
29536 * data-streamer-out.c (streamer_write_gcov_count_stream):
29538 * value-prof.c (stream_out_histogram_value): Make assert more
29539 precise based on type of counter.
29541 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
29544 * config/i386/i386.md (standard_x87sse_constant_load splitter):
29545 Use nonimmediate_operand instead of memory_operand for operand 1.
29546 (float-extend standard_x87sse_constant_load splitter): Ditto.
29548 2017-02-21 Jeff Law <law@redhat.com>
29550 PR tree-optimization/79621
29551 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
29552 blocks with edges to themselves.
29554 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29557 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
29558 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
29559 Use gimple_call_builtin_p.
29562 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
29563 on temporarily removed DEBUG_INSNs.
29565 PR tree-optimization/79649
29566 * tree-loop-distribution.c (classify_partition): Give up on
29567 non-generic address space loads/stores.
29569 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
29571 * doc/loop.texi (Loop manipulation): Remove nonexistent
29572 tree_ssa_loop_version from the documentation.
29573 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
29575 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29578 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
29579 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
29580 * config/rs6000/rs6000.c: Include except.h.
29581 (rs6000_expand_split_stack_prologue): Call
29582 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
29584 2017-02-21 Martin Jambor <mjambor@suse.cz>
29587 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
29588 have been analyzed.
29590 2017-02-21 Martin Jambor <mjambor@suse.cz>
29592 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
29593 for backward compatibility only.
29594 * doc/invoke.texi (Option Summary): Remove all references to
29595 -fipa-cp-alignment.
29597 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
29601 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29603 * lra-constraints.c (curr_insn_transform): Handle
29604 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
29606 2017-02-21 Martin Liska <mliska@suse.cz>
29608 * config/i386/i386.opt: Replace -masm-dialect with -masm.
29610 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
29612 PR translation/79638
29613 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
29615 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
29618 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
29619 (arm_function_ok_for_sibcall): Return false for an indirect call by
29620 descriptor if all the argument registers are used.
29621 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
29622 alignment of the function.
29624 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29626 PR tree-optimization/61441
29627 * simplify-rtx.c (simplify_const_unary_operation): For
29628 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
29629 the sNaN unmodified.
29631 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
29633 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
29634 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
29635 instead of SYSTEM_HEADER_DIR.
29637 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
29638 Martin Liška <mliska@suse.cz>
29640 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
29641 Fix typos and grammar, use active voice, and clarify.
29643 2017-02-20 Marek Polacek <polacek@redhat.com>
29645 PR middle-end/79537
29646 * gimplify.c (gimplify_expr): Handle unused *&&L;.
29649 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
29651 2017-02-20 Jakub Jelinek <jakub@redhat.com>
29654 * config/i386/i386.c (ix86_expand_builtin): Handle
29655 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
29656 ix86_builtins_isa[fcode].isa as a requirement of those
29657 flags and any other flag in the bitmask.
29658 (ix86_init_mmx_sse_builtins): Use 0 instead of
29659 ~OPTION_MASK_ISA_64BIT as mask.
29660 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
29661 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
29662 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
29663 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
29665 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29668 * lra-constraints.c (split_reg): Check requested split mode
29669 is supported by the register.
29671 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29673 * lra-constraints.c (simplify_operand_subreg): Remove early
29676 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29679 * lra-constraints.c (curr_insn_transform): Tighten condition
29680 for converting SUBREG reloads from OP_OUT to OP_INOUT.
29682 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29685 * lra-constraints.c (curr_insn_transform): Handle
29686 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
29688 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
29691 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
29693 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
29695 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
29698 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
29701 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29703 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
29706 2017-02-18 Jakub Jelinek <jakub@redhat.com>
29708 * final.c (last_columnnum, override_columnnum): New variables.
29709 (final_start_function): Set last_columnnum, pass it to begin_prologue
29710 hook and pass 0 to dwarf2out_begin_prologue.
29711 (final_scan_insn): Update override_columnnum. Pass last_columnnum
29712 to source_line debug hook.
29713 (notice_source_line): Compute last_columnnum and for debug_column_info
29714 return true on column changes.
29715 * debug.h (struct gcc_debug_hooks): Add column argument to
29716 source_line and begin_prologue hooks.
29717 (debug_nothing_int_charstar_int_bool): Remove prototype.
29718 (debug_nothing_int_int_charstar,
29719 debug_nothing_int_int_charstar_int_bool): New prototypes.
29720 (dwarf2out_begin_prologue): Add column argument.
29721 * debug.c (do_nothing_debug_hooks): Adjust source_line and
29722 begin_prologue hooks.
29723 (debug_nothing_int_charstar_int_bool): Remove.
29724 (debug_nothing_int_int_charstar,
29725 debug_nothing_int_int_charstar_int_bool): New functions.
29726 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
29727 through to dwarf2out_source_line.
29728 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
29729 (dwarf2out_source_line): Add column argument, emit it if requested.
29730 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
29732 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
29733 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
29734 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
29735 through to dwarf2out_begin_prologue.
29736 (vmsdbgout_source_line): Add column argument, pass it through to
29737 dwarf2out_source_line.
29738 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
29739 dbxout_source_line caller.
29740 (dbxout_source_line): Add column argument.
29742 * common.opt (gno-column-info, gcolumn-info): New options.
29743 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
29744 (check_die): Also test for multiple DW_AT_decl_column attributes.
29745 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
29746 DW_AT_decl_column if requested.
29747 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
29749 (gen_variable_die): Likewise.
29750 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
29751 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
29754 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
29755 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
29756 (ix86_handle_option): Handle OPT_m3dnowa.
29757 * doc/invoke.texi (-m3dnowa): Document.
29758 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
29759 -m3dnowa instead of -m3dnow -march=athlon.
29762 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
29763 instead of gcc_assert for K, r and R code checks. Formatting fixes.
29765 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29768 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
29769 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
29770 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
29771 generator for vsx_xxpermdi_<mode>_be.
29772 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
29773 force big-endian semantics.
29774 (vsx_xxpermdi_<mode>_be): New define_expand with same
29775 implementation as previous version of vsx_xxpermdi_<mode>.
29777 2017-02-17 Jakub Jelinek <jakub@redhat.com>
29779 PR tree-optimization/79327
29780 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
29781 variable, its initialization and use.
29783 2017-02-17 Julia Koval <julia.koval@intel.com>
29785 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
29786 (OPTION_MASK_ISA_PKU_UNSET): New.
29787 (ix86_handle_option): Handle -mrdpid.
29788 * config/i386/cpuid.h (bit_RDPID): New.
29789 * config/i386/driver-i386.c (host_detect_local_cpu):
29790 Detect RDPID feature.
29791 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
29792 * config/i386/i386-c.c (ix86_target_macros_internal):
29794 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
29795 (ix86_valid_target_attribute_inner_p): Add "rdpid".
29796 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
29797 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
29798 * config/i386/i386.md (define_insn "rdpid"): New.
29799 * config/i386/i386.opt Add -mrdpid.
29800 * config/i386/immintrin.h (_rdpid_u32): New.
29802 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
29804 PR rtl-optimization/79541
29805 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
29806 instead of transforming it into USE.
29808 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
29810 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
29811 If HONOR_SNANS (SFmode) force the input to a register.
29812 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
29813 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
29814 an frsp or similar insn.
29816 2017-02-17 Martin Liska <mliska@suse.cz>
29818 PR rtl-optimization/79577
29819 * params.def (selsched-max-sched-times): Increase minimum to 1.
29821 2017-02-17 Martin Liska <mliska@suse.cz>
29823 PR rtl-optimization/79574
29824 * gcse.c (want_to_gcse_p): Prevent integer overflow.
29826 2017-02-17 Martin Liska <mliska@suse.cz>
29828 PR tree-optimization/79529
29829 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
29830 ssa_defined_default_def_p to handle cases which are implicitly
29832 * tree-ssa.c (ssa_defined_default_def_p): New function.
29833 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
29834 which are implicitly defined.
29835 * tree-ssa.h (ssa_defined_default_def_p): Declare.
29837 2017-02-17 Richard Biener <rguenther@suse.de>
29839 PR middle-end/79576
29840 * params.def (max-ssa-name-query-depth): Limit to 10.
29842 2017-02-17 Richard Biener <rguenther@suse.de>
29844 PR tree-optimization/79552
29845 * tree-ssa-structalias.c (visit_loadstore): Properly verify
29848 2017-02-17 Richard Biener <rguenther@suse.de>
29851 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
29853 2017-02-17 Marek Polacek <polacek@redhat.com>
29855 PR middle-end/79536
29856 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
29857 (fold_negate_expr): New wrapper.
29859 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
29861 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
29862 Correct terminology and de-emphasize pre-standard behavior.
29864 2017-02-16 Alan Modra <amodra@gmail.com>
29866 PR rtl-optimization/79286
29867 * ira.c (def_dominates_uses): New function.
29868 (update_equiv_regs): Don't create an equivalence for insns that
29869 may trap where the register def does not dominate the use.
29871 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
29873 PR rtl-optimization/78127
29874 * lra.c (lra): Call lra_eliminate before finish the loop after
29877 2017-02-16 Richard Biener <rguenther@suse.de>
29879 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
29881 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
29882 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
29883 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
29884 (isl_val_int_from_wi): New function.
29885 (extract_affine_gmp): Rename to ...
29886 (extract_affine_wi): ... this, take a widest_int.
29887 (extract_affine_int): Just wrap extract_affine_wi.
29888 (add_param_constraints): Use isl_val_int_from_wi.
29889 (add_loop_constraints): Likewise, and extract_affine_wi.
29891 2017-02-15 Jeff Law <law@redhat.com>
29893 PR middle-end/79521
29894 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
29895 ira_init_register_move_cost_if_necessary.
29897 2017-02-15 Martin Sebor <msebor@redhat.com>
29899 PR middle-end/32003
29900 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
29901 removed in a prior commit.
29903 2017-02-15 Bin Cheng <bin.cheng@arm.com>
29905 PR tree-optimization/79347
29906 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
29907 counters during peeling.
29909 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
29911 * Makefile.in (site.exp): Remove "set ISLVER".
29913 2017-02-15 Jakub Jelinek <jakub@redhat.com>
29916 * real.c (real_from_integer): Call real_convert even for decimal.
29918 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
29921 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
29923 2017-02-14 Andrew Pinski <apinski@cavium.com>
29925 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
29926 cores and change the partno/implementer to be correct.
29927 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
29928 the 'B" as the implementer.
29929 * config/aarch64/aarch64-tune.md: Regenerate.
29931 2017-02-14 Carl Love <cel@us.ibm.com>
29933 * config/rs6000/rs6000.c: Add case statement entry to make the
29934 xvcvuxdsp built-in argument unsigned.
29935 * config/rs6000/vsx.md: Fix the source and return operand types so they
29936 match the instruction definitions from the ISA document. Fix typo
29937 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
29940 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
29943 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
29944 member early_clobber_alts.
29945 * lra-lives.c (reg_early_clobber_p): New.
29946 (process_bb_lives): Use it.
29947 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
29948 (debug_operand_data): Initialize early_clobber_alts.
29949 (setup_operand_alternative): Set up early_clobber_alts.
29950 (collect_non_operand_hard_regs): Ditto. Pass early clobber
29951 alternatives to new_insn_reg.
29952 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
29954 (lra_update_insn_regno_info): Pass the new arg.
29956 2017-02-14 Jakub Jelinek <jakub@redhat.com>
29958 PR middle-end/79505
29959 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
29960 (new_oacc_loop_raw): Don't clear already cleared fields.
29963 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
29964 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
29965 _mm512_prefetch_i64gather_ps): New inline functions and macros.
29967 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
29970 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
29972 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
29975 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
29976 the extra instruction to the right place to store 128-bit constant
29979 2017-02-14 Martin Sebor <msebor@redhat.com>
29981 PR middle-end/79448
29982 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
29983 warning for strings of unknown length.
29985 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
29987 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
29989 2017-02-14 Jeff Law <law@redhat.com>
29992 * ira-costs.c (scan_one_insn): Initialize register move costs
29993 for pseudos seen in USE/CLOBBER insns.
29995 PR tree-optimization/79095
29996 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
29997 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
29998 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
29999 if the operands are known to be not equal, then the resulting range
30001 (intersect_ranges): If the new range is ~[0,0] and the old range is
30002 wide, then prefer ~[0,0].
30003 * tree-vrp.c (overflow_comparison_p_1): New function.
30004 (overflow_comparison_p): New function.
30005 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
30006 if NAME is used in an overflow test.
30007 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
30008 overflow check that can be expressed as an equality test, then adjust
30009 ops to be that equality test.
30011 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30013 * config/s390/s390-builtin-types.def: Remove flags argument.
30014 * config/s390/s390.c (s390_init_builtins): Likewise.
30016 2017-02-14 Martin Liska <mliska@suse.cz>
30018 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
30019 vector. Fix trailing white spaces.
30021 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
30023 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
30026 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30028 PR rtl-optimization/68664
30029 * config/arm/arm.c (arm_sched_can_speculate_insn):
30030 New function. Declare prototype.
30031 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
30033 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30035 PR rtl-optimization/68664
30036 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
30038 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
30040 2017-02-14 Amit Pawar <amit.pawar@amd.com>
30042 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
30043 max skip bytes for function, loop and jump.
30045 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30047 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
30048 ABS_EXPR for gimple dump.
30050 2017-02-14 Jakub Jelinek <jakub@redhat.com>
30053 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
30055 PR tree-optimization/79408
30056 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
30057 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
30058 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
30059 also if rhs1 is INTEGER_CST.
30061 2017-02-14 Richard Biener <rguenther@suse.de>
30063 PR middle-end/79432
30064 * tree-into-ssa.c (insert_phi_nodes): When the function can
30065 have abnormal edges rewrite SSA names with broken use-def
30066 dominance out of SSA and register them for PHI insertion.
30068 2017-02-13 Martin Sebor <msebor@redhat.com>
30070 PR middle-end/79496
30071 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
30072 clearing info.nowrite flag when snprintf size argument is a range.
30074 2017-02-13 Jakub Jelinek <jakub@redhat.com>
30076 * cprop.c (cprop_jump): Add missing space in string literal.
30077 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
30078 (get_constraint_for_component_ref): Likewise.
30079 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
30080 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
30081 * lra-constraints.c (process_alt_operands): Likewise.
30082 * ipa-inline.c (inline_small_functions): Likewise.
30083 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
30084 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
30085 * trans-mem.c (diagnose_tm_1_op): Likewise.
30086 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
30087 (grid_parallel_clauses_gridifiable): Likewise.
30089 * config/nvptx/mkoffload.c (process): Add space in between
30092 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
30093 "MOD4_SSE_REGS" and "ALL_REGS".
30095 * spellcheck.c (test_data): Add , in between "foo" and "food".
30097 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30100 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
30101 boundary crossing check and subsequent code generation agree.
30103 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30105 * config/aarch64/aarch64.c (has_memory_op): Delete.
30106 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
30109 2017-02-13 Jakub Jelinek <jakub@redhat.com>
30111 PR rtl-optimization/79388
30112 PR rtl-optimization/79450
30113 * combine.c (distribute_notes): When removing TEM_INSN for which
30114 corresponding dest has last value recorded, invalidate that last
30117 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30119 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
30120 of explicit '@'. Add missing assembly comment marker on branch costs
30123 2017-02-13 Nathan Sidwell <nathan@acm.org>
30125 * gengtype-lex.l (<in_struct>): Add '/'.
30127 2017-02-13 Martin Liska <mliska@suse.cz>
30130 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
30132 2017-02-13 Richard Biener <rguenther@suse.de>
30134 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
30136 * configure: Re-generate.
30137 * config.in: Likewise.
30138 * graphite-dependences.c: Simplify as if
30139 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
30140 * graphite-isl-ast-to-gimple.c: Likewise.
30141 * graphite-optimize-isl.c: Likewise.
30142 * graphite-poly.c: Likewise.
30143 * graphite-sese-to-poly.c: Likewise.
30144 * graphite.h: Likewise.
30145 * toplev.c: Include isl/version.h and use isl_version () for
30146 printing the ISL version.
30147 * doc/install.texi: Update ISL requirement.
30149 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
30151 * doc/standards.texi (Standards): Update reference to
30154 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
30156 * doc/extend.texi (Named Address Spaces): sourceware.org now
30158 * doc/install.texi (Binaries): Ditto.
30161 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
30163 * doc/cpp.texi: Replace "stringify"/"stringification" with C
30164 standard terminology "stringize"/"stringizing" throughout.
30165 * doc/cppinternals.texi: Likewise.
30167 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
30169 * doc/extend.texi: Fix some spelling mistakes and typos.
30170 * doc/invoke.texi: Likewise.
30172 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
30175 * params.def (inline-min-speedup) Change from 10 to 8.
30177 2017-02-11 Jakub Jelinek <jakub@redhat.com>
30179 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
30182 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
30185 * ipa-inline-analysis.c (get_minimal_bb): New function.
30186 (record_modified): Use it.
30187 (remap_edge_change_prob): Handle also ancestor functions.
30189 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
30191 * doc/contrib.texi (Contributors): Remove broken link into
30192 the Mauve CVS repository.
30194 2017-02-11 Jakub Jelinek <jakub@redhat.com>
30196 PR middle-end/79454
30197 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
30198 result computation whenever lhs doesn't have vector mode, not
30199 just when it has BLKmode.
30201 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
30203 * doc/makefile.texi (profiledbootstrap): Refer to the
30204 installation instructions only in textual form.
30206 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30209 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
30211 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
30213 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
30214 (Specific): Update mingw-w64 reference.
30216 (Specific): Remove broken link to Renesas RX processor.
30218 2017-02-10 Richard Biener <rguenther@suse.de>
30220 * toplev.c (process_options): Do not mention obsolete graphite
30221 options when printing sorry message about missing graphite support.
30222 Mention -floop-nest-optimize.
30224 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
30226 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
30227 (vtst_p16): Likewise.
30228 (vtstq_p8): Likewise.
30229 (vtstq_p16): Likewise.
30231 (vtstq_p64): Likewise.
30232 * config/arm/arm_neon.h (vgetq_lane_p64): New.
30233 (vset_lane_p64): New.
30234 (vsetq_lane_p64): New.
30236 2017-02-10 Jakub Jelinek <jakub@redhat.com>
30238 PR tree-optimization/79411
30239 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
30240 stmt operands are SSA_NAMEs used in abnormal phis.
30241 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
30244 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
30247 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
30250 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
30252 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
30254 2017-02-09 Jakub Jelinek <jakub@redhat.com>
30256 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
30260 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
30261 not arbitrary TREE_CONSTANT.
30264 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
30265 "omp declare target link" attribute unless is_global_var.
30266 * omp-offload.c (find_link_var_op): Likewise.
30268 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
30269 Chung-Lin Tang <cltang@codesourcery.com>
30271 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
30273 (gimplify_adjust_omp_clauses): Don't delete TILE.
30274 (gimplify_omp_for): Deal with TILE.
30275 * internal-fn.c (expand_GOACC_TILE): New function.
30276 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
30278 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
30279 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
30281 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
30282 avoid DIV for outermost collapse var.
30283 (expand_oacc_for): Insert tile element loop as needed. Adjust.
30284 Remove out of date comments, fix whitespace.
30285 * omp-general.c (omp_extract_for_data): Deal with tiling.
30286 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
30287 adjust OLF_DIM_BASE value.
30288 (struct omp_for_data): Add tiling field.
30289 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
30290 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
30291 for auto loops. Remove default auto determining, moved to
30292 oacc_loop_fixed_partitions.
30293 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
30294 stmts, add e_mask field.
30295 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
30296 (oacc_thread_numbers): Use oacc_dim_call.
30297 (oacc_xform_tile): New.
30298 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
30299 (finish_oacc_loop): Adjust for ifns vector.
30300 (oacc_loop_discover_walk): Append loop abstraction sites to list,
30301 add case for GOACC_TILE fns.
30302 (oacc_loop_xform_loop): Delete.
30303 (oacc_loop_process): Iterate over call list directly, and add
30304 handling for GOACC_TILE fns.
30305 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
30307 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
30308 vector partitioning to outer loops. Assign 2 partitions to loops
30309 when available. Add TILE handling.
30310 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
30311 (execite_oacc_device_lower): Process GOACC_TILE fns,
30312 ignore unknown specs.
30313 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
30314 * tree.c (omp_clause_num_ops): Adjust TILE ops.
30315 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
30317 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
30319 * configure.ac (ACX_BUGURL): Update.
30320 * configure: Regenerate.
30322 2017-02-09 Richard Biener <rguenther@suse.de>
30324 PR tree-optimization/69823
30325 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
30326 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
30328 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
30330 * config/arc/arc-c.def: Add __NPS400__ definition.
30331 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
30332 (TARGET_NPS400): Define.
30334 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
30336 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
30338 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
30339 pointer, arch_info.
30340 (arc_cpu_types): Fill the arch_info field with a pointer into the
30341 arc_arch_types table.
30342 (arc_selected_cpu): Declare.
30343 * config/arc/arc.c (arc_selected_cpu): Make global.
30344 (arc_selected_arch): Delete.
30345 (arc_base_cpu): Delete.
30346 (arc_override_options): Remove references to deleted variables,
30347 update access to arch information.
30348 (ARC_OPT): Update access to arch information.
30349 (ARC_OPTX): Likewise.
30350 * config/arc/arc.h (arc_base_cpu): Remove declaration.
30351 (TARGET_ARC600): Update access to arch information.
30352 (TARGET_ARC601): Likewise.
30353 (TARGET_ARC700): Likewise.
30354 (TARGET_EM): Likewise.
30355 (TARGET_HS): Likewise.
30356 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
30359 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
30362 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
30363 condition/operands for integer GE/LE/GEU/LEU operations.
30365 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
30367 PR translation/79397
30368 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
30371 2017-02-08 Martin Jambor <mjambor@suse.cz>
30374 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
30375 whether allocation happened.
30376 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
30377 nothing was allocated.
30379 2017-02-08 Jakub Jelinek <jakub@redhat.com>
30381 PR tree-optimization/79408
30382 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
30383 constant, but SSA_NAME with a known integer range, use the minimum
30384 of that range instead of op1 to determine if modulo can be replaced
30385 with its first operand.
30387 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30389 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
30391 2017-02-08 Richard Biener <rguenther@suse.de>
30393 PR tree-optimization/71824
30394 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
30395 Check all loops contained in the merged region.
30397 2017-02-07 Andrew Pinski <apinski@cavium.com>
30399 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
30401 2017-02-07 Andrew Pinski <apinski@cavium.com>
30403 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
30404 (thunderxt88): Likewise.
30405 (thunderxt81): Disable LSE and change v8.1 to v8.
30406 (thunderxt83): Likewise.
30408 2017-02-07 Jakub Jelinek <jakub@redhat.com>
30409 Richard Biener <rguenther@suse.de>
30411 PR middle-end/79399
30412 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
30413 type from int to size_t.
30414 * ira-costs.c (struct_costs_size): Change type from int to size_t.
30416 2017-02-07 Jakub Jelinek <jakub@redhat.com>
30418 PR rtl-optimization/79386
30419 * cprop.c (bypass_conditional_jumps): Initialize
30420 bypass_last_basic_block already before splitting bbs after
30421 unconditional traps...
30422 (bypass_conditional_jumps): ... rather than here.
30425 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
30426 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
30427 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
30428 fix -masm=intel patterns.
30430 2017-02-07 Richard Biener <rguenther@suse.de>
30432 PR tree-optimization/79256
30433 PR middle-end/79278
30434 * builtins.c (get_object_alignment_2): Use min_align_of_type
30435 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
30436 and ADJUST_FIELD_ALIGN.
30438 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
30440 * doc/tm.texi: Regenerate.
30441 * stor-layout.c (layout_decl): Adjust.
30442 (update_alignment_for_field): Likewise.
30443 (place_field): Likewise.
30444 (min_align_of_type): Likewise.
30445 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
30446 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
30447 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
30448 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
30449 * config/frv/frv.c (frv_adjust_field_align): Likewise.
30450 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
30451 * config/i386/i386.c (x86_field_alignment): Likewise.
30452 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
30453 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
30454 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
30455 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
30456 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
30457 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
30461 2017-01-30 Richard Biener <rguenther@suse.de>
30463 PR tree-optimization/79256
30464 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
30465 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
30468 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
30470 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
30471 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
30472 builtins to SImode and emit a zero-extend, if necessary.
30474 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30476 * docs/invoke.texi (RISC-V Options): Alphabetize.
30478 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30480 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
30483 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30485 * config/riscv/riscv.c: New file.
30486 * common/config/riscv/riscv-common.c: Likewise.
30487 * config.gcc: Likewise.
30488 * config/riscv/constraints.md: Likewise.
30489 * config/riscv/elf.h: Likewise.
30490 * config/riscv/generic.md: Likewise.
30491 * config/riscv/linux.h: Likewise.
30492 * config/riscv/multilib-generator: Likewise.
30493 * config/riscv/peephole.md: Likewise.
30494 * config/riscv/pic.md: Likewise.
30495 * config/riscv/predicates.md: Likewise.
30496 * config/riscv/riscv-builtins.c: Likewise.
30497 * config/riscv/riscv-c.c: Likewise.
30498 * config/riscv/riscv-ftypes.def: Likewise.
30499 * config/riscv/riscv-modes.def: Likewise.
30500 * config/riscv/riscv-opts.h: Likewise.
30501 * config/riscv/riscv-protos.h: Likewise.
30502 * config/riscv/riscv.h: Likewise.
30503 * config/riscv/riscv.md: Likewise.
30504 * config/riscv/riscv.opt: Likewise.
30505 * config/riscv/sync.md: Likewise.
30506 * config/riscv/t-elf-multilib: Likewise.
30507 * config/riscv/t-linux: Likewise.
30508 * config/riscv/t-linux-multilib: Likewise.
30509 * config/riscv/t-riscv: Likewise.
30510 * configure.ac: Likewise.
30511 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
30512 Waterman as RISC-V maintainers.
30513 * doc/install.texi: Add RISC-V entries.
30514 * doc/invoke.texi: Add RISC-V options section.
30515 * doc/md.texi: Add RISC-V constraints section.
30516 * configure: Regenerated.
30518 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
30521 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
30522 false values to be constant vectors with all 0 or all 1 bits set.
30523 (vcondu<mode><mode>): Likewise.
30524 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
30526 (fpmask_comparison_operator): Update comment.
30527 (vecint_comparison_operator): New predicate.
30528 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
30529 vector conditionals when the true and false values are constant
30530 vectors with all 0 bits or all 1 bits set.
30532 2017-02-06 Martin Sebor <msebor@redhat.com>
30534 PR tree-optimization/79376
30535 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
30537 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
30539 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
30540 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
30541 to simplify split condition.
30543 2017-02-06 Jakub Jelinek <jakub@redhat.com>
30545 * omp-expand.c (oxpand_omp_atomic_fetch_op,
30546 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
30549 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
30551 PR rtl-optimization/68664
30552 * target.def (can_speculate_insn): New hook.
30553 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
30554 * doc/tm.texi: Regenerate.
30555 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
30556 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
30557 (rs6000_sched_can_speculate_insn): New function.
30559 2017-02-06 Jakub Jelinek <jakub@redhat.com>
30561 PR tree-optimization/79284
30562 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
30563 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
30564 vectorizable_mask_load_store, vectorizable_operation,
30565 vect_is_simple_cond, get_same_sized_vectype): Use it instead
30566 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
30567 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
30568 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
30569 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
30570 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
30571 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
30572 is_gimple_assign (stmt). Replace another such test with
30573 is_gimple_assign (stmt).
30575 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
30578 * config/avr/avr.c (rtl-iter.h): Include it.
30579 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
30580 (avr_legitimate_combined_insn): ...and implementation.
30582 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
30584 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
30585 * config/s390/s390.c (s390_const_operand_ok)
30586 (s390_canonicalize_comparison, s390_extract_part)
30587 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
30588 (s390_contiguous_bitmask_p, s390_rtx_costs)
30589 (legitimize_pic_address): Likewise.
30590 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
30591 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
30592 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
30593 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
30594 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
30596 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
30598 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
30599 REGNO($0) == REGNO($1).
30601 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30603 * config/s390/linux.h(SIZE_TYPE): Add comment.
30605 2017-02-06 Julian Brown <julian@codesourcery.com>
30606 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30607 Virendra Pathak <virendra.pathak@broadcom.com>
30609 * config/aarch64/aarch64-cores.def: Change the scheduler
30611 * config/aarch64/aarch64.md: Include thunderx2t99.md.
30612 * config/aarch64/thunderx2t99.md: New file.
30614 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30616 * doc/standards.texi (Go Language): Update link to language
30619 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
30621 * tree-eh.c (lower_resx): Sanitize profile.
30622 (cleanup_empty_eh_move_lp): Likewise.
30624 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
30627 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
30629 * cfgloopmanip.h (loop_version): Update prototype.
30630 * modulo-sched.c (sms_schedule): Update call of loop_version.
30631 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
30632 * tree-parloops.c (gen_parallel_loop): Likewise.
30633 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
30634 * tree-ssa-loop-split.c (split_loop): Likewise.
30635 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
30636 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
30638 2017-02-05 Martin Liska <mliska@suse.cz>
30641 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
30643 (print_operand_address): Initialize a struct to zero.
30645 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30647 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
30648 garbage collector only in textual form.
30650 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30652 * doc/extend.texi (x86 specific memory model extensions for
30653 transactional memory): Simplify a phrase.
30655 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
30658 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
30659 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
30660 (atomic_storedi_1): Likewise.
30662 2017-02-04 Jakub Jelinek <jakub@redhat.com>
30664 PR tree-optimization/79338
30665 * tree-parloops.c (gather_scalar_reductions): Don't call
30666 vect_analyze_loop_form for loop->inner before destroying loop's
30669 2017-02-03 Martin Sebor <msebor@redhat.com>
30671 PR tree-optimization/79327
30672 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
30673 when precision has resulted in leading zeros.
30674 (format_integer): Adjust the likely counter to assume an unknown
30675 argument that may be zero is non-zero.
30677 2017-02-03 Jason Merrill <jason@redhat.com>
30680 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
30681 avoid copying non-taken branch.
30683 2017-02-03 Jakub Jelinek <jakub@redhat.com>
30685 PR tree-optimization/79340
30686 * tree-vect-loop.c (vectorizable_reduction): Release
30687 vec_defs elements after safe_splicing them into other vectors.
30690 PR tree-optimization/79327
30691 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
30692 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
30694 (format_integer): Use wide_int_to_tree instead of build_int_cst
30695 + to_?hwi. If argmin is NULL, just set argmin and argmax to
30696 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
30697 of shortest and longest sequence.
30699 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
30701 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
30702 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
30704 2017-02-03 Walter Lee <walt@tilera.com>
30707 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
30708 after initial stackframe link reg save.
30709 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
30711 2017-02-03 Jakub Jelinek <jakub@redhat.com>
30714 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
30715 wu for stxssp alternative.
30717 2017-02-03 Martin Sebor <msebor@redhat.com>
30719 PR tree-optimization/79352
30720 * gimple-fold.c (get_range_strlen): Add argument.
30721 (get_range_strlen): Change return type to bool.
30722 (get_maxval_strlen): Pass in a dummy argument.
30723 * gimple-fold.h (get_range_strlen): Change return type to bool.
30724 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
30725 * tree.h (array_at_struct_end_p): Add argument.
30726 * tree.c (array_at_struct_end_p): Handle it.
30728 2017-02-03 Martin Liska <mliska@suse.cz>
30731 * multiple_target.c (create_dispatcher_calls): Redirect edge
30732 from a caller of a dispatcher.
30733 (expand_target_clones): Make the clones local.
30734 (ipa_target_clone): Do both target clones and resolvers.
30735 (ipa_dispatcher_calls): Remove the pass.
30736 (pass_dispatcher_calls::gate): Likewise.
30737 (make_pass_dispatcher_calls): Likewise.
30738 * passes.def (pass_target_clone): Put as very first IPA early
30741 2017-02-03 Martin Liska <mliska@suse.cz>
30743 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
30744 in case of a function with ifunc attribute.
30746 2017-02-03 Martin Liska <mliska@suse.cz>
30748 * cgraph.c (cgraph_node::dump): Dump function version info.
30749 * symtab.c (symtab_node::dump_base): Add missing new line.
30751 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
30753 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
30754 (ifcombine_ifandif): Use it.
30756 2017-02-03 Martin Liska <mliska@suse.cz>
30758 * doc/invoke.texi: Document default value for
30759 use-after-scope-direct-emission-threshold.
30761 2017-02-03 Martin Liska <mliska@suse.cz>
30763 PR tree-optimization/79339
30764 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
30765 (format_floating): Likewise.
30767 2017-02-03 Martin Liska <mliska@suse.cz>
30770 * ipa-prop.c (ipa_node_params_t::insert): Remove current
30772 (ipa_node_params_t::remove): Likewise.
30773 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
30774 initialization from removed ipa_node_params_t::insert.
30775 (ipa_node_params::~ipa_node_params): Move from removed
30776 ipa_node_params_t::release.
30777 * symbol-summary.h (symbol_summary::m_released): New member.
30778 Do not release a summary twice. Do not allow to call finalizer
30779 for types of a summary that live in GGC memory.
30781 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30783 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
30786 2017-02-02 Martin Sebor <msebor@redhat.com>
30788 PR middle-end/79275
30789 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
30790 (format_string): Tighten up the range of output for non-constant
30791 strings and correct the expected range for wide non-constant strings.
30793 2017-02-02 Martin Sebor <msebor@redhat.com>
30795 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
30797 PR middle-end/32003
30798 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
30800 (-fdump-tree-@var): Add to index and document how to come up
30801 with pass-specific option and dump file names.
30802 (-fdump-passes): Clarify where to look for output.
30804 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
30806 PR middle-end/77445
30807 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
30808 statistics of the analyzed path; allow threading for speed when
30809 any of BBs along the path are optimized for speed.
30811 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
30813 PR middle-end/78468
30814 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
30815 settings of the virtual registers.
30818 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
30820 * explow.c (get_dynamic_stack_size): Take known alignment of stack
30821 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
30824 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30826 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
30827 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
30829 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30831 * config/s390/s390.md: Add missing comments with the expanded
30833 * config/s390/vector.md: Likewise.
30834 * config/s390/vx-builtins.md: Likewise.
30836 2017-02-02 Jakub Jelinek <jakub@redhat.com>
30839 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
30840 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
30841 conditions on a single line.
30843 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30845 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
30846 __S390_VX__ to __VX__.
30848 2017-02-01 Andrew Pinski <apinski@cavium.com>
30850 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
30851 stmt_info to record_stmt_cost.
30852 (vect_get_known_peeling_cost): Pass stmt_info if known to
30854 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
30855 cpu_vector_cost field into
30856 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
30857 field into vec_int_stmt_cost and vec_fp_stmt_cost.
30858 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
30859 splitting of scalar_stmt_cost and vec_stmt_cost.
30860 (thunderx_vector_cost): Likewise.
30861 (cortexa57_vector_cost): LIkewise.
30862 (exynosm1_vector_cost): Likewise.
30863 (xgene1_vector_cost): Likewise.
30864 (thunderx2t99_vector_cost): Improve after the splitting of the two
30866 (aarch64_builtin_vectorization_cost): Update for the splitting of
30867 scalar_stmt_cost and vec_stmt_cost.
30869 2017-02-01 Torvald Riegel <triegel@redhat.com>
30870 Richard Henderson <rth@redhat.com>
30872 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
30873 conditional on existance of a fast atomic load.
30874 * optabs-query.c (can_atomic_load_p): New function.
30875 * optabs-query.h (can_atomic_load_p): Declare it.
30876 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
30877 no fast atomic load is available for the particular size of access.
30878 (expand_atomic_compare_and_swap): Likewise.
30879 (expand_atomic_load): Likewise.
30880 (expand_atomic_store): Likewise.
30881 (expand_atomic_fetch_op): Likewise.
30882 * testsuite/lib/target-supports.exp
30883 (check_effective_target_sync_int_128): Remove x86 because it provides
30884 no fast atomic load.
30885 (check_effective_target_sync_int_128_runtime): Likewise.
30887 2017-02-01 Richard Biener <rguenther@suse.de>
30889 * graphite.c: Include tree-vectorizer.h for find_loop_location.
30890 (graphite_transform_loops): Provide opt-info for optimized nests.
30891 * tree-parloop.c (parallelize_loops): Provide opt-info for
30892 parallelized loops.
30894 2017-02-01 Richard Biener <rguenther@suse.de>
30896 PR middle-end/79315
30897 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
30898 was not set before.
30900 2017-02-01 Richard Biener <rguenther@suse.de>
30902 PR tree-optimization/71824
30903 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
30904 Verify the loops are valid in the merged SESE region.
30905 (scop_detection::can_represent_loop_1): Check analyzing the
30906 evolution of the number of iterations in the region succeeds.
30908 2017-01-31 Ian Lance Taylor <iant@golang.org>
30910 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
30911 REG_ARGS_SIZE note to 32-bit push insns and call insn.
30913 2017-01-31 David Malcolm <dmalcolm@redhat.com>
30915 PR preprocessor/79210
30916 * input.c (get_substring_ranges_for_loc): Replace line_width
30917 assertion with error-handling.
30919 2017-01-31 Richard Biener <rguenther@suse.de>
30921 PR tree-optimization/77318
30922 * graphite-sese-to-poly.c (extract_affine): Fix assert.
30923 (create_pw_aff_from_tree): Take loop parameter.
30924 (add_condition_to_pbb): Pass loop of the condition to
30925 create_pw_aff_from_tree.
30927 2017-01-31 Jakub Jelinek <jakub@redhat.com>
30929 * config/s390/s390.c (s390_asan_shadow_offset): New function.
30930 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
30932 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
30936 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
30938 (convert_int_to_float128): Likewise.
30939 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
30940 (convert_int_to_float128): Likewise.
30941 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
30942 (UNSPEC_IEEE128_CONVERT): Likewise.
30943 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
30944 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
30945 Use local variables for IBM extended format.
30946 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
30947 (fix_trunc<mode>si2_fprs): Likewise.
30948 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
30949 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
30950 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
30951 to know that we can now have integers of all sizes in vector
30953 (fix<uns>_<mode>di2_hw): Likewise.
30954 (float<uns>_<mode>si2_hw): Likewise.
30955 (fix_<mode>si2_hw): Likewise.
30956 (fixuns_<mode>si2_hw): Likewise.
30957 (float<uns>_<mode>di2_hw): Likewise.
30958 (float_<mode>di2_hw): Likewise.
30959 (float_<mode>si2_hw): Likewise.
30960 (floatuns_<mode>di2_hw): Likewise.
30961 (floatuns_<mode>si2_hw): Likewise.
30962 (xscvqp<su>wz_<mode>): Delete, no longer used.
30963 (xscvqp<su>dz_<mode>): Likewise.
30964 (xscv<su>dqp_<mode>): Likewise.
30965 (ieee128_mfvsrd_64bit): Likewise.
30966 (ieee128_mfvsrd_32bit): Likewise.
30967 (ieee128_mfvsrwz): Likewise.
30968 (ieee128_mtvsrw): Likewise.
30969 (ieee128_mtvsrd_64bit): Likewise.
30970 (ieee128_mtvsrd_32bit): Likewise.
30972 2017-01-31 Martin Liska <mliska@suse.cz>
30975 * ipa-prop.c (ipa_free_all_node_params): Call release method
30976 instead of ~sumbol_summary to not to trigger double times
30979 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
30981 PR tree-optimization/71691
30982 * bitmap.h (class auto_bitmap): New.
30983 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
30984 is_maybe_undefined instead of ssa_undefined_value_p.
30986 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30988 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
30989 __S390_ARCH_LEVEL__ to __ARCH__.
30991 2017-01-31 Jakub Jelinek <jakub@redhat.com>
30993 PR tree-optimization/79267
30994 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
30995 if should_remove_lhs_p is true.
30997 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
31000 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
31001 (add_alignment_attribute): New.
31002 (base_type_die): Add alignment attribute.
31003 (subrange_type_die): Likewise.
31004 (modified_type_die): Likewise.
31005 (gen_array_type_die): Likewise.
31006 (gen_descr_array_type_die: Likewise.
31007 (gen_enumeration_type_die): Likewise.
31008 (gen_subprogram_die): Likewise.
31009 (gen_variable_die): Likewise.
31010 (gen_field_die): Likewise.
31011 (gen_ptr_to_mbr_type_die): Likewise.
31012 (gen_struct_or_union_type_die): Likewise.
31013 (gen_subroutine_type_die): Likewise.
31014 (gen_typedef_die): Likewise.
31015 (base_type_cmp): Compare alignment attribute.
31017 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31020 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
31021 (setb_unsigned) New pattern for setb with CCUNS.
31022 * config/rs6000/rs6000.c (expand_block_compare): Use a different
31023 subfc./subfe sequence to avoid overflow problems. Generate a
31024 shorter sequence with cmpld/setb for power9.
31025 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
31026 for generating subfc. instruction.
31027 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
31028 now uses this instruction.
31030 2017-01-30 Ian Lance Taylor <iant@google.com>
31033 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
31034 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
31036 2017-01-30 Martin Sebor <msebor@redhat.com>
31038 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
31039 Move constant to the right of a relational operator.
31040 (get_mpfr_format_length, format_character, format_string): Ditto.
31041 (should_warn_p, maybe_warn): Same.
31043 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
31045 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
31048 * asan.c (get_translation_unit_decl): Remove function.
31049 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
31051 2017-01-30 Martin Liska <mliska@suse.cz>
31053 PR gcov-profile/79259
31054 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
31055 -fprofile-generate.
31057 2017-01-30 Martin Liska <mliska@suse.cz>
31060 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
31061 Initialize variables with NULL value.
31063 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
31066 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
31068 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
31070 2017-01-30 Richard Biener <rguenther@suse.de>
31072 PR tree-optimization/79276
31073 * tree-vrp.c (process_assert_insertions): Properly adjust common
31074 when removing a duplicate.
31076 2017-01-30 Richard Biener <rguenther@suse.de>
31078 PR tree-optimization/79256
31079 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
31080 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
31082 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
31084 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
31087 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
31088 ("*r<noxa>sbg_<mode>_sll_bitmask")
31089 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
31090 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
31091 Use contiguous_bitmask_nowrap_operand.
31093 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31096 * config/rs6000/altivec.h (vec_xl): Revise #define.
31097 (vec_xst): Likewise.
31099 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
31101 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
31103 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
31105 PR rtl-optimization/79194
31106 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
31107 traps before call to bypass_conditional_jumps.
31109 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
31111 PR tree-optimization/71374
31112 * lra-constraints.c (check_conflict_input_operands): New.
31113 (match_reload): Use it.
31115 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
31118 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
31119 account to calculate conflict_set.
31121 2017-01-27 Bin Cheng <bin.cheng@arm.com>
31123 PR rtl-optimization/78559
31124 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
31125 other_insn in combine.
31127 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
31129 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
31130 uint16_type_node for BT_UINT16.
31132 2017-01-27 David Malcolm <dmalcolm@redhat.com>
31134 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
31135 "RTL Tests" to menu.
31136 (GIMPLE Tests): New node.
31137 (RTL Tests): New node.
31139 2017-01-27 Richard Biener <rguenther@suse.de>
31141 PR tree-optimization/79245
31142 * tree-loop-distribution.c (distribute_loop): Apply cost
31143 modeling also to detected patterns.
31145 2017-01-27 Richard Biener <rguenther@suse.de>
31147 PR tree-optimization/71433
31148 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
31149 (compare_assert_loc): New function.
31150 (process_assert_insertions): Sort and optimize assert locations
31151 to remove duplicates and push down identical assertions on
31152 edges to their destination block.
31154 2017-01-27 Richard Biener <rguenther@suse.de>
31156 PR tree-optimization/79244
31157 * tree-vrp.c (remove_range_assertions): Forcefully propagate
31158 out SSA names even if abnormal.
31160 2017-01-27 Jakub Jelinek <jakub@redhat.com>
31162 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
31163 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
31164 instead of MPFR_RNDN.
31166 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
31169 * arm.c (arm_option_override): Don't call build_target_option_node
31170 until after doing all option overrides.
31171 (arm_valid_target_attribute_tree): Likewise.
31173 2017-01-27 Martin Liska <mliska@suse.cz>
31175 * doc/invoke.texi (-fprofile-arcs): Document profiling support
31176 for {cd}tors and C++ {cd}tors.
31178 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
31180 * config/s390/s390.md ("*setmem_long_and")
31181 ("*setmem_long_and_31z"): Use zero_extend instead of and.
31183 2017-01-26 Martin Sebor <msebor@redhat.com>
31185 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
31188 2017-01-26 Martin Sebor <msebor@redhat.com>
31190 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
31191 HAVE_DFmode before using XFmode or DFmode.
31192 (parse_directive): Avoid using the z length modifier to avoid
31193 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
31195 PR middle-end/78703
31196 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
31197 to accept adjustment as an array.
31198 (get_int_range): New function.
31199 (struct directive): Make width and prec arrays.
31200 (directive::set_width, directive::set_precision): Call get_int_range.
31201 (format_integer, format_floating): Handle width and precision ranges.
31202 (format_string, parse_directive): Same.
31204 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31207 * dwarf2out.c (generate_skeleton_bottom_up): For children with
31208 comdat_type_p set, just clone them, but keep the children in the
31212 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
31213 which have direct callers with -fvar-tracking-assignments enabled
31215 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
31216 inside of type units.
31218 2017-01-26 Martin Sebor <msebor@redhat.com>
31220 PR middle-end/78703
31221 * gimple-ssa-sprintf.c (struct result_range): Add likely and
31223 (struct format_result): Replace number_chars, number_chars_min,
31224 and number_chars_max with a single member of struct result_range.
31226 (format_result::operator+=): Adjust.
31227 (struct fmtresult): Remove bounded. Handle likely and unlikely
31229 (fmtresult::adjust_for_width_or_precision): New function.
31230 (fmtresult:type_max_digits): New function.
31231 (bytes_remaining): Handle likely and unlikely counters.
31232 (min_bytes_remaining): Remove.
31233 (format_percent): Simplify.
31234 (format_integer, format_floating): Set likely and unlikely counters.
31235 (get_string_length, format_character, format_string): Same.
31236 (format_plain, should_warn_p): New function.
31237 (maybe_warn): Call should_warn_p. Update diagnostic messages
31238 and handle those for all directives, including plain strings.
31239 (format_directive): Handle likely and unlikely counters.
31240 Remove unnecessary quoting from diagnostics. Add an informational
31242 (add_bytes): Remove.
31243 (pass_sprintf_length::compute_format_length): Simplify.
31244 (try_substitute_return_value): Handle likely and unlikely counters.
31246 2017-01-26 Carl Love <cel@us.ibm.com>
31248 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
31249 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
31251 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
31254 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
31255 endianess for subregs into account.
31256 * lra-constraints.c (lra_constraints): Do risky transformations
31257 always on the first iteration.
31258 * lra-lives.c (check_pseudos_live_through_calls): Add arg
31259 last_call_used_reg_set.
31260 (process_bb_lives): Define and use last_call_used_reg_set.
31261 * lra.c (lra): Always continue after lra_constraints on the first
31264 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
31266 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
31268 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
31270 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31272 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
31273 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
31274 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
31275 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
31276 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
31277 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
31278 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
31279 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
31280 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
31282 2017-01-26 Marek Polacek <polacek@redhat.com>
31285 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
31286 for the third operand.
31288 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31290 PR middle-end/79236
31291 * omp-low.c (struct omp_context): Add simt_stmt field.
31292 (scan_omp_for): Return omp_context *.
31293 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
31294 context to the _simt_ SIMD stmt.
31295 (lower_omp_for): For combined SIMD with sibling _simt_
31296 SIMD, make sure to use the same decls in _looptemp_
31297 clauses as in the sibling.
31299 2017-01-26 David Sherwood <david.sherwood@arm.com>
31301 PR middle-end/79212
31302 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
31305 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31308 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
31309 emit fld b; fld a; if possible.
31311 * brig-builtins.def: Update copyright years.
31312 * config/arm/arm_acle_builtins.def: Update copyright years.
31314 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
31317 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
31318 constraint instead of o for the stxsd instruction.
31320 2017-01-25 Carl Love <cel@us.ibm.com>
31322 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
31323 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
31325 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
31327 * doc/invoke.texi (C++ Dialect Options): Fix typo.
31329 2017-01-25 Richard Biener <rguenther@suse.de>
31331 PR tree-optimization/69264
31332 * target.def (vector_alignment_reachable): Improve documentation.
31333 * doc/tm.texi: Regenerate.
31334 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
31336 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
31337 earlier changes with respect to TYPE_USER_ALIGN.
31338 (vector_alignment_reachable_p): Likewise. Improve dumping.
31340 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31343 * config/arm/arm.md (xordi3): Force constant operand into a register
31346 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31348 * doc/invoke.texi (-fstore-merging): Correct default optimization
31349 levels at which it is enabled.
31350 (-O): Move -fstore-merging from list to...
31353 2017-01-25 Richard Biener <rguenther@suse.de>
31356 * omp-expand.c: Include debug.h.
31357 (expand_omp_taskreg): Make sure to generate early debug before
31358 outlining anything from a function.
31359 (expand_omp_target): Likewise.
31360 (grid_expand_target_grid_body): Likewise.
31362 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
31365 * asan.c (get_translation_unit_decl): New function.
31366 (asan_add_global): Extract modules file name from globals
31367 TRANSLATION_UNIT_DECL name.
31369 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
31372 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
31373 for long calls with APCS frame and VFP.
31375 2017-01-24 David Malcolm <dmalcolm@redhat.com>
31377 * cfg.c (original_copy_tables_initialized_p): New function.
31378 * cfg.h (original_copy_tables_initialized_p): New decl.
31379 * cfgrtl.c (relink_block_chain): Guard the call to
31380 free_original_copy_tables with a call to
31381 original_copy_tables_initialized_p.
31382 * cgraph.h (symtab_node::native_rtl_p): New decl.
31383 * cgraphunit.c (symtab_node::native_rtl_p): New function.
31384 (symtab_node::needed_p): Don't assert for early assembly output
31385 for __RTL functions.
31386 (cgraph_node::finalize_function): Set "force_output" for __RTL
31388 (cgraph_node::analyze): Bail out early for __RTL functions.
31389 (analyze_functions): Update assertion to support __RTL functions.
31390 (cgraph_node::expand): Bail out early for __RTL functions.
31391 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
31393 * function.h (struct function): Update comment for field
31395 * gimple-expr.c: Include "tree-pass.h".
31396 (gimple_has_body_p): Return false for __RTL functions.
31397 * Makefile.in (OBJS): Add run-rtl-passes.o.
31398 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
31400 (gcc::pass_manager::get_clean_slate): New accessor.
31401 * passes.c: Include "insn-addr.h".
31402 (should_skip_pass_p): Add logging. Update logic for running
31403 "expand" to be compatible with both __GIMPLE and __RTL. Guard
31404 property-provider override so it is only done for gimple passes.
31406 (skip_pass): New function.
31407 (execute_one_pass): Call skip_pass when skipping passes.
31408 * read-md.c (md_reader::read_char): Support filtering
31409 the input to a subset of line numbers.
31410 (md_reader::md_reader): Initialize fields
31411 m_first_line and m_last_line.
31412 (md_reader::read_file_fragment): New function.
31413 * read-md.h (md_reader::read_file_fragment): New decl.
31414 (md_reader::m_first_line): New field.
31415 (md_reader::m_last_line): New field.
31416 * read-rtl-function.c (function_reader::create_function): Only
31417 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
31418 curr_properties. Set DECL_INITIAL to a dummy block.
31419 (read_rtl_function_body_from_file_range): New function.
31420 * read-rtl-function.h (read_rtl_function_body_from_file_range):
31422 * run-rtl-passes.c: New file.
31423 * run-rtl-passes.h: New file.
31425 2017-01-24 Jeff Law <law@redhat.com>
31427 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
31430 2017-01-24 Bin Cheng <bin.cheng@arm.com>
31432 PR tree-optimization/79159
31433 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
31434 (record_nonwrapping_iv): Improve boundary using above function if no
31435 value range information.
31437 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
31438 Martin Jambor <mjambor@suse.cz>
31440 * brig-builtins.def: New file.
31441 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
31442 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
31443 (DEF_HSAIL_SAT_BUILTIN): Likewise.
31444 (DEF_HSAIL_INTR_BUILTIN): Likewise.
31445 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
31446 * builtin-types.def (BT_INT8): New.
31447 (BT_INT16): Likewise.
31448 (BT_UINT8): Likewise.
31449 (BT_UINT16): Likewise.
31450 (BT_FN_ULONG): Likewise.
31451 (BT_FN_UINT_INT): Likewise.
31452 (BT_FN_UINT_ULONG): Likewise.
31453 (BT_FN_UINT_LONG): Likewise.
31454 (BT_FN_UINT_PTR): Likewise.
31455 (BT_FN_ULONG_PTR): Likewise.
31456 (BT_FN_INT8_FLOAT): Likewise.
31457 (BT_FN_INT16_FLOAT): Likewise.
31458 (BT_FN_UINT32_FLOAT): Likewise.
31459 (BT_FN_UINT16_FLOAT): Likewise.
31460 (BT_FN_UINT8_FLOAT): Likewise.
31461 (BT_FN_UINT64_FLOAT): Likewise.
31462 (BT_FN_UINT16_UINT32): Likewise.
31463 (BT_FN_UINT32_UINT16): Likewise.
31464 (BT_FN_UINT16_UINT16_UINT16): Likewise.
31465 (BT_FN_INT_PTR_INT): Likewise.
31466 (BT_FN_UINT_PTR_UINT): Likewise.
31467 (BT_FN_LONG_PTR_LONG): Likewise.
31468 (BT_FN_ULONG_PTR_ULONG): Likewise.
31469 (BT_FN_VOID_UINT64_UINT64): Likewise.
31470 (BT_FN_UINT8_UINT8_UINT8): Likewise.
31471 (BT_FN_INT8_INT8_INT8): Likewise.
31472 (BT_FN_INT16_INT16_INT16): Likewise.
31473 (BT_FN_INT_INT_INT): Likewise.
31474 (BT_FN_UINT_FLOAT_UINT): Likewise.
31475 (BT_FN_FLOAT_UINT_UINT): Likewise.
31476 (BT_FN_ULONG_UINT_UINT): Likewise.
31477 (BT_FN_ULONG_UINT_PTR): Likewise.
31478 (BT_FN_ULONG_ULONG_ULONG): Likewise.
31479 (BT_FN_UINT_UINT_UINT): Likewise.
31480 (BT_FN_VOID_UINT_PTR): Likewise.
31481 (BT_FN_UINT_UINT_PTR: Likewise.
31482 (BT_FN_UINT32_UINT64_PTR): Likewise.
31483 (BT_FN_INT_INT_UINT_UINT): Likewise.
31484 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
31485 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
31486 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
31487 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
31488 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
31489 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
31490 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
31491 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
31492 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
31493 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
31494 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
31495 * doc/frontends.texi: List BRIG FE.
31496 * doc/install.texi (Testing): Add BRIG tesring requirements.
31497 * doc/invoke.texi (Overall Options): Mention BRIG.
31498 * doc/standards.texi (Standards): Doucment BRIG HSA version.
31500 2017-01-24 Richard Biener <rguenther@suse.de>
31502 PR translation/79208
31503 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
31505 2017-01-24 Martin Jambor <mjambor@suse.cz>
31508 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
31509 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
31510 and known_contexts.
31512 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
31514 PR middle-end/79123
31515 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
31516 casts from signed to unsigned really don't have a range.
31518 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
31520 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
31521 GMP_RNDx for compatiblity.
31523 2017-01-24 Martin Liska <mliska@suse.cz>
31526 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
31527 that would prevent us to call alloca with -1 as argument.
31529 2017-01-24 Jakub Jelinek <jakub@redhat.com>
31531 * dwarf2out.c (output_compilation_unit_header, output_file_names):
31532 Avoid -Wformat-security warning.
31534 2017-01-23 Andrew Pinski <apinski@cavium.com>
31536 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
31539 2017-01-23 Martin Sebor <msebor@redhat.com>
31541 PR middle-end/78703
31542 * gimple-ssa-sprintf.c (warn_level): New global.
31543 (format_integer): Use it here and throughout the rest of the file.
31544 Use the same switch to compute sign as base.
31545 (maybe_warn): New function.
31546 (format_directive): Factor out warnings into maybe_warn.
31547 Add debugging output. Use warn_level.
31548 (add_bytes): Use warn_level.
31549 (pass_sprintf_length::compute_format_length): Add debugging output.
31550 (try_substitute_return_value): Same.
31551 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
31553 PR middle-end/78703
31554 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
31555 (struct fmtresult, format_integer, format_floating): Adjust.
31556 (fmtresult::fmtresult): Set max correctly in two argument ctor.
31557 (get_string_length, format_string,format_directive): Same.
31558 (pass_sprintf_length::compute_format_length): Same.
31559 (try_substitute_return_value): Simplify slightly.
31561 PR middle-end/78703
31562 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
31563 (fmtresult::operator+=): Outlined.
31564 (struct fmtresult): Add ctors.
31565 (struct conversion_spec): Rename...
31566 (struct directive): ...to this. Add and remove data members.
31567 (directive::set_width, directive::set_precision): New functions.
31568 (format_percent): Use fmtresult ctor.
31569 (get_width_and_precision): Remove.
31570 (format_integer): Make naming changes. Avoid computing width and
31572 (format_floating): Same. Adjust indentation.
31573 (format_character, format_none): New functions.
31574 (format_string): Moved character handling to format_character.
31575 (format_directive): Remove arguments, change return type.
31576 (parse_directive): New function.
31577 (pass_sprintf_length::compute_format_length): Move directive
31578 parsing to parse_directive.
31580 2017-01-23 Jakub Jelinek <jakub@redhat.com>
31582 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
31583 (assign_assembler_name_if_needed): ... this.
31584 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
31585 (assign_assembler_name_if_needed): ... this.
31586 (free_lang_data_in_cgraph): Adjust callers.
31587 * cgraphunit.c (cgraph_node::analyze): Likewise.
31588 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
31591 2017-01-23 Richard Biener <rguenther@suse.de>
31593 PR tree-optimization/79088
31594 PR tree-optimization/79188
31595 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
31596 resetting loop bounds after last path deletion. Reset loop
31597 bounds of the target loop, make code match the comments.
31598 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
31599 Make sure loops need no fixups.
31601 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
31603 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
31604 exponent support with double type for first argument.
31605 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
31606 type returned by __builtin_vec_extract_sig,
31607 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
31608 functions from "vector int" to "vector unsigned int" or from
31609 "vector long long int" to "vector unsigned long long int".
31610 Changed type returned by __builtin_vec_extract_exp,
31611 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
31612 functions from "vector int" to "vector unsigned int" or from
31613 "vector long long int" to "vector unsigned long long int".
31614 Changed return type of __builtin_vec_test_data_class,
31615 __builtin_vec_test_data_class_sp, and
31616 __builtin_vec_test_data_class_dp from "vector int" to
31617 "vector bool int" or from "vector long long int" to "vector bool
31618 long long int" and changed second argument type from "unsigned
31619 int" to "int". Added new overloaded function forms "vector float
31620 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
31621 "vector float __builtin_vec_insert_exp_sp (vector float, vector
31622 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
31623 double, vector unsigned long long int)" and "vector double
31624 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
31625 long int)". Changed return type of
31626 __builtin_scalar_test_data_class and
31627 __builtin_scalar_test_data_class_sp and
31628 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
31629 int" and changed second argument from "unsigned int" to "int".
31630 Changed type returned by __builtin_scalar_test_neg,
31631 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
31632 from "int" to "bool int". Added new overloaded function form
31633 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
31634 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
31635 exponent double-precision with floating point first argument.
31636 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
31637 documentation of scalar_test_data_class, scalar_test_neg,
31638 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
31639 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
31640 vec_test_data_class built-in functions to reflect refinements in
31641 their type signatures.
31643 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
31645 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
31647 (aarch64_elf_asm_destructor): Likewise.
31649 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
31651 PR rtl-optimization/78634
31652 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
31653 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
31654 * ifcvt.c (noce_try_cmove): Add missing cost check.
31656 PR rtl-optimization/71724
31657 * combine.c (if_then_else_cond): Look for situations where it is
31658 beneficial to undo the work of one of the recursive calls.
31660 2017-01-23 Bin Cheng <bin.cheng@arm.com>
31662 PR tree-optimization/70754
31663 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
31664 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
31665 combined stmt before it if not NULL.
31666 (combine_chains): Process refs reversely and compute dominance point
31669 2017-01-23 Martin Liska <mliska@suse.cz>
31671 PR tree-optimization/79196
31672 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
31673 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
31675 (strlen_optimize_stmt): Call the renamed function.
31677 2017-01-23 Michael Matz <matz@suse.de>
31679 PR tree-optimization/78384
31680 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
31682 2017-01-23 Richard Biener <rguenther@suse.de>
31684 PR tree-optimization/79186
31685 * tree-vrp.c (register_new_assert_for): Make sure we've seen
31686 both incoming edges before moving an assert.
31688 2017-01-23 Martin Jambor <mjambor@suse.cz>
31690 * ipa-prop.c (load_from_param_1): Removed.
31691 (load_from_unmodified_param): Bits from load_from_param_1 put back
31693 (load_from_param): Removed.
31694 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
31695 with stmt. Reverted back to use of load_from_unmodified_param.
31697 2017-01-23 Martin Jambor <mjambor@suse.cz>
31700 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
31701 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
31702 field a pointer to garbage collected vector, mark lattices and
31703 ipcp_orig_node with GTY((skip)).
31704 (ipa_get_param_count): Adjust to descriptors being a pointer.
31705 (ipa_get_param): Likewise.
31706 (ipa_get_type): Likewise.
31707 (ipa_get_param_move_cost): Likewise.
31708 (ipa_set_param_used): Likewise.
31709 (ipa_get_controlled_uses): Likewise.
31710 (ipa_set_controlled_uses): Likewise.
31711 (ipa_is_param_used): Likewise.
31712 (ipa_node_params_t): Move into garbage collector. New methods insert
31714 (ipa_node_params_sum): Annotate wth GTY(()).
31715 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
31717 (ipa_load_from_parm_agg): Adjust declaration.
31718 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
31719 * ipa-profile.c (ipa_profile): Likewise.
31720 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
31721 (ipa_populate_param_decls): Make descriptors parameter garbage
31723 (ipa_dump_param): Adjust to descriptors being a pointer.
31724 (ipa_alloc_node_params): Likewise.
31725 (ipa_initialize_node_params): Likewise.
31726 (load_from_param_1): Make descriptors parameter garbage collected.
31727 (load_from_unmodified_param): Likewise.
31728 (load_from_param): Likewise.
31729 (ipa_load_from_parm_agg): Likewise.
31730 (ipa_node_params::~ipa_node_params): Removed.
31731 (ipa_free_all_node_params): Remove call to delete operator.
31732 (ipa_node_params_t::insert): New.
31733 (ipa_node_params_t::remove): Likewise.
31734 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
31735 copy known_csts and known_contexts vectors.
31736 (ipa_read_node_info): Adjust to descriptors being a pointer.
31737 (ipcp_modif_dom_walker): Make m_descriptors field garbage
31739 (ipcp_transform_function): Make descriptors variable garbage
31742 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
31744 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
31745 * config/i386/avx512dqintrin.h: Ditto.
31746 * config/i386/avx512fintrin.h: Ditto.
31747 * config/i386/i386.c: Handle new builtins.
31748 * config/i386/i386-builtin.def: Add new builtins.
31749 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
31750 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
31752 2017-01-23 Jakub Jelinek <jakub@redhat.com>
31753 Martin Liska <mliska@suse.cz>
31755 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
31756 * asan.c (asan_expand_poison_ifn): Support stores and use
31757 appropriate ASAN report function.
31758 * internal-fn.c (expand_ASAN_POISON_USE): New function.
31759 * internal-fn.def (ASAN_POISON_USE): Declare.
31760 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
31761 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
31762 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
31763 ASAN_POISON calls w/o LHS.
31764 * tree-ssa.c (execute_update_addresses_taken): Create clobber
31765 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
31766 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
31767 * gimplify.c (asan_poison_variables): Add attribute
31768 use_after_scope_memory to variables that really needs to live
31770 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
31771 having the attribute.
31773 2017-01-23 Martin Liska <mliska@suse.cz>
31775 * asan.c (create_asan_shadow_var): New function.
31776 (asan_expand_poison_ifn): Likewise.
31777 * asan.h (asan_expand_poison_ifn): New declaration.
31778 * internal-fn.c (expand_ASAN_POISON): Likewise.
31779 * internal-fn.def (ASAN_POISON): New builtin.
31780 * sanopt.c (pass_sanopt::execute): Expand
31781 asan_expand_poison_ifn.
31782 * tree-inline.c (copy_decl_for_dup_finish): Make function
31784 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
31785 * tree-ssa.c (is_asan_mark_p): New function.
31786 (execute_update_addresses_taken): Rewrite local variables
31787 (identified just by use-after-scope as addressable) into SSA.
31789 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
31791 * doc/install.texi (Specific): opensource.apple.com uses https
31792 now. Remove trailing slash.
31794 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
31796 * README.Portability: Remove note on an Irix compatibility issue.
31798 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
31800 * gcov.c (INCLUDE_ALGORITHM): Define.
31801 (INCLUDE_VECTOR): Define.
31802 No longer include <vector> and <algorithm> directly.
31804 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
31806 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
31808 * doc/invoke.texi (Code Gen Options): Ditto.
31810 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
31813 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
31815 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
31817 rtl-optimization/79125
31818 * cprop.c (local_cprop_pass): Handle cases where we make an
31819 unconditional trap.
31821 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
31825 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
31826 read from, for big endian.
31828 2017-01-20 Jiong Wang <jiong.wang@arm.com>
31830 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
31831 register pauth builtins for LP64 only.
31833 2017-01-20 Marek Polacek <polacek@redhat.com>
31836 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
31839 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
31841 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
31843 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
31844 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
31845 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
31847 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31850 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
31851 in big-endian mode when they are not a single duplicated value.
31853 2017-01-20 Richard Biener <rguenther@suse.de>
31855 * BASE-VER: Bump to 7.0.1.
31857 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
31859 * omp-low.c (omplow_simd_context): New struct. Use it...
31860 (lower_rec_simd_input_clauses): ...here and...
31861 (lower_rec_input_clauses): ...here to hold common data. Adjust all
31862 references to idx, lane, max_vf, is_simt.
31864 2017-01-20 Graham Markall <graham.markall@embecosm.com>
31866 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
31869 2017-01-20 Martin Jambor <mjambor@suse.cz>
31871 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
31872 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
31874 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
31875 (GTFILES): Rename hsa.c to hsa-common.c.
31876 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
31877 * hsa-dump.c: Likewise.
31878 * hsa-gen.c: Likewise.
31879 * hsa-regalloc.c: Likewise.
31880 * ipa-hsa.c: Likewise.
31881 * omp-expand.c: Likewise.
31882 * omp-low.c: Likewise.
31883 * toplev.c: Likewise.
31885 2017-01-20 Marek Polacek <polacek@redhat.com>
31888 * doc/invoke.texi: Document -Wduplicated-branches.
31889 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
31890 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
31891 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
31892 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
31893 return 0 only when not OEP_LEXICOGRAPHIC.
31894 (fold_build_cleanup_point_expr): Use the expression
31895 location when building CLEANUP_POINT_EXPR.
31896 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
31897 * tree.c (add_expr): Handle error_mark_node.
31899 2017-01-20 Martin Liska <mliska@suse.cz>
31902 * tree-profile.c (init_ic_make_global_vars): Do not call
31904 (gimple_init_gcov_profiler): Likewise.
31906 2017-01-20 Martin Liska <mliska@suse.cz>
31909 * cgraph.h (maybe_create_reference): Remove argument and
31911 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
31913 * ipa-cp.c (create_specialized_node): Likewise.
31914 * symtab.c (symtab_node::maybe_create_reference): Handle
31915 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
31917 2017-01-20 Martin Liska <mliska@suse.cz>
31919 * read-rtl-function.c (function_reader::create_function): Use
31920 build_decl instread of build_decl_stat.
31922 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
31924 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
31925 * config/i386/avx512dqintrin.h: Ditto.
31926 * config/i386/avx512fintrin.h: Ditto.
31927 * config/i386/i386-builtin-types.def: Add new types.
31928 * config/i386/i386.c: Handle new types.
31929 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
31930 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
31931 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
31932 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
31933 (__builtin_ia32_kshiftridi): New.
31934 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
31936 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
31940 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
31941 define to rs6000_init_stack_protect_guard.
31942 (rs6000_init_stack_protect_guard): New function.
31944 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
31945 Yunqiang Su <yunqiang.su@imgtec.com>
31947 * config.gcc (supported_defaults): Add madd4.
31948 (with_madd4): Add validation.
31949 (all_defaults): Add madd4.
31950 * config/mips/mips.opt (mmadd4): New option.
31951 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
31953 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
31955 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
31956 (ISA_HAS_FUSED_MADD4): Likewise.
31957 * doc/invoke.texi (-mmadd4): Document the new option.
31958 * doc/install.texi (--with-madd4): Document the new option.
31960 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31962 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
31963 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
31964 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
31965 (aarch64_init_pauth_hint_builtins): New.
31966 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
31967 (aarch64_expand_builtin): Expand new builtins.
31969 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31971 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
31972 * combine-stack-adj.c (no_unhandled_cfa): Handle
31973 REG_CFA_TOGGLE_RA_MANGLE.
31974 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
31975 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
31976 info for return address signing.
31977 (aarch64_expand_epilogue): Likewise.
31979 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31981 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
31982 * config/aarch64/aarch64-protos.h
31983 (aarch64_return_address_signing_enabled): New declaration.
31984 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
31986 (aarch64_expand_prologue): Sign return address before it's pushed onto
31988 (aarch64_expand_epilogue): Authenticate return address fetched from
31990 (aarch64_override_options): Sanity check for ILP32 and ISA level.
31991 (aarch64_attributes): New function attributes for "sign-return-address".
31992 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
31993 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
31994 ("*do_return"): Generate combined instructions according to key index.
31995 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
31996 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
31998 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
31999 * config/aarch64/aarch64.opt (msign-return-address=): New.
32000 * doc/extend.texi (AArch64 Function Attributes): Documents
32001 "sign-return-address=".
32002 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
32004 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
32006 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
32007 overall option summary.
32009 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32011 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
32012 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
32013 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
32014 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
32016 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
32018 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
32019 -mpower9-minmax by default for -mcpu=power9.
32020 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
32021 128-bit floating point.
32023 2017-01-20 Alan Modra <amodra@gmail.com>
32025 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
32026 optimizing for size.
32028 2017-01-20 Alan Modra <amodra@gmail.com>
32031 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
32032 for strcmp and strncmp from corresponding builtin decl.
32034 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
32036 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
32037 instead of i386/rtems-64.h.
32038 * config/i386/rtems-64.h: Remove.
32040 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
32044 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
32046 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
32048 2017-01-19 Tamar Christina <tamar.christina@arm.com>
32050 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
32051 Change int to HOST_WIDE_INT.
32052 * config/aarch64/aarch64-protos.h
32053 (aarch64_simd_gen_const_vector_dup): Likewise.
32054 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
32056 2017-01-19 David Malcolm <dmalcolm@redhat.com>
32058 * langhooks-def.h (lhd_type_for_size): New decl.
32059 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
32060 * langhooks.c (lhd_type_for_size): New function, taken from
32063 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
32065 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
32066 define_bypass for CR latency.
32067 (power9-cracked-alu): Update bypass latency and remove power9-branch.
32068 (power9-alu2): Add define_bypass for CR latency.
32070 (power9-mul): Update insn latency.
32071 (power9-mul-compare): Update insn latency, bypass latency and remove
32074 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32076 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
32078 * config/aarch64/aarch64.md
32079 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
32080 aarch64_nopcrelative_literal_loads.
32081 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
32083 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
32085 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
32086 TARGET_LOONGSON_3A.
32087 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
32089 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
32092 * config.gcc (supported_defaults): Add lxc1-sxc1.
32093 (with_lxc1_sxc1): Add validation.
32094 (all_defaults): Add lxc1-sxc1.
32095 * config/mips/mips.opt (mlxc1-sxc1): New option.
32096 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
32098 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
32099 __mips_no_lxc1_sxc1.
32100 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
32101 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
32102 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
32104 2017-01-19 Richard Biener <rguenther@suse.de>
32106 PR tree-optimization/72488
32107 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
32108 sure to restore SSA info.
32109 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
32111 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
32113 PR rtl-optimization/79121
32114 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
32115 of the inner type when shifting an extended value.
32117 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
32120 * symtab.c (symtab_node::equal_address_to): Fix comparing of
32121 interposable aliases.
32123 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
32126 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
32127 Use the evmergelohi instruction.
32128 (mov_si<mode>_e500_subreg4_2_le): Likewise.
32129 (mov_sitf_e500_subreg8_2_be): Likewise.
32130 (mov_sitf_e500_subreg12_2_le): Likewise.
32131 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
32132 (mov_si<mode>_e500_subreg4_2_be): Likewise.
32133 (mov_sitf_e500_subreg8_2_le): Likewise.
32134 (mov_sitf_e500_subreg12_2_be): Likewise.
32136 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32138 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
32139 attribute from vecsimple to vecperm.
32140 (altivec_vbpermq2): Likewise.
32142 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32145 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
32147 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32148 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
32149 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
32150 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
32151 case where N arg is SIZE_MAX.
32152 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
32153 (cmpstrsi): Add pattern.
32155 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
32157 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
32158 __builtin_vec_revb builtins.
32159 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
32160 built-in functions to support generation of the ISA 3.0 XXBR<x>
32161 vector byte reverse instructions.
32162 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
32163 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
32164 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
32165 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
32166 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
32167 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
32168 (P9V_BUILTIN_VEC_REVB): Likewise.
32169 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
32170 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
32171 (p9_xxbrq_v16qi): Likewise.
32172 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
32173 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
32174 (p9_xxbrh_v8hi): Likewise.
32175 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
32176 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
32177 vec_revb built-in functions.
32179 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
32181 PR rtl-optimization/78952
32182 * config/i386/i386.md (any_extract): New code iterator.
32183 (*insvqi_2): Use any_extract for source operand.
32184 (*insvqi_3): Use any_shiftrt for source operand.
32186 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
32188 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
32190 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
32192 2017-01-18 Matthias Klose <doko@ubuntu.com>
32194 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
32196 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32198 * config/rs6000/altivec.h (vec_bperm): Change #define.
32199 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
32200 (altivec_vbpermq2): New define_insn.
32201 (altivec_vbpermd): Likewise.
32202 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
32203 function interface.
32204 (VBPERMD): Likewise.
32205 (VBPERM): New polymorphic function interface.
32206 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
32207 Add entries for P9V_BUILTIN_VEC_VBPERM.
32208 * doc/extend.texi: Add interfaces for vec_bperm.
32210 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32212 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
32213 first letter of error messages.
32214 (s390_resolve_overloaded_builtin): Likewise.
32215 * config/s390/s390.c (s390_expand_builtin): Likewise.
32216 (s390_invalid_arg_for_unprototyped_fn): Likewise.
32217 (s390_valid_target_attribute_inner_p): Likewise.
32218 * config/s390/s390.md ("tabort"): Likewise.
32220 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
32222 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
32223 (ISA_AVOID_DIV_HILO): New macro.
32224 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
32225 (ISA_HAS_DDIV): Likewise.
32227 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32229 * doc/invoke.texi (fabi-version): Correct number of occurrences.
32231 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32233 * doc/invoke.texi (fabi-version): Spelling fix.
32235 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32238 * doc/invoke.texi (fabi-version): Mention mangling fix for
32241 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32244 * doc/invoke.texi (fabi-version): Document discriminator mangling.
32246 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
32249 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
32250 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
32252 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
32254 (stack_protect_test): Ditto.
32255 * config/rs6000/rs6000.opt (mstack-protector-guard=,
32256 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
32258 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
32259 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
32260 -mstack-protector-guard-offset=.
32261 (RS/6000 and PowerPC Options): Ditto.
32263 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
32265 * config/i386/i386.h (MASK_CLASS_P): New define.
32266 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
32267 there are no registers from different register sets also when
32268 mask registers are used. Update function comment.
32269 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
32270 to (*k/*r) and (*k/*km) alternatives.
32272 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
32274 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
32275 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
32276 (EH_RETURN_HANDLER_RTX): New define.
32277 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
32278 Force frame pointer in EH return functions.
32279 (aarch64_expand_epilogue): Add barrier for eh_return.
32280 (aarch64_final_eh_return_addr): Remove.
32281 (aarch64_eh_return_handler_rtx): New function.
32282 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
32284 (aarch64_eh_return_handler_rtx): New prototype.
32286 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32288 * config/rs6000/altivec.h (vec_rlmi): New #define.
32289 (vec_vrlnm): Likewise.
32290 (vec_rlnm): Likewise.
32291 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
32292 (UNSPEC_VRLNM): Likewise.
32293 (VIlong): New mode iterator.
32294 (altivec_vrl<VI_char>mi): New define_insn.
32295 (altivec_vrl<VI_char>nm): Likewise.
32296 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
32298 (VRLDNM): Likewise.
32299 (RLNM): New polymorphic function entry.
32300 (VRLWMI): New monomorphic function entry.
32301 (VRLDMI): Likewise.
32302 (RLMI): New polymorphic function entry.
32303 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
32304 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
32305 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
32308 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32311 * dwarf2out.c (field_byte_offset): Restore the
32312 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
32313 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
32314 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
32317 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
32320 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
32322 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32325 * dwarf2out.c (add_data_member_location_attribute): For constant
32326 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
32327 instead of DW_AT_data_member_location, DW_AT_bit_offset and
32328 DW_AT_byte_size attributes.
32330 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
32332 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
32333 after forcing to constant memory when the code model is medium.
32335 2017-01-17 Julia Koval <julia.koval@intel.com>
32338 * config/i386/avx512fintrin.h
32339 (_mm512_i32gather_ps): Change __addr type to void const*.
32340 (_mm512_mask_i32gather_ps): Ditto.
32341 (_mm512_i32gather_pd): Ditto.
32342 (_mm512_mask_i32gather_pd): Ditto.
32343 (_mm512_i64gather_ps): Ditto.
32344 (_mm512_mask_i64gather_ps): Ditto.
32345 (_mm512_i64gather_pd): Ditto.
32346 (_mm512_mask_i64gather_pd): Ditto.
32347 (_mm512_i32gather_epi32): Ditto.
32348 (_mm512_mask_i32gather_epi32): Ditto.
32349 (_mm512_i32gather_epi64): Ditto.
32350 (_mm512_mask_i32gather_epi64): Ditto.
32351 (_mm512_i64gather_epi32): Ditto.
32352 (_mm512_mask_i64gather_epi32): Ditto.
32353 (_mm512_i64gather_epi64): Ditto.
32354 (_mm512_mask_i64gather_epi64): Ditto.
32355 (_mm512_i32scatter_ps): Change __addr type to void*.
32356 (_mm512_mask_i32scatter_ps): Ditto.
32357 (_mm512_i32scatter_pd): Ditto.
32358 (_mm512_mask_i32scatter_pd): Ditto.
32359 (_mm512_i64scatter_ps): Ditto.
32360 (_mm512_mask_i64scatter_ps): Ditto.
32361 (_mm512_i64scatter_pd): Ditto.
32362 (_mm512_mask_i64scatter_pd): Ditto.
32363 (_mm512_i32scatter_epi32): Ditto.
32364 (_mm512_mask_i32scatter_epi32): Ditto.
32365 (_mm512_i32scatter_epi64): Ditto.
32366 (_mm512_mask_i32scatter_epi64): Ditto.
32367 (_mm512_i64scatter_epi32): Ditto.
32368 (_mm512_mask_i64scatter_epi32): Ditto.
32369 (_mm512_i64scatter_epi64): Ditto.
32370 (_mm512_mask_i64scatter_epi64): Ditto.
32371 * config/i386/avx512pfintrin.h
32372 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
32373 (_mm512_mask_prefetch_i32gather_ps): Ditto.
32374 (_mm512_mask_prefetch_i64gather_pd): Ditto.
32375 (_mm512_mask_prefetch_i64gather_ps): Ditto.
32376 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
32377 (_mm512_prefetch_i32scatter_ps): Ditto.
32378 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
32379 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
32380 (_mm512_prefetch_i64scatter_pd): Ditto.
32381 (_mm512_prefetch_i64scatter_ps): Ditto.
32382 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
32383 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
32384 * config/i386/avx512vlintrin.h
32385 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
32386 (_mm_mmask_i32gather_ps): Ditto.
32387 (_mm256_mmask_i32gather_pd): Ditto.
32388 (_mm_mmask_i32gather_pd): Ditto.
32389 (_mm256_mmask_i64gather_ps): Ditto.
32390 (_mm_mmask_i64gather_ps): Ditto.
32391 (_mm256_mmask_i64gather_pd): Ditto.
32392 (_mm_mmask_i64gather_pd): Ditto.
32393 (_mm256_mmask_i32gather_epi32): Ditto.
32394 (_mm_mmask_i32gather_epi32): Ditto.
32395 (_mm256_mmask_i32gather_epi64): Ditto.
32396 (_mm_mmask_i32gather_epi64): Ditto.
32397 (_mm256_mmask_i64gather_epi32): Ditto.
32398 (_mm_mmask_i64gather_epi32): Ditto.
32399 (_mm256_mmask_i64gather_epi64): Ditto.
32400 (_mm_mmask_i64gather_epi64): Ditto.
32401 (_mm256_i32scatter_ps): Change __addr type to void*.
32402 (_mm256_mask_i32scatter_ps): Ditto.
32403 (_mm_i32scatter_ps): Ditto.
32404 (_mm_mask_i32scatter_ps): Ditto.
32405 (_mm256_i32scatter_pd): Ditto.
32406 (_mm256_mask_i32scatter_pd): Ditto.
32407 (_mm_i32scatter_pd): Ditto.
32408 (_mm_mask_i32scatter_pd): Ditto.
32409 (_mm256_i64scatter_ps): Ditto.
32410 (_mm256_mask_i64scatter_ps): Ditto.
32411 (_mm_i64scatter_ps): Ditto.
32412 (_mm_mask_i64scatter_ps): Ditto.
32413 (_mm256_i64scatter_pd): Ditto.
32414 (_mm256_mask_i64scatter_pd): Ditto.
32415 (_mm_i64scatter_pd): Ditto.
32416 (_mm_mask_i64scatter_pd): Ditto.
32417 (_mm256_i32scatter_epi32): Ditto.
32418 (_mm256_mask_i32scatter_epi32): Ditto.
32419 (_mm_i32scatter_epi32): Ditto.
32420 (_mm_mask_i32scatter_epi32): Ditto.
32421 (_mm256_i32scatter_epi64): Ditto.
32422 (_mm256_mask_i32scatter_epi64): Ditto.
32423 (_mm_i32scatter_epi64): Ditto.
32424 (_mm_mask_i32scatter_epi64): Ditto.
32425 (_mm256_i64scatter_epi32): Ditto.
32426 (_mm256_mask_i64scatter_epi32): Ditto.
32427 (_mm_i64scatter_epi32): Ditto.
32428 (_mm_mask_i64scatter_epi32): Ditto.
32429 (_mm256_i64scatter_epi64): Ditto.
32430 (_mm256_mask_i64scatter_epi64): Ditto.
32431 (_mm_i64scatter_epi64): Ditto.
32432 (_mm_mask_i64scatter_epi64): Ditto.
32433 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
32434 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
32435 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
32436 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
32437 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
32438 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
32439 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
32440 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
32441 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
32442 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
32443 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
32444 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
32445 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
32446 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
32447 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
32448 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
32449 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
32450 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
32451 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
32452 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
32453 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
32454 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
32455 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
32456 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
32457 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
32458 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
32459 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
32460 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
32461 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
32462 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
32463 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
32464 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
32465 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
32466 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
32467 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
32468 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
32469 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
32470 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
32471 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
32472 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
32473 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
32474 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
32475 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
32476 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
32477 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
32478 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
32479 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
32480 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
32481 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
32482 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
32483 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
32484 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
32485 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
32486 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
32487 definitions accordingly.
32489 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
32490 Kuan-Lin Chen <kuanlinchentw@gmail.com>
32493 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
32496 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
32499 * ira-conflicts.c (ira_build_conflicts): Update total conflict
32500 hard regs for inner regno.
32502 2017-01-17 Martin Liska <mliska@suse.cz>
32505 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
32506 assumption and add comment.
32508 2017-01-17 Nathan Sidwell <nathan@acm.org>
32510 * ipa-visibility.c (localize_node): New function, broken out of ...
32511 (function_and_variable_visibility): ... here. Call it.
32513 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
32515 PR middle-end/77445
32516 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
32517 correctly set frequency of oudgoing edge.
32518 (duplicate_thread_path): Fix profile updating.
32520 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32523 * configure.ac: Add GCC_BASE_VER.
32524 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
32525 version from BASE-VER file.
32526 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
32527 (gcc.o): Depend on $(BASEVER).
32528 * common.opt (dumpfullversion): New option.
32529 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
32530 * doc/invoke.texi: Document -dumpfullversion.
32531 * doc/install.texi: Document --with-gcc-major-version-only.
32532 * configure: Regenerated.
32534 2017-01-17 Richard Biener <rguenther@suse.de>
32536 PR tree-optimization/71433
32537 * tree-vrp.c (register_new_assert_for): Merge same asserts
32538 on all incoming edges.
32539 (process_assert_insertions_for): Handle insertions at the
32542 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
32544 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
32545 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
32547 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
32550 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
32553 2017-01-17 Alan Modra <amodra@gmail.com>
32556 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
32557 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
32558 symbolic stack limit when pic.
32560 2017-01-16 Martin Sebor <msebor@redhat.com>
32562 PR tree-optimization/78608
32563 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
32565 2017-01-16 Jeff Law <law@redhat.com>
32568 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
32569 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
32570 for several include directories that may be relative to sysroot.
32571 * config/i386/x-mingw32 (gplus_includedir): Define.
32572 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
32573 (native_system_includedir): Likewise.
32574 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
32575 override if TARGET_SYSTEM_ROOT is defined.
32576 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
32578 PR tree-optimization/79090
32579 PR tree-optimization/33562
32580 PR tree-optimization/61912
32581 PR tree-optimization/77485
32582 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
32583 and computed trims into the dump file.
32585 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
32587 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
32589 2017-01-16 Jakub Jelinek <jakub@redhat.com>
32592 * gimplify.c (gimplify_init_constructor): If want_value and
32593 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
32597 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
32598 sequence. Formatting fixes.
32599 (doloop_optimize): Formatting fixes.
32602 * gcc.c (debug_level_greater_than_spec_func): New function.
32603 (static_spec_functions): Add debug-level-gt spec function.
32604 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
32606 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
32607 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
32608 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
32609 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
32610 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
32611 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
32613 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
32615 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
32616 QImode fixups to general and mask registers only.
32618 2017-01-16 Carl Love <cel@us.ibm.com>
32620 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
32621 for built-in functions
32622 vector signed char vec_nabs (vector signed char)
32623 vector signed short vec_nabs (vector signed short)
32624 vector signed int vec_nabs (vector signed int)
32625 vector signed long long vec_nabs (vector signed long long)
32626 vector float vec_nabs (vector float)
32627 vector double vec_nabs (vector double)
32628 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
32630 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
32631 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
32632 * doc/extend.texi: Update the documentation file for the new built-in
32635 2017-01-16 Martin Sebor <msebor@redhat.com>
32637 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
32640 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32642 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
32643 UNSPEC_VSX__XXSPLTD to require special splat handling.
32645 2017-01-16 David Malcolm <dmalcolm@redhat.com>
32648 * system.h: Poison strndup.
32650 2017-01-16 Alan Modra <amodra@gmail.com>
32653 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
32656 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
32658 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
32660 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
32662 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
32663 call recog here. Assert that INSN_CODE (insn) is non-negative.
32665 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
32668 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
32670 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
32671 in the currently scheduled RTL fragment.
32673 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
32675 PR rtl-optimization/78751
32676 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
32679 2017-01-14 Jeff Law <law@redhat.com>
32681 PR tree-optimization/79090
32682 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
32683 variable length stores.
32684 (compute_trims): Delete dead assignment to *trim_tail.
32685 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
32688 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
32690 PR rtl-optimization/78626
32691 PR rtl-optimization/78727
32692 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
32693 of a block, and split such blocks after everything else is finished.
32695 2017-01-14 Alan Modra <amodra@gmail.com>
32698 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
32699 target legitimate_combined_insn.
32700 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
32701 (rs6000_legitimate_combined_insn): New function.
32702 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
32704 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
32705 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
32706 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
32708 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
32710 * doc/frontends.texi (G++ and GCC): Remove references to Java.
32712 2017-01-13 Jeff Law <law@redhat.com>
32714 PR tree-optimization/33562
32715 PR tree-optimization/61912
32716 PR tree-optimization/77485
32717 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
32719 (delete_dead_assignment): Likewise.
32720 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
32721 statement to delete_dead_call and delete_dead_assignment.
32723 2017-01-13 David Malcolm <dmalcolm@redhat.com>
32726 * substring-locations.c (format_warning_va): Strengthen case 1 so
32727 that both endpoints of the substring must be within the format
32728 range for just the substring to be printed.
32730 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
32732 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
32733 * config/i386/i386.c (ix86_target_string): Add missing options
32734 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
32735 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
32736 flags_other and ix86_target_other to flags2_other. Display unknown
32738 (ix86_valid_target_attribute_inner_p): Add missing options and
32739 reorder options by implied ISAs, as in ix86_target_string.
32741 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
32743 * hash-table.h (hash_table::too_empty_p): New function.
32744 (hash_table::expand): Use it.
32745 (hash_table::traverse): Likewise.
32746 (hash_table::empty_slot): Use sizeof (value_type) instead of
32747 sizeof (PTR) to convert bytes to elements. Shrink the table
32748 if the current size is excessive for the current number of
32751 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
32753 * ira-costs.c (record_reg_classes): Break from the inner loop
32754 early once alt_fail is known to be true. Update outer loop
32755 handling accordingly.
32757 2017-01-13 Jeff Law <law@redhat.com>
32759 * tree-ssa-dse.c (decrement_count): New function.
32760 (increment_start_addr, maybe_trim_memstar_call): Likewise.
32761 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
32762 when we know the partially dead statement is a mem* function.
32764 PR tree-optimization/61912
32765 PR tree-optimization/77485
32766 * tree-ssa-dse.c: Include expr.h.
32767 (maybe_trim_constructor_store): New function.
32768 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
32770 PR tree-optimization/33562
32771 PR tree-optimization/61912
32772 PR tree-optimization/77485
32773 * doc/invoke.texi: Document new dse-max-object-size param.
32774 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
32775 * tree-ssa-dse.c: Include params.h.
32776 (dse_store_status): New enum.
32777 (initialize_ao_ref_for_dse): New, partially extracted from
32779 (valid_ao_ref_for_dse, normalize_ref): New.
32780 (setup_live_bytes_from_ref, compute_trims): Likewise.
32781 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
32782 (maybe_trim_partially_dead_store): Likewise.
32783 (maybe_trim_complex_store): Likewise.
32784 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
32785 Track what bytes live from the original store. Return tri-state
32786 for dead, partially dead or live.
32787 (dse_dom_walker): Add constructor, destructor and new private members.
32788 (delete_dead_call, delete_dead_assignment): New extracted from
32790 (dse_optimize_stmt): Make a member of dse_dom_walker.
32791 Use initialize_ao_ref_for_dse.
32793 PR tree-optimization/33562
32794 PR tree-optimization/61912
32795 PR tree-optimization/77485
32796 * sbitmap.h (bitmap_count_bits): Prototype.
32797 (bitmap_clear_range, bitmap_set_range): Likewise.
32798 * sbitmap.c (bitmap_clear_range): New function.
32799 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
32801 2017-01-13 Martin Liska <mliska@suse.cz>
32804 * function.c (set_cfun): Add new argument force.
32805 * function.h (set_cfun): Likewise.
32806 * ipa-inline-transform.c (inline_call): Use the function when
32807 strict alising from is dropped for function we inline to.
32809 2017-01-13 Richard Biener <rguenther@suse.de>
32811 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
32812 for dumping GIMPLE INTEGER_CSTs.
32814 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
32816 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
32817 to 201112L since C++17.
32819 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
32822 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
32823 if -fsanitize=kernel-address is present.
32825 2017-01-13 Richard Biener <rguenther@suse.de>
32827 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
32828 as _Literal ( type ) number in case usual suffixes do not
32829 preserve all information.
32831 2017-01-13 Richard Biener <rguenther@suse.de>
32833 PR tree-optimization/77283
32834 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
32835 and ssa-iterators.h.
32836 (is_feasible_trace): Implement a cost model based on joiner
32839 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
32842 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
32843 char or short to __float128/_Float128 directly.
32845 2017-01-12 Martin Sebor <msebor@redhat.com>
32847 to -Wformat-overflow.
32848 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
32849 (min_bytes_remaining): Same.
32850 (get_string_length): Same.
32851 (format_string): Same.
32852 (format_directive): Same.
32854 (pass_sprintf_length::handle_gimple_call): Same.
32856 2017-01-12 Jakub Jelinek <jakub@redhat.com>
32858 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
32859 info.nowrite calls with no lhs that can't throw. Return bool
32860 whether gsi_remove has been called or not.
32861 (pass_sprintf_length::handle_gimple_call): Return bool whether
32862 try_substitute_return_value called gsi_remove. Formatting fix.
32863 (pass_sprintf_length::execute): Don't use gsi_remove if
32864 handle_gimple_call returned true.
32867 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
32868 be removed due to side-effects, don't remove following barrier nor
32869 turn the successor edge into fallthru edge.
32871 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32874 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
32875 element-reversing loads and stores as not swappable.
32877 2017-01-12 Nathan Sidwell <nathan@acm.org>
32878 Nicolai Stange <nicstange@gmail.com>
32880 * combine.c (try_combine): Don't ignore result of overlap checking
32881 loop. Combine overlap & asm check into single loop.
32883 2017-01-12 Richard Biener <rguenther@suse.de>
32885 * tree-pretty-print.c (dump_generic_node): Provide -gimple
32886 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
32888 2017-01-12 Richard Biener <rguenther@suse.de>
32890 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
32891 and TS_TARGET_OPTION directly derive from TS_BASE.
32892 * tree-core.h (tree_optimization_option): Derive from tree_base.
32893 (tree_target_option): Likewise.
32895 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
32897 * config/i386/i386.c (memory_address_length): Increase len
32898 only when rip_relative_addr_p returns false.
32900 2017-01-11 Julia Koval <julia.koval@intel.com>
32902 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
32903 (OPTION_MASK_ISA_SGX_SET): New.
32904 (ix86_handle_option): Handle OPT_msgx.
32905 * config.gcc: Added sgxintrin.h.
32906 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
32907 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
32908 * config/i386/i386.c (ix86_target_string): Add -msgx.
32910 (ix86_option_override_internal): Handle new options.
32911 (ix86_valid_target_attribute_inner_p): Add sgx.
32912 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
32913 * config/i386/i386.opt: Add msgx.
32914 * config/i386/sgxintrin.h: New file.
32915 * config/i386/x86intrin.h: Add sgxintrin.h.
32917 2017-01-11 Jakub Jelinek <jakub@redhat.com>
32920 * fold-const.c (maybe_nonzero_address): Return 1 for function
32922 (tree_single_nonzero_warnv_p): Don't handle function local objects
32926 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
32929 2017-01-11 David Malcolm <dmalcolm@redhat.com>
32932 * opts.c: Include "spellcheck.h"
32933 (struct string_fragment): New struct.
32934 (struct edit_distance_traits<const string_fragment &>): New
32936 (get_closest_sanitizer_option): New function.
32937 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
32939 2017-01-11 Jakub Jelinek <jakub@redhat.com>
32941 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
32943 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
32944 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
32945 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
32946 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
32947 for initial die_offset if dwarf_split_debug_info.
32948 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
32949 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
32951 (output_skeleton_debug_sections): Formatting fix. Use
32952 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
32953 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
32955 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
32957 * config/arm/cortex-a53.md: Add bypasses for
32958 cortex_a53_r2f_cvt.
32959 (cortex_a53_r2f): Only use for transfers.
32960 (cortex_a53_f2r): Likewise.
32961 (cortex_a53_r2f_cvt): Add reservation for conversions.
32962 (cortex_a53_f2r_cvt): Likewise.
32964 2017-01-11 Tamar Christina <tamar.christina@arm.com>
32966 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
32967 to all inlined functions, change static to extern.
32969 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
32972 * config/arm/arm.c (legitimize_pic_address): Handle reference to
32974 (arm_assemble_integer): Likewise.
32976 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
32978 * config.gcc: Use new awk script to check CPU, FPU and architecture
32979 parameters for --with-... options.
32980 * config/arm/parsecpu.awk: New file
32981 * config/arm/arm-cpus.in: New file.
32982 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
32984 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
32986 * config/arm/t-arm: Update dependency rules.
32987 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
32988 of processing .def files.
32989 * config/arm/genopt.sh: Deleted.
32990 * config/arm/gentune.sh: Deleted.
32991 * config/arm/arm-cores.def: Deleted.
32992 * config/arm/arm-arches.def: Deleted.
32993 * config/arm/arm-fpus.def: Deleted.
32994 * config/arm/arm-tune.md: Regenerated.
32995 * config/arm/arm-tables.opt: Regenerated.
32996 * config/arm/arm-cpu.h: New generated file.
32997 * config/arm/arm-cpu-data.h: New generated file.
32998 * config/arm/arm-cpu-cdata.h: New generated file.
33000 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
33003 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
33005 (input_varpool_node): Unpack dynamically_initialized bit.
33007 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
33009 PR rtl-optimization/79032
33010 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
33011 the alignment of the adjusted memory reference against that of MODE,
33012 instead of the alignment of the original memory reference.
33014 2017-01-11 Martin Jambor <mjambor@suse.cz>
33016 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
33018 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
33019 decorated functions.
33021 2017-01-11 Richard Biener <rguenther@suse.de>
33023 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
33024 set range/nonnull info for PHI results. Do not set it on
33025 stmts marked for removal.
33027 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
33029 * expr.c (store_field): In the bitfield case, fetch the return value
33030 from the registers before applying a single big-endian adjustment.
33031 Always do a final load for a BLKmode value not larger than a word.
33033 2017-01-10 David Malcolm <dmalcolm@redhat.com>
33036 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
33037 that we correctly handle column numbers greater than
33038 LINE_MAP_MAX_COLUMN_NUMBER.
33040 2017-01-10 Martin Sebor <msebor@redhat.com>
33042 PR middle-end/78245
33043 * gimple-ssa-sprintf.c (get_destination_size): Call
33044 {init,fini}object_sizes.
33045 * tree-object-size.c (addr_object_size): Adjust.
33046 (pass_through_call): Adjust.
33047 (pass_object_sizes::execute): Adjust.
33048 * tree-object-size.h (fini_object_sizes): Declare.
33050 2017-01-10 Martin Sebor <msebor@redhat.com>
33052 PR tree-optimization/78775
33053 * builtins.c (get_size_range): Move...
33054 * calls.c: ...to here.
33055 (alloc_max_size): Accept zero argument.
33056 (operand_signed_p): Remove.
33057 (maybe_warn_alloc_args_overflow): Call get_size_range.
33058 * calls.h (get_size_range): Declare.
33060 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
33062 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
33063 from TI's devices.csv file as of September 2016.
33064 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
33066 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
33068 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
33069 * doc/invoke.texi: Likewise.
33070 * doc/md.texi: Likewise.
33071 * doc/objc.texi: Likewise.
33073 2017-01-10 Joshua Conner <joshconner@google.com>
33075 * config/arm/fuchsia-elf.h: New file.
33076 * config/fuchsia.h: New file.
33077 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
33078 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
33080 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
33082 2016-01-10 Richard Biener <rguenther@suse.de>
33084 PR tree-optimization/79034
33085 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
33086 Propagate out degenerate PHIs in the joiner.
33088 2017-01-10 Martin Liska <mliska@suse.cz>
33090 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
33091 (sort_congruence_classes_by_decl_uid): Likewise.
33092 (sort_congruence_class_groups_by_decl_uid): Likewise.
33093 (sem_item_optimizer::merge_classes): Sort class, groups in these
33094 classes and members in the groups by DECL_UID of declarations.
33095 This would make merge operations stable.
33097 2017-01-10 Martin Liska <mliska@suse.cz>
33099 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
33100 usage of m_classes_vec.
33101 (sem_item_optimizer::~sem_item_optimizer): Likewise.
33102 (sem_item_optimizer::get_group_by_hash): Likewise.
33103 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
33104 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
33105 (sem_item_optimizer::verify_classes): Likewise.
33106 (sem_item_optimizer::process_cong_reduction): Likewise.
33107 (sem_item_optimizer::dump_cong_classes): Likewise.
33108 (sem_item_optimizer::merge_classes): Likewise.
33109 * ipa-icf.h (congruence_class_hash): Rename from
33110 congruence_class_group_hash. Remove declaration of m_classes_vec.
33112 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
33114 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
33115 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
33116 * config.gcc: Add avx512vpopcntdqintrin.h.
33117 * config/i386/avx512vpopcntdqintrin.h: New.
33118 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
33119 * config/i386/i386-builtin-types.def: Add new types.
33120 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
33121 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
33122 __builtin_ia32_vpopcountq_v8di_mask): New.
33123 * config/i386/i386-c.c (ix86_target_macros_internal): Define
33124 __AVX512VPOPCNTDQ__.
33125 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
33126 (PTA_AVX512VPOPCNTDQ): Define.
33127 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
33128 TARGET_AVX512VPOPCNTDQ_P): Define.
33129 * config/i386/i386.opt: Add mavx512vpopcntdq.
33130 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
33131 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
33133 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33135 PR middle-end/77484
33136 * predict.def (PRED_CALL): Set to 67.
33138 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
33140 * expr.c (store_field): In the bitfield case, if the value comes from
33141 a function call and is of an aggregate type returned in registers, do
33142 not modify the field mode; extract the value in all cases if the mode
33143 is BLKmode and the size is not larger than a word.
33145 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
33148 * config/i386/cpuid.h: Fix undefined behavior.
33150 2017-01-04 Jeff Law <law@redhat.com>
33152 PR tree-optimization/79007
33153 PR tree-optimization/67955
33154 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
33155 conservative for pt.null when flag_non_call_exceptions is on.
33157 2017-01-09 Jakub Jelinek <jakub@redhat.com>
33159 PR translation/79019
33160 PR translation/79020
33161 * params.def (PARAM_INLINE_MIN_SPEEDUP,
33162 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
33163 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
33165 * config/avr/avr.opt (maccumulate-args): Likewise.
33166 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
33167 * common.opt (freport-bug): Likewise.
33168 * cif-code.def (CIF_FINAL_ERROR): Likewise.
33169 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
33170 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
33171 translatable string.
33172 * config/i386/i386.c (function_value_32): Likewise.
33173 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
33174 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
33176 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
33177 * common/config/msp430/msp430-common.c (msp430_handle_option):
33179 * symtab.c (symtab_node::verify_base): Likewise.
33180 * opts.c (set_debug_level): Likewise.
33181 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
33182 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
33183 missing whitespace to translatable strings.
33184 * config/avr/avr.md (bswapsi2): Fix typo in comment.
33185 * config/sh/superh.h: Likewise.
33186 * config/i386/xopintrin.h: Likewise.
33187 * config/i386/znver1.md: Likewise.
33188 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
33189 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
33190 * double-int.h (struct double_int): Likewise.
33191 * double-int.c (div_and_round_double): Likewise.
33192 * wide-int.cc: Likewise.
33193 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
33194 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
33195 * cfgcleanup.c (crossjumps_occured): Renamed to ...
33196 (crossjumps_occurred): ... this.
33197 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
33200 PR tree-optimization/78899
33201 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
33202 returning bool return struct loop *, NULL for failure and the new
33204 (versionable_outer_loop_p): Don't version outer loop if it has
33205 dont_vectorized bit set.
33206 (tree_if_conversion): When versioning outer loop, ensure
33207 tree_if_conversion is performed also on the inner loop of the
33208 non-vectorizable outer loop copy.
33209 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
33210 LOOP_VECTORIZED in inner loop of the scalar outer loop and
33211 prevent vectorization of it.
33212 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
33213 the outer loop vectorization of the non-scalar version is attempted
33214 before vectorization of the inner loop in scalar version. If
33215 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
33216 vectorization of its inner loop.
33217 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
33218 has 2 inner loops, rename also on edges from bb whose single pred
33219 is outer_loop->header. Fix typo in function comment.
33221 2017-01-09 Martin Sebor <msebor@redhat.com>
33224 * asan.c (asan_emit_stack_protection): Increase local buffer size
33225 to avoid snprintf truncation warning.
33227 2017-01-09 Andrew Pinski <apinski@cavium.com>
33229 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
33230 to reference thunderx2t99 for the tuning structure
33231 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
33233 (thunderx2t99_extra_costs): This.
33234 * config/aarch64/aarch64-tune.md: Regenerate.
33235 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
33236 (vulcan_addrcost_table): This.
33237 (vulcan_regmove_cost): Rename to ...
33238 (thunderx2t99_regmove_cost): This.
33239 (vulcan_vector_cost): Rename to ...
33240 (thunderx2t99_vector_cost): this.
33241 (vulcan_branch_cost): Rename to ...
33242 (thunderx2t99_branch_cost): This.
33243 (vulcan_tunings): Rename to ...
33244 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
33245 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
33247 2017-01-09 Martin Jambor <mjambor@suse.cz>
33251 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
33252 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
33253 (propagate_vr_accross_jump_function): Use the above function for all
33254 value range computations for pass-through jump functions and type
33255 converasion from explicit value range values.
33256 (ipcp_propagate_stage): Do not attempt to deduce types of formal
33257 parameters from TYPE_ARG_TYPES.
33258 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
33259 (ipa_write_node_info): Stream type of the actual argument.
33260 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
33262 2017-01-09 Martin Liska <mliska@suse.cz>
33265 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
33266 (lookup_compiler): Do not show error message with have_E.
33268 2017-01-09 Jakub Jelinek <jakub@redhat.com>
33270 PR tree-optimization/78938
33271 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
33272 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
33273 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
33274 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
33277 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33279 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
33282 2017-01-09 Richard Biener <rguenther@suse.de>
33284 PR tree-optimization/78997
33285 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
33286 name condition properly.
33288 2017-01-09 Richard Biener <rguenther@suse.de>
33291 * dwarf2out.c (is_cxx): New overload with context.
33292 (is_naming_typedef_decl): Use it.
33294 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33296 * invoke.texi (Option Summary): Correct spacing in option lists
33297 and add line breaks to fix over-long lines.
33299 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33301 PR middle-end/17660
33303 * extend.texi (Common Variable Attributes): Add xref to GCC
33304 Internals manual to explain mode attribute keywords.
33306 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33309 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
33310 and Preprocessor Options.
33311 (Options for Linking): Document -pthread here....
33312 (RS/6000 and PowerPC Options): ...not here.
33313 (Solaris 2 Options): ...or here.
33314 * doc/cppopts.texi: Document -pthread.
33316 2017-01-08 Martin Sebor <msebor@redhat.com>
33318 PR middle-end/77708
33319 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
33320 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
33321 New member functions.
33322 (format_directive): Used them.
33324 (pass_sprintf_length::handle_gimple_call): Same.
33325 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
33326 to avoid truncation for any argument.
33327 (extract_affine_mul): Same.
33328 * tree.c (get_file_function_name): Same.
33330 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33332 PR middle-end/77484
33333 * predict.def (PRED_INDIR_CALL): Set to 86.
33335 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33337 PR preprocessor/54124
33338 * doc/cppopts.texi: Reformat -d subtable to list the full name
33339 of the options. Add cross-reference to the docs for the general
33340 compiler -d options.
33341 * doc/invoke.texi (Developer Options): Add cross-reference to the
33342 preprocessor-specific -d option documentation.
33344 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33346 PR preprocessor/13498
33347 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
33348 redudant material, and reflect new command-line options.
33349 (System Headers): Likewise.
33351 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33353 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
33354 -isystem, and -idirafter. Copy-edit.
33355 * doc/cppopts.texi: Copy-edit. Remove contradiction about
33356 default for -ftrack-macro-expansion. Delete obsolete and
33357 badly-formatted implementation details about -fdebug-cpp output.
33358 * doc/cppwarnopts.texi: Copy-edit.
33360 2017-01-07 David Malcolm <dmalcolm@redhat.com>
33363 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
33364 that the transition from a max line width >= 1<<10 to narrower
33365 lines works correctly.
33367 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
33369 * doc/options.texi (PerFunction): New.
33370 * opt-functions.awk (switch_flags): Map both Optimization and
33371 PerFunction to CL_OPTIMIZATION.
33372 * opth-gen.awk: Test for PerFunction flag along with
33374 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
33375 it only when the latter is present. Skip those that don't in
33376 the hash function generator.
33377 * common.opt (fvar-tracking): Mark as PerFunction instead of
33379 (fvar-tracking-assignments): Likewise.
33380 (fvar-tracking-assignments-toggle): Likewise.
33381 (fvar-tracking-uninit): Likewise.
33383 2017-01-07 Jakub Jelinek <jakub@redhat.com>
33385 PR translation/79018
33386 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
33389 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
33392 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
33393 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
33394 TARGET_LONG_JUMP_TABLE_OFFSETS.
33395 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
33396 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
33397 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
33398 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
33399 * config/m68k/m68k.md (tablejump expander): Likewise.
33400 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
33401 TARGET_LONG_JUMP_TABLE_OFFSETS.
33402 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
33403 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
33405 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
33406 David Holsgrove <david.holsgrove@xilinx.com>
33408 * common/config/microblaze/microblaze-common.c
33409 (TARGET_EXCEPT_UNWIND_INFO): Remove.
33410 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
33412 * config/microblaze/microblaze.c (microblaze_must_save_register)
33413 (microblaze_expand_epilogue, microblaze_return_addr): Handle
33415 (microblaze_eh_return): New function.
33416 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
33417 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
33418 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
33419 * config/microblaze/microblaze.md (eh_return): New pattern.
33421 2017-01-06 Jakub Jelinek <jakub@redhat.com>
33423 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
33424 GCC_DIAGNOSTIC_STRINGIFY): Define.
33426 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
33428 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33430 * config/arm/arm.md (<mcrr>): New.
33432 * config/arm/arm.c (arm_arch5te): New.
33433 (arm_option_override): Set arm_arch5te.
33434 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
33436 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
33437 (arm_mcrr_qualifiers): ... this. New.
33438 (MRRC_QUALIFIERS): Define to...
33439 (arm_mrrc_qualifiers): ... this. New.
33440 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
33442 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
33443 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
33444 (MRRCI, mrrc, MRRC): New.
33445 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
33446 VUNSPEC_MRRC2): New.
33448 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33450 * config/arm/arm.md (<mcr>): New.
33452 * config/arm/arm.c (arm_coproc_builtin_available): Add
33453 support for mcr, mrc, mcr2 and mrc2.
33454 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
33455 (arm_mcr_qualifiers): ... this. New.
33456 (MRC_QUALIFIERS): Define to ...
33457 (arm_mrc_qualifiers): ... this. New.
33458 (MCR_QUALIFIERS): Define to ...
33459 (arm_mcr_qualifiers): ... this. New.
33460 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
33462 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
33463 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
33464 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
33465 VUNSPEC_MRC2): New.
33467 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33469 * config/arm/arm.md (*ldc): New.
33473 * config/arm/arm.c (arm_coproc_builtin_available): Add
33474 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
33475 (arm_coproc_ldc_stc_legitimate_address): New.
33476 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
33477 'qualifier_const_pointer'.
33478 (LDC_QUALIFIERS): Define to...
33479 (arm_ldc_qualifiers): ... this. New.
33480 (STC_QUALIFIERS): Define to...
33481 (arm_stc_qualifiers): ... this. New.
33482 * config/arm/arm-protos.h
33483 (arm_coproc_ldc_stc_legitimate_address): New.
33484 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
33485 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
33486 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
33487 stc2, stcl, stc2l): New.
33488 * config/arm/constraints.md (Uz): New.
33489 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
33490 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
33491 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
33492 VUNSPEC_STC2L): New.
33494 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33496 * config/arm/arm.md (<cdp>): New.
33497 * config/arm/arm.c (neon_const_bounds): Rename this ...
33498 (arm_const_bounds): ... this.
33499 (arm_coproc_builtin_available): New.
33500 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
33501 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
33502 (CDP_QUALIFIERS): Define to...
33503 (arm_cdp_qualifiers): ... this. New.
33505 (arm_expand_builtin_args): Add case for 6 arguments.
33506 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
33507 (arm_const_bounds): ... this.
33508 (arm_coproc_builtin_available): New.
33509 * config/arm/arm_acle.h (__arm_cdp): New.
33511 * config/arm/arm_acle_builtins.def (cdp): New.
33513 * config/arm/iterators.md (CDPI,CDP,cdp): New.
33514 * config/arm/neon.md: Rename all 'neon_const_bounds' to
33515 'arm_const_bounds'.
33516 * config/arm/types.md (coproc): New.
33517 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
33518 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
33519 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
33520 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
33522 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33524 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
33525 (UBINOP_QUALIFIERS): New.
33527 (acle_builtin_data): New. Change comment.
33528 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
33529 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
33530 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
33531 arm_acle_builtins.def.
33532 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
33533 (arm_init_acle_builtins): New.
33534 (CRC32_BUILTIN): Remove.
33535 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
33536 crc32cb, crc32ch and crc32cw.
33537 (arm_init_crc32_builtins): Remove.
33538 (arm_init_builtins): Use arm_init_acle_builtins rather
33539 than arm_init_crc32_builtins.
33540 (arm_expand_acle_builtin): New.
33541 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
33542 * config/arm/arm_acle_builtins.def: New.
33544 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33546 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
33547 (arm_builtin_datum): ... this.
33548 (arm_init_neon_builtin): Rename to ...
33549 (arm_init_builtin): ... this. Add a new parameters PREFIX
33550 and USE_SIG_IN_NAME.
33551 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
33552 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
33553 'arm_builtin_datum'.
33554 (arm_init_vfp_builtins): Likewise.
33555 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
33556 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
33557 (arm_expand_neon_args): Rename to ...
33558 (arm_expand_builtin_args): ... this. Rename builtin_arg
33559 enum values and differentiate between ARG_BUILTIN_MEMORY
33560 and ARG_BUILTIN_NEON_MEMORY.
33561 (arm_expand_neon_builtin_1): Rename to ...
33562 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
33563 values, arm_expand_builtin_args and add bool parameter NEON.
33564 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
33565 (arm_expand_vfp_builtin): Likewise.
33566 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
33568 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33570 PR middle-end/77484
33571 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
33572 * predict.c (tree_estimate_probability_bb): Reverse direction of
33573 polymorphic call predictor.
33575 2017-01-06 David Malcolm <dmalcolm@redhat.com>
33577 * passes.c (execute_one_pass): Split out pass-skipping logic into...
33578 (determine_pass_name_match): ...this new function and...
33579 (should_skip_pass_p): ...this new function.
33581 2017-01-06 Nathan Sidwell <nathan@acm.org>
33583 * ipa-visibility.c (function_and_variable_visibility): Reformat
33584 comments and long lines. Remove extrneous if.
33585 * symtab.c (symtab_node::make_decl_local): Fix code format.
33586 (symtab_node::set_section_for_node): Fix comment typo.
33588 2017-01-06 Martin Liska <mliska@suse.cz>
33591 * lra-constraints.c: Rename invariant to lra_invariant.
33592 * predict.c (set_even_probabilities): Initialize e to NULL.
33594 2017-01-05 Martin Sebor <msebor@redhat.com>
33596 PR tree-optimization/78910
33597 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
33598 (format_integer): Correct off-by-one error in the handling
33599 of precision with negative numbers in signed conversions..
33601 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
33603 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
33605 2017-01-05 Jakub Jelinek <jakub@redhat.com>
33607 PR tree-optimization/71016
33608 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
33609 factor_out_conditional_conversion. Formatting fix.
33610 (factor_out_conditional_conversion): Add cond_stmt argument.
33611 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
33612 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
33615 2017-01-05 David Malcolm <dmalcolm@redhat.com>
33617 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
33618 read-rtl-function.o, and selftest-rtl.o.
33619 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
33620 (selftest::aarch64_test_loading_full_dump): New function.
33621 (selftest::aarch64_run_selftests): New function.
33622 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
33623 selftest::aarch64_run_selftests.
33624 * config/i386/i386.c
33625 (selftest::ix86_test_loading_dump_fragment_1): New function.
33626 (selftest::ix86_test_loading_call_insn): New function.
33627 (selftest::ix86_test_loading_full_dump): New function.
33628 (selftest::ix86_test_loading_unspec): New function.
33629 (selftest::ix86_run_selftests): Call the new functions.
33630 * emit-rtl.c (maybe_set_max_label_num): New function.
33631 * emit-rtl.h (maybe_set_max_label_num): New decl.
33632 * function.c (instantiate_decls): Guard call to
33633 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
33634 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
33636 * gensupport.c (gen_reader::gen_reader): Pass "false"
33637 for new "compact" param of rtx_reader.
33638 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
33639 rather than an empty string for NULL strings.
33640 * read-md.c: Potentially include config.h rather than bconfig.h.
33641 Wrap include of errors.h with #ifdef GENERATOR_FILE.
33642 (have_error): New global, copied from errors.c.
33643 (md_reader::read_name): Rename to...
33644 (md_reader::read_name_1): ...this, adding "out_loc" param,
33645 and converting "missing name or number" to returning false, rather
33647 (md_reader::read_name): Reimplement in terms of read_name_1.
33648 (md_reader::read_name_or_nil): New function.
33649 (md_reader::read_string): Handle "(nil)" by returning NULL.
33650 (md_reader::md_reader): Add new param "compact".
33651 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
33652 (md_reader::read_file): New method.
33653 * read-md.h (md_reader::md_reader): Add new param "compact".
33654 (md_reader::read_file): New method.
33655 (md_reader::is_compact): New accessor.
33656 (md_reader::read_name): Convert return type from void to file_location.
33657 (md_reader::read_name_or_nil): New decl.
33658 (md_reader::read_name_1): New decl.
33659 (md_reader::m_compact): New field.
33660 (noop_reader::noop_reader): Pass "false" for new "compact" param
33662 (rtx_reader::rtx_reader): Add new "compact" param.
33663 (rtx_reader::read_rtx_operand): Make virtual and convert return
33664 type from void to rtx.
33665 (rtx_reader::read_until): New decl.
33666 (rtx_reader::handle_any_trailing_information): New virtual function.
33667 (rtx_reader::postprocess): New virtual function.
33668 (rtx_reader::finalize_string): New virtual function.
33669 (rtx_reader::m_in_call_function_usage): New field.
33670 (rtx_reader::m_reuse_rtx_by_id): New field.
33671 * read-rtl-function.c: New file.
33672 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
33673 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
33674 (selftest::verify_three_block_rtl_cfg): New decl.
33675 * read-rtl-function.h: New file.
33676 * read-rtl.c: Potentially include config.h rather than bconfig.h.
33677 For host, include function.h, memmodel.h, and emit-rtl.h.
33678 (one_time_initialization): New function.
33679 (struct compact_insn_name): New struct.
33680 (compact_insn_names): New array.
33681 (find_code): Handle insn codes in compact dumps.
33682 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
33683 (bind_subst_iter_and_attr): Likewise.
33684 (add_condition_to_string): Likewise.
33685 (add_condition_to_rtx): Likewise.
33686 (apply_attribute_uses): Likewise.
33687 (add_current_iterators): Likewise.
33688 (apply_iterators): Likewise.
33689 (initialize_iterators): Guard usage of apply_subst_iterator with
33690 #ifdef GENERATOR_FILE.
33691 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
33692 (md_reader::read_mapping): Likewise.
33693 (add_define_attr_for_define_subst): Likewise.
33694 (add_define_subst_attr): Likewise.
33695 (read_subst_mapping): Likewise.
33696 (check_code_iterator): Likewise.
33697 (rtx_reader::read_rtx): Likewise. Move one-time initialization
33699 (one_time_initialization): New function.
33700 (rtx_reader::read_until): New method.
33701 (read_flags): New function.
33702 (parse_reg_note_name): New function.
33703 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
33704 Handle reuse_rtx ids.
33705 Wrap iterator lookup within #ifdef GENERATOR_FILE.
33706 Add parsing support for RTL dumps, mirroring the special-cases in
33707 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
33708 values, and calling handle_any_trailing_information.
33709 (rtx_reader::read_rtx_operand): Convert return type from void
33710 to rtx, returning return_rtx. Handle case 'e'. Call
33711 finalize_string on XSTR and XTMPL fields.
33712 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
33713 "(nil)" values were omitted. Call the postprocess vfunc on the
33715 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
33716 class ctor. Initialize m_in_call_function_usage. Call
33717 one_time_initialization.
33718 * rtl-tests.c (selftest::test_uncond_jump): Call
33719 set_new_first_and_last_insn.
33720 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
33721 * selftest-rtl.c: New file.
33722 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
33723 (selftest::get_insn_by_uid): New decl.
33724 * selftest-run-tests.c (selftest::run_tests): Call
33725 read_rtl_function_c_tests.
33726 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
33727 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
33730 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
33732 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
33733 operands in a special way. Assert that pos+len <= mode precision.
33735 2017-01-05 Jakub Jelinek <jakub@redhat.com>
33737 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
33738 3 argument Alias with unlimited for the negative form.
33739 (fno-vect-cost-model): Removed.
33741 2017-01-05 Martin Liska <mliska@suse.cz>
33743 * hsa-gen.c (gen_hsa_divmod): New function.
33744 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
33746 2017-01-05 Martin Liska <mliska@suse.cz>
33749 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
33752 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33754 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
33755 small constant length operands.
33757 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33759 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
33760 between loop iterations.
33762 2017-01-05 Martin Liska <mliska@suse.cz>
33765 * gimplify.c (gimplify_decl_expr): Compare to
33766 asan_poisoned_variables instread of checking flags.
33767 (gimplify_target_expr): Likewise.
33768 (gimplify_expr): Likewise.
33769 (gimplify_function_tree): Conditionally initialize
33770 asan_poisoned_variables.
33772 2017-01-04 Jeff Law <law@redhat.com>
33774 PR tree-optimizatin/78812
33775 * rtl.h (contains_mem_rtx_p): Prototype.
33776 * ifcvt.c (containts_mem_rtx_p): Move from here to...
33777 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
33778 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
33779 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
33780 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
33782 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33784 * input.c (assert_char_at_range): Default-initialize actual_range.
33786 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33788 * df-scan.c (df_ref_create_structure): Make regno unsigned,
33789 to match the caller.
33791 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33793 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
33794 insns after final jump in test to emit dummy move.
33796 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33798 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
33799 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
33801 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33803 * multiple_target.c (create_dispatcher_calls): Init e_next.
33804 * tree-ssa-loop-split.c (split_loop): Init border.
33805 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
33808 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
33813 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
33814 (altivec_register_operand): Do not return true if the operand
33815 contains a SUBREG mixing SImode and SFmode.
33816 (vsx_register_operand): Likewise.
33817 (vsx_reg_sfsubreg_ok): New predicate.
33818 (vfloat_operand): Do not return true if the operand contains a
33819 SUBREG mixing SImode and SFmode.
33820 (vint_operand): Likewise.
33821 (vlogical_operand): Likewise.
33822 (gpc_reg_operand): Likewise.
33823 (int_reg_operand): Likewise.
33824 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
33825 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
33826 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
33828 (rs6000_emit_move_si_sf_subreg): New helper function.
33829 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
33830 fixup SUBREGs involving SImode and SFmode.
33831 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
33832 numbers for the new peephole2 optimization.
33833 (peephole2 for SFmode unions): New peephole2 to optimize cases in
33834 the GLIBC math library that do AND/IOR/XOR operations on single
33835 precision floating point.
33836 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
33837 target macros to say whether we need to avoid SUBREGs mixing
33839 (TARGET_ALLOW_SF_SUBREG): Likewise.
33840 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
33841 (UNSPEC_SI_FROM_SF): Likewise.
33842 (iorxor): Change spacing.
33843 (and_ior_xor): New iterator for AND, IOR, and XOR.
33844 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
33845 (movdi_from_sf_zero_ext): Likewise.
33846 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
33847 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
33848 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
33849 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
33850 (fms<mode>4): Likewise.
33851 (fnma<mode>4): Likewise.
33852 (fnms<mode>4): Likewise.
33853 (nfma<mode>4): Likewise.
33854 (nfms<mode>4): Likewise.
33856 2017-01-04 Marek Polacek <polacek@redhat.com>
33859 * doc/invoke.texi: Document -Wpointer-compare.
33861 2017-01-04 Jakub Jelinek <jakub@redhat.com>
33863 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
33866 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
33867 descriptions for -gdwarf-5 and emit them as uleb128 instead of
33870 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
33873 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
33874 documentation of the powerpc_popcntb_ok attribute.
33875 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
33876 code to issue warning messages if a requested CPU configuration is
33877 not supported by the binary (assembler and loader) toolchain.
33878 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
33879 made to define a built-in function that has been disabled.
33880 (paired_init_builtins): Add assertion to prevent ICE if attempt is
33881 made to define a built-in function that has been disabled.
33882 (altivec_init_builtins): Add comment explaining why definition
33883 of the DST built-in functions is not preceded by an assertion
33884 check. Add assertions to prevent ICE if attempts are made to
33885 define an altivec predicate or an abs* built-in function that has
33887 (htm_init_builtins): Add comment explaining why definition of the
33888 htm built-in functions is not preceded by an assertion check.
33890 2017-01-04 Jeff Law <law@redhat.com>
33892 PR tree-optimizatin/67955
33893 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
33894 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
33895 the points-to solution does not include pt_null. Use DECL_PT_UID
33898 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
33900 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
33901 Use gen_int_mode instead of gen_lopwart for const_int operands.
33903 2017-01-04 Jakub Jelinek <jakub@redhat.com>
33905 PR tree-optimization/71563
33906 * match.pd: Simplify X << Y into X if Y is known to be 0 or
33907 out of range value - has low bits known to be zero.
33909 2017-01-04 Alan Modra <amodra@gmail.com>
33911 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
33912 * configure: Regenerate.
33913 * config.in: Regenerate.
33915 2017-01-04 Jakub Jelinek <jakub@redhat.com>
33918 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
33919 a substring of the message, but strcmp with the whole message. Ifdef
33920 ENABLE_NLS, translate the message first using dgettext.
33922 2017-01-03 Jeff Law <law@redhat.com>
33924 PR tree-optimizatin/78856
33925 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
33926 (mark_threaded_blocks): Remove code to truncate thread paths that
33927 cross multiple loop headers. Instead invalidate the cached loop
33928 iteration information and handle case of a thread path walking
33929 into an irreducible region.
33931 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
33934 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
33935 assertions. Add support for doing the signbit if the IEEE 128-bit
33936 floating point value is in a GPR.
33937 * config/rs6000/rs6000.md (Fsignbit): Delete.
33938 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
33939 Update the length attribute if the value is in a GPR.
33940 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
33941 the sign or zero extension instruction, since the value is always 0/1.
33942 (signbit<mode>2_dm2): Delete using <Fsignbit>.
33945 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
33946 extracting SImode to a GPR register so that we can generate a
33947 store, limit the vector to be in a traditional Altivec register
33948 for the vextuwrx instruction.
33950 2017-01-03 Ian Lance Taylor <iant@google.com>
33952 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
33954 2017-01-03 Martin Sebor <msebor@redhat.com>
33956 PR tree-optimization/78696
33957 * gimple-ssa-sprintf.c (format_floating): Correct handling of
33958 precision. Use MPFR for %f for greater fidelity. Correct handling
33960 (pass_sprintf_length::compute_format_length): Set width and precision
33961 specified by asrerisk to void_node for vararg functions.
33962 (try_substitute_return_value): Adjust dump output.
33964 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
33966 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
33968 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
33970 * doc/invoke.texi (SPARC options): Document -mlra as the default.
33971 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
33972 -mlra/-mno-lra was passed to the compiler.
33974 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
33976 PR rtl-optimization/65618
33977 * emit-rtl.c (try_split): Move initialization of "before" and
33978 "after" to just before the call to emit_insn_after_setloc.
33980 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
33982 * doc/md.texi (Standard Names): Remove reference to Java frontend.
33984 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
33986 * dwarf2out.c (gen_enumeration_type_die): When
33987 -gno-strict-dwarf, add a DW_AT_encoding attribute.
33989 2017-01-03 Jakub Jelinek <jakub@redhat.com>
33991 PR tree-optimization/78965
33992 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
33993 Change first argument from const call_info & to call_info &. For %n
33994 set info.nowrite to false.
33996 PR middle-end/78901
33997 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
33998 possibly throwing calls.
34000 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
34001 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
34002 and fns handling, rather than in a separate case SSA_NAME.
34004 2017-01-02 Jeff Law <law@redhat.com>
34006 * config/darwin-driver.c (darwin_driver_init): Const-correctness
34007 fixes for first_period and second_period variables.
34009 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
34012 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
34013 (*insvqi_1): New insn pattern.
34014 (*insvqi_1_mem_rex64): Ditto.
34015 (*insvqi_2): Ditto.
34016 (*insvqi_3): Rename from *insvqi.
34018 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
34020 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
34022 * doc/cfg.texi (Edges): Remove reference to Java.
34023 (Maintaining the CFG): Ditto.
34025 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34027 PR middle-end/77674
34028 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
34029 transparent aliases.
34031 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34033 PR middle-end/77484
34034 * predict.def (PRED_CALL): Update hitrate.
34035 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
34036 * predict.c (tree_estimate_probability_bb): Split CALL predictor
34037 into direct/indirect/polymorphic variants.
34039 2017-01-01 Jakub Jelinek <jakub@redhat.com>
34041 Update copyright years.
34043 * gcc.c (process_command): Update copyright notice dates.
34044 * gcov-dump.c (print_version): Ditto.
34045 * gcov.c (print_version): Ditto.
34046 * gcov-tool.c (print_version): Ditto.
34047 * gengtype.c (create_file): Ditto.
34048 * doc/cpp.texi: Bump @copying's copyright year.
34049 * doc/cppinternals.texi: Ditto.
34050 * doc/gcc.texi: Ditto.
34051 * doc/gccint.texi: Ditto.
34052 * doc/gcov.texi: Ditto.
34053 * doc/install.texi: Ditto.
34054 * doc/invoke.texi: Ditto.
34056 Copyright (C) 2017 Free Software Foundation, Inc.
34058 Copying and distribution of this file, with or without modification,
34059 are permitted in any medium without royalty provided the copyright
34060 notice and this notice are preserved.