1 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
3 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
5 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
6 config/riscv/riscv.c (predict.h): New include.
7 (riscv_slow_unaligned_access_p): No longer static.
8 (riscv_block_move_straight): Add require.
9 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
11 2017-11-08 Jakub Jelinek <jakub@redhat.com>
14 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
15 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
16 nonimmediate_operand predicate for operand 1 instead of
19 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
23 * config/aarch64/constraints.md (Uml): New constraint.
24 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
27 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
30 of two vec_duplicates into a vec_concat.
32 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
35 Simplify vec_merge of vec_duplicate and vec_concat.
36 * config/aarch64/constraints.md (Utq): New constraint.
37 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
40 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
42 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
43 Simplify vec_merge of vec_duplicate and const_vector.
44 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
46 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
47 mode iterator. Update predicate on operand 1 to
48 handle non-const_vec constants. Delete constraints.
49 (*aarch64_combinez_be<mode>): Likewise for operand 2.
51 2017-11-08 Jakub Jelinek <jakub@redhat.com>
53 PR tree-optimization/78821
54 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
56 (store_operand_info::store_operand_info): Initialize it to false.
57 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
58 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
59 store in the group, and if chain_info is non-NULL, to ignore altogether
61 (compatible_load_p): Fail if bit_not_p does not match.
62 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
63 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
64 (pass_store_merging::process_store): Adjust
65 terminate_all_aliasing_chains calls to pass NULL in all current spots,
66 call terminate_all_aliasing_chains newly when adding a store into
67 a chain with non-NULL chain_info.
69 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
71 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
73 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
75 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
77 (aarch64_layout_frame): Initialise emit_frame_chain.
78 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
79 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
81 2017-11-08 Martin Liska <mliska@suse.cz>
83 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
84 of gimple_call_internal_p.
86 2017-11-07 Tom de Vries <tom@codesourcery.com>
88 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
90 2017-11-07 Tom de Vries <tom@codesourcery.com>
92 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
95 2017-11-08 Martin Liska <mliska@suse.cz>
98 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
100 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
102 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
104 2017-11-07 Jakub Jelinek <jakub@redhat.com>
107 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
108 (*cmp<mode>_ccz_1): New insn with $k alternative.
111 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
112 RTX_COMM_COMPARE as commutative as well.
113 (ix86_binary_operator_ok): Formatting fix.
114 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
115 *<code><mode>3<mask_name><round_saeonly_name>,
116 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
117 *<s>mul<mode>3_highpart<mask_name>,
118 *vec_widen_umult_even_v16si<mask_name>,
119 *vec_widen_umult_even_v8si<mask_name>,
120 *vec_widen_umult_even_v4si<mask_name>,
121 *vec_widen_smult_even_v16si<mask_name>,
122 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
123 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
124 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
125 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
126 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
127 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
128 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
129 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
130 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
131 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
132 ix86_binary_operator_ok. Formatting fixes.
133 (*<plusminus_insn><mode>3<mask_name><round_name>,
134 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
137 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
139 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
140 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
141 eq and ne if TARGET_ISEL.
142 (cmp): New code_iterator.
143 (UNS, UNSU_, UNSIK): New code_attrs.
144 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
145 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
147 ("eq<mode>3"): ... this.
148 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
150 ("ne<mode>3"): ... this.
152 2017-11-07 Julia Koval <julia.koval@intel.com>
155 * common/config/i386/i386-common.c
156 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
157 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
158 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
159 * config/i386/i386.opt: Ditto.
160 * config/i386/i386.c (ix86_target_string): Ditto.
161 (ix86_option_override_internal): Ditto.
162 (ix86_init_mpx_builtins): Move MPX to args2.
163 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
164 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
165 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
166 __builtin_ia32_vgf2p8affineinvqb_v32qi,
167 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
168 __builtin_ia32_vgf2p8affineinvqb_v16qi,
169 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
171 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
174 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
175 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
176 (zero-extendsidi peephole2): Remove peephole.
178 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
181 * stor-layout.c: Include attribs.h.
182 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
183 explicit lookup of "aligned" attribute.
185 2017-11-07 Andrew Waterman <andrew@sifive.com>
187 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
188 (riscv_expand_block_move): Likewise.
189 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
191 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
192 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
193 * config/riscv/riscv.c (riscv_block_move_straight): New function.
194 (riscv_adjust_block_mem): Likewise.
195 (riscv_block_move_loop): Likewise.
196 (riscv_expand_block_move): Likewise.
197 gcc/config/riscv/riscv.md (movmemsi): New pattern.
199 2017-11-07 Michael Clark <michaeljclark@mac.com>
201 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
202 (MUSL_DYNAMIC_LINKER): Likewise.
204 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
206 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
207 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
210 2017-11-07 Richard Biener <rguenther@suse.de>
212 * match.pd: Fix build.
214 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
215 Jackson Woodruff <jackson.woodruff@arm.com>
217 PR tree-optimization/71026
218 * match.pd: Canonicalize negate in division.
220 2017-11-07 Sudakshina Das <sudi.das@arm.com>
223 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
225 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
227 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
228 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
230 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
232 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
233 non-scalar integral types.
234 * match.pd (negate_expr_p): Handle MINUS_EXPR.
235 (-(A-B), -(~A)): New transformations.
237 2017-11-07 Tom de Vries <tom@codesourcery.com>
239 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
240 semicolon after "do {} while (0)".
241 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
242 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
243 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
244 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
245 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
247 2017-11-07 Tom de Vries <tom@codesourcery.com>
249 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
250 after "do {} while (0)".
251 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
252 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
253 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
254 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
255 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
257 2017-11-07 Tom de Vries <tom@codesourcery.com>
260 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
262 (arm_rtx_costs_internal): Add missing semicolon after
263 HANDLE_NARROW_SHIFT_ARITH call.
265 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
267 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
268 disable isel if it was not set explicitly.
270 2017-11-06 James Bowman <james.bowman@ftdichip.com>
272 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
274 (add_type_attribute) likewise.
276 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
278 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
279 of struct ix86_frame.
280 (ix86_initial_elimination_offset): Likewise.
281 (ix86_expand_split_stack_prologue): Likewise.
283 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
285 * tree-vrp.h (enum value_range_type): Update stale comment.
287 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
288 Alan Hayward <alan.hayward@arm.com>
289 David Sherwood <david.sherwood@arm.com>
291 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
292 (aarch64_expand_vec_perm_const): Take the number of units too.
293 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
294 (aarch64_expand_vec_perm_const): Likewise.
295 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
296 (vec_perm<mode>): Update accordingly.
298 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
299 Alan Hayward <alan.hayward@arm.com>
300 David Sherwood <david.sherwood@arm.com>
302 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
303 Take the number of units too.
304 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
305 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
306 but check for a vector mode before rather than after the call.
307 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
308 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
309 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
310 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
311 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
312 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
313 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
314 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
315 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
316 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
317 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
318 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
319 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
320 (aarch64_sqdmull2_n<mode>): Update accordingly.
322 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
323 Alan Hayward <alan.hayward@arm.com>
324 David Sherwood <david.sherwood@arm.com>
326 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
327 the number of units too.
328 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
329 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
330 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
331 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
332 (vec_store_lanesxi<mode>): Update accordingly.
334 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
335 Alan Hayward <alan.hayward@arm.com>
336 David Sherwood <david.sherwood@arm.com>
338 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
339 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
340 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
341 of units rather than the mode.
342 * config/aarch64/iterators.md (nunits): New mode attribute.
343 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
344 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
345 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
346 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
347 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
348 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
349 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
350 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
351 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
352 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
353 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
354 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
355 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
356 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
357 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
358 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
359 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
360 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
361 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
362 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
363 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
364 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
365 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
366 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
367 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
368 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
369 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
370 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
371 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
372 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
373 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
374 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
375 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
376 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
377 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
378 (aarch64_simd_vec_setv2di): Likewise.
380 2017-11-06 Carl Love <cel@us.ibm.com>
382 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
384 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
385 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
387 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
389 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
390 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
391 Add power 8 macro expansions.
392 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
393 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
394 power 8 instructions. (VSX_XXBR): Add iterator.
396 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
398 * config/arm/arm.md (predicable_short_it): Change default to "no",
399 improve documentation, remove uses that are identical to the default.
400 (enabled_for_depr_it): Rename to enabled_for_short_it.
401 * gcc/config/arm/arm-fixed.md (predicable_short_it): Remove default uses.
402 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
403 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
404 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
405 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
407 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
410 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
411 float128 helper macros, which are no longer used after deleting
412 the old 'q' built-in functions, and moving the round to odd
413 built-in functions to being special built-in functions.
414 (BU_FLOAT128_2): Likewise.
415 (BU_FLOAT128_1_HW): Likewise.
416 (BU_FLOAT128_2_HW): Likewise.
417 (BU_FLOAT128_3_HW): Likewise.
418 (FABSQ): Delete old 'q' built-in functions.
419 (COPYSIGNQ): Likewise.
420 (SQRTF128_ODD): Move round to odd built-in functions to be
421 special built-in functions, so that we can handle
422 -mabi=ieeelongdouble.
423 (TRUNCF128_ODD): Likewise.
424 (ADDF128_ODD): Likewise.
425 (SUBF128_ODD): Likewise.
426 (MULF128_ODD): Likewise.
427 (DIVF128_ODD): Likewise.
428 (FMAF128_ODD): Likewise.
429 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
430 built-in names to 'f128'.
431 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
432 old 'q' built-in functions, as the machine independent code for
433 'f128' built-in functions handles this.
434 (rs6000_expand_builtin): Add expansion for float128 round to odd
435 functions, keying off on -mabi=ieeelongdouble of whether to use
436 the KFmode or TFmode variant.
437 (rs6000_init_builtins): Initialize the _Float128 round to odd
439 * doc/extend.texi (PowerPC Built-in Functions): Document the old
440 _Float128 'q' built-in functions are now mapped into the new
441 'f128' built-in functions.
443 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
445 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
446 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
448 2017-11-06 David Malcolm <dmalcolm@redhat.com>
451 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
452 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
453 * toplev.c: Include "ipa-fnsummary.h".
454 (toplev::finalize): Call ipa_fnsummary_c_finalize.
456 2017-11-06 Jakub Jelinek <jakub@redhat.com>
458 PR tree-optimization/82838
459 * gimple-ssa-store-merging.c
460 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
461 on a separate gimple_seq which is then appended to seq.
463 2017-11-06 Jeff Law <law@redhat.com>
466 * config/i386/i386.c (PROBE_INTERVAL): Remove.
467 (get_probe_interval): New functions.
468 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
469 (ix86_adjust_stack_and_probe): Likewise.
470 (output_adjust_stack_and_probe): Likewise.
471 (ix86_emit_probe_stack_range): Likewise.
472 (ix86_expand_prologue): Likewise.
474 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
476 PR tree-optimization/82816
477 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
478 if the modes of the two types are the same.
479 (convert_plusminus_to_widen): Likewise.
481 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
483 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
485 (usadv16qi): New define_expand.
486 (usadv8hi): New define_expand.
488 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
491 * ipa-inline-transform.c (update_noncloned_frequencies): Always
493 (inline_transform): Likewise.
494 * predict.c (counts_to_freqs): Remove useless conditional.
495 * profile-count.h (profile_count::apply_scale): Move sanity check.
496 * tree-inline.c (copy_bb): Always scale.
497 (copy_cfg_body): Likewise.
499 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
502 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
504 (*cmp_ite0): Add enabled_for_depr_it attribute.
505 (*cmp_ite1): Likewise.
507 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
509 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
512 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
514 * tree-vrp.c (vrp_int_const_binop): Return true on success and
515 return the value by pointer.
516 (extract_range_from_multiplicative_op_1): Update accordingly.
517 Return as soon as an operation fails.
519 2017-11-05 Tom de Vries <tom@codesourcery.com>
522 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
523 (DEF_SANITIZER_BUILTIN): ... here.
524 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
525 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
527 2017-11-05 Tom de Vries <tom@codesourcery.com>
530 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
532 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
533 ASM_OUTPUT_BEFORE_CASE_LABEL call.
534 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
536 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
537 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
538 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
540 2017-11-05 Tom de Vries <tom@codesourcery.com>
543 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
546 2017-11-04 Michael Clark <michaeljclark@mac.com>
548 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
549 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
551 (*addsi3_extended): Likewise.
552 (*addsi3_extended2): Likewise.
553 (<optab>si3): Likewise.
554 (<optab>di3): Likewise.
555 (<optab><mode>3): Likewise.
556 (<*optabe>si3_internal): Likewise.
557 (zero_extendqi<SUPERQI:mode>2): Likewise.
558 (*add<mode>hi3): Likewise.
559 (*xor<mode>hi3): Likewise.
560 (<optab>di3): Likewise.
561 (*<optab>si3_extend): Likewise.
562 (*sge<u>_<X:mode><GPR:mode>): Likewise.
563 (*slt<u>_<X:mode><GPR:mode>): Likewise.
564 (*sle<u>_<X:mode><GPR:mode>): Likewise.
566 2017-11-04 Andrew Waterman <andrew@sifive.com>
568 * config/riscv/riscv.c (riscv_option_override): Conditionally set
569 TARGET_STRICT_ALIGN based upon -mtune argument.
571 2017-11-04 Andrew Waterman <andrew@sifive.com>
573 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
575 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
577 config/i386/i386.c (choose_basereg): Use optional scratch
578 register and add assertion.
579 (x86_emit_outlined_ms2sysv_save): Use scratch register when
580 needed, and don't allocate stack.
581 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
582 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
583 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
585 2017-11-03 Jeff Law <law@redhat.com>
587 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
588 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
589 to probe at the start of a noreturn function.
591 2017-11-03 Jakub Jelinek <jakub@redhat.com>
593 PR tree-optimization/78821
594 * gimple-ssa-store-merging.c: Update the file comment.
595 (MAX_STORE_ALIAS_CHECKS): Define.
596 (struct store_operand_info): New type.
597 (store_operand_info::store_operand_info): New constructor.
598 (struct store_immediate_info): Add rhs_code and ops data members.
599 (store_immediate_info::store_immediate_info): Add rhscode, op0r
600 and op1r arguments to the ctor, initialize corresponding data members.
601 (struct merged_store_group): Add load_align_base and load_align
603 (merged_store_group::merged_store_group): Initialize them.
604 (merged_store_group::do_merge): Update them.
605 (merged_store_group::apply_stores): Pick the constant for
606 encode_tree_to_bitpos from one of the two operands, or skip
607 encode_tree_to_bitpos if neither operand is a constant.
608 (class pass_store_merging): Add process_store method decl. Remove
609 bool argument from terminate_all_aliasing_chains method decl.
610 (pass_store_merging::terminate_all_aliasing_chains): Remove
611 var_offset_p argument and corresponding handling.
612 (stmts_may_clobber_ref_p): New function.
613 (compatible_load_p): New function.
614 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
615 if there is overlap and rhs_code is not INTEGER_CST. For
616 non-overlapping stores terminate group if rhs is not mergeable.
617 (get_alias_type_for_stmts): Change first argument from
618 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
619 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
620 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
622 (get_location_for_stmts): Change first argument from
623 auto_vec<gimple *> & to vec<gimple *> &.
624 (struct split_store): Remove orig_stmts data member, add orig_stores.
625 (split_store::split_store): Create orig_stores rather than orig_stmts.
626 (find_constituent_stmts): Renamed to ...
627 (find_constituent_stores): ... this. Change second argument from
628 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
629 to info structures rather than the statements.
630 (split_group): Rename ALLOW_UNALIGNED argument to
631 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
632 it. Adjust find_constituent_stores caller.
633 (imm_store_chain_info::output_merged_store): Handle rhs_code other
634 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
635 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
636 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
637 (mem_valid_for_store_merging): New function.
638 (handled_load): New function.
639 (pass_store_merging::process_store): New method.
640 (pass_store_merging::execute): Use process_store method. Adjust
641 terminate_all_aliasing_chains caller.
643 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
645 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
646 Return true for more constants, symbols and label references.
647 (aarch64_valid_floating_const): Remove unused function.
649 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
652 * config/aarch64/aarch64.c (aarch64_layout_frame):
653 Undo forcing of LR at bottom of frame.
655 2017-11-03 Jeff Law <law@redhat.com>
658 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
659 for int_registers_saved.
661 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
662 extracted from tree-ssa-dom.c.
663 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
664 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
665 (record_equivalences_from_incoming_edge): Add additional argument
666 to single_pred_edge_ignoring_loop_edges call.
667 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
668 (uncprop_dom_walker::before_dom_children): Add additional argument
669 to single_pred_edge_ignoring_loop_edges call.
670 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
671 single_pred_edge_ignoring_loop_edges rather than open coding.
672 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
674 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
676 * match.pd (-(-A)): Rewrite.
678 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
680 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
681 (rs6000_emit_int_cmove): New declaration.
682 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
683 (rs6000_emit_sISEL): Delete.
684 (rs6000_emit_int_cmove): Make non-static.
685 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
686 instead of rs6000_emit_sISEL.
688 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
690 * asan.c (create_cond_insert_point): Maintain profile.
691 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
693 * basic-block.h (struct basic_block_def): Remove frequency.
694 (EDGE_FREQUENCY): Use to_frequency
695 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
697 (find_traces): Update to use to_frequency.
698 (find_traces_1_round): Likewise; use only IPA counts.
699 (bb_to_key): Likewise.
700 (connect_traces): Use IPA counts only.
701 (copy_bb_p): Update to use to_frequency.
702 (fix_up_crossing_landing_pad): Likewise.
703 (sanitize_hot_paths): Likewise.
704 * bt-load.c (basic_block_freq): Likewise.
705 * cfg.c (init_flow): Set count_max to uninitialized.
706 (check_bb_profile): Remove frequencies; check counts.
707 (dump_bb_info): Do not dump frequencies.
708 (update_bb_profile_for_threading): Update counts only.
709 (scale_bbs_frequencies_int): Likewise.
710 (MAX_SAFE_MULTIPLIER): Remove.
711 (scale_bbs_frequencies_gcov_type): Update counts only.
712 (scale_bbs_frequencies_profile_count): Update counts only.
713 (scale_bbs_frequencies): Update counts only.
714 * cfg.h (struct control_flow_graph): Add count-max.
715 (update_bb_profile_for_threading): Update prototype.
716 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
717 (find_many_sub_basic_blocks): Likewise.
718 * cfgcleanup.c (try_forward_edges): Likewise.
719 (try_crossjump_to_edge): Likewise.
720 * cfgexpand.c (expand_gimple_cond): Likewise.
721 (expand_gimple_tailcall): Likewise.
722 (construct_init_block): Likewise.
723 (construct_exit_block): Likewise.
724 * cfghooks.c (verify_flow_info): Check consistency of counts.
725 (dump_bb_for_graph): Do not dump frequencies.
726 (split_block_1): Do not update frequencies.
727 (split_edge): Do not update frequencies.
728 (make_forwarder_block): Do not update frequencies.
729 (duplicate_block): Do not update frequencies.
730 (account_profile_record): Do not update frequencies.
731 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
732 for global heuristics.
733 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
734 (expected_loop_iterations_unbounded): Use counts only.
735 * cfgloopmanip.c (scale_loop_profile): Simplify.
736 (create_empty_loop_on_edge): Simplify
738 (duplicate_loop_to_header_edge): Simplify
739 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
740 (update_br_prob_note): Take care of removing note when profile
742 (relink_block_chain): Do not dump frequency.
743 (rtl_account_profile_record): Use to_frequency.
744 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
745 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
746 (cgraph_update_edges_for_call_stmt_node): Likewise.
747 (cgraph_edge::verify_count_and_frequency): Update.
748 (cgraph_node::verify_node): Temporarily disable frequency verification.
749 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
751 (cgraph_edge::rebuild_edges): Convert to ipa counts.
752 * cgraphunit.c (init_lowered_empty_function): Do not initialize
754 (cgraph_node::expand_thunk): Update profile.
755 * except.c (dw2_build_landing_pads): Do not update frequency.
756 * final.c (compute_alignments): Use to_frequency.
757 (dump_basic_block_info): Do not dump frequency.
758 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
759 (dump_gimple_bb_header): Do not dump frequency.
760 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
762 * gimple-streamer-in.c (input_bb): Do not stream frequency.
763 * gimple-streamer-out.c (output_bb): Do not stream frequency.
764 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
765 (init_before_recovery): Do not update frequency.
766 (sched_create_recovery_edges): Do not update frequency.
767 * hsa-gen.c (convert_switch_statements): Do not update frequency.
768 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
769 (ipa_cp_c_finalize): Set max_count to uninitialized.
770 * ipa-fnsummary.c (get_minimal_bb): Use counts.
771 (param_change_prob): Use counts.
772 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
774 * ipa-split.c (consider_split): Use to_frequency.
775 (split_function): Use to_frequency.
776 * ira-build.c (loop_compare_func): Likewise.
777 (mark_loops_for_removal): Likewise.
778 (mark_all_loops_for_removal): Likewise.
779 * loop-doloop.c (doloop_modify): Do not update frequency.
780 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
782 * lto-streamer-in.c (input_function): Update count_max.
783 * omp-expand.c (expand_omp_taskreg): Update count_max.
784 * omp-simd-clone.c (simd_clone_adjust): Update profile.
785 * predict.c (maybe_hot_frequency_p): Use to_frequency.
786 (maybe_hot_count_p): Use ipa counts only.
787 (maybe_hot_bb_p): Simplify.
788 (maybe_hot_edge_p): Simplify.
789 (probably_never_executed): Do not take frequency argument.
790 (probably_never_executed_bb_p): Do not pass frequency.
791 (probably_never_executed_edge_p): Likewise.
792 (combine_predictions_for_bb): Check that profile is nonzero.
793 (propagate_freq): Do not set frequency.
794 (drop_profile): Simplify.
795 (counts_to_freqs): Simplify.
796 (expensive_function_p): Use to_frequency.
797 (propagate_unlikely_bbs_forward): Simplify.
798 (determine_unlikely_bbs): Simplify.
799 (estimate_bb_frequencies): Add hack to silence graphite issues.
800 (compute_function_frequency): Use ipa counts.
801 (pass_profile::execute): Update.
802 (rebuild_frequencies): Use counts only.
803 (force_edge_cold): Use counts only.
804 * profile-count.c (profile_count::dump): Dump new count types.
805 (profile_count::differs_from_p): Check compatiblity.
806 (profile_count::to_frequency): New function.
807 (profile_count::to_cgraph_frequency): New function.
808 * profile-count.h (struct function): Declare.
809 (enum profile_quality): Add profile_guessed_local and
810 profile_guessed_global0.
811 (class profile_proability): Decrease number of bits to 29;
812 update from_reg_br_prob_note and to_reg_br_prob_note.
813 (class profile_count: Update comment; decrease number of bits
814 to 61. Check compatibility.
815 (profile_count::compatible_p): New private member function.
816 (profile_count::ipa_p): New member function.
817 (profile_count::operator<): Handle global zero correctly.
818 (profile_count::operator>): Handle global zero correctly.
819 (profile_count::operator<=): Handle global zero correctly.
820 (profile_count::operator>=): Handle global zero correctly.
821 (profile_count::nonzero_p): New member function.
822 (profile_count::force_nonzero): New member function.
823 (profile_count::max): New member function.
824 (profile_count::apply_scale): Handle IPA scalling.
825 (profile_count::guessed_local): New member function.
826 (profile_count::global0): New member function.
827 (profile_count::ipa): New member function.
828 (profile_count::to_frequency): Declare.
829 (profile_count::to_cgraph_frequency): Declare.
830 * profile.c (OVERLAP_BASE): Delete.
831 (compute_frequency_overlap): Delete.
832 (compute_branch_probabilities): Do not use compute_frequency_overlap.
833 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
834 * sched-ebb.c (rank): Use counts only.
835 * shrink-wrap.c (handle_simple_exit): Use counts only.
836 (try_shrink_wrapping): Use counts only.
837 (place_prologue_for_one_component): Use counts only.
838 * tracer.c (find_best_predecessor): Use to_frequency.
839 (find_trace): Use to_frequency.
840 (tail_duplicate): Use to_frequency.
841 * trans-mem.c (expand_transaction): Do not update frequency.
842 * tree-call-cdce.c: Do not update frequency.
843 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
844 (gimple_merge_blocks): Likewise.
845 (gimple_split_edge): Likewise.
846 (gimple_duplicate_sese_region): Likewise.
847 (gimple_duplicate_sese_tail): Likewise.
848 (move_sese_region_to_fn): Likewise.
849 (gimple_account_profile_record): Likewise.
850 (insert_cond_bb): Likewise.
851 * tree-complex.c (expand_complex_div_wide): Likewise.
852 * tree-eh.c (lower_resx): Update profile.
853 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
855 (initialize_cfun): Do not initialize frequencies
856 (freqs_to_counts): Delete.
857 (copy_cfg_body): Ignore count parameter.
859 (expand_call_inline): Update count_max.
860 (optimize_inline_calls): Update count_max.
861 (tree_function_versioning): Update count_max.
862 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
863 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
865 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
866 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
867 (try_peel_loop): Likewise.
868 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
870 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
871 (tree_transform_and_unroll_loop): Do not use frequencies
872 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
873 Use reliable prediction only.
874 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
875 * tree-ssa-sink.c (select_best_block): Use to_frequency.
876 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
878 * tree-ssa-threadupdate.c (create_block_for_threading): Do
880 (any_remaining_duplicated_blocks): Likewise.
881 (update_profile): Likewise.
882 (estimated_freqs_path): Delete.
883 (freqs_to_counts_path): Delete.
884 (clear_counts_path): Delete.
885 (ssa_fix_duplicate_block_edges): Likewise.
886 (duplicate_thread_path): Likewise.
887 * tree-switch-conversion.c (gen_inbound_check): Use counts.
888 * tree-tailcall.c (decrease_profile): Do not update frequency.
889 (eliminate_tail_call): Likewise.
890 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
891 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
892 (optimize_mask_stores): Likewise.
893 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
894 * ubsan.c (ubsan_expand_null_ifn): Update profile.
895 (ubsan_expand_ptr_ifn): Update profile.
896 * value-prof.c (gimple_ic): Simplify.
897 * value-prof.h (gimple_ic): Update prototype.
898 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
899 * ipa-inline.c (compute_uninlined_call_time): Be sure that
901 (want_inline_self_recursive_call_p): Likewise.
902 (resolve_noninline_speculation): Only cummulate defined counts.
903 (inline_small_functions): Use nonzero_p.
904 (ipa_inline): Do not access freed node.
906 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
908 * config/aarch64/aarch64.c (aarch64_override_options_internal):
909 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
911 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
913 * config/riscv/riscv.c (riscv_legitimize_move): Handle
914 non-legitimate address.
916 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
918 * config/rs6000/rs6000.md (*lt0_disi): Delete.
919 (*lt0_<mode>di, *lt0_<mode>si): New.
921 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
923 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
926 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
927 Jim Wilson <jim.wilson@linaro.org>
929 * config/aarch64/aarch64-cores.def (saphira): New CPU.
930 * config/aarch64/aarch64-tune.md: Regenerated.
931 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
932 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
934 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
936 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
937 (arc_expand_prologue): Restore blink for millicode.
938 * config/arc/linux.h (LINK_EH_SPEC): Defined.
940 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
943 * config/i386/i386.c (ix86_vector_duplicate_value): Use
944 gen_vec_duplicate after forcing the scalar into a register.
946 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
948 * combine (try_combine): Print the insns input to try_combine to the
951 2017-11-02 Steve Ellcey <sellcey@cavium.com>
954 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
955 Remove second argument from aarch64_process_target_attr call.
956 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
958 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
960 (aarch64_handle_attr_arch): Remove second argument.
961 (aarch64_handle_attr_cpu): Ditto.
962 (aarch64_handle_attr_tune): Ditto.
963 (aarch64_handle_attr_isa_flags): Ditto.
964 (aarch64_process_one_target_attr): Ditto.
965 (aarch64_process_target_attr): Ditto.
966 (aarch64_option_valid_attribute_p): Remove second argument.
967 on aarch64_process_target_attr call.
969 2017-11-02 David Malcolm <dmalcolm@redhat.com>
971 * diagnostic.c: Include "selftest-diagnostic.h".
972 (selftest::assert_location_text): New function.
973 (selftest::test_diagnostic_get_location_text): New function.
974 (selftest::diagnostic_c_tests): Call it.
976 2017-11-02 David Malcolm <dmalcolm@redhat.com>
978 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
979 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
980 (class selftest::test_diagnostic_context): Move to...
981 * selftest-diagnostic.c: New file.
982 * selftest-diagnostic.h: New file.
984 2017-11-02 James Bowman <james.bowman@ftdichip.com>
986 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
987 offset range for FT32B.
988 * config/ft32/ft32.h: option "mcompress" enables relaxation.
989 * config/ft32/ft32.md: Add TARGET_NOPM.
990 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
991 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
993 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
995 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
997 2017-11-02 Jeff Law <law@redhat.com>
999 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
1000 virtual keyword on FINAL OVERRIDE members.
1002 * tree-ssa-propagate.h (ssa_propagation_engine): Group
1003 virtuals together. Add virtual destructor.
1004 (substitute_and_fold_engine): Similarly.
1006 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
1008 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
1010 2017-11-02 Richard Biener <rguenther@suse.de>
1012 PR tree-optimization/82795
1013 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
1015 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1017 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
1019 (gcc_GAS_CHECK_FEATURE): Remove.
1020 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
1021 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
1022 * configure: Regenerate.
1024 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
1026 * config/arc/arc.c (hwloop_optimize): Account for empty
1029 2017-11-02 Richard Biener <rguenther@suse.de>
1032 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
1033 Truncate ARRAY_REF index and element size.
1035 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
1037 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
1039 2017-11-01 Jeff Law <law@redhat.com>
1041 * tree-ssa-ccp.c (ccp_folder): New class derived from
1042 substitute_and_fold_engine.
1043 (ccp_folder::get_value): New member function.
1044 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
1045 (ccp_fold_stmt): Remove prototype.
1046 (ccp_finalize): Call substitute_and_fold from the ccp_class.
1047 * tree-ssa-copy.c (copy_folder): New class derived from
1048 substitute_and_fold_engine.
1049 (copy_folder::get_value): Renamed from get_value.
1050 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
1051 * tree-vrp.c (vrp_folder): New class derived from
1052 substitute_and_fold_engine.
1053 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
1054 (vrp_folder::get_value): New member function.
1055 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
1056 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
1057 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
1058 provide a class interface to folder/substitute routines.
1059 (ssa_prop_fold_stmt_fn): Remove typedef.
1060 (ssa_prop_get_value_fn): Likewise.
1061 (subsitute_and_fold): Remove prototype.
1062 (replace_uses_in): Likewise.
1063 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
1064 Renamed from replace_uses_in. Call the virtual member function
1065 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
1066 (substitute_and_fold_dom_walker): Remove initialization of
1067 data member entries for calbacks. Add substitute_and_fold_engine
1068 member and initialize it.
1069 (substitute_and_fold_dom_walker::before_dom_children0: Use the
1070 member functions for get_value, replace_phi_args_in c
1071 replace_uses_in, and fold_stmt calls.
1072 (substitute_and_fold_engine::substitute_and_fold): Renamed from
1073 substitute_and_fold. Remove assert. Update ctor call.
1075 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
1076 (ssa_prop_visit_phi_fn): Likewise.
1077 (class ssa_propagation_engine): New class to provide an interface
1079 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
1081 (ssa_prop_visit_phi): Likewise.
1082 (ssa_propagation_engine::simulate_stmt): Moved into class.
1083 Call visit_phi/visit_stmt from the class rather than via
1084 file scoped static variables.
1085 (ssa_propagation_engine::simulate_block): Moved into class.
1086 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
1087 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
1088 set file scoped statics for the visit_stmt/visit_phi callbacks.
1089 * tree-complex.c (complex_propagate): New class derived from
1090 ssa_propagation_engine.
1091 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
1092 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
1093 (tree_lower_complex): Call ssa_propagate via the complex_propagate
1095 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
1096 ssa_propagation_engine.
1097 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
1098 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
1099 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
1100 * tree-ssa-copy.c (copy_prop): New class derived from
1101 ssa_propagation_engine.
1102 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
1103 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
1104 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
1105 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
1106 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
1107 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
1108 (execute_vrp): Call ssa_propagate from the vrp_prop class.
1110 2017-11-01 Jakub Jelinek <jakub@redhat.com>
1112 PR rtl-optimization/82778
1113 PR rtl-optimization/82597
1114 * compare-elim.c (struct comparison): Add in_a_setter field.
1115 (find_comparison_dom_walker::before_dom_children): Remove killed
1116 bitmap and df_simulate_find_defs call, instead walk the defs.
1117 Compute last_setter and initialize in_a_setter. Merge definitions
1118 with first initialization for a few variables.
1119 (try_validate_parallel): Use insn_invalid_p instead of
1120 recog_memoized. Return insn rather than just the pattern.
1121 (try_merge_compare): Fix up comment. Don't uselessly test if
1122 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
1124 (execute_compare_elim_after_reload): Remove df_chain_add_problem
1127 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1128 Alan Hayward <alan.hayward@arm.com>
1129 David Sherwood <david.sherwood@arm.com>
1131 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
1132 aarch64_hard_regno_nregs to get the number of registers
1135 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1136 Alan Hayward <alan.hayward@arm.com>
1137 David Sherwood <david.sherwood@arm.com>
1139 * config/aarch64/constraints.md (Upl): Rename to...
1141 * config/aarch64/aarch64.md
1142 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
1145 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1146 Alan Hayward <alan.hayward@arm.com>
1147 David Sherwood <david.sherwood@arm.com>
1149 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
1150 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
1153 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1154 Alan Hayward <alan.hayward@arm.com>
1155 David Sherwood <david.sherwood@arm.com>
1157 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
1158 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
1159 (aarch64_evpc_dup): Generate rtl direcly, rather than using
1161 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
1162 of a single element.
1163 * config/aarch64/iterators.md: Add a comment above the permute
1164 unspecs to say that they are generated directly by
1165 aarch64_expand_vec_perm_const.
1166 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
1168 2017-11-01 Nathan Sidwell <nathan@acm.org>
1170 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
1172 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
1174 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
1175 and medany code models, and describe what they do.
1177 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1179 Revert accidental duplicate:
1181 * combine.c (can_change_dest_mode): Reject changes in
1182 REGMODE_NATURAL_SIZE.
1184 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
1186 PR rtl-optimization/64682
1187 PR rtl-optimization/69567
1188 PR rtl-optimization/69737
1189 PR rtl-optimization/82683
1190 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
1191 register mentioned in the note, drop the note, unless it came from I3,
1192 in which case it should go to I3 again.
1194 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1196 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
1197 and return false if not.
1198 (clear_bytes_written_by, live_bytes_read): Update accordingly.
1200 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1202 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
1203 range is known to be empty.
1205 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1206 Alan Hayward <alan.hayward@arm.com>
1207 David Sherwood <david.sherwood@arm.com>
1209 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
1210 and CONST_VECTOR_NUNITS instead of computing the number of units from
1211 the byte sizes of the vector and element.
1212 (simplify_binary_operation_1): Likewise.
1213 (simplify_const_binary_operation): Likewise.
1214 (simplify_ternary_operation): Likewise.
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 * var-tracking.c (INT_MEM_OFFSET): Replace with...
1221 (int_mem_offset): ...this new function.
1222 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
1223 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
1224 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
1227 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1228 Alan Hayward <alan.hayward@arm.com>
1229 David Sherwood <david.sherwood@arm.com>
1231 * lower-subreg.c (interesting_mode_p): New function.
1232 (compute_costs, find_decomposable_subregs, decompose_register)
1233 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
1234 (resolve_clobber, dump_choices): Use it.
1236 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1237 Alan Hayward <alan.hayward@arm.com>
1238 David Sherwood <david.sherwood@arm.com>
1240 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
1242 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1243 Alan Hayward <alan.hayward@arm.com>
1244 David Sherwood <david.sherwood@arm.com>
1246 * alias.c (find_base_value, find_base_term): Only process integer
1247 truncations. Check the precision rather than the size.
1249 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1250 Alan Hayward <alan.hayward@arm.com>
1251 David Sherwood <david.sherwood@arm.com>
1253 * machmode.h (is_narrower_int_mode): New function
1254 * optabs.c (expand_float, expand_fix): Use it.
1255 * dwarf2out.c (rotate_loc_descriptor): Likewise.
1257 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1258 Alan Hayward <alan.hayward@arm.com>
1259 David Sherwood <david.sherwood@arm.com>
1261 * rtl.h (narrower_subreg_mode): New function.
1262 * ira-color.c (update_costs_from_allocno): Use it.
1264 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1265 Alan Hayward <alan.hayward@arm.com>
1266 David Sherwood <david.sherwood@arm.com>
1268 * optabs-query.h (convert_optab_p): New function, split out from...
1269 (convert_optab_handler): ...here.
1270 (widening_optab_handler): Delete.
1271 (find_widening_optab_handler): Remove permit_non_widening parameter.
1272 (find_widening_optab_handler_and_mode): Likewise. Provide an
1273 override that operates on mode class wrappers.
1274 * optabs-query.c (widening_optab_handler): Delete.
1275 (find_widening_optab_handler_and_mode): Remove permit_non_widening
1276 parameter. Assert that the two modes are the same class and that
1277 the "from" mode is narrower than the "to" mode. Use
1278 convert_optab_handler instead of widening_optab_handler.
1279 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
1280 instead of widening_optab_handler.
1281 * expr.c (expand_expr_real_2): Update calls to
1282 find_widening_optab_handler.
1283 * optabs.c (expand_widen_pattern_expr): Likewise.
1284 (expand_binop_directly): Take the insn_code as a parameter.
1285 (expand_binop): Only call find_widening_optab_handler for
1286 conversion optabs; use optab_handler otherwise. Update calls
1287 to find_widening_optab_handler and expand_binop_directly.
1288 Use convert_optab_handler instead of widening_optab_handler.
1289 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
1290 find_widening_optab_handler and use scalar_mode rather than
1292 (convert_plusminus_to_widen): Likewise.
1294 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1295 Alan Hayward <alan.hayward@arm.com>
1296 David Sherwood <david.sherwood@arm.com>
1298 * machmode.h (fixed_size_mode): New class.
1299 * rtl.h (get_pool_mode): Return fixed_size_mode.
1300 * gengtype.c (main): Add fixed_size_mode.
1301 * target.def (get_raw_result_mode): Return a fixed_size_mode.
1302 (get_raw_arg_mode): Likewise.
1303 * doc/tm.texi: Regenerate.
1304 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
1305 * targhooks.c (default_get_reg_raw_mode): Likewise.
1306 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
1307 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
1308 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
1309 (msp430_get_raw_result_mode): Likewise.
1310 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
1311 * dbxout.c (dbxout_parms): Require fixed-size modes.
1312 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
1313 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
1314 * omp-low.c (lower_oacc_reductions): Likewise.
1315 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
1316 (simplify_subreg): Update accordingly.
1317 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
1318 (force_const_mem): Update accordingly. Return NULL_RTX for modes
1319 that aren't fixed-size.
1320 (get_pool_mode): Return a fixed_size_mode.
1321 (output_constant_pool_2): Take a fixed_size_mode.
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 * doc/rtl.texi (vec_series): Document.
1328 (const): Say that the operand can be a vec_series.
1329 * rtl.def (VEC_SERIES): New rtx code.
1330 * rtl.h (const_vec_series_p_1): Declare.
1331 (const_vec_series_p): New function.
1332 * emit-rtl.h (gen_const_vec_series): Declare.
1333 (gen_vec_series): Likewise.
1334 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
1335 (gen_vec_series): Likewise.
1336 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
1337 * simplify-rtx.c (simplify_unary_operation): Handle negations
1339 (simplify_binary_operation_series): New function.
1340 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
1341 (test_vector_ops_series): New function.
1342 (test_vector_ops): Call it.
1343 * config/powerpcspe/altivec.md (altivec_lvsl): Use
1344 gen_const_vec_series.
1345 (altivec_lvsr): Likewise.
1346 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
1348 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1349 Alan Hayward <alan.hayward@arm.com>
1350 David Sherwood <david.sherwood@arm.com>
1352 * doc/rtl.texi (const): Update description of address constants.
1353 Say that vector constants are allowed too.
1354 * common.md (E, F): Use CONSTANT_P instead of checking for
1356 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
1357 checking for CONST_VECTOR.
1358 * expmed.c (make_tree): Use build_vector_from_val for a CONST
1360 * expr.c (expand_expr_real_2): Check for vector modes instead
1361 of checking for CONST_VECTOR.
1362 * rtl.h (const_vec_p): New function.
1363 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
1364 (unwrap_const_vec_duplicate): Handle them here too.
1366 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1367 David Malcolm <dmalcolm@redhat.com>
1368 Alan Hayward <alan.hayward@arm.com>
1369 David Sherwood <david.sherwood@arm.com>
1371 * rtl.h (vec_duplicate_p): New function.
1372 * selftest-rtl.c (assert_rtx_eq_at): New function.
1373 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
1374 (assert_rtx_eq_at): Declare.
1375 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
1376 * selftest-run-tests.c (selftest::run_tests): Call it.
1377 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
1378 (simplify_unary_operation_1): Recursively handle vector duplicates.
1379 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
1381 (simplify_subreg): Handle subregs of vector duplicates.
1382 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
1383 (selftest::simplify_rtx_c_tests): New functions.
1385 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1386 Alan Hayward <alan.hayward@arm.com>
1387 David Sherwood <david.sherwood@arm.com>
1389 * emit-rtl.h (gen_const_vec_duplicate): Declare.
1390 (gen_vec_duplicate): Likewise.
1391 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
1393 (gen_const_vector): ...here.
1394 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
1395 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
1396 whose elements are all equal.
1397 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
1398 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
1399 (simplify_relational_operation): Likewise.
1400 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
1402 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
1403 (aarch64_expand_vector_init): Likewise.
1404 * config/arm/arm.c (neon_vdup_constant): Likewise.
1405 (neon_expand_vector_init): Likewise.
1406 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
1407 (arm_block_set_unaligned_vect): Likewise.
1408 (arm_block_set_aligned_vect): Likewise.
1409 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
1410 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
1411 (expand_vec_perm_even_odd_pack): Likewise.
1412 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
1413 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
1414 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
1415 gen_const_vec_duplicate.
1416 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
1417 * config/mips/mips.c (mips_gen_const_int_vector): Use
1418 gen_const_vec_duplicate.
1419 (mips_expand_vector_init): Use CONST0_RTX.
1420 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
1421 (define_split): Use gen_const_vec_duplicate.
1422 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
1423 (define_split): Use gen_const_vec_duplicate.
1424 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
1425 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
1426 * config/spu/spu.c (spu_const): Likewise.
1428 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1429 Alan Hayward <alan.hayward@arm.com>
1430 David Sherwood <david.sherwood@arm.com>
1432 * combine.c (can_change_dest_mode): Reject changes in
1433 REGMODE_NATURAL_SIZE.
1435 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
1437 * configure.ac (--enable-libssp): New.
1438 (gcc_cv_libc_provides_ssp): Check for explicit setting before
1439 trying to determine target-specific default. Adjust indentation.
1440 * configure: Regenerated.
1441 * doc/install.texi (Configuration): Expand --disable-libssp
1444 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
1446 config/i386/i386.c (ix86_expand_epilogue): Correct stack
1449 2017-10-31 Martin Jambor <mjambor@suse.cz>
1452 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
1454 2017-10-31 David Malcolm <dmalcolm@redhat.com>
1456 * auto-profile.c (autofdo_source_profile::read): Use
1457 UNKNOWN_LOCATION rather than 0.
1458 * diagnostic-core.h (warning_at_rich_loc): Rename to...
1459 (warning_at): ...this overload.
1460 (warning_at_rich_loc_n): Rename to...
1461 (warning_n): ...this overload.
1462 (error_at_rich_loc): Rename to...
1463 (error_at): ...this overload.
1464 (pedwarn_at_rich_loc): Rename to...
1465 (pedwarn): ...this overload.
1466 (permerror_at_rich_loc): Rename to...
1467 (permerror): ...this overload.
1468 (inform_at_rich_loc): Rename to...
1469 (inform): ...this overload.
1470 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
1471 (diagnostic_n_impl_richloc): Rename to...
1472 (diagnostic_n_impl): ...this rich_location *-based decl.
1473 (inform_at_rich_loc): Rename to...
1474 (inform): ...this, and add an assertion.
1475 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
1476 (warning_at_rich_loc): Rename to...
1477 (warning_at): ...this, and add an assertion.
1478 (warning_at_rich_loc_n): Rename to...
1479 (warning_n): ...this, and add an assertion.
1480 (warning_n): Update location_t-based implementation for removal of
1481 location_t-based diagnostic_n_impl.
1482 (pedwarn_at_rich_loc): Rename to...
1483 (pedwarn): ...this, and add an assertion.
1484 (permerror_at_rich_loc): Rename to...
1485 (permerror): ...this, and add an assertion.
1486 (error_n): Update for removal of location_t-based diagnostic_n_impl.
1487 (error_at_rich_loc): Rename to...
1488 (error_at): ...this, and add an assertion.
1489 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
1490 (driver::do_spec_on_infiles): Likewise.
1491 * substring-locations.c (format_warning_va): Update for renaming
1492 of inform_at_rich_loc.
1494 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
1496 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
1497 _Float<N>X built-in functions so that the variant without the
1498 "__builtin_" prefix is only enabled for the GNU C and Objective C
1499 languages when they are in non-strict ANSI/ISO mode.
1500 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
1501 * target.def (floatn_builtin_p): Add a target hook to control
1502 whether _Float<N> and _Float<N>X built-in functions without the
1503 "__builtin_" prefix are enabled, and return true for C and
1504 Objective C in the default hook. Include langhooks.h in
1506 * targhooks.h (default_floatn_builtin_p): Likewise.
1507 * targhooks.c (default_floatn_builtin_p): Likewise.
1508 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
1509 floatn_builtin_p target hook.
1510 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
1512 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
1513 Eric Botcazou <ebotcazou@adacore.com>
1515 PR rtl-optimization/81803
1516 * lra-constraints.c (curr_insn_transform): Also reload the whole
1517 register for a strict subreg no wider than a word if this is for
1518 a WORD_REGISTER_OPERATIONS target.
1520 2017-10-31 Jason Merrill <jason@redhat.com>
1522 * gdbinit.in: Skip over inlines from timevar.h.
1524 2017-10-31 Martin Liska <mliska@suse.cz>
1526 * doc/gcov.texi: Document new option.
1527 * gcov.c (print_usage): Likewise print it.
1528 (process_args): Support the argument.
1529 (format_count): New function.
1530 (format_gcov): Use the function.
1532 2017-10-31 Martin Liska <mliska@suse.cz>
1534 * gcov.c (struct name_map): do not use typedef.
1535 Define operator== and operator<.
1536 (name_search): Remove.
1537 (name_sort): Remove.
1538 (main): Do not allocate names.
1539 (process_file): Add vertical space.
1540 (generate_results): Use std::find.
1541 (release_structures): Do not release memory.
1542 (find_source): Use std::find.
1544 2017-10-31 Martin Liska <mliska@suse.cz>
1546 * gcov.c (struct line_info): Remove it's typedef.
1547 (line_info::line_info): Add proper ctor.
1548 (line_info::has_block): Do not use a typedef.
1549 (struct source_info): Do not use typedef.
1550 (circuit): Likewise.
1551 (get_cycles_count): Likewise.
1552 (output_intermediate_file): Iterate via vector iterator.
1553 (add_line_counts): Use std::vector methods.
1554 (accumulate_line_counts): Likewise.
1555 (output_lines): Likewise.
1557 2017-10-31 Martin Liska <mliska@suse.cz>
1559 * gcov.c (struct source_info): Remove typedef.
1560 (source_info::source_info): Add proper ctor.
1561 (accumulate_line_counts): Use struct, not it's typedef.
1562 (output_gcov_file): Likewise.
1563 (output_lines): Likewise.
1564 (main): Do not allocate an array.
1565 (output_intermediate_file): Use size of vector container.
1566 (process_file): Resize the vector.
1567 (generate_results): Do not preallocate, use newly added vector
1569 (release_structures): Do not release sources.
1570 (find_source): Use vector methods.
1571 (add_line_counts): Do not use typedef.
1573 2017-10-31 Martin Liska <mliska@suse.cz>
1575 * doc/gcov.texi: Document that.
1576 * gcov.c (add_line_counts): Mark lines with a non-executed
1578 (output_line_beginning): Handle such lines.
1579 (output_lines): Pass new argument.
1580 (output_intermediate_file): Print it in intermediate format.
1582 2017-10-31 Martin Liska <mliska@suse.cz>
1584 * color-macros.h: New file.
1585 * diagnostic-color.c: Factor out color related to macros to
1587 * doc/gcov.texi: Document -k option.
1588 * gcov.c (INCLUDE_STRING): Include string.h.
1589 (print_usage): Add -k option.
1590 (process_args): Parse it.
1591 (pad_count_string): New function.
1592 (output_line_beginning): Likewise.
1593 (DEFAULT_LINE_START): New macro.
1594 (output_lines): Support color output.
1596 2017-10-31 Martin Liska <mliska@suse.cz>
1598 PR gcov-profile/82633
1599 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
1600 their interaction with GCOV infrastructure.
1601 * configure.ac: Add -fkeep-{inline,static}-functions to
1603 * configure: Regenerate.
1605 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
1608 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
1610 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
1613 * config/rs6000/rs6000.md (allocate_stack): Force update interval
1614 into a register if it does not fit into an immediate offset field.
1616 2017-10-31 Olivier Hainque <hainque@adacore.com>
1618 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
1620 2017-10-31 Julia Koval <julia.koval@intel.com>
1622 * config.gcc: Add gfniintrin.h.
1623 * config/i386/gfniintrin.h: New.
1624 * config/i386/i386-builtin-types.def
1625 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
1626 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
1627 __builtin_ia32_vgf2p8affineinvqb_v32qi,
1628 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
1629 __builtin_ia32_vgf2p8affineinvqb_v16qi,
1630 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
1631 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
1632 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
1633 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
1634 V64QI_FTYPE_V64QI_V64QI_INT): New types.
1635 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1636 * config/i386/immintrin.h: Include gfniintrin.h.
1637 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
1639 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
1641 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
1643 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
1645 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
1646 (arm_ashldi3_1bit): Remove pattern.
1647 (ashrdi3): Remove shift by 1 expansion.
1648 (arm_ashrdi3_1bit): Remove pattern.
1649 (lshrdi3): Remove shift by 1 expansion.
1650 (arm_lshrdi3_1bit): Remove pattern.
1651 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
1652 cost of ashldi3 by 1.
1653 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
1654 (<shift>di3_neon): Likewise.
1656 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
1658 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
1659 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
1660 and (*aarch64_simd_mov<VQ:mode>).
1661 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
1662 pattern alternative.
1663 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
1664 attributes to match pattern alternative.
1666 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
1668 * config.gcc (powerpc*-*-*): Add emmintrin.h.
1669 * config/rs6000/emmintrin.h: New file.
1670 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
1672 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
1674 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
1675 * (movdi_vfp_cortexa8): Remove pattern.
1677 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1679 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
1681 (Specific, alpha*-dec-osf5.1): Remove.
1682 (Specific, mips-sgi-irix5): Remove.
1683 (Specific, mips-sgi-irix6): Remove.
1685 2017-10-30 Jakub Jelinek <jakub@redhat.com>
1688 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
1689 arguments to clear_bit_region_be.
1691 2017-10-30 Jim Wilson <wilson@tuliptree.org>
1693 * gimplify.c: Include memmodel.h.
1695 2017-10-30 Martin Jambor <mjambor@suse.cz>
1697 * omp-grid.c (grid_attempt_target_gridification): Also insert a
1698 condition whether loop should be executed at all.
1700 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
1702 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
1703 gimple folding of vec_madd() intrinsics.
1704 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
1705 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
1706 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
1708 2017-10-30 Richard Biener <rguenther@suse.de>
1710 PR tree-optimization/82762
1712 2017-10-23 Richard Biener <rguenther@suse.de>
1714 PR tree-optimization/82129
1716 2017-08-01 Richard Biener <rguenther@suse.de>
1718 PR tree-optimization/81181
1719 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
1720 (compute_antic): ... end of iteration here.
1722 2017-10-30 Joseph Myers <joseph@codesourcery.com>
1724 * doc/invoke.texi (C Dialect Options): Document -std=c17,
1725 -std=iso9899:2017 and -std=gnu17.
1726 * doc/standards.texi (C Language): Document C17 support.
1727 * doc/cpp.texi (Overview): Mention -std=c17.
1728 (Standard Predefined Macros): Document C11 and C17 values of
1729 __STDC_VERSION__. Do not refer to C99 support as incomplete.
1730 * doc/extend.texi (Inline): Do not list individual options for
1731 standards newer than C99.
1732 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
1734 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
1737 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
1739 * asan.c (asan_finish_file): Align asan globals array by shadow
1742 2017-10-30 Jakub Jelinek <jakub@redhat.com>
1745 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
1746 (struct store_immediate_info): Add bitregion_start and bitregion_end
1748 (store_immediate_info::store_immediate_info): Add brs and bre
1749 arguments and initialize bitregion_{start,end} from those.
1750 (struct merged_store_group): Add bitregion_start, bitregion_end,
1751 align_base and mask fields. Drop unnecessary struct keyword from
1752 struct store_immediate_info. Add do_merge method.
1753 (clear_bit_region_be): Use memset instead of loop storing zeros.
1754 (merged_store_group::do_merge): New method.
1755 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
1756 stores as long as the surrounding bitregions have no gaps.
1757 (merged_store_group::merge_overlapping): Use do_merge.
1758 (merged_store_group::apply_stores): Test that bitregion_{start,end}
1759 is byte aligned, rather than requiring that start and width are
1760 byte aligned. Drop unnecessary struct keyword from
1761 struct store_immediate_info. Allocate and populate also mask array.
1762 Make start of the arrays relative to bitregion_start rather than
1763 start and size them according to bitregion_{end,start} difference.
1764 (struct imm_store_chain_info): Drop unnecessary struct keyword from
1765 struct store_immediate_info.
1766 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
1767 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
1768 struct keyword from struct store_immediate_info.
1769 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
1770 between stores as long as the surrounding bitregions have no gaps.
1772 (struct split_store): Add orig non-static data member.
1773 (split_store::split_store): Initialize orig to false.
1774 (find_constituent_stmts): Return store_immediate_info *, non-NULL
1775 if there is exactly a single original stmt. Change stmts argument
1776 to pointer from reference, if NULL, don't push anything to it. Add
1777 first argument, use it to optimize skipping over orig stmts that
1778 are known to be before bitpos already. Simplify.
1779 (split_group): Return unsigned int count how many stores are or
1780 would be needed rather than a bool. Add allow_unaligned argument.
1781 Change split_stores argument from reference to pointer, if NULL,
1782 only do a dry run computing how many stores would be produced.
1783 Rewritten algorithm to use both alignment and misalign if
1784 !allow_unaligned and handle bitfield stores with gaps.
1785 (imm_store_chain_info::output_merged_store): Set start_byte_pos
1786 from bitregion_start instead of start. Compute allow_unaligned
1787 here, if true, do 2 split_group dry runs to compute which one
1788 produces fewer stores and prefer aligned if equal. Punt if
1789 new count is bigger or equal than original before emitting any
1790 statements, rather than during that. Remove no longer needed
1791 new_ssa_names tracking. Replace num_stmts with
1792 split_stores.length (). Use 32-bit stack allocated entries
1793 in split_stores auto_vec. Try to reuse original store lhs/rhs1
1794 if possible. Handle bitfields with gaps.
1795 (pass_store_merging::execute): Ignore bitsize == 0 stores.
1796 Compute bitregion_{start,end} for the stores and construct
1797 store_immediate_info with that. Formatting fixes.
1799 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
1802 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
1803 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
1805 2017-10-29 Jim Wilson <wilson@tuliptree.org>
1807 * gimplify.c: Include tm_p.h.
1809 * common.opt (gcoff): Re-add as ignored option.
1810 (gcoff1, gcoff2, gcoff3): Likewise.
1812 * Makefile.in (OBJS): Delete sdbout.o.
1813 (GTFILES): Delete $(srcdir)/sdbout.c.
1814 * debug.h: Delete sdb_debug_hooks.
1815 * final.c: Delete sdbout.h include.
1816 (final_scan_insn): Delete SDB_DEBUG check.
1817 (rest_of_clean_state): Likewise.
1818 * output.h: Delete sdb_begin_function_line.
1821 * toplev.c: Delete sdbout.h include.
1822 (process_options): Delete SDB_DEBUG check.
1823 * tree-core.h (tree_type_common): Delete pointer field of
1825 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
1826 TYPE_SYMTAB_POINTER.
1827 * tree.h (TYPE_SYMTAB_POINTER): Delete.
1828 (TYPE_SYMTAB_IS_POINTER): Delete.
1829 (TYPE_SYMTAB_IS_DIE): Renumber.
1830 * xcoffout.c: Refer to former sdbout.c file.
1831 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
1833 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
1834 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
1835 (Debugging Options): Delete -gcoff.
1836 (-gstabs): Delete SDB reference.
1838 (-gcoff@var{level}): Delete.
1839 * doc/passes.texi (Debugging information output): Delete SDB and
1840 sdbout.c references.
1841 * doc/tm.texi: Regenerate.
1842 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
1843 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
1845 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
1846 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
1847 (SDB_DEBUGGING_INFO): Delete.
1848 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
1849 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
1850 * target.def (output_source_filename): Delete COFF reference.
1852 * common.opt (gcoff): Delete.
1853 (gxcoff+): Update Negative chain.
1854 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
1856 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
1857 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
1858 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
1860 (expand_function_start): Change sdb reference to past tense.
1861 (expand_function_end): Change sdb reference to past tense.
1862 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
1863 * opts.c (debug_type_names): Delete coff entry.
1864 (common_handle_option): Delete OPT_gcoff case.
1865 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
1867 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
1868 * config/cris/cris.h: Delete SDB reference in comment.
1869 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
1870 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
1871 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
1872 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
1874 (ix86_expand_prologue): Likewise.
1875 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
1876 * config/ia64/ia64.h: Likewise.
1877 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
1878 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
1880 * config/mmix/mmix.h: Likewise.
1881 * config/nds32/nds32.c: Likewise.
1882 * config/stormy/storym16.h: Likewise.
1883 * config/visium/visium.h: Likewise.
1884 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
1886 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
1888 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
1889 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
1890 FRAME_POINTER_REGNUM point at high end of local var area.
1892 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
1894 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
1895 Move comment around. Do not reset best_edge for a copiable
1896 destination if the copy would cause a partition change.
1897 (better_edge_p): Remove redundant check.
1899 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
1901 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
1903 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
1905 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
1906 for math functions that have _Float<N> and _Float<N>X variants.
1907 (mathfn_built_in_2): Add support for math functions that have
1908 _Float<N> and _Float<N>X variants.
1909 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
1910 (expand_builtin_mathfn_ternary): Add support for fma with
1911 _Float<N> and _Float<N>X variants.
1912 (expand_builtin): Likewise.
1913 (fold_builtin_3): Likewise.
1914 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
1915 create math function _Float<N> and _Float<N>X variants as external
1917 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
1918 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
1919 the __builtin_ prefix and if not strict ansi, without the prefix.
1920 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
1921 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
1922 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
1923 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
1924 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
1925 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
1926 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
1927 function signatures for fma _Float<N> and _Float<N>X variants.
1928 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
1929 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
1930 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
1931 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
1932 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
1933 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
1934 * gencfn-macros.c (print_case_cfn): Add support for math functions
1935 that have _Float<N> and _Float<N>X variants.
1936 (print_define_operator_list): Likewise.
1937 (fltfn_suffixes): Likewise.
1939 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
1940 for math functions that have _Float<N> and _Float<N>X variants.
1941 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
1942 and _Float<N>X variants.
1943 (COPYSIGN): Likewise.
1946 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
1947 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
1949 (integer_valued_read_call_p): Likewise.
1950 * fold-const-call.c (fold_const_call_ss): Likewise.
1951 (fold_const_call_sss): Add support for copysign, fmin, and fmax
1952 _Float<N> and _Float<N>X variants.
1953 (fold_const_call_ssss): Add support for fma _Float<N> and
1954 _Float<N>X variants.
1955 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
1956 support for copysign and fma _Float<N> and _Float<N>X variants.
1957 (backprop::process_builtin_call_use): Likewise.
1958 * tree-call-cdce.c (can_test_argument_range); Add support for
1959 sqrt _Float<N> and _Float<N>X variants.
1960 (edom_only_function): Likewise.
1961 (get_no_error_domain): Likewise.
1962 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
1963 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
1964 copysign _Float<N> and _Float<N>X variants.
1965 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
1966 handled by machine independent code.
1967 (FMAF128): Likewise.
1968 * doc/cpp.texi (Common Predefined Macros): Document defining
1969 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
1970 fma _Float<N> and _Float<N>X variants.
1972 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
1975 * config/i386/i386-modes.def (CCFPU): Remove definition.
1976 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
1977 (ix86_cc_modes_compatible): Ditto.
1978 (ix86_expand_carry_flag_compare): Ditto.
1979 (ix86_expand_int_movcc): Ditto.
1980 (ix86_expand_int_addcc): Ditto.
1981 (ix86_reverse_condition): Ditto.
1982 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
1983 Return true/false for unordered/ordered fp comparisons.
1984 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
1985 (ix86_prepare_fp_compare_args): Update for rename.
1986 (ix86_expand_fp_compare): Update for rename. Generate unordered
1987 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
1988 (ix86_expand_sse_compare_and_jump): Ditto.
1989 * config/i386/predicates.md (fcmov_comparison_operator):
1990 Remove CCFPU mode handling.
1991 (ix86_comparison_operator): Ditto.
1992 (ix86_carry_flag_operator): Ditto.
1993 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
1994 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
1995 (*cmpu<mode>_cc_i387): Ditto.
1996 (FPCMP): Remove mode iterator.
1997 (unord): Remove mode attribute.
1998 (unord_subst): New define_subst transformation
1999 (unord): New define_subst attribute.
2001 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
2002 (*cmpi<unord>xf_i387): Ditto.
2003 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
2004 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
2005 using unord_subst transformation.
2006 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
2007 (round_saeonly): Also handle CCFP mode.
2008 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
2009 Remove UNSPEC_SAHF unspec handling.
2011 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
2013 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
2015 2017-10-27 Jeff Law <law@redhat.com>
2017 * gimple-ssa-sprintf.c: Include domwalk.h.
2018 (class sprintf_dom_walker): New class, derived from dom_walker.
2019 (sprintf_dom_walker::before_dom_children): New function.
2020 (struct call_info): Moved into sprintf_dom_walker class
2021 (compute_formath_length, handle_gimple_call): Likewise.
2022 (sprintf_length::execute): Call the dominator walker rather
2023 than walking the statements.
2025 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
2026 gimple statement locations.
2027 (check_array_bounds): Corresponding changes. Get the statement's
2028 location directly from wi->stmt.
2030 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
2033 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
2035 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
2037 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
2038 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
2040 2017-10-27 Jakub Jelinek <jakub@redhat.com>
2043 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
2044 * config/i386/i386.c (maybe_get_pool_constant): Removed.
2045 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
2046 maybe_get_pool_constant.
2047 * config/i386/predicates.md (zero_extended_scalar_load_operand):
2050 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2052 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
2053 2.26 caveat. Update gas and gld versions.
2054 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
2057 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2059 * cgraph.h (set_malloc_flag): Declare.
2060 * cgraph.c (set_malloc_flag_1): New function.
2061 (set_malloc_flag): Likewise.
2062 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
2063 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
2065 (read_ipa_call_summary): Add support for reading is_return_callee.
2066 (write_ipa_call_summary): Stream is_return_callee.
2067 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
2068 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
2069 ipa-prop.h, ipa-fnsummary.h.
2070 (pure_const_names): Change to static.
2071 (malloc_state_e): Define.
2072 (malloc_state_names): Define.
2073 (funct_state_d): Add field malloc_state.
2074 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
2075 (check_retval_uses): New function.
2076 (malloc_candidate_p): Likewise.
2077 (analyze_function): Add support for malloc attribute.
2078 (pure_const_write_summary): Stream malloc_state.
2079 (pure_const_read_summary): Add support for reading malloc_state.
2080 (dump_malloc_lattice): New function.
2081 (propagate_malloc): New function.
2082 (warn_function_malloc): New function.
2083 (ipa_pure_const::execute): Call propagate_malloc and
2084 ipa_free_fn_summary.
2085 (pass_local_pure_const::execute): Add support for malloc attribute.
2086 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
2087 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
2089 2017-10-27 Martin Liska <mliska@suse.cz>
2091 PR gcov-profile/82457
2092 * doc/invoke.texi: Document that one needs a non-strict ISO mode
2093 for fork-like functions to be properly instrumented.
2095 2017-10-27 Richard Biener <rguenther@suse.de>
2098 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
2099 info when we redirected EH.
2101 2017-10-26 Michael Collison <michael.collison@arm.com>
2103 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
2105 (<optab>_trunchf<GPI:mode>2: New pattern.
2106 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
2107 * config/aarch64/iterators.md (wv): New mode attribute.
2108 (vf, VF): New mode attributes.
2109 (vgp, VGP): New mode attributes.
2110 (s): Update attribute with SImode and DImode prefixes.
2112 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
2114 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
2115 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
2116 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
2117 (nios2_option_overide): Initialize it. Don't allow R0-relative
2118 addressing with PIC.
2119 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
2120 (nios2_symbolic_constant_p): Likewise.
2121 (nios2_legitimate_address_p): Likewise.
2122 (nios2_r0rel_section_name_p): New.
2123 (nios2_symbol_ref_in_r0rel_data_p): New.
2124 (nios2_emit_move_sequence): Handle r0rel_constant_p.
2125 (r0rel_constant_p): New.
2126 (nios2_print_operand_address): Handle r0rel_constant_p.
2127 (nios2_cdx_narrow_form_p): Likewise.
2128 * config/nios2/nios2.opt (mr0rel-sec=): New option.
2129 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
2130 (Nios II Options): Document -mr0rel-sec.
2132 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
2134 * config/nios2/nios2.c: Include xregex.h.
2135 (nios2_gprel_sec_regex): New.
2136 (nios2_option_overide): Initialize it. Don't allow GP-relative
2137 addressing with PIC.
2138 (nios2_small_section_name_p): Check for regex match.
2139 * config/nios2/nios2.opt (mgprel-sec=): New option.
2140 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
2141 (Nios II Options): Document -mgprel-sec.
2143 2017-10-26 Jim Wilson <wilson@tuliptree.org>
2145 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
2147 2017-10-26 Tom de Vries <tom@codesourcery.com>
2149 PR tree-optimization/82707
2150 * gimple.c (gimple_copy): Fix unsharing of
2151 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
2153 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
2155 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
2156 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
2157 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
2158 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
2159 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
2160 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
2161 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
2162 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
2163 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
2164 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
2165 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
2166 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
2167 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
2168 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
2169 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
2170 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
2171 _mm512_mask_cmpunord_ps_mask): New intrinsics.
2173 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
2175 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
2177 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
2178 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
2179 warning to rs6000.c. Remove the Undocumented flag, since it has
2181 (-mabi=ibmlongdouble): Likewise.
2182 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
2183 already set, set the default format for long double.
2184 (rs6000_debug_reg_global): Print whether long double is IBM or
2186 (rs6000_option_override_internal): Rework setting long double
2187 format. Only warn if the user is changing the long double default
2188 and they did not use -Wno-psabi.
2189 * doc/invoke.texi (PowerPC options): Update the documentation for
2190 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
2192 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2193 Alan Hayward <alan.hayward@arm.com>
2194 David Sherwood <david.sherwood@arm.com>
2196 * rtl.h (wider_subreg_mode): New function.
2197 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
2198 rather than an unsigned int *.
2199 * ira-color.c (regno_max_ref_width): Replace with...
2200 (regno_max_ref_mode): ...this new variable.
2201 (coalesced_pseudo_reg_slot_compare): Update accordingly.
2202 Use wider_subreg_mode.
2203 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
2204 rather than an unsigned int *.
2205 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
2206 (process_alt_operands): Likewise.
2207 (invariant_p): Likewise.
2208 * lra-spills.c (assign_mem_slot): Likewise.
2209 (add_pseudo_to_slot): Likewise.
2210 * lra.c (collect_non_operand_hard_regs): Likewise.
2211 (add_regs_to_insn_regno_info): Likewise.
2212 * reload1.c (regno_max_ref_width): Replace with...
2213 (regno_max_ref_mode): ...this new variable.
2214 (reload): Update accordingly. Update call to
2215 ira_sort_regnos_for_alter_reg.
2216 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
2217 (init_eliminable_invariants): Update to use regno_max_ref_mode.
2218 (scan_paradoxical_subregs): Likewise.
2220 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2222 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
2223 (aarch64_frame): Add emit_frame_chain boolean.
2224 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2225 Move eh_return case to aarch64_layout_frame.
2226 (aarch64_layout_frame): Initialize emit_frame_chain.
2227 (aarch64_expand_prologue): Use emit_frame_chain.
2229 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2231 * config/aarch64/aarch64.c (aarch64_layout_frame):
2232 Ensure LR is always stored at the bottom of the callee-saves.
2233 Remove rarely used frame layout which saves callee-saves at top of
2234 frame, so the store of LR can be used as a valid probe in all cases.
2236 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2238 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
2239 Improve unaligned TImode/TFmode base/offset split.
2241 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2242 Alan Hayward <alan.hayward@arm.com>
2243 David Sherwood <david.sherwood@arm.com>
2245 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
2246 * combine.c (find_single_use_1): Likewise.
2247 (expand_field_assignment): Likewise.
2248 (move_deaths): Likewise.
2249 * lra-constraints.c (simplify_operand_subreg): Likewise.
2250 (curr_insn_transform): Likewise.
2251 * lra.c (collect_non_operand_hard_regs): Likewise.
2252 (add_regs_to_insn_regno_info): Likewise.
2253 * rtlanal.c (reg_referenced_p): Likewise.
2254 (covers_regno_no_parallel_p): Likewise.
2256 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2258 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
2259 Don't print any bits outside the precision of the value.
2260 * wide-int.cc (test_printing): Add some new tests.
2262 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2264 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
2265 supports -xbrace_comment option.
2266 * configure: Regenerate.
2267 * config.in: Regenerate.
2268 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
2269 (ASM_CPU_SPEC): Use it.
2271 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2273 * target.def (static_rtx_alignment): New hook.
2274 * targhooks.h (default_static_rtx_alignment): Declare.
2275 * targhooks.c (default_static_rtx_alignment): New function.
2276 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
2277 * doc/tm.texi: Regenerate.
2278 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
2279 instead of targetm.constant_alignment. Remove call to
2281 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2282 (cris_preferred_mininum_alignment): New function, split out from...
2283 (cris_constant_alignment): ...here.
2284 (cris_static_rtx_alignment): New function.
2285 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
2287 (ix86_constant_alignment): ...here.
2288 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2289 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2290 (mmix_static_rtx_alignment): New function.
2291 * config/spu/spu.c (spu_static_rtx_alignment): New function.
2292 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2294 2017-10-26 Tamar Christina <tamar.christina@arm.com>
2297 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
2298 Add flag_trapping_math and flag_fp_int_builtin_inexact.
2300 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
2302 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
2303 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
2304 mark as a sign-extending load.
2305 (local_pic_load_u): Define.
2307 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
2310 * fold-const.c (operand_equal_for_comparison_p): Also return true
2311 if ARG0 is a simple variant of ARG1 with narrower precision.
2312 (fold_ternary_loc): Always pass unstripped operands to the predicate.
2314 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
2316 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
2318 * i386.h (processor_costs): Add gather_static, gather_per_elt,
2319 scatter_static, scatter_per_elt.
2320 * x86-tune-costs.h: Add new cost entries.
2322 2017-10-25 Richard Biener <rguenther@suse.de>
2324 * tree-ssa-sccvn.h (vn_eliminate): Declare.
2325 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
2326 class pass_fre): Move to ...
2327 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
2328 class pass_fre): ... here and adjust for statistics.
2330 2017-10-25 Jakub Jelinek <jakub@redhat.com>
2333 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
2334 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
2335 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
2336 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
2339 2017-10-25 Richard Biener <rguenther@suse.de>
2341 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
2342 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
2343 eliminate_push_avail, eliminate_insert): Move inside...
2344 (class eliminate_dom_walker): ... this class in preparation
2346 (fini_eliminate): Remove by merging with ...
2347 (eliminate): ... this function. Adjust for class changes.
2348 (pass_pre::execute): Remove fini_eliminate call.
2349 (pass_fre::execute): Likewise.
2351 2017-10-24 Jakub Jelinek <jakub@redhat.com>
2354 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
2355 (VPERMI2, VPERMI2I): New mode iterators.
2356 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
2357 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
2359 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
2360 mode iterator. Remove 3 old define_insn patterns.
2361 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
2362 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
2363 VPERMI2 mode iterator, remove the other two expanders.
2364 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
2365 to use VPERMI2 mode iterator, add another alternative for vpermi2*
2366 instructions, remove the other two patterns.
2367 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
2368 mode iterator, remove the other two patterns.
2369 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
2370 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
2371 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
2372 and adjust argument order accordingly.
2373 (ix86_expand_vec_perm): Adjust caller.
2374 (expand_vec_perm_1): Likewise.
2375 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
2376 (expand_vec_perm_vpermt2_vpshub2): ... this.
2377 (ix86_expand_vec_perm_const_1): Adjust caller.
2378 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
2381 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
2382 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
2383 (vec_shl_<mode>): Remove unused expander.
2384 (avx512bw_<shift_insn><mode>3): New define_insn.
2385 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
2386 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
2388 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
2391 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
2394 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
2396 PR rtl-optimization/82396
2397 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
2398 (autopref_multipass_init): Simplify initialization.
2399 (autopref_rank_data): Simplify sort order.
2400 * gcc/sched-int.h (autopref_multipass_data_): Remove
2401 multi_mem_insn_p, min_offset and max_offset.
2403 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
2406 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2407 Check special value of flag_omit_frame_pointer.
2408 (aarch64_can_eliminate): Likewise.
2409 (aarch64_override_options_after_change_1): Simplify handling of
2410 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
2412 2017-10-24 Richard Biener <rguenther@suse.de>
2414 PR tree-optimization/82697
2415 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
2416 zero for conditional load and unconditional store.
2418 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
2420 * doc/install.texi: Document bootstrap-cet.
2422 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
2425 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
2426 ENDBR instruction at function entrance if function is only
2429 2017-10-24 Jakub Jelinek <jakub@redhat.com>
2432 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
2433 patterns to better describe from which operation the CF is computed.
2434 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
2435 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
2436 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
2437 is 0, use _0 suffixed expanders instead of emitting a comparison
2440 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
2442 * config/i386/i386.md(*movsf_internal, *movdf_internal):
2443 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
2445 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
2448 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
2449 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
2450 * loop-iv.c (iv_get_reaching_def): Likewise.
2451 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
2452 variable is promoted and the partition contains undefined values.
2454 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2456 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
2458 (nios2_address_cost): Define.
2459 (nios2_legitimize_address): Recognize (exp + constant) directly.
2460 (TARGET_ADDRESS_COST): Define.
2462 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2464 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
2465 (nios2_symbolic_memory_operand_p): Declare.
2466 (nios2_split_large_constant): Declare.
2467 (nios2_split_symbolic_memory_operand): Declare.
2468 * config/nios2/nios2.c: Adjust includes.
2469 (nios2_symbolic_constant_allowed): New.
2470 (nios2_symbolic_constant_p): New.
2471 (nios2_plus_symbolic_constant_p): New.
2472 (nios2_valid_addr_expr_p): Recognize addresses involving
2474 (nios2_legitimate_address_p): Likewise, also LO_SUM.
2475 (nios2_symbolic_memory_operand_p): New.
2476 (nios2_large_constant_p): New.
2477 (nios2_split_large_constant): New.
2478 (nios2_split_plus_large_constant): New.
2479 (nios2_split_symbolic_memory_operand): New.
2480 (nios2_legitimize_address): Code refactoring. Handle addresses
2481 involving symbolic constants.
2482 (nios2_emit_move_sequence): Likewise.
2483 (nios2_print_operand): Improve error output.
2484 (nios2_print_operand_address): Handle LO_SUM.
2485 (nios2_cdx_narrow_form_p): Likewise.
2486 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
2487 operands involving symbolic constants.
2488 (movhi_internal, movsi_internal): Likewise.
2489 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
2490 (extendhisi2, extendqi<mode>2): Likewise.
2492 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2494 * tree-pass.h (PROP_rtl_split_insns): Define.
2495 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
2497 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2499 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
2501 2017-10-23 Jakub Jelinek <jakub@redhat.com>
2504 * target.def (const_not_ok_for_debug_p): Default to
2505 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
2506 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
2507 * targhooks.c (default_const_not_ok_for_debug_p): New function.
2508 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
2509 which targetm.const_not_ok_for_debug_p returned true.
2510 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
2512 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
2514 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
2515 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
2516 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
2518 (ix86_const_not_ok_for_debug_p): New function.
2519 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
2520 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
2522 2017-10-23 David Malcolm <dmalcolm@redhat.com>
2525 * system.h: Conditionally include "unique-ptr.h" if
2526 INCLUDE_UNIQUE_PTR is defined.
2527 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
2528 of defining INCLUDE_UNIQUE_PTR before including "system.h".
2530 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
2532 * config/rl78/rl78.md: New define_expand "subdi3".
2534 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
2537 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
2538 DF_REF_INSN if DF_REF_INSN_INFO is false.
2540 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
2542 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
2543 xmm_move instead of sse_move.
2544 (sse_store_index): New function.
2545 (ix86_register_move_cost): Be more sensible about mismatch stall;
2546 model AVX moves correctly; make difference between sse->integer and
2548 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
2549 moves; make difference between SSE and AVX.
2550 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
2551 and zmm_move. Increase size of sse load and store tables;
2552 add unaligned load and store tables; add ssemmx_to_integer.
2553 * x86-tune-costs.h: Update all entries according to real
2554 move latencies from Agner Fog's manual and chip documentation.
2556 2017-10-23 Jakub Jelinek <jakub@redhat.com>
2559 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
2560 * config/i386/constraints.md (Wf): New constraint.
2561 * config/i386/i386.md (UNSPEC_SBB): New unspec.
2562 (cmp<dwi>_doubleword): Removed.
2563 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
2564 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
2565 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
2566 expand with cmp<dwi>_doubleword. For LTU and GEU use
2567 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
2569 * common.opt (gcolumn-info): Enable by default.
2570 * doc/invoke.texi (gcolumn-info): Document new default.
2572 2017-10-23 Richard Biener <rguenther@suse.de>
2574 PR tree-optimization/82672
2575 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
2576 Fold the stmt if we propagated into it.
2578 2017-10-23 Richard Biener <rguenther@suse.de>
2580 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
2581 (bitmap_remove_expr_from_set): ... this. All callers call this
2582 for non-constant values.
2583 (bitmap_set_subtract): Rename to...
2584 (bitmap_set_subtract_expressions): ... this. Adjust and
2586 (bitmap_set_contains_value): Remove superfluous check.
2587 (bitmap_set_replace_value): Inline into single caller ...
2588 (bitmap_value_replace_in_set): ... here and simplify.
2589 (dependent_clean): Merge into ...
2590 (clean): ... this using an overload. Adjust.
2591 (prune_clobbered_mems): Adjust.
2592 (compute_antic_aux): Likewise.
2593 (compute_partial_antic_aux): Likewise.
2595 2017-10-23 Richard Biener <rguenther@suse.de>
2597 PR tree-optimization/82129
2599 2017-08-01 Richard Biener <rguenther@suse.de>
2601 PR tree-optimization/81181
2602 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
2603 (compute_antic): ... end of iteration here.
2605 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2607 * target.def (starting_frame_offset): New hook.
2608 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
2609 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
2610 * doc/tm.texi.in: Regenerate.
2611 * hooks.h (hook_hwi_void_0): Declare.
2612 * hooks.c (hook_hwi_void_0): New function.
2613 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
2614 STARTING_FRAME_OFFSET.
2615 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
2616 * reload1.c (reload): Likewise.
2617 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
2618 instead of STARTING_FRAME_OFFSET.
2619 * function.c (try_fit_stack_local): Likewise.
2620 (assign_stack_local_1): Likewise
2621 (instantiate_virtual_regs): Likewise.
2622 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
2623 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
2624 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
2625 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
2626 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
2627 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
2628 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
2629 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
2630 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
2631 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
2632 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
2633 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
2634 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
2635 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
2636 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
2637 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
2638 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
2639 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
2640 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
2641 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
2642 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
2643 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
2644 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
2645 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
2646 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
2647 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
2648 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
2649 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
2650 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
2651 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
2652 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
2653 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
2654 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
2655 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
2656 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
2657 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
2658 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
2659 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
2660 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
2661 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
2662 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
2663 * config/avr/avr.c (avr_starting_frame_offset): Make static and
2664 return a HOST_WIDE_INT.
2665 (avr_builtin_setjmp_frame_value): Use it instead of
2666 STARTING_FRAME_OFFSET.
2667 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2668 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
2669 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
2671 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2672 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
2673 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
2674 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2675 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
2676 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
2677 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2678 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
2679 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
2680 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2681 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
2682 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
2684 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2685 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
2686 * config/mips/mips.c (mips_compute_frame_info): Refer to
2687 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
2688 (mips_starting_frame_offset): New function.
2689 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2690 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
2691 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
2692 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
2693 and return a HOST_WIDE_INT.
2694 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2695 (mmix_initial_elimination_offset): Refer to
2696 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
2697 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
2698 * config/pa/pa.c (pa_starting_frame_offset): New function.
2699 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
2700 (pa_expand_prologue): Likewise.
2701 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2702 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
2703 !FRAME_GROWS_DOWNWARD handling to...
2704 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2705 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
2706 !FRAME_GROWS_DOWNWARD handling to...
2707 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2708 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
2709 !FRAME_GROWS_DOWNWARD handling to...
2710 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2711 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
2713 (rs6000_starting_frame_offset): New function.
2714 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
2715 !FRAME_GROWS_DOWNWARD handling to...
2716 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2717 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
2718 !FRAME_GROWS_DOWNWARD handling to...
2719 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2720 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
2721 !FRAME_GROWS_DOWNWARD handling to...
2722 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2723 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
2724 (rs6000_starting_frame_offset): New function.
2725 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
2726 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
2727 * config/vax/vax.c (vax_starting_frame_offset): New function.
2728 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
2729 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2730 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
2731 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
2732 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2733 * system.h (STARTING_FRAME_OFFSET): Poison.
2735 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2737 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
2738 SCALAR_TYPE_MODE instead of TYPE_MODE.
2740 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2741 Alan Hayward <alan.hayward@arm.com>
2742 David Sherwood <david.sherwood@arm.com>
2744 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
2746 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2747 Alan Hayward <alan.hayward@arm.com>
2748 David Sherwood <david.sherwood@arm.com>
2750 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
2752 2017-10-23 Richard Biener <rguenther@suse.de>
2754 PR tree-optimization/82129
2755 * tree-ssa-pre.c (bitmap_set_and): Remove.
2756 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
2757 canonicalizing expressions in the set to those with lowest
2758 ID rather than taking that from the first edge.
2760 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2762 * combine.c (rtx_equal_for_field_assignment_p): Use
2763 byte_lowpart_offset.
2765 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2766 Alan Hayward <alan.hayward@arm.com>
2767 David Sherwood <david.sherwood@arm.com>
2769 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
2770 to a SUBREG_PROMOTED_VAR.
2772 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2773 Alan Hayward <alan.hayward@arm.com>
2774 David Sherwood <david.sherwood@arm.com>
2776 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
2777 (expand_debug_source_expr): Likewise.
2778 * combine.c (combine_simplify_rtx): Likewise.
2779 * cse.c (fold_rtx): Likewise.
2780 * optabs.c (expand_float): Likewise.
2781 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2782 (simplify_binary_operation_1): Likewise.
2784 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2785 Alan Hayward <alan.hayward@arm.com>
2786 David Sherwood <david.sherwood@arm.com>
2788 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
2789 (record_promoted_value): Likewise.
2790 * expr.c (expand_expr_real_2): Likewise.
2791 * ree.c (update_reg_equal_equiv_notes): Likewise.
2792 (combine_set_extension): Likewise.
2793 * rtlanal.c (low_bitmask_len): Likewise.
2794 * simplify-rtx.c (neg_const_int): Likewise.
2795 (simplify_binary_operation_1): Likewise.
2797 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2798 Alan Hayward <alan.hayward@arm.com>
2799 David Sherwood <david.sherwood@arm.com>
2801 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
2802 * regcprop.c (maybe_mode_change): Likewise.
2803 * reload1.c (alter_reg): Likewise.
2805 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2807 * inchash.h (inchash::hash::add_wide_int): New function.
2808 * lto-streamer-out.c (hash_tree): Use it.
2810 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2812 * inchash.h (inchash::hash::add_wide_int): Rename to...
2813 (inchash::hash::add_hwi): ...this.
2814 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
2815 (polymorphic_call_target_hasher::hash): Likewise.
2816 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
2817 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
2818 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
2819 * lto-streamer-out.c (hash_tree): Likewise.
2820 * optc-save-gen.awk: Likewise.
2821 * tree.c (add_expr): Likewise.
2823 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
2826 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
2827 for ordered inequality comparisons even with TARGET_IEEE_FP.
2829 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
2832 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
2833 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
2834 Expand with cmp<dwi>_doubleword.
2836 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2838 * extend.texi: Add x86 specific to 'nocf_check' attribute.
2839 List CET intrinsics.
2840 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
2841 specific to -fcf-protection option.
2843 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2845 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
2846 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
2847 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
2848 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
2849 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
2850 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
2851 (extra_objs): Add cet.o for Linux/x86 targets.
2852 (tmake_file): Add i386/t-cet for Linux/x86 targets.
2853 * config/i386/cet.c: New file.
2854 * config/i386/cetintrin.h: Likewise.
2855 * config/i386/t-cet: Likewise.
2856 * config/i386/cpuid.h (bit_SHSTK): New.
2857 (bit_IBT): Likewise.
2858 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
2859 pass IBT and SHSTK bits.
2860 * config/i386/i386-builtin-types.def
2861 (VOID_FTYPE_UNSIGNED_PVOID): New.
2862 (VOID_FTYPE_UINT64_PVOID): Likewise.
2863 * config/i386/i386-builtin.def: Add CET intrinsics.
2864 * config/i386/i386-c.c (ix86_target_macros_internal): Add
2865 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
2866 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
2867 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
2869 * config/i386/i386.c (rest_of_insert_endbranch): New.
2870 (pass_data_insert_endbranch): Likewise.
2871 (pass_insert_endbranch): Likewise.
2872 (make_pass_insert_endbranch): Likewise.
2873 (ix86_notrack_prefixed_insn_p): Likewise.
2874 (ix86_target_string): Add -mibt, -mshstk flags.
2875 (ix86_option_override_internal): Add flag_cf_protection
2877 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
2878 (ix86_print_operand): Add 'notrack' prefix output.
2879 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
2880 (ix86_expand_builtin): Expand CET intrinsics.
2881 (x86_output_mi_thunk): Add 'endbranch' instruction.
2882 * config/i386/i386.h (TARGET_IBT): New.
2883 (TARGET_IBT_P): Likewise.
2884 (TARGET_SHSTK): Likewise.
2885 (TARGET_SHSTK_P): Likewise.
2886 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
2887 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
2888 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
2889 (builtin_setjmp_setup): New pattern.
2890 (builtin_longjmp): Likewise.
2891 (rdssp<mode>): Likewise.
2892 (incssp<mode>): Likewise.
2893 (saveprevssp): Likewise.
2894 (rstorssp): Likewise.
2895 (wrss<mode>): Likewise.
2896 (wruss<mode>): Likewise.
2897 (setssbsy): Likewise.
2898 (clrssbsy): Likewise.
2899 (nop_endbr): Likewise.
2900 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
2902 * config/i386/immintrin.h: Include <cetintrin.h>.
2903 * config/i386/linux-common.h
2904 (file_end_indicate_exec_stack_and_cet): New prototype.
2905 (TARGET_ASM_FILE_END): New.
2907 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
2909 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
2910 latencies instead of having separate table; make difference between
2911 integer and float costs.
2912 * i386.h (processor_costs): Remove scalar_stmt_cost,
2913 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
2914 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
2916 * x86-tune-costs.h: Remove entries which has been removed in
2917 procesor_costs from all tables; make cond_taken_branch_cost
2918 and cond_not_taken_branch_cost COST_N_INSNS based.
2920 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
2922 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
2924 2017-10-20 Jakub Jelinek <jakub@redhat.com>
2926 * config/i386/i386.md (isa): Remove fma_avx512f.
2927 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
2928 <avx512>_fmadd_<mode>_mask3<round_name>,
2929 <avx512>_fmsub_<mode>_mask<round_name>,
2930 <avx512>_fmsub_<mode>_mask3<round_name>,
2931 <avx512>_fnmadd_<mode>_mask<round_name>,
2932 <avx512>_fnmadd_<mode>_mask3<round_name>,
2933 <avx512>_fnmsub_<mode>_mask<round_name>,
2934 <avx512>_fnmsub_<mode>_mask3<round_name>,
2935 <avx512>_fmaddsub_<mode>_mask<round_name>,
2936 <avx512>_fmaddsub_<mode>_mask3<round_name>,
2937 <avx512>_fmsubadd_<mode>_mask<round_name>,
2938 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
2939 (*vec_widen_umult_even_v16si<mask_name>,
2940 *vec_widen_smult_even_v16si<mask_name>): Likewise.
2941 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
2943 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2945 * extend.texi: Add 'nocf_check' documentation.
2946 * gimple.texi: Add second parameter to
2947 gimple_build_call_from_tree.
2948 * invoke.texi: Add -fcf-protection documentation.
2949 * rtl.texi: Add REG_CALL_NOTRACK documenation.
2951 2017-10-20 Richard Biener <rguenther@suse.de>
2953 PR tree-optimization/82473
2954 * tree-vect-loop.c (vectorizable_reduction): Properly get at
2955 the largest input type.
2957 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2959 * c-attribs.c (handle_nocf_check_attribute): New function.
2960 (c_common_attribute_table): Add 'nocf_check' handling.
2961 * gimple-parser.c: Add second argument NULL to
2962 gimple_build_call_from_tree.
2963 * attrib.c (comp_type_attributes): Check nocf_check attribute.
2964 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
2966 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
2967 * common.opt: Add fcf-protection flag.
2968 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
2969 * flag-types.h: Add enum cf_protection_level.
2970 * gimple.c (gimple_build_call_from_tree): Add second parameter.
2971 Add 'nocf_check' attribute propagation to gimple call.
2972 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
2973 (gimple_build_call_from_tree): Update prototype.
2974 (gimple_call_nocf_check_p): New function.
2975 (gimple_call_set_nocf_check): Likewise.
2976 * gimplify.c: Add second argument to gimple_build_call_from_tree.
2977 * ipa-icf.c: Add nocf_check attribute in statement hash.
2978 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
2979 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
2980 * toplev.c (process_options): Add flag_cf_protection handling.
2982 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
2984 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
2986 2017-10-20 Richard Biener <rguenther@suse.de>
2988 PR tree-optimization/82603
2989 * tree-if-conv.c (predicate_mem_writes): Make sure to only
2990 remove false predicated stores.
2992 2017-10-20 Richard Biener <rguenther@suse.de>
2994 * graphite-isl-ast-to-gimple.c
2995 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
2996 Remove return value and simplify, dump copied stmt after lhs
2998 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
2999 Reduce dump verbosity.
3000 (gsi_insert_earliest): Likewise.
3001 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
3002 * graphite.c (print_global_statistics): Adjust dumping.
3003 (print_graphite_scop_statistics): Likewise.
3004 (print_graphite_statistics): Do not dump loops here.
3005 (graphite_transform_loops): But here.
3007 2017-10-20 Nicolas Roche <roche@adacore.com>
3009 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
3010 * configure: Regenerate.
3012 2017-10-20 Jakub Jelinek <jakub@redhat.com>
3015 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
3016 functions when optimizing replace GIMPLE_RETURN stmts with
3017 calls to __builtin_unreachable ().
3020 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
3021 for -fsanitize=thread link of executables.
3022 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
3023 link of executables.
3026 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
3028 (<shift_insn><mode>3<mask_name>): Change the last of the 3
3029 define_insns for logical vector shifts to use VI248_AVX512BW
3030 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
3031 condition, useless isa and prefix attributes. Change the first
3032 2 of these define_insns to ...
3033 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
3034 define_insn for avx512vl.
3035 (<shift_insn><mode>3): ... and this, new define_insn without
3036 masking for non-avx512vl.
3039 * config/i386/sse.md (*andnot<mode>3,
3040 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
3041 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
3042 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
3043 not applied use empty suffix even for TARGET_AVX512VL.
3044 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
3045 is applied, supply evex,evex or evex,evex,evex instead of just
3048 2017-10-20 Julia Koval <julia.koval@intel.com>
3050 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
3051 (OPTION_MASK_ISA_GFNI_UNSET): New.
3052 (ix86_handle_option): Handle OPT_mgfni.
3053 * config/i386/cpuid.h (bit_GFNI): New.
3054 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
3055 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
3056 * config/i386/i386.c (ix86_target_string): Add -mgfni.
3057 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
3058 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
3059 * config/i386/i386.opt: Add mgfni.
3061 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
3063 * config/msp430/msp430.c (msp430_option_override): Disable
3064 -fdelete-null-pointer-checks.
3065 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
3067 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3069 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
3070 of x87 and SSE instructions.
3072 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3074 * asan.c (create_cond_insert_point): Do not update edge count.
3075 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
3076 (afdo_propagate_circuit): Likewise.
3077 (afdo_calculate_branch_prob): Likewise.
3078 (afdo_annotate_cfg): Likewise.
3079 * basic-block.h (struct edge_def): Remove count.
3080 (edge_def::count): New accessor.
3081 * bb-reorder.c (rotate_loop): Update.
3082 (find_traces_1_round): Update.
3083 (connect_traces): Update.
3084 (sanitize_hot_paths): Update.
3085 * cfg.c (unchecked_make_edge): Update.
3086 (make_single_succ_edge): Update.
3087 (check_bb_profile): Update.
3088 (dump_edge_info): Update.
3089 (update_bb_profile_for_threading): Update.
3090 (scale_bbs_frequencies_int): Update.
3091 (scale_bbs_frequencies_gcov_type): Update.
3092 (scale_bbs_frequencies_profile_count): Update.
3093 (scale_bbs_frequencies): Update.
3094 * cfganal.c (connect_infinite_loops_to_exit): Update.
3095 * cfgbuild.c (compute_outgoing_frequencies): Update.
3096 (find_many_sub_basic_blocks): Update.
3097 * cfgcleanup.c (try_forward_edges): Update.
3098 (try_crossjump_to_edge): Update
3099 * cfgexpand.c (expand_gimple_cond): Update
3100 (expand_gimple_tailcall): Update
3101 (construct_exit_block): Update
3102 * cfghooks.c (verify_flow_info): Update
3103 (redirect_edge_succ_nodup): Update
3104 (split_edge): Update
3105 (make_forwarder_block): Update
3106 (duplicate_block): Update
3107 (account_profile_record): Update
3108 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
3109 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
3110 * cfgloopmanip.c (scale_loop_profile): Update.
3112 (lv_adjust_loop_entry_edge): Update.
3113 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
3114 (force_nonfallthru_and_redirect): Update.
3115 (purge_dead_edges): Update.
3116 (rtl_flow_call_edges_add): Update.
3117 * cgraphunit.c (init_lowered_empty_function): Update.
3118 (cgraph_node::expand_thunk): Update.
3119 * gimple-pretty-print.c (dump_probability): Update.
3120 (dump_edge_probability): Update.
3121 * gimple-ssa-isolate-paths.c (isolate_path): Update.
3122 * haifa-sched.c (sched_create_recovery_edges): Update.
3123 * hsa-gen.c (convert_switch_statements): Update.
3124 * ifcvt.c (dead_or_predicable): Update.
3125 * ipa-inline-transform.c (inline_transform): Update.
3126 * ipa-split.c (split_function): Update.
3127 * ipa-utils.c (ipa_merge_profiles): Update.
3128 * loop-doloop.c (add_test): Update.
3129 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
3130 * lto-streamer-in.c (input_cfg): Update.
3131 (input_function): Update.
3132 * lto-streamer-out.c (output_cfg): Update.
3133 * modulo-sched.c (sms_schedule): Update.
3134 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
3135 * predict.c (maybe_hot_edge_p): Update.
3136 (unlikely_executed_edge_p): Update.
3137 (probably_never_executed_edge_p): Update.
3138 (dump_prediction): Update.
3139 (drop_profile): Update.
3140 (propagate_unlikely_bbs_forward): Update.
3141 (determine_unlikely_bbs): Update.
3142 (force_edge_cold): Update.
3143 * profile.c (compute_branch_probabilities): Update.
3144 * reg-stack.c (better_edge): Update.
3145 * shrink-wrap.c (handle_simple_exit): Update.
3146 * tracer.c (better_p): Update.
3147 * trans-mem.c (expand_transaction): Update.
3148 (split_bb_make_tm_edge): Update.
3149 * tree-call-cdce.c: Update.
3150 * tree-cfg.c (gimple_find_sub_bbs): Update.
3151 (gimple_split_edge): Update.
3152 (gimple_duplicate_sese_region): Update.
3153 (gimple_duplicate_sese_tail): Update.
3154 (gimple_flow_call_edges_add): Update.
3155 (insert_cond_bb): Update.
3156 (execute_fixup_cfg): Update.
3157 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
3158 * tree-complex.c (expand_complex_div_wide): Update.
3159 * tree-eh.c (lower_resx): Update.
3160 (unsplit_eh): Update.
3161 (cleanup_empty_eh_move_lp): Update.
3162 * tree-inline.c (copy_edges_for_bb): Update.
3163 (freqs_to_counts): Update.
3164 (copy_cfg_body): Update.
3165 * tree-ssa-dce.c (remove_dead_stmt): Update.
3166 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
3167 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
3168 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
3169 (unloop_loops): Update.
3170 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
3171 * tree-ssa-loop-split.c (connect_loops): Update.
3172 (split_loop): Update.
3173 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
3174 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
3175 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
3176 * tree-ssa-reassoc.c (branch_fixup): Update.
3177 * tree-ssa-tail-merge.c (replace_block_by): Update.
3178 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
3179 (compute_path_counts): Update.
3180 (update_profile): Update.
3181 (recompute_probabilities): Update.
3182 (update_joiner_offpath_counts): Update.
3183 (estimated_freqs_path): Update.
3184 (freqs_to_counts_path): Update.
3185 (clear_counts_path): Update.
3186 (ssa_fix_duplicate_block_edges): Update.
3187 (duplicate_thread_path): Update.
3188 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
3189 (case_bit_test_cmp): Update.
3190 (collect_switch_conv_info): Update.
3191 (gen_inbound_check): Update.
3192 (do_jump_if_equal): Update.
3193 (emit_cmp_and_jump_insns): Update.
3194 * tree-tailcall.c (decrease_profile): Update.
3195 (eliminate_tail_call): Update.
3196 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
3197 (vect_do_peeling): Update.
3198 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
3199 * ubsan.c (ubsan_expand_null_ifn): Update.
3200 (ubsan_expand_ptr_ifn): Update.
3201 * value-prof.c (gimple_divmod_fixed_value): Update.
3202 (gimple_mod_pow2): Update.
3203 (gimple_mod_subtract): Update.
3204 (gimple_ic): Update.
3205 (gimple_stringop_fixed_value): Update.
3207 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
3210 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
3212 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
3214 PR rtl-optimization/82395
3215 * ira-color.c (allocno_priority_compare_func): Fix comparison step
3216 based on non_spilled_static_chain_regno_p.
3218 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
3220 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
3221 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
3222 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
3224 2017-10-19 Martin Sebor <msebor@redhat.com>
3226 PR tree-optimization/82596
3227 * tree.c (array_at_struct_end_p): Handle STRING_CST.
3229 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
3231 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
3232 (get_mem_refs_of_builtin_call): Likewise.
3233 * builtins.c (expand_builtin_apply): Adjust call to
3234 allocate_dynamic_stack_space.
3235 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
3236 the third argument to allocate_dynamic_stack_space, otherwise -1.
3237 (expand_builtin): Deal with all alloca variants.
3238 (is_inexpensive_builtin): Likewise.
3239 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
3240 * calls.c (special_function_p): Deal with all alloca variants.
3241 (initialize_argument_information): Adjust call to
3242 allocate_dynamic_stack_space.
3243 (expand_call): Likewise.
3244 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
3245 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
3246 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
3247 use it for the stack usage computation.
3248 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
3249 * function.c (gimplify_parameters): Call build_alloca_call_expr.
3250 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
3251 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
3252 (in_loop_p): Remove first argument and useless check.
3253 (pass_walloca::execute): Remove useless test and adjust call to above.
3254 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
3255 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
3256 (gimplify_call_expr): Deal with all alloca variants.
3257 * hsa-gen.c (gen_hsa_alloca): Likewise.
3258 (gen_hsa_insns_for_call): Likewise.
3259 * ipa-pure-const.c (special_builtin_state): Likewise.
3260 * tree-chkp.c (chkp_build_returned_bound): Likewise.
3261 * tree-object-size.c (alloc_object_size): Likewise.
3262 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
3263 (call_may_clobber_ref_p_1): Likewise.
3264 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
3265 (ccp_fold_stmt): Likewise.
3266 (optimize_stack_restore): Likewise.
3267 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
3268 (mark_all_reaching_defs_necessary_1): Likewise.
3269 (propagate_necessity): Likewise.
3270 (eliminate_unnecessary_stmts): Likewise.
3271 * tree.c (build_common_builtin_nodes): Build
3272 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
3273 (build_alloca_call_expr): New function.
3274 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
3275 (CASE_BUILT_IN_ALLOCA): Likewise.
3276 (build_alloca_call_expr): Declare.
3277 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
3279 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
3282 * dwarf2out.c (new_die_raw): New static inline function.
3283 (new_die): Use it to create the DIE.
3284 (add_AT_external_die_ref): Likewise.
3285 (clone_die): Likewise.
3286 (clone_as_declaration): Likewise.
3287 (dwarf2out_vms_debug_main_pointer): Likewise.
3288 (base_type_die): Likewise. Remove early return for corner cases.
3289 Do not call add_pubtype on the DIE here.
3290 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
3291 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
3292 typedefs for base types with DW_AT_endianity. Make sure a DIE with
3293 native order exists for base types, attach the DIE manually and call
3294 add_pubtype on it. Do not equate a reverse order DIE to the type.
3296 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
3298 * config/arm/arm.c (align_ok_ldrd_strd): New function.
3299 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
3301 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
3303 2017-10-19 Jakub Jelinek <jakub@redhat.com>
3305 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
3306 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
3307 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
3308 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
3309 * opts.c (sanitizer_opts): Add builtin.
3310 * ubsan.c (instrument_builtin): New function.
3311 (pass_ubsan::execute): Call it.
3312 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
3313 * doc/invoke.texi: Document -fsanitize=builtin.
3315 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
3316 builtins, store max (log2 (align), 0) into uchar field instead of
3317 align into uptr field.
3318 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
3319 store uchar 0 field instead of uptr 0 field.
3320 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
3321 instead of passing one address of struct with 2 locations pass
3322 two addresses of structs with 1 location each.
3323 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
3324 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
3325 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
3326 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
3327 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
3328 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
3329 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
3330 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
3332 2017-10-19 Martin Liska <mliska@suse.cz>
3335 * file-find.c (remove_prefix): Remove.
3336 * file-find.h (remove_prefix): Likewise.
3337 * gcc-ar.c: Remove smartness of lookup.
3339 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
3341 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
3342 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
3343 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
3345 2017-10-19 Jakub Jelinek <jakub@redhat.com>
3348 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
3349 (setcc + and to xor + setcc): New peephole2.
3351 2017-10-19 Tom de Vries <tom@codesourcery.com>
3353 * doc/sourcebuild.texi (Test Directives, Variants of
3354 dg-require-support): Add dg-require-stack-size.
3356 2017-10-19 Martin Liska <mliska@suse.cz>
3359 * gimplify.c (gimplify_decl_expr): Do not instrument variables
3360 that have a large alignment.
3361 (gimplify_target_expr): Likewise.
3363 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
3365 PR rtl-optimization/82602
3366 * ira.c (rtx_moveable_p): Return false for volatile asm.
3368 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
3371 * config/i386/i386-modes.def (CCGZ): New CC mode.
3372 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
3373 * config/i386/predicates.md (ix86_comparison_operator):
3375 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
3376 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
3377 with double-word subtraction.
3378 (put_condition_code): Handle CCGZmode.
3380 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
3382 * wide-int.cc (debug (const wide_int &)): New.
3383 (debug (const wide_int *)): New.
3384 (debug (const widest_int &)): New.
3385 (debug (const widest_int *)): New.
3387 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
3390 * lra-constraints.c (curr_insn_transform): Use non-input operand
3391 instead of output one for matched reload.
3393 2017-10-18 Bin Cheng <bin.cheng@arm.com>
3395 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
3396 (tree-ssa-loop-ivopts.h): New header file.
3397 (struct builtin_info): New fields.
3398 (classify_builtin_1): Compute and record base and offset parts for
3399 memset builtin partition by calling strip_offset.
3400 (offset_cmp, fuse_memset_builtins): New functions.
3401 (finalize_partitions): Fuse adjacent memset partitions by calling
3403 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
3404 Expose the interface.
3405 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
3407 2017-10-18 Bin Cheng <bin.cheng@arm.com>
3409 PR tree-optimization/82574
3410 * tree-loop-distribution.c (find_single_drs): New parameter. Check
3411 that data reference must be executed exactly once per iteration
3412 against the outermost loop in nest.
3413 (classify_partition): Update call to above function.
3415 2017-10-18 Richard Biener <rguenther@suse.de>
3417 PR tree-optimization/82591
3418 * graphite.c (graphite_transform_loops): Move code gen message
3420 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
3421 Here. Handle scop_to_isl_ast failing.
3422 (scop_to_isl_ast): Limit the number of ISL operations.
3424 2017-10-18 Richard Biener <rguenther@suse.de>
3426 * graphite-isl-ast-to-gimple.c
3427 (translate_isl_ast_to_gimple::set_rename): Simplify.
3428 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
3429 (graphite_copy_stmts_from_block): ... here.
3430 (copy_bb_and_scalar_dependences): Simplify.
3431 (add_parameters_to_ivs_params): Canonicalize.
3432 (generate_entry_out_of_ssa_copies): Simplify.
3433 * graphite-sese-to-poly.c (extract_affine_name): Simplify
3434 by passing in ISL dimension.
3435 (parameter_index_in_region_1): Rename to ...
3436 (parameter_index_in_region): ... this.
3437 (extract_affine): Adjust assert, pass down parameter index.
3438 (add_param_constraints): Use range-info when available.
3439 (build_scop_context): Adjust.
3440 * sese.c (new_sese_info): Adjust.
3441 (free_sese_info): Likewise.
3442 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
3443 Remove unused typedefs.
3444 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
3446 2017-10-18 Martin Liska <mliska@suse.cz>
3448 * combine.c (simplify_compare_const): Add gcc_fallthrough.
3450 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
3452 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
3453 (s390_sched_init): Do not reset s390_sched_state if we entered the
3454 current basic block via a fallthru edge and all others are unlikely.
3456 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
3458 * config/s390/s390.c (NUM_SIDES): New variable.
3459 (LONGRUNNING_THRESHOLD): New variable.
3460 (LATENCY_FACTOR): New variable.
3461 (s390_sched_score): Decrease score for long-running instructions on
3463 (s390_sched_variable_issue): Perform bookkeeping for long-running
3466 2017-10-18 Richard Biener <rguenther@suse.de>
3468 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
3469 Simplify with removal of the parameter rename map.
3470 (set_rename): Likewise.
3471 (should_copy_to_new_region): Likewise.
3472 (graphite_copy_stmts_from_block): Likewise.
3473 (copy_bb_and_scalar_dependences): Remove initialization of
3474 unused copied_bb_map.
3476 (copy_internal_parameters): Likewise.
3477 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
3478 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
3479 Use INTEGRAL_TYPE_P.
3480 (parameter_index_in_region_1): Rename to ...
3481 (assign_parameter_index_in_region): ... this. Assert we have
3482 a parameter we handle.
3483 (scan_tree_for_params): Adjust.
3484 * sese.h (parameter_rename_map_t): Remove.
3485 (struct sese_info_t): Remove unused parameter_rename_map and
3486 copied_bb_map members.
3487 * sese.c (new_sese_info): Adjust.
3488 (free_sese_info): Likewise.
3490 2017-10-18 Martin Liska <mliska@suse.cz>
3493 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
3494 on an abnormal edge.
3496 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
3498 * doc/invoke.texi (ffunction-sections and fdata-sections):
3501 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
3503 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
3504 the use statement can throw internally.
3506 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
3508 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
3509 any RTX present on the RHS of a SET.
3510 * compare-elim.c (try_eliminate_compare): Restore comment.
3512 2017-10-17 Jakub Jelinek <jakub@redhat.com>
3514 * langhooks.h (struct lang_hooks): Document that tree_size langhook
3515 may be also called on tcc_type nodes.
3516 * langhooks.c (lhd_tree_size): Likewise.
3518 2017-10-17 David Malcolm <dmalcolm@redhat.com>
3520 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
3521 format_warning_at_substring.
3522 (maybe_warn): Convert source_range * param to a location_t. Pass
3523 UNKNOWN_LOCATION rather than NULL to fmtwarn.
3524 (format_directive): Remove code to extract source_ranges and
3525 source_range * in favor of just a location_t.
3526 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
3528 * substring-locations.c (format_warning_va): Convert
3529 source_range * param to a location_t.
3530 (format_warning_at_substring): Likewise.
3531 * substring-locations.h (format_warning_va): Likewise.
3532 (format_warning_at_substring): Likewise.
3534 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
3536 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
3538 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
3539 and scatter/gather ops.
3541 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
3542 vec_gather_load and vec_scatter_store.
3543 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
3544 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
3545 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
3546 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
3547 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
3548 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
3550 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
3552 * reg-stack.c (compare_for_stack_reg): Add bool argument.
3553 Detect FTST instruction and handle its register pops. Only pop
3554 second operand if can_pop_second_op is true.
3555 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
3556 set can_pop_second_op to false in the compare_for_stack_reg call.
3558 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
3559 output_fp_compare for stack register operands.
3560 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
3561 instructions here. Do not emit stack register pops here. Assert
3562 that FCOMPP pops next to top stack register. Rewrite function.
3564 2017-10-17 Nathan Sidwell <nathan@acm.org>
3567 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
3568 use DECL_ASSEMBLER_NAME_RAW.
3571 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
3574 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
3575 Wilco Dijkstra <wilco.dijkstra@arm.com>
3577 * builtins.c (expand_builtin_update_setjmp_buf): Add a
3578 converstion to Pmode from the buf_addr.
3580 2017-10-17 Richard Biener <rguenther@suse.de>
3582 * graphite-dependences.c (scop_get_reads_and_writes): Change
3583 output parameters to references.
3585 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
3587 PR 71026/tree-optimization
3588 * fold-const.c (distribute_real_division): Removed.
3589 (fold_binary_loc): Remove calls to distribute_real_divison.
3591 2017-10-17 Richard Biener <rguenther@suse.de>
3593 * graphite-scop-detection.c
3594 (scop_detection::stmt_has_simple_data_refs_p): Always use
3595 the full nest as region.
3596 (try_generate_gimple_bb): Likewise.
3597 * sese.c (scalar_evolution_in_region): Simplify now that
3598 SCEV can handle instantiation in regions.
3599 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
3600 in the non-loop part of a function if requested.
3602 2017-10-17 Richard Biener <rguenther@suse.de>
3604 PR tree-optimization/82563
3605 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
3607 (graphite_regenerate_ast_isl): Call it.
3608 * graphite-scop-detection.c (build_scops): Remove entry edge split.
3610 2017-10-17 Jakub Jelinek <jakub@redhat.com>
3612 PR tree-optimization/82549
3613 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
3614 Formatting fixes. Instead of calling make_bit_field_ref with negative
3617 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
3619 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
3620 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
3621 _mm_maskz_reduce_ss): New.
3622 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
3623 __builtin_ia32_reducess_mask): Ditto..
3624 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
3625 * config/i386/sse.md (reduces<mode>): Renamed to ...
3626 (reduces<mode><mask_scalar_name>): ... this.
3627 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
3629 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
3630 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
3632 2017-10-16 David Malcolm <dmalcolm@redhat.com>
3634 * Makefile.in (OBJS): Add unique-ptr-tests.o.
3635 * selftest-run-tests.c (selftest::run_tests): Call
3636 selftest::unique_ptr_tests_cc_tests.
3637 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
3638 * unique-ptr-tests.cc: New file.
3640 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
3643 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
3645 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
3646 (make_hard_regno_born, make_hard_regno_dead): Update
3647 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
3649 2017-10-16 Jeff Law <law@redhat.com>
3651 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
3653 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
3655 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
3657 2017-10-16 Olivier Hainque <hainque@adacore.com>
3659 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
3660 with_cpu if we were configured for an e500v2 target cpu name.
3662 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
3664 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
3665 * doc/invoke.texi: Document +nodsp as a valid extension for
3668 2017-10-16 Martin Liska <mliska@suse.cz>
3670 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
3671 (test_set_range): Likewise.
3672 (test_range_functions): Rename to ...
3673 (test_bit_in_range): ... this.
3674 (sbitmap_c_tests): Add new test.
3676 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3678 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
3680 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
3681 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
3683 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3685 * config/aarch64/aarch64-builtins.c
3686 (aarch64_types_quadopu_lane_qualifiers): New.
3687 (TYPES_QUADOPU_LANE): New.
3688 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
3689 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
3690 (aarch64_<sur>dot_laneq<vsi2qi>): New.
3691 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
3692 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
3693 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
3694 (Vdottype, DOTPROD): New.
3695 (sur): Add SDOT and UDOT.
3697 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3699 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
3700 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
3701 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
3703 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
3704 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
3705 Enable TARGET_DOTPROD.
3706 (cortex-a75.cortex-a55): Likewise.
3707 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
3709 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3711 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
3712 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
3714 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
3716 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
3717 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
3718 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
3719 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
3720 * config/arm/types.md (neon_dot, neon_dot_q): New.
3721 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
3723 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3725 * config/arm/arm.h (TARGET_DOTPROD): New.
3726 * config/arm/arm.c (arm_arch_dotprod): New.
3727 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
3728 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
3729 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
3730 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
3731 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
3732 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
3733 * doc/invoke.texi (armv8.2-a): Document dotprod
3735 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
3737 * i386.c (ix86_vec_cost): New function.
3738 (ix86_rtx_costs): Handle vector operations better.
3739 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
3740 * x86-tune-costs.h: Add new costs to all tables.
3742 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
3744 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
3746 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
3747 divsd, sqrtss and sqrtsd
3748 * x86-tune-costs.h: Add new entries to all costs.
3749 (znver1_cost): Fix to match real instruction latencies.
3751 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3752 Michael Collison <michael.collison@arm.com>
3754 * compare-elim.c: Include emit-rtl.h.
3755 (can_merge_compare_into_arith): New function.
3756 (try_validate_parallel): Likewise.
3757 (try_merge_compare): Likewise.
3758 (try_eliminate_compare): Call the above when no previous clobber
3760 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
3763 2017-10-14 Jakub Jelinek <jakub@redhat.com>
3767 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
3768 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
3772 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
3773 to be any operand_equal_p operands. For & (B - 1) require
3774 B to be power of 2. Recognize
3775 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
3777 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
3780 * optabs.c (expand_memory_blockage): Fix call of
3781 targetm.have_memory_blockage.
3783 2017-10-14 Jakub Jelinek <jakub@redhat.com>
3786 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
3787 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
3788 objects to extra_objs instead of overwriting it.
3790 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
3792 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
3793 Use any_fp_register_operand as operand[3] predicate. Simplify
3794 equality test for operands[2] and operands[4] memory location.
3795 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
3796 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
3797 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
3798 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
3799 any_fp_register_operand as operand[1] predicate. Simplify
3800 equality test for operands[0] and operands[3] memory location.
3801 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
3802 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
3803 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
3805 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
3807 * target-insns.def: Add memory_blockage.
3808 * optabs.c (expand_memory_blockage): New function.
3809 (expand_asm_memory_barrier): Rename ...
3810 (expand_asm_memory_blockage): ... to this.
3811 (expand_mem_thread_fence): Call expand_memory_blockage
3812 instead of expand_asm_memory_barrier.
3813 (expand_mem_singnal_fence): Ditto.
3814 (expand_atomic_load): Ditto.
3815 (expand_atomic_store): Ditto.
3816 * doc/md.texi (Standard Pattern Names For Generation):
3817 Document memory_blockage instruction pattern.
3819 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
3821 * config/rl78/rl78.c (rl78_emit_libcall): New function.
3822 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
3823 * config/rl78/rl78.md: New define_expand "adddi3".
3825 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
3827 * cfghooks.c (verify_flow_info): Disable check that all probabilities
3830 2017-10-13 Jeff Law <law@redhat.com>
3832 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
3834 2017-10-13 Jakub Jelinek <jakub@redhat.com>
3837 * internal-fn.c (expand_mul_overflow): If both operands have
3838 the same highpart of -1 or 0 and the topmost bit of lowpart
3839 is different, overflow is if res <= 0 rather than res < 0.
3841 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
3843 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
3844 TARGET_P9_VECTOR code for unaligned_load case.
3846 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
3848 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
3850 2017-10-13 Nathan Sidwell <nathan@acm.org>
3852 * tree-core.h (tree_contains_struct): Make bool.
3853 * tree.c (tree_contains_struct): Likewise.
3854 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
3855 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
3856 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
3857 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
3858 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
3860 2017-10-13 Richard Biener <rguenther@suse.de>
3862 * graphite-isl-ast-to-gimple.c
3863 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
3864 parameters and dominance check.
3865 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
3866 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
3867 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
3868 Do not update SSA form here or do intermediate IL verification.
3869 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
3870 (graphite_initialize): Remove check on the number of loops in
3871 the function and inline into graphite_transform_loops.
3872 (graphite_finalize): Inline into graphite_transform_loops.
3873 (graphite_transform_loops): Perform SSA update and IL verification
3875 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
3877 2017-10-13 Richard Biener <rguenther@suse.de>
3879 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
3880 graphite_expression_type_precision): Avoid global constructor
3882 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
3883 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
3884 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
3885 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
3886 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
3887 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
3889 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
3892 * config/i386/i386.h (ix86_red_zone_size): New.
3893 * config/i386/i386.md (push peephole2s): Replace
3894 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
3896 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3897 Alan Hayward <alan.hayward@arm.com>
3898 David Sherwood <david.sherwood@arm.com>
3900 * combine.c (can_change_dest_mode): Reject changes in
3901 REGMODE_NATURAL_SIZE.
3903 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3904 Alan Hayward <alan.hayward@arm.com>
3905 David Sherwood <david.sherwood@arm.com>
3907 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
3908 (expand_debug_source_expr): Likewise.
3909 * combine.c (combine_simplify_rtx): Likewise.
3910 * cse.c (fold_rtx): Likewise.
3911 * fwprop.c (canonicalize_address): Likewise.
3912 * targhooks.c (default_shift_truncation_mask): Likewise.
3914 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3915 Alan Hayward <alan.hayward@arm.com>
3916 David Sherwood <david.sherwood@arm.com>
3918 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
3919 (widened_mode): Likewise.
3920 (expand_unop): Likewise.
3921 * ree.c (transform_ifelse): Likewise.
3922 (merge_def_and_ext): Likewise.
3923 (combine_reaching_defs): Likewise.
3924 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3926 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3927 Alan Hayward <alan.hayward@arm.com>
3928 David Sherwood <david.sherwood@arm.com>
3930 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
3931 * combine.c (gen_lowpart_for_combine): Likewise.
3932 * dwarf2out.c (rtl_for_decl_location): Likewise.
3933 * final.c (alter_subreg): Likewise.
3934 * rtlhooks.c (gen_lowpart_general): Likewise.
3935 (gen_lowpart_if_possible): Likewise.
3937 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3938 Alan Hayward <alan.hayward@arm.com>
3939 David Sherwood <david.sherwood@arm.com>
3941 * calls.c (expand_call): Use subreg_lowpart_offset.
3942 * cse.c (cse_insn): Likewise.
3943 * regcprop.c (copy_value): Likewise.
3944 (copyprop_hardreg_forward_1): Likewise.
3946 2017-10-13 Jakub Jelinek <jakub@redhat.com>
3949 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
3950 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
3951 =Q constraints to +Q and into insn condition add check
3952 that operands[0] and operands[1] are equal.
3953 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
3954 =Q constraints to +Q and into insn condition add check
3955 that operands[0] is equal to either operands[1] or operands[2].
3958 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
3959 instead of handling MINUS_EXPR twice (once for each argument),
3960 canonicalize operand order and handle just once, use rtype where
3961 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
3964 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
3965 any values of __C while still being pattern recognizable as a simple
3968 2017-10-13 Richard Biener <rguenther@suse.de>
3970 PR tree-optimization/82451
3972 2017-10-02 Richard Biener <rguenther@suse.de>
3974 PR tree-optimization/82355
3975 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
3976 a mapping for the enclosing loop but avoid generating one for
3978 (copy_bb_and_scalar_dependences): Remove premature codegen
3979 error on PHIs in blocks duplicated into multiple places.
3980 * graphite-scop-detection.c
3981 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
3982 in the region use it as loop and nest to analyze the DR in.
3983 (try_generate_gimple_bb): Likewise.
3984 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
3985 (add_loop_constraints): For blocks in a loop not in the region
3986 create a dimension with a single iteration.
3987 * sese.h (gbb_loop_at_index): Remove assert.
3989 * cfgloop.c (loop_preheader_edge): For the loop tree root
3990 return the single successor of the entry block.
3991 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
3992 Reset the SCEV hashtable and niters.
3993 * graphite-scop-detection.c
3994 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
3995 assert that we only have POLYNOMIAL_CHREC that vary in loops
3996 contained in the region.
3997 (scop_detection::graphite_can_represent_expr): Adjust.
3998 (scop_detection::stmt_has_simple_data_refs_p): For loops
3999 not in the region set loop to NULL. The nest is now the
4000 entry edge to the region.
4001 (try_generate_gimple_bb): Likewise.
4002 * sese.c (scalar_evolution_in_region): Adjust for
4003 instantiate_scev change.
4004 * tree-data-ref.h (graphite_find_data_references_in_stmt):
4005 Make nest parameter the edge into the region.
4006 (create_data_ref): Likewise.
4007 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
4008 entry edge into a region and adjust instantiate_scev calls.
4009 (create_data_ref): Likewise.
4010 (graphite_find_data_references_in_stmt): Likewise.
4011 (find_data_references_in_stmt): Pass the loop preheader edge
4012 from the nest argument.
4013 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
4014 parameter the edge into the region.
4015 (instantiate_parameters): Use the loop preheader edge as entry.
4016 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
4018 (get_instantiated_value_entry): Make instantiate_below parameter
4019 the edge into the region.
4020 (instantiate_scev_name): Likewise. Adjust dominance checks,
4021 when we cannot use loop-based instantiation instantiate by
4022 walking use-def chains.
4023 (instantiate_scev_poly): Adjust.
4024 (instantiate_scev_binary): Likewise.
4025 (instantiate_scev_convert): Likewise.
4026 (instantiate_scev_not): Likewise.
4027 (instantiate_array_ref): Remove.
4028 (instantiate_scev_3): Likewise.
4029 (instantiate_scev_2): Likewise.
4030 (instantiate_scev_1): Likewise.
4031 (instantiate_scev_r): Do not blindly handle N-operand trees.
4032 Do not instantiate array-refs. Handle all constants and invariants.
4033 (instantiate_scev): Make instantiate_below parameter
4034 the edge into the region.
4035 (resolve_mixers): Use the loop preheader edge for the region
4036 parameter to instantiate_scev_r.
4037 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
4039 2017-10-13 Richard Biener <rguenther@suse.de>
4041 PR tree-optimization/82525
4042 * graphite-isl-ast-to-gimple.c
4043 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
4045 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
4046 Fail code generation when we cannot represent the isl integer.
4047 (binary_op_to_tree): Elide modulo operations that are no-ops
4048 in the type we code generate. Remove now superfluous code
4051 2017-10-13 Richard Biener <rguenther@suse.de>
4053 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
4054 (scop_detection::harmful_loop_in_region): Remove premature
4055 IV type restriction.
4056 (scop_detection::graphite_can_represent_scev): We can handle
4057 pointer IVs just fine.
4059 2017-10-13 Alan Modra <amodra@gmail.com>
4061 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
4062 "Clobbers and Scratch Registers". Add paragraph on
4063 alternative to clobbers for scratch registers and OpenBLAS
4066 2017-10-13 Alan Modra <amodra@gmail.com>
4068 * doc/extend.texi (Clobbers): Correct vax example. Delete old
4069 example of a memory input for a string of known length. Move
4070 commentary out of table. Add a number of new examples
4071 covering array memory inputs.
4073 2017-10-12 Martin Liska <mliska@suse.cz>
4075 PR tree-optimization/82493
4076 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
4077 (test_range_functions): New function.
4078 (sbitmap_c_tests): Likewise.
4079 * selftest-run-tests.c (selftest::run_tests): Run new tests.
4080 * selftest.h (sbitmap_c_tests): New function.
4082 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
4084 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
4086 * config/rs6000/amo.h: Fix spacing issue.
4088 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4091 * config/i386/i386.md (*ashl<mode>3_mask_1,
4092 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
4093 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
4096 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
4098 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
4099 (profile_probability): Set max_probability
4100 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
4102 * profile-count.c (profile_probability::differs_from_p): Do not
4103 rely on max_probaiblity == 10000
4105 2017-10-12 Jeff Law <law@redhat.com>
4107 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
4110 2017-10-12 Martin Sebor <msebor@redhat.com>
4114 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
4115 (handle_alias_pairs): Call it.
4116 * common.opt (-Wattribute-alias): New option.
4117 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
4118 * doc/invoke.texi (-Wattribute-alias): Document.
4120 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
4123 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
4125 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4127 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4128 (make_hard_regno_born, make_hard_regno_dead): Update
4129 bb_killed_pseudos and bb_gen_pseudos.
4131 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
4133 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
4135 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
4137 * config/alpha/alpha.c (alpha_split_conditional_move):
4138 Use std::swap instead of manually swapping.
4139 (alpha_stdarg_optimize_hook): Ditto.
4140 (alpha_canonicalize_comparison): Ditto.
4142 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4144 * tree-loop-distribution.c (struct builtin_info): New struct.
4145 (struct partition): Refactor fields into struct builtin_info.
4146 (partition_free): Free struct builtin_info.
4147 (build_size_arg_loc, build_addr_arg_loc): Delete.
4148 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
4149 information from struct builtin_info.
4150 (find_single_drs): New function refactored from classify_partition.
4151 Also moved builtin validity checks to this function.
4152 (compute_access_range, alloc_builtin): New functions.
4153 (classify_builtin_st, classify_builtin_ldst): New functions.
4154 (classify_partition): Refactor code into functions find_single_drs,
4155 classify_builtin_st and classify_builtin_ldst.
4156 (distribute_loop): Don't do runtime alias check when distributing
4158 (find_seed_stmts_for_distribution): New function.
4159 (pass_loop_distribution::execute): Refactor code finding seed
4160 stmts into above function. Support distribution for the innermost
4161 two-level loop nest. Adjust dump information.
4163 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4165 * tree-loop-distribution.c: Adjust the general comment.
4166 (NUM_PARTITION_THRESHOLD): New macro.
4167 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
4168 (classify_partition): Skip builtin pattern of loop nest's inner loop.
4169 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
4170 in call to build_partition_graph.
4171 (finalize_partitions): New parameter. Make loop distribution more
4172 conservative by fusing more partitions.
4173 (distribute_loop): Don't do runtime alias check in case of loop nest
4175 (find_seed_stmts_for_distribution): New function.
4176 (prepare_perfect_loop_nest): New function.
4177 (pass_loop_distribution::execute): Refactor code finding seed stmts
4178 and loop nest into above functions. Support loop nest distribution.
4179 Adjust dump information accordingly.
4181 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4183 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
4184 and set PTYPE_SEQUENTIAL for merged partition.
4186 2017-10-12 Richard Biener <rguenther@suse.de>
4188 PR tree-optimization/69728
4190 2017-09-19 Richard Biener <rguenther@suse.de>
4192 PR tree-optimization/69728
4193 * graphite-sese-to-poly.c (schedule_error): New global.
4194 (add_loop_schedule): Handle empty domain by failing the
4196 (build_original_schedule): Handle schedule_error.
4198 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
4199 domain by returning an unchanged schedule.
4201 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4203 * genrecog.c (validate_pattern): For VEC_SELECT verify that
4204 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
4206 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
4208 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
4211 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4214 * expr.c (store_field): Don't optimize away bitsize == 0 store
4215 from CALL_EXPR with addressable return type.
4217 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
4219 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
4220 * config/rs6000/rs6000.md (sel): Delete mode attribute.
4221 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
4222 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
4223 TARGET_ISEL instead of TARGET_ISEL<sel>.
4225 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
4227 * config/rs6000/rs6000.c
4228 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
4230 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
4232 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
4234 (reg_or_cint_operand): Fix comment.
4235 (reg_or_zero_operand): New predicate.
4236 * config/rs6000/rs6000-protos.h (output_isel): Delete.
4237 * config/rs6000/rs6000.c (output_isel): Delete.
4238 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
4239 instead of reg_or_cint_operand. Output instruction directly (not via
4241 (isel_unsigned_<mode>): Ditto.
4242 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
4243 gpc_reg_operand. Add an instruction alternative for this. Output
4244 instruction directly.
4245 (*isel_reversed_unsigned_<mode>): Ditto.
4247 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4249 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
4250 (TARGET_CANONICALIZE_COMPARISON): Define.
4252 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
4255 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
4256 Check whether the dest is REG before adding REG_EQUIV note.
4258 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
4261 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4263 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4264 (make_hard_regno_born, make_hard_regno_dead): Update
4265 bb_killed_pseudos and bb_gen_pseudos.
4267 2017-10-11 Nathan Sidwell <nathan@acm.org>
4269 * incpath.h (enum incpath_kind): Name enum, prefix values.
4270 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
4271 * incpath.c (heads, tails): Use INC_MAX.
4272 (add_env_var_paths, add_standard_paths): Use incpath_kind.
4273 (merge_include_chains, split_quote_chain,
4274 register_include_chains): Update incpath_kind names.
4275 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
4276 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
4278 (add_framework_path, darwin_register_objc_includes): Likewise.
4279 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
4281 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4283 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
4284 Do not use float_operator operator predicate.
4285 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
4286 * config/i386/predicates.md (float_operator): Remove predicate.
4288 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4290 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
4291 (*jccxf_i387): Ditto.
4292 (*jcc<mode>_i387): Ditto.
4293 (*jccu<mode>_i387): Ditto.
4294 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
4295 (*jcc_*_i387 splitters): Remove.
4296 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
4297 * config/i386/i386.c (ix86_split_fp_branch): Remove.
4298 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
4301 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
4303 * profile-count.h (slow_safe_scale_64bit): New function.
4304 (safe_scale_64bit): New inline.
4305 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
4306 * profile-count.c: Include wide-int.h
4307 (slow_safe_scale_64bit): New.
4309 2017-10-11 Nathan Sidwell <nathan@acm.org>
4311 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
4312 HAS_DECL_ASSEMBLER_NAME_P.
4313 * gimple-expr.c (gimple_decl_printable_name: Check
4314 HAS_DECL_ASSEMBLER_NAME_P too.
4315 * ipa-utils.h (type_in_anonymous_namespace_p): Check
4316 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
4317 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
4318 * passes.c (rest_of_decl_compilation): Check
4319 HAS_DECL_ASSEMBLER_NAME_P too.
4320 * recog.c (verify_changes): Likewise.
4321 * tree-pretty-print.c (dump_decl_name): Likewise.
4322 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
4324 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
4325 (SET_DECL_ASSEMBLER_NAME): Use it.
4326 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
4327 (COPY_DECL_ASSEMBLER_NAME): Likewise.
4328 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
4330 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
4332 * config.gcc (i386, x86_64): Add extra objects.
4333 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
4334 (ix86_min_insn_size): Declare.
4335 (ix86_issue_rate): Declare.
4336 (ix86_adjust_cost): Declare.
4337 (ia32_multipass_dfa_lookahead): Declare.
4338 (ix86_macro_fusion_p): Declare.
4339 (ix86_macro_fusion_pair_p): Declare.
4340 (ix86_bd_has_dispatch): Declare.
4341 (ix86_bd_do_dispatch): Declare.
4342 (ix86_core2i7_init_hooks): Declare.
4343 (ix86_atom_sched_reorder): Declare.
4344 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
4345 (COSTS_N_BYTES): Move to x86-tune-costs.h.
4346 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
4347 (rip_relative_addr_p): Rename to ...
4348 (ix86_rip_relative_addr_p): ... this one; export.
4349 (memory_address_length): Update.
4350 (ix86_issue_rate): Move to x86-tune-sched.c.
4351 (ix86_flags_dependent): Move to x86-tune-sched.c.
4352 (ix86_agi_dependent): Move to x86-tune-sched.c.
4353 (exact_dependency_1): Move to x86-tune-sched.c.
4354 (exact_store_load_dependency): Move to x86-tune-sched.c.
4355 (ix86_adjust_cost): Move to x86-tune-sched.c.
4356 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
4357 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
4358 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
4359 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
4360 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
4361 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
4362 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
4363 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
4364 (min_insn_size): Rename to ...
4365 (ix86_min_insn_size): ... this one; export.
4366 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
4367 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
4368 (core2i7_first_cycle_multipass_backtrack): Move to
4369 x86-tune-sched-core.c.
4370 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
4371 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
4372 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
4373 (ix86_avoid_jump_mispredicts): Update.
4374 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
4375 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
4376 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
4377 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
4378 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
4379 (MAX_INSN): Move to ix86-tune-sched-bd.c.
4380 (MAX_IMM): Move to ix86-tune-sched-bd.c.
4381 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
4382 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
4383 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
4384 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
4385 (MAX_STORE): Move to ix86-tune-sched-bd.c.
4386 (BIG): Move to ix86-tune-sched-bd.c.
4387 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
4388 (enum insn_path): Move to ix86-tune-sched-bd.c.
4389 (get_mem_group): Move to ix86-tune-sched-bd.c.
4390 (is_cmp): Move to ix86-tune-sched-bd.c.
4391 (dispatch_violation): Move to ix86-tune-sched-bd.c.
4392 (is_branch): Move to ix86-tune-sched-bd.c.
4393 (is_prefetch): Move to ix86-tune-sched-bd.c.
4394 (init_window): Move to ix86-tune-sched-bd.c.
4395 (allocate_window): Move to ix86-tune-sched-bd.c.
4396 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
4397 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
4398 (process_end_window): Move to ix86-tune-sched-bd.c.
4399 (allocate_next_window): Move to ix86-tune-sched-bd.c.
4400 (find_constant): Move to ix86-tune-sched-bd.c.
4401 (get_num_immediates): Move to ix86-tune-sched-bd.c.
4402 (has_immediate): Move to ix86-tune-sched-bd.c.
4403 (get_insn_path): Move to ix86-tune-sched-bd.c.
4404 (get_insn_group): Move to ix86-tune-sched-bd.c.
4405 (count_num_restricted): Move to ix86-tune-sched-bd.c.
4406 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
4407 (add_insn_window): Move to ix86-tune-sched-bd.c.
4408 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
4409 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
4410 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
4411 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
4412 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
4413 (do_dispatch): Move to ix86-tune-sched-bd.c.
4414 (has_dispatch): Move to ix86-tune-sched-bd.c.
4415 * config/i386/t-i386: Add new object files.
4416 * config/i386/x86-tune-costs.h: New file.
4417 * config/i386/x86-tune-sched-atom.c: New file.
4418 * config/i386/x86-tune-sched-bd.c: New file.
4419 * config/i386/x86-tune-sched-core.c: New file.
4420 * config/i386/x86-tune-sched.c: New file.
4422 2017-10-11 Liu Hao <lh_mouse@126.com>
4424 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
4425 the generic version below instead.
4426 (should_colorize): Recognize Windows consoles as terminals
4428 * pretty-print.c [__MINGW32__] (write_all): New function.
4429 [__MINGW32__] (find_esc_head): Likewise.
4430 [__MINGW32__] (find_esc_terminator): Likewise.
4431 [__MINGW32__] (eat_esc_sequence): Likewise.
4432 [__MINGW32__] (mingw_ansi_fputs): New function that handles
4434 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
4437 2017-10-11 Richard Biener <rguenther@suse.de>
4439 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
4440 Properly call analyze_scalar_evolution with the loop of the stmt.
4442 2017-10-11 Richard Biener <rguenther@suse.de>
4444 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
4445 * tree-core.h (tree_base): Add chrec_var union member.
4446 * tree.h (CHREC_VAR): Remove.
4447 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
4448 * tree-chrec.h (build_polynomial_chrec): Adjust.
4449 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
4450 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
4452 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
4454 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
4455 * match.pd: ... here.
4456 ((T) X == (T) Y): Relax condition.
4458 2017-10-11 Bin Cheng <bin.cheng@arm.com>
4460 PR tree-optimization/82472
4461 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
4463 (break_alias_scc_partitions): Update postorder number.
4465 2017-10-11 Martin Liska <mliska@suse.cz>
4468 * opts.c (parse_no_sanitize_attribute): Do not use error_value
4470 * opts.h (parse_no_sanitize_attribute): Remove last argument.
4472 2017-10-11 Martin Liska <mliska@suse.cz>
4474 * print-rtl.c (print_insn): Move declaration of idbuf
4475 to same scope as name.
4477 2017-10-11 Martin Liska <mliska@suse.cz>
4482 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
4485 2017-10-11 Martin Liska <mliska@suse.cz>
4488 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
4491 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
4493 * config.gcc (default_gnu_indirect_function): Default to yes for
4494 arm*-*-linux* with glibc.
4496 2017-10-11 Richard Biener <rguenther@suse.de>
4498 * tree-scalar-evolution.c (get_scalar_evolution): Handle
4499 default-defs and types we do not want to analyze.
4500 (interpret_loop_phi): Replace unreachable code with an assert.
4501 (compute_scalar_evolution_in_loop): Remove and inline ...
4502 (analyze_scalar_evolution_1): ... here, replacing condition with
4503 what makes the intent clearer. Remove handling of cases
4504 get_scalar_evolution now handles.
4506 2017-10-10 Jim Wilson <wilson@tuliptree.org>
4508 PR rtl-optimization/81434
4509 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
4510 comment for main loop. In sched_group_found if, also add checks for
4511 pass and min_cost_group.
4513 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
4515 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
4516 (rs6000_insn_cost): New function.
4517 * config/rs6000/rs6000.md (cost): New attribute.
4519 2017-10-10 Jakub Jelinek <jakub@redhat.com>
4520 H.J. Lu <hongjiu.lu@intel.com>
4524 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
4525 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
4526 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
4527 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
4528 to OPTION_MASK_ISA_AVX512VL - builtins that have both
4529 OPTION_MASK_ISA_MMX and some other bit set require both
4530 mmx and the ISAs without the mmx bit.
4531 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
4532 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
4533 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
4534 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
4535 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
4536 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
4537 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
4538 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
4539 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
4540 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
4541 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
4542 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
4543 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
4544 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
4545 Add OPTION_MASK_ISA_MMX.
4547 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
4549 * config.gcc (armv7*-*-freebsd*): New target.
4550 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
4552 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
4554 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
4555 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
4558 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
4560 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
4561 a template parameter.
4562 (WIDE_INT_REF_FOR): Update accordingly.
4563 * tree.h (wi::int_traits <const_tree>): Delete.
4564 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
4565 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
4566 (wi::tree_to_wide_ref): New typedef.
4567 (wi::to_wide): New function.
4568 * calls.c (get_size_range): Use wi::to_wide when operating on
4570 * cgraph.c (cgraph_node::create_thunk): Likewise.
4571 * config/i386/i386.c (ix86_data_alignment): Likewise.
4572 (ix86_local_alignment): Likewise.
4573 * dbxout.c (stabstr_O): Likewise.
4574 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
4575 * expr.c (const_vector_from_tree): Likewise.
4576 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
4577 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
4578 (fold_negate_expr_1, int_const_binop_1, const_binop)
4579 (fold_convert_const_int_from_real, optimize_bit_field_compare)
4580 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
4581 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
4582 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
4583 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
4584 (fold_not_const, round_up_loc): Likewise.
4585 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
4586 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
4587 (alloca_call_type): Likewise.
4588 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
4589 * godump.c (go_output_typedef): Likewise.
4590 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
4591 * internal-fn.c (get_min_precision): Likewise.
4592 * ipa-cp.c (ipcp_store_vr_results): Likewise.
4593 * ipa-polymorphic-call.c
4594 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
4595 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
4596 (ipa_modify_call_arguments): Likewise.
4597 * match.pd: Likewise.
4598 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
4599 * print-tree.c (print_node_brief, print_node): Likewise.
4600 * stmt.c (expand_case): Likewise.
4601 * stor-layout.c (layout_type): Likewise.
4602 * tree-affine.c (tree_to_aff_combination): Likewise.
4603 * tree-cfg.c (group_case_labels_stmt): Likewise.
4604 * tree-data-ref.c (dr_analyze_indices): Likewise.
4605 (prune_runtime_alias_test_list): Likewise.
4606 * tree-dump.c (dequeue_and_dump): Likewise.
4607 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
4608 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
4609 * tree-pretty-print.c (dump_generic_node): Likewise.
4610 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
4611 (simple_iv_with_niters): Likewise.
4612 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
4613 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
4614 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
4615 * tree-ssa-loop-niter.c (split_to_var_and_offset)
4616 (refine_value_range_using_guard, number_of_iterations_ne_max)
4617 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
4618 (get_cst_init_from_scev, record_nonwrapping_iv)
4619 (scev_var_range_cant_overflow): Likewise.
4620 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
4621 * tree-ssa-pre.c (compute_avail): Likewise.
4622 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
4623 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
4624 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
4625 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
4626 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
4627 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
4628 (dump_case_nodes, try_switch_expansion): Likewise.
4629 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
4630 (vect_do_peeling): Likewise.
4631 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
4632 * tree-vect-stmts.c (vectorizable_load): Likewise.
4633 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
4634 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
4635 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
4636 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
4637 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
4638 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
4639 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
4640 (evrp_dom_walker::before_dom_children): Likewise.
4641 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
4642 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
4643 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
4644 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
4645 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
4646 (get_range_pos_neg): Likewise.
4647 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
4648 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
4649 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
4650 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
4651 * config/avr/avr.c (avr_fold_builtin): Likewise.
4652 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
4653 * config/msp430/msp430.c (msp430_attr): Likewise.
4654 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
4655 * config/powerpcspe/powerpcspe-c.c
4656 (altivec_resolve_overloaded_builtin): Likewise.
4657 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
4658 (rs6000_expand_ternop_builtin): Likewise.
4659 * config/rs6000/rs6000-c.c
4660 (altivec_resolve_overloaded_builtin): Likewise.
4661 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
4662 (rs6000_expand_ternop_builtin): Likewise.
4663 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
4665 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4667 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
4668 when copying loop nest with only one inner loop.
4670 2017-10-10 Richard Biener <rguenther@suse.de>
4672 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
4673 blocks if SCEV is active.
4674 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
4676 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
4677 (scev_initialize): Assert we are not yet initialized.
4679 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4681 * tree-loop-distribution.c (generate_loops_for_partition): Remove
4682 inner loop's exit stmt by making it always exit the loop, otherwise
4683 we would generate an infinite empty loop.
4685 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4687 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
4688 renaming variables in new preheader if it's deleted.
4690 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4692 * tree-loop-distribution.c (struct partition): Remove unused field
4693 loops of the structure.
4694 (partition_alloc, partition_free): Ditto.
4695 (build_rdg_partition_for_vertex): Ditto.
4697 2017-10-09 Jeff Law <law@redhat.com>
4699 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
4700 return type to match prototype and documentation.
4702 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4704 * config/rs6000/rs6000.c (processor_costs): Move to ...
4705 * config/rs6000/rs6000.h: ... here.
4706 (rs6000_cost): Declare.
4708 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
4710 * except.c (setjmp_fn): New global variable.
4711 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
4712 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
4713 if DONT_USE_BUILTIN_SETJMP is defined.
4715 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4717 * target.def (insn_cost): New hook.
4718 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
4719 * doc/tm.texi: Regenerate.
4720 * rtlanal.c (insn_cost): Use the new hook.
4722 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4724 * combine.c (combine_validate_cost): Compute the new insn_cost,
4725 not just pattern_cost.
4726 (try_combine): Adjust comment.
4728 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4730 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
4732 * combine.c (uid_insn_cost): Adjust comment.
4733 (combine_validate_cost): Adjust comment. Use pattern_cost instead
4735 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
4736 * dse.c (find_shift_sequence): Ditto.
4737 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
4738 (bb_valid_for_noce_process_p): Use pattern_cost.
4739 * rtl.h (insn_rtx_cost): Delete.
4740 (pattern_cost): New prototype.
4741 (insn_cost): New prototype.
4742 * rtlanal.c (insn_rtx_cost): Rename to...
4743 (pattern_cost): ... this.
4746 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
4748 * config/i386/i386.md (*jcc_2): Remove insn pattern.
4749 (*jcc<mode>_0_r_i387): Ditto.
4750 (*jccxf_r_i387): Ditto.
4751 (*jcc<mode>_r_i387): Ditto.
4752 (*jccu<mode>_r_i387): Ditto.
4753 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
4754 (*jcc): Rename from *jcc_1.
4756 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4758 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
4759 deferred rescans after the lvx/stvx recombination pre-pass.
4761 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
4763 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
4764 memory operation instruction support.
4765 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
4766 (rs6000-ibm-aix[789]*): Likewise.
4767 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
4768 Document new functions.
4770 2017-10-09 Richard Biener <rguenther@suse.de>
4772 PR tree-optimization/82397
4773 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
4774 equality only for semantically equal trees.
4776 2017-10-09 Richard Biener <rguenther@suse.de>
4778 PR tree-optimization/82449
4779 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
4780 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
4781 allow constant addresses.
4782 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
4785 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4787 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
4790 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4793 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
4796 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4799 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
4801 2017-10-09 Jakub Jelinek <jakub@redhat.com>
4804 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
4805 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
4807 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
4809 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
4810 (WI_BINARY_PREDICATE_RESULT): Likewise.
4811 (wi::binary_traits::operator_result): New type.
4812 (wi::binary_traits::predicate_result): Likewise.
4813 (generic_wide_int::operator~, unary generic_wide_int::operator-)
4814 (generic_wide_int::operator==, generic_wide_int::operator!=)
4815 (generic_wide_int::operator&, generic_wide_int::and_not)
4816 (generic_wide_int::operator|, generic_wide_int::or_not)
4817 (generic_wide_int::operator^, generic_wide_int::operator+
4818 (binary generic_wide_int::operator-, generic_wide_int::operator*):
4820 (operator~, unary operator-, operator==, operator!=, operator&)
4821 (operator|, operator^, operator+, binary operator-, operator*): New
4823 * expr.c (get_inner_reference): Use wi::bit_and_not.
4824 * fold-const.c (fold_binary_loc): Likewise.
4825 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
4826 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
4827 (bit_value_binop): Likewise.
4828 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
4829 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
4830 (extract_range_from_binary_expr_1): Likewise.
4831 (masked_increment): Likewise.
4832 (simplify_bit_ops_using_ranges): Likewise.
4834 2017-10-09 Martin Jambor <mjambor@suse.cz>
4837 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
4838 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
4839 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
4841 (hsa_fixup_mov_insn_type): New function.
4842 (hsa_op_with_type::get_in_type): Use it.
4843 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
4844 immediates in an assert.
4845 (hsa_op_with_type::extend_int_to_32bit): New method.
4846 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
4847 types. Convert to dest type if necessary.
4848 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
4849 (reg_for_gimple_ssa): Pass false as min32int to
4850 hsa_type_for_scalar_tree_type.
4851 (gen_hsa_addr): Fixup type when creating addresable temporary.
4852 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
4853 (gen_hsa_unary_operation): Extend operands and convert to dest type if
4854 necessary. Call hsa_fixup_mov_insn_type.
4855 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
4856 extend operands and convert to dest type if necessary.
4857 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
4858 to dest type if necessary.
4859 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
4860 if conversion nt necessary and size matches.
4861 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
4862 to dest type if necessary.
4863 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
4864 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
4866 (gen_hsa_clrsb): Likewise.
4867 (gen_hsa_ffs): Likewise.
4868 (gen_hsa_divmod): Extend operands and convert to dest type if
4870 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
4872 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
4874 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
4875 Remove empty default arguments. Use a brace block as output
4877 (conditional return): Ditto.
4879 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
4880 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
4882 (group_ending_nop): Ditto.
4883 (doloop_end): Ditto.
4884 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
4885 (splitters for those): Ditto.
4887 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
4889 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
4890 a conditional jump (and the compare for it) so that pc_rtx is the
4892 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
4893 for the deleted and renamed ctr<mode>_internal[234] patterns.
4894 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
4895 Delete second conditional return pattern.
4896 (ctr<mode>_internal2): Delete this second bdnz pattern.
4897 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
4898 (ctr<mode>_internal4): Delete this second bdz pattern.
4900 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
4902 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
4903 (always_initialized_rtx_for_ssa_name_p): New predicate.
4904 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
4905 (finish_out_of_ssa): Free new field of SA.
4906 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
4907 * tree-ssa-coalesce.c: Include tree-ssa.h.
4908 (get_parm_default_def_partitions): Remove extern keyword.
4909 (get_undefined_value_partitions): New function.
4910 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
4911 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
4913 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
4915 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
4917 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
4919 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
4921 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
4922 for targets that preffer 128bit.
4924 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
4926 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
4928 2017-10-08 Olivier Hainque <hainque@adacore.com>
4930 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
4931 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
4933 (thumb_set_return_address): Likewise.
4935 2017-10-08 Olivier Hainque <hainque@adacore.com>
4937 * common/config/arm/arm-common.c (arm_except_unwind_info):
4938 Handle DWARF2_UNWIND_INFO.
4940 2017-10-07 Michael Collison <michael.collison@arm.com>
4942 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
4945 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
4947 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
4948 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
4949 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
4950 defined, force the creation of a new block for a dispatch label.
4952 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
4954 * invoke.texi (Wsuggest-attribute=cold): Document.
4955 * common.opt (Wsuggest-attribute=cold): New
4956 * ipa-pure-const.c (warn_function_cold): New function.
4957 * predict.c (compute_function_frequency): Use it.
4958 * predict.h (warn_function_cold): Declare.
4960 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
4962 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
4965 2017-10-06 Martin Liska <mliska@suse.cz>
4967 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
4968 keyword for member functions.
4969 (struct sanopt_tree_couple): New struct.
4970 (struct sanopt_tree_couple_hash): New function.
4971 (struct sanopt_ctx): Add new hash_map.
4972 (has_dominating_ubsan_ptr_check): New function.
4973 (record_ubsan_ptr_check_stmt): Likewise.
4974 (maybe_optimize_ubsan_ptr_ifn): Likewise.
4975 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
4976 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
4978 2017-10-06 Sudakshina Das <sudi.das@arm.com>
4981 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
4982 aarch64_simd_valid_immediate on CONST_VECTORs.
4983 (aarch64_reg_or_bic_imm): Likewise.
4985 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
4987 PR rtl-optimization/82396
4988 * haifa-sched.c (ready_sort_real): Disable qsort checking.
4990 2017-10-06 Sebastian Pop <sebpop@gmail.com>
4992 * graphite-dependences.c (scop_get_reads): Move code to...
4993 (scop_get_must_writes): Move code to...
4994 (scop_get_may_writes): Move code to...
4995 (scop_get_reads_and_writes): ... here.
4996 (scop_get_dependences): Call scop_get_reads_and_writes.
4998 2017-10-06 Jakub Jelinek <jakub@redhat.com>
5000 PR tree-optimization/82434
5001 * fold-const.h (can_native_encode_type_p,
5002 can_native_encode_string_p): Remove.
5003 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
5004 don't encode anything, just return what would be otherwise returned.
5005 (native_encode_fixed, native_encode_complex, native_encode_vector):
5007 (native_encode_string): Likewise. Inline by hand
5008 can_native_encode_string_p.
5009 (can_native_encode_type_p): Remove.
5010 (can_native_encode_string_p): Remove.
5011 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
5012 STRING_CSTs using can_native_encode_string_p, test all
5013 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
5014 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
5015 argument from native_encode_expr.
5016 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
5017 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
5020 2017-10-06 Richard Biener <rguenther@suse.de>
5022 PR tree-optimization/82397
5023 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
5024 operand_equal_p but rely on data_ref_compare_tree for detecting
5026 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
5027 to match up with dr_group_sort_cmp.
5029 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5032 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
5034 * config/s390/s390-builtin-types.def: Regenerate.
5036 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5039 * config/s390/s390-builtin-types.def: Regenerate.
5040 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
5041 Change flag from B_VXE to B_VX.
5042 (s390_vec_min_dbl): Remove B_VXE flag.
5044 2017-10-06 Richard Biener <rguenther@suse.de>
5046 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
5047 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
5048 translate_isl_ast_to_gimple::is_valid_rename,
5049 translate_isl_ast_to_gimple::get_rename,
5050 translate_isl_ast_to_gimple::get_def_bb_for_const,
5051 translate_isl_ast_to_gimple::get_new_name,
5052 translate_isl_ast_to_gimple::collect_all_ssa_names,
5053 translate_isl_ast_to_gimple::copy_loop_phi_args,
5054 translate_isl_ast_to_gimple::collect_all_ssa_names,
5055 translate_isl_ast_to_gimple::copy_loop_phi_args,
5056 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
5057 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
5058 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
5059 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
5060 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
5061 translate_isl_ast_to_gimple::copy_cond_phi_args,
5062 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
5063 translate_isl_ast_to_gimple::edge_for_new_close_phis,
5064 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
5065 translate_isl_ast_to_gimple::rename_uses,
5066 translate_isl_ast_to_gimple::rename_all_uses): Remove.
5067 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
5068 (set_rename_for_each_def): Likewise.
5069 (graphite_copy_stmts_from_block): Handle debug stmt resetting
5070 here. Handle rewriting SCEV analyzable uses here.
5071 (copy_bb_and_scalar_dependences): Generate code for PHI
5073 (graphite_regenerate_ast_isl): Adjust.
5074 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
5075 (add_write, add_read): New functions.
5076 (build_cross_bb_scalars_def): Use it and simplify.
5077 (build_cross_bb_scalars_use): Likewise.
5078 (graphite_find_cross_bb_scalar_vars): Inline into...
5079 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
5080 simulating out-of-SSA. Compute liveout and add dependencies.
5081 (build_scops): Force an empty entry block.
5082 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
5084 (sese_build_liveouts): Declare.
5085 (sese_trivially_empty_bb_p): Likewise.
5086 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
5087 compute liveout and debug_liveout.
5088 (sese_bad_liveouts_use): Remove.
5089 (sese_reset_debug_liveouts_bb): Likewise.
5090 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
5091 (sese_build_liveouts): Build liveout and debug_liveout and store
5093 (new_sese_info): Adjust.
5094 (free_sese_info): Likewise.
5095 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
5096 do not build liveout here.
5097 (move_sese_in_condition): Adjust region entry.
5098 (scev_analyzable_p): Match up with chrec_apply requirements.
5099 (sese_trivially_empty_bb_p): New.
5100 * tree-into-ssa.c (get_reaching_def): Properly support generating
5101 default-defs for incremental rewrite of anonymous names.
5103 2017-10-06 Richard Biener <rguenther@suse.de>
5105 * graphite-sese-to-poly.c (extract_affine): For casts increasing
5106 precision do not perform modulo reduction.
5108 2017-10-06 Richard Biener <rguenther@suse.de>
5110 PR tree-optimization/82436
5111 * tree-vect-slp.c (vect_supported_load_permutation_p): More
5112 conservatively choose the vectorization factor when checking
5113 whether we can perform the required load permutation.
5114 (vect_transform_slp_perm_load): Assert when we may not fail.
5116 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
5118 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
5119 message for incompatible -msdata=* and -mcall-* options.
5121 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5123 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
5124 rate for post-reload scheduling.
5126 2017-10-05 Tamar Christina <tamar.christina@arm.com>
5128 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
5130 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5132 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
5133 to improve monte carlo in scimark.
5135 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5137 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
5138 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
5139 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
5140 pentium4_cost, nocona_cost): Set reassociation width to 1.
5141 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
5142 width to 2 for fp operations and 1 otherwise.
5143 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
5145 (atom_cost): Set reassociation width to 2.
5146 (slm_cost, generic_cost): Set fp reassociation width
5147 to 2 and 1 otherwise.
5148 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
5149 (core_cost): Set fp reassociation width to 4 and vector to 2.
5150 (ix86_reassociation_width): Rewrite using cost table; special case
5151 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
5152 and TARGET_AVX128_OPTIMAL.
5153 * config/i386/i386.h (processor_costs): Add
5154 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
5155 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
5156 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
5157 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
5158 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
5159 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
5161 2017-10-05 Nathan Sidwell <nathan@acm.org>
5163 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
5165 2017-10-05 Tamar Christina <tamar.christina@arm.com>
5167 * config/arm/arm.c (arm_test_fpu_data): New.
5168 (arm_run_selftests): Call arm_test_fpu_data.
5170 2017-10-04 Nathan Sidwell <nathan@acm.org>
5172 * toplev.c (toplev::main): Remove excess parens on pretty_printer
5174 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
5176 2017-10-04 Sudakshina Das <sudi.das@arm.com>
5178 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
5179 check type for aarch64_simd_valid_immediate.
5180 (aarch64_output_simd_mov_immediate): Update prototype.
5181 (aarch64_simd_valid_immediate): Update prototype.
5182 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
5183 support for ORR-immediate.
5184 (and<mode>3): modified pattern to add support for BIC-immediate.
5185 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
5186 now checks for valid immediate for BIC and ORR based on new enum
5188 (aarch64_output_simd_mov_immediate): Function now used to output
5189 BIC/ORR imm as well based on new enum argument.
5190 * config/aarch64/constraints.md (Do): New vector immediate constraint.
5192 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
5193 (aarch64_reg_or_bic_imm): Likewise.
5195 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5197 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
5198 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
5200 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
5204 PR rtl-optimization/82396
5205 * haifa-sched.c (autopref_multipass_init): Simplify
5207 (autopref_rank_data): Simplify sort order.
5208 * sched-int.h (autopref_multipass_data_): Remove
5209 multi_mem_insn_p, min_offset and max_offset.
5211 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5213 * doc/sourcebuild.texi: Document vect_peeling_profitable.
5215 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5217 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
5220 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5222 * doc/sourcebuild.texi: Document vect_long_mult.
5224 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
5226 PR tree-optimization/82413
5227 * fold-const.c (build_range_check): Use widest_int when comparing
5228 the maximum ETYPE value with HIGH.
5230 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
5232 PR rtl-optimization/82396
5233 * haifa-sched.c (autopref_multipass_init): Simplify
5235 (autopref_rank_data): Simplify sort order.
5236 * sched-int.h (autopref_multipass_data_): Remove
5237 multi_mem_insn_p, min_offset and max_offset.
5239 2017-10-04 Jakub Jelinek <jakub@redhat.com>
5241 PR tree-optimization/82381
5242 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
5243 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
5246 PR tree-optimization/82374
5247 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
5248 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
5249 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
5250 current_function_decl to the new decl.
5252 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
5254 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
5255 helper macro for IEEE float128 hardware built-in functions.
5256 (SQRTF128_ODD): Add built-in functions with the round-to-odd
5258 (TRUNCF128_ODD): Likewise.
5259 (ADDF128_ODD): Likewise.
5260 (SUBF128_ODD): Likewise.
5261 (MULF128_ODD): Likewise.
5262 (DIVF128_ODD): Likewise.
5263 (FMAF128_ODD): Likewise.
5264 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
5265 UNSPEC_TRUNC_ROUND_TO_ODD.
5266 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
5267 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
5268 floating point round to odd instructions.
5269 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
5270 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
5271 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
5272 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
5273 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
5274 (trunc<mode>sf2_hw): Change the truncate with round to odd
5275 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
5276 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
5277 to odd hardware instructions.
5278 (sub<mode>3_odd): Likewise.
5279 (mul<mode>3_odd): Likewise.
5280 (div<mode>3_odd): Likewise.
5281 (sqrt<mode>2_odd): Likewise.
5282 (fma<mode>4_odd): Likewise.
5283 (fms<mode>4_odd): Likewise.
5284 (nfma<mode>4_odd): Likewise.
5285 (nfms<mode>4_odd): Likewise.
5286 (trunc<mode>df2_odd): Change the truncate with round to odd
5287 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
5289 * doc/extend.texi (PowerPC built-in functions): Update documentation
5290 for existing IEEE float128-bit built-in functions. Add built-in
5291 functions that generate the IEEE 128-bit floating point round to
5294 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
5296 PR rtl-optimization/77729
5297 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
5298 to (X&(C1&~C2))|C2 transformations.
5300 2017-10-03 Martin Jambor <mjambor@suse.cz>
5302 PR tree-optimization/82363
5303 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
5304 mismatch, mark lacc written regardless of racc.
5306 2017-10-03 Jakub Jelinek <jakub@redhat.com>
5308 PR tree-optimization/82381
5309 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
5310 stmt_to_insert nor wheather SSA_NAMEs are default defs.
5311 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
5312 fallthrough into reassoc_stmt_dominates_stmt_p.
5315 * combine.c (combine_instructions): Don't combine in unreachable
5318 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
5321 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
5322 function to not use the have_cpu variable. Do not set cpu_index,
5323 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
5325 (rs6000_valid_attribute_p): Remove duplicate initializations of
5326 old_optimize and func_optimize.
5327 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
5328 (rs6000_activate_target_options): Make global.
5329 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
5332 2017-10-02 Jakub Jelinek <jakub@redhat.com>
5334 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
5335 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
5336 Set *poffset to 0 and *psize and *pmax_size to -1 if
5337 *poffset + *psize overflows in HOST_WIDE_INT.
5339 PR tree-optimization/82387
5340 PR tree-optimization/82388
5341 PR tree-optimization/82389
5342 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
5343 instead of live_bytes non-NULL.
5345 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
5348 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
5351 2017-10-02 Richard Biener <rguenther@suse.de>
5353 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
5354 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
5355 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
5357 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
5359 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
5360 requested precision matches the type's.
5361 * calls.c (alloc_max_size): Calculate the new candidate size as
5362 a widest_int and use wi::to_widest when comparing it with the
5363 current candidate size.
5364 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
5365 zero rather than integer_zero_node.
5366 * match.pd: Check for a no-op conversion before using wi::add
5367 rather than after. Use tree_to_uhwi when summing small shift
5368 counts into an unsigned int.
5370 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
5371 Alan Hayward <alan.hayward@arm.com>
5372 David Sherwood <david.sherwood@arm.com>
5375 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
5376 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
5377 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
5378 POINTER_AND_FP_REGS.
5380 2017-10-02 Richard Biener <rguenther@suse.de>
5382 PR tree-optimization/82355
5383 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5384 a mapping for the enclosing loop but avoid generating one for
5386 (copy_bb_and_scalar_dependences): Remove premature codegen
5387 error on PHIs in blocks duplicated into multiple places.
5388 * graphite-scop-detection.c
5389 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5390 in the region use it as loop and nest to analyze the DR in.
5391 (try_generate_gimple_bb): Likewise.
5392 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5393 (add_loop_constraints): For blocks in a loop not in the region
5394 create a dimension with a single iteration.
5395 * sese.h (gbb_loop_at_index): Remove assert.
5397 2017-10-01 Kevin Buettner <kevinb@redhat.com>
5399 * omp-expand.c (adjust_context_scope): New function.
5400 (expand_parallel_call): Call adjust_context_scope.
5402 2017-10-01 Jeff Law <law@redhat.com>
5404 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
5405 dom_opt_dom_walker class with direct access to private members.
5406 Add comments. Call test_for_singularity.
5407 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
5408 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
5409 m_dummy_cond anymore.
5410 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
5412 (pass_dominator:execute): Build the dummy_cond here and pass it
5413 to the dom_opt_dom_walker ctor.
5414 (test_for_singularity): New function.
5416 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
5417 Maya Rashish <coypu@sdf.org>
5419 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
5420 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
5421 (alpha*-*-netbsd*) Use nbsd_tm_file.
5422 (arm*-*-netbsdelf*) Likewise.
5423 (i[34567]86-*-netbsdelf*) Likewise.
5424 (x86_64-*-netbsd*) Likewise.
5425 (mips*-*-netbsd*) Likewise.
5426 (powerpc-*-netbsd*) Likewise.
5427 (sh*-*-netbsd*) Likewise.
5428 (sparc-*-netbsdelf*) Likewise.
5429 (sparc64-*-netbsd*) Likewise.
5430 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
5432 (vax-*-netbsdelf*) Likewise.
5433 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
5434 (UINT_FAST8_TYPE) Likewise.
5435 (INT_FAST16_TYPE) Check CHAR_FAST16.
5436 (UINT_FAST16_TYPE) Likewise.
5438 2017-09-30 Jakub Jelinek <jakub@redhat.com>
5441 * config/i386/i386.md
5442 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
5443 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
5444 *divmodsi4_zext_2): New define_insn_and_split.
5445 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
5446 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
5447 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
5448 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
5449 New define_insn_and_split.
5450 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
5451 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
5452 operands[1] having DImode when mode is SImode.
5454 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
5455 always SImode for DIV and MOD in REG_EQUAL notes.
5457 2017-09-29 Yury Gribov <tetra2005@gmail.com>
5460 * match.pd: Fix handling of NaNs in pattern.
5462 2017-09-29 Jeff Law <law@redhat.com>
5464 * sbitmap.c (bitmap_bit_in_range_p): New function.
5465 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
5466 * tree-ssa-dse.c (live_bytes_read): New function.
5467 (dse_classify_store): Ignore reads of dead bytes.
5469 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
5470 typos and whitespace errors.
5471 * config/i386/predicates.md (address_no_seg_operand): Likewise.
5472 * config/s390/s390.c (s390_emit_prologue): Likewise.
5474 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
5477 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
5478 with a symbol for LRA.
5480 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
5482 PR rtl-optimization/82338
5483 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
5485 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
5487 * genmodes.c (calc_wider_mode): Suppress qsort macro.
5488 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
5489 (qsort_chk): Declare.
5490 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
5491 (qsort_chk): New function.
5493 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5495 PR tree-optimization/82337
5496 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
5497 phi definition if the PHI result appears in an abnormal PHI.
5498 (find_basis_for_base_expr): Don't record a basis if the LHS of the
5499 basis appears in an abnormal PHI.
5501 2017-09-29 Richard Biener <rguenther@suse.de>
5503 * graphite-isl-ast-to-gimple.c
5504 (translate_isl_ast_to_gimple::set_codegen_error): New function.
5505 (binary_op_to_tree): Use it.
5506 (get_rename_from_scev): Likewise.
5507 (copy_loop_phi_nodes): Likewise.
5508 (copy_bb_and_scalar_dependences): Likewise.
5509 (translate_pending_phi_nodes): Likewise.
5511 2017-09-29 Jakub Jelinek <jakub@redhat.com>
5514 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
5515 for movabsq $(i32 << shift), r64.
5517 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
5519 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
5520 index when encoding %esp as %rsp to avoid 0x67 prefix.
5522 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
5524 * config/i386/i386.md (*movsf_internal, *movdf_internal):
5525 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
5527 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
5529 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
5530 Extensions with more than 16 double VFP registers.
5531 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
5532 to_clear_mask and all code related to it. Replace the remaining
5533 entry by a sbitmap and adapt code accordingly.
5535 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
5537 * brig-builtins.def: Change pure attributes to const.
5539 2017-09-28 Joseph Myers <joseph@codesourcery.com>
5541 * config.gcc (default_gnu_indirect_function): Default to yes for
5542 sparc*-*-linux* with glibc.
5544 2017-09-28 Joseph Myers <joseph@codesourcery.com>
5546 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
5547 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
5548 when creating .init_array and .fini_array sections with priority
5551 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
5554 * config/aarch64/aarch64.c
5555 (aarch64_builtin_support_vector_misalignment): Always return false
5556 when misalignment is unknown.
5558 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
5560 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
5561 this function to return false if the definition used by the swap
5562 instruction is artificial, or if the memory address from which the
5563 constant value is loaded is not represented by a base address held
5564 in a register or if the base address register is a frame or stack
5565 pointer. Additionally, return false if the base address of the
5566 loaded constant is a SYMBOL_REF but is not considered to be a
5568 (replace_swapped_load_constant): New function.
5569 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
5570 loaded constant vector with a load of a swapped constant vector.
5572 2017-09-27 Carl Love <cel@us.ibm.com>
5574 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
5575 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
5576 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
5579 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
5581 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
5582 first, always call autopref_rank_data otherwise.
5584 2017-09-27 Richard Biener <rguenther@suse.de>
5586 * graphite-scop-detection.c (find_scop_parameters): Move
5587 loop bound handling ...
5588 (gather_bbs::before_dom_children): ... here, avoiding the need
5589 to build scop_info->loop_nest.
5590 (record_loop_in_sese): Remove.
5591 * sese.h (sese_info_t::loop_nest): Remove.
5592 * sese.c (new_sese_info): Do not allocate loop_nest.
5593 (free_sese_info): Do not free loop_nest.
5595 2017-09-27 Jakub Jelinek <jakub@redhat.com>
5598 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
5599 lhs from calls if the lhs has addressable type.
5601 2017-09-27 Richard Biener <rguenther@suse.de>
5603 * graphite.h (scop::max_alias_set): New member.
5604 * graphite-scop-detection.c: Remove references to non-existing
5605 --param in comments.
5606 (build_alias_sets): Record the maximum alias set used for drs.
5607 (build_scops): Support zero as unlimited for
5608 --param graphite-max-arrays-per-scop.
5609 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
5611 (build_poly_sr_1): ... here. Compute alias set based on the
5612 maximum alias set used for drs rather than
5613 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
5615 2017-09-27 Richard Biener <rguenther@suse.de>
5617 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
5618 --param loop-block-tile-size=0 to disable tiling.
5620 2017-09-27 Richard Biener <rguenther@suse.de>
5622 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
5623 (graphite-max-nb-scop-params): Document special value zero.
5624 * domwalk.h (dom_walker::STOP): New symbolical constant.
5625 (dom_walker::dom_walker): Add optional parameter for bb to
5627 (dom_walker::~dom_walker): Declare.
5628 (dom_walker::before_dom_children): Document STOP return value.
5629 (dom_walker::m_user_bb_to_rpo): New member.
5630 (dom_walker::m_bb_to_rpo): Likewise.
5631 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
5632 mapping here if not provided by the user.
5633 (dom_walker::~dom_walker): Free bb to RPO mapping if not
5634 provided by the user.
5635 (dom_walker::STOP): Define.
5636 (dom_walker::walk): Do not compute bb to RPO mapping here.
5637 Support STOP return value from before_dom_children to stop
5639 * graphite-optimize-isl.c (optimize_isl): If the schedule
5640 is the same still generate code if -fgraphite-identity
5641 or -floop-parallelize-all are given.
5642 * graphite-scop-detection.c: Include cfganal.h.
5643 (gather_bbs::gather_bbs): Get and pass through bb to RPO
5645 (gather_bbs::before_dom_children): Return STOP for BBs
5647 (build_scops): Compute bb to RPO mapping and pass it to
5648 the domwalk. Treat --param graphite-max-nb-scop-params=0
5649 as not limiting the number of params.
5650 * graphite.c (graphite_initialize): Remove limit on the
5651 number of basic-blocks in a function.
5652 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
5653 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
5654 default value of 10.
5656 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5658 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
5659 Adjust code to eliminate needing to do the shift right 32-bits
5660 operation after XSCVDPSPN.
5662 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5664 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
5665 ((X / Y) != 0 -> X >= Y): Likewise.
5667 2017-09-26 Carl Love <cel@us.ibm.com>
5669 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
5670 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
5671 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
5672 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
5673 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
5674 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
5675 definitions and overloading.
5676 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
5677 statement for P9V_BUILTIN_XST_LEN_R.
5678 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
5679 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
5680 define_expand and define_insn for the instructions and builtins.
5681 * doc/extend.texi: Update the built-in documentation file for the new
5683 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
5684 define_insn for the instructions
5686 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
5689 * gcc/config/netbsd-protos.h: New file.
5690 * gcc/config/netbsd.c: New file.
5691 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
5692 * gcc/config/t-netbsd: New file.
5693 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
5694 (tmake_file) Add t-netbsd.
5695 (extra_objs) Add netbsd.o.
5697 2017-09-26 Janus Weil <janus@gcc.gnu.org>
5701 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
5703 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5705 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
5706 sign extension from a vector register to a GPR by doing a 32-bit
5707 direct move and then an EXTSW.
5708 (extendsi<mode>2 splitter): Likewise.
5709 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
5710 right or vector extract after doing XSCVDPSPN. Use
5711 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
5713 (movdi_from_sf_zero_ext): Likewise.
5714 (reload_gpr_from_vsxsf): Likewise.
5715 (p8_mfvsrd_4_disf): Delete, no longer used.
5716 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
5717 then needing to move the SFmode to a GPR to use the XSCVDPSP
5718 instruction instead of FRSP and XSCVDPSPN.
5719 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
5720 it is adjacent to the other XSCVSPDP insns.
5721 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
5722 SFmode to be in traditional Altivec registers.
5723 (vsx_xscvdpspn): Eliminate useless alternative constraint.
5724 (vsx_xscvspdpn): Likewise.
5725 (vsx_xscvspdpn_scalar): Likewise.
5727 2017-09-26 Martin Jambor <mjambor@suse.cz>
5729 * tree-sra.c (compare_access_positions): Put integral types first,
5730 stabilize sorting of integral types, remove conditions putting
5731 non-full-precision integers last.
5732 (sort_and_splice_var_accesses): Disable scalarization if a
5733 non-integert would be represented by a non-full-precision integer.
5735 2017-09-26 Joseph Myers <joseph@codesourcery.com>
5737 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
5738 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
5739 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
5740 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
5741 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
5742 conditionals inside the function instead of around it. Call
5743 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
5744 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
5746 2017-09-26 Richard Biener <rguenther@suse.de>
5748 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
5750 (scop_detection::build_scop_breadth): ... this. Removed.
5751 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
5752 (scop_detection::harmful_stmt_in_bb): Likewise.
5753 (scop_detection::graphite_can_represent_stmt): Likewise.
5754 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
5755 (scop_detection::can_represent_loop): Remove recursion, fold in ...
5756 (scop_detection::can_represent_loop_1): ... this. Removed.
5757 (scop_detection::harmful_loop_in_region): Simplify after inlining
5758 the above and remove more quadraticness.
5759 (build_scops): Adjust.
5760 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
5763 2017-09-26 Jakub Jelinek <jakub@redhat.com>
5766 * config/i386/i386.c (ix86_print_operand_address_as): Only test
5767 REGNO (base) == SP_REG if base is a REG.
5770 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
5771 if it is different SSA_NAME.
5772 (optimize_range_tests_cmp_bitwise): New function.
5773 (optimize_range_tests): Call it.
5775 2017-09-26 Richard Biener <rguenther@suse.de>
5777 PR tree-optimization/82321
5778 * graphite.c (canonicalize_loop_closed_ssa): Properly check
5779 for the def being inside the loop.
5781 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5783 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
5785 * config/s390/s390-builtins.def: Fix constraint on op4.
5787 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5789 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
5790 independent expanders.
5791 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
5792 ("vec_ordered", "vec_unordered"): New expanders.
5794 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5796 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
5799 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5801 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
5802 vec_unpacks_lo_v16qi.
5803 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
5805 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5807 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
5808 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
5809 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
5811 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5813 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
5815 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
5817 ("*vec_slb<mode>"): New insn pattern.
5818 ("vec_shr_<mode>"): New expander.
5819 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
5820 and force the shift count operand to V16QImode.
5821 ("vec_srb<mode>"): Set shift count mode to V16QI.
5823 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5825 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
5826 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
5827 ("vec_widen_smult_hi_<mode>"): New expander definitions.
5829 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
5832 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
5834 2017-09-26 Richard Biener <rguenther@suse.de>
5836 PR tree-optimization/82320
5837 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
5840 2017-09-25 Jeff Law <law@redhat.com>
5842 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
5843 prototype for new argument.
5844 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
5845 mostly extracted from rs6000_emit_allocate_stack.
5846 (rs6000_emit_probe_stack_range_stack_clash): New function.
5847 (rs6000_emit_allocate_stack): Call
5848 rs6000_emit_probe_stack_range_stack_clash as needed.
5849 (rs6000_emit_probe_stack_range): Add additional argument
5850 to call to gen_probe_stack_range{si,di}.
5851 (output_probe_stack_range): New.
5852 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
5853 (output_probe_stack_range_stack_clash): New.
5854 (rs6000_emit_prologue): Emit notes into dump file as requested.
5855 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
5856 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
5857 Add additional operand and pass it to output_probe_stack_range.
5859 2017-09-25 Bin Cheng <bin.cheng@arm.com>
5861 PR tree-optimization/82163
5862 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
5863 (checking_verify_loop_closed_ssa): New parameter.
5864 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
5865 (check_loop_closed_ssa_stmt): Delete.
5866 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
5867 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
5868 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
5871 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
5873 * brig-builtins.def: Treat HSAIL barrier builtins as
5874 setjmp/longjump style functions.
5876 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
5878 * target.def (constant_alignment): New hook.
5879 * defaults.h (CONSTANT_ALIGNMENT): Delete.
5880 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
5881 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
5882 * doc/tm.texi: Regenerate.
5883 * targhooks.h (default_constant_alignment): Declare.
5884 (constant_alignment_word_strings): Likewise.
5885 * targhooks.c (default_constant_alignment): New function.
5886 (constant_alignment_word_strings): Likewise.
5887 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
5888 instead of CONSTANT_ALIGNMENT.
5889 * varasm.c (align_variable, get_variable_align, build_constant_desc)
5890 (force_const_mem): Likewise.
5891 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
5892 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
5893 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
5894 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5895 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
5897 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
5898 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5899 constant_alignment_word_strings.
5900 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
5901 (CONSTANT_ALIGNMENT): Likewise.
5902 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5903 (arm_constant_alignment): New function.
5904 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
5905 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5906 constant_alignment_word_strings.
5907 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
5908 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5909 constant_alignment_word_strings.
5910 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
5911 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5912 (cris_constant_alignment): New function.
5913 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
5914 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5915 (epiphany_constant_alignment): New function.
5916 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
5917 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5918 constant_alignment_word_strings.
5919 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
5920 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5921 constant_alignment_word_strings.
5922 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
5923 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5924 constant_alignment_word_strings.
5925 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
5926 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
5927 * config/i386/i386.c (ix86_constant_alignment): Make static.
5928 Use the same interface as the target hook.
5929 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5930 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
5931 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5932 constant_alignment_word_strings.
5933 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
5934 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
5935 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5936 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
5937 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5938 constant_alignment_word_strings.
5939 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
5940 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5941 constant_alignment_word_strings.
5942 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
5943 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5944 constant_alignment_word_strings.
5945 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
5946 * config/microblaze/microblaze.c (microblaze_constant_alignment):
5948 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5949 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
5950 * config/mips/mips.c (mips_constant_alignment): New function.
5951 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5952 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
5953 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
5954 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5955 (mmix_constant_alignment): Make static. Use the same interface
5957 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
5958 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5959 constant_alignment_word_strings.
5960 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
5961 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5962 constant_alignment_word_strings.
5963 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
5964 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5965 constant_alignment_word_strings.
5966 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
5967 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5968 (rs6000_constant_alignment): New function.
5969 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
5970 * config/riscv/riscv.c (riscv_constant_alignment): New function.
5971 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5972 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
5973 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5974 (rs6000_constant_alignment): New function.
5975 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
5976 * config/s390/s390.c (s390_constant_alignment): New function.
5977 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5978 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
5979 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5980 constant_alignment_word_strings.
5981 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
5982 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5983 (sparc_constant_alignment): New function.
5984 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
5985 * config/spu/spu.c (spu_constant_alignment): New function.
5986 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5987 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
5988 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5989 constant_alignment_word_strings.
5990 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
5991 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5992 constant_alignment_word_strings.
5993 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
5994 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5995 constant_alignment_word_strings.
5996 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
5997 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5998 (visium_constant_alignment): New function.
5999 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
6000 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6001 (xtensa_constant_alignment): New function.
6002 * system.h (CONSTANT_ALIGNMENT): Poison.
6004 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
6006 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6007 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
6008 (rs6000_builtin_valid_without_lhs): New helper function.
6009 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6010 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
6012 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
6014 * target.h (vec_perm_indices): Use unsigned short rather than
6016 (auto_vec_perm_indices): Likewise.
6017 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
6018 Use unsigned int rather than unsigned char.
6019 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
6021 2017-09-25 Richard Biener <rguenther@suse.de>
6023 * cfgloop.h (sort_sibling_loops): Declare.
6024 * cfgloop.c (sort_sibling_loops_cmp): New helper.
6025 (sort_sibling_loops): New function sorting the sibling loop list
6027 * graphite.c (graphite_transform_loops): Sort sibling loops.
6029 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
6031 * target.def (vec_perm_const_ok): Change sel parameter to
6033 * optabs-query.c (can_vec_perm_p): Update accordingly.
6034 * doc/tm.texi: Regenerate.
6035 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
6036 auto_vec_perm_indices and remove separate nelt field.
6037 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
6038 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
6039 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
6040 (aarch64_expand_vec_perm_const): Update accordingly.
6041 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
6042 to vec_perm_indices.
6043 * config/arm/arm.c (expand_vec_perm_d): Change perm to
6044 auto_vec_perm_indices and remove separate nelt field.
6045 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
6046 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
6047 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
6049 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
6050 to vec_perm_indices.
6051 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
6052 sel to vec_perm_indices.
6053 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
6054 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
6055 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
6057 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
6060 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
6063 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
6064 on the FUNCTION_DECL function context if it has a DIE that is a
6067 2017-09-25 Richard Biener <rguenther@suse.de>
6069 PR tree-optimization/82285
6070 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
6073 2017-09-25 Tom de Vries <tom@codesourcery.com>
6077 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
6080 2017-09-25 Richard Biener <rguenther@suse.de>
6082 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
6083 ISL errors other than isl_error_quota happen. Dump if the
6084 schedule is the same.
6085 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
6086 errors instead of aborting inside ISL.
6088 2017-09-25 Iain Sandoe <iain@codesourcery.com>
6091 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
6092 of libgcc_eh for m64.
6093 * config/i386/darwin64.h: Likewise.
6095 2017-09-25 Richard Biener <rguenther@suse.de>
6098 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
6099 attribute for incomplete types nor twice for complete ones.
6101 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
6104 * config/i386/i386.c (ix86_print_operand_address_as): Encode
6105 %esp as %rsp to avoid 0x67 prefix if there is no index or base
6108 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
6111 * config/i386/i386.opt (mprefer-avx256): Use
6112 ix86_target_flags variable.
6113 * config/i386/i386.c (ix86_target_string): Move
6114 -mprefer-avx256 to flag2_opts.
6116 2017-09-22 Jakub Jelinek <jakub@redhat.com>
6119 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
6120 and x != -1 | y != -1 into (x & y) != -1.
6122 2017-09-22 Steve Ellcey <sellcey@cavium.com>
6124 * config.gcc: Add new case statement to set
6125 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
6126 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
6127 s390x-*-linux* case statements. Added aarch64 to the list of
6128 supported architectures.
6130 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6132 PR tree-optimization/82289
6133 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
6134 STMT_VINFO_RELEVANT_P.
6136 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6137 Alan Hayward <alan.hayward@arm.com>
6138 David Sherwood <david.sherwood@arm.com>
6140 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
6141 for VR_RANGE only; don't allow VR_ANTI_RANGE.
6142 (extract_range_from_binary_expr_1): Don't call
6143 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
6145 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6146 Alan Hayward <alan.hayward@arm.com>
6147 David Sherwood <david.sherwood@arm.com>
6149 * target.def (preferred_vector_alignment): New hook.
6150 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
6152 * doc/tm.texi: Regenerate.
6153 * targhooks.h (default_preferred_vector_alignment): Declare.
6154 * targhooks.c (default_preferred_vector_alignment): New function.
6155 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
6157 (DR_TARGET_ALIGNMENT): New macro.
6158 (aligned_access_p): Update commentary.
6159 (vect_known_alignment_in_bytes): New function.
6160 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
6162 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
6163 Calculate the misalignment based on the target alignment rather than
6165 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
6166 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
6167 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
6168 the target alignment, rather than masking the element misalignment
6169 with the number of elements in a vector. Also use the target
6170 alignment when calculating the maximum number of peels.
6171 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
6172 instead of TYPE_ALIGN_UNIT.
6173 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
6174 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
6175 (vect_create_addr_base_for_vector_ref): Update call accordingly.
6176 (vect_create_data_ref_ptr): Likewise.
6177 (vect_setup_realignment): Realign by ANDing with
6178 -DR_TARGET_MISALIGNMENT.
6179 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
6180 the number of peels based on DR_TARGET_ALIGNMENT.
6181 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
6182 with the guaranteed alignment boundary when deciding whether
6184 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
6185 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
6186 (ensure_base_align): Remove stmt_info parameter. Get the
6187 target base alignment from DR_TARGET_ALIGNMENT.
6188 (vectorizable_store): Update call accordingly. Interpret
6189 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
6191 (vectorizable_load): Likewise.
6193 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6194 Alan Hayward <alan.hayward@arm.com>
6195 David Sherwood <david.sherwood@arm.com>
6197 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
6198 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
6199 (vect_enhance_data_refs_alignment): Likewise.
6201 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
6203 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
6204 error. Only quit immediately if parsing is complete.
6205 (BEGIN): Initialize fatal_err and parse_done.
6206 (begin fpu, end fpu): Check number of arguments.
6207 (begin arch, end arch): Likewise.
6208 (begin cpu, end cpu): Likewise.
6209 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
6210 (optalias): Likewise.
6212 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
6214 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
6215 * config/arm/arm-isa.h: Delete. Move definitions to ...
6216 * arm-cpus.in: ... here. Use new feature and fgroup values.
6217 * config/arm/arm.c (arm_option_override): Use lower case for feature
6219 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
6220 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
6221 * config/arm/parsecpu.awk (END): Add new command 'isa'.
6223 (print_isa_bits_for): New function.
6224 (gen_isa): New function.
6225 (gen_comm_data): Use print_isa_bits_for.
6226 (define feature): New keyword.
6227 (define fgroup): New keyword.
6228 * config/arm/t-arm (TM_H): Remove.
6229 (GTM_H): Add arm-isa.h.
6230 (arm-isa.h): Add rule to generate file.
6231 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
6232 case for feature bit names.
6234 2017-09-22 Richard Biener <rguenther@suse.de>
6236 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
6238 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
6239 print of no dependency loops ...
6240 * graphite.c (graphite_transform_loops): ... here.
6241 (canonicalize_loop_closed_ssa_form): Work from inner to outer
6243 (same_close_phi_node, remove_duplicate_close_phi,
6244 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
6245 (canonicalize_loop_closed_ssa): ... here and simplify.
6246 * graphite-optimize-isl.c: Include tree-vectorizer.h.
6247 (optimize_isl): Use dump_printf_loc to tell when we stopped
6248 optimizing because of an ISL timeout.
6250 2017-09-22 Richard Biener <rguenther@suse.de>
6252 PR tree-optimization/82291
6253 * tree-if-conv.c (predicate_mem_writes): Make sure to
6254 remove writes in blocks predicated with false.
6256 2017-09-22 Richard Biener <rguenther@suse.de>
6258 * sese.c: Include cfganal.h.
6259 (if_region_set_false_region): Remove.
6260 (create_if_region_on_edge): Likewise.
6261 (move_sese_in_condition): Re-implement without destroying
6264 2017-09-22 Richard Biener <rguenther@suse.de>
6266 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
6267 Verify both BBs contain loop PHI nodes before dispatching to
6269 (graphite_regenerate_ast_isl): Do not recompute dominators,
6270 do not verify three times. Restructure for clarity.
6271 * graphite-scop-detection.c (same_close_phi_node,
6272 remove_duplicate_close_phi, make_close_phi_nodes_unique,
6273 defined_in_loop_p, canonicalize_loop_closed_ssa,
6274 canonicalize_loop_closed_ssa_form): Simplify, remove excess
6275 checking and SSA rewrite, move to ...
6276 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
6277 (graphite_initialize): Do not pass in ctx, do not reset the
6278 SCEV cache, compute only dominators.
6279 (graphite_transform_loops): Allocate ISL ctx after
6280 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
6281 Maintain post-dominators only around build_scops.
6282 * sese.c (if_region_set_false_region): Make static. Free
6283 and recompute dominators.
6284 (move_sese_in_condition): Assert we don't get called with
6285 post-dominators computed.
6286 * sese.h (if_region_set_false_region): Remove.
6288 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
6290 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
6291 mode attribute for TARGET_AVX512VL.
6293 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
6295 * config/i386/i386.opt (mprefer-avx256): New option.
6296 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
6298 (ix86_preferred_simd_mode): Return 256-bit AVX modes
6299 for TARGET_PREFER_AVX256.
6300 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
6302 2017-09-21 Jeff Law <law@redhat.com>
6304 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
6305 Fix dump output if the only stack space is for pushed registers.
6307 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6309 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
6312 2017-09-21 Martin Sebor <msebor@redhat.com>
6315 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
6316 code (in C++) or code that triggers warnings.
6318 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
6320 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
6322 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
6324 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
6325 * sched-rgn.c: Ditto.
6326 * sel-sched-ir.c: Ditto.
6328 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
6330 * toplev.h (set_random_seed): Adjust return type.
6331 * toplev.c (init_local_tick): Move eager initialization of random_seed
6332 to get_random_seed. Adjust comment.
6333 (init_random_seed): Inline to get_random_seed, delete.
6334 (get_random_seed): Initialize random_seed lazily.
6335 (set_random_seed): Do not return previous value.
6336 (print_switch_value): Do not call get_random_seed.
6338 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
6340 * cgraph.c (delete_function_version): New, broken out from...
6341 (cgraph_node::delete_function_version): ...here. Rename to
6342 cgraph_node::delete_function_version_by_decl. Update all uses.
6343 (cgraph_node::remove): Call delete_function_version.
6345 2017-09-21 Jakub Jelinek <jakub@redhat.com>
6348 * tree-inline.c (expand_call_inline): Emit clobber stmts for
6349 VAR_DECLs to which addressable non-volatile parameters are mapped
6350 and for id->retvar after the return value assignment. Clear
6351 id->retval and id->retbnd after inlining.
6353 2017-09-21 Richard Biener <rguenther@suse.de>
6355 PR tree-optimization/82276
6356 PR tree-optimization/82244
6357 * tree-vrp.c (build_assert_expr_for): Set
6358 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
6360 (remove_range_assertions): Revert earlier change.
6362 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
6365 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
6367 2017-09-21 Richard Biener <rguenther@suse.de>
6369 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
6370 Restore valid IL after code generation errors.
6371 * graphite.c (graphite_transform_loops): Diagnose code
6372 generation issues as MSG_MISSED_OPTIMIZATION and continue
6373 with processing SCOPs.
6375 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6376 Alan Hayward <alan.hayward@arm.com>
6377 David Sherwood <david.sherwood@arm.com>
6379 * calls.c (compute_argument_addresses): Use simplify_gen_binary
6380 rather than choosing between plus_constant and gen_rtx_<CODE>.
6381 * expr.c (emit_push_insn): Likewise.
6382 (expand_expr_real_2): Likewise.
6384 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6385 Alan Hayward <alan.hayward@arm.com>
6386 David Sherwood <david.sherwood@arm.com>
6388 * loop-unroll.c (split_iv): Call copy_rtx on the step.
6390 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6391 Alan Hayward <alan.hayward@arm.com>
6392 David Sherwood <david.sherwood@arm.com>
6394 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
6395 calling tree_to_uhwi.
6397 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6398 Alan Hayward <alan.hayward@arm.com>
6399 David Sherwood <david.sherwood@arm.com>
6401 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
6402 INTEGER_CST rather than a negative test for ADDR_EXPR.
6404 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6405 Alan Hayward <alan.hayward@arm.com>
6406 David Sherwood <david.sherwood@arm.com>
6408 * tree-vrp.c (extract_range_from_binary_expr_1): Check
6409 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
6411 2017-09-21 Richard Biener <rguenther@suse.de>
6413 PR tree-optimization/71351
6414 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
6415 graphite_create_new_loop_guard): Remove, fold remaining parts
6417 (translate_isl_ast_node_for): ... here and simplify.
6419 2017-09-21 Jakub Jelinek <jakub@redhat.com>
6422 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
6423 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
6424 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
6425 alternative always use QI mode, for -Os imov (=R,R) alternative
6426 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
6429 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6430 Jeff Law <law@redhat.com>
6432 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
6433 (allocate_stack_space): New function, partially extracted from
6435 (s390_emit_prologue): Track offset to most recent stack probe.
6436 Code to allocate space moved into allocate_stack_space.
6437 Dump actions when no stack is allocated.
6438 (s390_prologue_plus_offset): New function.
6439 (s390_emit_stack_probe): Likewise.
6441 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
6443 * common.opt (Wa, Wl, Wp, g, gz=): Add
6445 (gno-column-info): Remove.
6446 (gcolumn-info): Drop RejectNegative.
6448 (gno-record-gcc-switches): Remove.
6449 (grecord-gcc-switches): Drop RejectNegative.
6450 (gno-split-dwarf): Remove.
6451 (gsplit-dwarf): Drop RejectNegative.
6452 (gno-strict-dwarf): Remove.
6453 (gstrict-dwarf): Drop RejectNegative.
6454 * config/darwin.opt (gfull, gused): Add RejectNegative.
6455 * dwarf2out.c (gen_producer_string): Drop
6456 gno-record-gcc-switches handler.
6457 * optc-gen.awk: Add g to prefixes with negative forms.
6458 * opts-common.c (remapping_prefix_p): New.
6459 (find_opt): Check it.
6460 (generate_canonical_option): Test g prefix.
6461 (option_map): Add -gno- mapping.
6462 (add_misspelling_candidates): Check remapping_prefix_p.
6464 2017-09-20 Jeff Law <law@redhat.com>
6466 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
6467 thinko in stack clash protection support.
6469 * explow.c (compute_stack_clash_protection_loop_data): Use
6470 CONST_INT_P instead of explicit test. Verify object is a
6471 CONST_INT_P before looking at INTVAL.
6472 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
6473 instead of explicit test.
6475 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
6478 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
6479 address instead of to r1 and r11.
6481 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
6483 * config.gcc: Support "knm".
6484 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
6485 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
6487 * config/i386/i386.c (m_KNM): Define.
6488 (processor_target_table): Add "knm".
6490 (ix86_option_override_internal): Add "knm".
6491 (ix86_issue_rate): Add PROCESSOR_KNM.
6492 (ix86_adjust_cost): Ditto.
6493 (ia32_multipass_dfa_lookahead): Ditto.
6494 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
6495 (fold_builtin_cpu): Add M_INTEL_KNM.
6496 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
6497 (processor_type): Add PROCESSOR_KNM.
6498 * config/i386/x86-tune.def: Add m_KNM.
6499 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
6501 2017-09-20 Richard Biener <rguenther@suse.de>
6503 PR tree-optimization/80213
6504 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
6505 are allowed in empty BBs as well.
6506 (canonicalize_loop_closed_ssa): Also look for other complex
6508 (scop_detection::get_sese): Include the loop-closed PHI block
6510 (scop_detection::merge_sese): Remove code adding extra blocks.
6511 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
6512 (build_scops): Assert the final returned scop is invalid.
6514 2017-09-20 Richard Biener <rguenther@suse.de>
6516 PR tree-optimization/82264
6517 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
6519 (vn_phi_lookup): Likewise.
6520 (vn_phi_insert): Likewise.
6522 2017-09-20 Jakub Jelinek <jakub@redhat.com>
6524 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
6525 that fits into uhwi or shwi, add DW_AT_const_value regardless
6526 of early_dwarf without going through RTL, using add_AT_unsigned
6529 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
6530 (DEBUG_LTO_ABBREV_SECTION): Likewise.
6531 (DEBUG_LTO_MACINFO_SECTION): Likewise.
6532 (DEBUG_MACRO_SECTION): Likewise.
6533 (DEBUG_LTO_MACRO_SECTION): Likewise.
6534 (DEBUG_STR_DWO_SECTION): Likewise.
6535 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
6536 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
6537 (DEBUG_LTO_DWO_LINE_SECTION): Define.
6538 (DEBUG_LTO_LINE_STR_SECTION): Define.
6539 (init_sections_and_labels): Initialize debug_line_str_section
6540 variable. Initialize debug_loc_section for -gdwarf-5 to
6541 DEBUG_LOCLISTS_SECTION. Formatting fixes.
6543 2017-09-20 Richard Biener <rguenther@suse.de>
6545 * graphite-sese-to-poly.c (extract_affine): Properly handle
6546 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
6548 2017-09-20 Richard Biener <rguenther@suse.de>
6550 PR tree-optimization/81373
6551 * graphite-scop-detection.c (build_cross_bb_scalars_def):
6552 Force SESE live-out defs to be handled even if they are
6555 2017-09-19 Jeff Law <law@redhat.com>
6557 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
6558 nothing for stack adjustments with REG_STACK_CHECK.
6559 * sched-deps.c (parse_add_or_inc): Reject insns with
6560 REG_STACK_CHECK from dependency breaking.
6561 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
6562 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
6563 * reg-notes.def (STACK_CHECK): New note.
6565 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
6566 (ix86_expand_prologue): Dump stack clash info as needed.
6567 Call ix86_adjust_stack_and_probe_stack_clash as needed.
6569 * function.c (dump_stack_clash_frame_info): New function.
6570 * function.h (dump_stack_clash_frame_info): Prototype.
6571 (enum stack_clash_probes): New enum.
6573 * config/alpha/alpha.c (alpha_expand_prologue): Also check
6574 flag_stack_clash_protection.
6575 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
6576 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
6577 (arm_frame_pointer_required): Likewise.
6578 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
6579 (ia64_expand_prologue): Likewise.
6580 * config/mips/mips.c (mips_expand_prologue): Likewise.
6581 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
6582 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
6583 (sparc_flat_expand_prologue): Likewise.
6584 * config/spu/spu.c (spu_expand_prologue): Likewise.
6586 * explow.c: Include "params.h".
6587 (anti_adjust_stack_and_probe_stack_clash): New function.
6588 (get_stack_check_protect): Likewise.
6589 (compute_stack_clash_protection_loop_data): Likewise.
6590 (emit_stack_clash_protection_loop_start): Likewise.
6591 (emit_stack_clash_protection_loop_end): Likewise.
6592 (allocate_dynamic_stack_space): Use get_stack_check_protect.
6593 Use anti_adjust_stack_and_probe_stack_clash.
6594 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
6595 (emit_stack_clash_protection_loop_start): Likewise.
6596 (emit_stack_clash_protection_loop_end): Likewise.
6597 * rtl.h (get_stack_check_protect): Prototype.
6598 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
6599 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
6600 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
6602 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
6604 * doc/tm.texi: Rebuilt.
6605 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
6606 get_stack_check_protect.
6607 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
6608 * config/arm/arm.c (arm_expand_prologue): Likewise.
6609 (arm_frame_pointer_required): Likewise.
6610 * config/i386/i386.c (ix86_expand_prologue): Likewise.
6611 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
6612 * config/mips/mips.c (mips_expand_prologue): Likewise.
6613 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
6614 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
6615 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
6616 (sparc_flat_expand_prologue): Likewise.
6618 * common.opt (-fstack-clash-protection): New option.
6619 * flag-types.h (enum stack_check_type): Note difference between
6620 -fstack-check= and -fstack-clash-protection.
6621 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
6622 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
6623 * toplev.c (process_options): Issue warnings/errors for cases
6624 not handled with -fstack-clash-protection.
6625 * doc/invoke.texi (-fstack-clash-protection): Document new option.
6626 (-fstack-check): Note additional problem with -fstack-check=generic.
6627 Note that -fstack-check is primarily for Ada and refer users
6628 to -fstack-clash-protection for stack-clash-protection.
6629 Document new params for stack clash protection.
6631 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
6633 * config/i386/i386.c (ix86_split_long_move): Do not handle
6634 address used for LEA in a special way.
6636 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
6638 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
6640 2017-09-19 Martin Sebor <msebor@redhat.com>
6643 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
6644 of incompatible types.
6646 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
6648 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6649 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
6650 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6651 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
6653 2017-09-19 Richard Biener <rguenther@suse.de>
6655 PR tree-optimization/82244
6656 * tree-vrp.c (remove_range_assertions): Do not propagate
6657 a constant to abnormals but replace the assert with a copy.
6659 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
6661 PR rtl-optimization/57878
6662 PR rtl-optimization/68988
6663 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
6664 avoidance test involving non_reload_pseudos. Move frequency test
6665 below the general fragmentation avoidance test.
6667 2017-09-19 Richard Biener <rguenther@suse.de>
6669 PR tree-optimization/69728
6670 * graphite-sese-to-poly.c (schedule_error): New global.
6671 (add_loop_schedule): Handle empty domain by failing the
6673 (build_original_schedule): Handle schedule_error.
6675 2017-09-19 Richard Biener <rguenther@suse.de>
6677 * graphite-scop-detection.c (scop_detection::can_represent_loop):
6678 Do not iterate to sibling loops but only to siblings of inner
6681 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
6684 * config/m68k/m68k.md (moveq feeding equality comparison): Check
6685 that the registers are different.
6687 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
6689 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
6690 to processor_model and "amdfam17h" to arch_names_table.
6691 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
6693 2017-09-18 Jakub Jelinek <jakub@redhat.com>
6696 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
6698 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6699 Alan Hayward <alan.hayward@arm.com>
6700 David Sherwood <david.sherwood@arm.com>
6702 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
6704 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
6706 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
6707 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
6708 vect_schedule_slp_instance.
6709 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
6710 Update call to vect_slp_analyze_node_operations. Simplify return
6712 (vect_slp_analyze_bb_1): Update call accordingly.
6713 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
6714 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
6715 (vect_schedule_slp): Update call accordingly.
6717 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6718 Alan Hayward <alan.hayward@arm.com>
6719 David Sherwood <david.sherwood@arm.com>
6721 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
6722 with types that aren't in fact scalar.
6724 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6726 * tree-vect-slp.c (vect_record_max_nunits): New function,
6728 (vect_build_slp_tree_1): ...here.
6729 (vect_build_slp_tree_2): Call it for phis too.
6731 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6733 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
6734 to vect_get_vec_def_for_operand when getting the mask operand.
6736 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6737 Alan Hayward <alan.hayward@arm.com>
6738 David Sherwood <david.sherwood@arm.com>
6740 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
6743 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6744 Alan Hayward <alan.hayward@arm.com>
6745 David Sherwood <david.sherwood@arm.com>
6747 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
6748 calculation for vector booleans.
6750 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6751 Alan Hayward <alan.hayward@arm.com>
6752 David Sherwood <david.sherwood@arm.com>
6754 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
6756 (vect_transform_stmt): ...here.
6757 (vect_analyze_stmt): Use it instead of calling
6758 vectorizable_live_operation directly.
6760 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
6762 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
6763 non-SIMT targets in acc vector loops.
6765 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
6767 * configure.ac: Add arc and check if assembler supports gdwarf2.
6768 * configure: Regenerate.
6770 2017-09-18 Richard Biener <rguenther@suse.de>
6772 PR tree-optimization/82220
6773 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
6774 epilogue niters from the min_profitable_iters compute.
6776 2017-09-18 Jakub Jelinek <jakub@redhat.com>
6779 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
6780 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
6781 (ix86_init_pic_reg): Revert 2017-09-01 changes.
6783 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
6786 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
6787 switching to a new text section.
6789 2017-09-18 Richard Biener <rguenther@suse.de>
6791 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
6793 (build_alias_set): Reject aliases with no access function.
6795 2017-09-18 Richard Biener <rguenther@suse.de>
6797 PR tree-optimization/79622
6798 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
6800 (build_cross_bb_scalars_use): Likewise.
6802 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
6804 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
6806 2017-09-18 Alan Modra <amodra@gmail.com>
6809 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
6810 stack_pointer_rtx for count 0. Update comments. Break up
6811 large rtl expression.
6813 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
6815 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
6816 Increase to 20 bytes.
6817 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
6818 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
6819 or avx version of the stub.
6821 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
6824 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
6825 compute the minimum stack alignment. Also update preferred stack
6826 boundary for leaf functions.
6828 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
6830 PR tree-optimization/82228
6831 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
6834 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
6836 * common/config/nds32/nds32-common.c
6837 (nds32_option_optimization_table): Refine formatting.
6838 (nds32_option_optimization_table): Use -fsched-pressure and
6839 -fomit-frame-pointer for specific optimization level.
6841 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
6843 * config/nds32/nds32.c: Refine formatting and comments.
6844 * config/nds32/nds32.h: Likewise.
6845 * config/nds32/nds32.md: Likewise.
6846 * config/nds32/nds32-cost.c: Likewise.
6847 * config/nds32/nds32-isr.c: Likewise.
6848 * config/nds32/nds32-md-auxiliary.c: Likewise.
6849 * config/nds32/nds32-multiple.md: Likewise.
6850 * config/nds32/nds32-predicates.c: Likewise.
6852 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
6853 Jakub Jelinek <jakub@redhat.com>
6855 Add support for -std=c++2a.
6856 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
6858 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
6860 2017-09-15 Steve Ellcey <sellcey@cavium.com>
6863 * doc/extend.texi (Common Function Attributes): Add
6864 references to ARM, AArch64, and S/390 specific attributes.
6865 (Function Specific Option Pragmas): Add AArch64 and S/390
6866 to list of back ends that support the target pragma.
6868 2017-09-15 Nathan Sidwell <nathan@acm.org>
6870 * doc/standards.texi: Fix C++17 description. Update URLs for
6873 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
6875 * common.opt (Wcast-align=strict): New warning option.
6876 * doc/invoke.texi: Document -Wcast-align=strict.
6878 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
6880 * cgraph.h (cgraph_thunk_info): Add comments.
6881 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
6882 assert for VIRTUAL_* arguments stricter.
6884 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
6886 PR tree-optimization/71026
6887 * match.pd: Move RDIV patterns from fold-const.c
6888 * fold-const.c (distribute_real_division): Removed.
6889 (fold_binary_loc): Remove calls to distribute_real_divison.
6891 2017-09-15 Jakub Jelinek <jakub@redhat.com>
6893 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
6894 c++1z and gnu++1z as deprecated. Change other references to
6895 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
6896 Change -Wc++1z-compat to -Wc++17-compat.
6897 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
6898 * dwarf2out.c (highest_c_language): Handle C++17.
6899 (gen_compile_unit_die): Likewise.
6901 2017-09-15 Jakub Jelinek <jakub@redhat.com>
6903 PR rtl-optimization/82192
6904 * combine.c (make_extraction): Don't look through non-paradoxical
6905 SUBREGs or TRUNCATE if pos + len is or might be bigger than
6908 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
6909 Alan Hayward <alan.hayward@arm.com>
6910 David Sherwood <david.sherwood@arm.com>
6912 * target.def (function_arg_offset): New hook.
6913 * targhooks.h (default_function_arg_offset): Declare.
6914 * targhooks.c (default_function_arg_offset): New function.
6915 * function.c (locate_and_pad_parm): Use
6916 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
6917 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
6918 (TARGET_FUNCTION_ARG_OFFSET): ...this.
6919 * doc/tm.texi: Regenerate.
6920 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
6921 * config/spu/spu.c (spu_function_arg_offset): New function.
6922 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
6923 * system.h (FUNCTION_ARG_OFFSET): Poison.
6925 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
6926 Alan Hayard <alan.hayward@arm.com>
6927 David Sherwood <david.sherwood@arm.com>
6929 * target.def (truly_noop_truncation): New hook.
6930 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
6931 than TRULY_NOOP_TRUNCATION.
6932 * hooks.h (hook_bool_uint_uint_true): Declare.
6933 * hooks.c (hook_bool_uint_uint_true): New function.
6934 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
6935 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
6936 * doc/tm.texi: Regenerate.
6937 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
6938 rather than TRULY_NOOP_TRUNCATION in comments.
6939 (simplify_comparison): Likewise.
6940 (record_truncated_value): Likewise.
6941 * expmed.c (extract_bit_field_1): Likewise.
6942 (extract_split_bit_field): Likewise.
6943 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
6944 instead of TRULY_NOOP_TRUNCATION.
6945 * function.c (assign_parm_setup_block): Likewise.
6946 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
6947 * rtlhooks.c: Include target.h.
6948 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
6949 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
6950 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
6951 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
6952 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
6953 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
6954 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
6955 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
6956 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
6957 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
6958 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
6959 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
6960 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
6961 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
6962 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
6963 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
6964 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
6965 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
6966 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
6967 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
6968 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
6969 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
6970 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
6971 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
6972 * config/mips/mips.c (mips_truly_noop_truncation): New function.
6973 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
6974 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
6975 rather than TRULY_NOOP_TRUNCATION in comments.
6976 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
6977 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
6978 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
6979 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
6980 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
6981 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
6982 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
6983 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
6984 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
6985 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
6986 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
6987 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
6988 rather than TRULY_NOOP_TRUNCATION in comments.
6989 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
6990 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
6991 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
6992 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
6993 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
6994 TRULY_NOOP_TRUNCATION condition.
6995 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
6996 (TRULY_NOOP_TRUNCATION): Delete.
6997 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
6998 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
6999 * config/spu/spu.c (spu_truly_noop_truncation): New function.
7000 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7001 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
7002 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
7003 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
7004 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7005 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7006 rather than TRULY_NOOP_TRUNCATION in comments.
7007 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
7008 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
7009 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
7010 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
7011 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
7012 * system.h (TRULY_NOOP_TRUNCATION): Poison.
7014 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
7017 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
7018 (*cmp_ior): Likewise.
7019 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
7020 (*ior_scc_scc_cmp): Likewise.
7021 (*and_scc_scc): Likewise.
7022 (*and_scc_scc_cmp): Likewise.
7024 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7025 Alan Hayard <alan.hayward@arm.com>
7026 David Sherwood <david.sherwood@arm.com>
7028 * target.def (can_change_mode_class): New hook.
7029 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
7030 (hard_regno_nregs): Likewise.
7031 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
7032 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
7033 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
7034 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
7035 (LOAD_EXTEND_OP): Update accordingly.
7036 * doc/tm.texi: Regenerate.
7037 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
7038 CANNOT_CHANGE_MODE_CLASS.
7039 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
7040 (REG_CAN_CHANGE_MODE_P): ...this new macro.
7041 * combine.c (simplify_set): Update accordingly.
7042 * emit-rtl.c (validate_subreg): Likewise.
7043 * recog.c (general_operand): Likewise.
7044 * regcprop.c (mode_change_ok): Likewise.
7045 * reload1.c (choose_reload_regs): Likewise.
7046 (inherit_piecemeal_p): Likewise.
7047 * rtlanal.c (simplify_subreg_regno): Likewise.
7048 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
7049 instead of CANNOT_CHANGE_MODE_CLASS.
7050 (reload_cse_simplify_operands): Likewise.
7051 * reload.c (push_reload): Use targetm.can_change_mode_class
7052 instead of CANNOT_CHANGE_MODE_CLASS.
7053 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
7054 REG_CANNOT_CHANGE_MODE_P.
7055 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7056 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
7057 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7058 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7059 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7060 (arm_can_change_mode_class): New function.
7061 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
7062 than CANNOT_CHANGE_MODE_CLASS in comments.
7063 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7064 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
7065 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
7066 (ix86_can_change_mode_class): ...this new function, inverting the
7067 sense of the return value.
7068 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7069 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7070 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7071 (ia64_can_change_mode_class): New function.
7072 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7073 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
7074 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
7075 (m32c_can_change_mode_class): ...this new function, inverting the
7076 sense of the return value.
7077 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7078 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7079 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
7080 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
7081 (mips_can_change_mode_class): ...this new function, inverting the
7082 sense of the return value.
7083 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7084 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7085 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7086 (msp430_can_change_mode_class): New function.
7087 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7088 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
7089 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7090 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7091 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7092 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
7093 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7094 (pa_cannot_change_mode_class): Replace with...
7095 (pa_can_change_mode_class): ...this new function, inverting the
7096 sense of the return value.
7097 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
7098 than CANNOT_CHANGE_MODE_CLASS in comments.
7099 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7100 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
7101 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7102 (pdp11_cannot_change_mode_class): Replace with...
7103 (pdp11_can_change_mode_class): ...this new function, inverting the
7104 sense of the return value.
7105 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7106 * config/powerpcspe/powerpcspe-protos.h
7107 (rs6000_cannot_change_mode_class_ptr): Delete.
7108 * config/powerpcspe/powerpcspe.c
7109 (rs6000_cannot_change_mode_class_ptr): Delete.
7110 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7111 (rs6000_option_override_internal): Assign to
7112 targetm.can_change_mode_class instead of
7113 rs6000_cannot_change_mode_class_ptr.
7114 (rs6000_cannot_change_mode_class): Replace with...
7115 (rs6000_can_change_mode_class): ...this new function, inverting the
7116 sense of the return value.
7117 (rs6000_debug_cannot_change_mode_class): Replace with...
7118 (rs6000_debug_can_change_mode_class): ...this new function.
7119 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7120 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
7121 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7122 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7123 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
7125 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
7126 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7127 (rs6000_option_override_internal): Assign to
7128 targetm.can_change_mode_class instead of
7129 rs6000_cannot_change_mode_class_ptr.
7130 (rs6000_cannot_change_mode_class): Replace with...
7131 (rs6000_can_change_mode_class): ...this new function, inverting the
7132 sense of the return value.
7133 (rs6000_debug_cannot_change_mode_class): Replace with...
7134 (rs6000_debug_can_change_mode_class): ...this new function.
7135 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7136 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
7137 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
7138 (s390_can_change_mode_class): ...this new function, inverting the
7139 sense of the return value.
7140 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7141 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7142 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
7143 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7144 (sh_cannot_change_mode_class): Replace with...
7145 (sh_can_change_mode_class): ...this new function, inverting the
7146 sense of the return value.
7147 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7148 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7149 (sparc_can_change_mode_class): New function.
7150 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7151 * config/spu/spu.c (spu_can_change_mode_class): New function.
7152 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7153 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7154 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7155 (visium_can_change_mode_class): New function.
7156 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
7158 2017-09-15 Richard Biener <rguenther@suse.de>
7160 PR tree-optimization/82217
7161 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
7162 but not undefined case.
7164 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7167 * postreload.c (reload_cse_simplify_operands): Skip
7168 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
7170 2017-09-15 Richard Biener <rguenther@suse.de>
7172 PR tree-optimization/68823
7173 * graphite-scop-detection.c (build_alias_set): If we have a
7174 possible dependence check whether we can handle them by just
7175 looking at the DRs DR_ACCESS_FNs.
7176 (build_scops): If build_alias_set fails, fail the SCOP.
7178 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
7180 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
7181 to support float128 built-in functions that require the ISA 3.0
7183 (BU_FLOAT128_3_HW): Likewise.
7184 (SQRTF128): Add support for the IEEE 128-bit square root and fma
7186 (FMAF128): Likewise.
7188 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
7189 support for built-in functions that need the ISA 3.0 IEEE 128-bit
7190 floating point instructions.
7191 (rs6000_invalid_builtin): Likewise.
7192 (rs6000_builtin_mask_names): Likewise.
7193 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
7194 (RS6000_BTM_FLOAT128_HW): Likewise.
7195 (RS6000_BTM_COMMON): Likewise.
7196 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
7198 * doc/extend.texi (RS/6000 built-in functions): Document the
7199 IEEE 128-bit floating point square root and fused multiply-add
7202 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
7204 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
7205 reg (r2) isn't in the set of registers defined in the prologue.
7207 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7208 Alan Hayward <alan.hayward@arm.com>
7209 David Sherwood <david.sherwood@arm.com>
7211 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
7212 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
7213 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
7214 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
7215 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
7217 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
7218 max_vectorization_factor.
7219 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
7221 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7222 Alan Hayward <alan.hayward@arm.com>
7223 David Sherwood <david.sherwood@arm.com>
7225 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
7226 (vect_worthwhile_without_simd_p): Declare.
7227 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
7228 (vectorizable_reduction): Use it.
7229 * tree-vect-stmts.c (vectorizable_shift): Likewise.
7230 (vectorizable_operation): Likewise.
7232 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7233 Alan Hayward <alan.hayward@arm.com>
7234 David Sherwood <david.sherwood@arm.com>
7236 * tree-vectorizer.h (vect_get_num_copies): New function.
7237 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
7238 * tree-vect-loop.c (vectorizable_reduction): Likewise.
7239 (vectorizable_induction): Likewise.
7240 (vectorizable_live_operation): Likewise.
7241 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
7242 (vectorizable_bswap): Likewise.
7243 (vectorizable_call): Likewise.
7244 (vectorizable_conversion): Likewise.
7245 (vectorizable_assignment): Likewise.
7246 (vectorizable_shift): Likewise.
7247 (vectorizable_operation): Likewise.
7248 (vectorizable_store): Likewise.
7249 (vectorizable_load): Likewise.
7250 (vectorizable_condition): Likewise.
7251 (vectorizable_comparison): Likewise.
7252 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
7254 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7255 Alan Hayward <alan.hayward@arm.com>
7256 David Sherwood <david.sherwood@arm.com>
7258 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
7259 of vect_init_vector.
7261 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7262 Alan Hayward <alan.hayward@arm.com>
7263 David Sherwood <david.sherwood@arm.com>
7265 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
7266 an inline wrapper that provides a location.
7267 (gimple_build_vector): Likewise.
7268 * gimple-fold.c (gimple_build_vector_from_val): New function.
7269 (gimple_build_vector): Likewise.
7270 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
7271 functions to build the initial value. Always return a gimple value.
7272 (get_initial_defs_for_reduction): Likewise. Only compute
7274 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
7275 vect_init_vector on the results from get_initial_def(s)_for_reduction.
7276 (vectorizable_induction): Use gimple_build_vector rather than
7279 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7280 Alan Hayward <alan.hayward@arm.com>
7281 David Sherwood <david.sherwood@arm.com>
7283 * target.h (vec_perm_indices): New typedef.
7284 (auto_vec_perm_indices): Likewise.
7285 * optabs-query.h: Include target.h
7286 (can_vec_perm_p): Take a vec_perm_indices *.
7287 * optabs-query.c (can_vec_perm_p): Likewise.
7288 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
7289 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
7290 * tree-vect-generic.c (lower_vec_perm): Likewise.
7291 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
7292 (vect_grouped_load_supported): Likewise.
7293 (vect_shift_permute_load_chain): Likewise.
7294 (vect_permute_store_chain): Use auto_vec_perm_indices.
7295 (vect_permute_load_chain): Likewise.
7296 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
7297 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
7298 Update uses of can_vec_perm_p.
7299 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
7300 mode with a number of elements. Take a vec_perm_indices *.
7301 (vect_create_epilog_for_reduction): Update accordingly.
7302 Use auto_vec_perm_indices.
7303 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
7304 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
7305 (vect_transform_slp_perm_load): Likewise.
7306 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
7307 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
7308 (vect_gen_perm_mask_checked): Likewise.
7309 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
7310 (vect_gen_perm_mask_checked): Likewise.
7311 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
7312 (vectorizable_store): Likewise.
7313 (vectorizable_load): Likewise.
7314 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
7315 (vectorizable_bswap): Likewise.
7317 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7318 Alan Hayward <alan.hayward@arm.com>
7319 David Sherwood <david.sherwood@arm.com>
7321 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
7322 * tree.c (build_vector): Likewise.
7323 (build_vector_from_ctor): Update accordingly.
7324 (build_vector_from_val): Likewise.
7325 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
7326 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
7327 * tree-vect-generic.c (add_rshift): Likewise.
7328 (expand_vector_divmod): Likewise.
7329 (optimize_vector_constructor): Likewise.
7330 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
7331 (vect_transform_slp_perm_load): Likewise.
7332 (vect_schedule_slp_instance): Likewise.
7333 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
7334 (vectorizable_call): Likewise.
7335 (vect_gen_perm_mask_any): Likewise. Add elements in order.
7336 * expmed.c (make_tree): Likewise.
7337 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
7338 a vector passed to build_vector.
7339 (fold_convert_const): Likewise.
7340 (exact_inverse): Likewise.
7341 (fold_ternary_loc): Likewise.
7342 (fold_relational_const): Likewise.
7343 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
7344 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
7345 (const_unop): Likewise. Store the reduction accumulator in a
7346 variable rather than an array.
7347 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
7348 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
7349 the new vector, rather than constructing it after the input arrays.
7350 (native_interpret_vector): Use auto_vec<tree> when building
7351 a vector passed to build_vector. Add elements in order.
7352 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
7353 auto_vec<tree> when building a vector passed to build_vector.
7354 (vect_create_epilog_for_reduction): Likewise.
7355 (vectorizable_induction): Likewise.
7356 (get_initial_def_for_reduction): Likewise. Fix indentation of
7358 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
7360 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
7361 passed to build_vector.
7363 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7364 Alan Hayward <alan.hayward@arm.com>
7365 David Sherwood <david.sherwood@arm.com>
7367 * tree-core.h (tree_base::u): Add an "nelts" field.
7368 (tree_vector): Use VECTOR_CST_NELTS as the length.
7369 * tree.c (tree_size): Likewise.
7370 (make_vector): Initialize VECTOR_CST_NELTS.
7371 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
7372 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
7373 TYPE_VECTOR_SUBPARTS.
7374 * expr.c (const_vector_mask_from_tree): Consistently use "units"
7375 as the number of units, setting it from VECTOR_CST_NELTS.
7376 (const_vector_from_tree): Likewise.
7377 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
7378 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
7379 (fold_negate_expr_1): Likewise.
7380 (fold_convert_const): Likewise.
7381 (const_binop): Likewise. Differentiate the number of output and
7383 (const_unop): Likewise.
7384 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
7385 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
7386 in cases that did the opposite.
7388 2017-09-14 Richard Biener <rguenther@suse.de>
7390 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
7393 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
7395 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
7397 2017-09-14 Jakub Jelinek <jakub@redhat.com>
7400 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
7401 if and where to split a bb, except for splitting before debug insn
7402 sequences followed by non-label real insn. Delete debug insns
7403 in between basic blocks.
7405 * combine.c (make_compound_operation_int): Formatting fixes.
7407 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
7408 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7409 * config/netbsd.h (LINK_EH_SPEC): Likewise.
7410 * config/sol2.h (LINK_EH_SPEC): Likewise.
7411 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7412 * config/s390/linux.h (LINK_SPEC): Likewise.
7413 * config/freebsd.h (LINK_EH_SPEC): Likewise.
7414 * config/openbsd.h (LINK_EH_SPEC): Likewise.
7415 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7416 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
7417 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
7418 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7419 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
7420 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
7422 2017-09-13 Jakub Jelinek <jakub@redhat.com>
7424 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
7426 (ENDFILE_LINUX_SPEC): Likewise.
7427 (LINK_EH_SPEC): Likewise.
7428 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
7429 (LINK_OS_LINUX_SPEC32): Likewise.
7430 (LINK_OS_LINUX_SPEC64): Likewise.
7431 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
7432 (LINK_OS_LINUX_SPEC): Likewise.
7434 2017-09-13 Martin Liska <mliska@suse.cz>
7437 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
7438 CASE_HIGH is NULL_TREE.
7440 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
7441 Alan Hayward <alan.hayward@arm.com>
7442 David Sherwood <david.sherwood@arm.com>
7444 * target.def (secondary_memory_needed): New hook.
7445 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
7446 instead of SECONDARY_MEMORY_NEEDED.
7447 (secondary_memory_needed_mode): Likewise.
7448 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
7449 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
7450 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
7451 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
7452 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
7453 * doc/tm.texi: Regenerate.
7454 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
7455 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
7456 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7457 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
7458 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
7459 * config/i386/i386.c (inline_secondary_memory_needed): Put the
7460 mode argument first and change the reg_class arguments to reg_class_t.
7461 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
7462 Make static. Update the call to inline_secondary_memory_needed.
7463 (ix86_register_move_cost): Update the call to
7464 inline_secondary_memory_needed.
7465 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7466 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
7468 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
7469 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7471 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
7472 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
7473 * config/mips/mips.c (mips_secondary_memory_needed): Make static
7474 and match hook interface. Add comment from mips.h.
7475 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7476 * config/mmix/mmix.md (truncdfsf2): Refer to
7477 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7479 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
7480 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
7481 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7482 (pa_secondary_memory_needed): New function.
7483 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
7484 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
7485 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7486 (pdp11_secondary_memory_needed): Make static and match hook interface.
7487 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
7488 * config/powerpcspe/powerpcspe-protos.h
7489 (rs6000_secondary_memory_needed_ptr): Delete.
7490 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
7492 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7493 (rs6000_option_override_internal): Assign to
7494 targetm.secondary_memory_needed rather than
7495 rs6000_secondary_memory_needed_ptr.
7496 (rs6000_secondary_memory_needed): Match hook interface.
7497 (rs6000_debug_secondary_memory_needed): Likewise.
7498 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
7499 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
7500 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
7501 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7502 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
7503 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
7505 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
7506 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7507 (rs6000_option_override_internal): Assign to
7508 targetm.secondary_memory_needed rather than
7509 rs6000_secondary_memory_needed_ptr.
7510 (rs6000_secondary_memory_needed): Match hook interface.
7511 (rs6000_debug_secondary_memory_needed): Likewise.
7512 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
7513 * config/s390/s390.c (s390_secondary_memory_needed): New function.
7514 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7515 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
7516 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7517 (sparc_secondary_memory_needed): New function.
7518 * lra-constraints.c (check_and_process_move): Refer to
7519 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7521 (curr_insn_transform): Likewise.
7522 (process_alt_operands): Use targetm.secondary_memory_needed
7523 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7524 (check_secondary_memory_needed_p): Likewise.
7525 (choose_split_class): Likewise.
7526 * reload.c: Unconditionally include code that was previously
7527 conditional on SECONDARY_MEMORY_NEEDED.
7528 (push_secondary_reload): Use targetm.secondary_memory_needed
7529 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7530 (push_reload): Likewise.
7531 * reload1.c: Unconditionally include code that was previously
7532 conditional on SECONDARY_MEMORY_NEEDED.
7533 (choose_reload_regs): Use targetm.secondary_memory_needed
7534 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7535 (gen_reload): Likewise.
7536 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
7538 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
7539 Alan Hayward <alan.hayward@arm.com>
7540 David Sherwood <david.sherwood@arm.com>
7542 * target.def (secondary_memory_needed_mode): New hook:
7543 * targhooks.c (default_secondary_memory_needed_mode): Declare.
7544 * targhooks.h (default_secondary_memory_needed_mode): New function.
7545 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
7546 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
7547 * doc/tm.texi: Regenerate.
7548 * lra-constraints.c (check_and_process_move): Use
7549 targetm.secondary_memory_needed_mode instead of
7550 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
7551 (curr_insn_transform): Likewise.
7552 * reload.c (get_secondary_mem): Likewise.
7553 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7554 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
7556 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7557 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7558 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
7559 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7560 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
7562 * config/powerpcspe/powerpcspe-protos.h
7563 (rs6000_secondary_memory_needed_mode): Delete.
7564 * config/powerpcspe/powerpcspe.c
7565 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7566 (rs6000_secondary_memory_needed_mode): Make static.
7567 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7568 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
7570 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
7572 (rs6000_secondary_memory_needed_mode): Make static.
7573 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7574 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
7575 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7576 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7577 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
7579 (sparc_secondary_memory_needed_mode): New function.
7580 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
7582 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
7584 * config/aarch64/constraints.md (Umq): New constraint.
7585 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
7587 (mov<mode>): Update condition.
7589 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7591 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
7592 when bitposition is the same.
7594 2017-09-13 Richard Biener <rguenther@suse.de>
7596 * dwarf2out.c (output_die_symbol): Remove.
7597 (output_die): Do not output a DIEs symbol.
7599 2017-09-13 Richard Biener <rguenther@suse.de>
7602 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
7603 default-def to avoid breaking iterator update with the weird
7604 interaction with cgraph_update_edges_for_call_stmt_node.
7606 2017-09-13 Richard Biener <rguenther@suse.de>
7608 * tree-cfg.c (verify_gimple_assign_binary): Add verification
7609 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
7610 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
7611 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
7613 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
7615 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
7616 Disable pc relative literal load irrespective of
7617 TARGET_FIX_ERR_A53_84341 for default.
7619 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
7621 * config/sparc/sparc.c (output_return): Output the source location of
7622 the insn in the delay slot, if any.
7623 (output_sibcall): Likewise.
7625 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
7628 * common.opt (-static-pie): New alias.
7629 (shared): Negate static-pie.
7630 (-no-pie): Update help text.
7632 (static-pie): New option.
7633 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
7634 -static-pie support.
7635 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
7636 (LINK_EH_SPEC): Likewise.
7637 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7638 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
7639 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
7640 * gcc.c (LINK_COMMAND_SPEC): Likewise.
7641 (init_gcc_specs): Likewise.
7642 (init_spec): Likewise.
7643 (display_help): Update help message for -pie.
7644 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
7647 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
7649 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
7650 (movdi_aarch64): Likewise.
7651 (movti_aarch64): Likewise.
7653 2017-09-12 Simon Wright <simon@pushface.org>
7656 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
7657 calculation of the minor version, always output as 0.
7659 2017-09-12 Jakub Jelinek <jakub@redhat.com>
7662 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
7663 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
7664 on it early, rather than manual conversion late. For
7665 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
7666 instead of performing manual conversion.
7668 2017-09-12 Carl Love <cel@us.ibm.com>
7670 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
7671 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
7673 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
7674 VMULOSW): Add definitions.
7675 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7676 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
7677 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
7678 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
7679 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
7681 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
7683 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
7685 (movti_aarch64): Likewise.
7686 (movdf_aarch64): Likewise.
7687 (movtf_aarch64): Likewise.
7688 (load_pairdi): Likewise.
7689 (store_pairdi): Likewise.
7690 (load_pairdf): Likewise.
7691 (store_pairdf): Likewise.
7692 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
7693 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
7694 (ldr_got_small_<mode>): Likewise.
7695 (ldr_got_small_28k_<mode>): Likewise.
7696 (ldr_got_tiny): Likewise.
7697 * config/aarch64/iterators.md (ldst_sz): New.
7698 (ldpstp_sz): Likewise.
7699 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
7701 (thunderx_load): Split load_8 to load_16.
7702 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
7704 (thunderx2t99_storepair_basic): Split store_8 to store_16.
7705 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
7706 (xgene1_store_pair): Split store_8 to store_16.
7707 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
7708 (falkor_st_0_st_sd): Split store_8 to store_16.
7710 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
7712 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
7713 and store1/2/3/4 to store_4/8/12/16.
7714 * config/aarch64/aarch64.md: Update for rename.
7715 * config/arm/arm.md: Likewise.: Likewise.
7716 * config/arm/arm.c: Likewise.
7717 * config/arm/thumb1.md: Likewise.
7718 * config/arm/thumb2.md: Likewise.
7719 * config/arm/vfp.md: Likewise.
7720 * config/arm/arm-generic.md: Likewise.
7721 * config/arm/arm1020e.md: Likewise.
7722 * config/arm/arm1026ejs.md: Likewise.
7723 * config/arm/arm1136jfs.md: Likewise.
7724 * config/arm/arm926ejs.md: Likewise.
7725 * config/arm/cortex-a15.md: Likewise.
7726 * config/arm/cortex-a17.md: Likewise.
7727 * config/arm/cortex-a5.md: Likewise.
7728 * config/arm/cortex-a53.md: Likewise.
7729 * config/arm/cortex-a57.md: Likewise.
7730 * config/arm/cortex-a7.md: Likewise.
7731 * config/arm/cortex-a8.md: Likewise.
7732 * config/arm/cortex-a9.md: Likewise.
7733 * config/arm/cortex-m4.md: Likewise.
7734 * config/arm/cortex-m7.md: Likewise.
7735 * config/arm/cortex-r4.md: Likewise.
7736 * config/arm/exynos-m1.md: Likewise.
7737 * config/arm/fa526.md: Likewise.
7738 * config/arm/fa606te.md: Likewise.
7739 * config/arm/fa626te.md: Likewise.
7740 * config/arm/fa726te.md: Likewise.
7741 * config/arm/fmp626.md: Likewise.
7742 * config/arm/iwmmxt.md: Likewise.
7743 * config/arm/ldmstm.md: Likewise.
7744 * config/arm/marvell-pj4.md: Likewise.
7745 * config/arm/xgene1.md: Likewise.
7746 * config/aarch64/thunderx.md: Likewise.
7747 * config/aarch64/thunderx2t99.md: Likewise.
7748 * config/aarch64/falkor.md: Likewise.
7750 2017-09-12 Martin Liska <mliska@suse.cz>
7752 * attribs.c (private_lookup_attribute): New function.
7753 * attribs.h (private_lookup_attribute): Declared here.
7754 (lookup_attribute): Called from this place.
7756 2017-09-12 Richard Biener <rguenther@suse.de>
7758 PR tree-optimization/82157
7759 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
7760 stmts with side-effects.
7762 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7763 Alan Hayward <alan.hayward@arm.com>
7764 David Sherwood <david.sherwood@arm.com>
7766 * target.def (hard_regno_nregs): New hook.
7767 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
7768 * targhooks.h (default_hard_regno_nregs): Declare.
7769 * targhooks.c (default_hard_regno_nregs): New function.
7770 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
7771 (TARGET_HARD_REGNO_NREGS): ...this hook.
7772 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
7773 (CLASS_MAX_NREGS): Likewise.
7774 * doc/tm.texi: Regenerate.
7775 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
7776 instead of HARD_REGNO_NREGS.
7777 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
7778 HARD_REGNO_NREGS in the comment.
7779 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
7780 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
7781 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
7782 Return an unsigned int.
7783 (TARGET_HARD_REGNO_NREGS): Redefine.
7784 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
7785 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
7786 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
7787 (arc_hard_regno_nregs): New function.
7788 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
7789 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
7790 (arm_hard_regno_nregs): New function.
7791 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
7792 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
7793 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
7794 (TARGET_HARD_REGNO_NREGS): Redefine.
7795 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
7796 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
7797 (HARD_REGNO_NREGS): Delete.
7798 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
7799 (cr16_hard_regno_nregs): New function.
7800 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
7801 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
7802 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
7803 (cris_hard_regno_nregs): New function.
7804 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
7805 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
7806 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
7807 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
7808 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
7809 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
7810 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
7811 (frv_hard_regno_nregs): Make static. Take and return an
7813 (frv_class_max_nregs): Remove outdated copy of documentation.
7814 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
7815 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
7816 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
7817 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
7818 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
7819 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
7820 (TARGET_HARD_REGNO_NREGS): Redefine.
7821 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
7822 (CLASS_MAX_NREGS): Update comment.
7823 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
7824 (ia64_hard_regno_nregs): New function.
7825 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
7826 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
7827 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
7828 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
7829 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
7831 (m32c_hard_regno_nregs): Likewise. Make static.
7832 (TARGET_HARD_REGNO_NREGS): Redefine.
7833 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
7834 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
7835 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
7836 (m68k_hard_regno_nregs): New function.
7837 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
7838 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
7839 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
7840 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
7841 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
7842 Take and return an unsigned int.
7843 (TARGET_HARD_REGNO_NREGS): Redefine.
7844 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
7845 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
7846 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
7847 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
7848 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
7849 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
7850 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
7851 (msp430_hard_regno_nregs): Make static. Take and return an
7853 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
7854 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
7855 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
7856 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
7857 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
7858 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
7859 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
7860 (TARGET_HARD_REGNO_NREGS): Redefine.
7861 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
7862 (PA_HARD_REGNO_NREGS): ...this.
7863 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
7864 (PA_HARD_REGNO_NREGS): ...this.
7865 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
7866 (pa_hard_regno_nregs): New function.
7867 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
7868 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
7869 (pdp11_hard_regno_nregs): New function.
7870 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
7871 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
7872 (rs6000_hard_regno_nregs_hook): New function.
7873 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
7874 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
7875 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
7876 Take and return an unsigned int. Move earlier in file.
7877 (TARGET_HARD_REGNO_NREGS): Redefine.
7878 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
7879 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
7880 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
7881 (rl78_hard_regno_nregs): Make static. Take and return an
7883 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
7884 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
7885 (rs6000_hard_regno_nregs_hook): New function.
7886 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
7887 * config/rx/rx.c (rx_hard_regno_nregs): New function.
7888 (TARGET_HARD_REGNO_NREGS): Redefine.
7889 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
7890 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
7891 instead of HARD_REGNO_NREGS.
7892 (s390_hard_regno_nregs): New function.
7893 (s390_hard_regno_mode_ok): Add comment from s390.h.
7894 (TARGET_HARD_REGNO_NREGS): Redefine.
7895 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
7896 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
7897 (sh_hard_regno_nregs): New function.
7898 (sh_pass_in_reg_p): Use it.
7899 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
7900 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
7901 (sparc_hard_regno_nregs): New function.
7902 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
7903 * config/spu/spu.c (spu_hard_regno_nregs): New function.
7904 (spu_function_arg_advance): Use it, supplying a valid register number.
7905 (TARGET_HARD_REGNO_NREGS): Redefine.
7906 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
7907 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
7908 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
7909 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
7910 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
7911 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
7912 (CLASS_MAX_NREGS): Remove copy of old documentation.
7913 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
7914 (visium_hard_regno_nregs): New function.
7915 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
7916 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
7917 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
7918 xtensa_hard_regno_nregs): New function.
7919 * system.h (HARD_REGNO_NREGS): Poison.
7921 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7923 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
7924 hard_regno_nregs instead of HARD_REGNO_NREGS.
7925 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
7926 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
7927 (c6x_expand_epilogue): Likewise.
7928 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
7929 (frv_read_iacc_argument): Likewise.
7930 * config/sh/sh.c: Include regs.h.
7931 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
7932 (regs_used): Likewise.
7933 (output_stack_adjust): Likewise.
7934 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
7935 * expmed.c: Include regs.h.
7936 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
7937 * ree.c: Include regs.h.
7938 (combine_reaching_defs): Use hard_regno_nregs instead of
7940 (add_removable_extension): Likewise.
7942 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7944 * regs.h (hard_regno_nregs): Turn into a function.
7945 (end_hard_regno): Update accordingly.
7946 * caller-save.c (setup_save_areas): Likewise.
7947 (save_call_clobbered_regs): Likewise.
7948 (replace_reg_with_saved_mem): Likewise.
7949 (insert_restore): Likewise.
7950 (insert_save): Likewise.
7951 * combine.c (can_change_dest_mode): Likewise.
7952 (move_deaths): Likewise.
7953 (distribute_notes): Likewise.
7954 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
7955 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
7956 (rs6000_split_multireg_move): Likewise.
7957 (rs6000_register_move_cost): Likewise.
7958 (rs6000_memory_move_cost): Likewise.
7959 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
7960 (rs6000_split_multireg_move): Likewise.
7961 (rs6000_register_move_cost): Likewise.
7962 (rs6000_memory_move_cost): Likewise.
7963 * cselib.c (cselib_reset_table): Likewise.
7964 (cselib_lookup_1): Likewise.
7965 * emit-rtl.c (set_mode_and_regno): Likewise.
7966 * function.c (aggregate_value_p): Likewise.
7967 * ira-color.c (setup_profitable_hard_regs): Likewise.
7968 (check_hard_reg_p): Likewise.
7969 (calculate_saved_nregs): Likewise.
7970 (assign_hard_reg): Likewise.
7971 (improve_allocation): Likewise.
7972 (calculate_spill_cost): Likewise.
7973 * ira-emit.c (modify_move_list): Likewise.
7974 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
7975 (ira_hard_reg_in_set_p): Likewise.
7976 * ira.c (setup_reg_mode_hard_regset): Likewise.
7977 (clarify_prohibited_class_mode_regs): Likewise.
7978 (check_allocation): Likewise.
7979 * lra-assigns.c (find_hard_regno_for_1): Likewise.
7980 (lra_setup_reg_renumber): Likewise.
7981 (setup_try_hard_regno_pseudos): Likewise.
7982 (spill_for): Likewise.
7983 (assign_hard_regno): Likewise.
7984 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
7985 * lra-constraints.c (in_class_p): Likewise.
7986 (lra_constraint_offset): Likewise.
7987 (simplify_operand_subreg): Likewise.
7988 (lra_constraints): Likewise.
7989 (split_reg): Likewise.
7990 (split_if_necessary): Likewise.
7991 (invariant_p): Likewise.
7992 (inherit_in_ebb): Likewise.
7993 * lra-lives.c (process_bb_lives): Likewise.
7994 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
7995 (get_hard_regs): Likewise.
7996 (do_remat): Likewise.
7997 * lra-spills.c (assign_spill_hard_regs): Likewise.
7998 * mode-switching.c (create_pre_exit): Likewise.
7999 * postreload.c (reload_combine_recognize_pattern): Likewise.
8000 * recog.c (peep2_find_free_register): Likewise.
8001 * regcprop.c (kill_value_regno): Likewise.
8002 (set_value_regno): Likewise.
8003 (copy_value): Likewise.
8004 (maybe_mode_change): Likewise.
8005 (find_oldest_value_reg): Likewise.
8006 (copyprop_hardreg_forward_1): Likewise.
8007 * regrename.c (check_new_reg_p): Likewise.
8008 (regrename_do_replace): Likewise.
8009 * reload.c (push_reload): Likewise.
8010 (combine_reloads): Likewise.
8011 (find_dummy_reload): Likewise.
8012 (operands_match_p): Likewise.
8013 (find_reloads): Likewise.
8014 (find_equiv_reg): Likewise.
8015 (reload_adjust_reg_for_mode): Likewise.
8016 * reload1.c (count_pseudo): Likewise.
8017 (count_spilled_pseudo): Likewise.
8018 (find_reg): Likewise.
8019 (clear_reload_reg_in_use): Likewise.
8020 (free_for_value_p): Likewise.
8021 (allocate_reload_reg): Likewise.
8022 (choose_reload_regs): Likewise.
8023 (reload_adjust_reg_for_temp): Likewise.
8024 (emit_reload_insns): Likewise.
8025 (delete_output_reload): Likewise.
8026 * rtlanal.c (subreg_get_info): Likewise.
8027 * sched-deps.c (sched_analyze_reg): Likewise.
8028 * sel-sched.c (init_regs_for_mode): Likewise.
8029 (mark_unavailable_hard_regs): Likewise.
8030 (choose_best_reg_1): Likewise.
8031 (verify_target_availability): Likewise.
8032 * valtrack.c (dead_debug_insert_temp): Likewise.
8033 * var-tracking.c (track_loc_p): Likewise.
8034 (emit_note_insn_var_location): Likewise.
8035 * varasm.c (make_decl_rtl): Likewise.
8036 * reginfo.c (choose_hard_reg_mode): Likewise.
8037 (init_reg_modes_target): Refer directly to
8038 this_target_regs->x_hard_regno_nregs.
8040 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8042 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
8043 instead of hard_regno_nregs.
8045 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8047 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
8048 end_hard_regno instead of hard_regno_nregs.
8049 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
8050 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
8051 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
8052 * ira-color.c (improve_allocation): Likewise.
8053 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8054 * lra-lives.c (mark_regno_live): Likewise.
8055 (mark_regno_dead): Likewise.
8056 * lra-remat.c (operand_to_remat): Likewise.
8057 * lra.c (collect_non_operand_hard_regs): Likewise.
8058 * postreload.c (reload_combine_note_store): Likewise.
8059 (move2add_valid_value_p): Likewise.
8060 * reload.c (regno_clobbered_p): Likewise.
8062 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8064 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
8066 * config/v850/v850.c (v850_reorg): Likewise.
8067 * reload.c (refers_to_regno_for_reload_p): Likewise.
8068 (find_equiv_reg): Likewise.
8069 * reload1.c (reload_reg_reaches_end_p): Likewise.
8071 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8073 * caller-save.c (add_used_regs): Use REG_NREGS instead of
8075 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
8076 * config/arm/arm.c (output_move_neon): Likewise.
8077 (arm_attr_length_move_neon): Likewise.
8078 (neon_split_vcombine): Likewise.
8079 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
8080 (c6x_mark_reg_written): Likewise.
8081 (c6x_dwarf_register_span): Likewise.
8082 * config/i386/i386.c (ix86_save_reg): Likewise.
8083 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
8084 (rws_access_reg): Likewise.
8085 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
8086 * mode-switching.c (create_pre_exit): Likewise.
8087 * ree.c (combine_reaching_defs): Likewise.
8088 (add_removable_extension): Likewise.
8089 * regcprop.c (find_oldest_value_reg): Likewise.
8090 (copyprop_hardreg_forward_1): Likewise.
8091 * reload.c (reload_inner_reg_of_subreg): Likewise.
8092 (push_reload): Likewise.
8093 (combine_reloads): Likewise.
8094 (find_dummy_reload): Likewise.
8095 (reload_adjust_reg_for_mode): Likewise.
8096 * reload1.c (find_reload_regs): Likewise.
8097 (forget_old_reloads_1): Likewise.
8098 (reload_reg_free_for_value_p): Likewise.
8099 (reload_adjust_reg_for_temp): Likewise.
8100 (emit_reload_insns): Likewise.
8101 (delete_output_reload): Likewise.
8102 * sel-sched.c (choose_best_reg_1): Likewise.
8103 (choose_best_pseudo_reg): Likewise.
8105 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8106 Alan Hayward <alan.hayward@arm.com>
8107 David Sherwood <david.sherwood@arm.com>
8109 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
8110 * target.def (slow_unaligned_access): New hook.
8111 * targhooks.h (default_slow_unaligned_access): Declare.
8112 * targhooks.c (default_slow_unaligned_access): New function.
8113 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
8114 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
8115 * doc/tm.texi: Regenerate.
8116 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
8117 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
8118 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
8120 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
8121 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
8123 (rs6000_slow_unaligned_access): New function.
8124 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
8125 (expand_block_compare): Likewise.
8126 (expand_strn_compare): Likewise.
8127 (rs6000_rtx_costs): Likewise.
8128 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
8129 (riscv_slow_unaligned_access): Likewise.
8130 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
8131 (riscv_slow_unaligned_access_p): ...this and make static.
8132 (riscv_option_override): Update accordingly.
8133 (riscv_slow_unaligned_access): New function.
8134 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
8135 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
8136 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
8137 (rs6000_slow_unaligned_access): New function.
8138 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
8139 (rs6000_rtx_costs): Likewise.
8140 * config/rs6000/rs6000-string.c (expand_block_compare)
8141 (expand_strn_compare): Use targetm.slow_unaligned_access instead
8142 of SLOW_UNALIGNED_ACCESS.
8143 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
8144 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
8145 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
8146 of SLOW_UNALIGNED_ACCESS.
8147 * expmed.c (simple_mem_bitfield_p): Likewise.
8148 * expr.c (alignment_for_piecewise_move): Likewise.
8149 (emit_group_load_1): Likewise.
8150 (emit_group_store): Likewise.
8151 (copy_blkmode_from_reg): Likewise.
8152 (emit_push_insn): Likewise.
8153 (expand_assignment): Likewise.
8154 (store_field): Likewise.
8155 (expand_expr_real_1): Likewise.
8156 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
8157 * lra-constraints.c (simplify_operand_subreg): Likewise.
8158 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
8159 * gimple-ssa-store-merging.c: Likewise in block comment at start
8161 * tree-ssa-strlen.c: Include target.h.
8162 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
8163 of SLOW_UNALIGNED_ACCESS.
8164 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
8166 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8168 PR rtl-optimization/82185
8169 * expmed.c (emit_store_flag_int): Only test tem if it has been
8172 2017-09-12 Richard Biener <rguenther@suse.de>
8175 * match.pd ((FTYPE) N CMP CST): Fix typo.
8177 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
8179 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
8181 (mips_near_type_p): Add 'short_call' attribute as a synonym
8183 * doc/extend.texi (short_call): Document new function attribute.
8185 2017-09-12 Jakub Jelinek <jakub@redhat.com>
8188 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
8189 assertion check that in the condition.
8190 (get_atomic_generic_size): Likewise. Before testing if parameter
8191 has pointer type, if it has array type, call for C++
8192 default_conversion to perform array-to-pointer conversion.
8194 2017-09-12 Richard Biener <rguenther@suse.de>
8196 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
8197 for operations we cannot scalarize.
8199 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
8201 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
8202 vectors heap vectors. Clean up comments.
8203 Make visited_bbs a reference.
8204 (profitable_jump_thread_path): Make GC
8205 vectors heap vectors. Clean up comments.
8207 (convert_and_register_jump_thread_path): Make GC vectors heap
8209 (check_subpath_and_update_thread_path): Same. Clean up comments.
8210 Make visited_bbs a reference.
8211 (handle_phi): Abstract common code to to
8212 register_jump_thread_path_if_profitable.
8213 Rename VAR_BB to DEF_BB.
8215 Make GC vectors heap vectors.
8216 Make visited_bbs a reference.
8217 (handle_assignment): Same.
8218 (register_jump_thread_path_if_profitable): New.
8219 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
8221 Make GC vectors heap vectors. Clean up comments.
8222 Make visited_bbs a reference.
8223 (find_jump_threads_backwards): Make visited_bbs live in the stack.
8224 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
8227 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
8230 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
8231 words of E_DImode object are reachable by xtensa_uimm8x4 access.
8233 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
8235 Revert r251800 and r251799.
8237 2017-09-11 Martin Jambor <mjambor@suse.cz>
8240 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
8241 arguments in advance.
8242 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
8243 use it to find predecessor edges.
8244 (naive_outof_ssa): Collect vector of predecessors.
8246 2017-09-08 Jason Merrill <jason@redhat.com>
8248 PR c++/70029 - ICE with ref-qualifier and -flto
8249 * langhooks.h (struct lang_hooks_for_types): Add
8250 copy_lang_qualifiers.
8251 * attribs.c (build_type_attribute_qual_variant): Use it.
8252 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
8254 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
8255 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
8257 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
8260 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
8261 (*mulsi3_sp64): New instruction.
8262 (mulsi3): New expander.
8264 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
8266 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
8268 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8270 * sancov.c: Include memmodel.h.
8272 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
8275 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
8278 2017-09-07 Carl Love <cel@us.ibm.com>
8280 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
8281 the sldi instruction.
8283 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
8285 * sancov.c: Include tm_p.h.
8287 2017-09-07 Jakub Jelinek <jakub@redhat.com>
8290 * output.h (switch_to_other_text_partition): New declaration.
8291 * varasm.c (switch_to_other_text_partition): New function.
8292 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
8293 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
8294 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
8295 to the other text partition before emitting LCL label and switch back
8296 after emitting the word after it.
8298 2017-09-07 Richard Biener <rguenther@suse.de>
8300 * passes.def (pass_split_crit_edges): Remove instance before PRE.
8301 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
8302 critical edges here, after loop init.
8303 (pass_data_pre): Remove PROP_no_crit_edges flags.
8304 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
8305 for valueization of call args to avoid leaking VN_TOP.
8306 (visit_use): Assert we do not visit default defs.
8307 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
8308 Use error_mark_node to more easily detect leaking VN_TOP.
8309 All default-defs are varying, not VN_TOP. Mark them visited.
8310 (run_scc_vn): Make code match comment.
8312 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
8314 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
8315 OPTION_MASK_FLOAT128_KEYWORD.
8316 (POWERPC_MASKS): Likewise.
8317 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
8318 support for the -mfloat128-type option, and make -mfloat128
8319 default on PowerPC Linux systems. Define or undefine
8320 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
8321 Define __float128 to be __ieee128 if IEEE 128-bit support is
8322 enabled, or undefine it.
8323 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
8324 Delete defining __FLOAT128_TYPE__.
8325 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
8326 -mfloat128-type option and make -mfloat128 default on PowerPC
8328 (TARGET_FLOAT128_TYPE): Likewise.
8329 (-mfloat128-type): Likewise.
8330 * config/rs6000/rs6000.c (rs6000_option_override_internal):
8331 Delete the -mfloat128-type option and make -mfloat128 default on
8332 PowerPC Linux systems. Always use __ieee128 to be the keyword for
8333 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
8334 128-bit floating point is enabled. Change tests from using
8335 -mfloat128-type to -mfloat128.
8336 (rs6000_mangle_type): Use the correct mangling for the __float128
8337 type even if normal long double is restricted to 64-bits.
8338 (floatn_mode): Enable the _Float128 type by default on VSX Linux
8340 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
8341 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
8342 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
8344 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
8345 documentation for -mfloat128.
8347 2017-09-06 Olivier Hainque <hainque@adacore.com>
8349 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
8351 2017-09-06 Wish Wu <wishwu007@gmail.com>
8352 Jakub Jelinek <jakub@redhat.com>
8354 * asan.c (initialize_sanitizer_builtins): Add
8355 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
8356 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
8357 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
8358 BT_FN_VOID_UINT64_PTR variables.
8359 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
8360 (BT_FN_VOID_UINT16_UINT16): Likewise.
8361 (BT_FN_VOID_UINT32_UINT32): Likewise.
8362 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
8363 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
8364 (BT_FN_VOID_UINT64_PTR): Likewise.
8365 * common.opt (flag_sanitize_coverage): New variable.
8366 (fsanitize-coverage=trace-pc): Remove.
8367 (fsanitize-coverage=): Add.
8368 * flag-types.h (enum sanitize_coverage_code): New enum.
8369 * fold-const.c (fold_range_test): Disable non-short-circuit
8370 optimization if flag_sanitize_coverage.
8371 (fold_truth_andor): Likewise.
8372 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
8373 * opts.c (COVERAGE_SANITIZER_OPT): Define.
8374 (coverage_sanitizer_opts): New array.
8375 (get_closest_sanitizer_option): Add OPTS argument, handle also
8376 OPT_fsanitize_coverage_.
8377 (parse_sanitizer_options): Adjusted to also handle
8378 OPT_fsanitize_coverage_.
8379 (common_handle_option): Add OPT_fsanitize_coverage_.
8380 * sancov.c (instrument_comparison, instrument_switch): New function.
8381 (sancov_pass): Add trace-cmp support.
8382 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
8383 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
8384 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
8385 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
8386 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
8387 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
8388 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
8389 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
8390 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
8391 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
8393 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
8395 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
8396 error. Only quit immediately if parsing is complete.
8397 (BEGIN): Initialize fatal_err and parse_done.
8398 (begin fpu, end fpu): Check number of arguments.
8399 (begin arch, end arch): Likewise.
8400 (begin cpu, end cpu): Likewise.
8401 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
8402 (optalias): Likewise.
8404 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
8406 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
8407 * config/arm/arm-isa.h: Delete. Move definitions to ...
8408 * arm-cpus.in: ... here. Use new feature and fgroup values.
8409 * config/arm/arm.c (arm_option_override): Use lower case for feature
8411 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
8412 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
8413 * config/arm/parsecpu.awk (END): Add new command 'isa'.
8415 (print_isa_bits_for): New function.
8416 (gen_isa): New function.
8417 (gen_comm_data): Use print_isa_bits_for.
8418 (define feature): New keyword.
8419 (define fgroup): New keyword.
8420 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
8421 (arm-isa.h): Add rule to generate file.
8422 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
8423 case for feature bit names.
8425 2017-09-06 Richard Biener <rguenther@suse.de>
8427 * tree-ssa-pre.c (NECESSARY): Remove.
8428 (create_expression_by_pieces): Do not touch pass-local flags.
8429 (insert_into_preds_of_block): Likewise.
8430 (do_pre_regular_insertion): Likewise.
8431 (eliminate_insert): Likewise.
8432 (eliminate_dom_walker::before_dom_children): Likewise.
8433 (fini_eliminate): Do not look at inserted_exprs.
8434 (mark_operand_necessary): Remove.
8435 (remove_dead_inserted_code): Replace with simple work-list
8436 algorithm based on inserted_exprs and SSA uses.
8437 (pass_pre::execute): Re-order fini_eliminate and
8438 remove_dead_inserted_code.
8440 2017-09-06 Olivier Hainque <hainque@adacore.com>
8442 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
8443 for VxWorks 7. Adjust surrounding comments.
8445 2017-09-06 Richard Biener <rguenther@suse.de>
8447 * gimple-ssa-strength-reduction.c
8448 (find_candidates_dom_walker::before_dom_children): Also allow
8451 2017-09-06 Richard Biener <rguenther@suse.de>
8453 PR tree-optimization/82108
8454 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
8455 for gap in the non-permutation SLP case.
8457 2017-09-06 Martin Jambor <mjambor@suse.cz>
8459 PR tree-optimization/82078
8460 * tree-sra.c (sort_and_splice_var_accesses): Move call to
8461 add_access_to_work_queue...
8462 (build_accesses_from_assign): ...here.
8463 (propagate_all_subaccesses): Make sure racc is the group
8464 representative, if there is one.
8466 2017-09-06 Jakub Jelinek <jakub@redhat.com>
8469 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
8472 2017-09-06 Richard Biener <rguenther@suse.de>
8474 * gimple-ssa-strength-reduction.c
8475 (find_candidates_dom_walker::before_doom_children): Use a
8476 type and not a mode check.
8478 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
8481 * config/arm/predicates.md (arm_general_adddi_operand): Create new
8483 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
8485 2017-09-05 Jeff Law <law@redhat.com>
8487 PR tree-optimization/64910
8488 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
8489 cases where we have 3 or more operands.
8491 2017-09-05 Jakub Jelinek <jakub@redhat.com>
8494 * omp-low.c (lower_omp_for): Recompute tree invariant if
8495 gimple_omp_for_initial/final is ADDR_EXPR.
8498 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
8499 into gimple val before gimplification fo the COND_EXPR.
8501 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
8503 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
8504 REGION_COPY argument.
8505 (thread_through_all_blocks): Remove unused argument to
8506 duplicate_thread_path.
8508 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8509 Alan Hayward <alan.hayward@arm.com>
8510 David Sherwood <david.sherwood@arm.com>
8512 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
8513 Take a scalar_mode rather than a machine_mode.
8514 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8515 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
8516 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8517 (aarch64_gen_adjusted_ldpstp): Likewise.
8518 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
8520 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8521 Alan Hayward <alan.hayward@arm.com>
8522 David Sherwood <david.sherwood@arm.com>
8524 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
8525 Take a scalar_int_mode instead of a machine_mode.
8526 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
8527 (aarch64_output_scalar_simd_mov_immediate): Likewise.
8528 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
8529 (aarch64_simd_attr_length_rglist): Delete.
8530 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
8531 a scalar_int_mode instead of a machine_mode.
8532 (aarch64_add_offset): Likewise.
8533 (aarch64_internal_mov_immediate): Likewise
8534 (aarch64_add_constant_internal): Likewise.
8535 (aarch64_add_constant): Likewise.
8536 (aarch64_movw_imm): Likewise.
8537 (aarch64_rtx_arith_op_extract_p): Likewise.
8538 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
8539 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
8540 Remove assert that the mode isn't a vector.
8541 (aarch64_output_scalar_simd_mov_immediate): Likewise.
8542 (aarch64_expand_mov_immediate): Update calls after above changes.
8543 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
8544 (aarch64_and_bitmask_imm): Check for scalar integer modes.
8545 (aarch64_move_imm): Likewise.
8546 (aarch64_can_const_movi_rtx_p): Likewise.
8547 (aarch64_strip_extend): Likewise.
8548 (aarch64_extr_rtx_p): Likewise.
8549 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
8550 a CONST_INT when the mode parameter is VOIDmode.
8551 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
8553 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8555 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
8556 * stor-layout.c (bitwise_mode_for_mode): Likewise.
8557 (bitwise_type_for_mode): Update accordingly.
8559 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8561 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
8562 * stor-layout.c (mode_for_size_tree): Likewise.
8563 (mode_for_array): Update accordingly.
8564 (layout_decl): Likewise.
8565 (compute_record_mode): Likewise. Only set the mode once.
8567 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8569 * target.def (get_mask_mode): Change return type to opt_mode.
8571 * doc/tm.texi: Regenerate.
8572 * targhooks.h (default_get_mask_mode): Return an opt_mode.
8573 * targhooks.c (default_get_mask_mode): Likewise.
8574 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
8575 * optabs-query.c (can_vec_mask_load_store_p): Update use of
8576 targetm.get_mask_mode.
8577 * tree.c (build_truth_vector_type): Likewise.
8579 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8581 * machmode.h (mode_for_vector): Return an opt_mode.
8582 * stor-layout.c (mode_for_vector): Likewise.
8583 (mode_for_int_vector): Update accordingly.
8584 (layout_type): Likewise.
8585 * config/i386/i386.c (emit_memmov): Likewise.
8586 (ix86_expand_set_or_movmem): Likewise.
8587 (ix86_expand_vector_init): Likewise.
8588 (ix86_get_mask_mode): Likewise.
8589 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
8591 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
8592 * expmed.c (extract_bit_field_1): Likewise.
8593 * expr.c (expand_expr_real_2): Likewise.
8594 * optabs-query.c (can_vec_perm_p): Likewise.
8595 (can_vec_mask_load_store_p): Likewise.
8596 * optabs.c (expand_vec_perm): Likewise.
8597 * targhooks.c (default_get_mask_mode): Likewise.
8598 * tree-vect-stmts.c (vectorizable_store): Likewise.
8599 (vectorizable_load): Likewise.
8600 (get_vectype_for_scalar_type_and_size): Likewise.
8602 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8604 * machmode.h (mode_for_int_vector): New function.
8605 * stor-layout.c (mode_for_int_vector): Likewise.
8606 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
8607 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
8608 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
8609 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
8610 (s390_expand_vcond): Likewise.
8612 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8614 * machmode.h (opt_machine_mode): New type.
8615 (opt_mode<T>): Allow construction from anything that can be
8617 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
8618 (mode_for_size): Return an opt_machine_mode.
8619 * stor-layout.c (mode_for_size): Likewise.
8620 (mode_for_size_tree): Update call accordingly.
8621 (bitwise_mode_for_mode): Likewise.
8622 (make_fract_type): Likewise.
8623 (make_accum_type): Likewise.
8624 * caller-save.c (replace_reg_with_saved_mem): Update call
8626 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8627 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8628 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8629 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8630 * expmed.c (extract_bit_field_1): Likewise.
8631 * reload.c (get_secondary_mem): Likewise.
8632 * varasm.c (assemble_integer): Likewise.
8633 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
8636 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8638 * machmode.h (decimal_float_mode_for_size): New function.
8639 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
8640 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
8641 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
8642 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
8643 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
8644 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
8646 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8648 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
8649 (get_builtin_sync_mode): Likewise.
8650 (expand_ifn_atomic_compare_exchange): Likewise.
8651 (expand_builtin_atomic_clear): Likewise.
8652 (expand_builtin_atomic_test_and_set): Likewise.
8653 (fold_builtin_atomic_always_lock_free): Likewise.
8654 * calls.c (compute_argument_addresses): Likewise.
8655 (emit_library_call_value_1): Likewise.
8656 (store_one_arg): Likewise.
8657 * combine.c (combine_instructions): Likewise.
8658 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
8659 * config/arm/arm.c (arm_function_value): Likewise.
8660 (aapcs_allocate_return_reg): Likewise.
8661 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
8662 * config/i386/i386.c (construct_container): Likewise.
8663 (ix86_gimplify_va_arg): Likewise.
8664 (ix86_expand_sse_cmp): Likewise.
8665 (emit_memmov): Likewise.
8666 (emit_memset): Likewise.
8667 (expand_small_movmem_or_setmem): Likewise.
8668 (ix86_expand_pextr): Likewise.
8669 (ix86_expand_pinsr): Likewise.
8670 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
8671 * config/microblaze/microblaze.c (microblaze_block_move_straight):
8673 * config/mips/mips.c (mips_function_value_1) Likewise.
8674 (mips_block_move_straight): Likewise.
8675 (mips_expand_ins_as_unaligned_store): Likewise.
8676 * config/powerpcspe/powerpcspe.c
8677 (rs6000_darwin64_record_arg_advance_flush): Likewise.
8678 (rs6000_darwin64_record_arg_flush): Likewise.
8679 * config/rs6000/rs6000.c
8680 (rs6000_darwin64_record_arg_advance_flush): Likewise.
8681 (rs6000_darwin64_record_arg_flush): Likewise.
8682 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
8683 (sparc_function_value_1): Likewise.
8684 * config/spu/spu.c (adjust_operand): Likewise.
8685 (spu_emit_branch_or_set): Likewise.
8686 (arith_immediate_p): Likewise.
8687 * emit-rtl.c (gen_lowpart_common): Likewise.
8688 * expr.c (expand_expr_real_1): Likewise.
8689 * function.c (assign_parm_setup_block): Likewise.
8690 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
8691 * reload1.c (alter_reg): Likewise.
8692 * stor-layout.c (mode_for_vector): Likewise.
8693 (layout_type): Likewise.
8695 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8697 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
8698 (spu_convert_move): Likewise.
8699 * lower-subreg.c (resolve_simple_move): Likewise.
8701 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8704 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
8705 define_insn to a define_expand.
8706 (altivec_vsum2sws_direct): New define_insn.
8707 (altivec_vsumsws): Convert from a define_insn to a define_expand.
8709 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
8711 * config/arm/arm.c (arm_option_params_internal): Improve setting of
8714 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
8719 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
8720 realign stack if stack alignment needed is less than incoming
8723 2017-09-05 Marek Polacek <polacek@redhat.com>
8726 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
8729 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
8731 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
8733 2017-09-05 Richard Biener <rguenther@suse.de>
8735 PR tree-optimization/82084
8736 * fold-const.c (can_native_encode_string_p): Handle wide characters.
8738 2017-09-05 Richard Biener <rguenther@suse.de>
8740 PR tree-optimization/82102
8741 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
8743 2017-09-05 Martin Liska <mliska@suse.cz>
8745 PR tree-optimization/82032
8746 * tree-cfg.c (generate_range_test): New function.
8747 * tree-cfg.h (generate_range_test): Declared here.
8748 * tree-cfgcleanup.c (convert_single_case_switch): New function.
8749 (cleanup_control_expr_graph): Use it.
8750 * tree-switch-conversion.c (try_switch_expansion): Remove
8752 (emit_case_nodes): Use generate_range_test.
8754 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
8757 * config/i386/i386.md (*<btsc><mode>_mask): Add
8758 TARGET_USE_BT to insn constraint.
8759 (*btr<mode>_mask): Ditto.
8761 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
8763 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
8764 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
8766 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
8769 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
8770 TARGET_NEON and TARGET_IWMMXT.
8771 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
8772 TARGET_NEON and TARGET_IWMMXT.
8773 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
8775 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
8777 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
8778 (ix86_rewrite_tls_address): Ditto.
8779 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
8780 (ix86_rewrite_tls_address_1): Ditto.
8781 (ix86_rewrite_tls_address): Ditto.
8782 * config/i386/predicates.md (tls_address_pattern): New predicate.
8783 * config/i386/i386.md (TLS address splitter): New splitter.
8785 2017-09-04 Richard Biener <rguenther@suse.de>
8787 PR tree-optimization/82084
8788 * fold-const.h (can_native_encode_string_p): Declare.
8789 * fold-const.c (can_native_encode_string_p): Factor out from ...
8790 (native_encode_string): ... here.
8791 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
8792 vectorizing stores from constants we later cannot handle.
8794 2017-09-04 Marek Polacek <polacek@redhat.com>
8797 * doc/invoke.texi: Update -Wtautological-compare documentation.
8799 2017-09-04 Jeff Law <law@redhat.com>
8801 PR tree-optimization/64910
8802 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
8803 swap the first and last operand if the last is a constant.
8805 2017-09-04 Marek Polacek <polacek@redhat.com>
8808 * convert.c (do_narrow): When sanitizing signed integer overflows,
8809 bail out for signed types.
8810 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
8812 2017-09-04 Richard Biener <rguenther@suse.de>
8814 PR tree-optimization/82060
8815 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
8816 Move devirtualization after stmt folding and before EH/AB/noreturn
8817 cleanup to get the stmt refs canonicalized. Use a bool instead
8818 of gimple_modified_p since that doesn't work for NOPs. Schedule
8819 NOPs generated by folding for removal.
8821 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8822 Alan Hayward <alan.hayward@arm.com>
8823 David Sherwood <david.sherwood@arm.com>
8825 * coretypes.h (pad_direction): New enum.
8826 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
8827 (FUNCTION_ARG_PADDING): Likewise.
8828 * target.def (function_arg_padding): New hook.
8829 * targhooks.h (default_function_arg_padding): Declare.
8830 * targhooks.c (default_function_arg_padding): New function.
8831 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
8832 (TARGET_FUNCTION_ARG_PADDING): ...this.
8833 * doc/tm.texi: Regenerate.
8834 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
8835 instead of direction.
8836 (compute_argument_addresses): Likewise.
8837 (load_register_parameters): Likewise.
8838 (emit_library_call_value_1): Likewise.
8839 (store_one_arg): Use targetm.calls.function_arg_padding instead
8840 of FUNCTION_ARG_PADDING.
8841 (must_pass_in_stack_var_size_or_pad): Likewise.
8842 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
8843 (emit_group_store): Likewise.
8844 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
8845 instead of FUNCTION_ARG_PADDING.
8846 (emit_push_insn): Likewise, and propagate enum change throughout
8848 * function.h (direction): Delete.
8849 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
8851 * function.c (assign_parm_find_stack_rtl): Likewise.
8852 (assign_parm_setup_block_p): Likewise.
8853 (assign_parm_setup_block): Likewise.
8854 (gimplify_parameters): Likewise.
8855 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
8856 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
8858 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
8859 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
8860 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
8861 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
8862 (aarch64_function_arg_padding): ...this new function.
8863 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
8864 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8865 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
8866 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
8867 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
8868 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8869 (arm_pad_arg_upward): Replace with...
8870 (arm_function_arg_padding): ...this new function.
8871 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
8873 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
8874 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
8875 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8876 (ia64_hpux_function_arg_padding): Replace with...
8877 (ia64_function_arg_padding): ...this new function. Use pad_direction
8878 instead of direction. Check for TARGET_HPUX.
8879 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
8880 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8881 (iq2000_function_arg_padding): New function.
8882 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
8883 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
8884 (mips_function_arg_padding): ...this new function.
8885 (mips_pad_reg_upward): Update accordingly.
8886 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8887 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
8888 targetm.calls.function_arg_padding.
8889 (FUNCTION_ARG_PADDING): Delete.
8890 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
8891 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
8892 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
8893 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
8894 (nios2_block_reg_padding): Return pad_direction instead of direction.
8895 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
8896 instead of direction.
8897 (nios2_function_arg_padding): Likewise. Make static.
8898 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8899 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
8900 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
8901 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
8902 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8903 (pa_function_arg_padding): Make static. Return pad_direction instead
8905 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
8906 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
8907 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
8908 instead of direction. Use targetm.calls.function_arg_padding.
8909 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
8910 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
8911 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
8912 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
8913 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
8915 (function_arg_padding): Rename to...
8916 (rs6000_function_arg_padding): ...this. Make static. Return
8917 pad_direction instead of direction.
8918 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
8919 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
8920 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
8921 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
8922 instead of direction. Use targetm.calls.function_arg_padding.
8923 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
8924 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
8925 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
8926 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
8927 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8928 (function_arg_padding): Rename to...
8929 (rs6000_function_arg_padding): ...this. Make static. Return
8930 pad_direction instead of direction.
8931 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
8932 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
8933 * config/s390/s390.c (s390_function_arg_padding): New function.
8934 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8935 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
8936 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
8937 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8938 (function_arg_padding): Rename to...
8939 (sparc_function_arg_padding): ...this. Make static. Return
8940 pad_direction instead of direction.
8941 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
8942 * config/spu/spu.c (spu_function_arg_padding): New function.
8943 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8944 * system.h (FUNCTION_ARG_PADDING): Poison.
8946 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8947 Alan Hayward <alan.hayward@arm.com>
8948 David Sherwood <david.sherwood@arm.com>
8950 * target.def (modes_tieable_p): New hook.
8951 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
8952 (TARGET_MODES_TIEABLE_P): ...this.
8953 * doc/tm.texi.in: Regenerate.
8954 * hooks.h (hook_bool_mode_mode_true): Declare.
8955 * hooks.c (hook_bool_mode_mode_true): New function.
8956 * combine.c (subst): Use targetm.modes_tieable_p instead of
8958 * dse.c (find_shift_sequence): Likewise.
8959 * expmed.c (extract_low_bits): Likewise.
8960 * lower-subreg.c: Include target.h.
8961 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
8963 * rtlanal.c (rtx_cost): Likewise.
8964 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
8965 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
8966 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
8967 (TARGET_MODES_TIEABLE_P): Redefine.
8968 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
8969 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
8970 (TARGET_MODES_TIEABLE_P): Redefine.
8971 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
8972 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
8973 (arc_modes_tieable_p): New function.
8974 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
8975 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
8976 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
8977 (arm_modes_tieable_p): Make static.
8978 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
8979 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
8980 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
8981 (TARGET_MODES_TIEABLE_P): Redefine.
8982 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
8983 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
8984 (TARGET_MODES_TIEABLE_P): Redefine.
8985 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
8986 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
8987 (cr16_modes_tieable_p): New function.
8988 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
8989 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
8990 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
8991 (TRULY_NOOP_TRUNCATION): Update comment.
8992 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
8993 (TRULY_NOOP_TRUNCATION): Update comment.
8994 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
8995 (frv_modes_tieable_p): New function.
8996 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
8997 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
8998 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
8999 (TARGET_MODES_TIEABLE_P): Redefine.
9000 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
9001 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
9002 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
9003 (TARGET_MODES_TIEABLE_P): Redefine.
9004 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
9005 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
9006 (ia64_modes_tieable_p): New function.
9007 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
9008 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
9009 (iq2000_modes_tieable_p): New function.
9010 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
9011 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
9012 (lm32_modes_tieable_p): New function.
9013 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
9014 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
9015 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
9016 (TARGET_MODES_TIEABLE_P): Redefine.
9017 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
9018 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
9019 (m32r_modes_tieable_p): New function.
9020 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
9021 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
9022 (m68k_modes_tieable_p): New function.
9023 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
9024 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
9025 (mcore_modes_tieable_p): New function.
9026 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
9027 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
9029 (TARGET_MODES_TIEABLE_P): Redefine.
9030 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
9031 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
9032 * config/mips/mips.c (mips_modes_tieable_p): Make static.
9033 (TARGET_MODES_TIEABLE_P): Redefine.
9034 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
9035 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
9036 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
9037 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
9038 (mn10300_modes_tieable_p): ...this and make static.
9039 (TARGET_MODES_TIEABLE_P): Redefine.
9040 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
9041 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
9042 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
9043 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
9044 (msp430_modes_tieable_p): Make static.
9045 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
9046 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
9047 (TARGET_MODES_TIEABLE_P): Redefine.
9048 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
9049 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
9050 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
9051 (TARGET_MODES_TIEABLE_P): Redefine.
9052 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
9053 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
9054 * config/pa/pa.c (pa_modes_tieable_p): Make static.
9055 (TARGET_MODES_TIEABLE_P): Redefine.
9056 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
9057 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
9058 (pdp11_modes_tieable_p): New function.
9059 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
9060 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
9061 (rs6000_modes_tieable_p): New function.
9062 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
9063 * config/powerpcspe/powerpcspe.md: Update comment.
9064 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
9065 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
9066 (TARGET_MODES_TIEABLE_P): Redefine.
9067 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
9068 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
9069 (rl78_modes_tieable_p): New function.
9070 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
9071 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
9072 (rs6000_modes_tieable_p): New function.
9073 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
9074 * config/rs6000/rs6000.md: Update comment.
9075 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
9076 * config/rx/rx.c (rx_modes_tieable_p): New function.
9077 (TARGET_MODES_TIEABLE_P): Redefine.
9078 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
9079 * config/s390/s390.c (s390_modes_tieable_p): New function.
9080 (TARGET_MODES_TIEABLE_P): Redefine.
9081 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
9082 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
9083 (sh_modes_tieable_p): New function.
9084 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
9085 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
9086 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
9087 (sparc_modes_tieable_p): Make static.
9088 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
9089 * config/spu/spu.c (spu_modes_tieable_p): New function.
9090 (TARGET_MODES_TIEABLE_P): Redefine.
9091 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
9092 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
9093 (TARGET_MODES_TIEABLE_P): Redefine.
9094 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
9095 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
9096 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
9097 * config/v850/v850.c (v850_modes_tieable_p): New function.
9098 (TARGET_MODES_TIEABLE_P): Redefine.
9099 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
9100 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
9101 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
9102 (visium_modes_tieable_p): New function.
9103 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
9104 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
9105 (xtensa_modes_tieable_p): New function.
9106 * system.h (MODES_TIEABLE_P): Poison.
9108 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9109 Alan Hayward <alan.hayward@arm.com>
9110 David Sherwood <david.sherwood@arm.com>
9112 * target.def (hard_regno_mode_ok): New hook.
9113 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
9114 (TARGET_HARD_REGNO_MODE_OK): ...this.
9115 * doc/tm.texi.in: Regenerate.
9116 * hooks.h (hook_bool_uint_mode_true): Declare.
9117 * hooks.c (hook_bool_uint_mode_true): New function.
9118 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
9120 * genpreds.c (write_insn_preds_c): Add an include of target.h.
9121 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
9122 instead of HARD_REGNO_MODE_OK.
9123 * caller-save.c: Include target.h.
9124 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
9126 * combine.c (can_combine_p): Likewise.
9127 (combinable_i3pat): Likewise.
9128 (can_change_dest_mode): Likewise.
9129 * expr.c (init_expr_target): Likewise.
9130 (convert_move): Likewise.
9131 (convert_modes): Likewise.
9132 * ira.c (setup_prohibited_class_mode_regs): Likewise.
9133 (setup_prohibited_mode_move_regs): Likewise.
9134 * ira.h (target_ira): Likewise.
9135 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9136 * lra-constraints.c (process_alt_operands): Likewise.
9137 (split_reg): Likewise.
9138 * recog.c (peep2_find_free_register): Likewise.
9139 * ree.c (combine_reaching_defs): Likewise.
9140 * regcprop.c (maybe_mode_change): Likewise.
9141 * reginfo.c (init_reg_sets_1): Likewise.
9142 (choose_hard_reg_mode): Likewise.
9143 (simplifiable_subregs): Likewise.
9144 * regrename.c (check_new_reg_p): Likewise.
9145 * reload.c (find_valid_class): Likewise.
9146 (find_valid_class_1): Likewise.
9147 (reload_inner_reg_of_subreg): Likewise.
9148 (push_reload): Likewise.
9149 (combine_reloads): Likewise.
9150 (find_dummy_reload): Likewise.
9151 (find_reloads): Likewise.
9152 * reload1.c (find_reg): Likewise.
9153 (set_reload_reg): Likewise.
9154 (allocate_reload_reg): Likewise.
9155 (choose_reload_regs): Likewise.
9156 (reload_adjust_reg_for_temp): Likewise.
9157 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
9158 (simplify_subreg_regno): Likewise.
9159 * sel-sched.c (init_regs_for_mode): Likewise.
9160 * varasm.c (make_decl_rtl): Likewise.
9161 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
9162 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
9164 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
9165 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
9166 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9167 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
9168 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
9169 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9170 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
9171 (arc_mode_class): Delete.
9172 (HARD_REGNO_MODE_OK): Delete.
9173 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9174 (arc_hard_regno_mode_ok): Rename old array to...
9175 (arc_hard_regno_mode_ok_modes): ...this.
9176 (arc_conditional_register_usage): Update accordingly.
9177 (arc_mode_class): Make static.
9178 (arc_hard_regno_mode_ok): New function.
9179 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
9180 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
9181 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9182 (arm_hard_regno_mode_ok): Make static.
9183 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
9185 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
9186 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
9187 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
9189 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9190 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
9191 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
9192 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
9193 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
9194 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9195 * config/bfin/predicates.md (valid_reg_operand): Use
9196 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
9197 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
9198 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
9199 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9200 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
9201 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
9202 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9203 (cr16_hard_regno_mode_ok): Make static and return a bool.
9204 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
9205 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9206 (cris_hard_regno_mode_ok): New function.
9207 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
9208 (epiphany_mode_class): Delete.
9209 (HARD_REGNO_MODE_OK): Delete.
9210 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
9211 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9212 (hard_regno_mode_ok): Rename to...
9213 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
9214 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
9215 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
9217 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
9218 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
9219 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9220 (frv_hard_regno_mode_ok): Make static and return a bool.
9221 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
9223 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
9224 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
9225 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
9226 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
9228 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9229 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
9230 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
9231 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
9233 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9234 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
9235 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9236 (ia64_hard_regno_mode_ok): New function.
9237 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
9238 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9239 (iq2000_hard_regno_mode_ok): New function.
9240 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
9241 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9242 (lm32_hard_regno_mode_ok): New function.
9243 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
9244 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
9245 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
9246 instead of HARD_REGNO_MODE_OK.
9247 (m32c_hard_regno_ok): Rename to...
9248 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
9249 (m32c_cannot_change_mode_class): Update accordingly.
9250 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9251 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
9252 (m32r_mode_class): Delete.
9253 (HARD_REGNO_MODE_OK): Delete.
9254 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9255 (m32r_hard_regno_mode_ok): Rename to...
9256 (m32r_hard_regno_modes): ...this.
9257 (m32r_mode_class): Make static.
9258 (m32r_hard_regno_mode_ok): New function.
9259 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
9260 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
9261 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9262 (m68k_hard_regno_mode_ok): Make static.
9263 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
9264 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9265 (mcore_hard_regno_mode_ok): New function.
9266 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
9267 (HARD_REGNO_MODE_OK): Delete.
9268 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
9270 (microblaze_hard_regno_mode_ok_p): ...this and make static.
9271 (microblaze_hard_regno_mode_ok): New function.
9272 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9273 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
9274 (mips_hard_regno_mode_ok): Delete.
9275 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
9276 (mips_hard_regno_mode_ok_p): ...this and make static.
9277 (mips_hard_regno_mode_ok_p): Rename to...
9278 (mips_hard_regno_mode_ok_uncached): ...this.
9279 (mips_hard_regno_mode_ok): New function.
9280 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
9281 of HARD_REGNO_MODE_OK.
9282 (mips_option_override): Update after above name changes.
9283 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9284 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
9285 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
9286 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
9287 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
9288 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9289 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
9290 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
9291 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
9292 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9293 (msp430_hard_regno_mode_ok): Make static and return a bool.
9294 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
9295 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
9296 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
9298 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9299 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
9300 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
9301 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
9302 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
9303 (PA_HARD_REGNO_MODE_OK): ...this
9304 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
9305 (PA_HARD_REGNO_MODE_OK): ...this.
9306 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9307 (pa_hard_regno_mode_ok): New function.
9308 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
9309 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9310 (pdp11_hard_regno_mode_ok): New function.
9311 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
9312 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
9314 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
9316 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9317 (rs6000_hard_regno_mode_ok): Rename to...
9318 (rs6000_hard_regno_mode_ok_uncached): ...this.
9319 (rs6000_init_hard_regno_mode_ok): Update accordingly.
9320 (rs6000_hard_regno_mode_ok): New function.
9321 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
9322 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
9323 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
9324 (riscv_hard_regno_mode_ok): ...this and make static.
9325 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9326 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
9327 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
9328 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9329 (rl78_hard_regno_mode_ok): Make static and return bool.
9330 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
9331 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
9333 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
9334 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9335 (rs6000_hard_regno_mode_ok): Rename to...
9336 (rs6000_hard_regno_mode_ok_uncached): ...this.
9337 (rs6000_init_hard_regno_mode_ok): Update accordingly.
9338 (rs6000_hard_regno_mode_ok): New function.
9339 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
9340 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
9341 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9342 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
9343 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
9344 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
9345 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9346 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
9347 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
9348 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9349 (sh_hard_regno_mode_ok): Make static.
9350 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
9351 instead of HARD_REGNO_MODE_OK.
9352 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
9353 (sparc_mode_class): Delete.
9354 (HARD_REGNO_MODE_OK): Delete.
9355 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9356 (hard_regno_mode_classes): Make static.
9357 (sparc_mode_class): Likewise.
9358 (sparc_hard_regno_mode_ok): New function.
9359 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
9360 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
9361 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
9363 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9364 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
9365 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
9366 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
9367 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
9368 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9369 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
9370 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
9371 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9372 (visium_hard_regno_mode_ok): New function.
9373 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
9374 instead of HARD_REGNO_MODE_OK.
9375 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
9376 (HARD_REGNO_MODE_OK): Delete.
9377 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
9378 (xtensa_hard_regno_mode_ok_p): ...this and make static.
9379 (xtensa_option_override): Update accordingly.
9380 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9381 (xtensa_hard_regno_mode_ok): New function.
9382 * system.h (HARD_REGNO_MODE_OK): Poison.
9384 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9385 Alan Hayward <alan.hayward@arm.com>
9386 David Sherwood <david.sherwood@arm.com>
9388 * target.def (hard_regno_call_part_clobbered): New hook.
9389 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
9390 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
9391 * doc/tm.texi: Regenerate.
9392 * hooks.h (hook_bool_uint_mode_false): Declare.
9393 * hooks.c (hook_bool_uint_mode_false): New function.
9394 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9395 * cselib.c (cselib_process_insn): Use
9396 targetm.hard_regno_call_part_clobbered instead of
9397 HARD_REGNO_CALL_PART_CLOBBERED.
9398 * ira-conflicts.c (ira_build_conflicts): Likewise.
9399 * ira-costs.c (ira_tune_allocno_costs): Likewise.
9400 * lra-constraints.c (need_for_call_save_p): Likewise.
9401 * lra-lives.c: Include target.h.
9402 (check_pseudos_live_through_calls): Use
9403 targetm.hard_regno_call_part_clobbered instead of
9404 HARD_REGNO_CALL_PART_CLOBBERED.
9405 * regcprop.c: Include target.h.
9406 (copyprop_hardreg_forward_1): Use
9407 targetm.hard_regno_call_part_clobbered instead of
9408 HARD_REGNO_CALL_PART_CLOBBERED.
9409 * reginfo.c (choose_hard_reg_mode): Likewise.
9410 * regrename.c (check_new_reg_p): Likewise.
9411 * reload.c (find_equiv_reg): Likewise.
9412 * reload1.c (emit_reload_insns): Likewise.
9413 * sched-deps.c (deps_analyze_insn): Likewise.
9414 * sel-sched.c (init_regs_for_mode): Likewise.
9415 (mark_unavailable_hard_regs): Likewise.
9416 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
9417 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9418 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
9420 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9421 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9422 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
9424 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
9426 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9427 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9428 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
9430 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9431 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9432 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
9434 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9435 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
9437 * config/powerpcspe/powerpcspe.c
9438 (rs6000_hard_regno_call_part_clobbered): New function.
9439 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9440 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9441 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
9443 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9444 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9445 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
9447 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9448 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9449 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
9451 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9452 Alan Hayward <alan.hayward@arm.com>
9453 David Sherwood <david.sherwood@arm.com>
9455 * rtl.h (subreg_memory_offset): Declare.
9456 * emit-rtl.c (subreg_memory_offset): New function.
9457 * expmed.c (store_bit_field_1): Use it.
9458 * expr.c (undefined_operand_subword_p): Likewise.
9459 * simplify-rtx.c (simplify_subreg): Likewise.
9461 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
9463 PR rtl-optimization/57448
9466 * optabs.c (expand_atomic_load): Place compiler memory barriers if
9467 using atomic_load pattern.
9468 (expand_atomic_store): Likewise.
9470 2017-09-04 Jakub Jelinek <jakub@redhat.com>
9473 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
9474 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
9475 handling. Use replace_call_with_value with NULL instead of
9476 gsi_replace, unlink_stmt_vdef and release_defs.
9478 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
9481 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
9483 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9486 * rtl.h (emit_library_call_value_1): Declare.
9487 (emit_library_call): Replace declaration with a series of overloads.
9488 Remove the parameter count argument.
9489 (emit_library_call_value): Likewise.
9490 * calls.c (emit_library_call_value_1): Make global. Replace varargs
9491 with an "rtx_mode_t *".
9492 (emit_library_call_value): Delete.
9493 (emit_library_call): Likewise.
9494 * asan.c (asan_emit_stack_protection): Update calls accordingly.
9495 (asan_emit_allocas_unpoison): Likewise.
9496 * builtins.c (expand_builtin_powi): Likewise.
9497 (expand_asan_emit_allocas_unpoison): Likewise.
9498 * cfgexpand.c (expand_main_function): Likewise.
9499 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
9500 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
9501 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
9502 * config/arm/arm.c (arm_trampoline_init): Likewise.
9503 (arm_call_tls_get_addr): Likewise.
9504 (arm_expand_divmod_libfunc): Likewise.
9505 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
9506 (smulsi3_highpart): Likewise.
9507 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
9508 (c6x_expand_compare): Likewise.
9509 (c6x_expand_movmem): Likewise.
9510 * config/frv/frv.c (frv_trampoline_init): Likewise.
9511 * config/i386/i386.c (ix86_trampoline_init): Likewise.
9512 (ix86_expand_divmod_libfunc): Likewise.
9513 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
9514 (ia64_expand_compare): Likewise.
9515 (ia64_profile_hook): Likewise.
9516 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
9517 (nonlocal_goto): Likewise.
9518 (restore_stack_nonlocal): Likewise.
9519 * config/m32r/m32r.c (block_move_call): Likewise.
9520 (m32r_trampoline_init): Likewise.
9521 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
9522 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
9523 (m68k_call_m68k_read_tp): Likewise.
9524 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
9525 (microblaze_expand_divide): Likewise.
9526 * config/mips/mips.h (mips_args): Likewise.
9527 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
9528 (MIPS_ICACHE_SYNC): Likewise.
9529 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
9530 (nios2_trampoline_init): Likewise.
9531 * config/pa/pa.c (hppa_tls_call): Likewise.
9532 (pa_trampoline_init): Likewise.
9533 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
9534 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
9535 (expand_strn_compare): Likewise.
9536 (rs6000_generate_compare): Likewise.
9537 (rs6000_expand_float128_convert): Likewise.
9538 (output_profile_hook): Likewise.
9539 (rs6000_trampoline_init): Likewise.
9540 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
9541 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
9542 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
9543 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
9544 (rs6000_generate_compare): Likewise.
9545 (rs6000_expand_float128_convert): Likewise.
9546 (output_profile_hook): Likewise.
9547 (rs6000_trampoline_init): Likewise.
9548 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
9549 * config/sh/sh.c (sh_trampoline_init): Likewise.
9550 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
9551 (sparc_emit_float_lib_cmp): Likewise.
9552 (sparc32_initialize_trampoline): Likewise.
9553 (sparc64_initialize_trampoline): Likewise.
9554 (sparc_profile_hook): Likewise.
9555 * config/spu/spu.c (ea_load_store): Likewise.
9556 * config/spu/spu.md (floatunssidf2): Likewise.
9557 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
9558 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
9559 * config/visium/visium.c (expand_block_move_4): Likewise.
9560 (expand_block_move_2): Likewise.
9561 (expand_block_move_1): Likewise.
9562 (expand_block_set_4): Likewise.
9563 (expand_block_set_2): Likewise.
9564 (expand_block_set_1): Likewise.
9565 (visium_trampoline_init): Likewise.
9566 (visium_profile_hook): Likewise.
9567 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
9568 (xtensa_setup_frame_addresses): Likewise.
9569 (xtensa_trampoline_init): Likewise.
9570 * except.c (sjlj_emit_function_enter): Likewise.
9571 (sjlj_emit_function_exit): Likewise.
9572 * explow.c (allocate_dynamic_stack_space): Likewise.
9573 (probe_stack_range): Likewise.
9574 * expr.c (convert_mode_scalar): Likewise.
9575 * optabs.c (expand_binop): Likewise.
9576 (expand_twoval_binop_libfunc): Likewise.
9577 (expand_unop): Likewise.
9578 (prepare_cmp_insn): Likewise.
9579 (prepare_float_lib_cmp): Likewise.
9580 (expand_float): Likewise.
9581 (expand_fix): Likewise.
9582 (expand_fixed_convert): Likewise.
9583 (maybe_emit_sync_lock_test_and_set): Likewise.
9584 (expand_atomic_compare_and_swap): Likewise.
9585 (expand_mem_thread_fence): Likewise.
9586 (expand_atomic_fetch_op): Likewise.
9588 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
9590 * doc/generic.texi (OpenACC): Adjust URL.
9591 * doc/invoke.texi (C Dialect Options): Ditto.
9593 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
9595 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
9596 predicate for operand 1. Add (m,<S>) constraint.
9597 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
9598 Prevent memory operand 1 with register operand 2.
9600 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
9602 PR rtl-optimization/82024
9603 * combine.c (try_combine): If the combination result is a PARALLEL,
9604 and we only need to retain the SET in there that would be placed
9605 at I2, check that we can place that at I3 instead, before doing so.
9607 2017-09-01 Jakub Jelinek <jakub@redhat.com>
9610 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
9612 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
9613 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
9616 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
9617 Jeff Law <law@redhat.com>
9619 * varasm.c (bss_initializer_p): Do not put constants into .bss
9620 (categorize_decl_for_section): Handle bss_initializer_p returning
9621 false when DECL_INITIAL is NULL.
9623 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9626 * config/s390/s390.c (s390_can_inline_p): New function.
9628 2017-09-01 Jeff Law <law@redhat.com>
9630 PR tree-optimization/82052
9631 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
9632 Always initialize the returned slot after a hash table miss
9633 when INSERT is true.
9635 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
9637 * config/s390/s390.md (mem_signal_fence): Remove.
9638 * doc/md.texi (mem_signal_fence): Remove.
9639 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
9641 * target-insns.def (mem_signal_fence): Remove.
9643 2017-09-01 Jakub Jelinek <jakub@redhat.com>
9646 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
9649 * asan.c (create_odr_indicator): Strip name encoding from assembler
9650 name before appending it after __odr_asan_.
9652 2017-09-01 Martin Liska <mliska@suse.cz>
9654 PR tree-optimization/82059
9655 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
9656 frequency only when an edge is redirected.
9658 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9660 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
9661 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
9662 (arc_conditional_register_usage): Remove ARC600 lp_count
9664 (arc_file_start): Emit Tag_ARC_CPU_variation.
9665 (arc_can_use_doloop_p): New conditions to use ZOLs.
9666 (hwloop_fail): New function.
9667 (hwloop_optimize): Likewise.
9668 (hwloop_pattern_reg): Likewise.
9669 (arc_doloop_hooks): New struct, to be used with reorg_loops.
9670 (arc_reorg_loops): New function, calls reorg_loops.
9671 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
9672 (arc600_corereg_hazard): Remove ZOL checking, case handled by
9674 (arc_loop_hazard): Remove function, functionality moved into
9676 (arc_hazard): Remove arc_loop_hazard call.
9677 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
9678 into hwloop_optimize.
9679 (arc_label_align): Remove ZOL handling.
9680 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
9681 * config/arc/arc.md (doloop_begin): Remove pattern.
9682 (doloop_begin_i): Likewise.
9683 (doloop_end_i): Likewise.
9684 (doloop_fallback): Likewise.
9685 (doloop_fallback_m): Likewise.
9686 (doloop_end): Reimplement expand.
9687 (arc_lp): New pattern for LP instruction.
9688 (loop_end): New pattern.
9689 (loop_fail): Likewise.
9690 (decrement_and_branch_until_zero): Likewise.
9691 * config/arc/arc.opt (mlpc-width): New option.
9692 * doc/invoke.texi (mlpc-width): Document option.
9694 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9696 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
9697 (arc_ccfsm_advance): Fix checking for delay slots.
9698 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
9700 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9702 * config/arc/arc.md (movqi_insn): Add stores to save constant long
9704 (movhi_insn): Update store instruction constraint which are saving
9705 6-bit short immediates.
9706 (movsi_insn): Consider also short scaled load operations.
9707 (zero_extendhisi2_i): Use Usd constraint instead of T.
9708 (extendhisi2_i): Add q constraint.
9709 (arc_clzsi2): Add type and length attributes.
9710 (arc_ctzsi2): Likewise.
9711 * config/arc/constraints.md (Usc): Update constraint, the
9712 assembler can parse two relocations for a single instruction.
9714 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9716 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
9717 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
9719 2017-08-31 Olivier Hainque <hainque@adacore.com>
9721 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
9722 match as powerpc-wrs-vxworks*.
9724 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
9726 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
9727 register constraint for by-element operand.
9728 (aarch64_mls_elt_merge<mode>): Likewise.
9730 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9732 * config/arc/arc.c (arc_can_follow_jump): Check for short
9735 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9737 * config.gcc: Use g.opt for arc.
9738 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
9739 functionality moved to ...
9740 (legitimate_scaled_address_p): New function, ...here.
9741 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
9742 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
9743 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
9745 (arc_override_options): Handle G option.
9746 (arc_output_pic_addr_const): Correct function definition.
9747 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
9748 (arc_decl_anon_ns_mem_p): Delete.
9749 (arc_in_small_data_p): Overhaul this function to take into
9750 consideration the value given via G option.
9751 (arc_rewrite_small_data_1): Renamed and corrected old
9752 arc_rewrite_small_data function.
9753 (arc_rewrite_small_data): New function.
9754 (small_data_pattern): Don't use pic_offset_table_rtx.
9755 * config/arc/arc.h (CC1_SPEC): Recognize G option.
9756 * config/arc/simdext.md (movmisalignv2hi): Use
9757 prepare_move_operands function.
9759 (movmisalign*): Likewise.
9760 * doc/invoke.texi (ARC options): Document -G option.
9762 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9764 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
9766 * config/arc/arc.c (arc_print_operand): Output scalled address for
9767 sdata whenever is possible.
9768 (arc_in_small_data_p): Allow sdata for 64bit datum when double
9769 load/stores are available.
9770 (compact_sda_memory_operand): Check for the alignment required by
9771 code density instructions.
9772 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
9774 * config/arc/constraints.md (Usd): Update constraint.
9775 (Us0): New constraint.
9776 (Usc): Update constraint.
9778 2017-08-31 Richard Biener <rguenther@suse.de>
9781 * dwarf2out.c (dwarf2out_early_global_decl): Process each
9784 2017-08-31 Tamar Christina <tamar.christina@arm.com>
9786 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
9787 Resize type_signature.
9789 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
9790 Alan Hayward <alan.hayward@arm.com>
9791 David Sherwood <david.sherwood@arm.com>
9793 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
9794 subregs whose inner modes can be stored in GPRs.
9795 (aarch64_classify_index): Likewise.
9797 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
9798 Alan Hayward <alan.hayward@arm.com>
9799 David Sherwood <david.sherwood@arm.com>
9801 * config/aarch64/iterators.md (V_cmp_result): Rename to...
9802 (V_INT_EQUIV): ...this.
9803 (v_cmp_result): Rename to...
9804 (v_int_equiv): ...this.
9805 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
9806 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
9807 (copysign<mode>3): Likewise.
9808 (aarch64_simd_bsl<mode>_internal): Likewise.
9809 (aarch64_simd_bsl<mode>): Likewise.
9810 (vec_cmp<mode><mode>): Likewise.
9811 (vcond<mode><mode>): Likewise.
9812 (vcond<v_cmp_mixed><mode>): Likewise.
9813 (vcondu<mode><v_cmp_mixed>): Likewise.
9814 (aarch64_cm<optab><mode>): Likewise.
9815 (aarch64_cmtst<mode>): Likewise.
9816 (aarch64_fac<optab><mode>): Likewise.
9817 (vec_perm_const<mode>): Likewise.
9818 (vcond_mask_<mode><v_cmp_result>): Rename to...
9819 (vcond_mask_<mode><v_int_equiv>): ...this.
9820 (vec_cmp<mode><v_cmp_result>): Rename to...
9821 (vec_cmp<mode><v_int_equiv>): ...this.
9823 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
9824 Alan Hayward <alan.hayward@arm.com>
9825 David Sherwood <david.sherwood@arm.com>
9827 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
9829 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
9830 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
9831 (UNSPEC_LD4_DREG): New unspecs.
9832 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
9833 (aarch64_ld2<mode>_dreg_be): Replace with...
9834 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
9836 (aarch64_ld3<mode>_dreg_le)
9837 (aarch64_ld3<mode>_dreg_be): Replace with...
9838 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
9840 (aarch64_ld4<mode>_dreg_le)
9841 (aarch64_ld4<mode>_dreg_be): Replace with...
9842 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
9845 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9847 PR tree-optimization/81987
9848 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
9849 insert an initializer in a location not dominated by the stride
9852 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
9854 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
9855 on the entire header of the finally block in the fallthru case.
9857 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
9859 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
9861 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
9863 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
9864 rs6000_emit_move_from_cr and call renamed function.
9865 (rs6000_emit_prologue): Call renamed functions.
9866 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
9867 movesi_from_cr, remove volatile CRs.
9869 2017-08-30 Jon Beniston <jon@beniston.com>
9870 Richard Biener <rguenther@suse.de>
9872 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
9873 of VECTOR_MODE_P check.
9874 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
9875 element vector types.
9877 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9879 * df.h (df_read_modify_subreg_p): Remove in favor of...
9880 * rtl.h (read_modify_subreg_p): ...this new function. Take a
9881 const_rtx instead of an rtx.
9882 * cprop.c (local_cprop_find_used_regs): Update accordingly.
9883 * df-problems.c (df_word_lr_mark_ref): Likewise.
9884 * ira-lives.c (mark_pseudo_reg_live): Likewise.
9885 (mark_pseudo_reg_dead): Likewise.
9886 (mark_ref_dead): Likewise.
9887 * reginfo.c (init_subregs_of_mode): Likewise.
9888 * sched-deps.c (sched_analyze_1): Likewise.
9889 * df-scan.c (df_def_record_1): Likewise.
9890 (df_uses_record): Likewise.
9891 (df_read_modify_subreg_p): Remove in favor of...
9892 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
9893 const_rtx instead of an rtx.
9895 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9896 Alan Hayward <alan.hayward@arm.com>
9897 David Sherwood <david.sherwood@arm.com>
9899 * rtl.h (partial_subreg_p): New function.
9900 * caller-save.c (save_call_clobbered_regs): Use it.
9901 * calls.c (expand_call): Likewise.
9902 * combine.c (combinable_i3pat): Likewise.
9903 (simplify_set): Likewise.
9904 (make_extraction): Likewise.
9905 (make_compound_operation_int): Likewise.
9906 (gen_lowpart_or_truncate): Likewise.
9907 (force_to_mode): Likewise.
9908 (make_field_assignment): Likewise.
9909 (reg_truncated_to_mode): Likewise.
9910 (record_truncated_value): Likewise.
9911 (move_deaths): Likewise.
9912 * cse.c (record_jump_cond): Likewise.
9913 (cse_insn): Likewise.
9914 * cselib.c (cselib_lookup_1): Likewise.
9915 * expmed.c (extract_bit_field_using_extv): Likewise.
9916 * function.c (assign_parm_setup_reg): Likewise.
9917 * ifcvt.c (noce_convert_multiple_sets): Likewise.
9918 * ira-build.c (create_insn_allocnos): Likewise.
9919 * lra-coalesce.c (merge_pseudos): Likewise.
9920 * lra-constraints.c (match_reload): Likewise.
9921 (simplify_operand_subreg): Likewise.
9922 (curr_insn_transform): Likewise.
9923 * lra-lives.c (process_bb_lives): Likewise.
9924 * lra.c (new_insn_reg): Likewise.
9925 (lra_substitute_pseudo): Likewise.
9926 * regcprop.c (mode_change_ok): Likewise.
9927 (maybe_mode_change): Likewise.
9928 (copyprop_hardreg_forward_1): Likewise.
9929 * reload.c (push_reload): Likewise.
9930 (find_reloads): Likewise.
9931 (find_reloads_subreg_address): Likewise.
9932 * reload1.c (alter_reg): Likewise.
9933 (eliminate_regs_1): Likewise.
9934 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9936 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
9938 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
9939 back to if statements, including unpack.
9941 2017-08-30 Martin Liska <mliska@suse.cz>
9944 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
9946 (func_checker::compare_asm_inputs_outputs): ... this function.
9947 (func_checker::compare_gimple_asm): Use the function to compare
9948 also ASM constrains.
9949 * ipa-icf-gimple.h: Rename the function.
9951 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9952 Alan Hayward <alan.hayward@arm.com>
9953 David Sherwood <david.sherwood@arm.com>
9955 * coretypes.h (complex_mode): New type.
9956 * gdbhooks.py (build_pretty_printer): Handle it.
9957 * machmode.h (complex_mode): New class.
9958 (complex_mode::includes_p): New function.
9959 (is_complex_int_mode): Likewise.
9960 (is_complex_float_mode): Likewise.
9961 * genmodes.c (get_mode_class): Handle complex mode classes.
9962 * function.c (expand_function_end): Use is_complex_int_mode.
9964 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9965 Alan Hayward <alan.hayward@arm.com>
9966 David Sherwood <david.sherwood@arm.com>
9968 * coretypes.h (scalar_mode_pod): New typedef.
9969 * gdbhooks.py (build_pretty_printer): Handle it.
9970 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
9971 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
9972 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
9973 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
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 * machmode.h (mode_for_vector): Take a scalar_mode instead
9982 * stor-layout.c (mode_for_vector): Likewise.
9983 * explow.c (promote_mode): Use as_a <scalar_mode>.
9984 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
9986 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9987 Alan Hayward <alan.hayward@arm.com>
9988 David Sherwood <david.sherwood@arm.com>
9990 * target.def (preferred_simd_mode): Take a scalar_mode
9991 instead of a machine_mode.
9992 * targhooks.h (default_preferred_simd_mode): Likewise.
9993 * targhooks.c (default_preferred_simd_mode): Likewise.
9994 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
9995 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
9996 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
9997 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
9998 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
9999 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
10000 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
10001 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
10003 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
10004 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
10005 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
10006 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
10007 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
10008 * doc/tm.texi: Regenerate.
10009 * optabs-query.c (can_vec_mask_load_store_p): Return false for
10012 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10013 Alan Hayward <alan.hayward@arm.com>
10014 David Sherwood <david.sherwood@arm.com>
10016 * target.def (scalar_mode_supported_p): Take a scalar_mode
10017 instead of a machine_mode.
10018 * targhooks.h (default_scalar_mode_supported_p): Likewise.
10019 * targhooks.c (default_scalar_mode_supported_p): Likewise.
10020 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
10021 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
10022 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
10023 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
10024 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
10025 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
10026 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
10027 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
10028 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
10029 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
10030 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
10031 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
10033 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
10034 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
10035 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
10036 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
10037 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
10039 * doc/tm.texi: Regenerate.
10041 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10042 Alan Hayward <alan.hayward@arm.com>
10043 David Sherwood <david.sherwood@arm.com>
10045 * coretypes.h (opt_scalar_mode): New typedef.
10046 * gdbhooks.py (build_pretty_printers): Handle it.
10047 * machmode.h (mode_iterator::get_2xwider): Add overload for
10049 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
10051 * expr.c (convert_mode_scalar): Likewise.
10052 * omp-low.c (omp_clause_aligned_alignment): Likewise.
10053 * optabs.c (expand_float): Likewise.
10054 (expand_fix): Likewise.
10055 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
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 * optabs.c (expand_float): Explicitly check for scalars before
10062 using a branching expansion.
10063 (expand_fix): Likewise.
10065 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10066 Alan Hayward <alan.hayward@arm.com>
10067 David Sherwood <david.sherwood@arm.com>
10069 * expr.c (convert_mode): Split scalar handling out into...
10070 (convert_mode_scalar): ...this new function. Treat the modes
10073 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10074 Alan Hayward <alan.hayward@arm.com>
10075 David Sherwood <david.sherwood@arm.com>
10077 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
10079 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
10081 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10082 Alan Hayward <alan.hayward@arm.com>
10083 David Sherwood <david.sherwood@arm.com>
10085 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
10086 rather than a machine_mode.
10087 (fixed_from_string): Likewise.
10088 (fixed_convert): Likewise.
10089 (fixed_convert_from_int): Likewise.
10090 (fixed_convert_from_real): Likewise.
10091 (real_convert_from_fixed): Likewise.
10092 * fixed-value.c (fixed_from_double_int): Likewise.
10093 (fixed_from_string): Likewise.
10094 (fixed_convert): Likewise.
10095 (fixed_convert_from_int): Likewise.
10096 (fixed_convert_from_real): Likewise.
10097 (real_convert_from_fixed): Likewise.
10098 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
10100 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10101 Alan Hayward <alan.hayward@arm.com>
10102 David Sherwood <david.sherwood@arm.com>
10104 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
10105 of separate mode class checks. Do not allow vector modes here.
10106 (immed_wide_int_const): Use as_a <scalar_mode>.
10107 * explow.c (trunc_int_for_mode): Likewise.
10108 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
10109 (wi::shwi): Likewise.
10110 (wi::min_value): Likewise.
10111 (wi::max_value): Likewise.
10112 * dwarf2out.c (loc_descriptor): Likewise.
10113 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
10114 for CONST_WIDE_INT.
10116 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10117 Alan Hayward <alan.hayward@arm.com>
10118 David Sherwood <david.sherwood@arm.com>
10120 * tree.h (SCALAR_TYPE_MODE): New macro.
10121 * expr.c (expand_expr_addr_expr_1): Use it.
10122 (expand_expr_real_2): Likewise.
10123 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
10124 (fold_convert_const_fixed_from_int): Likewise.
10125 (fold_convert_const_fixed_from_real): Likewise.
10126 (native_encode_fixed): Likewise
10127 (native_encode_complex): Likewise
10128 (native_encode_vector): Likewise.
10129 (native_interpret_fixed): Likewise.
10130 (native_interpret_real): Likewise.
10131 (native_interpret_complex): Likewise.
10132 (native_interpret_vector): Likewise.
10133 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
10134 (simd_clone_adjust_argument_types): Likewise.
10135 (simd_clone_init_simd_arrays): Likewise.
10136 (simd_clone_adjust): Likewise.
10137 * stor-layout.c (layout_type): Likewise.
10138 * tree.c (build_minus_one_cst): Likewise.
10139 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
10140 * tree-inline.c (estimate_move_cost): Likewise.
10141 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
10142 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
10143 (vectorizable_reduction): Likewise.
10144 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
10145 (vect_recog_mixed_size_cond_pattern): Likewise.
10146 (check_bool_pattern): Likewise.
10147 (adjust_bool_pattern): Likewise.
10148 (search_type_for_mask_1): Likewise.
10149 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
10150 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10151 (vectorizable_load): Likewise.
10152 (vectorizable_store): Likewise.
10153 * ubsan.c (ubsan_encode_value): Likewise.
10154 * varasm.c (output_constant): Likewise.
10156 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10157 Alan Hayward <alan.hayward@arm.com>
10158 David Sherwood <david.sherwood@arm.com>
10160 * coretypes.h (scalar_mode): New class.
10161 * machmode.h (scalar_mode): Likewise.
10162 (scalar_mode::includes_p): New function.
10163 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
10164 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
10165 * genmodes.c (get_mode_class): Handle remaining scalar modes.
10166 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
10167 * expmed.c (store_bit_field_1): Likewise.
10168 (extract_bit_field_1): Likewise.
10169 * expr.c (write_complex_part): Likewise.
10170 (read_complex_part): Likewise.
10171 (emit_move_complex_push): Likewise.
10172 (expand_expr_real_2): Likewise.
10173 * function.c (assign_parm_setup_reg): Likewise.
10174 (assign_parms_unsplit_complex): Likewise.
10175 * optabs.c (expand_binop): Likewise.
10176 * rtlanal.c (subreg_get_info): Likewise.
10177 * simplify-rtx.c (simplify_immed_subreg): Likewise.
10178 * varasm.c (output_constant_pool_2): Likewise.
10180 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10181 Alan Hayward <alan.hayward@arm.com>
10182 David Sherwood <david.sherwood@arm.com>
10184 * expmed.c (extract_high_half): Use scalar_int_mode and remove
10186 (expmed_mult_highpart_optab): Likewise.
10187 (expmed_mult_highpart): Likewise.
10189 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10190 Alan Hayward <alan.hayward@arm.com>
10191 David Sherwood <david.sherwood@arm.com>
10193 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
10194 instead of a machine_mode.
10195 (builtin_memset_read_str): Likewise.
10196 * builtins.c (c_readstr): Likewise.
10197 (builtin_memcpy_read_str): Likewise.
10198 (builtin_strncpy_read_str): Likewise.
10199 (builtin_memset_read_str): Likewise.
10200 (builtin_memset_gen_str): Likewise.
10201 (expand_builtin_signbit): Use scalar_int_mode for local variables.
10202 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
10203 instead of a machine_mode.
10204 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
10206 (make_extraction): Likewise.
10207 (try_widen_shift_mode): Take and return scalar_int_modes instead
10209 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
10210 a scalar_int_mode instead of a machine_mode.
10211 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
10212 (avr_addr_space_pointer_mode): Likewise.
10213 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
10214 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
10215 (msp430_unwind_word_mode): Likewise.
10216 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
10217 (spu_addr_space_pointer_mode): Likewise.
10218 (spu_addr_space_address_mode): Likewise.
10219 (spu_libgcc_cmp_return_mode): Likewise.
10220 (spu_libgcc_shift_count_mode): Likewise.
10221 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
10222 (rl78_addr_space_pointer_mode): Likewise.
10223 (fl78_unwind_word_mode): Likewise.
10224 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
10226 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
10227 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
10228 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
10229 (mips_valid_pointer_mode): Likewise.
10230 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
10231 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
10232 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
10234 (ft32_addr_space_address_mode): Likewise.
10235 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
10236 scalar_int_mode instead of a machine_mode.
10237 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
10239 (m32c_addr_space_address_mode): Likewise.
10240 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
10241 (rs6000_eh_return_filter_mode): Likewise.
10242 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
10243 (rs6000_eh_return_filter_mode): Likewise.
10244 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
10245 (s390_libgcc_shift_count_mode): Likewise.
10246 (s390_unwind_word_mode): Likewise.
10247 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
10249 * target.def (mode_rep_extended): Likewise.
10250 (valid_pointer_mode): Likewise.
10251 (addr_space.valid_pointer_mode): Likewise.
10252 (eh_return_filter_mode): Return a scalar_int_mode rather than
10254 (libgcc_cmp_return_mode): Likewise.
10255 (libgcc_shift_count_mode): Likewise.
10256 (unwind_word_mode): Likewise.
10257 (addr_space.pointer_mode): Likewise.
10258 (addr_space.address_mode): Likewise.
10259 * doc/tm.texi: Regenerate.
10260 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
10262 (do_jump): Use scalar_int_mode for local variables.
10263 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
10264 rather than a machine_mode.
10265 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
10266 (scompare_loc_descriptor_wide): Likewise.
10267 (scompare_loc_descriptor_narrow): Likewise.
10268 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
10270 * except.c (sjlj_emit_dispatch_table): Likewise.
10271 (expand_builtin_eh_copy_values): Likewise.
10272 * explow.c (convert_memory_address_addr_space_1): Likewise.
10273 Take a scalar_int_mode rather than a machine_mode.
10274 (convert_memory_address_addr_space): Take a scalar_int_mode rather
10275 than a machine_mode.
10276 (memory_address_addr_space): Use scalar_int_mode for local variables.
10277 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
10278 rather than a machine_mode.
10279 * expmed.c (mask_rtx): Likewise.
10280 (init_expmed_one_conv): Likewise.
10281 (expand_mult_highpart_adjust): Likewise.
10282 (extract_high_half): Likewise.
10283 (expmed_mult_highpart_optab): Likewise.
10284 (expmed_mult_highpart): Likewise.
10285 (expand_smod_pow2): Likewise.
10286 (expand_sdiv_pow2): Likewise.
10287 (emit_store_flag_int): Likewise.
10288 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
10290 (extract_low_bits): Likewise.
10291 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
10293 * expr.c (pieces_addr::adjust): Likewise.
10294 (can_store_by_pieces): Likewise.
10295 (store_by_pieces): Likewise.
10296 (clear_by_pieces_1): Likewise.
10297 (expand_expr_addr_expr_1): Likewise.
10298 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
10299 (expand_expr_real_1): Likewise.
10300 (try_casesi): Likewise.
10301 * final.c (shorten_branches): Likewise.
10302 * fold-const.c (fold_convert_const_int_from_fixed): Change the
10303 type of "mode" to machine_mode.
10304 * internal-fn.c (expand_arith_overflow_result_store): Take a
10305 scalar_int_mode rather than a machine_mode.
10306 (expand_mul_overflow): Use scalar_int_mode for local variables.
10307 * loop-doloop.c (doloop_modify): Likewise.
10308 (doloop_optimize): Likewise.
10309 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
10310 than a machine_mode.
10311 (expand_doubleword_shift_condmove): Likewise.
10312 (expand_doubleword_shift): Likewise.
10313 (expand_doubleword_clz): Likewise.
10314 (expand_doubleword_popcount): Likewise.
10315 (expand_doubleword_parity): Likewise.
10316 (expand_absneg_bit): Use scalar_int_mode for local variables.
10317 (prepare_float_lib_cmp): Likewise.
10318 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
10319 rather than a machine_mode.
10320 (convert_memory_address_addr_space): Likewise.
10321 (get_mode_bounds): Likewise.
10322 (get_address_mode): Return a scalar_int_mode rather than a
10324 * rtlanal.c (get_address_mode): Likewise.
10325 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
10326 than a machine_mode.
10327 * targhooks.c (default_mode_rep_extended): Likewise.
10328 (default_valid_pointer_mode): Likewise.
10329 (default_addr_space_valid_pointer_mode): Likewise.
10330 (default_eh_return_filter_mode): Return a scalar_int_mode rather
10331 than a machine_mode.
10332 (default_libgcc_cmp_return_mode): Likewise.
10333 (default_libgcc_shift_count_mode): Likewise.
10334 (default_unwind_word_mode): Likewise.
10335 (default_addr_space_pointer_mode): Likewise.
10336 (default_addr_space_address_mode): Likewise.
10337 * targhooks.h (default_eh_return_filter_mode): Likewise.
10338 (default_libgcc_cmp_return_mode): Likewise.
10339 (default_libgcc_shift_count_mode): Likewise.
10340 (default_unwind_word_mode): Likewise.
10341 (default_addr_space_pointer_mode): Likewise.
10342 (default_addr_space_address_mode): Likewise.
10343 (default_mode_rep_extended): Take a scalar_int_mode rather than
10345 (default_valid_pointer_mode): Likewise.
10346 (default_addr_space_valid_pointer_mode): Likewise.
10347 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
10349 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
10350 rather than a machine_mode.
10351 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
10352 for local variables.
10353 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10354 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
10355 than a machine_mode.
10357 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10358 Alan Hayward <alan.hayward@arm.com>
10359 David Sherwood <david.sherwood@arm.com>
10361 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
10362 the mode argument to scalar_int_mode.
10363 (do_jump_by_parts_zero_rtx): Likewise.
10364 (do_jump_by_parts_equality_rtx): Likewise.
10365 (do_jump_by_parts_greater): Take a mode argument.
10366 (do_jump_by_parts_equality): Likewise.
10367 (do_jump_1): Update calls accordingly.
10369 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10370 Alan Hayward <alan.hayward@arm.com>
10371 David Sherwood <david.sherwood@arm.com>
10373 * is-a.h (safe_dyn_cast): New function.
10374 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
10375 (jump_table_for_label): Likewise.
10376 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
10377 instead of an rtx_insn *.
10378 (shorten_branches): Use dyn_cast instead of LABEL_P and
10379 JUMP_TABLE_DATA_P. Use jump_table_for_label and
10380 rtx_jump_table_data::get_data_mode.
10381 (final_scan_insn): Likewise.
10383 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10384 Alan Hayward <alan.hayward@arm.com>
10385 David Sherwood <david.sherwood@arm.com>
10387 * combine.c (try_combine): Use is_a <scalar_int_mode> when
10388 trying to combine a full-register integer set with a subreg
10391 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10392 Alan Hayward <alan.hayward@arm.com>
10393 David Sherwood <david.sherwood@arm.com>
10395 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
10396 that is always either address_mode or pointer_mode.
10398 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10399 Alan Hayward <alan.hayward@arm.com>
10400 David Sherwood <david.sherwood@arm.com>
10402 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
10403 when the two are known to be equal.
10405 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10406 Alan Hayward <alan.hayward@arm.com>
10407 David Sherwood <david.sherwood@arm.com>
10409 * simplify-rtx.c (simplify_const_unary_operation): Use
10410 is_a <scalar_int_mode> instead of checking for a nonzero
10411 precision. Forcibly convert op_mode to a scalar_int_mode
10412 in that case. More clearly differentiate the operand and
10413 result modes and use the former when deciding what the value
10414 of a count-bits operation should be. Use is_int_mode instead
10415 of checking for a MODE_INT. Remove redundant check for whether
10416 this mode has a zero precision.
10418 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10419 Alan Hayward <alan.hayward@arm.com>
10420 David Sherwood <david.sherwood@arm.com>
10422 * optabs.c (widen_leading): Change the type of the mode argument
10423 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
10424 (widen_bswap): Likewise.
10425 (expand_parity): Likewise.
10426 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
10427 (expand_ffs): Likewise.
10428 (epand_unop): Check for scalar integer modes before calling the
10431 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10432 Alan Hayward <alan.hayward@arm.com>
10433 David Sherwood <david.sherwood@arm.com>
10435 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
10437 (expand_expr_real_1): Update call accordingly.
10439 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10440 Alan Hayward <alan.hayward@arm.com>
10441 David Sherwood <david.sherwood@arm.com>
10443 * expmed.c (store_bit_field_using_insv): Add op0_mode and
10444 value_mode arguments. Use scalar_int_mode internally.
10445 (store_bit_field_1): Rename the new integer mode from imode
10446 to op0_mode and use it instead of GET_MODE (op0). Update calls
10447 to store_split_bit_field, store_bit_field_using_insv and
10448 store_fixed_bit_field.
10449 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
10450 Use scalar_int_mode internally. Use a bit count rather than a mode
10451 when calculating the largest bit size for get_best_mode.
10452 Update calls to store_split_bit_field and store_fixed_bit_field_1.
10453 (store_fixed_bit_field_1): Add mode and value_mode arguments.
10454 Remove assertion that OP0 has a scalar integer mode.
10455 (store_split_bit_field): Add op0_mode and value_mode arguments.
10456 Update calls to extract_fixed_bit_field.
10457 (extract_bit_field_using_extv): Add an op0_mode argument.
10458 Use scalar_int_mode internally.
10459 (extract_bit_field_1): Rename the new integer mode from imode to
10460 op0_mode and use it instead of GET_MODE (op0). Update calls to
10461 extract_split_bit_field, extract_bit_field_using_extv and
10462 extract_fixed_bit_field.
10463 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
10464 to extract_split_bit_field and extract_fixed_bit_field_1.
10465 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
10466 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
10467 on the target mode.
10468 (extract_split_bit_field): Add an op0_mode argument. Update call
10469 to extract_fixed_bit_field.
10471 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10472 Alan Hayward <alan.hayward@arm.com>
10473 David Sherwood <david.sherwood@arm.com>
10475 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
10476 * explow.c (hard_function_value): Likewise.
10477 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
10478 convert_to_mode call outside the loop.
10479 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
10480 for the mode iterator. Require the mode specified by max_pieces
10482 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
10484 (copy_blkmode_to_reg): Likewise.
10485 (set_storage_via_setmem): Likewise.
10486 * optabs.c (prepare_cmp_insn): Likewise.
10487 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
10488 * stor-layout.c (finish_bitfield_representative): Likewise.
10490 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10491 Alan Hayward <alan.hayward@arm.com>
10492 David Sherwood <david.sherwood@arm.com>
10494 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
10495 * expr.c (convert_move): Use them.
10496 (convert_modes): Likewise.
10497 (store_expr_with_bounds): Likewise.
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 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
10504 parameter for the mode of "x". Remove the "known_x", "known_mode"
10505 and "known_ret" arguments. Change the type of the mode argument
10506 to scalar_int_mode.
10507 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
10508 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
10509 (reg_num_sign_bit_copies_for_combine): Likewise.
10510 * rtlanal.c (nonzero_bits1): Likewise.
10511 (num_sign_bit_copies1): Likewise.
10512 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
10513 (reg_num_sign_bit_copies_general): Likewise.
10514 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
10515 (reg_nonzero_bits_general): Likewise.
10517 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10518 Alan Hayward <alan.hayward@arm.com>
10519 David Sherwood <david.sherwood@arm.com>
10521 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
10522 than in subroutines. Return 1 for non-integer modes.
10523 (cached_num_sign_bit_copies): Change the type of the mode parameter
10524 to scalar_int_mode.
10525 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
10526 classes. Handle CONST_INT_P first and then check whether X also
10527 has a scalar integer mode. Check the same thing for inner registers
10528 of a SUBREG and for values that are being extended or truncated.
10530 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10531 Alan Hayward <alan.hayward@arm.com>
10532 David Sherwood <david.sherwood@arm.com>
10534 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
10535 in subroutines. Return the mode mask for non-integer modes.
10536 (cached_nonzero_bits): Change the type of the mode parameter
10537 to scalar_int_mode.
10538 (nonzero_bits1): Likewise. Remove early exit for other mode
10539 classes. Handle CONST_INT_P first and then check whether X
10540 also has a scalar integer mode.
10542 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10543 Alan Hayward <alan.hayward@arm.com>
10544 David Sherwood <david.sherwood@arm.com>
10546 * expr.c (widest_int_mode_for_size): Make the comment match the code.
10547 Return a scalar_int_mode and assert that the size is greater than
10549 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
10550 (op_by_pieces_d::op_by_pieces_d): Likewise.
10551 (op_by_pieces_d::run): Likewise.
10552 (can_store_by_pieces): Likewise.
10554 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10555 Alan Hayward <alan.hayward@arm.com>
10556 David Sherwood <david.sherwood@arm.com>
10558 * combine.c (extract_left_shift): Add a mode argument and update
10560 (make_compound_operation_int): Change the type of the mode parameter
10561 to scalar_int_mode and update the call to extract_left_shift.
10563 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10564 Alan Hayward <alan.hayward@arm.com>
10565 David Sherwood <david.sherwood@arm.com>
10567 * combine.c (simplify_and_const_int): Change the type of the mode
10568 parameter to scalar_int_mode.
10569 (simplify_and_const_int_1): Likewise. Update recursive call.
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 * combine.c (simplify_compare_const): Check that the mode is a
10576 scalar_int_mode (rather than VOIDmode) before testing its
10578 (simplify_comparison): Move COMPARISON_P handling out of the
10579 loop and restrict the latter part of the loop to scalar_int_modes.
10580 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
10581 and when considering SUBREG_REGs. Use is_int_mode instead of
10582 checking GET_MODE_CLASS against MODE_INT.
10584 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10585 Alan Hayward <alan.hayward@arm.com>
10586 David Sherwood <david.sherwood@arm.com>
10588 * combine.c (try_widen_shift_mode): Move check for equal modes to...
10589 (simplify_shift_const_1): ...here. Use scalar_int_mode for
10590 shift_unit_mode and for modes involved in scalar shifts.
10592 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10593 Alan Hayward <alan.hayward@arm.com>
10594 David Sherwood <david.sherwood@arm.com>
10596 * combine.c (force_int_to_mode): New function, split out from...
10597 (force_to_mode): ...here. Keep xmode up-to-date and use it
10598 instead of GET_MODE (x).
10600 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10601 Alan Hayward <alan.hayward@arm.com>
10602 David Sherwood <david.sherwood@arm.com>
10604 * optabs-query.h (extraction_insn::struct_mode): Change type to
10605 opt_scalar_int_mode and update comment.
10606 (extraction_insn::field_mode): Change type to scalar_int_mode.
10607 (extraction_insn::pos_mode): Likewise.
10608 * combine.c (make_extraction): Update accordingly.
10609 * optabs-query.c (get_traditional_extraction_insn): Likewise.
10610 (get_optab_extraction_insn): Likewise.
10611 * recog.c (simplify_while_replacing): Likewise.
10612 * expmed.c (narrow_bit_field_mem): Change the type of the mode
10613 parameter to opt_scalar_int_mode.
10615 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10616 Alan Hayward <alan.hayward@arm.com>
10617 David Sherwood <david.sherwood@arm.com>
10619 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
10620 to a scalar_int_mode instead of a machine_mode.
10621 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
10622 (get_best_mode): Return a boolean and use a pointer argument to store
10623 the selected mode. Replace the limit mode parameter with a bit limit.
10624 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
10625 for the values returned by bit_field_mode_iterator::next_mode.
10626 (store_bit_field): Update call to get_best_mode.
10627 (store_fixed_bit_field): Likewise.
10628 (extract_fixed_bit_field): Likewise.
10629 * expr.c (optimize_bitfield_assignment_op): Likewise.
10630 * fold-const.c (optimize_bit_field_compare): Likewise.
10631 (fold_truth_andor_1): Likewise.
10632 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
10633 Update for new type of m_mode.
10634 (get_best_mode): As above.
10636 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10637 Alan Hayward <alan.hayward@arm.com>
10638 David Sherwood <david.sherwood@arm.com>
10640 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
10641 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
10642 (store_bit_field): Check is_a <scalar_int_mode> before calling
10643 strict_volatile_bitfield_p.
10644 (extract_bit_field): Likewise.
10646 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10647 Alan Hayward <alan.hayward@arm.com>
10648 David Sherwood <david.sherwood@arm.com>
10650 * target.def (cstore_mode): Return a scalar_int_mode.
10651 * doc/tm.texi: Regenerate.
10652 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
10653 * targhooks.h (default_cstore_mode): Likewise.
10654 * targhooks.c (default_cstore_mode): Likewise, using a forced
10656 * expmed.c (emit_cstore): Expect the target of the cstore to be
10659 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10660 Alan Hayward <alan.hayward@arm.com>
10661 David Sherwood <david.sherwood@arm.com>
10663 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
10665 (niter_desc): Likewise mode.
10666 (iv_analyze): Add a mode parameter.
10668 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
10669 and change its type to scalar_int_mode.
10670 * loop-iv.c: Update commentary at head of file.
10671 (iv_constant): Pass the mode paraeter before the rtx it describes
10672 and change its type to scalar_int_mode. Remove VOIDmode handling.
10673 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
10674 (iv_extend): Likewise.
10675 (shorten_into_mode): Likewise.
10676 (iv_add): Use scalar_int_mode.
10677 (iv_mult): Likewise.
10678 (iv_shift): Likewise.
10679 (canonicalize_iv_subregs): Likewise.
10680 (get_biv_step_1): Pass the outer_mode parameter before the rtx
10681 it describes and change its mode to scalar_int_mode. Also change
10682 the type of the returned inner_mode to scalar_int_mode.
10683 (get_biv_step): Likewise, turning outer_mode from a pointer
10684 into a direct parameter. Update call to get_biv_step_1.
10685 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
10686 iv_constant and get_biv_step.
10687 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
10688 and change its type to scalar_int_mode. Don't initialise iv->mode
10689 to VOIDmode and remove later checks for its still being VOIDmode.
10690 Update calls to iv_analyze_op and iv_analyze_expr. Check
10691 is_a <scalar_int_mode> when changing the mode under consideration.
10692 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
10693 Update call to iv_analyze_expr.
10694 (iv_analyze_op): Add a mode parameter. Reject subregs whose
10695 inner register is not also a scalar_int_mode. Update call to
10697 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
10698 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
10699 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
10700 separate mode class checks. Update calls to iv_analyze. Remove
10701 fix-up of VOIDmodes after iv_analyze_biv.
10702 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
10703 don't have a scalar_int_mode. Update call to biv_p.
10705 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10706 Alan Hayward <alan.hayward@arm.com>
10707 David Sherwood <david.sherwood@arm.com>
10709 * cfgexpand.c (convert_debug_memory_address): Use
10710 as_a <scalar_int_mode>.
10711 * combine.c (expand_compound_operation): Likewise.
10712 (make_extraction): Likewise.
10713 (change_zero_ext): Likewise.
10714 (simplify_comparison): Likewise.
10715 * cse.c (cse_insn): Likewise.
10716 * dwarf2out.c (minmax_loc_descriptor): Likewise.
10717 (mem_loc_descriptor): Likewise.
10718 (loc_descriptor): Likewise.
10719 * expmed.c (init_expmed_one_mode): Likewise.
10720 (synth_mult): Likewise.
10721 (emit_store_flag_1): Likewise.
10722 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
10723 of a comparison with size.
10724 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
10725 (reduce_to_bit_field_precision): Likewise.
10726 * function.c (expand_function_end): Likewise.
10727 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
10728 * loop-doloop.c (doloop_modify): Likewise.
10729 * optabs.c (expand_binop): Likewise.
10730 (expand_unop): Likewise.
10731 (expand_copysign_absneg): Likewise.
10732 (prepare_cmp_insn): Likewise.
10733 (maybe_legitimize_operand): Likewise.
10734 * recog.c (const_scalar_int_operand): Likewise.
10735 * rtlanal.c (get_address_mode): Likewise.
10736 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10737 (simplify_cond_clz_ctz): Likewise.
10738 * tree-nested.c (get_nl_goto_field): Likewise.
10739 * tree.c (build_vector_type_for_mode): Likewise.
10740 * var-tracking.c (use_narrower_mode): Likewise.
10742 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10743 Alan Hayward <alan.hayward@arm.com>
10744 David Sherwood <david.sherwood@arm.com>
10746 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
10747 * builtins.c (expand_builtin_signbit): Use it.
10748 * cfgexpand.c (expand_debug_expr): Likewise.
10749 * dojump.c (do_jump): Likewise.
10750 (do_compare_and_jump): Likewise.
10751 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
10752 * expmed.c (make_tree): Likewise.
10753 * expr.c (expand_expr_real_2): Likewise.
10754 (expand_expr_real_1): Likewise.
10755 (try_casesi): Likewise.
10756 * fold-const-call.c (fold_const_call_ss): Likewise.
10757 * fold-const.c (unextend): Likewise.
10758 (extract_muldiv_1): Likewise.
10759 (fold_single_bit_test): Likewise.
10760 (native_encode_int): Likewise.
10761 (native_encode_string): Likewise.
10762 (native_interpret_int): Likewise.
10763 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
10764 * internal-fn.c (expand_addsub_overflow): Likewise.
10765 (expand_neg_overflow): Likewise.
10766 (expand_mul_overflow): Likewise.
10767 (expand_arith_overflow): Likewise.
10768 * match.pd: Likewise.
10769 * stor-layout.c (layout_type): Likewise.
10770 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
10771 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
10772 * tree-ssanames.c (get_range_info): Likewise.
10773 * tree-switch-conversion.c (array_value_type) Likewise.
10774 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
10775 (vect_recog_divmod_pattern): Likewise.
10776 (vect_recog_mixed_size_cond_pattern): Likewise.
10777 * tree-vrp.c (extract_range_basic): Likewise.
10778 (simplify_float_conversion_using_ranges): Likewise.
10779 * tree.c (int_fits_type_p): Likewise.
10780 * ubsan.c (instrument_bool_enum_load): Likewise.
10781 * varasm.c (mergeable_string_section): Likewise.
10782 (narrowing_initializer_constant_valid_p): Likewise.
10783 (output_constant): Likewise.
10785 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10786 Alan Hayward <alan.hayward@arm.com>
10787 David Sherwood <david.sherwood@arm.com>
10789 * machmode.h (NARROWEST_INT_MODE): New macro.
10790 * expr.c (alignment_for_piecewise_move): Use it instead of
10791 GET_CLASS_NARROWEST_MODE (MODE_INT).
10792 (push_block): Likewise.
10793 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
10795 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10797 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10798 Alan Hayward <alan.hayward@arm.com>
10799 David Sherwood <david.sherwood@arm.com>
10801 * postreload.c (move2add_valid_value_p): Change the type of the
10802 mode parameter to scalar_int_mode.
10803 (move2add_use_add2_insn): Add a mode parameter and use it instead
10805 (move2add_use_add3_insn): Likewise.
10806 (reload_cse_move2add): Update accordingly.
10808 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10809 Alan Hayward <alan.hayward@arm.com>
10810 David Sherwood <david.sherwood@arm.com>
10812 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
10814 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
10815 * optabs.c (expand_unop): Likewise.
10817 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10818 Alan Hayward <alan.hayward@arm.com>
10819 David Sherwood <david.sherwood@arm.com>
10821 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
10822 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
10823 (popcount_loc_descriptor): Likewise.
10824 (bswap_loc_descriptor): Likewise.
10825 (rotate_loc_descriptor): Likewise.
10826 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
10827 calling the functions above.
10829 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10830 Alan Hayward <alan.hayward@arm.com>
10831 David Sherwood <david.sherwood@arm.com>
10833 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
10835 (try_combine): Likewise.
10836 (simplify_if_then_else): Likewise.
10837 * cse.c (cse_insn): Likewise.
10838 * dwarf2out.c (mem_loc_descriptor): Likewise.
10839 * emit-rtl.c (gen_lowpart_common): Likewise.
10840 * simplify-rtx.c (simplify_truncation): Likewise.
10841 (simplify_binary_operation_1): Likewise.
10842 (simplify_const_relational_operation): Likewise.
10843 (simplify_ternary_operation): Likewise.
10844 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
10846 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10847 Alan Hayward <alan.hayward@arm.com>
10848 David Sherwood <david.sherwood@arm.com>
10850 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
10851 * reload.c (push_reload): Likewise.
10852 (find_reloads): Likewise.
10854 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10855 Alan Hayward <alan.hayward@arm.com>
10856 David Sherwood <david.sherwood@arm.com>
10858 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
10859 (make_compound_operation_int): Likewise.
10860 (change_zero_ext): Likewise.
10861 * expr.c (convert_move): Likewise.
10862 (convert_modes): Likewise.
10863 * fwprop.c (forward_propagate_subreg): Likewise.
10864 * loop-iv.c (get_biv_step_1): Likewise.
10865 * optabs.c (widen_operand): Likewise.
10866 * postreload.c (move2add_valid_value_p): Likewise.
10867 * recog.c (simplify_while_replacing): Likewise.
10868 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10869 (simplify_binary_operation_1): Likewise. Remove redundant
10870 mode equality check.
10872 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10873 Alan Hayward <alan.hayward@arm.com>
10874 David Sherwood <david.sherwood@arm.com>
10876 * combine.c (combine_simplify_rtx): Add checks for
10877 is_a <scalar_int_mode>.
10878 (simplify_if_then_else): Likewise.
10879 (make_field_assignment): Likewise.
10880 (simplify_comparison): Likewise.
10881 * ifcvt.c (noce_try_bitop): Likewise.
10882 * loop-invariant.c (canonicalize_address_mult): Likewise.
10883 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10885 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10886 Alan Hayward <alan.hayward@arm.com>
10887 David Sherwood <david.sherwood@arm.com>
10889 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
10890 is_a <scalar_int_mode> instead of != BLKmode.
10892 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10893 Alan Hayward <alan.hayward@arm.com>
10894 David Sherwood <david.sherwood@arm.com>
10896 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
10897 instead of != VOIDmode.
10898 * combine.c (if_then_else_cond): Likewise.
10899 (change_zero_ext): Likewise.
10900 * dwarf2out.c (mem_loc_descriptor): Likewise.
10901 (loc_descriptor): Likewise.
10902 * rtlanal.c (canonicalize_condition): Likewise.
10903 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
10905 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10906 Alan Hayward <alan.hayward@arm.com>
10907 David Sherwood <david.sherwood@arm.com>
10909 * simplify-rtx.c (simplify_binary_operation_1): Use
10910 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
10912 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10913 Alan Hayward <alan.hayward@arm.com>
10914 David Sherwood <david.sherwood@arm.com>
10916 * wide-int.h (int_traits<unsigned char>) New class.
10917 (int_traits<unsigned short>) Likewise.
10918 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
10919 Use GET_MODE_UNIT_PRECISION and remove redundant test for
10921 * combine.c (set_nonzero_bits_and_sign_copies): Use
10922 is_a <scalar_int_mode>.
10923 (find_split_point): Likewise.
10924 (combine_simplify_rtx): Likewise.
10925 (simplify_logical): Likewise.
10926 (expand_compound_operation): Likewise.
10927 (expand_field_assignment): Likewise.
10928 (make_compound_operation): Likewise.
10929 (extended_count): Likewise.
10930 (change_zero_ext): Likewise.
10931 (simplify_comparison): Likewise.
10932 * dwarf2out.c (scompare_loc_descriptor): Likewise.
10933 (ucompare_loc_descriptor): Likewise.
10934 (minmax_loc_descriptor): Likewise.
10935 (mem_loc_descriptor): Likewise.
10936 (loc_descriptor): Likewise.
10937 * expmed.c (init_expmed_one_mode): Likewise.
10938 * lra-constraints.c (lra_constraint_offset): Likewise.
10939 * optabs.c (prepare_libcall_arg): Likewise.
10940 * postreload.c (move2add_note_store): Likewise.
10941 * reload.c (operands_match_p): Likewise.
10942 * rtl.h (load_extend_op): Likewise.
10943 * rtlhooks.c (gen_lowpart_general): Likewise.
10944 * simplify-rtx.c (simplify_truncation): Likewise.
10945 (simplify_unary_operation_1): Likewise.
10946 (simplify_binary_operation_1): Likewise.
10947 (simplify_const_binary_operation): Likewise.
10948 (simplify_const_relational_operation): Likewise.
10949 (simplify_subreg): Likewise.
10950 * stor-layout.c (bitwise_mode_for_mode): Likewise.
10951 * var-tracking.c (adjust_mems): Likewise.
10952 (prepare_call_arguments): Likewise.
10954 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10955 Alan Hayward <alan.hayward@arm.com>
10956 David Sherwood <david.sherwood@arm.com>
10958 * machmode.h (is_int_mode): New fuction.
10959 * combine.c (find_split_point): Use it.
10960 (combine_simplify_rtx): Likewise.
10961 (simplify_if_then_else): Likewise.
10962 (simplify_set): Likewise.
10963 (simplify_shift_const_1): Likewise.
10964 (simplify_comparison): Likewise.
10965 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
10966 * cse.c (notreg_cost): Likewise.
10967 (cse_insn): Likewise.
10968 * cselib.c (cselib_lookup_1): Likewise.
10969 * dojump.c (do_jump_1): Likewise.
10970 (do_compare_rtx_and_jump): Likewise.
10971 * dse.c (get_call_args): Likewise.
10972 * dwarf2out.c (rtl_for_decl_init): Likewise.
10973 (native_encode_initializer): Likewise.
10974 * expmed.c (emit_store_flag_1): Likewise.
10975 (emit_store_flag): Likewise.
10976 * expr.c (convert_modes): Likewise.
10977 (store_field): Likewise.
10978 (expand_expr_real_1): Likewise.
10979 * fold-const.c (fold_read_from_constant_string): Likewise.
10980 * gimple-ssa-sprintf.c (get_format_string): Likewise.
10981 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
10982 * optabs.c (expand_binop): Likewise.
10983 (expand_unop): Likewise.
10984 (expand_abs_nojump): Likewise.
10985 (expand_one_cmpl_abs_nojump): Likewise.
10986 * simplify-rtx.c (mode_signbit_p): Likewise.
10987 (val_signbit_p): Likewise.
10988 (val_signbit_known_set_p): Likewise.
10989 (val_signbit_known_clear_p): Likewise.
10990 (simplify_relational_operation_1): Likewise.
10991 * tree.c (vector_type_mode): Likewise.
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 * machmode.h (smallest_mode_for_size): Fix formatting.
10998 (smallest_int_mode_for_size): New function.
10999 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
11000 instead of smallest_mode_for_size.
11001 * combine.c (make_extraction): Likewise.
11002 * config/arc/arc.c (arc_expand_movmem): Likewise.
11003 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
11004 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
11005 * config/s390/s390.c (s390_expand_insv): Likewise.
11006 * config/sparc/sparc.c (assign_int_registers): Likewise.
11007 * config/spu/spu.c (spu_function_value): Likewise.
11008 (spu_function_arg): Likewise.
11009 * coverage.c (get_gcov_type): Likewise.
11010 (get_gcov_unsigned_t): Likewise.
11011 * dse.c (find_shift_sequence): Likewise.
11012 * expmed.c (store_bit_field_1): Likewise.
11013 * expr.c (convert_move): Likewise.
11014 (store_field): Likewise.
11015 * internal-fn.c (expand_arith_overflow): Likewise.
11016 * optabs-query.c (get_best_extraction_insn): Likewise.
11017 * optabs.c (expand_twoval_binop_libfunc): Likewise.
11018 * stor-layout.c (layout_type): Likewise.
11019 (initialize_sizetypes): Likewise.
11020 * targhooks.c (default_get_mask_mode): Likewise.
11021 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
11023 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11024 Alan Hayward <alan.hayward@arm.com>
11025 David Sherwood <david.sherwood@arm.com>
11027 * machmode.h (opt_mode::else_blk): New function.
11028 (int_mode_for_mode): Declare.
11029 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
11030 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
11032 * cfgexpand.c (expand_debug_expr): Likewise.
11033 * combine.c (gen_lowpart_or_truncate): Likewise.
11034 (gen_lowpart_for_combine): Likewise.
11035 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
11036 * config/avr/avr.c (avr_to_int_mode): Likewise.
11037 (avr_out_plus_1): Likewise.
11038 (avr_out_plus): Likewise.
11039 (avr_out_round): Likewise.
11040 * config/i386/i386.c (ix86_split_to_parts): Likewise.
11041 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
11042 (s390_expand_vcond): Likewise.
11043 * config/spu/spu.c (spu_split_immediate): Likewise.
11044 (spu_expand_mov): Likewise.
11045 * dse.c (get_stored_val): Likewise.
11046 * expmed.c (store_bit_field_1): Likewise.
11047 (convert_extracted_bit_field): Use int_mode_for_mode instead of
11049 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
11050 (extract_low_bits): Likewise.
11051 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
11052 handling rather than repeating the check.
11053 (emit_group_store): Likewise.
11054 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
11055 * optabs.c (expand_absneg_bit): Likewise.
11056 (expand_copysign_absneg): Likewise.
11057 (expand_copysign_bit): Likewise.
11058 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
11059 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
11060 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
11061 * var-tracking.c (prepare_call_arguments): Likewise.
11062 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
11063 int_mode_for_mode instead of mode_for_size.
11064 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
11066 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11067 Alan Hayward <alan.hayward@arm.com>
11068 David Sherwood <david.sherwood@arm.com>
11070 * machmode.h (int_mode_for_size): New function.
11071 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
11072 instead of mode_for_size.
11073 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
11075 * combine.c (expand_field_assignment): Use int_mode_for_size
11076 instead of mode_for_size.
11077 (make_extraction): Likewise.
11078 (simplify_shift_const_1): Likewise.
11079 (simplify_comparison): Likewise.
11080 * dojump.c (do_jump): Likewise.
11081 * dwarf2out.c (mem_loc_descriptor): Likewise.
11082 * emit-rtl.c (init_derived_machine_modes): Likewise.
11083 * expmed.c (flip_storage_order): Likewise.
11084 (convert_extracted_bit_field): Likewise.
11085 * expr.c (copy_blkmode_from_reg): Likewise.
11086 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
11087 * internal-fn.c (expand_mul_overflow): Likewise.
11088 * lower-subreg.c (simple_move): Likewise.
11089 * optabs-libfuncs.c (init_optabs): Likewise.
11090 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11091 * tree.c (vector_type_mode): Likewise.
11092 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
11093 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
11094 * tree-vect-generic.c (expand_vector_parallel): Likewise.
11095 * tree-vect-stmts.c (vectorizable_load): Likewise.
11096 (vectorizable_store): Likewise.
11098 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11099 Alan Hayward <alan.hayward@arm.com>
11100 David Sherwood <david.sherwood@arm.com>
11102 * coretypes.h (pod_mode): New type.
11103 (scalar_int_mode_pod): New typedef.
11104 * machmode.h (pod_mode): New class.
11105 (int_n_data_t::m): Change type to scalar_int_mode_pod.
11106 * genmodes.c (emit_mode_int_n): Update accordingly.
11107 * lower-subreg.h (target_lower_subreg): Change type to
11108 scalar_int_mode_pod.
11109 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
11110 scalar_int_mode_pod.
11112 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11113 Alan Hayward <alan.hayward@arm.com>
11114 David Sherwood <david.sherwood@arm.com>
11116 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
11117 machine_mode to scalar_int_mode.
11118 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
11119 (rs6000_option_override_internal): Remove cast to int.
11120 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
11121 machine_mode to scalar_int_mode.
11122 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
11123 (rs6000_option_override_internal): Remove cast to int.
11124 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
11125 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
11127 * config/s390/s390.c (s390_expand_builtin): Likewise.
11128 * coretypes.h (scalar_int_mode): New type.
11129 (opt_scalar_int_mode): New typedef.
11130 * machmode.h (scalar_int_mode): New class.
11131 (scalar_int_mode::includes_p): New function.
11132 (byte_mode): Change type to scalar_int_mode.
11133 (word_mode): Likewise.
11134 (ptr_mode): Likewise.
11135 * emit-rtl.c (byte_mode): Likewise.
11136 (word_mode): Likewise.
11137 (ptr_mode): Likewise.
11138 (init_derived_machine_modes): Update accordingly.
11139 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
11140 and MODE_PARTIAL_INT.
11141 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
11142 opt_scalar_int_mode.
11144 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11145 Alan Hayward <alan.hayward@arm.com>
11146 David Sherwood <david.sherwood@arm.com>
11148 * target.def (libgcc_floating_mode_supported_p): Take a
11150 * doc/tm.texi: Regenerate.
11151 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
11153 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
11154 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
11157 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11158 Alan Hayward <alan.hayward@arm.com>
11159 David Sherwood <david.sherwood@arm.com>
11161 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
11162 * doc/tm.texi: Regenerate.
11163 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
11164 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
11165 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
11166 * targhooks.h (default_floatn_mode): Likewise.
11167 * targhooks.c (default_floatn_mode): Likewise.
11168 * tree.c (build_common_tree_nodes): Update accordingly.
11170 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11171 Alan Hayward <alan.hayward@arm.com>
11172 David Sherwood <david.sherwood@arm.com>
11174 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
11175 (mode_iterator::iterate_p): Likewise.
11176 (mode_iterator::get_wider): Likewise.
11177 * expr.c (init_expr_target): Use opt_scalar_float_mode.
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 (opt_scalar_float_mode): New typedef.
11184 * machmode.h (float_mode_for_size): New function.
11185 * emit-rtl.c (double_mode): Delete.
11186 (init_emit_once): Use float_mode_for_size.
11187 * stor-layout.c (layout_type): Likewise.
11188 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
11190 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11191 Alan Hayward <alan.hayward@arm.com>
11192 David Sherwood <david.sherwood@arm.com>
11194 * output.h (assemble_real): Take a scalar_float_mode.
11195 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
11196 * config/arm/arm.md (consttable_4): Likewise.
11197 (consttable_8): Likewise.
11198 (consttable_16): Likewise.
11199 * config/mips/mips.md (consttable_float): Likewise.
11200 * config/s390/s390.c (s390_output_pool_entry): Likewise.
11201 * varasm.c (assemble_real): Take a scalar_float_mode.
11202 (output_constant_pool_2): Update accordingly.
11203 (output_constant): Likewise.
11205 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11206 Alan Hayward <alan.hayward@arm.com>
11207 David Sherwood <david.sherwood@arm.com>
11209 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
11210 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
11211 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
11212 (native_encode_real): Likewise.
11213 (native_interpret_real): Likewise.
11214 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
11215 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11217 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11218 Alan Hayward <alan.hayward@arm.com>
11219 David Sherwood <david.sherwood@arm.com>
11221 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
11222 <scalar_float_mode>. Simplify.
11223 (gen_extend_conv_libfunc): Likewise.
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 * coretypes.h (scalar_float_mode): New type.
11230 * machmode.h (mode_traits::from_int): Use machine_mode if
11231 USE_ENUM_MODES is defined.
11232 (is_a): New function.
11234 (dyn_cast): Likewise.
11235 (scalar_float_mode): New class.
11236 (scalar_float_mode::includes_p): New function.
11237 (is_float_mode): Likewise.
11238 * gdbhooks.py (MachineModePrinter): New class.
11239 (build_pretty_printer): Use it for scalar_float_mode.
11240 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
11241 (format_helper::format_helper): Turn into a template.
11242 * genmodes.c (get_mode_class): New function.
11243 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
11244 or machine_mode if none.
11245 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
11246 as_a <scalar_float_mode>.
11247 * dwarf2out.c (mem_loc_descriptor): Likewise.
11248 (insert_float): Likewise.
11249 (add_const_value_attribute): Likewise.
11250 * simplify-rtx.c (simplify_immed_subreg): Likewise.
11251 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
11252 (expand_unop): Update accordingly.
11253 (expand_abs_nojump): Likewise.
11254 (expand_copysign_absneg): Take a scalar_float_mode.
11255 (expand_copysign_bit): Likewise.
11256 (expand_copysign): Update accordingly.
11258 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11259 Alan Hayward <alan.hayward@arm.com>
11260 David Sherwood <david.sherwood@arm.com>
11262 * coretypes.h (opt_mode): New class.
11263 * machmode.h (opt_mode): Likewise.
11264 (opt_mode::else_void): New function.
11265 (opt_mode::require): Likewise.
11266 (opt_mode::exists): Likewise.
11267 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
11268 (GET_MODE_2XWIDER_MODE): Likewise.
11269 (mode_iterator::get_wider): Update accordingly.
11270 (mode_iterator::get_2xwider): Likewise.
11271 (mode_iterator::get_known_wider): Likewise, turning into a template.
11272 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
11273 forcing a wider mode to exist.
11274 * config/cr16/cr16.h (LONG_REG_P): Likewise.
11275 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11276 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
11277 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
11278 * lower-subreg.c (init_lower_subreg): Likewise.
11279 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
11280 on the final iteration.
11281 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
11282 a wider mode exists before asking for a move pattern.
11283 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
11284 forcing a wider mode to exist.
11285 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
11286 returning false if no such mode exists.
11287 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
11288 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
11289 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
11290 Avoid checking for a MODE_INT if we already know the mode is not a
11292 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
11293 forcing a wider mode to exist.
11294 (expmed_mult_highpart_optab): Likewise.
11295 (expmed_mult_highpart): Likewise.
11296 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
11298 * lto-streamer-in.c (lto_input_mode_table): Likewise.
11299 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
11300 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
11301 * internal-fn.c (expand_mul_overflow): Update use of
11302 GET_MODE_2XWIDER_MODE.
11303 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11304 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
11305 GET_MODE_WIDER_MODE.
11306 (convert_plusminus_to_widen): Likewise.
11307 * tree-switch-conversion.c (array_value_type): Likewise.
11308 * var-tracking.c (emit_note_insn_var_location): Likewise.
11309 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11310 Return false inside rather than outside the loop if no wider mode
11312 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
11313 and GET_MODE_2XWIDER_MODE
11314 (can_compare_p): Use else_void.
11315 * gdbhooks.py (OptMachineModePrinter): New class.
11316 (build_pretty_printer): Use it for opt_mode.
11318 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11319 Alan Hayward <alan.hayward@arm.com>
11320 David Sherwood <david.sherwood@arm.com>
11322 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
11323 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
11325 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11326 Alan Hayward <alan.hayward@arm.com>
11327 David Sherwood <david.sherwood@arm.com>
11329 * machmode.h (mode_traits): New structure.
11330 (get_narrowest_mode): New function.
11331 (mode_iterator::start): Likewise.
11332 (mode_iterator::iterate_p): Likewise.
11333 (mode_iterator::get_wider): Likewise.
11334 (mode_iterator::get_known_wider): Likewise.
11335 (mode_iterator::get_2xwider): Likewise.
11336 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
11337 (FOR_EACH_MODE): Likewise.
11338 (FOR_EACH_MODE_FROM): Likewise.
11339 (FOR_EACH_MODE_UNTIL): Likewise.
11340 (FOR_EACH_WIDER_MODE): Likewise.
11341 (FOR_EACH_2XWIDER_MODE): Likewise.
11342 * builtins.c (expand_builtin_strlen): Use new mode iterators.
11343 * combine.c (simplify_comparison): Likewise
11344 * config/i386/i386.c (type_natural_mode): Likewise.
11345 * cse.c (cse_insn): Likewise.
11346 * dse.c (find_shift_sequence): Likewise.
11347 * emit-rtl.c (init_derived_machine_modes): Likewise.
11348 (init_emit_once): Likewise.
11349 * explow.c (hard_function_value): Likewise.
11350 * expmed.c (extract_fixed_bit_field_1): Likewise.
11351 (extract_bit_field_1): Likewise.
11352 (expand_divmod): Likewise.
11353 (emit_store_flag_1): Likewise.
11354 * expr.c (init_expr_target): Likewise.
11355 (convert_move): Likewise.
11356 (alignment_for_piecewise_move): Likewise.
11357 (widest_int_mode_for_size): Likewise.
11358 (emit_block_move_via_movmem): Likewise.
11359 (copy_blkmode_to_reg): Likewise.
11360 (set_storage_via_setmem): Likewise.
11361 (compress_float_constant): Likewise.
11362 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11363 * optabs-query.c (get_best_extraction_insn): Likewise.
11364 * optabs.c (expand_binop): Likewise.
11365 (expand_twoval_unop): Likewise.
11366 (expand_twoval_binop): Likewise.
11367 (widen_leading): Likewise.
11368 (widen_bswap): Likewise.
11369 (expand_parity): Likewise.
11370 (expand_unop): Likewise.
11371 (prepare_cmp_insn): Likewise.
11372 (prepare_float_lib_cmp): Likewise.
11373 (expand_float): Likewise.
11374 (expand_fix): Likewise.
11375 (expand_sfix_optab): Likewise.
11376 * postreload.c (move2add_use_add2_insn): Likewise.
11377 * reg-stack.c (reg_to_stack): Likewise.
11378 * reginfo.c (choose_hard_reg_mode): Likewise.
11379 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11380 * stor-layout.c (mode_for_size): Likewise.
11381 (smallest_mode_for_size): Likewise.
11382 (mode_for_vector): Likewise.
11383 (finish_bitfield_representative): Likewise.
11384 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
11385 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
11386 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11387 * var-tracking.c (prepare_call_arguments): Likewise.
11389 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11390 Alan Hayward <alan.hayward@arm.com>
11391 David Sherwood <david.sherwood@arm.com>
11393 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
11394 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
11395 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
11396 * machmode.h (mode_size): Move earlier in file.
11397 (mode_precision): Likewise.
11398 (mode_inner): Likewise.
11399 (mode_nunits): Likewise.
11400 (mode_unit_size): Likewise.
11401 (unit_unit_precision): Likewise.
11402 (mode_wider): Likewise.
11403 (mode_2xwider): Likewise.
11404 (machine_mode): New class.
11405 (mode_to_bytes): New function.
11406 (mode_to_bits): Likewise.
11407 (mode_to_precision): Likewise.
11408 (mode_to_inner): Likewise.
11409 (mode_to_unit_size): Likewise.
11410 (mode_to_unit_precision): Likewise.
11411 (mode_to_nunits): Likewise.
11412 (GET_MODE_SIZE): Use mode_to_bytes.
11413 (GET_MODE_BITSIZE): Use mode_to_bits.
11414 (GET_MODE_PRECISION): Use mode_to_precision.
11415 (GET_MODE_INNER): Use mode_to_inner.
11416 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
11417 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
11418 (GET_MODE_NUNITS): Use mode_to_nunits.
11419 * system.h (ALWAYS_INLINE): New macro.
11420 * config/powerpcspe/powerpcspe-c.c
11421 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
11422 int for arg1_mode and arg2_mode.
11424 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11425 Alan Hayward <alan.hayward@arm.com>
11426 David Sherwood <david.sherwood@arm.com>
11428 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
11429 Prefix mode names with E_ in case statements.
11430 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11431 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
11432 (aarch64_split_simd_move): Likewise.
11433 (aarch64_gen_storewb_pair): Likewise.
11434 (aarch64_gen_loadwb_pair): Likewise.
11435 (aarch64_gen_store_pair): Likewise.
11436 (aarch64_gen_load_pair): Likewise.
11437 (aarch64_get_condition_code_1): Likewise.
11438 (aarch64_constant_pool_reload_icode): Likewise.
11439 (get_rsqrte_type): Likewise.
11440 (get_rsqrts_type): Likewise.
11441 (get_recpe_type): Likewise.
11442 (get_recps_type): Likewise.
11443 (aarch64_gimplify_va_arg_expr): Likewise.
11444 (aarch64_simd_container_mode): Likewise.
11445 (aarch64_emit_load_exclusive): Likewise.
11446 (aarch64_emit_store_exclusive): Likewise.
11447 (aarch64_expand_compare_and_swap): Likewise.
11448 (aarch64_gen_atomic_cas): Likewise.
11449 (aarch64_emit_bic): Likewise.
11450 (aarch64_emit_atomic_swap): Likewise.
11451 (aarch64_emit_atomic_load_op): Likewise.
11452 (aarch64_evpc_trn): Likewise.
11453 (aarch64_evpc_uzp): Likewise.
11454 (aarch64_evpc_zip): Likewise.
11455 (aarch64_evpc_ext): Likewise.
11456 (aarch64_evpc_rev): Likewise.
11457 (aarch64_evpc_dup): Likewise.
11458 (aarch64_gen_ccmp_first): Likewise.
11459 (aarch64_gen_ccmp_next): Likewise.
11460 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
11461 (alpha_emit_xfloating_libcall): Likewise.
11462 (emit_insxl): Likewise.
11463 (alpha_arg_type): Likewise.
11464 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
11465 (arc_preferred_simd_mode): Likewise.
11466 (arc_secondary_reload): Likewise.
11467 (get_arc_condition_code): Likewise.
11468 (arc_print_operand): Likewise.
11469 (arc_legitimate_constant_p): Likewise.
11470 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11471 * config/arc/arc.md (casesi_load): Likewise.
11472 (casesi_compact_jump): Likewise.
11473 * config/arc/predicates.md (proper_comparison_operator): Likewise.
11474 (cc_use_register): Likewise.
11475 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11476 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
11477 (arm_init_iwmmxt_builtins): Likewise.
11478 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
11479 (neon_expand_vector_init): Likewise.
11480 (arm_attr_length_move_neon): Likewise.
11481 (maybe_get_arm_condition_code): Likewise.
11482 (arm_emit_vector_const): Likewise.
11483 (arm_preferred_simd_mode): Likewise.
11484 (arm_output_iwmmxt_tinsr): Likewise.
11485 (thumb1_output_casesi): Likewise.
11486 (thumb2_output_casesi): Likewise.
11487 (arm_emit_load_exclusive): Likewise.
11488 (arm_emit_store_exclusive): Likewise.
11489 (arm_expand_compare_and_swap): Likewise.
11490 (arm_evpc_neon_vuzp): Likewise.
11491 (arm_evpc_neon_vzip): Likewise.
11492 (arm_evpc_neon_vrev): Likewise.
11493 (arm_evpc_neon_vtrn): Likewise.
11494 (arm_evpc_neon_vext): Likewise.
11495 (arm_validize_comparison): Likewise.
11496 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
11497 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
11498 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
11499 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
11500 (c6x_preferred_simd_mode): Likewise.
11501 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
11502 (epiphany_rtx_costs): Likewise.
11503 * config/epiphany/predicates.md (proper_comparison_operator):
11505 * config/frv/frv.c (condexec_memory_operand): Likewise.
11506 (frv_emit_move): Likewise.
11507 (output_move_single): Likewise.
11508 (output_condmove_single): Likewise.
11509 (frv_hard_regno_mode_ok): Likewise.
11510 (frv_matching_accg_mode): Likewise.
11511 * config/h8300/h8300.c (split_adds_subs): Likewise.
11512 (h8300_rtx_costs): Likewise.
11513 (h8300_print_operand): Likewise.
11514 (compute_mov_length): Likewise.
11515 (output_logical_op): Likewise.
11516 (compute_logical_op_length): Likewise.
11517 (compute_logical_op_cc): Likewise.
11518 (h8300_shift_needs_scratch_p): Likewise.
11519 (output_a_shift): Likewise.
11520 (compute_a_shift_length): Likewise.
11521 (compute_a_shift_cc): Likewise.
11522 (expand_a_rotate): Likewise.
11523 (output_a_rotate): Likewise.
11524 * config/i386/i386.c (classify_argument): Likewise.
11525 (function_arg_advance_32): Likewise.
11526 (function_arg_32): Likewise.
11527 (function_arg_64): Likewise.
11528 (function_value_64): Likewise.
11529 (ix86_gimplify_va_arg): Likewise.
11530 (ix86_legitimate_constant_p): Likewise.
11531 (put_condition_code): Likewise.
11532 (split_double_mode): Likewise.
11533 (ix86_avx256_split_vector_move_misalign): Likewise.
11534 (ix86_expand_vector_logical_operator): Likewise.
11535 (ix86_split_idivmod): Likewise.
11536 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
11537 (ix86_build_const_vector): Likewise.
11538 (ix86_build_signbit_mask): Likewise.
11539 (ix86_match_ccmode): Likewise.
11540 (ix86_cc_modes_compatible): Likewise.
11541 (ix86_expand_branch): Likewise.
11542 (ix86_expand_sse_cmp): Likewise.
11543 (ix86_expand_sse_movcc): Likewise.
11544 (ix86_expand_int_sse_cmp): Likewise.
11545 (ix86_expand_vec_perm_vpermi2): Likewise.
11546 (ix86_expand_vec_perm): Likewise.
11547 (ix86_expand_sse_unpack): Likewise.
11548 (ix86_expand_int_addcc): Likewise.
11549 (ix86_split_to_parts): Likewise.
11550 (ix86_vectorize_builtin_gather): Likewise.
11551 (ix86_vectorize_builtin_scatter): Likewise.
11552 (avx_vpermilp_parallel): Likewise.
11553 (inline_memory_move_cost): Likewise.
11554 (ix86_tieable_integer_mode_p): Likewise.
11555 (x86_maybe_negate_const_int): Likewise.
11556 (ix86_expand_vector_init_duplicate): Likewise.
11557 (ix86_expand_vector_init_one_nonzero): Likewise.
11558 (ix86_expand_vector_init_one_var): Likewise.
11559 (ix86_expand_vector_init_concat): Likewise.
11560 (ix86_expand_vector_init_interleave): Likewise.
11561 (ix86_expand_vector_init_general): Likewise.
11562 (ix86_expand_vector_set): Likewise.
11563 (ix86_expand_vector_extract): Likewise.
11564 (emit_reduc_half): Likewise.
11565 (ix86_emit_i387_round): Likewise.
11566 (ix86_mangle_type): Likewise.
11567 (ix86_expand_round_sse4): Likewise.
11568 (expand_vec_perm_blend): Likewise.
11569 (canonicalize_vector_int_perm): Likewise.
11570 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
11571 (expand_vec_perm_1): Likewise.
11572 (expand_vec_perm_interleave3): Likewise.
11573 (expand_vec_perm_even_odd_pack): Likewise.
11574 (expand_vec_perm_even_odd_1): Likewise.
11575 (expand_vec_perm_broadcast_1): Likewise.
11576 (ix86_vectorize_vec_perm_const_ok): Likewise.
11577 (ix86_expand_vecop_qihi): Likewise.
11578 (ix86_expand_mul_widen_hilo): Likewise.
11579 (ix86_expand_sse2_abs): Likewise.
11580 (ix86_expand_pextr): Likewise.
11581 (ix86_expand_pinsr): Likewise.
11582 (ix86_preferred_simd_mode): Likewise.
11583 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
11584 * config/i386/sse.md (*andnot<mode>3): Likewise.
11585 (<mask_codefor><code><mode>3<mask_name>): Likewise.
11586 (*<code><mode>3): Likewise.
11587 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
11588 (ia64_expand_atomic_op): Likewise.
11589 (ia64_arg_type): Likewise.
11590 (ia64_mode_to_int): Likewise.
11591 (ia64_scalar_mode_supported_p): Likewise.
11592 (ia64_vector_mode_supported_p): Likewise.
11593 (expand_vec_perm_broadcast): Likewise.
11594 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
11595 (iq2000_function_arg_advance): Likewise.
11596 (iq2000_function_arg): Likewise.
11597 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
11598 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
11599 (m68k_libcall_value): Likewise.
11600 (m68k_function_value): Likewise.
11601 (sched_attr_op_type): Likewise.
11602 * config/mcore/mcore.c (mcore_output_move): Likewise.
11603 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
11605 (microblaze_function_arg): Likewise.
11606 * config/mips/mips.c (mips16_build_call_stub): Likewise.
11607 (mips_print_operand): Likewise.
11608 (mips_mode_ok_for_mov_fmt_p): Likewise.
11609 (mips_vector_mode_supported_p): Likewise.
11610 (mips_preferred_simd_mode): Likewise.
11611 (mips_expand_vpc_loongson_even_odd): Likewise.
11612 (mips_expand_vec_unpack): Likewise.
11613 (mips_expand_vi_broadcast): Likewise.
11614 (mips_expand_vector_init): Likewise.
11615 (mips_expand_vec_reduc): Likewise.
11616 (mips_expand_msa_cmp): Likewise.
11617 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
11618 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
11619 (cc_flags_for_mode): Likewise.
11620 * config/msp430/msp430.c (msp430_print_operand): Likewise.
11621 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
11622 (nds32_output_casesi_pc_relative): Likewise.
11623 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11624 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
11625 (nvptx_gen_unpack): Likewise.
11626 (nvptx_gen_pack): Likewise.
11627 (nvptx_gen_shuffle): Likewise.
11628 (nvptx_gen_wcast): Likewise.
11629 (nvptx_preferred_simd_mode): Likewise.
11630 * config/pa/pa.c (pa_secondary_reload): Likewise.
11631 * config/pa/predicates.md (base14_operand): Likewise.
11632 * config/powerpcspe/powerpcspe-c.c
11633 (altivec_resolve_overloaded_builtin): Likewise.
11634 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
11636 (rs6000_preferred_simd_mode): Likewise.
11637 (output_vec_const_move): Likewise.
11638 (rs6000_expand_vector_extract): Likewise.
11639 (rs6000_split_vec_extract_var): Likewise.
11640 (reg_offset_addressing_ok_p): Likewise.
11641 (rs6000_legitimate_offset_address_p): Likewise.
11642 (rs6000_legitimize_address): Likewise.
11643 (rs6000_emit_set_const): Likewise.
11644 (rs6000_const_vec): Likewise.
11645 (rs6000_emit_move): Likewise.
11646 (spe_build_register_parallel): Likewise.
11647 (rs6000_darwin64_record_arg_recurse): Likewise.
11648 (swap_selector_for_mode): Likewise.
11649 (spe_init_builtins): Likewise.
11650 (paired_init_builtins): Likewise.
11651 (altivec_init_builtins): Likewise.
11652 (do_load_for_compare): Likewise.
11653 (rs6000_generate_compare): Likewise.
11654 (rs6000_expand_float128_convert): Likewise.
11655 (emit_load_locked): Likewise.
11656 (emit_store_conditional): Likewise.
11657 (rs6000_output_function_epilogue): Likewise.
11658 (rs6000_handle_altivec_attribute): Likewise.
11659 (rs6000_function_value): Likewise.
11660 (emit_fusion_gpr_load): Likewise.
11661 (emit_fusion_p9_load): Likewise.
11662 (emit_fusion_p9_store): Likewise.
11663 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
11664 (fusion_gpr_mem_load): Likewise.
11665 (fusion_addis_mem_combo_load): Likewise.
11666 (fusion_addis_mem_combo_store): Likewise.
11667 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
11668 (fusion_gpr_mem_load): Likewise.
11669 (fusion_addis_mem_combo_load): Likewise.
11670 (fusion_addis_mem_combo_store): Likewise.
11671 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11673 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
11674 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
11675 (rs6000_preferred_simd_mode): Likewise.
11676 (output_vec_const_move): Likewise.
11677 (rs6000_expand_vector_extract): Likewise.
11678 (rs6000_split_vec_extract_var): Likewise.
11679 (reg_offset_addressing_ok_p): Likewise.
11680 (rs6000_legitimate_offset_address_p): Likewise.
11681 (rs6000_legitimize_address): Likewise.
11682 (rs6000_emit_set_const): Likewise.
11683 (rs6000_const_vec): Likewise.
11684 (rs6000_emit_move): Likewise.
11685 (rs6000_darwin64_record_arg_recurse): Likewise.
11686 (swap_selector_for_mode): Likewise.
11687 (paired_init_builtins): Likewise.
11688 (altivec_init_builtins): Likewise.
11689 (rs6000_expand_float128_convert): Likewise.
11690 (emit_load_locked): Likewise.
11691 (emit_store_conditional): Likewise.
11692 (rs6000_output_function_epilogue): Likewise.
11693 (rs6000_handle_altivec_attribute): Likewise.
11694 (rs6000_function_value): Likewise.
11695 (emit_fusion_gpr_load): Likewise.
11696 (emit_fusion_p9_load): Likewise.
11697 (emit_fusion_p9_store): Likewise.
11698 * config/rx/rx.c (rx_gen_move_template): Likewise.
11699 (flags_from_mode): Likewise.
11700 * config/s390/predicates.md (s390_alc_comparison): Likewise.
11701 (s390_slb_comparison): Likewise.
11702 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
11703 (s390_vector_mode_supported_p): Likewise.
11704 (s390_cc_modes_compatible): Likewise.
11705 (s390_match_ccmode_set): Likewise.
11706 (s390_canonicalize_comparison): Likewise.
11707 (s390_emit_compare_and_swap): Likewise.
11708 (s390_branch_condition_mask): Likewise.
11709 (s390_rtx_costs): Likewise.
11710 (s390_secondary_reload): Likewise.
11711 (__SECONDARY_RELOAD_CASE): Likewise.
11712 (s390_expand_cs): Likewise.
11713 (s390_preferred_simd_mode): Likewise.
11714 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
11715 * config/sh/sh.c (sh_print_operand): Likewise.
11716 (dump_table): Likewise.
11717 (sh_secondary_reload): Likewise.
11718 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11719 * config/sh/sh.md (casesi_worker_1): Likewise.
11720 (casesi_worker_2): Likewise.
11721 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
11722 (fcc_comparison_operator): Likewise.
11723 * config/sparc/sparc.c (sparc_expand_move): Likewise.
11724 (emit_soft_tfmode_cvt): Likewise.
11725 (sparc_preferred_simd_mode): Likewise.
11726 (output_cbranch): Likewise.
11727 (sparc_print_operand): Likewise.
11728 (sparc_expand_vec_perm_bmask): Likewise.
11729 (vector_init_bshuffle): Likewise.
11730 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
11731 (spu_vector_mode_supported_p): Likewise.
11732 (spu_expand_insv): Likewise.
11733 (spu_emit_branch_or_set): Likewise.
11734 (spu_handle_vector_attribute): Likewise.
11735 (spu_builtin_splats): Likewise.
11736 (spu_builtin_extract): Likewise.
11737 (spu_builtin_promote): Likewise.
11738 (spu_expand_sign_extend): Likewise.
11739 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
11740 (tilegx_simd_int): Likewise.
11741 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
11742 (tilepro_simd_int): Likewise.
11743 * config/v850/v850.c (const_double_split): Likewise.
11744 (v850_print_operand): Likewise.
11745 (ep_memory_offset): Likewise.
11746 * config/vax/vax.c (vax_rtx_costs): Likewise.
11747 (vax_output_int_move): Likewise.
11748 (vax_output_int_add): Likewise.
11749 (vax_output_int_subtract): Likewise.
11750 * config/visium/predicates.md (visium_branch_operator): Likewise.
11751 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
11752 (visium_print_operand_address): Likewise.
11753 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11754 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
11755 (xtensa_expand_conditional_branch): Likewise.
11756 (xtensa_copy_incoming_a7): Likewise.
11757 (xtensa_output_literal): Likewise.
11758 * dfp.c (decimal_real_maxval): Likewise.
11759 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
11761 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11762 Alan Hayward <alan.hayward@arm.com>
11763 David Sherwood <david.sherwood@arm.com>
11765 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
11766 (mode_nunits_inline): Likewise.
11767 (mode_inner_inline): Likewise.
11768 (mode_unit_size_inline): Likewise.
11769 (mode_unit_precision_inline): Likewise.
11770 (emit_insn_modes_h): Likewise. Also emit a #define of the
11772 (emit_mode_wider): Add an E_ prefix to mode names.
11773 (emit_mode_complex): Likewise.
11774 (emit_mode_inner): Likewise.
11775 (emit_mode_adjustments): Likewise.
11776 (emit_mode_int_n): Likewise.
11777 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
11778 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
11779 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
11780 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
11781 (CRC32_BUILTIN, ENTRY): Likewise.
11782 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
11783 (aarch64_pop_regs): Likewise.
11784 (aarch64_process_components): Likewise.
11785 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
11786 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
11787 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
11788 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
11789 * config/arm/arm.c (arm_init_libfuncs): Likewise.
11790 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
11792 * config/i386/i386-builtin.def (pcmpestr): Likewise.
11793 (pcmpistr): Likewise.
11794 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
11795 * config/mmix/mmix.c (mmix_output_condition): Likewise.
11796 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
11798 * config/rl78/rl78.c (mduc_regs): Likewise.
11799 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
11800 (htm_expand_builtin): Likewise.
11801 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
11802 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
11803 * config/xtensa/xtensa.c (print_operand): Likewise.
11804 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
11805 (NUM_MODE_VECTOR_INT): Likewise.
11806 * genoutput.c (null_operand): Likewise.
11807 (output_operand_data): Likewise.
11808 * genrecog.c (print_parameter_value): Likewise.
11809 * lra.c (debug_operand_data): Likewise.
11811 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11813 * dwarf2out.c (scompare_loc_descriptor_wide)
11814 (scompare_loc_descriptor_narrow): New functions, split out from...
11815 (scompare_loc_descriptor): ...here.
11816 * expmed.c (emit_store_flag_int): New function, split out from...
11817 (emit_store_flag): ...here.
11819 2017-08-30 Richard Biener <rguenther@suse.de>
11821 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
11822 (dwarf2out_early_finish): Move setting of AT_pubnames from
11823 early debug output to early finish.
11825 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11827 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
11828 and -mdata-region to the assembler.
11830 2017-08-30 Richard Biener <rguenther@suse.de>
11832 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
11834 (gen_subprogram_die): Add DW_AT_object_pointer only early.
11835 (dwarf2out_early_global_decl): Only generate a DIE for the
11836 abstract origin if it doesn't already exist or is a declaration DIE.
11837 (resolve_addr): Do not add the linkage name twice when
11838 generating a stub DIE for the DW_TAG_GNU_call_site target.
11840 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11842 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11843 Use machine_mode rather than int for arg1_mode.
11845 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
11848 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
11849 that the second argument of the built-in functions to unpack
11850 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
11851 switch statement instead a lot of if statements.
11852 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
11853 Allow 64-bit values to be in Altivec registers as well as
11854 traditional floating point registers.
11855 (pack<mode>, FMOVE128_VSX iterator): Likewise.
11857 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
11859 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
11860 MAX_REGS_PER_ADDRESS == 1.
11862 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
11864 * config/i386/i386.opt (flag_fentry): Do not init to -1.
11865 * config/i386/i386.c (ix86_option_override_internal): Simplify
11866 setting of opts->x_flag_entry.
11868 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11869 Jakub Jelinek <jakub@redhat.com>
11870 Richard Biener <rguenther@suse.de>
11872 PR tree-optimization/81503
11873 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
11874 folded constant fits in the target type; reorder tests for clarity.
11876 2017-08-29 Martin Liska <mliska@suse.cz>
11878 * passes.def: Include pass_lower_switch.
11879 * stmt.c (dump_case_nodes): Remove and move to
11880 tree-switch-conversion.
11881 (case_values_threshold): Likewise.
11882 (expand_switch_as_decision_tree_p): Likewise.
11883 (emit_case_decision_tree): Likewise.
11884 (expand_case): Likewise.
11885 (balance_case_nodes): Likewise.
11886 (node_has_low_bound): Likewise.
11887 (node_has_high_bound): Likewise.
11888 (node_is_bounded): Likewise.
11889 (emit_case_nodes): Likewise.
11890 (struct simple_case_node): New struct.
11891 (add_case_node): Remove.
11892 (emit_case_dispatch_table): Use vector instead of case_list.
11893 (reset_out_edges_aux): Remove.
11894 (compute_cases_per_edge): Likewise.
11895 (expand_case): Build list of simple_case_node.
11896 (expand_sjlj_dispatch_table): Use it.
11897 * tree-switch-conversion.c (struct case_node): Moved from
11898 stmt.c and adjusted.
11899 (emit_case_nodes): Likewise.
11900 (node_has_low_bound): Likewise.
11901 (node_has_high_bound): Likewise.
11902 (node_is_bounded): Likewise.
11903 (case_values_threshold): Likewise.
11904 (reset_out_edges_aux): Likewise.
11905 (compute_cases_per_edge): Likewise.
11906 (add_case_node): Likewise.
11907 (dump_case_nodes): Likewise.
11908 (balance_case_nodes): Likewise.
11909 (expand_switch_as_decision_tree_p): Likewise.
11910 (emit_jump): Likewise.
11911 (emit_case_decision_tree): Likewise.
11912 (try_switch_expansion): Likewise.
11913 (do_jump_if_equal): Likewise.
11914 (emit_cmp_and_jump_insns): Likewise.
11915 (fix_phi_operands_for_edge): New function.
11916 (record_phi_operand_mapping): Likewise.
11917 (class pass_lower_switch): New pass.
11918 (pass_lower_switch::execute): New function.
11919 (make_pass_lower_switch): Likewise.
11920 (conditional_probability):
11921 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
11922 * tree-pass.h: Add make_pass_lower_switch.
11924 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11927 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
11930 2017-08-29 Richard Biener <rguenther@suse.de>
11932 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
11933 we do not add a DW_AT_inline attribute twice.
11934 (gen_subprogram_die): Remove code setting DW_AT_inline on
11935 DECL_ABSTRACT_P nodes.
11937 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
11939 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
11940 calls to internal functions.
11941 (gimplify_modify_expr): Likewise.
11942 * tree-call-cdce.c (use_internal_fn): Likewise.
11943 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
11944 (convert_to_divmod): Set the nothrow flag.
11945 * tree-if-conv.c (predicate_mem_writes): Likewise.
11946 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
11947 (vectorizable_call): Likewise.
11948 (vectorizable_store): Likewise.
11949 (vectorizable_load): Likewise.
11950 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
11951 (vect_recog_mask_conversion_pattern): Likewise.
11953 2017-08-29 Martin Liska <mliska@suse.cz>
11956 * gcc.c (driver_handle_option): Add new argument.
11957 * opts-common.c (handle_option): Pass
11958 target_option_override_hook.
11959 * opts-global.c (lang_handle_option): Add new option.
11960 (set_default_handlers): Add new argument.
11961 (decode_options): Likewise.
11962 * opts.c (target_handle_option): Likewise.
11963 (common_handle_option): Call target_option_override_hook.
11964 * opts.h (struct cl_option_handler_func): Add hook for
11965 target option override.
11966 (struct cl_option_handlers): Likewise.
11967 (set_default_handlers): Add new argument.
11968 (decode_options): Likewise.
11969 (common_handle_option): Likewise.
11970 (target_handle_option): Likewise.
11971 * toplev.c (toplev::main): Pass targetm.target_option.override
11974 2017-08-29 Richard Biener <rguenther@suse.de>
11975 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
11977 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
11978 life to the active subtree.
11980 2017-08-28 Jeff Law <law@redhat.com>
11982 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
11983 derive_equivalences.
11984 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
11985 Code moved into....
11986 (edge_info::derive_equivalences): New private member function
11988 * tree-ssa-dom.c (class edge_info): Changed from a struct
11989 to a class. Add ctor/dtor, methods and data members.
11990 (edge_info::edge_info): Renamed from allocate_edge_info.
11991 Initialize additional members.
11992 (edge_info::~edge_info): New.
11993 (free_dom_edge_info): Delete the edge info.
11994 (record_edge_info): Use new class & associated member functions.
11995 Tighten forms for testing for edge equivalences.
11996 (record_temporary_equivalences): Iterate over the simple
11997 equivalences rather than assuming there's only one per edge.
11998 (cprop_into_successor_phis): Iterate over the simple
11999 equivalences rather than assuming there's only one per edge.
12000 (optimize_stmt): Use operand_equal_p rather than pointer
12001 equality for mini-DSE code.
12003 2017-08-28 Nathan Sidwell <nathan@acm.org>
12005 * gcc.c (execute): Fold SIGPIPE handling into switch
12006 statement. Adjust internal error message.
12008 2017-08-28 Richard Biener <rguenther@suse.de>
12011 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
12012 Do nothing for removed DIEs.
12014 2017-08-28 Richard Biener <rguenther@suse.de>
12016 PR tree-optimization/81977
12017 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
12020 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
12023 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
12024 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
12025 using targetm.gen_mem_thread_fence.
12027 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
12030 * config/i386/i386.md (*<btsc><mode>): Change operand 2
12031 predicate to register_operand. Reorder operands.
12032 (*btr<mode>): Ditto.
12033 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
12034 (*btr<mode>_mask): Ditto.
12036 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
12038 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
12039 * config/rs6000/xmmintrin.h: New file.
12040 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
12042 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12045 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
12046 parameter and_insn and return it.
12047 (recombine_lvx_pattern): Insert a copy to ensure availability of
12048 the base register of the copied masking operation at the point of
12049 the instruction replacement.
12050 (recombine_stvx_pattern): Likewise.
12052 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
12054 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
12055 undocumented switches.
12056 (-mpower9-dform-vector): Likewise.
12057 (-mpower9-dform): Likewise.
12058 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
12059 comments to delete references to -mpower9-dform* switches.
12060 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
12061 Delete reference to -mpower9-dform* switches, test for
12062 -mpower9-vector instead.
12063 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
12064 (OTHER_P9_VECTOR_MASKS): Likewise.
12065 (POWERPC_MASKS): Likewise.
12066 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
12067 tests against -mpower9-dform* to -mpower9-vector. Delete code
12068 that checked for -mpower9-dform* consistancy with other options.
12069 Add test for -mpower9-misc to enable other power9 switches.
12070 (rs6000_init_hard_regno_mode_ok): Likewise.
12071 (rs6000_option_override_internal): Likewise.
12072 (rs6000_emit_prologue): Likewise.
12073 (rs6000_emit_epilogue): Likewise.
12074 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
12075 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
12076 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
12078 (emit_fusion_p9_store): Likewise.
12079 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
12080 resetting these macros if the assembler does not support ISA 3.0
12082 (TARGET_P9_DFORM_VECTOR): Likewise.
12083 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
12084 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
12086 2017-08-25 Alan Modra <amodra@gmail.com>
12088 PR rtl-optimization/81747
12089 * cse.c (cse_extended_basic_block): Don't attempt to record
12090 equivalences for degenerate conditional branches that branch
12091 to their fall-through.
12093 2017-08-24 Martin Sebor <msebor@redhat.com>
12095 PR middle-end/81908
12096 * gimple-fold.c (size_must_be_zero_p): New function.
12097 (gimple_fold_builtin_memory_op): Call it.
12099 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
12101 * config/rs6000/mm_malloc.h: New file.
12103 2017-08-24 Bin Cheng <bin.cheng@arm.com>
12105 PR tree-optimization/81913
12106 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
12107 analysis when either IVs in condition can wrap.
12109 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
12111 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
12112 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
12114 2017-08-24 Richard Biener <rguenther@suse.de>
12117 * targhooks.c (default_target_can_inline_p): Properly
12118 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
12119 is present and always compare.
12120 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
12121 infer -mfpmath=sse from TARGET_SSE_P.
12122 (ix86_can_inline_p): Properly use target_option_default_node when
12123 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
12125 2017-08-24 Richard Biener <rguenther@suse.de>
12128 * dwarf2out.c (output_die): Handle flag_generate_offload like
12130 (output_comp_unit): Likewise.
12131 (gen_array_type_die): Likewise.
12132 (dwarf2out_early_finish): Likewise.
12133 (note_variable_value_in_expr): Likewise.
12134 (dwarf2out_finish): Likewise. Adjust assert.
12135 * cgraphunit.c (symbol_table::compile): Move setting of
12136 flag_generate_offload earlier ...
12137 (symbol_table::finalize_compilation_unit): ... here, before
12138 early debug finalization.
12140 2017-08-24 Richard Biener <rguenther@suse.de>
12142 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
12143 and ipa-fnsummary.h.
12144 (ix86_can_inline_p): When ix86_fpmath flags do not match
12145 check whether the callee uses FP math at all.
12147 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
12149 PR middle-end/81931
12150 * tree-ssanames.c (get_nonzero_bits): Use element_precision
12151 instead of TYPE_PRECISION.
12153 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
12154 Alan Hayward <alan.hayward@arm.com>
12155 David Sherwood <david.sherwood@arm.com>
12157 * combine.c (make_extraction): Use subreg_offset_from_lsb.
12159 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
12161 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
12163 (ix86_frame::stack_realign_allocate): New field.
12164 (struct machine_frame_state): Modify comments.
12165 (machine_frame_state::sp_realigned_fp_end): New field.
12166 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
12167 layout calculation.
12168 (sp_valid_at): Add assertion to assure no attempt to access invalid
12169 offset of a realigned stack.
12170 (fp_valid_at): Likewise.
12171 (choose_baseaddr): Modify comments.
12172 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
12173 ix86_expand_prologue.
12174 (ix86_expand_prologue): Modify stack realignment and allocation.
12175 (ix86_expand_epilogue): Modify comments.
12176 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
12177 avx2_runtime, avx512f, and avx512f_runtime.
12179 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
12181 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
12182 (mstackrealign): Do not init to -1.
12183 * config/i386/i386.c (ix86_option_override_internal):
12184 Check opts_set, not opts when setting default value of
12185 opts->x_ix86_force_align_arg_pointer.
12187 2017-08-23 Richard Biener <rguenther@suse.de>
12189 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
12190 lang_hooks.decl_printable_name.
12191 * print-rtl-function.c (print_rtx_function): Likewise.
12192 * tree-pretty-print.c (dump_function_header): Likewise.
12194 2017-08-23 Richard Biener <rguenther@suse.de>
12197 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
12198 -g0 at compile-time.
12200 2017-08-23 Tamar Christina <tamar.christina@arm.com>
12202 PR middle-end/19706
12203 * doc/sourcebuild.texi (Other hardware attributes):
12206 2017-08-23 Tamar Christina <tamar.christina@arm.com>
12208 PR middle-end/19706
12209 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
12210 Fix single-use check.
12212 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
12214 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
12216 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
12218 * doc/install.texi: Modify to add more details on running selected
12221 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
12223 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
12224 is combined with -mabi=ms.
12225 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
12228 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12230 PR tree-optimization/81488
12231 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
12232 and cached_basis fields.
12233 (MAX_SPREAD): New constant.
12234 (alloc_cand_and_find_basis): Initialize new fields.
12235 (clear_visited): New function.
12236 (create_phi_basis_1): Rename from create_phi_basis, set visited
12237 and cached_basis fields.
12238 (create_phi_basis): New wrapper function.
12239 (phi_add_costs_1): Rename from phi_add_costs, add spread
12240 parameter, set visited field, short-circuit when limits reached.
12241 (phi_add_costs): New wrapper function.
12242 (record_phi_increments_1): Rename from record_phi_increments, set
12244 (record_phi_increments): New wrapper function.
12245 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
12246 (phi_incr_cost): New wrapper function.
12247 (all_phi_incrs_profitable_1): Rename from
12248 all_phi_incrs_profitable, set visited field.
12249 (all_phi_incrs_profitable): New wrapper function.
12251 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
12252 Alan Hayward <alan.hayward@arm.com>
12253 David Sherwood <david.sherwood@arm.com>
12255 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
12256 that takes the outer and inner modes.
12257 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
12258 comparison as the canonical test for a paradoxical subreg.
12259 * combine.c (simplify_set): Use paradoxical_subreg_p.
12260 (make_extraction): Likewise.
12261 (force_to_mode): Likewise.
12262 (rtx_equal_for_field_assignment_p): Likewise.
12263 (gen_lowpart_for_combine): Likewise.
12264 (simplify_comparison): Likewise.
12265 * cse.c (equiv_constant): Likewise.
12266 * expmed.c (store_bit_field_1): Likewise.
12267 * final.c (alter_subreg): Likewise.
12268 * fwprop.c (propagate_rtx): Likewise.
12269 (forward_propagate_subreg): Likewise.
12270 * ira-conflicts.c (ira_build_conflicts): Likewise.
12271 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
12272 * lra-constraints.c (curr_insn_transform): Likewise.
12273 (split_reg): Likewise.
12274 * lra-eliminations.c (move_plus_up): Likewise.
12275 (lra_eliminate_regs_1): Likewise.
12276 * recog.c (general_operand): Likewise.
12277 * ree.c (combine_reaching_defs): Likewise.
12278 * reload.c (push_reload): Likewise.
12279 (find_reloads): Likewise.
12280 * reload1.c (elimination_effects): Likewise.
12281 (compute_reload_subreg_offset): Likewise.
12282 (choose_reload_regs): Likewise.
12283 * rtlanal.c (subreg_lsb_1): Likewise.
12284 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12285 (simplify_subreg): Likewise.
12286 * var-tracking.c (track_loc_p): Likewise.
12287 * emit-rtl.c (byte_lowpart_offset): Likewise.
12288 (paradoxical_subreg_p): Delete out-of-line definition.
12290 2017-08-22 Jeff Law <law@redhat.com>
12292 PR tree-optimization/81741
12293 PR tree-optimization/71947
12294 * tree-ssa-dom.c: Include tree-inline.h.
12295 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
12296 equivalences if one is more expensive to compute than the other.
12297 * tree-ssa-scopedtables.h (class const_or_copies): Make
12298 record_const_or_copy_raw method private.
12299 (class avail_exprs_stack): New method simplify_binary_operation.
12300 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
12301 avail_exprs_stack::simplify_binary_operation as needed.
12302 (avail_exprs_stack::simplify_binary_operation): New function.
12304 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
12306 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
12307 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
12308 (DOT_SYMBOLS): Likewise.
12309 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
12310 (RELOCATABLE_NEEDS_FIXUP): Likewise.
12311 (RS6000_ABI_NAME): Likewise.
12312 (TARGET_CMODEL): Likewise.
12313 (TOC_SECTION_ASM_OP): Likewise.
12314 (SET_CMODEL): New macro.
12315 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
12317 2017-08-22 Richard Biener <rguenther@suse.de>
12319 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
12320 to allow for free-lang-data replacements similar to verify_type_variant.
12322 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
12323 Alan Hayward <alan.hayward@arm.com>
12324 David Sherwood <david.sherwood@arm.com>
12326 * config/aarch64/aarch64.md (casesi): Use DImode rather than
12327 VOIDmode for the LABEL_REF.
12329 2017-08-22 Richard Biener <rguenther@suse.de>
12331 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
12333 2017-08-22 Richard Biener <rguenther@suse.de>
12335 * common.opt (feliminate-dwarf2-dups): Ignore.
12336 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
12337 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
12338 same_die_p_wrap, compute_section_prefix,
12339 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
12340 (comdat_symbol_id, comdat_symbol_number): Likewise.
12341 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
12343 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
12344 (output_die): Mark unreachable path unreachable.
12345 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
12346 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
12347 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
12348 (dwarf2out_early_finish): Likewise.
12350 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
12352 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
12354 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
12357 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
12358 not VAR_P. Filter attribute warnings with OPT_Wattributes.
12359 (avr_attribute_table) <io, io_low, address>: Initialize
12360 .decl_required with true.
12362 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
12364 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
12365 undocumented debugging options.
12366 (-mvsx-scalar-double): Likewise.
12367 (-mallow-df-permute): Likewise.
12368 (-mvectorize-builtins): Likewise.
12369 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12370 (rs6000_builtin_vectorized_function): Likewise.
12371 (rs6000_builtin_md_vectorized_function): Likewise.
12372 (rs6000_opt_vars): Likewise.
12374 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
12377 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
12378 (*btrq_imm): Rename from *btrq.
12379 (*btcq_imm): Rename from *btcq.
12380 (btsc): New code attribute.
12381 (*<btsc><mode>): New insn pattern.
12382 (*btr<mode>): Ditto.
12383 (*<btsc><mode>_mask): New insn_and_split pattern.
12384 (*btr<mode>_mask): Ditto.
12386 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12387 Alan Hayward <alan.hayward@arm.com>
12388 David Sherwood <david.sherwood@arm.com>
12390 * function.c (pad_below): Simplify padding calculation.
12392 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12393 Alan Hayward <alan.hayward@arm.com>
12394 David Sherwood <david.sherwood@arm.com>
12396 * target.def (function_prologue): Remove frame size argument.
12397 (function_epilogue): Likewise.
12398 * doc/tm.texi: Regenerate.
12399 * final.c (final_start_function): Update call to function_prologue.
12400 (final_end_function): Update call to function_epilogue.
12401 (default_function_pro_epilogue): Remove frame size argument.
12402 * output.h (default_function_pro_epilogue): Likewise.
12403 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
12404 (arm_output_function_prologue): Likewise.
12405 * config/frv/frv.c (frv_function_prologue): Likewise.
12406 (frv_function_epilogue): Likewise.
12407 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
12408 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
12409 (ia64_output_function_epilogue): Likewise.
12410 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
12411 (m32r_output_function_epilogue): Likewise.
12412 * config/microblaze/microblaze.c (microblaze_function_prologue)
12413 (microblaze_function_epilogue): Likewise.
12414 * config/mips/mips.c (mips_output_function_prologue): Likewise.
12415 (mips_output_function_epilogue): Likewise.
12416 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
12417 (mmix_target_asm_function_epilogue): Likewise.
12418 * config/msp430/msp430.c (msp430_start_function): Likewise.
12419 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
12420 (nds32_asm_function_epilogue): Likewise.
12421 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
12422 * config/pa/pa.c (pa_output_function_prologue): Likewise.
12423 (pa_output_function_epilogue): Likewise.
12424 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
12425 (rs6000_output_function_epilogue): Likewise.
12426 * config/rl78/rl78.c (rl78_start_function): Likewise.
12427 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
12428 (rs6000_output_function_epilogue): Likewise.
12429 * config/rx/rx.c (rx_output_function_prologue): Likewise.
12430 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
12431 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
12432 (sparc_asm_function_epilogue): Likewise.
12434 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12436 * tree.h (type_has_mode_precision_p): New function.
12437 * convert.c (convert_to_integer_1): Use it.
12438 * expr.c (expand_expr_real_2): Likewise.
12439 (expand_expr_real_1): Likewise.
12440 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
12441 * match.pd: Likewise.
12442 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
12443 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
12444 * tree-tailcall.c (process_assignment): Likewise.
12445 * tree-vect-loop.c (vectorizable_reduction): Likewise.
12446 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
12447 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
12448 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12449 (vectorizable_assignment): Likewise.
12450 (vectorizable_shift): Likewise.
12451 (vectorizable_operation): Likewise.
12452 * tree-vrp.c (register_edge_assert_for_2): Likewise.
12454 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
12456 * match.pd: Add pow (C, x) simplification.
12458 2017-08-21 Richard Biener <rguenther@suse.de>
12460 PR tree-optimization/81900
12461 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
12462 for blocks with abnormal predecessors.
12463 (compute_antic): Do not set visited flag prematurely.
12465 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
12468 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
12470 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12472 * stor-layout.h (vector_type_mode): Move to...
12473 * tree.h (vector_type_mode): ...here.
12474 * stor-layout.c (vector_type_mode): Move to...
12475 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
12477 2017-08-21 Richard Biener <rguenther@suse.de>
12479 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
12480 register_external_die hooks.
12481 (debug_false_tree_charstarstar_uhwistar): Declare.
12482 (debug_nothing_tree_charstar_uhwi): Likewise.
12483 * debug.c (do_nothing_debug_hooks): Adjust.
12484 (debug_false_tree_charstarstar_uhwistar): New do nothing.
12485 (debug_nothing_tree_charstar_uhwi): Likewise.
12486 * dbxout.c (dbx_debug_hooks): Adjust.
12487 (xcoff_debug_hooks): Likewise.
12488 * sdbout.c (sdb_debug_hooks): Likewise.
12489 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
12490 * dwarf2out.c (macinfo_label_base): New global.
12491 (dwarf2out_register_external_die): New function for the
12492 register_external_die hook.
12493 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
12494 (dwarf2_debug_hooks): Use them.
12495 (dwarf2_lineno_debug_hooks): Adjust.
12496 (struct die_struct): Add with_offset flag.
12497 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
12498 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
12499 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
12500 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
12501 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
12502 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
12503 defining section names for the early LTO debug variants.
12504 (reset_indirect_string): New helper.
12505 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
12506 (print_dw_val): Add support for offsetted symbol references.
12507 (get_ultimate_context): Split out from is_cxx.
12508 (is_cxx): Use get_ultimate_context.
12509 (is_fortran): Add decl overload.
12510 (compute_comp_unit_symbol): Split out worker from
12511 compute_section_prefix.
12512 (compute_section_prefix): Call compute_comp_unit_symbol and
12513 set comdat_type_p here.
12514 (output_die): Skip DIE symbol output for the LTO added one.
12515 Handle DIE symbol references with offset.
12516 (output_comp_unit): Guard section name mangling properly.
12517 For LTO debug sections emit a symbol at the section beginning
12518 which we use to refer to its DIEs.
12519 (add_abstract_origin_attribute): For DIEs registered via
12520 dwarf2out_register_external_die directly refer to the early
12521 DIE rather than indirectly through the shadow one we created.
12522 Remove obsolete call to dwarf2out_abstract_function for
12523 non-function/block origins.
12524 (gen_array_type_die): When generating early LTO debug do
12525 not emit DW_AT_string_length.
12526 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
12527 late when in LTO. As suggested place a gcc_unreachable for
12528 the DECL_ABSTRACT_P case.
12529 (gen_subprogram_die): Avoid another specification DIE
12530 for early built declarations/definitions for the late LTO case.
12531 (gen_variable_die): Add type references for late duplicated VLA dies
12533 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
12534 we have the abstract instance already.
12535 (process_scope_var): Adjust decl DIE contexts in LTO which
12536 first puts them in limbo.
12537 (gen_decl_die): Do not generate type DIEs late apart from
12538 types for VLAs or for decls we do not yet have a DIE. Do not
12539 call dwarf2out_abstract_function late.
12540 (dwarf2out_early_global_decl): Make sure to create DIEs
12541 for abstract instances of a decl first.
12542 (dwarf2out_late_global_decl): Adjust comment.
12543 (output_macinfo_op): With multiple macro sections use
12544 macinfo_label_base to distinguish labels.
12545 (output_macinfo): Likewise. Update macinfo_label_base.
12546 Pass in the line info label.
12547 (note_variable_value_in_expr): When generating LTO resolve
12548 all variable values here by generating DIEs as needed.
12549 (init_sections_and_labels): Add early LTO debug flag parameter
12550 and generate different sections and names if set. Add generation
12551 counter for the labels so we can have multiple of them.
12552 (reset_dies): Helper to allow DIEs to be output multiple times.
12553 (dwarf2out_finish): When outputting DIEs to the fat part of an
12554 LTO object first reset DIEs.
12555 (dwarf2out_early_finish): Output early DIEs when generating LTO.
12556 (modified_type_die): Check for decl_ultimate_origin being self
12558 (gen_type_die_with_usage): Likewise.
12559 (gen_typedef_die): Allow decl_ultimate_origin being self.
12560 (set_decl_abstract_flags): Remove.
12561 (set_block_abstract_flags): Likewise.
12562 (dwarf2out_abstract_function): Treat the early generated DIEs
12563 as the abstract copy and only add DW_AT_inline and
12564 DW_AT_artificial here and call set_decl_origin_self.
12565 If the DIE has an abstract origin don't do anything.
12566 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
12567 if we have none yet (Go fails to build one, PR78628).
12568 (variably_modified_type_p): Prevent endless recursion for Ada
12569 cyclic pointer types.
12570 * lto-streamer-in.c: Include debug.h.
12571 (dref_queue): New global.
12572 (lto_read_tree_1): Stream in DIE references.
12573 (lto_input_tree): Register DIE references.
12574 (input_function): Stream DECL_DEBUG_ARGS.
12575 * lto-streamer-out.c: Include debug.h.
12576 (lto_write_tree_1): Output DIE references.
12577 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
12578 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
12579 (output_function): Stream DECL_DEBUG_ARGS.
12580 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
12581 Stream DECL_ABSTRACT_ORIGIN.
12582 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
12583 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
12584 DECL_CONTEXT for file-scope decls.
12585 * lto-streamer.h (struct dref_entry): Declare.
12586 (dref_queue): Likewise.
12587 * cfgexpand.c (pass_expand::execute): Do not call the
12588 outlining_inline_function hook here.
12589 * lto-wrapper.c (debug_obj): New global.
12590 (tool_cleanup): Unlink it if required.
12591 (debug_objcopy): New function.
12592 (run_gcc): Handle early debug sections in the IL files by
12593 extracting them to separate files, partially linkin them and
12594 feeding the result back as result to the linker.
12595 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
12596 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
12597 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
12598 sections into a separate segment.
12599 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
12601 (darwin_asm_dwarf_section): Likewise.
12602 (darwin_asm_output_dwarf_offset): Likewise.
12603 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
12605 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12606 Alan Hayward <alan.hayward@arm.com>
12607 David Sherwood <david.sherwood@arm.com>
12609 * read-md.h (md_reader::record_potential_iterator_use): Replace
12610 pointer argument with an rtx and an index.
12611 * read-rtl.c (iterator_group::apply_iterator): Likewise.
12612 (apply_mode_iterator): Likewise.
12613 (apply_code_iterator): Likewise.
12614 (apply_int_iterator): Likewise.
12615 (apply_subst_iterator): Likewise.
12616 (record_iterator_use): Likewise.
12617 (record_attribute_use): Likewise.
12618 (md_reader::record_potential_iterator_use): Likewise. Update calls
12619 to record_iterator_use and apply_iterator.
12620 (iterator_use): Replace ptr with x and index.
12621 (attribute_use): Likewise.
12622 (apply_attribute_uses): Update calls to apply_iterator.
12623 (apply_iterators): Likewise. Update initialization of iterator_use.
12624 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
12625 and record_potential_iterator_use.
12626 (rtx_reader::read_rtx_operand): Likewise.
12628 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12629 Alan Hayward <alan.hayward@arm.com>
12630 David Sherwood <david.sherwood@arm.com>
12632 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
12635 2017-08-21 Richard Biener <rguenther@suse.de>
12637 PR middle-end/81884
12638 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
12639 at struct end conservatively when comparing common bases.
12641 2017-08-21 Richard Biener <rguenther@suse.de>
12643 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
12644 (mem_ref_in_stmt): Remove.
12645 (determine_max_movement): Use ref index to get at the reference.
12646 (invariantness_dom_walker::before_dom_children): Deal with
12647 lim data already initialized.
12648 (gather_mem_refs_stmt): Initialize lim data and record ref index.
12650 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
12652 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
12653 (TARGET_ISA_ROUND): Ditto.
12654 (TARGET_ROUND): Ditto.
12655 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
12656 * config/i386/i386.md: Ditto.
12657 * config/i386/sse.md: Ditto.
12658 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
12659 with OPTION_MASK_ISA_SSE4_1.
12661 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
12664 * doc/extend.texi (x86 Built-in Functions): Correct the name of
12665 __builtin_ia32_lzcnt_u16.
12667 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
12670 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
12671 (rs6000_set_current_function): Rewrite function to use it.
12673 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
12676 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
12677 and TYPE_WARN_IF_NOT_ALIGN.
12678 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
12679 (handle_warn_if_not_align): New.
12680 (place_union_field): Call handle_warn_if_not_align.
12681 (place_field): Call handle_warn_if_not_align.
12682 Copy TYPE_WARN_IF_NOT_ALIGN.
12683 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
12684 (layout_type): Likewise.
12685 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
12687 (tree_decl_common): Add warn_if_not_align.
12688 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
12689 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
12690 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
12691 (DECL_WARN_IF_NOT_ALIGN): Likewise.
12692 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
12693 * doc/extend.texi: Document warn_if_not_aligned attribute.
12694 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
12696 2017-08-17 Martin Liska <mliska@suse.cz>
12699 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
12700 (get_data_dependence): Use it as pointer type.
12701 (distribute_loop): Likewise.
12703 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12705 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
12706 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
12707 (p8_vmrgow_<mode>_direct): New define_insn.
12708 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
12709 handle endianness for vmrgew and vmrgow permute patterns.
12711 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
12713 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
12714 * config/rs6000/rs6000-cpus.def: Remove comment.
12715 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
12716 (POWERPC_MASKS): Likewise.
12717 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
12718 use of TARGET_VSX_TIMODE.
12719 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
12720 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
12721 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
12722 (rs6000_option_override_internal): Remove dead code.
12723 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
12724 (rs6000_legitimize_reload_address): Likewise.
12725 (rs6000_legitimate_address_p): Likewise.
12726 (rs6000_opt_masks): Delete "vsx-timode".
12727 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
12728 from function comment.
12729 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
12730 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
12731 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
12733 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
12734 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
12735 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
12736 (VSX_TI): Likewise.
12738 (define_peephole2): Likewise.
12740 2017-08-17 Martin Sebor <msebor@redhat.com>
12743 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
12744 past the end of an array.
12745 (test_pp_format): Add test cases.
12747 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
12749 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
12750 missing ECF_NOTHROW flags.
12752 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
12755 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
12756 operands residing in integer registers.
12757 (*vsx_le_perm_load_<mode>): Likewise.
12758 (*vsx_le_perm_store_<mode>): Likewise.
12759 (define_peephole2): Add peepholes to optimize the above.
12761 2017-08-17 Marek Polacek <polacek@redhat.com>
12763 PR middle-end/81814
12764 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
12765 to mimic what shorten_compare did. Change the return type to bool.
12766 (fold_cond_expr_with_comparison): Update call to
12767 operand_equal_for_comparison_p.
12768 (fold_ternary_loc): Likewise.
12770 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
12772 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
12774 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
12776 2017-08-17 Richard Biener <rguenther@suse.de>
12778 * tree-ssa-structalias.c (solve_graph): When propagating
12779 to successors update the graphs succ edges and avoid duplicate work.
12781 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
12784 * config/i386/i386.c (ix86_option_override_internal): Save target
12785 specific options after ix86_stack_protector_guard_reg was changed.
12787 2017-08-17 Richard Biener <rguenther@suse.de>
12789 PR tree-optimization/81827
12790 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
12791 (new_var_info): Initialize it conservatively.
12792 (get_call_vi): Mark register vars.
12793 (new_scalar_tmp_constraint_exp): Likewise.
12794 (handle_rhs_call): Likewise.
12795 (handle_const_call): Likewise.
12796 (create_function_info_for): Likewise.
12797 (solve_constraints): Sort varinfos to separate register from
12798 non-register vars to pack points-to solution bitmaps during
12801 2017-08-17 Marek Polacek <polacek@redhat.com>
12803 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
12805 2017-08-17 Richard Biener <rguenther@suse.de>
12807 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
12808 to true when overflow is undefined and we saturated the result.
12810 2017-08-17 Alan Modra <amodra@gmail.com>
12813 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
12814 Don't use store multiple if only one reg needs saving.
12815 (interesting_frame_related_regno): New function.
12816 (rs6000_frame_related): Don't emit frame info for regs that
12818 (rs6000_emit_epilogue): Likewise.
12820 2017-08-16 Nathan Sidwell <nathan@acm.org>
12822 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
12823 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
12824 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
12825 * tree.c (free_lang_data_in_type): Use else-if chain. Always
12826 clear TYPE_LANG_1. Remove obsolete member-function stripping.
12827 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
12828 (verify_type): Adjust for TYPE_BINFO move.
12829 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
12830 process TYPE_BINFO directly.
12831 (hash_tree): Likewise.
12832 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
12834 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
12837 2017-08-16 David Malcolm <dmalcolm@redhat.com>
12839 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
12841 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
12844 * config/i386/i386.md (*anddi_1_btr): Change predicates of
12845 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
12846 Add ix86_binary_operator_ok to insn constraint.
12847 (*iordi_1_bts): Ditto.
12848 (*xordi_1_btc): Ditto.
12849 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
12850 Update corresponding peephole2 pattern.
12854 2017-08-16 Bin Cheng <bin.cheng@arm.com>
12856 PR tree-optimization/81832
12857 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
12858 copy loop header which has IFN_LOOP_DIST_ALIAS call.
12860 2017-08-16 Marek Polacek <polacek@redhat.com>
12863 * fold-const.c (fold_indirect_ref_1): Restore original behavior
12864 regarding size_zero_node.
12866 2017-08-16 Martin Liska <mliska@suse.cz>
12869 * config.gcc: Respect previously set extra_objs in case
12872 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
12874 PR tree-optimization/81835
12875 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
12876 the phi SSA_NAME. Check that the condition in a COND_EXPR does
12877 not depend on the phi.
12879 2017-08-16 Alan Modra <amodra@gmail.com>
12881 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
12884 2017-08-16 Alan Modra <amodra@gmail.com>
12886 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
12887 (save_reg_p): ..into this. Update all callers.
12888 (first_reg_to_save): Simplify.
12890 2017-08-16 Alan Modra <amodra@gmail.com>
12892 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
12895 2017-08-15 Joseph Myers <joseph@codesourcery.com>
12899 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
12900 constant count if that count is less than 32.
12902 2017-08-15 Nathan Sidwell <nathan@acm.org>
12904 * gcc.c (execute): Emit friendlier message if inferior is killed
12905 by an external cause.
12907 2017-08-15 Richard Biener <rguenther@suse.de>
12909 PR tree-optimization/81790
12910 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
12911 CONSTRUCTORs from simplifying and VN.
12913 2017-08-14 Martin Sebor <msebor@redhat.com>
12915 * builtin-attrs.def: Add comments.
12917 2017-08-14 Martin Sebor <msebor@redhat.com>
12920 * doc/extend.texi (attribute nonstring): Document new attribute.
12922 2017-08-14 Martin Sebor <msebor@redhat.com>
12925 * tree-diagnostic.c (default_tree_printer): Handle %G.
12926 * gimple-pretty-print.h (percent_G_format): Declare new function.
12927 * gimple-pretty-print.c (percent_G_format): Define.
12928 * tree-pretty-print.c (percent_K_format): Add argument.
12930 2017-08-14 Martin Sebor <msebor@redhat.com>
12932 PR translation/79998
12933 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
12934 Remove a stray space.
12936 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
12939 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
12940 (*iordi_1_bts): Ditto.
12941 (*xordi_1_btc): Ditto.
12943 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12946 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
12947 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12949 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
12950 quoted strings, and make more translator-friendly.
12951 (darwin_rs6000_override_options): Likewise.
12952 (rs6000_option_override_internal): Likewise.
12953 (rs6000_return_in_memory): Fix overlong line.
12954 (init_cmulative_args): Use quoted strings, and make more
12955 translator-friendly.
12956 (rs6000_pass_by_reference): Fix overlong line.
12957 (def_builtin): Use quoted strings.
12958 (altivec_expand_predicate_builtin): Use quoted strings, and make
12959 more translator-friendly.
12960 (htm_expand_builtin): Use quoted strings.
12961 (cpu_expand_builtin): Use quoted strings, and make more
12962 translator-friendly.
12963 (altivec_expand_builtin): Likewise.
12964 (paired_expand_predicate_builtin): Likewise.
12965 (rs6000_invalid_builtin): Likewise.
12966 (builtin_function_type): Use quoted strings.
12967 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
12968 more translator-friendly.
12969 (rs6000_trampoline_init): Likewise.
12970 (rs6000_handle_altivec_attribute): Likewise.
12971 (rs6000_inner_target_options): Use quoted strings.
12972 (rs6000_disable_incompatible_switches): Likewise.
12973 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
12974 strings, and make more translator-friendly.
12975 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
12977 2017-08-14 Bin Cheng <bin.cheng@arm.com>
12979 PR tree-optimization/81799
12980 * tree-loop-distribution.c (version_loop_by_alias_check): Force
12981 cond_expr to simple gimple operand.
12983 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
12985 PR middle-end/46932
12986 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
12988 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
12992 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
12993 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
12994 TARGET_GASISR_PROLOGUES.
12995 * config/avr/avr.c (avr_option_override): Same.
12996 (avr_pass_pre_proep::execute): Same.
12998 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
13001 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
13002 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
13004 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
13006 * config/i386/i386.md (*load_tp_<mode>): Redefine as
13007 define_insn_and_split. Split to a memory load from 0 in
13008 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
13009 using PTR mode iterator.
13010 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
13011 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
13012 (*add_tp_<mode>): Redefine as define_insn_and_split.
13013 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
13014 address space. Merge with *add_tp_x32 using PTR mode iterator.
13015 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
13016 Split to an add with a memory load from 0 in
13017 DEFAULT_TLS_SEG_REG address space.
13019 2017-08-12 Andrew Pinski <apinski@cavium.com>
13021 * config/aarch64/aarch64-option-extensions.def (rdma):
13022 Fix feature string to what Linux prints out in /proc/cpuinfo.
13024 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
13027 * dwarf2out.c (modified_type_die): For C typedef types that have
13028 an ultimate origin, process the ultimate origin instead of the
13030 (gen_typedef_die): Assert that input DECLs have no ultimate
13032 (gen_type_die_with_usage): For typedef variants that have an
13033 ultimate origin, just call gen_decl_die on the original DECL.
13034 (process_scope_var): Avoid creating DIEs for local typedefs and
13035 concrete static variables.
13037 2017-08-12 Alan Modra <amodra@gmail.com>
13041 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
13042 match gnu-user.h startfile.
13043 (ENDFILE_LINUX_SPEC): Similarly.
13045 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
13048 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
13050 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
13052 2017-08-11 Tamar Christina <tamar.christina@arm.com>
13053 * config/aarch64/aarch64.md (mov<mode>): Change.
13054 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
13055 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
13056 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
13058 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
13060 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
13061 for storage order barriers.
13063 2017-08-11 Martin Liska <mliska@suse.cz>
13066 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
13067 variables of void type.
13069 2017-08-11 Martin Liska <mliska@suse.cz>
13071 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
13072 TARGET_SUPPORTS_ALIASES.
13073 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
13074 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
13075 (optimize_weakref): Likewise.
13076 * symtab.c (symtab_node::noninterposable_alias): Likewise.
13077 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
13078 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
13080 2017-08-11 Martin Liska <mliska@suse.cz>
13083 * config/i386/i386.c (make_resolver_func): Do complete
13084 refactoring of the function.
13086 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
13089 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
13090 * config/i386/i386.c (ix86_stack_protect_guard): Use
13091 ix86_stack_protect_guard_symbol_str to generate varible declaration.
13092 * doc/invoke.texi (x86 Options): Document
13093 -mstack-protector-guard-symbol= option.
13095 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
13097 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
13098 * config/i386/i386.c (ix86_split_stack_guard): New function.
13099 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
13100 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
13101 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
13102 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
13103 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
13104 (split_stack_space_check): Call ix86_split_stack_guard.
13106 2017-08-10 Martin Sebor <msebor@redhat.com>
13108 * print-tree.c (print_node): Print location using the established
13110 Replace spaces with colons.
13111 (debug_raw, debug): Ditto.
13113 2017-08-10 Martin Sebor <msebor@redhat.com>
13116 * pretty-print.c (pp_format): Correct the handling of %s precision.
13118 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
13121 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
13123 (ix86_finalize_stack_frame_flags): This. Also clear
13124 frame_pointer_needed if -fno-omit-frame-pointer is used without
13126 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
13127 with ix86_finalize_stack_frame_flags.
13128 (ix86_expand_epilogue): Likewise.
13129 (ix86_expand_split_stack_prologue): Likewise.
13130 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
13132 2017-08-10 Martin Liska <mliska@suse.cz>
13135 * c-attribs.c (handle_target_attribute):
13136 Report warning for an empty string argument of target attribute.
13138 2017-08-09 Jakub Jelinek <jakub@redhat.com>
13141 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
13143 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
13144 or DECL_NONLOCAL labels.
13145 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
13146 or DECL_NONLOCAL labels here.
13148 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
13150 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
13151 to indicate when early gimple folding has been disabled.
13152 (rs6000_gimple_fold_builtin): Add debug content.
13153 (rs6000_invalid_builtin): Fix whitespace.
13154 (rs6000_expand_builtin): Fix whitespace.
13155 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
13157 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
13160 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
13161 SAVE_MULTIPLE if not all the registers that saves, should be saved.
13163 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
13165 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
13166 (qdf24xx): Likewise.
13167 * config/aarch64/aarch64.md: Include falkor.md.
13168 * config/aarch64/falkor.md: New.
13170 2017-08-09 Marek Polacek <polacek@redhat.com>
13173 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
13174 * diagnostic.c (emit_diagnostic): Add a comment.
13175 (emit_diagnostic_valist): New function.
13177 2017-08-09 Marek Polacek <polacek@redhat.com>
13180 * input.c (make_location): New overload.
13181 * input.h (make_location): Declare.
13183 2017-08-08 Alan Modra <amodra@gmail.com>
13184 H.J. Lu <hongjiu.lu@intel.com>
13187 * gcc.c (NO_PIE_SPEC): Delete.
13188 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
13190 (LINK_PIE_SPEC): ..to here.
13191 (LINK_COMMAND_SPEC): Support -no-pie.
13192 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
13193 chain of crtbegin*.o selection, update for PIE_SPEC changes and
13195 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
13196 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
13197 (ENDFILE_CRTEND_SPEC): Similarly.
13199 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
13202 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
13203 (mstack-protector-guard-offset=): Ditto.
13204 * config/i386/i386.c (ix86_option_override): Handle
13205 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
13207 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
13208 ix86_stack_protect_guard_offset variables.
13209 (TARGET_STACK_PROTECT_GUARD): Always define.
13210 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
13211 and -mstack-protector-guard-offset= options.
13213 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13215 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
13216 boundary case for the last candidate.
13218 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13220 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
13221 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
13223 2017-08-08 Tamar Christina <tamar.christina@arm.com>
13225 PR middle-end/19706
13226 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
13227 * config/aarch64/aarch64-builtins.c
13228 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
13229 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
13230 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
13232 2017-08-08 Tamar Christina <tamar.christina@arm.com>
13233 Andrew Pinski <pinskia@gmail.com>
13235 PR middle-end/19706
13236 * internal-fn.def (XORSIGN): New.
13237 * optabs.def (xorsign_optab): New.
13238 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
13239 (convert_expand_mult_copysign): New.
13240 (pass_optimize_widening_mul::execute): Call
13241 convert_expand_mult_copysign.
13243 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13245 PR tree-optimization/81354
13246 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
13247 Insert on edges rather than explicitly creating landing pads.
13248 (analyze_candidates_and_replace): Commit edge inserts.
13250 2017-08-08 Richard Biener <rguenther@suse.de>
13252 PR middle-end/81719
13253 * tree-ssa-loop-niter.c: Include tree-dfa.h.
13254 (expand_simple_operations): Also look through ADDR_EXPRs with
13255 MEM_REF bases treating them as POINTER_PLUS_EXPR.
13257 2017-08-08 Richard Biener <rguenther@suse.de>
13259 PR tree-optimization/81723
13260 * tree-vect-slp.c (struct bst_traits): New hash traits.
13261 (bst_fail): New global.
13262 (vect_build_slp_tree_2): New worker, split out from ...
13263 (vect_build_slp_tree): ... this now wrapping it with using
13264 bst_fail set to cache SLP tree build fails. Properly handle
13266 (vect_analyze_slp_instance): Allocate and free bst_fail.
13268 2017-08-08 Martin Liska <mliska@suse.cz>
13271 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
13272 LABEL_DECLs that can be from a different function.
13274 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13276 PR tree-optimization/81744
13277 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
13278 loop's number of iterations.
13280 2017-08-08 Martin Liska <mliska@suse.cz>
13282 * asan.c: Include header files.
13283 * attribs.c (build_decl_attribute_variant): New function moved
13285 (build_type_attribute_qual_variant): Likewise.
13286 (cmp_attrib_identifiers): Likewise.
13287 (simple_cst_list_equal): Likewise.
13288 (omp_declare_simd_clauses_equal): Likewise.
13289 (attribute_value_equal): Likewise.
13290 (comp_type_attributes): Likewise.
13291 (build_type_attribute_variant): Likewise.
13292 (lookup_ident_attribute): Likewise.
13293 (remove_attribute): Likewise.
13294 (merge_attributes): Likewise.
13295 (merge_type_attributes): Likewise.
13296 (merge_decl_attributes): Likewise.
13297 (merge_dllimport_decl_attributes): Likewise.
13298 (handle_dll_attribute): Likewise.
13299 (attribute_list_equal): Likewise.
13300 (attribute_list_contained): Likewise.
13301 * attribs.h (lookup_attribute): New function moved from tree.[ch].
13302 (lookup_attribute_by_prefix): Likewise.
13303 * bb-reorder.c: Include header files.
13304 * builtins.c: Likewise.
13305 * calls.c: Likewise.
13306 * cfgexpand.c: Likewise.
13307 * cgraph.c: Likewise.
13308 * cgraphunit.c: Likewise.
13309 * convert.c: Likewise.
13310 * dwarf2out.c: Likewise.
13311 * final.c: Likewise.
13312 * fold-const.c: Likewise.
13313 * function.c: Likewise.
13314 * gimple-expr.c: Likewise.
13315 * gimple-fold.c: Likewise.
13316 * gimple-pretty-print.c: Likewise.
13317 * gimple.c: Likewise.
13318 * gimplify.c: Likewise.
13319 * hsa-common.c: Likewise.
13320 * hsa-gen.c: Likewise.
13321 * internal-fn.c: Likewise.
13322 * ipa-chkp.c: Likewise.
13323 * ipa-cp.c: Likewise.
13324 * ipa-devirt.c: Likewise.
13325 * ipa-fnsummary.c: Likewise.
13326 * ipa-inline.c: Likewise.
13327 * ipa-visibility.c: Likewise.
13329 * lto-cgraph.c: Likewise.
13330 * omp-expand.c: Likewise.
13331 * omp-general.c: Likewise.
13332 * omp-low.c: Likewise.
13333 * omp-offload.c: Likewise.
13334 * omp-simd-clone.c: Likewise.
13335 * opts-global.c: Likewise.
13336 * passes.c: Likewise.
13337 * predict.c: Likewise.
13338 * sancov.c: Likewise.
13339 * sanopt.c: Likewise.
13340 * symtab.c: Likewise.
13341 * toplev.c: Likewise.
13342 * trans-mem.c: Likewise.
13343 * tree-chkp.c: Likewise.
13344 * tree-eh.c: Likewise.
13345 * tree-into-ssa.c: Likewise.
13346 * tree-object-size.c: Likewise.
13347 * tree-parloops.c: Likewise.
13348 * tree-profile.c: Likewise.
13349 * tree-ssa-ccp.c: Likewise.
13350 * tree-ssa-live.c: Likewise.
13351 * tree-ssa-loop.c: Likewise.
13352 * tree-ssa-sccvn.c: Likewise.
13353 * tree-ssa-structalias.c: Likewise.
13354 * tree-ssa.c: Likewise.
13355 * tree-streamer-in.c: Likewise.
13356 * tree-vectorizer.c: Likewise.
13357 * tree-vrp.c: Likewise.
13358 * tsan.c: Likewise.
13359 * ubsan.c: Likewise.
13360 * varasm.c: Likewise.
13361 * varpool.c: Likewise.
13362 * tree.c: Remove functions moved to attribs.[ch].
13363 * tree.h: Likewise.
13364 * config/aarch64/aarch64.c: Add attrs.h header file.
13365 * config/alpha/alpha.c: Likewise.
13366 * config/arc/arc.c: Likewise.
13367 * config/arm/arm.c: Likewise.
13368 * config/avr/avr.c: Likewise.
13369 * config/bfin/bfin.c: Likewise.
13370 * config/c6x/c6x.c: Likewise.
13371 * config/cr16/cr16.c: Likewise.
13372 * config/cris/cris.c: Likewise.
13373 * config/darwin.c: Likewise.
13374 * config/epiphany/epiphany.c: Likewise.
13375 * config/fr30/fr30.c: Likewise.
13376 * config/frv/frv.c: Likewise.
13377 * config/ft32/ft32.c: Likewise.
13378 * config/h8300/h8300.c: Likewise.
13379 * config/i386/winnt.c: Likewise.
13380 * config/ia64/ia64.c: Likewise.
13381 * config/iq2000/iq2000.c: Likewise.
13382 * config/lm32/lm32.c: Likewise.
13383 * config/m32c/m32c.c: Likewise.
13384 * config/m32r/m32r.c: Likewise.
13385 * config/m68k/m68k.c: Likewise.
13386 * config/mcore/mcore.c: Likewise.
13387 * config/microblaze/microblaze.c: Likewise.
13388 * config/mips/mips.c: Likewise.
13389 * config/mmix/mmix.c: Likewise.
13390 * config/mn10300/mn10300.c: Likewise.
13391 * config/moxie/moxie.c: Likewise.
13392 * config/msp430/msp430.c: Likewise.
13393 * config/nds32/nds32-isr.c: Likewise.
13394 * config/nds32/nds32.c: Likewise.
13395 * config/nios2/nios2.c: Likewise.
13396 * config/nvptx/nvptx.c: Likewise.
13397 * config/pa/pa.c: Likewise.
13398 * config/pdp11/pdp11.c: Likewise.
13399 * config/powerpcspe/powerpcspe.c: Likewise.
13400 * config/riscv/riscv.c: Likewise.
13401 * config/rl78/rl78.c: Likewise.
13402 * config/rx/rx.c: Likewise.
13403 * config/s390/s390.c: Likewise.
13404 * config/sh/sh.c: Likewise.
13405 * config/sol2.c: Likewise.
13406 * config/sparc/sparc.c: Likewise.
13407 * config/spu/spu.c: Likewise.
13408 * config/stormy16/stormy16.c: Likewise.
13409 * config/tilegx/tilegx.c: Likewise.
13410 * config/tilepro/tilepro.c: Likewise.
13411 * config/v850/v850.c: Likewise.
13412 * config/vax/vax.c: Likewise.
13413 * config/visium/visium.c: Likewise.
13414 * config/xtensa/xtensa.c: Likewise.
13416 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13419 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
13420 constraints since the -mupper-regs-* switches have been
13422 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
13423 into a vector from a double word element that was extracted from
13424 another vector, and eliminate extra XXPERMDI instructions.
13425 (vsx_concat_<mode>_2): Likewise.
13426 (vsx_concat_<mode>_3): Likewise.
13427 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
13428 concat to allow optimizing inserts from previous extracts.
13430 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
13432 * config/i386/i386.c (ix86_stack_protect_guard): Generate
13433 memory reference to a SSP offset in TLS address space.
13434 (ix86_print_operand) <case '@'>: Remove.
13435 (ix86_print_operand_punct_valid_p): Remove '@' code.
13436 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
13437 UNSPEC_SP_TLS_TEST.
13438 (stack_tls_protect_set_<mode>): Remove.
13439 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
13440 (stack_tls_protect_test_<mode>): Remove.
13441 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
13443 2017-08-07 Olivier Hainque <hainque@adacore.com>
13446 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
13448 2017-08-07 Douglas Rupp <rupp@adacore.com>
13450 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
13451 variable was referenced as multidir in command.
13453 2017-08-07 Jakub Jelinek <jakub@redhat.com>
13456 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
13459 2017-08-07 Martin Liska <mliska@suse.cz>
13461 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
13462 * config/rl78/rl78.c: Add include of attribs.h.
13463 * config/sh/sh.c: Likewise.
13464 * config/v850/v850.c: Likewise.
13466 2017-08-07 Tom de Vries <tom@codesourcery.com>
13468 PR middle-end/78266
13469 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
13471 2017-08-07 Martin Liska <mliska@suse.cz>
13473 * config/mips/mips.c: Include attribs.h.
13475 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
13478 * doc/invoke.texi: Document change in behvaior for -Ofast for
13481 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
13483 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
13485 (aarch64_pop_regs): Likewise.
13486 (aarch64_gen_load_pair): Likewise.
13487 (aarch64_save_callee_saves): Likewise.
13488 (aarch64_restore_callee_saves): Likewise.
13490 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13492 * config/i386/i386.c: Revert the last change.
13494 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13497 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
13499 (ix86_finalize_stack_frame_flags): This. Also clear
13500 frame_pointer_needed if -fno-omit-frame-pointer is used without
13502 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
13503 with ix86_finalize_stack_frame_flags.
13504 (ix86_expand_epilogue): Likewise.
13505 (ix86_expand_split_stack_prologue): Likewise.
13507 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13510 * config/i386/i386.c (get_builtin_code_for_version): Set priority
13511 to P_AES for Westmere.
13513 2017-08-07 Jonathan Yong <10walls@gmail.com>
13515 * config/i386/mingw.opt (fset-stack-executable): Removed.
13516 * config/i386/cygming.opt (fset-stack-executable): Moved
13518 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
13520 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
13522 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
13524 2017-08-07 Marek Polacek <polacek@redhat.com>
13526 PR middle-end/81737
13527 * fold-const.c (fold_indirect_ref_1): Check type_domain.
13529 2017-08-07 Martin Liska <mliska@suse.cz>
13531 * attribs.h (canonicalize_attr_name): New function.
13532 (cmp_attribs): Move from c-format.c and adjusted.
13533 (is_attribute_p): Moved from tree.h.
13534 * tree-inline.c: Add new includes.
13535 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
13536 (private_is_attribute_p): Remove.
13537 (private_lookup_attribute): Likewise.
13538 (private_lookup_attribute_by_prefix): Simplify.
13539 (remove_attribute): Use is_attribute_p.
13540 * tree.h: Remove removed declarations.
13542 2017-08-07 Jakub Jelinek <jakub@redhat.com>
13544 PR middle-end/81698
13545 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
13546 instead of computing it in the function. Formatting fix.
13547 (expand_case): Don't rely on default_edge being the first edge,
13548 clear it if removing it, pass default_edge to
13549 emit_case_dispatch_table.
13550 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
13553 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
13555 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
13556 insn in the function, emit NOP after the insn.
13558 2017-08-06 Tom de Vries <tom@codesourcery.com>
13560 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
13563 2017-08-06 Tom de Vries <tom@codesourcery.com>
13565 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
13568 2017-08-04 Yury Gribov <tetra2005@gmail.com>
13570 PR tree-optimization/57371
13571 * match.pd: New pattern.
13573 2017-08-04 Marek Polacek <polacek@redhat.com>
13575 PR middle-end/81695
13576 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
13577 perform the computation in offset_int.
13579 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13581 PR tree-optimization/81136
13582 * tree-vectorizer.h: Include tree-hash-traits.h.
13583 (vec_base_alignments): New typedef.
13584 (vec_info): Add a base_alignments field.
13585 (vect_record_base_alignments): Declare.
13586 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
13588 (DR_IS_CONDITIONAL_IN_STMT): New macro.
13589 (create_data_ref): Add an is_conditional_in_stmt argument.
13590 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
13591 the is_conditional_in_stmt field.
13592 (data_ref_loc): Add an is_conditional_in_stmt field.
13593 (get_references_in_stmt): Set the is_conditional_in_stmt field.
13594 (find_data_references_in_stmt): Update call to create_data_ref.
13595 (graphite_find_data_references_in_stmt): Likewise.
13596 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
13597 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
13598 (vect_record_base_alignment): New function.
13599 (vect_record_base_alignments): Likewise.
13600 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
13601 for nested statements even if we fail to compute a misalignment.
13602 Use pooled base alignments for unconditional references.
13603 (vect_find_same_alignment_drs): Compare base addresses instead
13605 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
13606 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
13608 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13610 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
13611 Add an explicit name for the enum. Use auto_vec for slp_instances
13612 and grouped_stores.
13613 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
13615 (_bb_vec_info): Add a constructor and destructor.
13616 (vinfo_for_stmt): Return NULL for uids of -1 as well.
13617 (destroy_loop_vec_info): Delete.
13618 (vect_destroy_datarefs): Likewise.
13619 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
13620 (vec_info::vec_info): New function.
13621 (vec_info::~vec_info): Likewise.
13622 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
13623 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
13624 destroy_loop_vec_info.
13625 * tree-vect-loop.c (new_loop_vec_info): Replace with...
13626 (_loop_vec_info::_loop_vec_info): ...this.
13627 (destroy_loop_vec_info): Replace with...
13628 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
13629 the stmt_vec_infos. Leave handling of vec_info information to its
13630 destructor. Remove explicit vector releases.
13631 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
13632 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
13633 * tree-vect-slp.c (new_bb_vec_info): Replace with...
13634 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
13635 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
13636 (destroy_bb_vec_info): Replace with...
13637 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
13638 information to its destructor.
13639 (vect_slp_analyze_bb_1): Use new and delete instead of
13640 new_bb_vec_info and destroy_bb_vec_info.
13641 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
13644 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13646 * tree-data-ref.h (subscript): Add access_fn field.
13647 (data_dependence_relation): Add could_be_independent_p.
13648 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
13649 (same_access_functions): Move to tree-data-ref.c.
13650 * tree-data-ref.c (ref_contains_union_access_p): New function.
13651 (access_fn_component_p): Likewise.
13652 (access_fn_components_comparable_p): Likewise.
13653 (dr_analyze_indices): Add a reference to access_fn_component_p.
13654 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
13656 (constant_access_functions): Likewise.
13657 (add_other_self_distances): Likewise.
13658 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
13659 (initialize_data_dependence_relation): Use XCNEW and remove
13660 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
13661 of access functions that have the same type. Allow the
13662 subsequence to end with different bases in some circumstances.
13663 Record the chosen access functions in SUB_ACCESS_FN.
13664 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
13665 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
13666 (subscript_dependence_tester_1): Likewise dra and drb.
13667 (build_classic_dist_vector): Update calls accordingly.
13668 (subscript_dependence_tester): Likewise.
13669 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
13670 DDR_COULD_BE_INDEPENDENT_P.
13671 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
13672 comp_alias_ddrs instead of may_alias_ddrs.
13673 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
13675 (vect_analyze_data_ref_dependence): Use it if
13676 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
13677 distance vectors if that fails.
13678 (dependence_distance_ge_vf): New function.
13679 (vect_prune_runtime_alias_test_list): Use it. Don't clear
13680 LOOP_VINFO_MAY_ALIAS_DDRS.
13682 2017-08-04 Richard Biener <rguenther@suse.de>
13684 PR middle-end/81705
13685 * fold-const.c (fold_binary_loc): Properly restrict
13686 minus_var0 && minus_var1 case when associating undefined overflow
13689 2017-08-04 Tom de Vries <tom@codesourcery.com>
13691 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
13693 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13695 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13696 Don't start diagnostic messages with a capital letter.
13697 * config/rs6000/rs6000.c (rs6000_option_override_internal):
13699 (rs6000_invalid_builtin): Likewise.
13700 (rs6000_trampoline_init): Likewise.
13702 2017-08-03 Jakub Jelinek <jakub@redhat.com>
13705 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
13706 after setting changeable df flags.
13708 2017-08-03 Richard Biener <rguenther@suse.de>
13710 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
13711 up if the use is in USE - X.
13713 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
13715 * toplev.c (dumpfile.h): New include.
13716 (internal_error_reentered): New static function. Use it...
13717 (internal_error_function): ...here to handle reentered internal_error.
13719 2017-08-03 Richard Biener <rguenther@suse.de>
13721 PR middle-end/81148
13722 * fold-const.c (split_tree): Add minus_var and minus_con
13723 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
13724 here but always use minus_*.
13725 (associate_trees): Assert we never associate with MINUS_EXPR
13726 and NULL first operand. Do not recurse for PLUS_EXPR operands
13727 when associating as MINUS_EXPR either.
13728 (fold_binary_loc): Track minus_var and minus_con.
13730 2017-08-03 Tom de Vries <tom@codesourcery.com>
13733 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
13734 ACCEL_COMPILER, apply finish_options on
13735 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
13737 2017-08-03 Tom de Vries <tom@codesourcery.com>
13740 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
13741 function_entry_patch_area_size > 0.
13743 2017-08-03 Jakub Jelinek <jakub@redhat.com>
13746 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
13747 instead of 10??LU, perform unit multiplication in wide_int,
13748 don't change alloc_object_size_limit if the limit is larger
13751 PR tree-optimization/81655
13752 PR tree-optimization/81588
13753 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
13754 the case when ranges[i].low and high are 1 for unsigned type with
13757 PR middle-end/81052
13758 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
13759 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
13761 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13763 * tree-vrp.h: Add include guard.
13765 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
13768 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
13769 (ud2): New insn pattern.
13770 * config/i386/i386.c (ix86_expand_epilogue):
13771 For naked functions, generate ud2 instead of trap insn.
13773 2017-08-02 Marek Polacek <polacek@redhat.com>
13776 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
13778 2017-08-02 Tom de Vries <tom@codesourcery.com>
13779 Cesar Philippidis <cesar@codesourcery.com>
13781 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
13782 Add missing edge probabilities.
13784 2017-08-02 Tamar Christina <tamar.christina@arm.com>
13786 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
13787 Correct endianness.
13789 2017-08-02 Jakub Jelinek <jakub@redhat.com>
13791 PR middle-end/79499
13792 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
13793 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
13794 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
13795 prologue_seq sequences - if any.
13797 2017-08-02 Richard Biener <rguenther@suse.de>
13799 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
13800 via vectors if supported, integer extracts via punning if supported
13801 or otherwise vector extracts.
13803 2017-08-02 Richard Biener <rguenther@suse.de>
13805 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
13807 (bitmap_insert_into_set): ... this.
13809 2017-08-02 Richard Biener <rguenther@suse.de>
13811 PR tree-optimization/81633
13813 2015-08-17 Alan Hayward <alan.hayward@arm.com>
13815 PR tree-optimization/71752
13816 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
13818 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
13820 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
13821 (machine_function::call_ms2sysv_pad_out): Remove field.
13822 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
13823 (ix86_compute_frame_layout): Likewise.
13825 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
13828 * config/i386/i386.c (ix86_set_func_type): Disallow naked
13829 attribute with interrupt attribute.
13831 2017-08-01 Andrew Pinski <apinski@cavium.com>
13833 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
13834 BIT_INSERT_EXPR's operand 1
13835 to see if the types precision matches.
13837 2017-08-01 Martin Liska <mliska@suse.cz>
13839 PR middle-end/70140
13840 * builtins.c (expand_builtin_memcpy_args): Remove.
13841 (expand_builtin_memcpy): Call newly added function
13842 expand_builtin_memory_copy_args.
13843 (expand_builtin_memcpy_with_bounds): Likewise.
13844 (expand_builtin_mempcpy): Remove last argument.
13845 (expand_builtin_mempcpy_with_bounds): Likewise.
13846 (expand_builtin_memory_copy_args): New function created from
13847 expand_builtin_mempcpy_args with small modifications.
13848 (expand_builtin_mempcpy_args): Remove.
13849 (expand_builtin_stpcpy): Remove unused argument.
13850 (expand_builtin): Likewise.
13851 (expand_builtin_with_bounds): Likewise.
13853 2017-08-01 Martin Liska <mliska@suse.cz>
13856 Make mempcpy more optimal (PR middle-end/70140).
13858 2017-08-01 Jakub Jelinek <jakub@redhat.com>
13861 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
13862 __builtin_vec_cmpne verify both arguments are compatible vectors
13863 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
13864 verify arg1_type is a pointer or array type. For __builtin_vec_st,
13865 move computation of aligned to after checking the argument types.
13869 * config/rs6000/vsx.md (vextract_fp_from_shorth,
13870 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
13873 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
13874 Doug Rupp <rupp@adacore.com>
13875 Olivier Hainque <hainque@adacore.com>
13877 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
13878 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
13880 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
13881 for TARGET_OS_CPP_BUILTIN.
13882 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
13883 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
13885 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
13886 passing required abi options to the assembler for EABI configurations.
13887 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
13888 of .text.hot and .text.unlikely sections for kernel modules when
13889 using ARM style exceptions.
13890 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
13891 options. Add EXTRA_CC1_SPEC.
13892 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
13894 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
13896 (ARM_TARGET2_DWARF_FORMAT): Define.
13897 * config/arm/t-vxworks: Adjust multilib control to removal of the
13898 Diab command line options.
13900 2017-08-01 Martin Liska <mliska@suse.cz>
13902 PR gcov-profile/81561
13903 * gcov.c (unblock): Make unblocking safe as we need to preserve
13904 index correspondence of blocks and block_lists.
13906 2017-08-01 Richard Biener <rguenther@suse.de>
13908 PR tree-optimization/81181
13909 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
13910 (compute_antic): ... end of iteration here.
13912 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
13914 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
13915 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
13916 (ftree-slp-vectorize): Likewise.
13917 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
13918 can no longer be set independent of flag_tree_loop_vectorize.
13919 * omp-general.c (emp_max_vf): Likewise.
13920 * opts.c (enable_fdo_optimizations): Remove references to
13921 flag_tree_vectorize, these are now implicit.
13922 (common_handle_option): Remove handling for OPT_ftree_vectorize,
13923 and leave it for the options machinery.
13925 2017-08-01 Martin Liska <mliska@suse.cz>
13927 PR middle-end/70140
13928 * builtins.c (expand_builtin_memcpy_args): Remove.
13929 (expand_builtin_memcpy): Call newly added function
13930 expand_builtin_memory_copy_args.
13931 (expand_builtin_memcpy_with_bounds): Likewise.
13932 (expand_builtin_mempcpy): Remove last argument.
13933 (expand_builtin_mempcpy_with_bounds): Likewise.
13934 (expand_builtin_memory_copy_args): New function created from
13935 expand_builtin_mempcpy_args with small modifications.
13936 (expand_builtin_mempcpy_args): Remove.
13937 (expand_builtin_stpcpy): Remove unused argument.
13938 (expand_builtin): Likewise.
13939 (expand_builtin_with_bounds): Likewise.
13941 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
13944 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
13945 print "ds:" only for immediates in generic address space.
13947 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
13950 * config/i386/i386.c (ix86_funciton_naked): New prototype.
13951 (ix86_function_ok_for_sibcall): Return false for naked functions.
13953 2017-08-01 Richard Biener <rguenther@suse.de>
13955 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
13956 (compute_antic): Seed worklist with exit block predecessors.
13957 * cfganal.c (dfs_find_deadend): For a cycle return the source
13958 of the edge closing it.
13960 2017-08-01 Tamar Christina <tamar.christina@arm.com>
13962 * config/aarch64/aarch64.c
13963 (aarch64_can_const_movi_rtx_p): Move 0 check.
13965 2017-08-01 Bin Cheng <bin.cheng@arm.com>
13967 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
13968 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
13970 * match.pd: Ditto in address comparison pattern.
13972 2017-08-01 Bin Cheng <bin.cheng@arm.com>
13974 PR tree-optimization/81627
13975 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
13976 closed ssa form for store-store chain.
13978 2017-08-01 Bin Cheng <bin.cheng@arm.com>
13980 PR tree-optimization/81620
13981 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
13982 for store-store chain.
13984 2017-08-01 Jakub Jelinek <jakub@redhat.com>
13986 PR tree-optimization/81588
13987 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
13988 ranges[i].in_p, invert comparison code ccode. For >/>=,
13989 swap rhs1 and rhs2 and comparison code unconditionally,
13990 for </<= don't do that. Don't swap rhs1/rhs2 again if
13991 ranges[i].in_p, instead invert comparison code ccode if
13992 opcode or oe->rank is BIT_IOR_EXPR.
13995 * optabs.def (vec_extract_optab, vec_init_optab): Change from
13996 a direct optab to conversion optab.
13997 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
13998 with GET_MODE_INNER as last argument instead of optab_handler.
13999 * expmed.c (extract_bit_field_1): Likewise. Use vector from
14000 vector extraction if possible and optab is available.
14001 * expr.c (store_constructor): Use convert_optab_handler instead
14002 of optab_handler. Use vector initialization from smaller
14003 vectors if possible and optab is available.
14004 * tree-vect-stmts.c (vectorizable_load): Likewise.
14005 * doc/md.texi (vec_extract, vec_init): Document that the optabs
14006 now have two modes.
14007 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
14008 of vec_init from half-sized vectors with the same element mode.
14009 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
14010 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
14011 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
14012 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
14013 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
14014 after mode in gen_vec_extract* calls.
14015 (vec_extract<mode>): Renamed to ...
14016 (vec_extract<mode><ssescalarmodelower>): ... this.
14017 (vec_extract<mode><ssehalfvecmodelower>): New expander.
14018 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
14019 element mode after mode in gen_vec_init* calls.
14020 (VEC_INIT_HALF_MODE): New mode iterator.
14021 (vec_init<mode>): Renamed to ...
14022 (vec_init<mode><ssescalarmodelower>): ... this.
14023 (vec_init<mode><ssehalfvecmodelower>): New expander.
14024 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
14025 (vec_extractv2sfsf): ... this.
14026 (vec_initv2sf): Renamed to ...
14027 (vec_initv2sfsf): ... this.
14028 (vec_extractv2si): Renamed to ...
14029 (vec_extractv2sisi): ... this.
14030 (vec_initv2si): Renamed to ...
14031 (vec_initv2sisi): ... this.
14032 (vec_extractv4hi): Renamed to ...
14033 (vec_extractv4hihi): ... this.
14034 (vec_initv4hi): Renamed to ...
14035 (vec_initv4hihi): ... this.
14036 (vec_extractv8qi): Renamed to ...
14037 (vec_extractv8qiqi): ... this.
14038 (vec_initv8qi): Renamed to ...
14039 (vec_initv8qiqi): ... this.
14040 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
14041 (vec_init<mode>): Renamed to ...
14042 (vec_init<mode><VEC_base_l>): ... this.
14043 (vec_extract<mode>): Renamed to ...
14044 (vec_extract<mode><VEC_base_l>): ... this.
14045 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
14046 (vec_initv2sfsf): ... this.
14047 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
14048 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
14049 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
14050 element mode after mode in gen_vec_init* calls.
14051 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
14052 (vec_init<mode><Vel>): ... this.
14053 (vec_extract<mode>): Renamed to ...
14054 (vec_extract<mode><Vel>): ... this.
14055 * config/aarch64/iterators.md (Vel): New mode attribute.
14056 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
14057 Add element mode after mode in gen_vec_extract* calls.
14058 * config/s390/vector.md (non_vec_l): New mode attribute.
14059 (vec_extract<mode>): Renamed to ...
14060 (vec_extract<mode><non_vec_l>): ... this.
14061 (vec_init<mode>): Renamed to ...
14062 (vec_init<mode><non_vec_l>): ... this.
14063 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
14064 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
14066 * config/arm/iterators.md (V_elem_l): New mode attribute.
14067 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
14068 (vec_extract<mode><V_elem_l>): ... this.
14069 (vec_extractv2di): Renamed to ...
14070 (vec_extractv2didi): ... this.
14071 (vec_init<mode>): Renamed to ...
14072 (vec_init<mode><V_elem_l>): ... this.
14073 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
14074 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
14075 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
14076 Add element mode after gen_vec_extract* calls.
14077 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
14078 (vec_init<mode><unitmode>): ... this.
14079 (vec_extract<mode>): Renamed to ...
14080 (vec_extract<mode><unitmode>): ... this.
14081 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
14082 (vec_init<mode><unitmode>): ... this.
14083 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
14084 (vec_initv2sfsf): ... this.
14085 (vec_extractv2sf): Renamed to ...
14086 (vec_extractv2sfsf): ... this.
14087 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
14088 Add element mode after gen_vec_extract* calls.
14089 * config/mips/mips.md (unitmode): New mode iterator.
14090 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
14091 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
14092 * config/spu/spu.md (inner_l): New mode attribute.
14093 (vec_init<mode>): Renamed to ...
14094 (vec_init<mode><inner_l>): ... this.
14095 (vec_extract<mode>): Renamed to ...
14096 (vec_extract<mode><inner_l>): ... this.
14097 * config/sparc/sparc.md (veltmode): New mode iterator.
14098 (vec_init<VMALL:mode>): Renamed to ...
14099 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
14100 * config/ia64/vect.md (vec_initv2si): Renamed to ...
14101 (vec_initv2sisi): ... this.
14102 (vec_initv2sf): Renamed to ...
14103 (vec_initv2sfsf): ... this.
14104 (vec_extractv2sf): Renamed to ...
14105 (vec_extractv2sfsf): ... this.
14106 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
14107 (vec_init<mode>): Renamed to ...
14108 (vec_init<mode><VEC_base_l>): ... this.
14109 (vec_extract<mode>): Renamed to ...
14110 (vec_extract<mode><VEC_base_l>): ... this.
14111 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
14112 (vec_initv2sfsf): ... this.
14113 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
14114 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
14115 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
14116 gen_vec_init* calls.
14118 2017-08-01 Richard Biener <rguenther@suse.de>
14120 PR tree-optimization/81297
14121 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
14122 TREE_OVERFLOW from INTEGER_CSTs.
14124 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
14126 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
14128 2017-07-31 Carl Love <cel@us.ibm.com>
14130 * config/rs6000/rs6000-c: Add support for built-in functions
14131 vector signed char vec_xl_be (signed long long, signed char *);
14132 vector unsigned char vec_xl_be (signed long long, unsigned char *);
14133 vector signed int vec_xl_be (signed long long, signed int *);
14134 vector unsigned int vec_xl_be (signed long long, unsigned int *);
14135 vector signed long long vec_xl_be (signed long long, signed long long *);
14136 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
14137 vector signed short vec_xl_be (signed long long, signed short *);
14138 vector unsigned short vec_xl_be (signed long long, unsigned short *);
14139 vector double vec_xl_be (signed long long, double *);
14140 vector float vec_xl_be (signed long long, float *);
14141 * config/rs6000/altivec.h (vec_xl_be): Add #define.
14142 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
14143 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
14145 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
14146 (altivec_expand_builtin): Add switch statement to call
14147 altivec_expand_xl_be for each builtin.
14148 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
14149 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
14150 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
14151 __builtin_vsx_le_be_v16qi.
14152 * doc/extend.texi: Update the built-in documentation file for the
14153 new built-in functions.
14155 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
14158 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
14160 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
14162 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14164 * config.gcc: Add z14.
14165 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
14166 CPU model numbers for z13s and z14.
14167 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
14169 * config/s390/s390-opts.h (enum processor_type): Rename
14170 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
14171 * config/s390/s390.c (processor_table): Add field for CPU name to
14172 be passed to Binutils.
14173 (s390_asm_output_machine_for_arch): Use the new field in
14174 processor_table for Binutils.
14175 (s390_expand_builtin): Replace arch12 with z14.
14176 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
14177 (s390_get_sched_attrmask): Likewise.
14178 (s390_get_unit_mask): Likewise.
14179 * config/s390/s390.opt: Add z14 to processor_type enum.
14181 2017-07-31 Martin Jambor <mjambor@suse.cz>
14184 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
14185 regardless of optimization level.
14187 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
14188 Martin Liska <mliska@suse.cz>
14190 * predict.def: Remove old comment and adjust probability.
14191 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
14192 PREDICT statements.
14194 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
14197 * config/i386/i386.c (ix86_function_naked): New function.
14198 (ix86_can_use_return_insn_p): Return false for naked functions.
14199 (ix86_expand_prologue): Skip prologue for naked functions.
14200 (ix86_expand_epilogue): Skip epilogue for naked functions
14201 and emit trap instruction.
14202 (ix86_warn_func_return): New function.
14203 (ix86_attribute_table): Add "naked" attribute specification.
14204 (TARGET_WARN_FUNC_RETURN): Define.
14205 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
14207 2017-07-31 Martin Liska <mliska@suse.cz>
14209 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
14210 (dump_gimple_bb_header): Always dump BB info.
14211 (pp_cfg_jump): Do not append info about BB when dumping a jump.
14213 2017-07-31 Martin Liska <mliska@suse.cz>
14216 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
14217 also with current_function_decl non-null equality.
14219 2017-07-31 Jakub Jelinek <jakub@redhat.com>
14222 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
14223 change type to the element type, instead add eltype variable and
14224 use it where we are interested in the element type.
14226 PR tree-optimization/81603
14227 * ipa-polymorphic-call.c
14228 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
14229 offset arithmetic in offset_int, bail out if the resulting bit offset
14230 doesn't fit into shwi.
14232 2017-07-31 Martin Liska <mliska@suse.cz>
14234 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
14235 (gimplify_save_expr): Fix comment.
14237 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
14240 * config/i386/i386.c (ix86_function_arg): Update arguments for
14242 (ix86_compute_frame_layout): Set the initial stack offset to
14243 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
14244 INCOMING_FRAME_SP_OFFSET.
14245 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
14246 stack before exception handler returns.
14247 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
14248 the 'ERROR_CODE' for exception handler.
14250 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
14252 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
14253 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
14254 (ASM_OUTPUT_REG_POP): Ditto.
14255 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
14256 instead of asm_fprintf to output pure string.
14258 2017-07-29 Jakub Jelinek <jakub@redhat.com>
14260 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
14261 to imported_module_or_decl hook.
14262 (debug_nothing_tree_tree_tree_bool): Remove.
14263 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
14264 * debug.c (do_nothing_debug_hooks): Use
14265 debug_nothing_tree_tree_tree_bool_bool instead of
14266 debug_nothing_tree_tree_tree_bool.
14267 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
14268 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
14269 * sdbout.c (sdb_debug_hooks): Likewise.
14270 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
14271 (gen_namespace_die): Add DW_AT_export_symbols attribute if
14273 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
14274 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
14275 attribute, don't add anything.
14277 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14279 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
14280 (fold_build2_stat_loc): Likewise.
14281 (fold_build3_stat_loc): Likewise.
14282 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
14283 (fold_build1_loc): Remove macro.
14284 (fold_build2_loc): Likewise.
14285 (fold_build3_loc): Likewise.
14287 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14289 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
14290 (gimple_build_debug_bind_source_stat): Likewise.
14291 * gimple.h (gimple_build_debug_bind): Remove macro.
14292 (gimple_build_debug_bind_source): Likewise.
14294 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14296 * bitmap.c (bitmap_alloc): Adjust.
14297 (bitmap_gc_alloc): Likewise.
14298 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
14300 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14302 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
14303 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
14304 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
14305 (bitmap_gc_alloc_stat): Likewise.
14306 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
14308 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14310 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
14311 * rtl.h (shallow_copy_rtx): Remove macro.
14313 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14315 * emit-rtl.c (gen_raw_REG): Adjust.
14316 * gengenrtl.c (gendef): Likewise.
14317 * rtl.c (rtx_alloc_stat): Remove _stat from name.
14318 * rtl.h (rtx_alloc): Remove macro.
14320 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14322 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
14323 (build_tree_list_stat): Likewise.
14324 * tree.h (build_tree_list): Remove macro.
14325 (build_tree_list_vec): Likewise.
14327 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14329 * tree.c (make_vector_stat): Remove _stat from name.
14330 (build_vector_stat): Likewise.
14331 * tree.h (make_vector_stat): Remove macro.
14332 (build_vector_stat): Likewise.
14334 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14336 * tree.h (build_var_debug_value): Remove prototype.
14338 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14340 * tree.c (tree_cons_stat): Remove _stat from name.
14341 * tree.h (tree_cons): Remove macro.
14343 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14345 * tree.c (build_vl_exp_stat): Remove _stat from name.
14346 * tree.h (build_vl_exp): Remove macro.
14348 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14350 * tree.c (build_decl_stat): Remove _stat from name.
14351 * tree.h (build_decl): Remove macro.
14353 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14355 * gimple.c (gimple_build_with_ops_stat): Adjust.
14356 (gimple_alloc_stat): Remove _stat from name.
14357 * gimple.h (gimple_alloc): Remove macro.
14359 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14361 * tree.c (make_tree_vec_stat): Remove _stat from name.
14362 (grow_tree_vec_stat): Likewise.
14363 * tree.h (make_tree_vec_stat): Adjust prototype.
14364 (grow_tree_vec_stat): Likewise.
14365 (make_tree_vec): Remove macro.
14366 (grow_tree_vec): Likewise.
14368 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14370 * fold-const.c (fold_build1_stat_loc): Adjust.
14371 (fold_build2_stat_loc): Likewise.
14372 (fold_build3_stat_loc): Likewise.
14373 * tree.c (build0_stat): Remove _stat from name.
14374 (build1_stat): Likewise.
14375 (build2_stat): Likewise.
14376 (build3_stat): Likewise.
14377 (build4_stat): Likewise.
14378 (build5_stat): Likewise.
14379 * tree.h (build1_loc): Remove macro, and rename _stat function
14381 (build2_loc): Likewise.
14382 (build3_loc): Likewise.
14383 (build4_loc): Likewise.
14384 (build5_loc): Likewise.
14386 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14388 * tree.c (make_int_cst_stat): Remove _stat from name.
14389 * tree.h (make_int_cst_stat): Adjust prototype.
14390 (make_int_cst): Remove macro.
14392 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14394 * tree.c (make_tre_binfo_stat): Remove _stat from name.
14395 * tree.h (make_tree_binfo_stat): Adjust prototype.
14396 (make_tree_binfo): Remove.
14398 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14400 * tree.c (copy_node_stat): Rename to copy_node.
14401 (build_distinct_type_copy): Adjust.
14402 * tree.h (copy_node_stat): Adjust prototype.
14403 (copy_node): Remove macro.
14405 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14407 * tree.c (make_node_stat): rename to make_node.
14408 (build_tree_list_stat): Adjust.
14409 (build0_stat): Likewise.
14410 (build2_stat): Likewise.
14411 (build3_stat): Likewise.
14412 (build4_stat): Likewise.
14413 (build5_stat): Likewise.
14414 (build_decl_stat): Likewise.
14415 * tree.h (make_node_stat): Adjust prototype.
14416 (make_node): remove macro.
14418 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
14420 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
14421 (PPC_FEATURE2_SCV): Likewise.
14422 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
14424 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14426 * config/aarch64/aarch64.c
14427 (aarch64_internal_mov_immediate): Add new special pattern.
14428 * config/aarch64/aarch64.md (*movdi_aarch64):
14429 Add reg/32bit const mov case.
14431 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14432 Richard Sandiford <richard.sandiford@linaro.org>
14434 * config/aarch64/aarch64.md (mov<mode>): Generalize.
14435 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
14436 Add integer and movi cases.
14437 (movi-split-hf-df-sf split, fp16): New.
14438 (enabled): Added TARGET_FP_F16INST.
14439 * config/aarch64/iterators.md (GPF_HF): New.
14440 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
14442 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14444 * config/aarch64/aarch64.c
14445 (aarch64_simd_container_mode): Add prototype.
14446 (aarch64_expand_mov_immediate): Add HI support.
14447 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
14448 (aarch64_can_const_movi_rtx_p): New.
14449 (aarch64_preferred_reload_class):
14450 Remove restrictions of using FP registers for certain SIMD operations.
14451 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
14452 (aarch64_valid_floating_const): Add integer move validation.
14453 (aarch64_simd_imm_scalar_p): Remove.
14454 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
14455 (aarch64_legitimate_constant_p): Expand list of supported cases.
14456 * config/aarch64/aarch64-protos.h
14457 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
14458 (aarch64_reinterpret_float_as_int): New.
14459 (aarch64_simd_imm_scalar_p): Remove.
14460 * config/aarch64/constraints.md (Uvi): New.
14461 (Dd): Split into Ds and new Dd.
14462 * config/aarch64/aarch64.md (*movsi_aarch64):
14464 (*movdi_aarch64): Add SIMD mov case.
14466 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14468 * tree-predcom.c: (struct chain): Handle store-store chain in which
14469 stores for elimination only store loop invariant values.
14470 (execute_pred_commoning_chain): Ditto.
14471 (prepare_initializers_chain_store_elim): Ditto.
14472 (prepare_finalizers): Ditto.
14473 (is_inv_store_elimination_chain): New function.
14474 (initialize_root_vars_store_elim_1): New function.
14476 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14478 * tree-predcom.c: Revise general description of the pass.
14479 (enum chain_type): New enum type for store elimination.
14480 (struct chain): New field supporting store elimination.
14481 (struct component): Ditto.
14482 (dump_chain): Dump store-stores chain.
14483 (release_chain): Release resources.
14484 (split_data_refs_to_components): Compute and create component
14485 contains only stores for elimination.
14486 (get_chain_last_ref_at): New function.
14487 (make_invariant_chain): Initialization.
14488 (make_rooted_chain): Specify chain type in parameter and record it.
14489 (add_looparound_copies): Skip for store-stores chain.
14490 (determine_roots_comp): Compute type of chain and pass it to
14492 (initialize_root_vars_store_elim_2): New function.
14493 (finalize_eliminated_stores): New function.
14494 (remove_stmt): Handle store for elimination.
14495 (execute_pred_commoning_chain): Execute predictive commoning on
14496 store-store chains.
14497 (determine_unroll_factor): Skip unroll for store-stores chain.
14498 (prepare_initializers_chain_store_elim): New function.
14499 (prepare_initializers_chain): Hanlde store-store chain.
14500 (prepare_finalizers_chain, prepare_finalizers): New function.
14501 (tree_predictive_commoning_loop): Return integer value indicating
14502 if loop is unrolled or lcssa form is corrupted.
14503 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
14505 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14507 * tree-predcom.c (initialize_root): Delete.
14508 (execute_pred_commoning_chain): Initialize root vars and replace
14509 reference of non-combined chain directly, rather than call above
14512 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14514 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
14515 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
14517 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14519 * tree-predcom.c (struct chain): New field init_seq.
14520 (release_chain): Release init_seq.
14521 (prepare_initializers_chain): Record intialization stmts in above
14523 (insert_init_seqs): New function.
14524 (tree_predictive_commoning_loop): Call insert_init_seqs.
14526 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14528 * tree-predcom.c (determine_roots_comp): Skip trivial components.
14530 2017-07-28 Richard Biener <rguenther@suse.de>
14532 * match.pd: Remove superfluous :c.
14533 * genmatch.c (simplify::id): Add member.
14534 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
14536 (current_id): New global.
14537 (dt_node::parent): Move from ...
14538 (dt_operand::parent): ... here. Add for_id member.
14539 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
14540 (decision_tree::find_node): Relax order requirement when
14541 merging DT_TRUE nodes to ones inbetween the current simplify
14542 and the one we try to merge with. Add diagnostic whenever
14543 we need to enforce pattern order by not merging.
14544 (decision_tree::insert): Set current_id.
14545 (decision_tree::print_node): Dump parent node and for_id.
14546 (parser::last_id): Add member.
14547 (parser::push_simplify): Assign unique id.
14548 (parser::parser): Initialize last_id.
14550 2017-07-28 Martin Liska <mliska@suse.cz>
14553 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
14554 gimple_build_debug_bind.
14556 2017-07-28 Richard Biener <rguenther@suse.de>
14558 PR tree-optimization/81502
14559 * match.pd: Add pattern combining BIT_INSERT_EXPR with
14561 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
14563 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
14564 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
14565 for BIT_FIELD_REF args.
14566 * fold-const.c (make_bit_field_ref): Likewise.
14567 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
14569 2017-07-28 Jakub Jelinek <jakub@redhat.com>
14572 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
14573 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
14574 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
14575 Or it into SANITIZER_UNDEFINED.
14576 * ubsan.c: Include gimple-fold.h and varasm.h.
14577 (ubsan_expand_ptr_ifn): New function.
14578 (instrument_pointer_overflow): New function.
14579 (maybe_instrument_pointer_overflow): New function.
14580 (instrument_object_size): Formatting fix.
14581 (pass_ubsan::execute): Call instrument_pointer_overflow
14582 and maybe_instrument_pointer_overflow.
14583 * internal-fn.c (expand_UBSAN_PTR): New function.
14584 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
14585 * sanitizer.def (__ubsan_handle_pointer_overflow,
14586 __ubsan_handle_pointer_overflow_abort): New builtins.
14587 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
14588 * internal-fn.def (UBSAN_PTR): New internal function.
14589 * opts.c (sanitizer_opts): Add pointer-overflow.
14590 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
14591 * fold-const.c (build_range_check): Compute pointer range check in
14592 integral type if pointer arithmetics would be needed. Formatting
14595 2017-07-28 Martin Liska <mliska@suse.cz>
14598 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
14599 parameters that are of a variable-length.
14601 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
14603 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
14605 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
14606 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
14607 (CRT_CALL_STATIC_FUNCTION): Likewise.
14608 (ASM_DEFAULT_SPEC): New define.
14609 (ASM_SPEC32): Likewise.
14610 (ASM_SPEC64): Likewise.
14611 (ASM_SPEC_COMMON): Likewise.
14612 (ASM_SPEC): Likewise.
14613 (INVALID_64BIT): Likewise.
14614 (LINK_OS_DEFAULT_SPEC): Likewise.
14615 (LINK_OS_SPEC32): Likewise.
14616 (LINK_OS_SPEC64): Likewise.
14617 (POWERPC_LINUX): Likewise.
14618 (PTRDIFF_TYPE): Likewise.
14619 (RESTORE_FP_PREFIX): Likewise.
14620 (RESTORE_FP_SUFFIX): Likewise.
14621 (SAVE_FP_PREFIX): Likewise.
14622 (SAVE_FP_SUFFIX): Likewise.
14623 (SIZE_TYPE): Likewise.
14624 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
14625 (TARGET_64BIT): Likewise.
14626 (TARGET_64BIT): Likewise.
14627 (TARGET_AIX): Likewise.
14628 (WCHAR_TYPE_SIZE): Likewise.
14629 (WCHAR_TYPE): Undefine.
14630 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
14631 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
14632 (CPP_OS_RTEMS_SPEC): Delete.
14633 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
14634 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
14636 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
14638 2017-07-28 Jakub Jelinek <jakub@redhat.com>
14640 PR tree-optimization/81578
14641 * tree-parloops.c (build_new_reduction): Bail out if
14642 reduction_code isn't one of the standard OpenMP reductions.
14643 Move the details printing after that decision.
14645 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14647 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
14648 related to reload_in_progress.
14649 (splat_input_operand): Likewise.
14650 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
14652 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
14654 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
14655 (TARGET_INSTANTIATE_DECLS): Likewise.
14656 (legitimate_indexed_address_p): Delete reload_in_progress code.
14657 (rs6000_debug_legitimate_address_p): Likewise.
14658 (rs6000_eliminate_indexed_memrefs): Likewise.
14659 (rs6000_emit_le_vsx_store): Likewise.
14660 (rs6000_emit_move_si_sf_subreg): Likewise.
14661 (rs6000_emit_move): Likewise.
14662 (register_to_reg_type): Likewise.
14663 (rs6000_pre_atomic_barrier): Likewise.
14664 (rs6000_machopic_legitimize_pic_address): Likewise.
14665 (rs6000_allocate_stack_temp): Likewise.
14666 (rs6000_address_for_fpconvert): Likewise.
14667 (rs6000_address_for_altivec): Likewise.
14668 (rs6000_secondary_memory_needed_rtx): Delete function.
14669 (rs6000_check_sdmode): Likewise.
14670 (rs6000_alloc_sdmode_stack_slot): Likewise.
14671 (rs6000_instantiate_decls): Likewise.
14672 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
14673 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
14674 Delete reload_in_progress.
14675 (*vec_reload_and_plus_<mptrsize>): Likewise.
14676 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
14677 (vsx_div_v2di): Likewise.
14678 (vsx_udiv_v2di): Likewise.
14680 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14682 * config/rs6000/rs6000.opt (mlra): Replace with stub.
14683 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
14684 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
14685 (rs6000_debug_reg_global): Delete print of LRA status.
14686 (rs6000_option_override_internal): Delete dead LRA related code.
14687 (rs6000_lra_p): Delete function.
14688 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
14690 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
14692 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
14693 * config/riscv/rtems.h: New file.
14695 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14696 Sudakshina Das <sudi.das@arm.com>
14698 * config/aarch64/aarch64.md
14699 (define_split for and<mode>3nr_compare): Move
14700 non aarch64_logical_operand to a register.
14701 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
14702 register immediate operand to a register.
14703 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
14705 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14707 PR middle-end/81564
14708 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
14710 2017-07-27 Richard Biener <rguenther@suse.de>
14712 PR tree-optimization/81573
14713 PR tree-optimization/81494
14714 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
14715 multi defuse cycle case.
14717 2017-07-27 Richard Biener <rguenther@suse.de>
14719 PR tree-optimization/81571
14720 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
14723 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
14725 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
14726 earlier and only if MASK_FPU is set. Adjust formatting.
14728 2017-07-27 Martin Liska <mliska@suse.cz>
14730 * opt-functions.awk: Add validation of value of Init.
14731 * optc-gen.awk: Pass new argument.
14733 2017-07-27 Martin Liska <mliska@suse.cz>
14735 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
14736 Fix wrong condition.
14738 2017-07-27 Martin Liska <mliska@suse.cz>
14740 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
14741 BBs and edges seen by autoFDO.
14743 2017-07-27 Richard Biener <rguenther@suse.de>
14745 PR tree-optimization/81502
14746 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
14747 with incompatible but same sized type.
14748 (execute_update_addresses_taken): Likewise.
14750 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
14752 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
14753 flag_tree_loop_vectorize rather than flag_tree_vectorize.
14755 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14758 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
14759 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
14760 Change s_operand to memory_operand.
14762 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
14764 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
14765 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
14766 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
14767 Emit instructions rather than returning an expression. Handle TFmode
14768 and KFmode by casting to TImode.
14769 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
14770 (rs6000_emit_le_vsx_store): Likewise.
14771 * config/rs6000/vsx.md (VSX_TI): New iterator.
14772 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
14773 (*vsx_le_undo_permute_<mode>): Likewise.
14774 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
14775 emit the split sequence.
14776 (*vsx_le_perm_store_<mode>): Likewise.
14778 2017-07-27 Jakub Jelinek <jakub@redhat.com>
14780 PR tree-optimization/81555
14781 PR tree-optimization/81556
14782 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
14783 if true, force CHANGED for the recursive invocation.
14784 (reassociate_bb): Remember original length of ops array, pass
14785 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
14787 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
14788 attributes for noipa attribute. For naked attribute use
14789 lookup_attribute first before lookup_attribute_spec.
14790 * final.c (rest_of_handle_final): Disable IPA RA for functions with
14792 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
14793 for functions with noipa attribute.
14794 (cgraph_externally_visible_p): Return true for functions with noipa
14796 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
14797 for functions with noipa attribute.
14798 * doc/extend.texi: Document noipa function attribute.
14799 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
14800 also for functions with noipa attribute.
14801 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
14803 2017-07-26 Andrew Pinski <apinski@cavium.com>
14805 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
14806 vec_unalign_load_cost and vec_unalign_store_cost.
14808 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
14810 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
14811 -mvsx-small-integer option.
14812 (ISA_3_0_MASKS_IEEE): Likewise.
14813 (OTHER_VSX_VECTOR_MASKS): Likewise.
14814 (POWERPC_MASKS): Likewise.
14815 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
14816 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
14817 code, only testing for DImode being allowed in non-VSX floating
14819 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
14820 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
14822 (rs6000_option_override_internal): Delete -mvsx-small-integer.
14823 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
14824 TARGET_P8_VECTOR test.
14825 (rs6000_secondary_reload_simple_move): Likewise.
14826 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
14827 since TARGET_P9_VECTOR was already tested.
14828 (rs6000_opt_masks): Remove -mvsx-small-integer.
14829 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
14830 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
14832 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
14833 test for TARGET_VEXTRACTUB was used, and that uses
14835 (p9 extract splitter): Likewise.
14836 (vsx_extract_<mode>_di_p9): Likewise.
14837 (vsx_extract_<mode>_store_p9): Likewise.
14838 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
14839 for TARGET_P9_VECTOR was used. Delete code that is now dead with
14840 the elimination of TARGET_VSX_SMALL_INTEGER.
14841 (vsx_extract_<mode>_p8): Likewise.
14842 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
14843 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
14844 (vsx_set_<mode>_p9): Likewise.
14845 (vsx_set_v4sf_p9): Likewise.
14846 (vsx_set_v4sf_p9_zero): Likewise.
14847 (vsx_insert_extract_v4sf_p9): Likewise.
14848 (vsx_insert_extract_v4sf_p9_2): Likewise.
14849 * config/rs6000/rs6000.md (sign extend splitter): Change
14850 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
14851 (floatsi<mode>2_lfiwax_mem): Likewise.
14852 (floatunssi<mode>2_lfiwzx_mem): Likewise.
14853 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
14854 since a test for TARGET_P9_VECTOR was used.
14855 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
14856 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
14857 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
14858 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
14859 TARGET_P8_VECTOR test.
14860 (fix_trunc<mode>si2_stfiwx): Likewise.
14861 (fix_trunc<mode>si2_internal): Likewise.
14862 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
14863 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
14865 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
14866 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
14867 TARGET_P8_VECTOR test.
14868 (fixuns_trunc<mode>si2_stfiwx): Likewise.
14869 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
14870 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
14872 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
14873 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
14874 since a test for TARGET_P9_VECTOR was used.
14875 (splitter for loading small constants): Likewise.
14877 2017-07-26 Andrew Pinski <apinski@cavium.com>
14879 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
14882 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
14885 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
14886 (fp_valid_at): Likewise.
14888 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
14890 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
14891 (qdf24xx_addrcost_table): Likewise.
14892 (cortexa57_tunings): Update to use generic_branch_cost.
14893 (cortexa72_tunings): Likewise.
14894 (cortexa73_tunings): Likewise.
14895 (qdf24xx_tunings): Likewise.
14897 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
14899 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
14900 (thunderx2t99_branch_cost): Likewise.
14901 (cortexa35_tunings): Update to use generic_branch_cost.
14902 (cortexa53_tunings): Likewise.
14903 (cortexa57_tunings): Likewise.
14904 (cortexa72_tunings): Likewise.
14905 (cortexa73_tunings): Likewise.
14906 (thunderx2t99_tunings): Likewise.
14908 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
14910 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
14911 (sparc_option_override): Honour MASK_FSMULD.
14912 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
14913 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
14914 * config/sparc/sparc.opt (mfsmuld): New option.
14915 * doc/invoke.texi (mfsmuld): Document option.
14917 2017-07-26 Marek Polacek <polacek@redhat.com>
14919 PR middle-end/70992
14920 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
14922 2017-07-26 Richard Biener <rguenther@suse.de>
14924 * gimple-match-head.c (do_valueize): Return OP if valueize
14926 (get_def): New helper to get at the def stmt of a SSA name
14927 if valueize allows.
14928 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
14929 do_valueize to get at the def stmt.
14930 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
14932 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
14934 PR middle-end/46932
14935 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
14937 2017-07-26 Martin Liska <mliska@suse.cz>
14940 * function.c (expand_function_start): Make expansion of
14941 nonlocal_goto_save_area after parm_birth_insn.
14943 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
14945 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
14946 from all CPU target flags enable members.
14948 2017-07-26 Richard Biener <rguenther@suse.de>
14950 * genmatch.c (dt_simplify::gen): Make iterator vars const.
14951 (decision_tree::gen): Make 'type' const.
14952 (write_predicate): Likewise.
14954 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
14956 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
14957 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
14958 (rs6000_option_override_internal): Likewise.
14959 (rs6000_expand_vector_set): Likewise.
14960 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
14961 (TARGET_UPPER_REGS_SF): Likewise.
14962 (TARGET_UPPER_REGS_DI): Likewise.
14963 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
14964 (TARGET_DIRECT_MOVE_64BIT): Likewise.
14965 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
14966 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
14967 (Splitters for DI constants in Altivec registers): Likewise.
14968 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
14969 (vsx_set_v4sf_p9): Likewise.
14970 (vsx_set_v4sf_p9_zero): Likewise.
14971 (vsx_insert_extract_v4sf_p9): Likewise.
14972 (vsx_insert_extract_v4sf_p9_2): Likewise.
14974 2017-07-25 Carl Love <cel@us.ibm.com>
14976 * doc/extend.texi: Update the built-in documentation file for the
14977 existing built-in functions
14978 vector signed char vec_cnttz (vector signed char);
14979 vector unsigned char vec_cnttz (vector unsigned char);
14980 vector signed short vec_cnttz (vector signed short);
14981 vector unsigned short vec_cnttz (vector unsigned short);
14982 vector signed int vec_cnttz (vector signed int);
14983 vector unsigned int vec_cnttz (vector unsigned int);
14984 vector signed long long vec_cnttz (vector signed long long);
14985 vector unsigned long long vec_cnttz (vector unsigned long long);
14987 2017-07-25 Andrew Pinski <apinski@cavium.com>
14989 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
14990 accesses where the use is for the first operand of a BIT_INSERT.
14992 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
14995 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
14996 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
14998 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
15000 * config/i386/gstabs.h: Delete.
15001 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
15003 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
15005 * config/i386/i386.c (ix86_decompose_address): Do not check for
15006 register RTX when looking at index_reg or base_reg.
15007 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
15009 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
15011 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
15012 to update EH info here.
15014 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15016 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
15018 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15020 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
15022 2017-07-25 Torsten Duwe <duwe@suse.de>
15024 * common.opt: Introduce -fpatchable-function-entry
15025 command line option, and its variables function_entry_patch_area_size
15026 and function_entry_patch_area_start.
15027 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
15028 including a two-value parser.
15029 * target.def (print_patchable_function_entry): New target hook.
15030 * targhooks.h (default_print_patchable_function_entry): New function.
15031 * targhooks.c (default_print_patchable_function_entry): Likewise.
15032 * toplev.c (process_options): Switch off IPA-RA if
15033 patchable function entries are being generated.
15034 * varasm.c (assemble_start_function): Look at the
15035 patchable-function-entry command line switch and current
15036 function attributes and maybe generate NOP instructions by
15037 calling the print_patchable_function_entry hook.
15038 * doc/extend.texi: Document patchable_function_entry attribute.
15039 * doc/invoke.texi: Document -fpatchable_function_entry
15040 command line option.
15041 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
15043 * doc/tm.texi: Re-generate.
15045 2017-07-25 Jakub Jelinek <jakub@redhat.com>
15048 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
15049 TARGET_AVX512DQ rather than TARGET_AVX512BW.
15051 2017-07-25 Tamar Christina <tamar.christina@arm.com>
15053 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
15055 2017-07-25 Richard Biener <rguenther@suse.de>
15057 PR tree-optimization/81455
15058 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
15059 not walk in cycles when looking for guards.
15061 2017-07-25 Richard Biener <rguenther@suse.de>
15063 PR tree-optimization/81529
15064 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
15065 when optimizing backedge uses.
15067 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
15069 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
15071 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
15072 to dl_section_ref. On AIX, append an expression to subtract
15073 the size of the section length to dl_section_ref.
15075 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
15077 * configure.ac: If any of the config.* scripts fail, exit 1.
15078 * configure: Regenerate.
15080 2017-07-25 Richard Biener <rguenther@suse.de>
15082 PR middle-end/81546
15083 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
15084 of immediate uses, be more verbose on errors.
15086 2017-07-25 Richard Biener <rguenther@suse.de>
15088 PR tree-optimization/81510
15089 * tree-vect-loop.c (vect_is_simple_reduction): When the
15090 reduction stmt is not inside the loop bail out.
15092 2017-07-25 Richard Biener <rguenther@suse.de>
15094 PR tree-optimization/81303
15095 * tree-vect-loop-manip.c (vect_loop_versioning): Build
15096 profitability check against LOOP_VINFO_NITERSM1.
15098 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15100 * domwalk.c (cmp_bb_postorder): Simplify.
15101 (sort_bbs_postorder): New function. Use it...
15102 (dom_walker::walk): ...here to optimize common cases.
15104 2017-07-25 Martin Liska <mliska@suse.cz>
15107 * ipa-visibility.c (function_and_variable_visibility): Make the
15108 redirection just on target that supports aliasing.
15109 Fix GNU coding style.
15111 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
15114 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
15116 * config/arm/rtems.h: Likewise.
15117 * config/bfin/rtems.h: Likewise.
15118 * config/i386/rtemself.h: Likewise.
15119 * config/lm32/rtems.h: Likewise.
15120 * config/m32c/rtems.h: Likewise.
15121 * config/m68k/rtemself.h: Likewise.
15122 * config/microblaze/rtems.h: Likewise.
15123 * config/mips/rtems.h: Likewise.
15124 * config/moxie/rtems.h: Likewise.
15125 * config/nios2/rtems.h: Likewise.
15126 * config/powerpcspe/rtems.h: Likewise.
15127 * config/rs6000/rtems.h: Likewise.
15128 * config/rtems.h: Likewise.
15129 * config/sh/rtems.h: Likewise.
15130 * config/sh/rtemself.h: Likewise.
15131 * config/sparc/rtemself.h: Likewise.
15133 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
15136 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
15137 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
15138 * tree-ssa-structalias.c (alias_get_name): Same.
15140 2017-07-25 Bin Cheng <bin.cheng@arm.com>
15143 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
15144 instructions if no du chain is found.
15146 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
15148 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
15150 2017-07-25 Richard Biener <rguenther@suse.de>
15152 PR middle-end/81505
15153 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
15156 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
15158 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
15159 upper-regs options.
15160 (ISA_2_7_MASKS_SERVER): Likewise.
15161 (ISA_3_0_MASKS_IEEE): Likewise.
15162 (OTHER_P8_VECTOR_MASKS): Likewise.
15163 (OTHER_VSX_VECTOR_MASKS): Likewise.
15164 (POWERPC_MASKS): Likewise.
15165 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
15166 duplicate list of options.
15167 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
15168 explicit -mupper-regs options.
15169 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
15170 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
15171 alias for -mupper-regs-df.
15172 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
15173 (rs6000_init_hard_regno_mode_ok): Likewise.
15174 (rs6000_option_override_internal): Likewise.
15175 (rs6000_opt_masks): Likewise.
15176 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
15177 options in terms of whether -mvsx or -mpower8-vector was used.
15178 (TARGET_UPPER_REGS_DI): Likewise.
15179 (TARGET_UPPER_REGS_SF): Likewise.
15180 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
15181 -mupper-regs-* options.
15183 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
15185 * passes.c (emergency_dump_function): Print some empty lines and a
15186 header before the RTL dump.
15188 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
15190 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
15192 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
15195 * config/aarch64/aarch64.c (aarch64_classify_symbol):
15196 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
15198 2017-07-24 Carl Love <cel@us.ibm.com>
15200 * config/rs6000/rs6000-c.c: Add support for built-in functions
15201 vector float vec_extract_fp32_from_shorth (vector unsigned short);
15202 vector float vec_extract_fp32_from_shortl (vector unsigned short);
15203 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
15204 vec_extract_fp_from_shortl): Add defines for the two builtins.
15205 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
15206 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
15208 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
15209 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
15210 * doc/extend.texi: Update the built-in documentation file for the
15211 new built-in function.
15213 2017-07-24 Jakub Jelinek <jakub@redhat.com>
15216 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
15218 * doc/generic.texi: Likewise.
15219 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
15220 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
15222 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
15224 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
15225 (aarch64_mls_elt_merge<mode>): Likewise.
15227 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
15229 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
15230 having __cxa_atexit.
15232 2017-07-23 Michael Collison <michael.collison@arm.com>
15234 * config/arm/arm.c (arm_option_override): Deprecate
15235 use of -mstructure-size-boundary.
15236 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
15237 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
15239 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15242 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
15243 Reduce cost estimate for direct moves.
15245 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
15248 * config/i386/i386.c (ix86_option_override_internal): Disable
15249 BMI, BMI2 and TBM instructions for -m16.
15251 2017-07-21 Carl Love <cel@us.ibm.com>
15253 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15254 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15255 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15256 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15257 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15258 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
15259 VMULOSW): New enum "unspec" values.
15260 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
15261 altivec_vmulosw): New patterns.
15262 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15263 VMULOSW): Add definitions.
15265 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
15267 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
15268 (qdf24xx): Likewise.
15269 * config/aarch64/aarch64-options-extensions.def (rdma); New.
15270 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
15271 (AARCH64_FL_V8_1): Renumber.
15272 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
15273 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
15274 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
15275 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
15276 rdma to feature modifiers list.
15278 2017-07-21 Yury Gribov <tetra2005@gmail.com>
15280 PR middle-end/56727
15281 * ipa-visibility (function_and_variable_visibility): Convert
15282 recursive PLT call to direct call if appropriate.
15284 2017-07-21 Andrew Pinski <apinski@cavium.com>
15286 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
15287 operand 1 to see if the types precision matches.
15288 * fold-const.c (operand_equal_p): Likewise.
15290 2017-07-21 Richard Biener <rguenther@suse.de>
15292 PR tree-optimization/81303
15293 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
15294 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
15295 (vect_peeling_hash_get_lowest_cost): Adjust.
15296 (vect_enhance_data_refs_alignment): Likewise. Use
15297 vect_get_peeling_costs_all_drs to compute the penalty for no
15298 peeling to match up costs.
15300 2017-07-21 Richard Biener <rguenther@suse.de>
15302 PR tree-optimization/81500
15303 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
15304 we didn't identify a reduction path.
15306 2017-07-21 Tom de Vries <tom@codesourcery.com>
15307 Cesar Philippidis <cesar@codesourcery.com>
15309 PR gcov-profile/81442
15310 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
15313 2017-07-21 Tom de Vries <tom@codesourcery.com>
15316 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
15318 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
15319 nvptx_override_options_after_change.
15321 2017-07-21 Ulrich Drepper <drepper@redhat.com>
15323 * dwarf2out.c (output_file_names): Avoid double testing for
15324 dwarf_version >= 5.
15326 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
15328 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
15330 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
15332 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
15334 (try_crossjump_to_edge): Do not punt on partitioned functions.
15336 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
15338 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
15339 Put all BBs reachable only via paths crossing cold region to cold
15341 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
15343 2016-07-21 Richard Biener <rguenther@suse.de>
15345 PR tree-optimization/81303
15346 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
15347 into account prologue and epilogue iterations when raising
15348 min_profitable_iters to sth at least covering one vector iteration.
15350 2017-07-21 Tamar Christina <tamar.christina@arm.com>
15352 * config/arm/arm.c (arm_test_cpu_arch_dat):
15355 2017-07-20 Nathan Sidwell <nathan@acm.org>
15357 Remove TYPE_METHODS.
15358 * tree.h (TYPE_METHODS): Delete.
15359 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
15360 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
15361 (dbxout_type_methods): Scan TYPE_FIELDS.
15362 (dbxout_type): Don't check TYPE_METHODS here.
15363 * function.c (use_register_for_decl): Always ignore register for
15364 class types when not optimizing.
15365 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
15366 * tree.c (free_lang_data_in_type): Stitch out member functions and
15367 templates from TYPE_FIELDS.
15368 (build_distinct_type_copy, verify_type_variant,
15369 verify_type): Member fns are on TYPE_FIELDS.
15370 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
15371 * tree-pretty-print.c (dump_generic_node): Likewise.
15373 2017-07-20 Jakub Jelinek <jakub@redhat.com>
15376 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
15377 V2TImode and V4TImode.
15378 (ix86_expand_vector_extract): Likewise.
15379 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
15380 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
15381 (ssescalarmode): Handle V4TImode and V2TImode.
15382 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
15383 (*vec_extractv2ti, *vec_extractv4ti): New insns.
15384 (VEXTRACTI128_MODE): New mode iterator.
15385 (splitter for *vec_extractv?ti first element): New.
15386 (VEC_INIT_MODE): New mode iterator.
15387 (vec_init<mode>): Consolidate 3 expanders into one using
15388 VEC_INIT_MODE mode iterator.
15390 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
15392 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
15393 non_spilled_static_chain_regno_p.
15395 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
15397 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
15399 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
15401 * bb-reorder.c (connect_traces): Allow copying of blocks within
15404 2017-07-20 Richard Biener <rguenther@suse.de>
15406 * gimple.h (gimple_phi_result): Add gphi * overload.
15407 (gimple_phi_result_ptr): Likewise.
15408 (gimple_phi_arg): Likewise. Adjust index assert to only
15409 allow actual argument accesses rather than all slots available
15411 (gimple_phi_arg_def): Add gphi * overload.
15412 * tree-phinodes.c (make_phi_node): Initialize only actual
15414 (resize_phi_node): Clear memory not covered by old node,
15415 do not initialize excess argument slots.
15416 (reserve_phi_args_for_new_edge): Initialize new argument slot
15419 2017-07-20 Bin Cheng <bin.cheng@arm.com>
15421 PR tree-optimization/81388
15423 2016-07-21 Bin Cheng <bin.cheng@arm.com>
15425 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
15426 by removing computation of may_be_zero.
15428 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15429 Tom de Vries <tom@codesourcery.com>
15431 PR middle-end/81030
15432 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
15433 when gimple level profile disagrees with what RTL expander did.
15435 2017-07-20 Richard Biener <rguenther@suse.de>
15437 PR tree-optimization/61171
15438 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
15439 (vect_analyze_stmt): Add slp instance parameter.
15440 (vectorizable_reduction): Likewise.
15441 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
15442 (vect_is_simple_reduction): Deal with chains not detected
15443 as SLP reduction chain, specifically not properly associated
15444 chains containing a mix of plus/minus.
15445 (get_reduction_op): Remove.
15446 (get_initial_defs_for_reduction): Simplify, pass in whether
15447 this is a reduction chain, pass in the SLP node for the PHIs.
15448 (vect_create_epilog_for_reduction): Get the SLP instance as
15450 (vectorizable_reduction): Get the SLP instance as arg.
15451 During analysis remember the SLP node with the PHIs in the
15452 instance. Simplify getting at the vectorized reduction PHIs.
15453 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
15454 through SLP instance.
15455 (vect_slp_analyze_operations): Likewise.
15456 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
15457 (vect_transform_stmt): Likewise.
15459 2017-07-20 Tom de Vries <tom@codesourcery.com>
15461 PR tree-optimization/81489
15462 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
15463 read of phi arg location to before loop that modifies phi.
15465 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15467 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
15470 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15472 PR middle-end/81331
15473 * except.c (execute): Fix ordering issue.
15475 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
15477 PR rtl-optimization/81423
15478 * combine.c (make_compound_operation_int): Don't try to optimize
15479 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
15481 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
15483 PR rtl-optimization/81423
15484 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
15485 with a constant that is -1 in the truncated to mode.
15487 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15489 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
15490 (determine_unlikely_bbs): ... here.
15491 * predict.h (propagate_unlikely_bbs_forward): Declare.
15492 * cfgexpand.c (pass_expand::execute): Use it.
15493 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
15495 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
15496 propagate_unlikely_bbs_forward.
15498 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15500 PR middle-end/81331
15501 * except.c (maybe_add_nop_after_section_switch): New function.
15504 2017-07-19 Tom de Vries <tom@codesourcery.com>
15506 * gimple.h (gimple_phi_set_arg): Make assert more strict.
15508 2017-07-19 Tom de Vries <tom@codesourcery.com>
15510 * gimple.h (gimple_phi_arg): Make assert more strict.
15512 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
15514 * config.gcc (powerpc*-*-*): Add mmintrin.h.
15515 * config/rs6000/mmintrin.h: New file.
15516 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
15518 2017-07-19 Jakub Jelinek <jakub@redhat.com>
15520 PR tree-optimization/81346
15521 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
15523 2017-07-19 Tom de Vries <tom@codesourcery.com>
15525 * config/nvptx/nvptx.md (VECIM): Add V2DI.
15527 2017-07-19 Tom de Vries <tom@codesourcery.com>
15529 * config/nvptx/nvptx-modes.def: Add V2DImode.
15530 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
15531 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
15532 (nvptx_output_mov_insn): Handle lack of mov.b128.
15533 (nvptx_print_operand): Handle 'H' and 'L' codes.
15534 (nvptx_vector_mode_supported): Allow V2DImode.
15535 (nvptx_preferred_simd_mode): New function.
15536 (nvptx_data_alignment): New function.
15537 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
15538 nvptx_preferred_simd_mode.
15539 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
15541 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
15543 2017-07-19 Tom de Vries <tom@codesourcery.com>
15545 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
15546 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
15547 (nvptx_vector_mode_supported): New function. Allow V2SImode.
15548 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
15549 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
15550 (mov<VECIM>_insn): New define_insn.
15551 (define_expand "mov<VECIM>): New define_expand.
15553 2017-07-19 Tom de Vries <tom@codesourcery.com>
15555 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
15557 2017-07-19 Jakub Jelinek <jakub@redhat.com>
15559 PR tree-optimization/81346
15560 * fold-const.h (fold_div_compare, range_check_type): Declare.
15561 * fold-const.c (range_check_type): New function.
15562 (build_range_check): Use range_check_type.
15563 (fold_div_compare): No longer static, rewritten into
15564 a match.pd helper function.
15565 (fold_comparison): Don't call fold_div_compare here.
15566 * match.pd (X / C1 op C2): New optimization using fold_div_compare
15567 as helper function.
15569 2017-07-19 Nathan Sidwell <nathan@acm.org>
15571 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
15572 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
15573 * tree.c (find_decls_types_r, verify_type): Use
15574 TYPE_{MIN,MAX}_VALUE_RAW.
15575 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
15576 (hash_tree): Likewise.
15577 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
15579 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
15582 2017-07-18 Tom de Vries <tom@codesourcery.com>
15584 PR middle-end/81464
15585 * omp-expand.c (expand_omp_for_static_chunk): Handle
15586 equal-argument loop exit phi.
15588 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
15591 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
15592 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
15593 operand 2 predicate.
15594 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
15595 operand 2 predicate.
15596 (ror,rol -> rorx splitters): Use const_int_operand as
15597 operand 2 predicate.
15599 2017-06-18 Richard Biener <rguenther@suse.de>
15601 PR tree-optimization/81410
15602 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
15603 the gap in the ! slp_perm SLP case after each group.
15605 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15607 PR middle-end/81463
15608 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
15611 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15613 PR middle-end/81462
15614 * predict.c (set_even_probabilities): Cleanup; do not affect
15615 probabilities that are already known.
15616 (combine_predictions_for_bb): Call even when count is set.
15618 2017-07-18 Nathan Sidwell <nathan@acm.org>
15620 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
15623 2017-07-18 Bin Cheng <bin.cheng@arm.com>
15626 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
15627 optimization for loop niter analysis.
15629 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
15632 * config/avr/avr.c (avr_optimize_casesi): Don't use
15633 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
15635 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
15637 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
15638 body_cost_vec from _vect_peel_extended_info.
15639 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
15640 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
15643 2017-07-18 Bin Cheng <bin.cheng@arm.com>
15645 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
15647 2017-07-18 Richard Biener <rguenther@suse.de>
15649 PR tree-optimization/80620
15650 PR tree-optimization/81403
15651 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
15652 info when re-using a VN table entry.
15654 2017-07-18 Richard Biener <rguenther@suse.de>
15656 PR tree-optimization/81418
15657 * tree-vect-loop.c (vectorizable_reduction): Properly compute
15658 vectype_in. Verify that with lane-reducing reduction operations
15659 we have a single def-use cycle.
15661 2017-07-17 Carl Love <cel@us.ibm.com>
15663 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
15665 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15666 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15667 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15668 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15669 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15670 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
15671 VMULOSW): New enum "unspec" values.
15672 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
15673 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
15674 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
15675 altivec_vmulosw): New patterns.
15676 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15677 VMULOSW): Add definitions.
15678 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
15680 * config/alpha/alpha.c: Include predict.h.
15682 2017-07-17 Yury Gribov <tetra2005@gmail.com>
15684 * tree-vrp.c (compare_assert_loc): Fix comparison function
15685 to return predictable results.
15687 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15689 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
15691 (subdi3): Likewise.
15692 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
15693 * doc/invoke.texi (mexpand-adddi): Update text.
15695 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15697 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
15698 that also clobbers the CC register. The old expand code is moved
15700 (*arc_clzsi2): ... here.
15701 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
15702 the CC register. The old expand code is moved to ...
15703 (arc_ctzsi2): ... here.
15705 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15707 * config/arc/arc.opt (mindexed-loads): Use initial value
15708 TARGET_INDEXED_LOADS_DEFAULT.
15709 (mauto-modify-reg): Use initial value
15710 TARGET_AUTO_MODIFY_REG_DEFAULT.
15711 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
15712 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
15713 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
15714 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
15716 2017-07-17 Martin Liska <mliska@suse.cz>
15719 * opts.c (finish_options): Do not allow -fgnu-tm
15720 w/ -fsanitize={kernel-,}address. Say sorry.
15722 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15725 * tree-loop-distribution.c (classify_partition): Only assert on
15726 numer of iterations.
15727 (merge_dep_scc_partitions): Delete prameter. Update function call.
15728 (distribute_loop): Remove code handling loop with unknown niters.
15729 (pass_loop_distribution::execute): Skip loop with unknown niters.
15731 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15734 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
15735 function sort_partitions_by_post_order.
15737 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15739 PR tree-optimization/81374
15740 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
15741 the max index of basic blocks, rather than number of basic blocks.
15743 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15745 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
15747 (arc_legitimate_pic_operand_p): Likewise.
15748 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
15750 (arc_needs_pcl_p): Likewise.
15751 (arc_legitimate_pc_offset_p): Likewise.
15752 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
15753 function is also used in constrains.md.
15754 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
15755 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
15756 PLUS. Only return true/false in known cases, otherwise assert.
15757 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
15758 is already called in arc_legitimate_constant_p.
15759 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
15761 (LEGITIMATE_PIC_OPERAND_P): Use
15762 arc_raw_symbolic_reference_mentioned_p function.
15763 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
15768 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15769 Andrew Burgess <andrew.burgess@embecosm.com>
15771 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
15772 (arc_return_address_register): New function.
15773 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
15774 (arc_handle_fndecl_attribute): Add naked attribute.
15775 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
15776 (TARGET_WARN_FUNC_RETURN): Likewise.
15777 (arc_allocate_stack_slots_for_args): New function.
15778 (arc_warn_func_return): Likewise.
15779 (machine_function): Change type fn_type.
15780 (arc_compute_function_type): Consider new naked function type,
15781 change function return type.
15782 (arc_must_save_register): Adapt to handle new
15783 arc_compute_function_type's return type.
15784 (arc_expand_prologue): Likewise.
15785 (arc_expand_epilogue): Likewise.
15786 (arc_return_address_regs): Delete.
15787 (arc_return_address_register): New function.
15788 (arc_epilogue_uses): Use above function.
15789 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
15790 (arc_function_type): Change encoding, add naked type.
15791 (ARC_INTERRUPT_P): Change to handle the new encoding.
15792 (ARC_FAST_INTERRUPT_P): Likewise.
15793 (ARC_NORMAL_P): Define.
15794 (ARC_NAKED_P): Likewise.
15795 (arc_compute_function_type): Delete prototype.
15796 * config/arc/arc.md (in_ret_delay_slot): Use
15797 arc_return_address_register function.
15798 (simple_return): Likewise.
15799 (p_return_i): Likewise.
15801 2017-07-17 Jakub Jelinek <jakub@redhat.com>
15803 PR tree-optimization/81428
15804 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
15805 can't be built for those types.
15807 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
15809 Remove stuff dead since r239246.
15811 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
15812 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
15813 (avr_inform_devices): Remove dead stuff.
15815 2017-07-17 Tamar Christina <tamar.christina@arm.com>
15817 * config/arm/arm_neon.h: Fix softp typo.
15819 2017-07-17 Jakub Jelinek <jakub@redhat.com>
15821 PR tree-optimization/81365
15822 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
15823 aggregate moves onto bb predecessor edges, make sure there are no
15824 loads that could alias the lhs in between the start of bb and the
15827 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
15830 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
15831 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
15832 [LSHIFTRT, outer_code = TRUNCATE]: Same.
15834 2017-07-17 Jakub Jelinek <jakub@redhat.com>
15836 PR tree-optimization/81396
15837 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
15838 (init_symbolic_number): Initialize it to 1.
15839 (perform_symbolic_merge): Add n_ops from both operands into the new
15841 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
15842 without base_addr as useless if they need more than one operation.
15843 (bswap_replace): Handle !bswap case for NULL base_addr.
15845 2017-07-17 Tom de Vries <tom@codesourcery.com>
15848 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
15851 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
15853 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
15854 conditional builtin define __FIX_LEON3FT_B2BST.
15856 2017-07-17 Daniel Cederman <cederman@gaisler.com>
15858 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
15859 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
15862 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
15864 PR rtl-optimization/81424
15865 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
15866 to remove potential trapping from operands if -fnon-call-exceptions.
15868 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15870 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
15871 profile_proability for scalling.
15872 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
15874 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15876 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
15878 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15880 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
15881 fixpoint arithmetics.
15883 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15885 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
15886 fixpoint arithmetics.
15888 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15890 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
15891 fixpoint arithmetics.
15893 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
15895 * profile-count.h (profile_probability::from_reg_br_prob_note,
15896 profile_probability::to_reg_br_prob_note): New functions.
15897 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
15898 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
15899 * predict.c (probability_reliable_p): Update.
15900 (edge_probability_reliable_p): Update.
15901 (br_prob_note_reliable_p): Update.
15902 (invert_br_probabilities): Update.
15903 (add_reg_br_prob_note): New function.
15904 (combine_predictions_for_insn): Update.
15905 * asan.c (asan_clear_shadow): Update.
15906 * cfgbuild.c (compute_outgoing_frequencies): Update.
15907 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
15908 (update_br_prob_note): Update.
15909 (rtl_verify_edges): Update.
15910 (purge_dead_edges): Update.
15911 (fixup_reorder_chain): Update.
15912 * emit-rtl.c (try_split): Update.
15913 * ifcvt.c (cond_exec_process_insns): Update.
15914 (cond_exec_process_if_block): Update.
15915 (dead_or_predicable): Update.
15916 * internal-fn.c (expand_addsub_overflow): Update.
15917 (expand_neg_overflow): Update.
15918 (expand_mul_overflow): Update.
15919 * loop-doloop.c (doloop_modify): Update.
15920 * loop-unroll.c (compare_and_jump_seq): Update.
15921 * optabs.c (emit_cmp_and_jump_insn_1): Update.
15922 * predict.h: Update.
15923 * reorg.c (mostly_true_jump): Update.
15925 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
15926 * config/alpha/alpha.c (emit_unlikely_jump): Update.
15927 * config/arc/arc.c: (emit_unlikely_jump): Update.
15928 * config/arm/arm.c: (emit_unlikely_jump): Update.
15929 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
15930 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
15931 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
15932 (ix86_print_operand): Update.
15933 (ix86_split_fp_branch): Update.
15934 (predict_jump): Update.
15935 * config/ia64/ia64.c (ia64_print_operand): Update.
15936 * config/mmix/mmix.c (mmix_print_operand): Update.
15937 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
15938 (rs6000_expand_split_stack_prologue): Update.
15939 * config/rs6000/rs6000.c: Update.
15940 * config/s390/s390.c (s390_expand_vec_strlen): Update.
15941 (s390_expand_vec_movstr): Update.
15942 (s390_expand_cs_tdsi): Update.
15943 (s390_expand_split_stack_prologue): Update.
15944 * config/sh/sh.c (sh_print_operand): Update.
15945 (expand_cbranchsi4): Update.
15946 (expand_cbranchdi4): Update.
15947 * config/sparc/sparc.c (output_v9branch): Update.
15948 * config/spu/spu.c (get_branch_target): Update.
15949 (ea_load_store_inline): Update.
15950 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
15951 * config/tilepro/tilepro.c: Update.
15953 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
15955 * gimplify.c (mostly_copy_tree_r): Revert latest change.
15956 (gimplify_save_expr): Likewise.
15958 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15960 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
15962 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15964 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
15966 * timevar.def (TV_IPA_FNSUMMARY): Define.
15968 2017-07-16 Daniel Cederman <cederman@gaisler.com>
15970 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
15971 to back store errata sensitive sequence from being generated.
15972 (sqrtdf2_fix): Likewise.
15974 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15976 * tree-ssa-threadupdate.c (compute_path_counts,
15977 update_joiner_offpath_counts): Use profile_probability.
15979 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
15982 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
15984 * config/arm/arm-c.c (arm_cpu_builtins): Define
15985 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
15987 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
15989 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15990 array entries to represent __ieee128 versions of the
15991 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
15992 scalar_extract_sig, and scalar_insert_exp built-in functions.
15993 (altivec_resolve_overloaded_builtin): Add special case handling
15994 for the __builtin_scalar_insert_exp function, as represented by
15995 the P9V_BUILTIN_VEC_VSIEDP constant.
15996 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
15997 exponent support for __ieee128 argument.
15998 (VSESQP): Add scalar extract signature support for __ieee128
16000 (VSTDCNQP): Add scalar test negative support for __ieee128
16002 (VSIEQP): Add scalar insert exponent support for __int128 argument
16003 with __ieee128 result.
16004 (VSIEQPF): Add scalar insert exponent support for __ieee128
16005 argument with __ieee128 result.
16006 (VSTDCQP): Add scalar test data class support for __ieee128
16008 (VSTDCNQP): Add overload support for scalar test negative with
16009 __ieee128 argument.
16010 (VSTDCQP): Add overload support for scalar test data class
16011 __ieee128 argument.
16012 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
16013 UNSPEC_VSX_SXSIGDP.
16014 (UNSPEC_VSX_SIEXPQP): New constant.
16015 (xsxexpqp): New insn for VSX scalar extract exponent quad
16017 (xsxsigqp): New insn for VSX scalar extract significand quad
16019 (xsiexpqpf): New insn for VSX scalar insert exponent quad
16020 precision with floating point argument.
16021 (xststdcqp): New expand for VSX scalar test data class quad
16023 (xststdcnegqp): New expand for VSX scalar test negative quad
16025 (xststdcqp): New insn to match expansions for VSX scalar test data
16026 class quad precision and VSX scalar test negative quad precision.
16027 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
16028 special case operand checking to enforce that second operand of
16029 VSX scalar test data class with quad precision argument is a 7-bit
16031 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
16032 prototypes and descriptions of __ieee128 versions of
16033 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
16034 scalar_test_data_class, and scalar_test_neg built-in functions.
16036 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16038 PR tree-optimization/81162
16039 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
16040 replace a negate with an add.
16042 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
16044 * doc/invoke.texi (arm/-mcpu): Document +crypto.
16046 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16048 * config/arm/arm-c.c (arm_cpu_builtins): Define
16049 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
16051 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16053 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
16054 (armv8-r): Set ARM Cortex-R52 as default CPU.
16055 * config/arm/arm-tables.opt: Regenerate.
16056 * config/arm/arm-tune.md: Regenerate.
16057 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
16059 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
16060 extension for -mcpu=cortex-r52.
16062 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16064 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
16065 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
16066 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
16067 (fp-armv8): Define it as FP_ARMv8 only.
16068 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
16069 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
16071 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
16072 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
16073 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
16074 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
16075 than TARGET_FPU_ARMV8.
16076 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
16077 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
16078 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
16080 * config/arm/neon.md (neon_vrint): Likewise.
16081 (neon_vcvt): Likewise.
16082 (neon_<fmaxmin_op><mode>): Likewise.
16083 (<fmaxmin><mode>3): Likewise.
16084 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
16085 * config/arm/predicates.md (arm_cond_move_operator): Check against
16086 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
16088 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
16090 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
16091 to top of function.
16093 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16095 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
16096 loop in comment with memset.
16098 2017-07-14 Martin Liska <mliska@suse.cz>
16100 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
16101 * dwarf2out.c (is_java): Remove the function.
16102 (output_pubname): Remove usage of the function.
16103 (lower_bound_default): Remove usage of DW_LANG_Java.
16104 (gen_compile_unit_die): Likewise.
16105 * gcc.c: Remove compiler defaults for .java and .zip files.
16106 * gimple-expr.c (remove_suffix): Change as there's no longer
16107 extension than 4-letter one.
16108 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
16109 (gimplify_save_expr): Likewise.
16110 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
16111 as it's possible even for other languages than Java.
16112 * langhooks.h (struct lang_hooks): Remove Java from a comment.
16113 * lto-opts.c (lto_write_options): Remove reference to Java.
16114 * opts.c (strip_off_ending): Update file extension handling.
16115 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
16116 * tree-eh.c (lower_resx): Likewise.
16117 * tree.c (free_lang_data_in_type): Remove dead code.
16118 (find_decls_types_r): Likewise.
16119 (build_common_builtin_nodes): Remove Java from a comment.
16120 (verify_type): Remove dead code.
16121 * varasm.c (assemble_external): Remove Java from a comment.
16123 2017-07-14 Martin Liska <mliska@suse.cz>
16125 * opts.c (finish_options): Add quotes.
16126 (common_handle_option): Likewise.
16128 2017-07-14 Martin Liska <mliska@suse.cz>
16130 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
16131 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
16132 Remove N_SO_PASCAL.
16133 * dwarf2out.c (lower_bound_default): Do not handle
16135 (gen_compile_unit_die): Likewise.
16136 * gcc.c: Remove default extension binding for GNU Pascal.
16137 * stmt.c: Remove Pascal language from a comment.
16138 * xcoffout.c: Likewise.
16140 2017-07-13 David Malcolm <dmalcolm@redhat.com>
16143 * diagnostic-show-locus.c (fixit_cmp): New function.
16144 (layout::layout): Sort m_fixit_hints.
16145 (column_range::column_range): Assert that the values are valid.
16146 (struct char_span): New struct.
16147 (correction::overwrite): New method.
16148 (struct source_line): New struct.
16149 (line_corrections::add_hint): Add assertions. Reimplement memcpy
16150 calls in terms of classes source_line and char_span, and
16151 correction::overwrite.
16152 (selftest::test_overlapped_fixit_printing_2): New function.
16153 (selftest::diagnostic_show_locus_c_tests): Call it.
16155 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
16157 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
16158 early if there is no lhs.
16160 2017-07-13 Martin Liska <mliska@suse.cz>
16162 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
16163 (gen_reference_type_die): Likewise.
16164 * stor-layout.c: Remove Pascal-related comment.
16166 2017-07-13 Martin Liska <mliska@suse.cz>
16168 * opts.c (finish_options): Add quotes to error messages.
16169 (parse_sanitizer_options): Likewise.
16171 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
16173 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
16175 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
16177 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
16179 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
16181 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
16183 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
16185 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
16188 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
16189 provides the hardware capability bits, define the macro
16190 __BUILTIN_CPU_SUPPORTS__.
16191 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
16192 if GLIBC does not provide the hardware capability bits. Add a
16193 gcc_unreachable call if the built-in cpu function is neither
16194 __builtin_cpu_is nor __builtin_cpu_supports.
16195 (rs6000_get_function_versions_dispatcher): Change the warning
16196 that an old GLIBC is used which does not export the capability
16197 bits to be an error.
16198 * doc/extend.texi (target_clones attribute): Document the
16199 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
16200 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
16201 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
16202 the macros defined by GCC if the newer GLIBC is available.
16204 2017-07-12 Jeff Law <law@redhat.com>
16206 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
16207 remaining includes slightly.
16208 * config/riscv/riscv-builtins.c: Include profile-count.h.
16210 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
16213 * config/avr/avr.c (avr_set_current_function): In diagnostic
16214 messages: Quote keywords and (parts of) identifiers.
16215 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
16218 2017-07-12 Carl Love <cel@us.ibm.com>
16220 * config/rs6000/rs6000-c.c: Add support for built-in functions
16221 vector bool char vec_revb (vector bool char);
16222 vector bool short vec_revb (vector short char);
16223 vector bool int vec_revb (vector bool int);
16224 vector bool long long vec_revb (vector bool long long);
16225 * doc/extend.texi: Update the built-in documentation file for the
16226 new built-in functions.
16228 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16230 * config/s390/s390.md: Remove movcc splitter.
16232 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16234 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
16235 load/store on condition.
16237 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
16240 * config/avr/avr.c (avr_encode_section_info)
16241 [progmem && !TREE_READONLY]: Error if progmem object needs
16244 2017-07-11 Michael Collison <michael.collison@arm.com>
16246 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
16249 2017-07-11 Carl Love <cel@us.ibm.com>
16251 * config/rs6000/rs6000-c.c: Add support for builtins
16252 vector unsigned int vec_parity_lsbb (vector signed int);
16253 vector unsigned int vec_parity_lsbb (vector unsigned int);
16254 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
16255 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
16256 vector unsigned long long vec_parity_lsbb (vector signed long long);
16257 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
16258 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
16259 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
16260 * doc/extend.texi: Update the built-in documentation file for the
16261 new built-in functions.
16263 2017-07-11 David Malcolm <dmalcolm@redhat.com>
16265 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
16266 (layout::m_primary_loc): New field.
16267 (layout::layout): Initialize new field. Move location filtering
16268 logic from here to...
16269 (layout::maybe_add_location_range): ...this new method. Add
16270 support for filtering to just the lines already specified by other
16272 (layout::will_show_line_p): New method.
16273 (gcc_rich_location::add_location_if_nearby): New method.
16274 (selftest::test_add_location_if_nearby): New test function.
16275 (selftest::diagnostic_show_locus_c_tests): Call it.
16276 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
16279 2017-07-11 Tom de Vries <tom@codesourcery.com>
16281 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
16282 (bb_first_real_insn): New function.
16283 (nvptx_single): Add extra initialization of broadcasted condition
16286 2017-07-11 Nathan Sidwell <nathan@acm.org>
16288 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
16290 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
16292 * doc/extend.texi (AVR Function Attributes): Remove weblink to
16293 Binutils doc as TEXI will mess them up.
16294 * doc/invoke.texi (AVR Options): Same here.
16296 2017-07-11 Daniel Cederman <cederman@gaisler.com>
16298 * config/sparc/sparc.opt (mfix-ut700): New option.
16299 (mfix-gr712rc): Likewise.
16300 (sparc_fix_b2bst): New variable.
16301 * doc/invoke.texi (SPARC options): Document them.
16302 (ARM options): Fix warnings.
16303 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
16304 instructions to prevent sequences that can trigger the store-store
16305 errata for certain LEON3FT processors.
16306 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
16307 (sparc_option_override): Set sparc_fix_b2bst appropriately.
16308 * config/sparc/sparc.md (fix_b2bst): New attribute.
16309 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
16311 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
16314 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
16316 (*rsqrtsf2_sse): Ditto.
16318 (div<mode>3): Macroize insn from divdf3 and divsf3
16319 using MODEF mode iterator.
16321 2017-07-10 Martin Sebor <msebor@redhat.com>
16323 PR tree-optimization/80397
16324 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
16325 instead of testing for equality to INTEGER_TYPE.
16327 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
16329 * config.gcc: Remove uclibc from arc target spec.
16331 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
16333 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
16335 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16338 * lto-wrapper.c (remove_option): New function.
16339 (merge_and_complain): Merge PIC/PIE options more realistically.
16341 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
16343 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
16347 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
16348 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
16349 * config.in: Regenerate.
16350 * configure: Regenerate.
16351 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
16352 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
16353 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
16354 (TARGET_GASISR_PROLOGUES): ...target mask.
16355 * common/config/avr/avr-common.c
16356 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
16357 Set -mgas-isr-prologues.
16358 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
16359 INSERT_PASS_BEFORE for it.
16360 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
16361 * config/avr/avr.c (avr_option_override)
16362 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
16363 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
16364 (avr_attribute_table) <no_gccisr>: Add new function attribute.
16365 (avr_set_current_function) <is_no_gccisr>: Init machine field.
16366 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
16368 (make_avr_pass_pre_proep): New function.
16369 (emit_push_sfr) <treg>: Add argument to function and use it
16370 instead of TMP_REG.
16371 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
16372 and set machine->gasisr.yes.
16373 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
16374 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
16375 __gcc_isr.n_pushed to .L__stack_usage.
16376 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
16377 (avr_asm_final_postscan_insn): ...this new static function.
16378 * config/avr/avr.h (machine_function)
16379 <is_no_gccisr, use_L__stack_usage>: New fields.
16380 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
16381 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
16382 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
16383 (gasisr, *gasisr): New expander and insn.
16384 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
16385 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
16386 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
16388 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
16390 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
16393 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
16395 Move jump-tables out of .text again.
16398 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
16399 (ASM_OUTPUT_ADDR_VEC): New function.
16400 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
16401 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
16402 INSN_ADDRESSes as asm comment.
16403 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
16404 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
16405 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
16406 * config/avr/avr.md (*tablejump): Adjust comment.
16407 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
16408 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
16410 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
16411 (avr_output_addr_vec): New proto.
16412 (avr_log_t) <insn_addresses>: New field.
16414 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
16417 * config/i386/i386.c (ix86_function_arg_advance): Set
16418 outgoing_args_on_stack to true if there are outgoing arguments
16420 (ix86_function_arg): Likewise.
16421 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
16422 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
16423 * config/i386/i386.h (machine_function): Add
16424 outgoing_args_on_stack.
16426 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
16428 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
16429 supporting pthreds.
16430 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
16432 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
16434 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
16435 (REAL_H): Remove $(MACHMODE_H).
16436 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
16438 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
16439 $(MACHMODE_H) and double-int.h.
16440 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
16442 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
16445 2017-07-07 Andrew Pinski <apinski@cavium.com>
16447 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
16448 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
16450 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16452 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
16453 Add warning if GCC was not configured to link against a GLIBC that
16454 exports the hardware capability bits.
16455 (make_resolver_func): Make resolver function private and not a
16456 COMDAT function. Create the name with clone_function_name instead
16457 of make_unique_name.
16460 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
16461 correct operand in doing the split.
16463 2017-07-07 Carl Love <cel@us.ibm.com>
16465 * config/rs6000/rs6000-c: Add support for built-in function
16466 vector unsigned short vec_pack_to_short_fp32 (vector float,
16468 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
16469 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
16470 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
16471 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
16472 (convert_4f32_8i16): Add define_expand.
16473 * doc/extend.texi: Update the built-in documentation file for the
16474 new built-in function.
16476 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16478 * config/sparc/m8.md: New file.
16479 * config/sparc/sparc.md: Include m8.md.
16481 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16483 * config/sparc/sparc.opt: New option -mvis4b.
16484 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
16485 (sparc_option_override): Handle VIS4B.
16486 (enum sparc_builtins): Define
16487 SPARC_BUILTIN_DICTUNPACK{8,16,32},
16488 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
16489 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
16490 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
16491 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
16492 (check_constant_argument): New function.
16493 (sparc_vis_init_builtins): Define builtins
16494 __builtin_vis_dictunpack{8,16,32},
16495 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
16496 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
16497 __builtin_vis_fpcmpde{8,16,32}shl and
16498 __builtin_vis_fpcmpur{8,16,32}shl.
16499 (sparc_expand_builtin): Check that the constant operands to
16500 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
16501 constant and in range.
16502 * config/sparc/sparc-c.c (sparc_target_macros): Handle
16504 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
16505 (SPARC_IMM5_P): Likewise.
16506 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
16507 (enabled): Handle vis4b.
16508 (UNSPEC_DICTUNPACK): New unspec.
16509 (UNSPEC_FPCMPSHL): Likewise.
16510 (UNSPEC_FPUCMPSHL): Likewise.
16511 (UNSPEC_FPCMPDESHL): Likewise.
16512 (UNSPEC_FPCMPURSHL): Likewise.
16513 (cpu_feature): New CPU feature `vis4b'.
16514 (dictunpack{8,16,32}): New insns.
16515 (FPCSMODE): New mode iterator.
16516 (fpcscond): New code iterator.
16517 (fpcsucond): Likewise.
16518 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
16519 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
16520 (fpcmpde{8,16,32}{si,di}shl): Likewise.
16521 (fpcmpur{8,16,32}{si,di}shl): Likewise.
16522 * config/sparc/constraints.md: Define constraints `q' for unsigned
16523 2-bit integer constants and `t' for unsigned 5-bit integer
16525 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
16527 (imm5_operand_dictunpack16): Likewise.
16528 (imm5_operand_dictunpack32): Likewise.
16529 (imm2_operand): Likewise.
16530 * doc/invoke.texi (SPARC Options): Document -mvis4b.
16531 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
16532 ditunpack* and fpcmp*shl builtins.
16534 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16536 * config.gcc: Handle m8 in --with-{cpu,tune} options.
16537 * config.in: Add HAVE_AS_SPARC6 define.
16538 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
16540 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
16542 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
16543 (CPP_CPU_SPEC): Handle m8.
16544 (ASM_CPU_SPEC): Likewise.
16545 * config/sparc/sparc-opts.h (enum processor_type): Add
16547 * config/sparc/sparc.c (m8_costs): New struct.
16548 (sparc_option_override): Handle TARGET_CPU_m8.
16549 (sparc32_initialize_trampoline): Likewise.
16550 (sparc64_initialize_trampoline): Likewise.
16551 (sparc_issue_rate): Likewise.
16552 (sparc_register_move_cost): Likewise.
16553 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
16554 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
16555 (ASM_CPU64_DEFAULT_SPEC): Likewise.
16556 (CPP_CPU_SPEC): Handle M8.
16557 (ASM_CPU_SPEC): Likewise.
16558 (AS_M8_FLAG): Define.
16559 * config/sparc/sparc.md: Add m8 to the cpu attribute.
16560 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
16561 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
16563 * configure: Regenerate.
16564 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
16567 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16569 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
16571 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
16572 ("*movdi_insn_sp32"): Do not set v3pipe.
16573 ("*movsi_insn"): Likewise.
16574 ("*movdi_insn_sp64"): Likewise.
16575 ("*movsf_insn"): Likewise.
16576 ("*movdf_insn_sp32"): Likewise.
16577 ("*movdf_insn_sp64"): Likewise.
16578 ("*zero_extendsidi2_insn_sp64"): Likewise.
16579 ("*sign_extendsidi2_insn"): Likewise.
16580 ("*mov<VM32:mode>_insn"): Likewise.
16581 ("*mov<VM64:mode>_insn_sp64"): Likewise.
16582 ("*mov<VM64:mode>_insn_sp32"): Likewise.
16583 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
16584 ("<vlop:code><VL:mode>3"): Likewise.
16585 ("*not_<vlop:code><VL:mode>3"): Likewise.
16586 ("*nand<VL:mode>_vis"): Likewise.
16587 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
16588 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
16589 ("one_cmpl<VL:mode>2"): Likewise.
16590 ("faligndata<VM64:mode>_vis"): Likewise.
16591 ("alignaddrsi_vis"): Likewise.
16592 ("alignaddrdi_vis"): Likweise.
16593 ("alignaddrlsi_vis"): Likewise.
16594 ("alignaddrldi_vis"): Likewise.
16595 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
16596 ("bmaskdi_vis"): Likewise.
16597 ("bmasksi_vis"): Likewise.
16598 ("bshuffle<VM64:mode>_vis"): Likewise.
16599 ("cmask8<P:mode>_vis"): Likewise.
16600 ("cmask16<P:mode>_vis"): Likewise.
16601 ("cmask32<P:mode>_vis"): Likewise.
16602 ("pdistn<P:mode>_vis"): Likewise.
16603 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
16605 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16607 * config/sparc/sparc.md ("subtype"): New insn attribute.
16608 ("*wrgsr_sp64"): Set insn subtype.
16609 ("*rdgsr_sp64"): Likewise.
16610 ("alignaddrsi_vis"): Likewise.
16611 ("alignaddrdi_vis"): Likewise.
16612 ("alignaddrlsi_vis"): Likewise.
16613 ("alignaddrldi_vis"): Likewise.
16614 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
16615 ("fexpand_vis"): Likewise.
16616 ("fpmerge_vis"): Likewise.
16617 ("faligndata<VM64:mode>_vis"): Likewise.
16618 ("bshuffle<VM64:mode>_vis"): Likewise.
16619 ("cmask8<P:mode>_vis"): Likewise.
16620 ("cmask16<P:mode>_vis"): Likewise.
16621 ("cmask32<P:mode>_vis"): Likewise.
16622 ("fchksm16_vis"): Likewise.
16623 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
16624 ("fmean16_vis"): Likewise.
16625 ("fp<plusminus_insn>64_vis"): Likewise.
16626 ("<plusminus_insn>v8qi3"): Likewise.
16627 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
16628 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
16629 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
16630 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
16631 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
16632 ("*movqi_insn"): Likewise.
16633 ("*movhi_insn"): Likewise.
16634 ("*movsi_insn"): Likewise.
16635 ("movsi_pic_gotdata_op"): Likewise.
16636 ("*movdi_insn_sp32"): Likewise.
16637 ("*movdi_insn_sp64"): Likewise.
16638 ("movdi_pic_gotdata_op"): Likewise.
16639 ("*movsf_insn"): Likewise.
16640 ("*movdf_insn_sp32"): Likewise.
16641 ("*movdf_insn_sp64"): Likewise.
16642 ("*zero_extendhisi2_insn"): Likewise.
16643 ("*zero_extendqihi2_insn"): Likewise.
16644 ("*zero_extendqisi2_insn"): Likewise.
16645 ("*zero_extendqidi2_insn"): Likewise.
16646 ("*zero_extendhidi2_insn"): Likewise.
16647 ("*zero_extendsidi2_insn_sp64"): Likewise.
16648 ("ldfsr"): Likewise.
16649 ("prefetch_64"): Likewise.
16650 ("prefetch_32"): Likewise.
16651 ("tie_ld32"): Likewise.
16652 ("tie_ld64"): Likewise.
16653 ("*tldo_ldub_sp32"): Likewise.
16654 ("*tldo_ldub1_sp32"): Likewise.
16655 ("*tldo_ldub2_sp32"): Likewise.
16656 ("*tldo_ldub_sp64"): Likewise.
16657 ("*tldo_ldub1_sp64"): Likewise.
16658 ("*tldo_ldub2_sp64"): Likewise.
16659 ("*tldo_ldub3_sp64"): Likewise.
16660 ("*tldo_lduh_sp32"): Likewise.
16661 ("*tldo_lduh1_sp32"): Likewise.
16662 ("*tldo_lduh_sp64"): Likewise.
16663 ("*tldo_lduh1_sp64"): Likewise.
16664 ("*tldo_lduh2_sp64"): Likewise.
16665 ("*tldo_lduw_sp32"): Likewise.
16666 ("*tldo_lduw_sp64"): Likewise.
16667 ("*tldo_lduw1_sp64"): Likewise.
16668 ("*tldo_ldx_sp64"): Likewise.
16669 ("*mov<VM32:mode>_insn"): Likewise.
16670 ("*mov<VM64:mode>_insn_sp64"): Likewise.
16671 ("*mov<VM64:mode>_insn_sp32"): Likewise.
16673 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16675 * config/sparc/sparc.md ("type"): New insn type viscmp.
16676 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
16678 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
16679 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
16680 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
16681 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
16683 ("n7_vis_logical_11cycle"): Likewise.
16684 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
16685 * config/sparc/niagara2.md ("niag3_vis": Likewise.
16686 * config/sparc/niagara.md ("niag_vis"): Likewise.
16687 * config/sparc/ultra3.md ("us3_fga"): Likewise.
16688 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
16690 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16692 * config/sparc/sparc.md: New instruction type `bmask'.
16693 (bmaskdi_vis): Use the `bmask' type.
16694 (bmasksi_vis): Likewise.
16695 * config/sparc/ultra3.md (us3_array): Likewise.
16696 * config/sparc/niagara7.md (n7_array): Likewise.
16697 * config/sparc/niagara4.md (n4_array): Likewise.
16698 * config/sparc/niagara2.md (niag2_vis): Likewise.
16699 (niag3_vis): Likewise.
16700 * config/sparc/niagara.md (niag_vis): Likewise.
16702 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16704 * ipa-comdats.c: Remove optimize check from gate.
16705 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
16706 for functions not optimized.
16707 (ipa_fn_summary_read): Skip optimize check.
16708 (ipa_fn_summary_write): Likewise.
16709 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
16711 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
16713 (can_inline_edge_p): Check flag_pcc_struct_return for match.
16714 (check_callers): Give up on caller which is not optimized.
16715 (inline_small_functions): Likewise.
16716 (ipa_inline): Do not give up when not optimizing.
16717 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
16718 away unoptimizes cdtors.
16719 (whole_program_function_and_variable_visibility): Do
16720 ipa_discover_readonly_nonaddressable_vars in LTO mode.
16721 * ipa.c (process_references): Do not check optimize.
16722 (symbol_table::remove_unreachable_nodes): Update optimize check.
16723 (set_writeonly_bit): Update optimize check.
16724 (pass_ipa_cdtor_merge::gate): Do not check optimize.
16725 (pass_ipa_single_use::gate): Remove.
16727 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16729 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
16730 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
16731 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
16732 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
16733 permute_load, permute_store, adjust_extract, adjust_splat,
16734 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
16735 replace_swap_with_copy, dump_swap_insn_table,
16736 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
16737 recombine_lvx_pattern, recombine_stvx_pattern,
16738 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
16739 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
16740 to file rs6000-p8swap.c.
16741 * config/rs6000/rs6000-p8swap.c: New file.
16742 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
16743 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
16744 and rs6000*-*-* targets.
16746 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16748 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
16750 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16752 * lto-wrapper.c (merge_and_complain): Do not merge
16753 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
16754 fsigned_zeros, ftrapping_math, fwrapv.
16755 (append_compiler_options): Do not track these options.
16756 (append_linker_options): Likewie
16758 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16760 * cgraphunit.c (cgraph_node::finalize_function): When
16761 !flag_toplevel_reorde set no_reorder flag.
16762 (varpool_node::finalize_decl): Likewise.
16763 (symbol_table::compile): Drop no toplevel reorder path.
16765 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16767 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
16768 edges; zero probability is not better than uninitialized.
16770 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
16772 * asan.h (asan_sanitize_allocas_p): Declare.
16773 * asan.c (asan_sanitize_allocas_p): New function.
16774 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
16775 (handle_builtin_alloca): Likewise.
16776 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
16777 if !asan_sanitize_allocas_p.
16778 * params.def (asan-instrument-allocas): Add new option.
16779 * params.h (ASAN_PROTECT_ALLOCAS): Define.
16780 * opts.c (common_handle_option): Disable allocas sanitization for
16783 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
16785 * asan.c: Include gimple-fold.h.
16786 (get_last_alloca_addr): New function.
16787 (handle_builtin_stackrestore): Likewise.
16788 (handle_builtin_alloca): Likewise.
16789 (asan_emit_allocas_unpoison): Likewise.
16790 (get_mem_refs_of_builtin_call): Add new parameter, remove const
16791 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
16792 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
16793 (instrument_builtin_call): Pass gimple iterator to
16794 get_mem_refs_of_builtin_call.
16795 (last_alloca_addr): New global.
16796 * asan.h (asan_emit_allocas_unpoison): Declare.
16797 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
16798 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
16799 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
16800 if function calls alloca.
16801 * gimple-fold.c (replace_call_with_value): Remove static keyword.
16802 * gimple-fold.h (replace_call_with_value): Declare.
16803 * internal-fn.c: Include asan.h.
16804 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
16805 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
16807 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16809 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
16810 (C_SELFTEST_FLAGS): New.
16811 (CPP_SELFTEST_FLAGS): New.
16812 (SELFTEST_DEPS): New, from deps of s-selftest.
16813 (C_SELFTEST_DEPS): New, from deps of s-selftest.
16814 (CPP_SELFTEST_DEPS): New.
16815 (selftest): Add dependency on s-selftest-c++.
16816 (s-selftest): Rename to...
16817 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
16818 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
16819 than SELFTEST_FLAGS.
16820 (selftest-gdb): Rename to...
16821 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
16823 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
16824 (selftest-valgrind): Rename to...
16825 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
16827 (selftest-valgrind): Reintroduce as an alias for
16828 selftest-c-valgrind.
16829 (s-selftest-c++): New.
16830 (selftest-c++-gdb): New.
16831 (selftest-c++-valgrind): New.
16833 2017-07-06 Olivier Hainque <hainque@adacore.com>
16835 * gcc.c (process_command): When deciding if undefined variables
16836 should be ignored when processing specs, accept "gcc -v" as well.
16838 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16840 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
16841 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
16843 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
16845 * config/arm/arm-cpus.in (armv8-r): Add new entry.
16846 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
16847 * config/arm/arm-tables.opt: Regenerate.
16848 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
16850 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
16852 2017-07-06 Carl Love <cel@us.ibm.com>
16854 * ChangeLog: Clean up from mid air collision
16856 2017-07-06 Carl Love <cel@us.ibm.com>
16858 * config/rs6000/rs6000-c.c: Add support for built-in functions
16859 vector signed int vec_subc (vector signed int, vector signed int);
16860 vector signed __int128 vec_subc (vector signed __int128,
16861 vector signed __int128);
16862 vector unsigned __int128 vec_subc (vector unsigned __int128,
16863 vector unsigned __int128);
16864 vector signed int vec_sube (vector signed int, vector signed int,
16865 vector signed int);
16866 vector unsigned int vec_sube (vector unsigned int,
16867 vector unsigned int,
16868 vector unsigned int);
16869 vector signed __int128 vec_sube (vector signed __int128,
16870 vector signed __int128,
16871 vector signed__int128);
16872 vector unsigned __int128 vec_sube (vector unsigned __int128,
16873 vector unsigned __int128,
16874 vector unsigned __int128);
16875 vector signed int vec_subec (vector signed int, vector signed int,
16876 vector signed int);
16877 vector unsigned int vec_subec (vector unsigned int,
16878 vector unsigned int,
16879 vector unsigned int);
16880 vector signed __int128 vec_subec (vector signed __int128,
16881 vector signed __int128,
16882 vector signed__int128);
16883 vector unsigned __int128 vec_subec (vector unsigned __int128,
16884 vector unsigned __int128,
16885 vector unsigned __int128);
16886 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
16887 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
16888 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
16889 BU_ALTIVEC_OVERLOAD_X definitions.
16890 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
16891 * doc/extend.texi: Update the built-in documentation file for the new
16892 built-in functions.
16894 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16897 * diagnostic-show-locus.c (layout::layout): Use start and finish
16898 spelling location for the start and finish of each range.
16899 * genmatch.c (linemap_client_expand_location_to_spelling_point):
16900 Add unused aspect param.
16901 * input.c (expand_location_1): Add "aspect" param, and use it
16902 to access the correct part of the location.
16903 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
16905 (expand_location_to_spelling_point): Likewise.
16906 (linemap_client_expand_location_to_spelling_point): Add "aspect"
16907 param, and pass it to expand_location_1.
16909 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
16911 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
16912 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
16913 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
16914 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
16915 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
16916 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
16917 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
16918 _mm_maskz_getmant_ss): New intrinsics.
16919 (__builtin_ia32_getexpss128_mask): Changed to ...
16920 __builtin_ia32_getexpss128_round ... this.
16921 (__builtin_ia32_getexpsd128_mask): Changed to ...
16922 __builtin_ia32_getexpsd128_round ... this.
16923 * config/i386/i386-builtin-types.def
16924 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
16925 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
16926 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
16927 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
16928 __builtin_ia32_getmantss_mask_round): New builtins.
16929 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
16930 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
16931 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
16932 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
16933 * config/i386/sse.md
16934 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
16935 avx512f_sgetexp<mode><mask_scalar_name>
16936 <round_saeonly_scalar_name> ... this.
16937 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
16938 %0, %1, %2<round_saeonly_op3>}): Changed to ...
16939 vgetexp<ssescalarmodesuffix>
16940 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16941 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
16942 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
16943 avx512f_vgetmant<mode><mask_scalar_name>
16944 <round_saeonly_scalar_name> ... this.
16945 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
16946 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
16947 vgetmant<ssescalarmodesuffix>
16948 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
16949 %0<mask_scalar_operand4>, %1, %2
16950 <round_saeonly_scalar_mask_op4>, %3} ... this.
16951 * config/i386/subst.md (mask_scalar_operand4,
16952 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
16953 round_saeonly_scalar_nimm_predicate): New subst attributes.
16955 2017-07-06 Julia Koval <julia.koval@intel.com>
16957 * config/i386/i386.c (ix86_erase_embedded_rounding):
16958 Remove code for old rounding pattern.
16960 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
16962 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
16964 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
16966 * doc/sourcebuild.texi (Test Directives, Variants of
16967 dg-require-support): Add documentation for dg-require-stack-check.
16969 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
16971 * config/i386/subst.md (mask_scalar, round_scalar,
16972 round_saeonly_scalar): New meta-templates.
16973 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
16974 round_scalar_mask_operand3, round_scalar_mask_op3,
16975 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
16976 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
16977 round_saeonly_scalar_constraint,
16978 round_saeonly_scalar_prefix): New subst attribute.
16979 * config/i386/sse.md
16980 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
16981 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
16982 <round_scalar_name> ... this.
16983 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
16984 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
16985 <round_scalar_name> ... this.
16986 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
16987 <sse>_vm<code><mode>3<mask_scalar_name>
16988 <round_saeonly_scalar_name> ... this.
16989 (v<plusminus_mnemonic><ssescalarmodesuffix>
16990 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
16991 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
16992 v<plusminus_mnemonic><ssescalarmodesuffix>
16993 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16994 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
16995 (v<multdiv_mnemonic><ssescalarmodesuffix>
16996 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
16997 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
16998 v<multdiv_mnemonic><ssescalarmodesuffix>
16999 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17000 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
17001 (v<maxmin_float><ssescalarmodesuffix>
17002 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
17003 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
17004 v<maxmin_float><ssescalarmodesuffix>
17005 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17006 %0<mask_scalar_operand3>, %1, %<iptr>2
17007 <round_saeonly_scalar_mask_op3>} ... this.
17009 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
17011 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
17012 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
17014 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
17015 Alan Hayward <alan.hayward@arm.com>
17016 David Sherwood <david.sherwood@arm.com>
17018 * combine.c (simplify_if_then_else): Remove "enum" before
17020 * compare-elim.c (can_eliminate_compare): Likewise.
17021 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
17023 (aarch64_lookup_simd_builtin_type): Likewise.
17024 (aarch64_simd_builtin_type): Likewise.
17025 (aarch64_init_simd_builtin_types): Likewise.
17026 (aarch64_simd_expand_args): Likewise.
17027 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
17029 (aarch64_reverse_mask): Likewise.
17030 (aarch64_simd_emit_reg_reg_move): Likewise.
17031 (aarch64_gen_adjusted_ldpstp): Likewise.
17032 (aarch64_ccmp_mode_to_code): Likewise.
17033 (aarch64_operands_ok_for_ldpstp): Likewise.
17034 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17035 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
17037 (aarch64_min_divisions_for_recip_mul): Likewise.
17038 (aarch64_reassociation_width): Likewise.
17039 (aarch64_get_condition_code_1): Likewise.
17040 (aarch64_simd_emit_reg_reg_move): Likewise.
17041 (aarch64_simd_attr_length_rglist): Likewise.
17042 (aarch64_reverse_mask): Likewise.
17043 (aarch64_operands_ok_for_ldpstp): Likewise.
17044 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17045 (aarch64_gen_adjusted_ldpstp): Likewise.
17046 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
17048 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
17049 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
17050 (arm_lookup_simd_builtin_type): Likewise.
17051 (arm_simd_builtin_type): Likewise.
17052 (arm_init_simd_builtin_types): Likewise.
17053 (arm_expand_builtin_args): Likewise.
17054 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
17055 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
17056 (ft32_setup_incoming_varargs): Likewise.
17057 (ft32_function_arg): Likewise.
17058 (ft32_function_arg_advance): Likewise.
17059 (ft32_pass_by_reference): Likewise.
17060 (ft32_arg_partial_bytes): Likewise.
17061 (ft32_valid_pointer_mode): Likewise.
17062 (ft32_addr_space_pointer_mode): Likewise.
17063 (ft32_addr_space_legitimate_address_p): Likewise.
17064 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
17066 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
17067 (ix86_emit_outlined_ms2sysv_restore): Likewise.
17068 (iamcu_alignment): Likewise.
17069 (canonicalize_vector_int_perm): Likewise.
17070 (ix86_noce_conversion_profitable_p): Likewise.
17071 (ix86_mpx_bound_mode): Likewise.
17072 (ix86_operands_ok_for_move_multiple): Likewise.
17073 * config/microblaze/microblaze-protos.h
17074 (microblaze_expand_conditional_branch_reg): Likewise.
17075 * config/microblaze/microblaze.c
17076 (microblaze_expand_conditional_branch_reg): Likewise.
17077 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
17079 (rs6000_reassociation_width): Likewise.
17080 (rs6000_invalid_binary_op): Likewise.
17081 (fusion_p9_p): Likewise.
17082 (emit_fusion_p9_load): Likewise.
17083 (emit_fusion_p9_store): Likewise.
17084 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
17086 (riscv_hard_regno_mode_ok_p): Likewise.
17087 (riscv_address_insns): Likewise.
17088 (riscv_split_symbol): Likewise.
17089 (riscv_legitimize_move): Likewise.
17090 (riscv_function_value): Likewise.
17091 (riscv_hard_regno_nregs): Likewise.
17092 (riscv_expand_builtin): Likewise.
17093 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
17094 (riscv_build_integer): Likewise.
17095 (riscv_split_integer): Likewise.
17096 (riscv_legitimate_constant_p): Likewise.
17097 (riscv_cannot_force_const_mem): Likewise.
17098 (riscv_regno_mode_ok_for_base_p): Likewise.
17099 (riscv_valid_base_register_p): Likewise.
17100 (riscv_valid_offset_p): Likewise.
17101 (riscv_valid_lo_sum_p): Likewise.
17102 (riscv_classify_address): Likewise.
17103 (riscv_legitimate_address_p): Likewise.
17104 (riscv_address_insns): Likewise.
17105 (riscv_load_store_insns): Likewise.
17106 (riscv_force_binary): Likewise.
17107 (riscv_split_symbol): Likewise.
17108 (riscv_force_address): Likewise.
17109 (riscv_legitimize_address): Likewise.
17110 (riscv_move_integer): Likewise.
17111 (riscv_legitimize_const_move): Likewise.
17112 (riscv_legitimize_move): Likewise.
17113 (riscv_address_cost): Likewise.
17114 (riscv_subword): Likewise.
17115 (riscv_output_move): Likewise.
17116 (riscv_canonicalize_int_order_test): Likewise.
17117 (riscv_emit_int_order_test): Likewise.
17118 (riscv_function_arg_boundary): Likewise.
17119 (riscv_pass_mode_in_fpr_p): Likewise.
17120 (riscv_pass_fpr_single): Likewise.
17121 (riscv_pass_fpr_pair): Likewise.
17122 (riscv_get_arg_info): Likewise.
17123 (riscv_function_arg): Likewise.
17124 (riscv_function_arg_advance): Likewise.
17125 (riscv_arg_partial_bytes): Likewise.
17126 (riscv_function_value): Likewise.
17127 (riscv_pass_by_reference): Likewise.
17128 (riscv_setup_incoming_varargs): Likewise.
17129 (riscv_print_operand): Likewise.
17130 (riscv_elf_select_rtx_section): Likewise.
17131 (riscv_save_restore_reg): Likewise.
17132 (riscv_for_each_saved_reg): Likewise.
17133 (riscv_register_move_cost): Likewise.
17134 (riscv_hard_regno_mode_ok_p): Likewise.
17135 (riscv_hard_regno_nregs): Likewise.
17136 (riscv_class_max_nregs): Likewise.
17137 (riscv_memory_move_cost): Likewise.
17138 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
17139 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
17140 (rl78_addr_space_address_mode): Likewise.
17141 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17143 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17144 (rs6000_reassociation_width): Likewise.
17145 (rs6000_invalid_binary_op): Likewise.
17146 (fusion_p9_p): Likewise.
17147 (emit_fusion_p9_load): Likewise.
17148 (emit_fusion_p9_store): Likewise.
17149 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
17150 (ok_for_simple_move_operands): Likewise.
17151 (ok_for_simple_move_strict_operands): Likewise.
17152 (ok_for_simple_arith_logic_operands): Likewise.
17153 (visium_legitimize_reload_address): Likewise.
17154 (visium_select_cc_mode): Likewise.
17155 (output_cbranch): Likewise.
17156 (visium_split_double_move): Likewise.
17157 (visium_expand_copysign): Likewise.
17158 (visium_expand_int_cstore): Likewise.
17159 (visium_expand_fp_cstore): Likewise.
17160 * config/visium/visium.c (visium_pass_by_reference): Likewise.
17161 (visium_function_arg): Likewise.
17162 (visium_function_arg_advance): Likewise.
17163 (visium_libcall_value): Likewise.
17164 (visium_setup_incoming_varargs): Likewise.
17165 (visium_legitimate_constant_p): Likewise.
17166 (visium_legitimate_address_p): Likewise.
17167 (visium_legitimize_address): Likewise.
17168 (visium_secondary_reload): Likewise.
17169 (visium_register_move_cost): Likewise.
17170 (visium_memory_move_cost): Likewise.
17171 (prepare_move_operands): Likewise.
17172 (ok_for_simple_move_operands): Likewise.
17173 (ok_for_simple_move_strict_operands): Likewise.
17174 (ok_for_simple_arith_logic_operands): Likewise.
17175 (visium_function_value_1): Likewise.
17176 (rtx_ok_for_offset_p): Likewise.
17177 (visium_legitimize_reload_address): Likewise.
17178 (visium_split_double_move): Likewise.
17179 (visium_expand_copysign): Likewise.
17180 (visium_expand_int_cstore): Likewise.
17181 (visium_expand_fp_cstore): Likewise.
17182 (visium_split_cstore): Likewise.
17183 (visium_select_cc_mode): Likewise.
17184 (visium_split_cbranch): Likewise.
17185 (output_cbranch): Likewise.
17186 (visium_print_operand_address): Likewise.
17187 * expmed.c (flip_storage_order): Likewise.
17188 * expmed.h (emit_cstore): Likewise.
17189 (flip_storage_order): Likewise.
17190 * genrecog.c (validate_pattern): Likewise.
17191 * hsa-gen.c (gen_hsa_addr): Likewise.
17192 * internal-fn.c (expand_arith_overflow): Likewise.
17193 * ira-color.c (allocno_copy_cost_saving): Likewise.
17194 * lra-assigns.c (find_hard_regno_for_1): Likewise.
17195 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
17196 (process_invariant_for_inheritance): Likewise.
17197 * lra-eliminations.c (move_plus_up): Likewise.
17198 * omp-low.c (lower_oacc_reductions): Likewise.
17199 * simplify-rtx.c (simplify_subreg): Likewise.
17200 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
17201 (TARGET_CHKP_BOUND_MODE): Likewise..
17202 * targhooks.c (default_chkp_bound_mode): Likewise.
17203 (default_setup_incoming_vararg_bounds): Likewise.
17204 * targhooks.h (default_chkp_bound_mode): Likewise.
17205 (default_setup_incoming_vararg_bounds): Likewise.
17206 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
17207 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
17208 (have_whole_vector_shift): Likewise.
17209 * tree-vect-stmts.c (vectorizable_load): Likewise.
17210 * doc/tm.texi: Regenerate.
17212 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
17214 Graceful degrade if Binutils PR21472 is not available.
17217 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
17218 .rodata in flash test fails.
17219 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
17220 * confgure: Regenerate.
17221 * config.in: Regenerate.
17222 * config/avr/avr.c (avr_asm_named_section)
17223 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
17224 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
17225 (avr_asm_init_sections): Same.
17227 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17229 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
17230 (fma<VH:mode>4_intrinsic): Likewise.
17231 (*fmsub<VCVTF:mode>4): Likewise.
17232 (*fmsub<VH:mode>4_intrinsic): Likewise.
17234 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
17237 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
17238 Don't depend on "optimize > 0".
17239 (out_movhi_r_mr, out_movqi_mr_r): Same.
17240 (out_movhi_mr_r, out_movqi_r_mr): Same.
17241 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
17242 io_address_operand on "optimize > 0".
17244 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17246 * tree-loop-distribution.c: Add general explanantion on the pass.
17247 (generate_loops_for_partition): Mark distributed loop.
17248 (pg_add_dependence_edges): New parameter. Handle alias data
17249 dependence specially and record it in the parameter if asked.
17250 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
17251 (init_partition_graph_vertices, add_partition_graph_edge): New.
17252 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
17253 (free_partition_graph_vdata, build_partition_graph): New.
17254 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
17255 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
17256 (data_ref_segment_size, latch_dominated_by_data_ref): New.
17257 (compute_alias_check_pairs, version_loop_by_alias_check): New.
17258 (version_for_distribution_p, finalize_partitions): New.
17259 (distribute_loop): Handle alias data dependence specially. Factor
17260 out loop fusion code as functions and call these functions.
17262 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17264 * tree-loop-distribution.c (classify_partition): New parameter and
17265 better handle reduction statement.
17266 (rdg_build_partitions): Revise comment.
17267 (distribute_loop): Compute statements in all partitions and pass it
17268 to classify_partition.
17270 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17272 * tree-loop-distribution.c (enum partition_type): New.
17273 (struct partition): New field type.
17274 (partition_merge_into): Add parameter. Update partition type.
17275 (data_dep_in_cycle_p, update_type_for_merge): New functions.
17276 (build_rdg_partition_for_vertex): Compute partition type.
17277 (rdg_build_partitions): Dump partition type.
17278 (distribute_loop): Update calls to partition_merge_into.
17280 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17282 * tree-loop-distribution.c (struct ddr_hasher): New.
17283 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
17285 (classify_partition): Call get_data_dependence.
17286 (pg_add_dependence_edges): Ditto.
17287 (distribute_loop): Release data dependence hash table.
17289 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17291 * tree-loop-distribution.c (ref_base_address): Delete.
17292 (similar_memory_accesses): Rename ...
17293 (share_memory_accesses): ... to this. Check if partitions access
17294 the same memory reference.
17295 (distribute_loop): Call share_memory_accesses.
17297 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17299 * tree-loop-distribution.c (struct partition): New field recording
17300 its data reference.
17301 (partition_alloc, partition_free): Init and release data refs.
17302 (partition_merge_into): Merge data refs.
17303 (build_rdg_partition_for_vertex): Collect data refs for partition.
17304 (pg_add_dependence_edges): Change parameters from vector to bitmap.
17306 (distribute_loop): Remve data refs from vertice data of partition
17309 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17311 * tree-loop-distribution.c (params.h): Include header file.
17312 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
17313 (datarefs_vec): New global var.
17314 (create_rdg_vertices): Use datarefs_vec directly.
17315 (free_rdg): Don't free data references.
17316 (build_rdg): Update use. Don't free data references.
17317 (distribute_loop): Compute global variable for data references.
17318 Bail out if there are too many data references.
17320 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17322 * tree-loop-distribution.c (loop_nest): New global var.
17323 (build_rdg): Use loop directly, rather than loop nest.
17324 (pg_add_dependence_edges): Remove loop nest parameter. Use global
17326 (distribute_loop): Compute global variable loop nest. Update use.
17328 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17330 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
17331 (partition_merge_into): New parameter. Dump reason for fusion.
17332 (distribute_loop): Update use of partition_merge_into.
17334 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17336 * tree-loop-distribution.c (bb_top_order_index): New.
17337 (bb_top_order_index_size, bb_top_order_cmp): New.
17338 (stmts_from_loop): Use topological order.
17339 (pass_loop_distribution::execute): Compute and release topological
17340 order for basic blocks.
17342 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17344 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
17347 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17349 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
17350 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
17351 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
17352 * internal-fn.def (LOOP_DIST_ALIAS): New.
17353 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
17354 (fold_loop_internal_call): ... this.
17355 (vect_loop_dist_alias_call): New function.
17356 (set_uid_loop_bbs): Call fold_loop_internal_call.
17357 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
17360 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
17363 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
17364 Require dead FLAGS_REG at the beginning of a peephole.
17366 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
17369 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
17370 arguments in the call to __builtin_ia32_sbb_u32.
17371 (_subborrow_u64): Swap _X and _Y arguments in the call to
17372 __builtin_ia32_sbb_u64.
17374 2017-07-04 Jakub Jelinek <jakub@redhat.com>
17377 * tree-vrp.c (compare_assert_loc): Turn into a function template
17378 with stable template parameter. Only test if a->e is NULL,
17379 !a->e == !b->e has been verified already. Use e == NULL or
17380 e != NULL instead of e or ! e tests. If stable is true, don't use
17381 iterative_hash_expr, on the other side allow a or b or both NULL
17382 and sort the NULLs last.
17383 (process_assert_insertions): Sort using compare_assert_loc<false>
17384 instead of compare_assert_loc, later sort using
17385 compare_assert_loc<true> before calling process_assert_insertions_for
17386 in a loop. Use break instead of continue once seen NULL pointer.
17388 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
17390 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
17391 Cortex-R7 and Cortex-R8 processors.
17393 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17395 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
17396 uninitialized while src is not.
17398 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
17400 * common/config/arm/arm-common.c: Adjust include path for
17402 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
17403 (arm-cpu.h): Create in build directory. Adjust dependency rules.
17404 (arm-cpu-data.h): Likewise.
17405 (arm-cpu-cdata.h): Likewise.
17406 * config/arm/arm-cpu.h: Delete.
17407 * config/arm/arm-cpu-cdata.h: Delete.
17408 * config/arm/arm-cpu-data.h: Delete.
17410 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
17412 * config/arm/arm-cpus.in (cortex-a55): New.
17413 (cortex-a75): Likewise.
17414 (cortex-a75.cortex-a55): Likewise.
17415 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
17417 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
17418 * config/arm/arm-cpu-cdata.h: Regenerate.
17419 * config/arm/arm-cpu-data.h: Regenerate.
17420 * config/arm/arm-cpu.h: Regenerate.
17421 * config/arm/arm-tables.opt: Regenerate.
17422 * config/arm/arm-tune.md: Regenerate.
17424 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17426 * haifa-sched.c (sched_create_recovery_edges): Update profile.
17428 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17430 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
17433 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
17435 PR tree-optimization/81292
17436 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
17437 full_string_p, also call adjust_related_strinfos if the adjustment
17438 is simple, otherwise invalidate related strinfos.
17440 2017-07-04 Martin Liska <mliska@suse.cz>
17443 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
17444 newly created variable as DECL_IGNORED_P.
17446 2017-07-04 Martin Liska <mliska@suse.cz>
17449 * ipa-inline.c (inline_small_functions):
17450 Use xstrdup_for_dump.
17452 2017-07-04 Tom de Vries <tom@codesourcery.com>
17454 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
17456 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
17459 * config/darwin.c (darwin_function_switched_text_sections):
17462 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
17464 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
17466 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
17468 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
17470 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17472 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
17473 min_profitable_iters, and th as inclusive lower bounds.
17474 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
17475 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
17476 for min_profitable_iters and min_profitable_estimate.
17477 (vect_transform_loop): Treat th as an inclusive lower bound.
17478 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
17480 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
17483 * config/darwin.c (darwin_function_switched_text_sections):
17484 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
17485 in two pieces, and suppress the use of buf.
17487 2017-07-03 Nathan Sidwell <nathan@acm.org>
17489 * hash-table.h (hash_table_mod1): Fix indentation.
17491 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17493 PR middle-end/81290
17494 * predict.c (force_edge_cold): Be more careful about propagation
17496 * profile-count.h (profile_probability::guessed,
17497 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
17499 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
17501 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
17503 * doc/invoke.texi (rcpc architecture extension): Document it.
17505 2017-07-03 Richard Biener <rguenther@suse.de>
17507 PR tree-optimization/60510
17508 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
17509 the scalar reduction PHI and use it.
17510 (vectorizable_reduction): Properly guard the single_defuse_cycle
17511 path for non-SLP reduction chains where we cannot use it.
17512 Rework reduc_def/index and vector type deduction. Rework
17513 vector operand gathering during reduction op code-gen.
17514 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
17515 chains dissolve the chain and leave it to non-SLP reduction
17518 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17520 * tree-data-ref.h (dr_alignment): Declare.
17521 * tree-data-ref.c (dr_alignment): New function.
17522 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
17523 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
17525 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
17527 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17529 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
17530 and base_misalignment fields.
17531 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
17532 * tree-data-ref.c: Include builtins.h.
17533 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
17534 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
17535 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
17536 * tree-vect-data-refs.c: Include tree-cfg.h.
17537 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
17538 fields instead of calculating an alignment here.
17539 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
17540 innermost_loop_behavior fields.
17542 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17544 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
17546 (DR_STEP_ALIGNMENT): New macro.
17547 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
17548 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
17549 (create_data_ref): Print it.
17550 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
17551 to tell whether the step preserves vector (mis)alignment.
17552 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
17553 Move the check for an integer step and generalise to all INTEGER_CST.
17554 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
17555 Print the outer step alignment.
17557 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17559 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
17560 with offset_alignment.
17561 (DR_ALIGNED_TO): Delete.
17562 (DR_OFFSET_ALIGNMENT): New macro.
17563 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
17564 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
17565 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
17566 (create_data_ref): Likewise.
17567 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
17568 (vect_analyze_data_refs): Likewise.
17569 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
17570 creating dummy innermost behavior.
17572 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17574 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
17575 with a "innermost_loop_behavior *" and refeence tree.
17576 * tree-data-ref.c (dr_analyze_innermost): Likewise.
17577 (create_data_ref): Update call accordingly.
17578 * tree-predcom.c (find_looparound_phi): Likewise.
17580 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17582 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
17583 fields with dr_wrt_vec_loop.
17584 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
17585 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
17586 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
17587 (vect_dr_behavior): New function.
17588 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
17589 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
17590 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
17591 track whether the step preserves the misalignment.
17592 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
17593 Use vect_dr_behavior.
17594 (vect_setup_realignment): Update call accordingly.
17595 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
17596 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
17597 call to vect_create_addr_base_for_vector_ref.
17598 (vect_create_cond_for_align_checks): Likewise.
17599 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
17600 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
17601 (vect_recog_mask_conversion_pattern): Likewise.
17602 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
17603 (new_stmt_vec_info): Remove redundant zeroing.
17605 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
17607 * common/config/arm/arm-common.c (arm_be8_option): New function.
17608 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
17609 (ISA_ARMv6): Add isa_bit_be8.
17610 * config/arm/arm.h (arm_be8_option): Add prototype.
17611 (BE8_SPEC_FUNCTION): New define.
17612 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
17613 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
17614 (mlittle-endian): Similarly.
17615 (mbe8, mbe32): New options.
17616 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
17617 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
17619 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17621 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
17623 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17625 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
17626 (cleanup_tree_cfg_bb): Use it.
17627 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
17629 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
17631 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17634 * loop-doloop.c (add_test): Update profile.
17636 2017-07-03 Martin Liska <mliska@suse.cz>
17639 * sanopt.c (rewrite_usage_of_param): New function.
17640 (sanitize_rewrite_addressable_params): Likewise.
17641 (pass_sanopt::execute): Call rewrite_usage_of_param.
17643 2017-07-03 Richard Biener <rguenther@suse.de>
17645 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
17646 back to using VIEW_CONVERT_EXPR.
17648 2017-07-03 Martin Liska <mliska@suse.cz>
17651 * doc/extend.texi: Document when a resolver function is
17652 generated for target_clones.
17654 2017-07-03 Martin Liska <mliska@suse.cz>
17656 * asan.c (asan_emit_stack_protection): Unpoison just red zones
17657 and shadow memory of auto variables which are subject of
17658 use-after-scope sanitization.
17659 (asan_expand_mark_ifn): Add do set only when is_poison.
17661 2016-07-03 Richard Biener <rguenther@suse.de>
17663 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
17665 (vect_force_simple_reduction): Record reduction def -> phi mapping.
17666 (vectorizable_reduction): Perform reduction PHI creation when
17667 visiting a reduction PHI and adjust and simplify code generation
17668 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
17669 (vect_transform_loop): Visit reduction PHIs.
17670 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
17671 defs into the SLP tree.
17672 (vect_build_slp_tree): Reduction defs terminate the recursion.
17673 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
17675 (vect_get_vec_defs_for_stmt_copy): Export.
17676 (vect_get_vec_defs): Likewise.
17677 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
17679 (vect_get_vec_defs_for_stmt_copy): Declare.
17680 (vect_get_vec_defs): Likewise.
17682 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17684 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
17685 parameter with a "loop" parameter and use it instead of the
17686 loop containing DR_STMT. Don't check simple_iv when doing
17687 BB analysis. Describe the two analysis modes in the comment.
17689 2017-07-03 Tom de Vries <tom@codesourcery.com>
17691 PR tree-optimization/69468
17692 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
17693 (find_same_succ_bb): Handle ignore_edge_flags.
17695 2017-07-03 Tom de Vries <tom@codesourcery.com>
17697 PR tree-optimization/81192
17698 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
17700 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
17702 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
17704 2017-07-03 Tom de Vries <tom@codesourcery.com>
17706 PR tree-optimization/81192
17707 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
17708 BB_SAME_SUCC (bb) == NULL.
17710 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17712 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
17715 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17717 * dumpfile.c: Include profile-count.h
17718 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
17720 (insert_cond_bb): Update profile.
17721 * tree-cfg.h (insert_cond_bb): Update prototype.
17722 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
17723 * tree-dump.c: Do not include tree-cfg.
17725 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17727 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
17729 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17731 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
17734 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17736 * tree-complex.c (expand_complex_div_wide): update profile.
17738 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17739 Alan Hayward <alan.hayward@arm.com>
17740 David Sherwood <david.sherwood@arm.com>
17742 * Makefile.in (MACHMODE_H): Remove insn-modes.h
17743 (CORETYPES_H): New define.
17744 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
17745 (insn-modes-inline.h, s-modes-inline-h): New rules.
17746 (generated_files): Add insn-modes-inline.h.
17747 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
17748 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
17749 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
17750 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
17751 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
17752 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
17753 (build/gencodes.o, build/genconditions.o): Likewise.
17754 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
17755 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
17756 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
17757 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
17758 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
17759 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
17760 * coretypes.h: Include everything up to real.h for generators.
17761 Include insn-modes.h first. Include wide-int-print.h after
17762 wide-int.h. Include insn-modes-inline.h and then machmode.h.
17763 * machmode.h: Don't include insn-modes.h here.
17764 * function-tests.c: Remove includes of signop.h, machmode.h,
17765 double-int.h and wide-int.h.
17767 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
17769 * optc-save-gen.awk: Likewise.
17770 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
17771 * godump.c: Remove include of wide-int-print.h.
17772 * pretty-print.h: Likewise.
17773 * wide-int-print.cc: Likewise.
17774 * wide-int.cc: Likewise.
17775 * hash-map-tests.c: Remove include of signop.h.
17776 * hash-set-tests.c: Likewise.
17777 * rtl-tests.c: Likewise.
17778 * mkconfig.sh: Remove include of machmode.h.
17779 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
17781 (emit_insn_modes_inline_h): ...this new function. Emit the code
17782 into an insn-modes-inline.h header file, adding appropriate
17783 include guards and end comments.
17784 (emit_insn_modes_c_header): Remove include of machmode.h.
17785 (emit_min_insn_modes_c_header): Include coretypes.h rather than
17787 (main): Handle -i flag and call emit_insn_modes_inline_h when
17790 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17792 * tree-ssa-strlen.c (strinfo): Rename the length field to
17793 nonzero_chars. Add a full_string_p field.
17794 (compare_nonzero_chars, zero_length_string_p): New functions.
17795 (get_addr_stridx): Add an offset_out parameter.
17796 Use compare_nonzero_chars.
17797 (get_stridx): Update accordingly. Use compare_nonzero_chars.
17798 (new_strinfo): Update after above changes to strinfo.
17799 (set_endptr_and_length): Set full_string_p.
17800 (get_string_length): Update after above changes to strinfo.
17801 (unshare_strinfo): Update call to new_strinfo.
17802 (maybe_invalidate): Likewise.
17803 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
17804 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
17805 as a uhwi instead of an shwi. Update after above changes to
17806 strinfo and new_strinfo.
17807 (zero_length_string): Assert that chainsi contains full strings.
17808 Use zero_length_string_p. Update call to new_strinfo.
17809 (adjust_related_strinfos): Update after above changes to strinfo.
17810 Copy full_string_p from origsi.
17811 (adjust_last_stmt): Use zero_length_string_p.
17812 (handle_builtin_strlen): Update after above changes to strinfo and
17813 new_strinfo. Install the lhs as the string length if the previous
17814 entry didn't describe a full string.
17815 (handle_builtin_strchr): Update after above changes to strinfo
17817 (handle_builtin_strcpy): Likewise.
17818 (handle_builtin_strcat): Likewise.
17819 (handle_builtin_malloc): Likewise.
17820 (handle_pointer_plus): Likewise.
17821 (handle_builtin_memcpy): Likewise. Track nonzero characters
17822 that aren't necessarily followed by a nul terminator.
17823 (handle_char_store): Likewise.
17825 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17827 PR tree-optimization/80769
17828 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
17829 for malloc and calloc. Document the new invariant that all related
17830 strinfos have delayed lengths or none do.
17831 (verify_related_strinfos): Move earlier in file.
17832 (set_endptr_and_length): New function, split out from...
17833 (get_string_length): ...here. Also set the lengths of related
17835 (zero_length_string): Assert that chainsi has known (rather than
17837 (adjust_related_strinfos): Likewise.
17839 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17841 PR tree-optimization/81136
17842 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
17843 assert that two references with the same misalignment have the same
17844 compile-time misalignment if those compile-time misalignments
17847 2017-07-01 Andi Kleen <ak@linux.intel.com>
17849 * print-tree.c (print_node): Print all attributes.
17851 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17853 * cfg.c (scale_bbs_frequencies): New function.
17854 * cfg.h (scale_bbs_frequencies): Declare it.
17855 * cfgloopanal.c (single_likely_exit): Cleanup.
17856 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
17858 (scale_loop_profile): Likewise.
17859 (loop_version): Likewise.
17860 (create_empty_loop_on_edge): Update.
17861 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
17862 scale_loop_frequencies, scale_loop_profile, loopify,
17863 loop_version): Update prototypes.
17864 * modulo-sched.c (sms_schedule): Update.
17865 * predict.c (unlikely_executed_edge_p): Also check probability.
17866 (probably_never_executed_edge_p): Fix typo.
17867 * tree-if-conv.c (version_loop_for_if_conversion): Update.
17868 * tree-parloops.c (gen_parallel_loop): Update.
17869 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
17870 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
17871 * tree-ssa-loop-split.c (split_loop): Update.
17872 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
17873 * tree-vect-loop-manip.c (vect_do_peeling): Update.
17874 (vect_loop_versioning): Update.
17875 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
17877 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17879 * trans-mem.c (split_bb_make_tm_edge): Update profile.
17881 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17883 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
17884 to keep profile consistent.
17886 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17888 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
17889 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
17890 * profile-count.h (max_safe_multiplier): Make unsigned.
17891 (profile_count::guessed_zero): New.
17893 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17895 * bb-reorder.c (fix_up_crossing_landing_pad,
17896 fix_crossing_conditional_branches): Use make_single_succ_edge
17897 to keep profile consistent.
17899 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
17901 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
17904 2017-07-01 Jakub Jelinek <jakub@redhat.com>
17907 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
17908 the right scopes, make sure cond_jump isn't preserved between multiple
17909 iterations. Search for fallthru edge whenever there are 3+ edges and
17910 use find_fallthru_edge for it.
17912 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17914 Patch by Alexander Monakov <amonakov@ispras.ru>
17915 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
17916 probabilities consistently.
17918 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17920 * pa.c (pa_expand_compare_and_swap_loop): Update call of
17921 emit_cmp_and_jump_insns.
17923 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17926 * tree-inline.c (expand_call_inline): Combine profile statuses.
17928 2017-06-30 Andrew Pinski <apinski@cavium.com>
17930 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
17931 fold_stmt returned true.
17933 2017-06-30 Nathan Sidwell <nathan@acm.org>
17935 * ggc.h (empty_string): Delete.
17936 * cfgexpand.c (expand_asm_stmt): Use plain "".
17937 * optabs.c (expand_asm_memory_barrier): Likewise.
17938 * stringpool.c (empty_string): Delete.
17939 (digit_vector, digit_string): Delete.
17940 (ggc_alloc_string): Use plain "", don't optimize single digit
17941 strings. Use ggc_alloc_atomic.
17943 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
17945 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
17946 comparison set and one other set, use the cost of the non-comparison
17949 2017-06-30 Nathan Sidwell <nathan@acm.org>
17951 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
17954 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
17956 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
17957 loops. Remove now unneeded calls to gimple_switch_set_label() that
17958 just set removed labels to NULL_TREE.
17960 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
17962 * tree-ssanames.c (set_range_info_raw): Abstract from ...
17963 (set_range_info): ...here. Only call set_range_info_raw if domain
17965 (set_nonzero_bits): Call set_range_info_raw.
17966 * tree-ssanames.h (set_range_info_raw): New.
17968 2017-06-30 Jakub Jelinek <jakub@redhat.com>
17971 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
17972 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
17973 of nonimmediate_operand and <store_mask_constraint> instead of m
17974 for the input operand. For V8FI iterator, always split if input
17975 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
17976 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
17977 <store_mask_predicate> instead of register_operand and
17978 <store_mask_constraint> instead of v for the input operand. Make
17979 sure both operands aren't MEMs for if not <mask_applied>.
17981 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
17983 * lto-wrapper.c (copy_file) Close both file descriptors before
17986 2017-06-30 Martin Liska <mliska@suse.cz>
17989 * multiple_target.c (create_dispatcher_calls): Make ifunc
17990 also for function that don't have calls or are not referenced.
17992 2017-06-30 Richard Biener <rguenther@suse.de>
17994 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
17995 analyze the first scalar stmt. Move vector type computation
17996 for the BB case here from ...
17997 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
17998 live operation processing in the SLP case properly.
18000 2017-06-30 Richard Biener <rguenther@suse.de>
18002 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
18004 2017-06-30 Martin Liska <mliska@suse.cz>
18007 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
18008 before BUILT_IN_UNWIND_RESUME when ASAN is used.
18010 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
18012 * doc/invoke.texi (AArch64): Add missing options and remove redundant
18015 2017-06-30 Richard Biener <rguenther@suse.de>
18017 PR tree-optimization/81249
18018 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
18019 condition reduction result to original scalar type.
18021 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18023 * profile-count.h (enum profile_quality): Fix typos and whitespace
18026 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18028 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
18029 type for branch probabilities.
18031 2017-06-29 Julian Brown <julian@codesourcery.com>
18032 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18034 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
18035 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
18036 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
18037 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
18039 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18041 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
18042 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
18043 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
18044 CC usage from generic code to here.
18045 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
18046 CC usage into the target macros.
18048 2017-06-29 Maya Rashish <coypu@sdf.org>
18050 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
18053 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18055 * arm/arm-builtins.c: Include profile-count.h
18056 * except.c (sjlj_emit_function_enter): Use
18057 profile_probability::unlikely.
18059 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18061 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
18062 and tocrel_offset be pointer args rather than implicitly using
18064 (legitimate_constant_pool_address_p, rs6000_emit_move,
18065 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
18066 tocrel_offset and use in toc_relative_expr_p call.
18067 (print_operand, print_operand_address): Use static tocrel_base_oac
18068 and tocrel_offset_oac.
18069 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
18072 2017-06-29 Maya Rashish <coypu@sdf.org>
18074 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
18076 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
18078 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
18079 objects, take into account only the alignment of 'op0' and 'mode1' if
18082 2017-06-29 Steve Ellcey <sellcey@cavium.com>
18084 * ccmp.c (ccmp_tree_comparison_p): New function.
18085 (ccmp_candidate_p): Update to use above function.
18086 (get_compare_parts): New function.
18087 (expand_ccmp_next): Update to use new functions.
18088 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
18090 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
18091 take mode as argument.
18092 * ccmp.h (expand_ccmp_expr): Add mode as argument.
18093 * expr.c (expand_expr_real_1): Pass mode as argument.
18095 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
18097 * combine.c (combine_instructions): Print insns to dump_file, together
18100 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18102 * asan.c (asan_emit_stack_protection): Update.
18103 (create_cond_insert_point): Update.
18104 * auto-profile.c (afdo_propagate_circuit): Update.
18105 * basic-block.h (struct edge_def): Turn probability to
18106 profile_probability.
18107 (EDGE_FREQUENCY): Update.
18108 * bb-reorder.c (find_traces_1_round): Update.
18109 (better_edge_p): Update.
18110 (sanitize_hot_paths): Update.
18111 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
18112 (make_single_succ_edge): Update.
18113 (check_bb_profile): Update.
18114 (dump_edge_info): Update.
18115 (update_bb_profile_for_threading): Update.
18116 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
18117 probabilitycount to 0.
18118 * cfgbuild.c (compute_outgoing_frequencies): Update.
18119 * cfgcleanup.c (try_forward_edges): Update.
18120 (outgoing_edges_match): Update.
18121 (try_crossjump_to_edge): Update.
18122 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
18123 (expand_gimple_tailcall): Update.
18124 (construct_init_block): Use make_single_succ_edge.
18125 (construct_exit_block): Use make_single_succ_edge.
18126 * cfghooks.c (verify_flow_info): Update.
18127 (redirect_edge_succ_nodup): Update.
18128 (split_edge): Update.
18129 (account_profile_record): Update.
18130 * cfgloopanal.c (single_likely_exit): Update.
18131 * cfgloopmanip.c (scale_loop_profile): Update.
18132 (set_zero_probability): Remove.
18133 (duplicate_loop_to_header_edge): Update.
18134 * cfgloopmanip.h (loop_version): Update prototype.
18135 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
18136 (force_nonfallthru_and_redirect): Update.
18137 (update_br_prob_note): Update.
18138 (rtl_verify_edges): Update.
18139 (purge_dead_edges): Update.
18140 (rtl_lv_add_condition_to_bb): Update.
18141 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
18142 * cgraphunit.c (init_lowered_empty_function): Update.
18143 (cgraph_node::expand_thunk): Update.
18144 * cilk-common.c: Include profile-count.h
18145 * dojump.c (inv): Remove.
18146 (jumpifnot): Update.
18147 (jumpifnot_1): Update.
18148 (do_jump_1): Update.
18150 (do_jump_by_parts_greater_rtx): Update.
18151 (do_compare_rtx_and_jump): Update.
18152 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
18153 do_jump_1. do_compare_rtx_and_jump): Update prototype.
18154 * dwarf2cfi.c: Include profile-count.h
18155 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
18156 (sjlj_emit_dispatch_table): Likewise.
18157 * explow.c: Include profile-count.h
18158 * expmed.c (emit_store_flag_force): Update.
18159 (do_cmp_and_jump): Update.
18160 * expr.c (compare_by_pieces_d::generate): Update.
18161 (compare_by_pieces_d::finish_mode): Update.
18162 (emit_block_move_via_loop): Update.
18163 (store_expr_with_bounds): Update.
18164 (store_constructor): Update.
18165 (expand_expr_real_2): Update.
18166 (expand_expr_real_1): Update.
18167 * expr.h (try_casesi, try_tablejump): Update prototypes.
18168 * gimple-pretty-print.c (dump_probability): Update.
18169 (dump_profile): New.
18170 (dump_gimple_label): Update.
18171 (dump_gimple_bb_header): Update.
18172 * graph.c (draw_cfg_node_succ_edges): Update.
18173 * hsa-gen.c (convert_switch_statements): Update.
18174 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
18175 (find_if_case_1): Update.
18176 (find_if_case_2): Update.
18177 * internal-fn.c (expand_arith_overflow_result_store): Update.
18178 (expand_addsub_overflow): Update.
18179 (expand_neg_overflow): Update.
18180 (expand_mul_overflow): Update.
18181 (expand_vector_ubsan_overflow): Update.
18182 * ipa-cp.c (good_cloning_opportunity_p): Update.
18183 * ipa-split.c (split_function): Use make_single_succ_edge.
18184 * ipa-utils.c (ipa_merge_profiles): Update.
18185 * loop-doloop.c (add_test): Update.
18186 (doloop_modify): Update.
18187 * loop-unroll.c (compare_and_jump_seq): Update.
18188 (unroll_loop_runtime_iterations): Update.
18189 * lra-constraints.c (lra_inheritance): Update.
18190 * lto-streamer-in.c (input_cfg): Update.
18191 * lto-streamer-out.c (output_cfg): Update.
18192 * mcf.c (adjust_cfg_counts): Update.
18193 * modulo-sched.c (sms_schedule): Update.
18194 * omp-expand.c (expand_omp_for_init_counts): Update.
18195 (extract_omp_for_update_vars): Update.
18196 (expand_omp_ordered_sink): Update.
18197 (expand_omp_for_ordered_loops): Update.
18198 (expand_omp_for_generic): Update.
18199 (expand_omp_for_static_nochunk): Update.
18200 (expand_omp_for_static_chunk): Update.
18201 (expand_cilk_for): Update.
18202 (expand_omp_simd): Update.
18203 (expand_omp_taskloop_for_outer): Update.
18204 (expand_omp_taskloop_for_inner): Update.
18205 * omp-simd-clone.c (simd_clone_adjust): Update.
18206 * optabs.c (expand_doubleword_shift): Update.
18207 (expand_abs): Update.
18208 (emit_cmp_and_jump_insn_1): Update.
18209 (expand_compare_and_swap_loop): Update.
18210 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
18211 * predict.c (predictable_edge_p): Update.
18212 (edge_probability_reliable_p): Update.
18213 (set_even_probabilities): Update.
18214 (combine_predictions_for_insn): Update.
18215 (combine_predictions_for_bb): Update.
18216 (propagate_freq): Update.
18217 (estimate_bb_frequencies): Update.
18218 (force_edge_cold): Update.
18219 * profile-count.c (profile_count::dump): Add missing space into dump.
18220 (profile_count::debug): Add newline.
18221 (profile_count::differs_from_p): Explicitly convert to unsigned.
18222 (profile_count::stream_in): Update.
18223 (profile_probability::dump): New member function.
18224 (profile_probability::debug): New member function.
18225 (profile_probability::differs_from_p): New member function.
18226 (profile_probability::differs_lot_from_p): New member function.
18227 (profile_probability::stream_in): New member function.
18228 (profile_probability::stream_out): New member function.
18229 * profile-count.h (profile_count_quality): Rename to ...
18230 (profile_quality): ... this one.
18231 (profile_probability): New.
18232 (profile_count): Update.
18233 * profile.c (compute_branch_probabilities): Update.
18234 * recog.c (peep2_attempt): Update.
18235 * sched-ebb.c (schedule_ebbs): Update.
18236 * sched-rgn.c (find_single_block_region): Update.
18237 (compute_dom_prob_ps): Update.
18238 (schedule_region): Update.
18239 * sel-sched-ir.c (compute_succs_info): Update.
18240 * stmt.c (struct case_node): Update.
18241 (do_jump_if_equal): Update.
18242 (get_outgoing_edge_probs): Update.
18243 (conditional_probability): Update.
18244 (emit_case_dispatch_table): Update.
18245 (expand_case): Update.
18246 (expand_sjlj_dispatch_table): Update.
18247 (emit_case_nodes): Update.
18248 * targhooks.c: Update.
18249 * tracer.c (better_p): Update.
18250 (find_best_successor): Update.
18251 * trans-mem.c (expand_transaction): Update.
18252 * tree-call-cdce.c: Update.
18253 * tree-cfg.c (gimple_split_edge): Upate.
18254 (move_sese_region_to_fn): Upate.
18255 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
18256 * tree-eh.c (lower_resx): Upate.
18257 (cleanup_empty_eh_move_lp): Upate.
18258 * tree-if-conv.c (version_loop_for_if_conversion): Update.
18259 * tree-inline.c (copy_edges_for_bb): Update.
18260 (copy_cfg_body): Update.
18261 * tree-parloops.c (gen_parallel_loop): Update.
18262 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
18263 (gimple_gen_time_profiler): Update.
18264 * tree-ssa-dce.c (remove_dead_stmt): Update.
18265 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
18266 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
18267 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
18268 (unloop_loops): Update.
18269 (try_peel_loop): Update.
18270 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18271 * tree-ssa-loop-split.c (connect_loops): Update.
18272 (split_loop): Update.
18273 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
18274 (hoist_guard): Update.
18275 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
18276 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
18277 (value_replacement): Update.
18278 * tree-ssa-reassoc.c (branch_fixup): Update.
18279 * tree-ssa-tail-merge.c (replace_block_by): Update.
18280 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
18281 (create_edge_and_update_destination_phis): Update.
18282 (compute_path_counts): Update.
18283 (recompute_probabilities): Update.
18284 (update_joiner_offpath_counts): Update.
18285 (freqs_to_counts_path): Update.
18286 (duplicate_thread_path): Update.
18287 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
18288 (struct switch_conv_info): Update.
18289 (gen_inbound_check): Update.
18290 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
18291 (vect_do_peeling): Update.
18292 (vect_loop_versioning): Update.
18293 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18294 (optimize_mask_stores): Update.
18295 * ubsan.c (ubsan_expand_null_ifn): Update.
18296 * value-prof.c (gimple_divmod_fixed_value): Update.
18297 (gimple_divmod_fixed_value_transform): Update.
18298 (gimple_mod_pow2): Update.
18299 (gimple_mod_pow2_value_transform): Update.
18300 (gimple_mod_subtract): Update.
18301 (gimple_mod_subtract_transform): Update.
18302 (gimple_ic): Update.
18303 (gimple_stringop_fixed_value): Update.
18304 (gimple_stringops_transform): Update.
18305 * value-prof.h: Update.
18307 2017-06-29 Carl Love <cel@us.ibm.com>
18309 * config/rs6000/rs6000-c.c: Add support for built-in functions
18310 vector signed int vec_signed (vector float);
18311 vector signed long long vec_signed (vector double);
18312 vector signed int vec_signed2 (vector double, vector double);
18313 vector signed int vec_signede (vector double);
18314 vector signed int vec_signedo (vector double);
18315 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
18316 instruction generator.
18317 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
18318 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
18319 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
18321 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
18322 vunsignede_v2df): Add define_expands.
18323 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
18324 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
18325 VEC_UNSIGNEDO): Add definitions.
18326 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
18327 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
18328 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
18329 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
18330 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
18331 * config/rs6000/altivec.h (vec_signed, vec_signed2,
18332 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
18333 vec_unsignede, vec_unsignedo): Add builtin defines.
18334 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
18336 * doc/extend.texi: Update the built-in documentation file for the
18337 new built-in functions.
18339 2017-06-29 Richard Biener <rguenther@suse.de>
18341 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
18342 reduction chains to LOOP_VINFO_REDUCTIONS.
18343 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
18344 SLP reductions after processing reduction chains.
18346 2017-06-29 Nathan Sidwell <nathan@acm.org>
18348 * builtins.c (fold_builtin_FUNCTION): Use
18349 lang_hooks.decl_printable_name.
18351 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
18353 PR middle-end/81194
18354 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
18355 with only one label.
18356 * stmt.c (expand_case): Assert NCASES is greater than one.
18358 2017-06-29 Richard Biener <rguenther@suse.de>
18360 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
18362 (group_case_labels): Likewise.
18363 (find_taken_edge): Push sanity checking on val to workers...
18364 (find_taken_edge_cond_expr): ... here
18365 (find_taken_edge_switch_expr): ... and here, handle cases
18366 with just a default label.
18367 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
18368 (group_case_labels): Likewise.
18369 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
18370 group_case_labels does anything cleanup the CFG again.
18372 2017-06-29 Bin Cheng <bin.cheng@arm.com>
18374 PR tree-optimization/81196
18375 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
18376 exit condition comparing two IVs.
18378 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
18380 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
18381 profile to the dummy entry at the end of the list of architectures.
18382 * config/arm/arm-cpu-cdata.h: Regenerated.
18384 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18385 Michael Collison <michael.collison@arm.com>
18388 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
18390 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
18391 (*aarch64_reg_<mode>3_minus_mask): New pattern.
18392 (*aarch64_<optab>_reg_di3_mask2): New pattern.
18393 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
18394 of shift when the shift amount is masked with constant equal to
18395 the size of the mode.
18396 * config/aarch64/predicates.md (subreg_lowpart_operator): New
18399 2017-06-29 Martin Liska <mliska@suse.cz>
18401 * config/i386/i386.opt: Change range from [1,5] to [0,5].
18403 2017-06-29 Yury Gribov <tetra2005@gmail.com>
18406 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
18409 (edge_growth_cache_entry::edge_growth_cache_entry): New
18411 (reset_edge_growth_cache): Update to use constructor.
18413 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18415 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
18416 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
18417 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
18419 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
18421 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
18422 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
18424 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
18426 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
18427 (*-linux-uclibc*): Add t-uclibc tmake_file.
18428 * config/t-musl: New.
18429 * config/t-uclibc: New.
18431 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
18433 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
18435 (gen_comm_data): Emit architectural setting of arch_prof.
18436 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
18438 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
18439 (armv8-m.base, armv8-m.main): Likewise.
18440 * arm-protos.h (arm_build_target): Add profile field.
18441 (arch_option): Likewise.
18442 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
18444 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
18445 arm_active_target.profile.
18447 2017-06-28 Richard Biener <rguenther@suse.de>
18449 PR middle-end/81227
18450 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
18451 TYPE_OVERFLOW_WRAPS.
18452 * match.pd (negate_expr_p): Likewise.
18453 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
18454 fold_build2, not fold_binary.
18456 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
18458 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
18459 Convert memory address to Pmode.
18460 (aarch64_print_operand): Assert MEM operands are always Pmode.
18462 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
18465 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
18466 Remove redundant if.
18467 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
18468 * config/arm/aarch-common-protos.h
18469 (aarch_forward_to_shift_is_not_shifted_re): Remove.
18470 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
18472 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
18475 * multiple_target.c (create_dispatcher_calls): Set the default
18476 clone to be static, not public.
18478 2017-06-28 Richard Biener <rguenther@suse.de>
18480 * tree-vect-loop.c (vectorizable_reduction): Move special
18481 cond reduction IV var creation ...
18482 (vect_create_epilog_for_reduction): ... here. Remove induction_index
18483 parameter. Use STMT_VINFO_VECTYPE.
18484 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
18487 2017-06-28 Martin Liska <mliska@suse.cz>
18490 * ipa-visibility.c (non_local_p): Handle visibility.
18492 2017-06-28 Martin Liska <mliska@suse.cz>
18495 * common.opt: Add IntegerRange to various options.
18496 * opt-functions.awk (integer_range_info): New function.
18497 * optc-gen.awk: Add integer_range_info to cl_options struct.
18498 * opts-common.c (decode_cmdline_option): Handle
18499 CL_ERR_INT_RANGE_ARG.
18500 (cmdline_handle_error): Likewise.
18501 * opts.c (print_filtered_help): Show valid interval in
18502 when --help is provided.
18503 * opts.h (struct cl_option): Add range_min and range_max fields.
18504 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
18506 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
18508 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
18509 (x * C EQ/NE y * C): New transformation.
18511 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
18513 * genmultilib (combination_space): Accept '+' in option names.
18515 2017-06-28 Martin Liska <mliska@suse.cz>
18518 * asan.c (instrument_derefs): Bail out inner references
18519 that are hard register variables.
18521 2017-06-28 Jakub Jelinek <jakub@redhat.com>
18524 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
18525 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
18527 2017-06-28 Richard Biener <rguenther@suse.de>
18529 * tree-vectorizer.h (vect_get_vec_defs): Remove.
18530 (vect_get_slp_defs): Adjust.
18531 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
18533 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
18535 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
18536 get_initial_defs_for_reduction instead of vect_get_vec_defs.
18537 (vectorizable_reduction): Adjust.
18538 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
18540 (vect_get_slp_defs): Likewise.
18541 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
18542 (vectorizable_bswap): Adjust.
18543 (vectorizable_call): Likewise.
18544 (vectorizable_conversion): Likewise.
18545 (vectorizable_assignment): Likewise.
18546 (vectorizable_shift): Likewise.
18547 (vectorizable_operation): Likewise.
18548 (vectorizable_store): Likewise.
18549 (vectorizable_condition): Likewise.
18550 (vectorizable_comparison): Likewise.
18552 2017-06-28 Michael Collison <michael.collison@arm.com>
18555 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
18557 (arm_gen_movmemqi): Removed unused variable 'i'.
18558 Convert 'for' loop into 'while' loop.
18559 (arm_expand_prologue): Remove last unnecessary set of insn.
18560 (thumb_pop): Remove unused variable 'pushed_words'.
18561 (thumb_exit): Remove last unnecessary set of regs_to_pop.
18563 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18565 * config/s390/predicates.md: Use s390_rel_address_ok_p.
18566 * config/s390/s390-protos.h: Add prototype of
18567 s390_rel_address_ok_p.
18568 * config/s390/s390.c (s390_got_symbol): New function.
18569 (s390_rel_address_ok_p): New function.
18570 (legitimize_pic_address): Use s390_rel_address_ok_p.
18571 (s390_load_got): Use s390_got_symbol.
18572 (s390_option_override): Issue error if
18573 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
18574 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
18576 * config/s390/s390.opt: New option mpic-data-is-text-relative.
18578 2017-06-27 Andrew Pinski <apinski@cavium.com>
18580 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
18581 (X * copysign (1.0, X)): New pattern.
18582 (X * copysign (1.0, -X)): New pattern.
18583 (copysign (-1.0, CST)): New pattern.
18585 2017-06-27 Joseph Myers <joseph@codesourcery.com>
18587 * genmultilib (combination_space): Remove variable.
18588 Validate reuse rules against regular expression for any sequence
18589 of multilib options in any order.
18591 2017-06-27 Michael Collison <michael.collison@arm.com>
18593 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
18594 call aarch64_split_simd_combine.
18595 * (aarch64_combine_internal<mode>): Delete pattern.
18596 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
18597 Allow register and subreg operands.
18599 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18601 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
18602 specific need, just fallback on defaults.
18603 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
18605 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18606 Olivier Hainque <hainque@adacore.com>
18608 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
18610 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
18611 targets. Pick a default if no particular attempt applied.
18612 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
18615 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18617 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
18618 (x86_64-wrs-vxworks7): Likewise.
18620 2017-06-27 Marek Polacek <polacek@redhat.com>
18623 * ubsan.c (instrument_null): Check get_base_address's result for null.
18625 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
18627 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
18629 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
18631 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
18632 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
18633 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
18634 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
18635 New function types.
18636 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
18637 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
18638 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
18639 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
18640 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
18641 BUILT_IN_FEUPDATEENV): New builtins.
18642 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
18643 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
18644 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
18645 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
18647 (builtin_structptr_types): Adjust size.
18648 * tree.c (builtin_structptr_types): Add four entries.
18650 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18651 Olivier Hainque <hainque@adacore.com>
18653 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
18654 (TLS_SYM): New local macro, forcing reference to __tls__ on
18655 link command lines for VxWorks 7 RTPs, triggering initialization
18657 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
18658 OS features TLS support, true for RTPs on VxWorks 7.
18659 * config/vxworks.c (vxworks_override_options): Setup emutls
18662 2017-06-27 Jakub Jelinek <jakub@redhat.com>
18664 * predict.c (test_prediction_value_range): Use -1U instead of -1
18665 to avoid narrowing conversion warning.
18666 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
18667 to avoid narrowing conversion warning.
18668 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
18670 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
18672 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18674 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
18675 64bit configurations.
18676 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
18677 (SIZE_TYPE): Likewise.
18678 * config/vxworks.c (vxworks_emutls_var_fields): Use
18679 long_unsigned_type_node instead of unsigned_type_node as the offset
18680 field type, which is "pointer" mode in emutls.c.
18682 2017-06-27 Jakub Jelinek <jakub@redhat.com>
18685 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
18687 PR middle-end/81207
18688 * gimple-fold.c (replace_call_with_call_and_fold): Handle
18689 gimple_vuse copying separately from gimple_vdef copying.
18691 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18693 * value-prof.c (free_hist): Remove call to memset and the enclosing if
18696 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
18697 Olivier Hainque <hainque@adacore.com>
18699 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
18700 for all vxworks7 targets.
18701 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
18702 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
18703 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
18704 variations for VX6/VX7 and 32/64bits later on in ...
18705 (VXWORKS_LIB_SPEC): Leverage new macros.
18706 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
18707 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
18709 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
18711 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
18712 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
18714 2017-06-26 Carl Love <cel@us.ibm.com>
18716 * config/rs6000/rs6000-c.c: Add support for built-in functions
18717 vector bool char vec_reve (vector bool char);
18718 vector signed char vec_reve (vector signed char);
18719 vector unsigned char vec_reve (vector unsigned char);
18720 vector bool int vec_reve (vector bool int);
18721 vector signed int vec_reve (vector signed int);
18722 vector unsigned int vec_reve (vector unsigned int);
18723 vector bool long long vec_reve (vector bool long long);
18724 vector signed long long vec_reve (vector signed long long);
18725 vector unsigned long long vec_reve (vector unsigned long long);
18726 vector bool short vec_reve (vector bool short);
18727 vector signed short vec_reve (vector signed short);
18728 vector double vec_reve (vector double);
18729 vector float vec_reve (vector float);
18730 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
18731 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
18732 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
18733 (altivec_vreve): New pattern.
18734 * config/rs6000/altivec.h (vec_reve): New define.
18735 * doc/extend.texi (vec_rev): Update the built-in documentation file
18736 for the new built-in functions.
18738 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18740 PR tree-optimization/71815
18741 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
18743 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
18745 (slsr_process_phi): Likewise.
18746 (replace_uncond_cands_and_profitable_phis): Don't replace a
18747 multiply candidate with a stride of 1 (copy or cast).
18748 (phi_incr_cost): Call uses_consumed_by_stmt rather than
18750 (lowest_cost_path): Likewise.
18751 (total_savings): Likewise.
18753 2017-06-26 Richard Biener <rguenther@suse.de>
18756 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
18757 Use def_builtin_pure for all gather builtins.
18759 2017-06-26 Richard Biener <rguenther@suse.de>
18761 PR tree-optimization/81203
18762 * tree-tailcall.c (find_tail_calls): Do not move stmts into
18763 non-dominating BBs.
18765 2017-06-26 Marek Polacek <polacek@redhat.com>
18768 * doc/invoke.texi: Document -Wmultistatement-macros.
18770 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
18772 * doc/sourcebuild.texi (ARM-specific attributes): Document new
18773 arm_neon_ok_no_float_abi effective target.
18775 2017-06-26 Richard Biener <rguenther@suse.de>
18777 PR tree-optimization/80928
18778 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
18779 (copy_bbs): Set BB_DUPLICATED flag early.
18780 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
18782 (execute_on_shrinking_pred): Likewise.
18783 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
18784 BB_DUPLICATED blocks.
18785 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
18786 iterate over all PHIs considering removal of *gsi.
18788 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
18790 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
18793 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18795 * config/rs6000/rs6000-string.c: (expand_block_clear,
18796 do_load_for_compare, select_block_compare_mode,
18797 compute_current_alignment, expand_block_compare,
18798 expand_strncmp_align_check, expand_strn_compare,
18799 expand_block_move, rs6000_output_load_multiple)
18800 Move functions related to string/block move/compare
18801 to a separate file.
18802 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
18803 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
18804 for this function which is now used in two files.
18805 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
18806 * config.gcc: Add rs6000-string.o to extra_objs for
18807 targets powerpc*-*-* and rs6000*-*-*.
18809 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
18812 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
18813 32-bit, since indexed is not valid for DImode.
18814 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
18815 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
18816 (define_peephole2 for Altivec d-form load): Add 32-bit support.
18817 (define_peephole2 for Altivec d-form store): Likewise.
18820 * multiple_target.c (create_dispatcher_calls): Only create the
18821 dispatcher call if the function is the default clone of a
18822 versioned function.
18824 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
18826 PR middle-end/80902
18827 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
18828 a call, force the call to not be a tail call.
18830 2017-06-23 Jeff Law <law@redhat.com>
18832 * doc/contrib.texi: Add entry for Steven Pemberton's work on
18835 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
18837 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
18838 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
18839 handling for early expansion of vector shifts (sl,sr,sra,rl).
18840 (builtin_function_type): Add vector shift right instructions
18841 to the unsigned argument list.
18843 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
18845 rtl-optimizatoin/79286
18846 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
18847 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
18848 trap. PIC register plus a const unspec without offset can never trap.
18850 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
18852 * tree.h (builtin_structptr_type): New type.
18853 (builtin_structptr_types): Declare new array.
18854 * tree.c (builtin_structptr_types): New array.
18855 (free_lang_data, build_common_tree_nodes): Use it.
18857 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
18860 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
18863 2017-06-22 Matt Turner <mattst88@gmail.com>
18865 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
18866 Lake models to skylake case. Assume skylake for unknown
18867 models with clflushopt.
18869 2017-06-22 Jeff Law <law@redhat.com>
18871 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
18872 frame sizes that do not satisfy aarch64_uimm12_shift.
18874 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
18876 * profile-count.h (apply_probability,
18877 apply_scale, probability_in): Fix checks for zero.
18879 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18881 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
18882 * doc/cppdiropts.texi (-I @var{dir}): Document it.
18884 2016-06-22 Richard Biener <rguenther@suse.de>
18886 * tree-vect-loop.c (vect_model_reduction_cost): Handle
18887 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
18888 REDUC_MAX_EXPR support.
18889 (vectorizable_reduction): Likewise.
18890 (vect_create_epilog_for_reduction): Likewise.
18892 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
18894 * match.pd (A / (1 << B) -> A >> B): New.
18895 * generic-match-head.c: Include optabs-tree.h.
18896 * gimple-match-head.c: Likewise.
18897 * optabs-tree.h (target_supports_op_p): New.
18898 * optabs-tree.c (target_supports_op_p): New.
18900 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18902 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
18903 $gcc_cv_ld --help output.
18904 (gcc_cv_ld_demangle): Likewise.
18905 (gcc_cv_ld_eh_frame_hdr): Likewise.
18906 (gcc_cv_ld_pie): Likewise.
18907 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
18908 (gcc_cv_ld_buildid): Likewise.
18909 (gcc_cv_ld_sysroot): Likewise.
18910 (ld_bndplt_support): Likewise.
18911 (ld_pushpopstate_support): Likewise.
18912 * configure: Regenerate.
18913 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
18915 2017-06-21 Jakub Jelinek <jakub@redhat.com>
18918 * config/i386/sse.md (round<mode>2): Renumber match_dup and
18919 operands indexes to avoid gap between operands and match_dups.
18921 2017-06-21 Andrew Pinski <apinski@cavium.com>
18923 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
18924 Increment Arith_shift and Arith_shift_reg by 1.
18925 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
18927 * config/aarch64/aarch64.c (thunderx_tunings): Enable
18928 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
18929 (aarch64_strip_extend): Add new argument and test for it.
18930 (aarch64_cheap_mult_shift_p): New function.
18931 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
18932 add a cost if it is true.
18933 Update calls to aarch64_strip_extend.
18934 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
18936 2017-06-21 Andrew Pinski <apinski@cavium.com>
18938 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
18940 (thunderxt88): Likewise.
18941 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
18942 (thunderx_prefetch_tune): New variable.
18943 (thunderx2t99_prefetch_tune): Update for the correct values.
18944 (thunderxt88_tunings): New variable.
18945 (thunderx_tunings): Use thunderx_prefetch_tune instead of
18946 generic_prefetch_tune.
18947 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
18949 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18951 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
18952 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
18953 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
18954 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
18955 (aarch64_atomic_cas<mode>, GPI): Likewise.
18957 2017-06-21 Martin Liska <mliska@suse.cz>
18959 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
18960 statements on cold and hot labels.
18961 * predict.c (tree_estimate_probability_bb): Remove the
18962 prediction from this place.
18964 2017-06-21 Martin Liska <mliska@suse.cz>
18966 PR tree-optimization/79489
18967 * gimplify.c (maybe_add_early_return_predict_stmt): New
18969 (gimplify_return_expr): Call the function.
18970 * predict.c (tree_estimate_probability_bb): Remove handling
18972 * predict.def: Update comment about early return predictor.
18973 * gimple-predict.h (is_gimple_predict): New function.
18974 * predict.def: Change default value of early return to 66.
18975 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
18977 * passes.def: Put pass_strip_predict_hints to the beginning of
18980 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
18982 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
18983 FUNCTION_DECL declarations.
18984 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
18986 (dwaf2out_decl): Likewise.
18987 * godump.c (go_early_global_decl): Skip call to the real debug hook
18988 for FUNCTION_DECL declarations.
18989 * passes.c (rest_of_decl_compilation): Skip call to the
18990 early_global_decl debug hook for FUNCTION_DECL declarations, unless
18991 -fdump-go-spec is passed.
18993 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
18995 * config/i386/i386.c (struct builtin_isa): New field pure_p.
18996 Reorder for compactness.
18997 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
18998 (def_builtin_pure, def_builtin_pure2): New functions.
18999 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
19001 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
19003 * match.pd (nop_convert): New predicate.
19004 ((A +- CST1) +- CST2): Allow some NOP conversions.
19006 2017-06-21 Jakub Jelinek <jakub@redhat.com>
19009 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
19010 with ctors/dtors if GOVD_SHARED is set.
19012 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19014 * config/aarch64/aarch64.md (movti_aarch64):
19015 Emit mov rather than orr.
19016 (movtf_aarch64): Likewise.
19017 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
19018 Emit mov rather than orr.
19020 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19022 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
19023 Swap alternatives, make integer dup more expensive.
19025 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19027 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
19028 Return true for non-tls symbols.
19030 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
19032 * config/aarch64/aarch64-cores.def (cortex-a55): New.
19033 (cortex-a75): Likewise.
19034 (cortex-a75.cortex-a55): Likewise.
19035 * config/aarch64/aarch64-tune.md: Regenerate.
19036 * doc/invoke.texi (-mtune): Document new values for -mtune.
19038 2017-06-21 Tom de Vries <tom@codesourcery.com>
19040 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
19041 stack_size feature.
19042 (Effective-Target Keywords, Other attributes): Suggest using
19043 dg-add-options stack_size feature to get stack limit in stack_size
19044 effective target documentation.
19046 2017-06-21 Julian Brown <julian@codesourcery.com>
19047 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19049 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
19050 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
19051 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
19053 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
19054 attribute type list for neon_multiply.
19055 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
19056 attribute type list for neon_multiply.
19057 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
19058 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
19059 attribute type list for neon_multiply.
19060 * config/arm/types.md (crypto_pmull): Add.
19061 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
19062 attribute type list.
19064 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
19066 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
19069 2017-06-20 Jakub Jelinek <jakub@redhat.com>
19071 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
19072 to make sure not to dereference a NULL cost_classes_ptr pointer.
19074 2017-06-20 Carl Love <cel@us.ibm.com>
19076 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19077 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
19078 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
19079 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
19080 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
19081 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
19082 VMULOSW): New enum "unspec" values.
19083 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
19084 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
19085 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
19086 altivec_vmulosw): New patterns.
19087 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
19088 VMULOSW): Add definitions.
19090 2017-06-20 Julia Koval <julia.koval@intel.com>
19092 * config/i386/i386.c: Fix rounding expand for new pattern.
19093 * config/i386/subst.md: Fix pattern (parallel -> unspec).
19095 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19097 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
19098 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
19100 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19102 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
19105 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19107 * config/aarch64/aarch64-cores.def: Rearrange to sort by
19108 architecture, then by implementer ID.
19109 * config/aarch64/aarch64-tune.md: Regenerate.
19111 2017-06-20 Richard Biener <rguenther@suse.de>
19113 PR middle-end/81097
19114 * fold-const.c (split_tree): Fold to type before negating.
19116 2017-06-20 David Malcolm <dmalcolm@redhat.com>
19118 * diagnostic-show-locus.c
19119 (selftest::test_fixit_deletion_affecting_newline): New function.
19120 (selftest::diagnostic_show_locus_c_tests): Call it.
19122 2017-06-20 Andreas Schwab <schwab@suse.de>
19125 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
19128 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19130 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
19131 __ARM_FEATURE_COPROC according to support.
19133 2017-06-20 Jakub Jelinek <jakub@redhat.com>
19135 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
19136 Rewritten to avoid overflow for > 32-bit pointers.
19139 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
19140 by removing enum keyword.
19141 (ubsan_type_descriptor): Likewise. Formatting fix.
19144 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
19145 splitter): Require TARGET_SSE2 in the condition.
19147 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
19150 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
19151 for doing vector set of SFmode on ISA 3.0.
19152 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
19153 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
19155 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
19156 SFmode value into a V4SF variable that was extracted from another
19157 V4SF variable without converting the element to double precision
19158 and back to single precision vector format.
19159 (vsx_insert_extract_v4sf_p9_2): Likewise.
19161 2017-06-19 Jakub Jelinek <jakub@redhat.com>
19163 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
19164 in UWHI to avoid undefined overflow.
19167 * ubsan.h (enum ubsan_encode_value_phase): New.
19168 (ubsan_encode_value): Change second argument to
19169 enum ubsan_encode_value_phase with default value of
19170 UBSAN_ENCODE_VALUE_GENERIC.
19171 * ubsan.c (ubsan_encode_value): Change second argument to
19172 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
19173 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
19174 create_tmp_var_raw instead of create_tmp_var and use a
19176 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
19177 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
19178 ubsan_encode_value callers.
19181 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
19182 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
19183 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
19185 2017-06-19 Richard Biener <rguenther@suse.de>
19187 PR middle-end/81118
19188 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
19189 estimates if we changed anything.
19191 2017-06-19 Richard Biener <rguenther@suse.de>
19193 PR tree-optimization/80887
19194 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
19195 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
19196 simplified lookups, then reset mprts_hook.
19197 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
19199 (try_to_simplify): Likewise.
19201 2017-06-19 Martin Liska <mliska@suse.cz>
19204 * gimplify.c (gimplify_switch_expr):
19205 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
19207 2017-06-19 Martin Liska <mliska@suse.cz>
19209 * doc/install.texi: Document that PGO runs in 4 stages.
19211 2017-06-19 Martin Liska <mliska@suse.cz>
19214 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
19215 to dispatcher function name.
19216 * multiple_target.c (replace_function_decl): New function.
19217 (create_dispatcher_calls): Redirect both edges and references.
19219 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
19221 * profile-count.c (profile_count::dump): Dump quality.
19222 (profile_count::differs_from_p): Update for unsigned val.
19223 * profile-count.h (profile_count_quality): New enum.
19224 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
19226 2017-06-19 Richard Biener <rguenther@suse.de>
19228 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
19229 struct function as arg.
19230 (estimate_numbers_of_iterations): Export overload with loop arg.
19231 (free_numbers_of_iterations_estimates_loop): Use an overload of
19232 free_numbers_of_iterations_estimates instead.
19233 * tree-cfg.c (remove_bb): Adjust.
19234 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
19235 * tree-parloops.c (gen_parallel_loop): Likewise.
19236 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
19238 (tree_unroll_loops_completely): Likewise.
19239 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
19240 Use an overload instead and export.
19241 (estimated_loop_iterations): Adjust.
19242 (max_loop_iterations): Likewise.
19243 (likely_max_loop_iterations): Likewise.
19244 (estimate_numbers_of_iterations): Take struct function as arg
19246 (loop_exits_before_overflow): Adjust.
19247 (free_numbers_of_iterations_estimates_loop): Use an overload.
19248 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
19249 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
19251 2017-06-19 Richard Biener <rguenther@suse.de>
19254 * ipa-prop.c (find_constructor_constant_at_offset): Handle
19255 RANGE_EXPR conservatively.
19257 2017-06-16 Carl Love <cel@us.ibm.com>
19259 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19260 definitions for vec_float, vec_float2, vec_floato,
19261 vec_floate built-ins.
19262 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
19263 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
19265 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
19266 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
19267 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
19268 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
19269 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
19270 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
19271 vec_floato): Add builtin defines.
19272 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
19273 Update the built-in documentation file for the new built-in
19276 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19278 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
19279 (mthumb): Mark as the negative of -marm.
19281 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19283 * doc/invoke.texi (ARM Options, -mcpu): Document supported
19285 (ARM Options, -mtune): Document that this accepts the same
19286 extension options as -mcpu.
19287 (ARM Options, -mfpu): Document addition of -mfpu=auto.
19289 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19291 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
19292 permitted extensions.
19294 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19296 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
19297 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
19298 (armv8-m.main): Add option +nodsp.
19299 * config/arm/arm-cpu-cdata.h: Regenerated.
19301 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19303 * config/arm/t-fuchsia: New file.
19304 * config.gcc (arm*-*-fuchsia*): Use it.
19306 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19308 * config/arm/t-symbian: Rewrite for new option infrastructure.
19310 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19312 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
19313 (MULTILIB_REQUIRED): Likewise.
19315 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19317 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
19318 (MULTILIB_RESUE): Likewise.
19319 (MULTILIB_MATCHES): Likewise.
19320 (MULTLIB_REQUIRED): Likewise.
19322 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19324 * config/arm/t-rtems: Rewrite for new option framework.
19326 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19328 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
19329 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
19330 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
19331 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
19332 * config/arm/t-multilib: ... here.
19333 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
19334 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
19335 armv7-a and armv8*-a when A-profile libraries have not been built.
19336 * config/arm/t-rmprofile: Rewrite.
19338 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19340 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
19341 with a backslash. Remove the backslash after substituting unescaped
19343 * doc/fragments.texi (MULTILIB_REUSE): Document it.
19345 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19347 * config.gcc: (arm*-*-*): When building a-profile libraries, force
19348 the driver to pass through the default setting of -mfloat-abi.
19349 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
19351 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
19352 (all_feat_combs): New rule.
19353 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
19355 * config/arm/t-aprofile: Rewrite.
19357 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19359 * config/arm/arm.h (FPUTYPE_AUTO): Define.
19360 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
19361 fpu is not specified by the user/command-line.
19362 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
19363 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
19364 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
19365 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
19366 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
19367 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
19369 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19371 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
19372 * config/arm/t-arm-elf: Rewritten.
19374 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19376 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
19377 have some floating-point instructions.
19378 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
19379 (TARGET_MAYBE_HARD_FLOAT): New macro.
19380 * config/arm/arm-builtins.c (arm_init_builtins): Use
19381 TARGET_MAYBE_HARD_FLOAT.
19382 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
19384 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19386 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
19387 (configargs.h): Include it.
19388 (arm_print_hint_for_fpu_option): New function.
19389 (arm_parse_fpu_option): New function.
19390 (candidate_extension): New class.
19391 (arm_canon_for_multilib): New function.
19392 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
19393 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
19394 (ARCH_CANONICAL_SPECS): New macro.
19395 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
19397 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19399 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
19400 are set after handling multilib fragments. Set target_cpu_default2
19403 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19405 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
19407 (arm*-*-*): Set target_cpu_default2 to a quoted string.
19408 * config/arm/parsecpu.awk (check_cpu): Validate any extension
19410 (check_arch): Likewise.
19411 * config/arm/arm.c (arm_configure_build_target): Handle
19412 TARGET_CPU_DEFAULT being a string constant. Scan any feature
19413 options in the default.
19415 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19417 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
19418 when an option is an alias of another.
19419 * config/arm/parsecpu.awk (optalias): New parser token.
19420 (gen_comm_data): Mark non-alias options as such. Emit entries
19421 for extension aliases.
19422 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
19423 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
19424 (armv6kz, armv6zk, armv6t2): Likewise.
19425 (armv7): Make vfpv3-d16 an alias.
19426 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
19428 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
19429 Sort in canonical order.
19430 (armv8-a): Sort in canonical order.
19431 (armv8.1-a, armv8.2-a): Likewise.
19432 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
19434 (cortex-a9): Sort in canonical order.
19435 * config/arm/arm.c (selftests.h): Include it.
19436 (arm_test_cpu_arch_data): New function.
19437 (arm_run_self_tests): New function.
19438 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
19439 (targetm): Move declaration to the end of the file.
19440 * arm-cpu-cdata.h: Regenerated.
19442 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19444 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
19445 call to target_mode_check describing the type of option passed.
19446 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
19447 (arm_target_thumb_only): Use arm_parse_arch_option_name or
19448 arm_parse_cpu_option_name to match parameters against list of
19450 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
19451 arm_arch_core_flags data structure.
19452 * config/arm/arm-cpu_cdata.h: Regenerated.
19454 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19456 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
19458 (arm_print_hint_for_cpu_option): Likewise.
19459 (arm_print_hint_for_arch_option): Likewise.
19460 (arm_parse_cpu_option_name): Likewise.
19461 (arm_parse_arch_option_name): Likewise.
19462 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
19463 of entries in the all_fpus list.
19464 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
19465 (arm_parse_cpu_option_name): Declare.
19466 (arm_parse_arch_option_name): Declare.
19467 (arm_parse_option_features): Declare.
19468 (arm_intialize_isa): Declare.
19469 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
19471 (gen_comm_data): ... here. Make definitions non-static.
19472 * config/arm/arm-cpu-data.h: Regenerated.
19473 * config/arm/arm-cpu-cdata.h: Regenerated.
19475 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19477 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
19478 (cpu_arch_extension): New structure.
19479 (cpu_arch_option, arch_option, cpu_option): New structures.
19480 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
19481 architecture types.
19482 (gen_data): Generate new format data tables.
19483 * config/arm/arm.c (cpu_tune): New structure.
19484 (cpu_option, processors): Delete.
19485 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
19486 (arm_print_hint_for_cpu_option): ... this and ...
19487 (arm_print_hint_for_arch_option): ... this.
19488 (arm_parse_arch_cpu_name): Delete. Replace with ...
19489 (arm_parse_cpu_option_name): ... this and ...
19490 (arm_parse_arch_option_name): ... this.
19491 (arm_unrecognized_feature): Change type of target parameter to
19493 (arm_parse_arch_cpu_features): Delete. Replace with ...
19494 (arm_parse_option_features): ... this.
19495 (arm_configure_build_target): Rework to use new configuration data
19497 (arm_print_tune_info): Rework for new configuration data tables.
19498 * config/arm/arm-cpu-data.h: Regenerated.
19499 * config/arm/arm-cpu.h: Regenerated.
19501 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19503 * Makefile.in (OBJS): Move sbitmap.o from here ...
19504 (OBJS-libcommon): ... to here.
19506 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19508 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
19509 (ISA_ALL_CRYPTO): New macro.
19510 (ISA_ALL_SIMD): New macro
19511 (ISA_ALL_FP): New macro.
19512 * config/arm/arm.c (fpu_bitlist): Update initializer.
19513 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
19515 (arm9e): Add fpu. Add option for nofp
19516 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
19517 (arm926ej-s, arm1026ej-s): Likewise.
19518 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
19519 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
19520 neon-fp16, neon-vfpv4, nofp and nosimd.
19521 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
19522 (cortex-a8): Add fpu. Add option for nofp.
19523 (cortex-a9): Add fpu. Add options for nosimd and nofp.
19524 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
19525 (cortex-r4f): Add fpu.
19526 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
19527 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
19529 (cortex-r8): Likewise.
19530 (cortex-m4): Add fpu. Add option for nofp.
19531 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
19532 (cortex-a17.cortex-a7): Likewise.
19533 (cortex-a32): Add fpu. Add options for crypto and nofp.
19534 (cortex-a35, cortex-a53): Likewise.
19535 (cortex-a57): Add fpu. Add option for crypto.
19536 (cortex-a72, cortex-a73): Likewise.
19537 (exynos-m1): Likewise.
19538 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
19539 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
19540 (cortex-m33): Add fpu. Add option for nofp.
19541 * config/arm/arm-cpu-cdata.h: Regenerated
19542 * config/arm/arm-cpu-data.h: Regenerated.
19544 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19546 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
19547 (armv5te, armv5tej): Likewise.
19548 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
19549 (armv7): Add options fp and vfpv3-d16.
19550 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
19551 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
19553 (armv7ve): Likewise.
19554 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
19555 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
19556 (armv8-a): Add nocrypto option.
19557 (armv8.1-a, armv8.2-a): Likewise.
19558 (armv8-m.main): add options fp, fp.dp and nofp.
19560 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19562 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
19564 (armv8-a+crc): Delete.
19565 (armv8.1-a): Add options simd, crypto and nofp.
19566 (armv8.2-a): Add options fp16, simd, crypto and nofp.
19567 (armv8.2-a+fp16): Delete.
19568 (armv8-m.main): Add option dsp.
19569 (armv8-m.main+dsp): Delete.
19570 (cortex-a8): Add fpu. Add nofp option.
19571 (cortex-a9): Add fpu. Add nofp and nosimd options.
19572 * config/arm/parsecpu.awk (gen_data): Generate option tables and
19573 link to main cpu and architecture data structures.
19574 (gen_comm_data): Only put isa attributes from the main architecture
19576 (option): New statement for architecture and CPU entries.
19577 * arm.c (struct cpu_option): New structure.
19578 (struct processors): Add entry for options.
19579 (arm_unrecognized_feature): New function.
19580 (arm_parse_arch_cpu_name): Ignore any characters after the first
19582 (arm_parse_arch_cpu_feature): New function.
19583 (arm_configure_build_target): Separate out any CPU and architecture
19584 features and parse separately. Don't error out if -mfpu=auto is
19585 used with only an architecture string.
19586 (arm_print_asm_arch_directives): New function.
19587 (arm_file_start): Call it.
19588 * config/arm/arm-cpu-cdata.h: Regenerated.
19589 * config/arm/arm-cpu-data.h: Likewise.
19590 * config/arm/arm-tables.opt: Likewise.
19592 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19594 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
19595 assembler when it is not -mfpu=auto.
19597 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19599 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
19600 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
19601 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
19602 (ASM_CPU_SPEC): Rewrite.
19603 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
19604 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
19605 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
19606 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
19607 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
19608 copied string is NUL-terminated. Also strip any characters prefixed
19610 (arm_rewrite_selected_arch): New function.
19611 (arm_rewrite_march): New function.
19613 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19615 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
19616 (x_arm_cpu_string, x_arm_tune_string): Likewise.
19617 (march, mcpu, mtune): Convert to string-based options.
19618 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
19619 (arm_parse_arch_cpu_name): New function.
19620 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
19621 identify selected architecture or CPU.
19622 (arm_option_save): New function.
19623 (TARGET_OPTION_SAVE): Redefine.
19624 (arm_option_restore): Restore string options.
19625 (arm_option_print): Print string options.
19627 2017-06-16 Martin Sebor <msebor@redhat.com>
19629 PR tree-optimization/80933
19630 PR tree-optimization/80934
19631 * builtins.c (fold_builtin_3): Do not handle bcmp here.
19632 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
19633 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
19634 (gimple_fold_builtin): Call them.
19636 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19638 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
19639 as unlikely; update profile.
19641 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19643 * predict.c (force_edge_cold): Handle declaring edges impossible
19646 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19648 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
19650 (try_unroll_loop_completely): Fix reporting.
19652 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19654 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
19656 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
19659 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
19660 if given a non-constant argument for an intrinsic which requires a
19663 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19665 * profile.c (compare_freqs): New function.
19666 (branch_prob): Sort edge list.
19667 (find_spanning_tree): Assume that the list is priority sorted.
19669 2017-06-16 Richard Biener <rguenther@suse.de>
19671 PR tree-optimization/81090
19672 * passes.def (pass_record_bounds): Remove.
19673 * tree-pass.h (make_pass_record_bounds): Likewise.
19674 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
19675 make_pass_record_bounds): Likewise.
19676 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
19677 not free niter estimates at the beginning but at the end.
19678 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
19680 2017-06-16 Richard Biener <rguenther@suse.de>
19682 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
19683 initializer to workaround ICE in host GCC 4.8.
19685 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19687 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
19689 (clone_inlined_nodes): Update.
19691 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19693 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
19694 prefetch settings, and enable prefetching by default at -O3.
19696 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19698 * config/aarch64/aarch64.c (aarch64_override_options_internal):
19699 Set flag_prefetch_loop_arrays according to tuning data.
19701 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19703 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
19704 New tune structure.
19705 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
19706 [Unrelated to main purpose of the patch] Place the pointer field last
19707 to enable type checking errors when tune structure are wrongly merged.
19708 * config/aarch64/aarch64.c (generic_prefetch_tune,)
19709 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
19710 (thunderx2t99_prefetch_tune): New tune constants.
19711 (tune_params *_tunings): Update all tunings (no functional change).
19712 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
19713 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
19714 from tunings structures.
19716 2017-06-16 Jakub Jelinek <jakub@redhat.com>
19719 * ubsan.c (instrument_null): Add T argument, use it instead
19720 of computing it based on IS_LHS.
19721 (instrument_object_size): Likewise.
19722 (pass_ubsan::execute): Adjust instrument_null and
19723 instrument_object_size callers to pass gimple_get_lhs or
19724 gimple_assign_rhs1 result to it. Use instrument_null instead of
19725 calling get_base_address and instrument_mem_ref. Handle
19726 aggregate call arguments for object-size sanitization.
19728 2017-06-16 Yury Gribov <tetra2005@gmail.com>
19730 PR tree-optimization/81089
19731 * tree-vrp.c (is_masked_range_test): Validate operands of
19734 2017-06-15 Martin Sebor <msebor@redhat.com>
19737 * dumpfile.c (dump_register): Avoid calling memset to initialize
19738 a class with a default ctor.
19739 * gcc.c (struct compiler): Remove const qualification.
19740 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
19741 * hash-table.h: Ditto.
19742 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
19744 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
19745 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
19747 * params.h (struct param_info): Make struct members non-const.
19748 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
19749 with default initialization.
19750 * vec.h (vec_copy_construct, vec_default_construct): New helper
19752 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
19753 with vec_copy_construct.
19754 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
19755 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
19756 * doc/invoke.texi (-Wclass-memaccess): Document.
19758 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19760 * emit-rtl.h (is_leaf): Update comment about local
19761 register allocator.
19763 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19766 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
19767 for a variable to have a section before checking if the section has a
19769 Set section to.persistent if persistent attribute is set.
19770 Warn if .persistent attribute is used on an automatic variable.
19772 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
19774 PR rtl-optimization/80474
19775 * reorg.c (update_block): Do not ignore instructions in a delay slot.
19777 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
19779 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
19782 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
19784 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
19785 (casesi): Emit bounds checking as RTL.
19786 (casesi_internal_mips16_<mode>): Remove bounds checking.
19788 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
19790 * config/xtensa/xtensa.c (xtensa_option_override): Append
19791 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
19792 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
19793 xtensa_doloop_hooks): Define unconditionally.
19794 (xtensa_reorg_loops): Only call reorg_loops in the presence of
19796 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
19797 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
19798 for it in xtensa_option_override.
19799 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
19800 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
19802 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
19804 * doc/cppopts.texi: Document '-' special value to -MF.
19806 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
19808 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
19809 (cortex_a53_fconst): Likewise.
19810 (cortex_a53_fpmul): Likewise.
19811 (cortex_a53_f_load_64): Likewise.
19812 (cortex_a53_f_load_many): Likewise.
19813 (cortex_a53_advsimd_alu): Likewise.
19814 (cortex_a53_advsimd_alu_q): Likewise.
19815 (cortex_a53_advsimd_mul): Likewise.
19816 (cortex_a53_advsimd_mul_q): Likewise.
19817 (fpmac bypass): Add new bypass for fpmac-fpmac case.
19818 Add missing fmul, r2f_cvt and fconst cases.
19820 2017-06-14 Richard Biener <rguenther@suse.de>
19822 PR middle-end/81088
19823 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
19825 (fold_binary_loc): When associating do not treat pre-existing
19826 TREE_OVERFLOW on literal constants as a reason to allow
19827 TREE_OVERFLOW on associated literal constants.
19829 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
19831 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
19832 (MASK_FEATURES): New macro.
19833 * config/sparc/sparc.c (sparc_option_override): Remove the special
19834 handling of -mfpu and generalize it to all MASK_FEATURES switches.
19836 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
19838 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
19839 a division of 0 if non-call exceptions are enabled.
19841 2017-06-14 Andrew Pinski <apinski@cavium.com>
19842 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19845 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
19846 Improve vector initialization code gen for only variable case.
19848 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
19850 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
19852 2017-06-14 Richard Biener <rguenther@suse.de>
19854 PR tree-optimization/81083
19855 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
19858 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19860 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
19861 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
19862 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
19863 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
19864 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
19865 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
19867 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19869 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
19870 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
19872 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19874 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
19876 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19878 * config/rs6000/t-rtems: Don't handle SPE.
19880 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19882 * config/rs6000/t-linux: Don't handle SPE.
19884 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19886 * config/rs6000/eabispe.h: Delete file.
19888 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19890 * config/rs6000/t-spe: Delete file.
19892 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19894 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
19895 (rs6000_legitimate_offset_address_p): Return false for anything in
19896 V2SImode or V2SFmode.
19898 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
19900 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
19901 except V2SF and V2SI. Rearrange the vector modes, and add comments.
19902 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
19904 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
19905 (rs6000_legitimate_offset_address_p): Ditto.
19906 (rs6000_emit_move): Ditto.
19907 (rs6000_init_builtins): Remove V4HI_type_node.
19909 2017-06-13 Martin Liska <mliska@suse.cz>
19912 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
19913 (gate_asan): Likewise.
19914 * asan.h (asan_no_sanitize_address_p): Remove the function.
19915 (sanitize_flags_p): New function.
19916 * builtins.def: Fix coding style.
19917 * common.opt: Use renamed enum value.
19918 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
19919 * doc/extend.texi: Document no_sanitize attribute.
19920 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
19921 to SANITIZE_UNDEFINED_NONDEFAULT.
19922 * gcc.c (sanitize_spec_function): Use the renamed enum value.
19923 * gimple-fold.c (optimize_atomic_compare_exchange_p):
19924 Use sanitize_flags_p.
19925 * gimplify.c (gimplify_function_tree): Likewise.
19926 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
19927 * opts.c (parse_no_sanitize_attribute): New function.
19928 (common_handle_option): Use renamed enum value.
19929 * opts.h (parse_no_sanitize_attribute): Declare.
19930 * tree.c (sanitize_flags_p): New function.
19931 * tree.h: Declared here.
19932 * tsan.c: Use sanitize_flags_p.
19933 * ubsan.c (ubsan_expand_null_ifn): Likewise.
19934 (instrument_mem_ref): Likewise.
19935 (instrument_bool_enum_load): Likewise.
19936 (do_ubsan_in_current_function): Remove the function.
19937 (pass_ubsan::execute): Use sanitize_flags_p.
19938 * ubsan.h: Remove do_ubsan_in_current_function
19939 * tree-cfg.c (print_no_sanitize_attr_value): New function.
19940 (dump_function_to_file): Use it here.
19942 2017-06-13 Martin Jambor <mjambor@suse.cz>
19944 PR tree-optimization/80803
19945 PR tree-optimization/81063
19946 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
19947 (propagate_subaccesses_across_link): Enqueue subtree whenever
19948 necessary instead of relying on the caller.
19950 2017-06-13 Martin Jambor <mjambor@suse.cz>
19952 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
19953 that have a first_link.
19954 (sort_and_splice_var_accesses): Do not check first_link before
19956 (subtree_mark_written_and_enqueue): Likewise.
19957 (propagate_all_subaccesses): Likewise and do not stop at first
19958 parent with a first_link.
19960 2017-06-13 Martin Jambor <mjambor@suse.cz>
19962 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
19965 2017-06-13 Yury Gribov <tetra2005@gmail.com>
19967 * match.pd: New pattern.
19969 2017-06-13 Yury Gribov <tetra2005@gmail.com>
19971 * tree-vrp.c (is_masked_range_test): New function.
19972 (register_edge_assert_for): Determine ranges for
19975 2017-06-13 Yury Gribov <tetra2005@gmail.com>
19977 PR tree-optimization/67328
19978 * fold-const.c (maskable_range_p): New function.
19979 (build_range_check): Generate bittests if possible.
19981 2017-06-13 Martin Liska <mliska@suse.cz>
19983 * gimple-pretty-print.c (dump_probability): Add new argument.
19984 (dump_edge_probability): Dump both probability and count.
19985 (dump_gimple_label): Likewise.
19986 (dump_gimple_bb_header): Likewise.
19988 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
19991 * config/avr/avr-devices.c: Fix indentation.
19992 * config/avr/gen-avr-mmcu-specs.c: Dito.
19994 2017-06-13 Richard Biener <rguenther@suse.de>
19996 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
19997 instead get vector type from stmt_info.
19998 (vectorizable_reduction): Adjust. Remove dead code.
20000 2017-06-13 Richard Biener <rguenther@suse.de>
20002 PR middle-end/81065
20003 * fold-const.c (extract_muldiv_1): Remove bogus distribution
20004 case of C * (x * C2 + C3).
20005 (fold_addr_of_array_ref_difference): Properly fold index difference.
20007 2017-06-12 David S. Miller <davem@davemloft.net>
20010 * config/sparc/sparc.md (return expander): Emit frame blockage if
20011 function uses alloca.
20013 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
20015 * combine.c (make_field_assignment): Check len rather than the mode
20016 precision when calling force_to_mode.
20018 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
20020 Support multilibs and devices that see flash in RAM address range.
20023 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
20024 (avr_mcu_t) <flash_pm_offset>: New field.
20025 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
20026 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
20027 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
20028 (AVR_TINY_PM_OFFSET): Remove macro.
20029 * config/avr/avr.opt (-mshort-calls): New option.
20030 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
20031 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
20032 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
20033 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
20034 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
20035 instead of avr_arch->have_jmp_call.
20036 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
20037 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
20038 avr_arch->flash_pm_offset to define.
20039 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
20040 new field flash_pm_offset. Add entry for avrxmega3.
20041 (avr_texinfo): Add entry for avrxmega3.
20042 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
20043 attiny212, attiny214,
20044 attiny412, attiny414, attiny416, attiny417,
20045 attiny814, attiny816, attiny817,
20046 attiny1614, attiny1616, attiny1617,
20047 attiny3214, attiny3216, attiny3217.
20048 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
20049 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
20050 (avr_print_operand_address) [AVR_TINY]: Same.
20051 (avr_asm_init_sections) <readonly_data_section>: Only patch
20052 callback if avr_arch->flash_pm_offset = 0.
20053 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
20054 for rodata if avr_arch->flash_pm_offset != 0.
20055 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
20056 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
20057 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
20058 (m_options): Append opt_rcall.
20059 (m_dirnames): Append dir_rcall.
20060 * config/avr/t-multilib: Regenerate.
20062 * configure.ac [target=avr]: Check whether avrxmega3 default
20063 linker description file works as needed.
20064 * configure: Regenerate.
20065 * doc/avr-mmcu.texi: Regenerate.
20066 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
20067 <__AVR_ARCH__>: Document avrxmega3 and 103.
20068 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
20069 <__AVR_SHORT_CALLS__>: Document it.
20070 <__AVR_PM_BASE_ADDRESS__>: Document it.
20071 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
20072 (AVR Variable Attributes) <progmem>: Document this is
20073 not needed for avrxmega3.
20074 (AVR Named Address Spaces) <__flash>: Dito.
20076 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
20078 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
20080 2017-06-12 Doug Rupp <rupp@adacore.com>
20082 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
20083 Append vxworks-stdint.h to the tm_file list.
20084 * config/vxworks-stdint.h: New file.
20086 2017-06-12 Martin Liska <mliska@suse.cz>
20088 PR tree-optimization/81041
20089 * tree-profile.c (gimple_gen_ic_func_profiler):
20090 Create an extra BB in profile-generate
20091 (gimple_gen_time_profiler): Likewise.
20093 2017-06-12 Jakub Jelinek <jakub@redhat.com>
20095 PR tree-optimization/81003
20096 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
20097 (update_range_test): Use it instead of force_gimple_operand_gsi.
20099 2017-06-12 Richard Biener <rguenther@suse.de>
20101 PR tree-optimization/81053
20102 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
20103 with backedge value not defined in loop. Simplify def stmt
20106 2017-06-11 Tom de Vries <tom@codesourcery.com>
20109 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
20111 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
20112 nvptx_cannot_force_const_mem.
20114 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20116 * opts.c (finish_options): Move test for flag_split_stack after
20117 it has been initialized.
20119 2017-06-11 Jason Merrill <jason@redhat.com>
20121 * tree.h (id_equal): New.
20122 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
20123 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
20124 instead of strcmp of IDENTIFIER_POINTER.
20126 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20128 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
20129 (mark_all_inlined_calls_cdtor): Fix formating.
20130 (inline_transform): Rescale profile before inlining.
20132 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20134 * cgraph.h (cgraph_edge::clone): Update prototype.
20135 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
20136 (cgraph_node::create_clone): Update.
20137 (cgraph_node::create_version_clone): Update.
20138 * tree-inline.c (copy_bb): Update.
20139 (expand_call_inline): Update.
20141 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
20143 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
20144 factored out from ...
20145 (rs6000_emit_prologue): ... here.
20147 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
20149 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
20150 factored out from ...
20151 (rs6000_emit_prologue): ... here.
20153 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20155 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
20157 (handle_missing_profiles): Fix computation of tp_first_run.
20158 (counts_to_freqs): Do not touch freqs when count is 0.
20160 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20162 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
20165 2017-06-10 Tom de Vries <tom@codesourcery.com>
20167 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
20168 attributes): Document signal effective target.
20170 2017-06-10 Tom de Vries <tom@codesourcery.com>
20172 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
20173 Document effective target stack_size.
20175 2017-06-09 David Malcolm <dmalcolm@redhat.com>
20177 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
20178 to the edit_context if they can be auto-applied.
20180 2017-06-9 Ian Lance Taylor <iant@golang.org>
20182 * opts.c (finish_options): If -fsplit-stack, disable implicit
20183 -forder-blocks-and-partition.
20184 * doc/invoke.texi (Optimize Options): Document that when using
20185 -fsplit-stack -forder-blocks-and-partition is not implicitly
20188 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20190 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
20191 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
20192 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
20193 * builtins.def (abort, trap, unreachable): Declare cold.
20194 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
20195 * tree-core.h (ECF_COLD): New.
20196 * tree.c (set_call_expr_flags): Handle ECF_COLD.
20197 (build_common_builtin_nodes): Mark unreachable and abort as cold.
20199 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20201 * predict.c (unlikely_executed_stmt_p): Cleanup.
20203 2017-06-09 Richard Biener <rguenther@suse.de>
20205 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
20206 model if the ref is always written to.
20208 2017-06-09 Tamar Christina <tamar.christina@arm.com>
20210 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
20212 2017-06-09 Tamar Christina <tamar.christina@arm.com>
20214 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
20217 2017-06-09 Tom de Vries <tom@codesourcery.com>
20220 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
20221 "target cannot support label values" when encountering LABEL_REF.
20223 2017-06-09 Martin Liska <mliska@suse.cz>
20225 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
20226 (gimple_gen_ic_func_profiler): Emit direct comparison
20227 of __gcov_indirect_call_callee with NULL.
20228 (gimple_gen_time_profiler): Change probability from
20229 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
20231 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20233 * profile.c (edge_gcov_counts): Turn to pointer.
20234 (compute_branch_probabilities, compute_branch_probabilities): Update.
20235 (branch_prob): Do not clear edge_gcov_count.
20236 * profile.h (edge_gcov_counts): Turn to pointer.
20237 (edge_gcov_count): Update.
20239 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20241 * gimple.h (gimple_check_failed): Mark cold.
20243 2017-06-09 Richard Biener <rguenther@suse.de>
20245 PR tree-optimization/66623
20246 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
20247 refactor check_reduction into two parts, properly computing
20248 whether we have to check reduction validity for outer loop
20251 2017-06-09 Richard Biener <rguenther@suse.de>
20253 PR tree-optimization/79483
20254 * graphite-scop-detection.c (order): New global.
20255 (get_order): Compute bb to order mapping that satisfies code
20256 generation constraints.
20257 (cmp_pbbs): New helper.
20258 (build_scops): Start domwalk at entry block, sort generated
20261 2017-06-09 Richard Biener <rguenther@suse.de>
20263 PR middle-end/81007
20264 * ipa-polymorphic-call.c
20265 (ipa_polymorphic_call_context::restrict_to_inner_class):
20266 Skip FIELD_DECLs with error_mark_node type.
20267 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
20270 2017-06-09 Martin Liska <mliska@suse.cz>
20272 * predict.c (struct branch_predictor): New struct.
20273 (test_prediction_value_range): New test.
20274 (predict_c_tests): New function.
20275 * selftest-run-tests.c (selftest::run_tests): Run the function.
20276 * selftest.h: Declare new tests.
20278 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
20281 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
20282 gen_add3_insn did not fail.
20283 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
20284 r0, construct that number in a temporary reg and add that reg to r0.
20285 If asked to put the result in r0 as well, fail.
20287 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
20289 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
20290 for early expansion of vec_eqv.
20292 2017-06-08 Jakub Jelinek <jakub@redhat.com>
20294 PR middle-end/81005
20295 * ubsan.c (instrument_null): Avoid pointless code temporary.
20296 (pass_ubsan::execute): Instrument aggregate arguments of calls.
20298 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
20302 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
20305 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
20308 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20310 * predict.c (unlikely_executed_edge_p): Move ahead.
20311 (probably_never_executed_edge_p): Use it.
20313 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
20315 PR middle-end/79988
20316 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
20317 gimple_call_builtin_p call.
20319 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20321 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
20322 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
20323 rtl_check_failed_type2, rtl_check_failed_code1,
20324 rtl_check_failed_code2, rtl_check_failed_code_mode,
20325 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
20326 rtvec_check_failed_bounds, rtl_check_failed_flag,
20327 _fatal_insn_not_found, _fatal_insn): Likewise.
20328 * tree.h (tree_contains_struct_check_failed,
20329 tree_check_failed, tree_not_check_failed,
20330 tree_class_check_failed, tree_range_check_failed,
20331 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
20332 tree_vec_elt_check_failed, phi_node_elt_check_failed,
20333 tree_operand_check_failed, omp_clause_check_failed,
20334 omp_clause_operand_check_failed, omp_clause_range_check_failed):
20337 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20339 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
20340 flag_branch_probabilities.
20341 * ipa-inline.c (edge_badness): Likewise.
20342 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
20343 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
20344 * predict.c (maybe_hot_frequency_p): Likewise.
20345 (probably_never_executed): Likewise.
20346 * sched-ebb.c (schedule_ebbs): Likewise.
20347 * sched-rgn.c (find_single_block_region): Likewise.
20348 * tracer.c (tail_duplicate): Likewise.
20350 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20352 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
20353 longer requires x_flag_profile_use.
20355 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20357 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
20358 instead of flag_reorder_blocks_and_partition.
20359 * dbxout.c (dbxout_function_end): Likewise.
20360 * dwarf2out.c (gen_subprogram_die): Likewise.
20361 * haifa-sched.c (sched_create_recovery_edges): Likewise.
20362 * hw-doloop.c (reorg_loops): Likewise.
20363 * varasm.c (assemble_start_function,
20364 assemble_end_function): Likewise.
20365 (decide_function_section): Do not check for
20366 flag_reorder_blocks_and_partition.
20368 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
20370 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
20372 (chkp_get_hard_register_fake_addr_expr): Ditto.
20373 (chkp_build_addr_expr): Add check for hard reg case.
20374 (chkp_parse_array_and_component_ref): Ditto.
20375 (chkp_find_bounds_1): Ditto.
20376 (chkp_process_stmt): Don't generate bounds store for
20379 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20381 * predict.c (maybe_hot_bb_p): Do not check profile status.
20382 (maybe_hot_edge_p): Likewise.
20383 (probably_never_executed): Check for zero counts even if profile
20385 (unlikely_executed_edge_p): New function.
20386 (unlikely_executed_stmt_p): New function.
20387 (unlikely_executed_bb_p): New function.
20388 (set_even_probabilities): Use unlikely predicates.
20389 (combine_predictions_for_bb): Likewise.
20390 (predict_paths_for_bb): Likewise.
20391 (predict_paths_leading_to_edge): Likewise.
20392 (determine_unlikely_bbs): New function.
20393 (estimate_bb_frequencies): Use it.
20394 (compute_function_frequency): Use zero counts even if profile is
20396 * profile-count.h: Fix typo.
20398 2017-08-08 Julia Koval <julia.koval@intel.com>
20400 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
20401 _mm512_mask_cvtsepi16_storeu_epi8,
20402 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
20403 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
20404 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
20405 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
20406 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
20407 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
20408 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
20409 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
20410 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
20411 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
20412 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
20413 __builtin_ia32_pmovuswb256mem_mask,
20414 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
20415 __builtin_ia32_pmovwb512mem_mask): New builtins.
20417 2017-08-08 Julia Koval <julia.koval@intel.com>
20419 PR target/73350,80862
20420 * config/i386/subst.md (round): Fix round pattern.
20421 * config/i386/i386.c (ix86_erase_embedded_rounding):
20422 Fix erasing rounding for the fixed pattern.
20424 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20426 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
20428 2017-06-08 Martin Liska <mliska@suse.cz>
20430 PR gcov-profile/80911
20431 * gcov.c (block_info::block_info): New constructor.
20433 2017-06-07 Carl Love <cel@us.ibm.com>
20435 * config/rs6000/rs6000-c: The return type of the following
20436 built-in functions was implemented as int not long long. Fix sign
20437 of return value for the unsigned version of vec_mulo and vec_mule.
20438 vector unsigned long long vec_bperm (vector unsigned long long,
20439 vector unsigned char)
20440 vector signed long long vec_mule (vector signed int,
20442 vector unsigned long long vec_mule (vector unsigned int,
20443 vector unsigned int)
20444 vector signed long long vec_mulo (vector signed int,
20446 vector unsigned long long vec_mulo (vector unsigned int,
20447 vector unsigned int)
20448 * doc/extend.texi: Fix the documentation for the built-in
20451 2017-06-07 Carl Love <cel@us.ibm.com>
20454 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
20457 2017-06-07 Carl Love <cel@us.ibm.com>
20459 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
20460 support, Generate doublehv for signed int/float for BE case only.
20462 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
20464 * doc/invoke.texi (mcx16): Rewrite.
20466 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20468 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
20469 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
20470 *mov<mode>_softfloat, and an anonymous splitter): Use
20471 nonimmediate_operand instead of rs6000_nonimmediate_operand.
20473 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20475 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
20477 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
20478 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
20479 (rs6000_debug_reg_global): Adjust.
20480 (rs6000_init_hard_regno_mode_ok): Adjust.
20481 (rs6000_dbx_register_number): Adjust.
20482 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
20483 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
20484 Remove SPE_ACC and SPEFSCR.
20485 (REG_ALLOC_ORDER): Ditto.
20486 (FRAME_POINTER_REGNUM): Change to 111.
20487 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
20488 (REG_CLASS_NAMES): Ditto.
20489 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
20490 (REGISTER_NAMES): Ditto.
20491 (ADDITIONAL_REG_NAMES): Ditto.
20492 (rs6000_reg_names): Ditto.
20493 * config/rs6000/rs6000.md: Renumber some register number
20496 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20498 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
20500 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
20501 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
20503 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
20504 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
20505 Delete the SPE high registers.
20506 (REG_ALLOC_ORDER): Ditto.
20507 (enum reg_class): Remove SPE_HIGH_REGS.
20508 (REG_CLASS_NAMES): Ditto.
20509 (REG_CLASS_CONTENTS): Delete the SPE high registers.
20510 (REGISTER_NAMES): Ditto.
20511 (rs6000_reg_names): Ditto.
20512 * doc/tm.texi.in: Remove SPE as example.
20513 * doc/tm.texi: Regenerate.
20515 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20517 * config/rs6000/8540.md (ppc8540_brinc): Delete.
20518 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
20519 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
20520 * config/rs6000/rs6000.md (type): Remove "brinc".
20522 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20524 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
20525 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
20526 * config/rs6000/linuxspe.h: Delete file.
20527 * config/rs6000/rs6000.md: Don't include spe.md.
20528 * config/rs6000/spe.h: Delete file.
20529 * config/rs6000/spe.md: Delete file.
20530 * config/rs6000/t-rs6000: Remove spe.md.
20532 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20534 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
20535 (reg_or_none500mem_operand): Delete.
20536 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
20537 instead of reg_or_none500mem_operand.
20539 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20541 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
20542 handling of SPE flags.
20543 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
20545 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20547 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
20549 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
20550 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
20551 paired_divv2sf3): Similar.
20552 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
20553 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
20554 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
20556 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
20557 Rename the paired_* instruction patterns.
20558 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
20560 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
20561 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
20562 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
20563 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
20564 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
20565 PAIRED_VECTOR_MODE.
20566 (struct machine_function): Delete field spe_insn_chain_scanned_p.
20567 (spe_func_has_64bit_regs_p): Delete.
20568 (spe_expand_predicate_builtin): Delete.
20569 (spe_expand_evsel_builtin): Delete.
20570 (TARGET_DWARF_REGISTER_SPAN): Do not define.
20571 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
20572 (invalid_e500_subreg): Delete.
20573 (rs6000_legitimize_address): Always force_reg op2 as well, for
20574 paired single memory accesses.
20575 (rs6000_member_type_forces_blk): Delete.
20576 (rs6000_spe_function_arg): Delete.
20577 (rs6000_expand_unop_builtin): Delete SPE handling.
20578 (rs6000_expand_binop_builtin): Ditto.
20579 (spe_expand_stv_builtin): Delete.
20580 (bdesc_2arg_spe): Delete.
20581 (spe_expand_builtin): Delete.
20582 (spe_expand_predicate_builtin): Delete.
20583 (spe_expand_evsel_builtin): Delete.
20584 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
20585 (spe_init_builtins): Delete.
20586 (spe_func_has_64bit_regs_p): Delete.
20587 (savres_routine_name): Delete "info" parameter. Adjust callers.
20588 (rs6000_emit_stack_reset): Ditto.
20589 (rs6000_dwarf_register_span): Delete.
20590 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
20591 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
20592 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
20594 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
20596 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
20597 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
20598 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
20599 mulv2sf3, divv2sf3): Delete expanders.
20601 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20603 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
20605 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20607 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
20608 * config/rs6000/rs6000.c: Ditto.
20610 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20612 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
20613 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
20614 comparison_operator.
20616 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20618 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
20619 * config/rs6000/rs6000.opt: Ditto.
20620 * config/rs6000/t-rtems: Ditto.
20622 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20624 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
20625 TARGET_E500_SINGLE by 0, simplify.
20626 * config/rs6000/rs6000.c: Ditto.
20627 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
20628 (spe_build_register_parallel): Delete.
20629 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
20630 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
20631 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
20632 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
20633 (E500_CONVERT): Delete.
20634 * config/rs6000/spe.md: Remove many patterns and all define_constants.
20636 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20638 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
20639 * config/rs6000/dfp.md: Ditto.
20640 (negdd2, *negdd2_fpr): Merge.
20641 (absdd2, *absdd2_fpr): Merge.
20642 (negtd2, *negtd2_fpr): Merge.
20643 (abstd2, *abstd2_fpr): Merge.
20644 * config/rs6000/e500.h: Delete file.
20645 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
20646 TARGET_FPRS by 1 and simplify.
20647 * config/rs6000/rs6000-c.c: Ditto.
20648 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
20649 TARGET_DF_SPE by 0.
20650 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
20652 * config/rs6000/rs6000.md: Ditto.
20653 (floatdidf2, *floatdidf2_fpr): Merge.
20654 (move_from_CR_gt_bit): Delete.
20655 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
20656 (E500_CR_IOR_COMPARE): Delete.
20657 (All patterns that require !TARGET_FPRS): Delete.
20658 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
20660 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20662 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
20664 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20666 * graphds.c (add_edge): Intitialize edge's attached data.
20667 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
20668 pointer parameter. Call pointed function on each edge during
20669 graph traversing. Skip traversing the edge when the function
20671 (graphds_dfs, graphds_scc): Ditto.
20672 (for_each_edge): New parameter. Pass the new parameter to callback
20674 * graphds.h (skip_edge_callback): New function pointer type.
20675 (graphds_dfs, graphds_scc): New function pointer parameter.
20676 (graphds_edge_callback, for_each_edge): New parameter.
20678 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20680 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
20681 out code checking if runtime alias check is possible to below ...
20682 Call the new function.
20683 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
20684 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
20686 2017-06-07 Marek Polacek <polacek@redhat.com>
20689 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
20690 TYPE_OVERFLOW_WRAPS check.
20692 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20694 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
20695 if versioning is required.
20696 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
20697 peeling with the check for versioning.
20699 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20701 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
20702 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
20703 Set true to new parameter if new ssa variable is defined.
20704 (vect_gen_vector_loop_niters): Refactor. Set range information
20705 for the new vector loop bound variable.
20706 (vect_do_peeling): Ditto.
20708 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20710 * tree-affine.c (ssa.h): Include header file.
20711 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
20712 has wrapping overflow behavior.
20714 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20716 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
20718 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20720 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
20721 (tree_to_aff_combination): ... here.
20723 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20725 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
20726 reg_pressure model function.
20727 (ivopts_global_cost_for_size): Delete.
20728 (determine_set_costs, iv_ca_recount_cost): Call new model function
20729 ivopts_estimate_reg_pressure.
20731 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20733 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
20734 expensive than udiv. Remove floating point cases from mod.
20736 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20738 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
20739 Increase idiv cost.
20741 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20743 * config/aarch64/aarch64.md
20744 (copysignsf3): Fix mask generation.
20746 2017-06-07 Jakub Jelinek <jakub@redhat.com>
20748 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
20750 (class dump_manager): Add register_dumps method.
20751 * dumpfile.c: Include langhooks.h.
20752 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
20753 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
20754 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
20755 (dump_manager::dump_register): Start with 512 entries instead of 32.
20756 (dump_manager::register_dumps): New method.
20757 * toplev.c (general_init): Instead of invoking register_dumps
20758 langhook, invoke register_dumps method on the dump manager.
20759 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
20762 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
20764 * doc/md.texi: Clarify the restrictions on a define_insn condition.
20765 Say that # requires an associated define_split to exist, and that
20766 the define_split must be suitable for use after register allocation.
20768 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20770 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
20771 (compute_outgoing_frequencies): Also initialize zero counts.
20772 (find_many_sub_basic_blocks): Do not produce uninitialized profile
20773 around loops; preserve more of profile when nothing changes.
20775 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
20777 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
20779 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
20780 * config/arm/arm-cpu-cdata.h: Regenerate.
20781 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
20782 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
20783 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
20784 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
20785 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
20787 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
20789 * config/arm/t-rmprofile: Likewise.
20790 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
20792 2017-06-06 David S. Miller <davem@davemloft.net>
20795 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
20796 blockage if function uses alloca.
20798 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20800 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
20801 New "uid" fields to hold pretty-print IDs of group and ref.
20802 Memory references are now identified as <group_id>:<ref_id>
20803 instead of using [random] addresses.
20804 (dump_mem_details): Simplify, no functional change.
20805 (dump_mem_ref): Simplify and make output more concise.
20806 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
20807 (find_or_create_group): Initialize group uid.
20808 (record_ref): Initialize ref uid. Improve debug output.
20809 (prune_group_by_reuse, should_issue_prefetch_p,)
20810 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
20811 (mark_nontemporal_store, determine_loop_nest_reuse):
20812 Improve debug output.
20814 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20816 * dbgcnt.def (prefetch): New debug counter.
20817 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
20818 (schedule_prefetches): Stop issueing prefetches if debug counter
20821 2017-06-06 Tom de Vries <tom@codesourcery.com>
20823 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
20824 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
20826 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20828 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
20829 Use aarch64_reg_or_zero predicate for operand 4.
20830 (aarch64_compare_and_swap<mode> define_insn_and_split):
20831 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
20832 (aarch64_store_exclusive<mode>): Likewise for operand 2.
20834 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
20836 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
20837 (arm_compute_save_core_reg_mask): This.
20838 (thumb1_compute_save_reg_mask): Rename into ...
20839 (thumb1_compute_save_core_reg_mask): This.
20840 (arm_compute_save_reg0_reg12_mask): Adapt comment.
20841 (arm_compute_frame_layout): Likewise.
20843 2017-06-06 Richard Biener <rguenther@suse.de>
20845 PR tree-optimization/80974
20846 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
20847 keep or clear leaders SSA info.
20849 2017-06-06 Tom de Vries <tom@codesourcery.com>
20851 * config/nvptx/nvptx.c (split_mode_p): New function.
20852 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
20854 2017-06-06 Tom de Vries <tom@codesourcery.com>
20856 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
20858 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20861 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
20864 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20866 * shrink-wrap.c (handle_simple_exit): Update profile.
20867 (try_shrink_wrapping): Upate profile.
20869 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20871 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
20872 (tree_guess_outgoing_edge_probabilities): New.
20873 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
20874 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
20876 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20878 * ipa-split.c (split_function): Initialize return bb profile.
20880 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20882 * profile.c (compute_branch_probabilities): Also initialize
20883 EXIT_BLOCK profile.
20885 2017-06-06 Richard Biener <rguenther@suse.de>
20887 PR tree-optimization/80928
20888 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
20889 (vect_analyze_loop_operations): Properly guard analysis for
20891 (vect_transform_loop): Likewise.
20892 (vect_analyze_loop_2): Also reset SLP type on PHIs.
20893 (vect_model_induction_cost): Do not cost for pure SLP.
20894 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
20895 of induction in inner loop vectorization.
20896 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
20897 (vect_get_and_check_slp_defs): Handle vect_induction_def.
20898 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
20900 (vect_analyze_slp_cost_1): Cost induction.
20901 (vect_detect_hybrid_slp_stmts): Handle PHIs.
20902 (vect_get_slp_vect_defs): Likewise.
20903 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
20904 (vect_transform_stmt): Handle SLP reductions.
20905 * tree-vectorizer.h (vectorizable_induction): Adjust.
20907 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
20909 * config/rs6000/rs6000.c (make_resolver_func): Update
20910 init_lowered_empty_function call.
20912 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
20914 * doc/invoke.texi: Document the -fprofile-abs-path option.
20915 * common.opt (fprofile-abs-path): New option.
20916 * gcov-io.h (gcov_write_filename): Declare.
20917 * gcov-io.c (gcov_write_filename): New function.
20918 * coverage.c (coverage_begin_function): Use gcov_write_filename.
20919 * profile.c (output_location): Likewise.
20921 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20923 * shring-wrap.c: Revert accidental commit.
20925 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
20927 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
20929 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20931 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
20933 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
20934 profile in callgraph edge.
20935 * profile-count.h (apply_probability): If THIS is 0, then result is 0
20936 (apply_scale): Likewise.
20937 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
20938 Also scale profile when inlining function with zero profile.
20939 (initialize_cfun): Update exit block profile even when it is zero.
20940 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
20941 when profile is read.
20943 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
20945 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
20946 (CLONE_*): New constants to define the processors we can generate
20947 code for with the target_clone attribute.
20948 (rs6000_clone_map): New array to identify which clone processors
20949 the current program is running on.
20950 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
20951 target_clone attribute.
20952 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
20953 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
20954 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
20955 (cpu_expand_builtin): Add support for target_clone attribute.
20956 (rs6000_valid_attribute_p): Allow "default" attribute.
20957 (get_decl_name): New debug function to simplify printing the
20958 current function name in debugging statements.
20959 (rs6000_clone_priority): New functions to support the target_clone
20960 attribute, and be able to generate code to switch between ISA 2.05
20961 through ISA 3.0 (power6 through power9).
20962 (rs6000_compare_version_priority): Likewise.
20963 (rs6000_get_function_versions_dispatcher): Likewise.
20964 (make_resolver_func): Likewise.
20965 (add_condition_to_bb): Likewise.
20966 (dispatch_function_versions): Likewise.
20967 (rs6000_generate_version_dispatcher_body): Likewise.
20968 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
20969 (fusion_gpr_load_p): Fix a spacing issue.
20970 * doc/extend.texi (Common Function Attributes): Document that the
20971 PowerPC supports the target_clone attribute.
20973 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
20975 * config/arm/arm.h: explain F symbol found in description of ARM
20976 register allocation in its legend.
20978 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20980 * config/mips/frame-header-opt.c: Include profile-count.h.
20981 * config/riscv/riscv.c: Include profile-count.h
20983 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20985 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
20987 (sm_set_flag_if_changed): Add bbs field.
20988 (execute_sm_if_changed_flag_set): Pass BBS.
20989 (execute_sm): Update.
20991 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20993 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
20996 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20998 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
20999 (peephole2): New peephole2 to emit the above.
21000 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
21002 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21004 * config/aarch64/aarch64.c (define_peephole2 above
21005 *sub_<shift>_<mode>): New peephole.
21007 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21009 * config/i386/i386.c (make_resolver_func): Update.
21010 * Makefile.in: Add profile-count.h and profile-count.o
21011 * auto-profile.c (afdo_indirect_call): Update to new API.
21012 (afdo_set_bb_count): Update.
21013 (afdo_propagate_edge): Update.
21014 (afdo_propagate_circuit): Update.
21015 (afdo_calculate_branch_prob): Update.
21016 (afdo_annotate_cfg): Update.
21017 * basic-block.h: Include profile-count.h
21018 (struct edge_def): Turn count to profile_count.
21019 (struct basic_block_def): Likewie.
21020 (REG_BR_PROB_BASE): Move to profile-count.h
21021 (RDIV): Move to profile-count.h
21022 * bb-reorder.c (max_entry_count): Turn to profile_count.
21023 (find_traces): Update.
21024 (rotate_loop):Update.
21025 (connect_traces):Update.
21026 (sanitize_hot_paths):Update.
21027 * bt-load.c (migrate_btr_defs): Update.
21028 * cfg.c (RDIV): Remove.
21029 (init_flow): Use alloc_block.
21030 (alloc_block): Uninitialize count.
21031 (unchecked_make_edge): Uninitialize count.
21032 (check_bb_profile): Update.
21033 (dump_edge_info): Update.
21034 (dump_bb_info): Update.
21035 (update_bb_profile_for_threading): Update.
21036 (scale_bbs_frequencies_int): Update.
21037 (scale_bbs_frequencies_gcov_type): Update.
21038 (scale_bbs_frequencies_profile_count): New.
21039 * cfg.h (update_bb_profile_for_threading): Update.
21040 (scale_bbs_frequencies_profile_count): Declare.
21041 * cfgbuild.c (compute_outgoing_frequencies): Update.
21042 (find_many_sub_basic_blocks): Update.
21043 * cfgcleanup.c (try_forward_edges): Update.
21044 (try_crossjump_to_edge): Update.
21045 * cfgexpand.c (expand_gimple_tailcall): Update.
21046 (construct_exit_block): Update.
21047 * cfghooks.c (verify_flow_info): Update.
21048 (dump_bb_for_graph): Update.
21049 (split_edge): Update.
21050 (make_forwarder_block): Update.
21051 (duplicate_block): Update.
21052 (account_profile_record): Update.
21053 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
21054 (get_estimated_loop_iterations): Update.
21055 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
21056 (single_likely_exit): Update.
21057 * cfgloopmanip.c (scale_loop_profile): Update.
21059 (set_zero_probability): Update.
21060 (lv_adjust_loop_entry_edge): Update.
21061 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
21062 (purge_dead_edges): Update.
21063 (rtl_account_profile_record): Update.
21064 * cgraph.c (cgraph_node::create): Uninitialize count.
21065 (symbol_table::create_edge): Uninitialize count.
21066 (cgraph_update_edges_for_call_stmt_node): Update.
21067 (cgraph_edge::dump_edge_flags): Update.
21068 (cgraph_node::dump): Update.
21069 (cgraph_edge::maybe_hot_p): Update.
21070 * cgraph.h: Include profile-count.h
21071 (create_clone), create_edge, create_indirect_edge): Update.
21072 (cgraph_node): Turn count to profile_count.
21073 (cgraph_edge0: Likewise.
21074 (make_speculative, clone): Update.
21075 (create_edge): Update.
21076 (init_lowered_empty_function): Update.
21077 * cgraphclones.c (cgraph_edge::clone): Update.
21078 (duplicate_thunk_for_node): Update.
21079 (cgraph_node::create_clone): Update.
21080 * cgraphunit.c (cgraph_node::analyze): Update.
21081 (cgraph_node::expand_thunk): Update.
21082 * final.c (dump_basic_block_info): Update.
21083 * gimple-streamer-in.c (input_bb): Update.
21084 * gimple-streamer-out.c (output_bb): Update.
21085 * graphite.c (print_global_statistics): Update.
21086 (print_graphite_scop_statistics): Update.
21087 * hsa-brig.c: Include basic-block.h.
21088 * hsa-dump.c: Include basic-block.h.
21089 * hsa-gen.c (T sum_slice): Update.
21090 (convert_switch_statements):Update.
21091 * hsa-regalloc.c: Include basic-block.h.
21092 * ipa-chkp.c (chkp_produce_thunks): Update.
21093 * ipa-cp.c (struct caller_statistics): Update.
21094 (init_caller_stats): Update.
21095 (gather_caller_stats): Update.
21096 (ipcp_cloning_candidate_p): Update.
21097 (good_cloning_opportunity_p): Update.
21098 (get_info_about_necessary_edges): Update.
21099 (dump_profile_updates): Update.
21100 (update_profiling_info): Update.
21101 (update_specialized_profile): Update.
21102 (perhaps_add_new_callers): Update.
21103 (decide_about_value): Update.
21104 (ipa_cp_c_finalize): Update.
21105 * ipa-devirt.c (struct odr_type_warn_count): Update.
21106 (struct decl_warn_count): Update.
21107 (struct final_warning_record): Update.
21108 (possible_polymorphic_call_targets): Update.
21109 (ipa_devirt): Update.
21110 * ipa-fnsummary.c (redirect_to_unreachable): Update.
21111 * ipa-icf.c (sem_function::merge): Update.
21112 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
21113 * ipa-inline.c (compute_uninlined_call_time): Update.
21114 (compute_inlined_call_time): Update.
21115 (want_inline_small_function_p): Update.
21116 (want_inline_self_recursive_call_p): Update.
21117 (edge_badness): Update.
21118 (lookup_recursive_calls): Update.
21119 (recursive_inlining): Update.
21120 (inline_small_functions): Update.
21121 (dump_overall_stats): Update.
21122 (dump_inline_stats): Update.
21123 * ipa-profile.c (ipa_profile_generate_summary): Update.
21124 (ipa_propagate_frequency): Update.
21125 (ipa_profile): Update.
21126 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
21127 * ipa-utils.c (ipa_merge_profiles): Update.
21128 * loop-doloop.c (doloop_modify): Update.
21129 * loop-unroll.c (report_unroll): Update.
21130 (unroll_loop_runtime_iterations): Update.
21131 * lto-cgraph.c (lto_output_edge): Update.
21132 (lto_output_node): Update.
21133 (input_node): Update.
21134 (input_edge): Update.
21135 (merge_profile_summaries): Update.
21136 * lto-streamer-in.c (input_cfg): Update.
21137 * lto-streamer-out.c (output_cfg): Update.
21138 * mcf.c (create_fixup_graph): Update.
21139 (adjust_cfg_counts): Update.
21140 (sum_edge_counts): Update.
21141 * modulo-sched.c (sms_schedule): Update.
21142 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
21143 * predict.c (maybe_hot_count_p): Update.
21144 (probably_never_executed): Update.
21145 (dump_prediction): Update.
21146 (combine_predictions_for_bb): Update.
21147 (propagate_freq): Update.
21148 (handle_missing_profiles): Update.
21149 (counts_to_freqs): Update.
21150 (rebuild_frequencies): Update.
21151 (force_edge_cold): Update.
21152 * predict.h: Include profile-count.h
21153 (maybe_hot_count_p, counts_to_freqs): UPdate.
21154 * print-rtl-function.c: Do not include cfg.h
21155 * print-rtl.c: Include basic-block.h
21156 * profile-count.c: New file.
21157 * profile-count.h: New file.
21158 * profile.c (is_edge_inconsistent): Update.
21159 (correct_negative_edge_counts): Update.
21160 (is_inconsistent): Update.
21161 (set_bb_counts): Update.
21162 (read_profile_edge_counts): Update.
21163 (compute_frequency_overlap): Update.
21164 (compute_branch_probabilities): Update; Initialize and deinitialize
21166 (branch_prob): Update.
21167 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
21168 (edge_gcov_count): New.
21169 (bb_gcov_count): New.
21170 * shrink-wrap.c (try_shrink_wrapping): Update.
21171 * tracer.c (better_p): Update.
21172 * trans-mem.c (expand_transaction): Update.
21173 (ipa_tm_insert_irr_call): Update.
21174 (ipa_tm_insert_gettmclone_call): Update.
21175 * tree-call-cdce.c: Update.
21176 * tree-cfg.c (gimple_duplicate_sese_region): Update.
21177 (gimple_duplicate_sese_tail): Update.
21178 (gimple_account_profile_record): Update.
21179 (execute_fixup_cfg): Update.
21180 * tree-inline.c (copy_bb): Update.
21181 (copy_edges_for_bb): Update.
21182 (initialize_cfun): Update.
21183 (freqs_to_counts): Update.
21184 (copy_cfg_body): Update.
21185 (expand_call_inline): Update.
21186 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
21187 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
21188 (try_unroll_loop_completely): Update.
21189 (try_peel_loop): Update.
21190 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
21191 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
21192 * tree-ssa-loop-split.c (connect_loops): Update.
21193 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
21194 * tree-ssa-reassoc.c (branch_fixup): Update.
21195 * tree-ssa-tail-merge.c (replace_block_by): Update.
21196 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
21197 (compute_path_counts): Update.
21198 (update_profile): Update.
21199 (recompute_probabilities): Update.
21200 (update_joiner_offpath_counts): Update.
21201 (estimated_freqs_path): Update.
21202 (freqs_to_counts_path): Update.
21203 (clear_counts_path): Update.
21204 (ssa_fix_duplicate_block_edges): Update.
21205 (duplicate_thread_path): Update.
21206 * tree-switch-conversion.c (case_bit_test_cmp): Update.
21207 (struct switch_conv_info): Update.
21208 * tree-tailcall.c (decrease_profile): Update.
21209 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
21210 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
21211 * value-prof.c (check_counter): Update.
21212 (gimple_divmod_fixed_value): Update.
21213 (gimple_mod_pow2): Update.
21214 (gimple_mod_subtract): Update.
21215 (gimple_ic_transform): Update.
21216 (gimple_stringop_fixed_value): Update.
21217 * value-prof.h (gimple_ic): Update.
21219 2017-06-02 Carl Love <cel@us.ibm.com>
21221 * config/rs6000/rs6000-c: Add support for built-in functions
21222 vector double vec_doublee (vector signed int);
21223 vector double vec_doublee (vector unsigned int);
21224 vector double vec_doublee (vector float);
21225 vector double vec_doubleh (vector signed int);
21226 vector double vec_doubleh (vector unsigned int);
21227 vector double vec_doubleh (vector float);
21228 vector double vec_doublel (vector signed int);
21229 vector double vec_doublel (vector unsigned int);
21230 vector double vec_doublel (vector float);
21231 vector double vec_doubleo (vector signed int);
21232 vector double vec_doubleo (vector unsigned int);
21233 vector double vec_doubleo (vector float);.
21234 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
21235 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
21237 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
21238 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
21239 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
21241 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
21242 vec_doublel, vec_doubleh.
21243 * doc/extend.texi: Update the built-in documentation file for the
21244 new built-in functions.
21246 2017-06-02 David Malcolm <dmalcolm@redhat.com>
21249 * ipa-inline-analysis.c (free_growth_caches): Set
21250 edge_removal_hook_holder to NULL after removing it.
21252 2017-06-02 Sudakshina Das <sudi.das@arm.com>
21254 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
21255 comparision with zero.
21257 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
21258 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21259 for early expansion of vec_min and vec_max builtins.
21260 (builtin_function_type): Add min/max unsigned variants to those
21261 identified as having unsigned arguments.
21263 2017-06-02 Olivier Hainque <hainque@adacore.com>
21265 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
21267 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21269 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
21270 Use VALL_F16 iterator rather than VALL.
21272 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21274 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
21275 Emit CBNZ inside loop when doing a strong exchange and comparing
21276 against zero. Generate the CC flags after the loop.
21278 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
21280 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
21281 (dl_section_ref): New.
21282 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
21283 On AIX, append an expression to subtract the size of the
21284 section length to dl_section_ref.
21286 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
21288 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21289 for early expansion of vector absolute builtins.
21291 2017-06-02 Richard Biener <rguenther@suse.de>
21293 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
21294 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
21296 2017-06-02 Richard Biener <rguenther@suse.de>
21298 PR tree-optimization/80948
21299 * tree-tailcall.c (find_tail_calls): Track stmts to move in
21300 stmt order as well.
21302 2017-06-02 Richard Biener <rguenther@suse.de>
21304 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
21306 * tree-vect-stmts.c (process_use): Do not mark backedge defs
21307 for inductions as relevant.
21309 2017-06-02 Richard Biener <rguenther@suse.de>
21311 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
21312 (vectorizable_induction): ... this. Remove dead code.
21314 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
21316 * builtins. (expand_builtin_alloca): Remove second parameter and
21317 infer its value from the first parameter instead.
21318 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
21320 2017-06-02 Jakub Jelinek <jakub@redhat.com>
21322 PR rtl-optimization/80903
21323 * loop-doloop.c (add_test): Unshare sequence.
21325 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
21327 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
21329 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
21331 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
21333 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
21334 xlogue_layout::get_instance, logue_layout::xlogue_layout,
21335 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
21336 (xlogue_layout::get_stub_rtx): Make static.
21337 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
21338 (xlogue_layout::compute_stub_managed_regs): Rename to...
21339 (xlogue_layout::count_stub_managed_regs): ...this.
21340 (xlogue_layout::is_stub_managed_reg): New function.
21341 (xlogue_layout::m_stub_names): Rename to...
21342 (xlogue_layout::s_stub_names): ...this, make static.
21343 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
21344 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
21345 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
21346 xlogue_layout::s_stub_names): Instantiate statics.
21347 (stub_managed_regs): Remove.
21348 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
21349 (disable_call_ms2sysv_xlogues): Rename to...
21350 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
21351 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
21353 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
21354 change after reload_completed.
21355 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
21357 (ix86_expand_prologue): Likewise.
21358 (ix86_expand_epilogue): Likewise.
21359 (ix86_expand_split_stack_prologue): Likewise.
21360 (ix86_compute_frame_layout): Remove frame parameter ...
21361 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
21362 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
21364 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
21365 (ix86_frame): Move from here ...
21366 * config/i386/i386.h (ix86_frame): ... to here.
21367 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
21368 complete ix86_frame data structure instead. Remove some_ld_name.
21370 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
21372 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
21373 symbols that hold a DECL_VALUE_EXPR.
21375 2017-06-01 Martin Jambor <mjambor@suse.cz>
21377 PR tree-optimization/80898
21378 * tree-sra.c (process_subtree_disqualification): Removed.
21379 (disqualify_candidate): Do not acll
21380 process_subtree_disqualification.
21381 (subtree_mark_written_and_enqueue): New function.
21382 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
21383 RHS has been disqualified and re-queue LHS if necessary. Apart
21384 from that, ignore disqualified RHS.
21386 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21388 * config/s390/s390.c (s390_emit_epilogue): Disable early return
21389 address fetch for z10 or later.
21391 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21393 * config/arc/arc.md (tst_movb): Add guard when splitting.
21395 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21397 * config/arc/arc.c (arc_can_eliminate): Test against
21398 arc_frame_pointer_needed.
21400 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21402 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
21403 to prevent store reordering.
21404 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
21405 (type): Add block type.
21406 (stack_tie): Define special instruction to be used in
21409 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21411 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
21412 constraint. It is not valid for the pattern.
21413 (noncommutative_binary_comparison): Likewise.
21415 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21417 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
21420 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21422 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
21423 be used by the reg-alloc.
21425 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21427 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
21428 reg-alloc when having mul64 or mul32x16 instructions.
21429 (mulsidi3): Likewise.
21430 (umulsidi3): Likewise.
21431 (mulsi32x16): New pattern.
21432 (mulsi64): Likewise.
21433 (mulsidi64): Likewise.
21434 (umulsidi64): Likewise.
21435 (MUL32x16_REG): Define.
21436 (mul64_600): Use MUL32x16_REG.
21437 (mac64_600): Likewise.
21438 (umul64_600): Likewise.
21439 (umac64_600): Likewise.
21441 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21443 * config/arc/arc.md (mulsi3_700): Make it commutative.
21445 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
21447 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
21449 (*sign_extendsidi2_insn): Likewise for movstosw.
21451 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
21453 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
21454 the type of the input discriminant value. Convert the
21455 discriminant value of signedness vary.
21457 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
21459 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
21460 Add to -Wall section.
21462 2017-06-01 Richard Biener <rguenther@suse.de>
21464 PR middle-end/66313
21465 * fold-const.c (fold_plusminus_mult_expr): If the factored
21466 factor may be zero use a wrapping type for the inner operation.
21467 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
21468 and handle moved defs.
21469 (process_assignment): Properly guard the unary op case. Return a
21470 tri-state indicating that moving the stmt before the call may allow
21471 to continue. Pass through to_move.
21472 (find_tail_calls): Handle moving unrelated defs before
21475 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
21478 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
21479 splitter result in the canonical way.
21481 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
21483 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
21484 also for 32bit target. Update insn attributes.
21485 (zero-extendsidi2 splitter): Allow all registers for operand 1.
21487 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
21489 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
21490 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
21491 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
21492 (_mm_maskz_min_ss): New intrinsics.
21494 2017-05-31 Martin Liska <mliska@suse.cz>
21496 * tree-vect-loop.c (vect_create_epilog_for_reduction):
21497 Change comment style to one we normally use.
21498 (vectorizable_reduction): Likewise.
21499 (vectorizable_induction): Likewise.
21500 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
21501 (vectorizable_call): Likewise.
21502 (vectorizable_simd_clone_call): Likewise.
21503 (vectorizable_conversion): Likewise.
21504 (vectorizable_assignment): Likewise.
21505 (vectorizable_shift): Likewise.
21506 (vectorizable_operation): Likewise.
21507 (vectorizable_store): Likewise.
21508 (vectorizable_load): Likewise.
21509 * tree-vectorizer.h: Likewise.
21511 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
21513 * passes.c (emergency_dump_function): New.
21514 * tree-pass.h (emergency_dump_function): Declare.
21515 * plugin.c (plugins_internal_error_function): Remove.
21516 * plugin.h (plugins_internal_error_function): Remove declaration.
21517 * toplev.c (internal_error_function): New static function. Use it...
21518 (general_init): ...here.
21520 2017-05-31 Graham Markall <graham.markall@embecosm.com>
21522 * config/arc/arc.c (arc_print_operand): Handle constant operands.
21523 (arc_rtx_costs): Add costs for new patterns.
21524 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
21525 * config/arc/predicates.md: Add _1_2_3_operand predicate.
21527 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
21529 * tree-ssa-strlen.c (get_next_strinfo): New function.
21530 (get_stridx_plus_constant): Use it.
21531 (zero_length_string): Likewise.
21532 (adjust_related_strinfos): Likewise.
21533 (adjust_last_stmt): Likewise.
21535 2017-05-31 Richard Biener <rguenther@suse.de>
21538 * config/i386/i386.c (ix86_expand_builtin): Remove assert
21539 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
21541 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
21543 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
21544 loop_vinfo argument and use of dependence distance vectors.
21545 Check instead whether the two references differ only in their
21546 initial value and assume that they have the same alignment if the
21547 difference is a multiple of the vector alignment.
21548 (vect_analyze_data_refs_alignment): Update call accordingly.
21550 2017-05-31 Martin Liska <mliska@suse.cz>
21553 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
21555 2017-05-31 Bin Cheng <bin.cheng@arm.com>
21557 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
21558 (create_intersect_range_checks): Move from ...
21559 * tree-data-ref.c (create_intersect_range_checks_index)
21560 (create_intersect_range_checks): ... to here.
21561 (create_runtime_alias_checks): New function factored from ...
21562 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
21563 here. Call above function.
21564 * tree-data-ref.h (create_runtime_alias_checks): New function.
21566 2017-05-31 Bin Cheng <bin.cheng@arm.com>
21568 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
21569 segment length for dr_b and compute it in wide_int.
21571 2017-05-31 Richard Biener <rguenther@suse.de>
21573 PR tree-optimization/80906
21574 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
21575 and pass through iv_map.
21576 (copy_bb_and_scalar_dependences): Adjust.
21577 (translate_pending_phi_nodes): Likewise.
21578 (copy_loop_close_phi_args): Handle code-generating IVs instead
21581 2017-05-30 David Malcolm <dmalcolm@redhat.com>
21583 * diagnostic-color.c (color_dict): Add "type-diff".
21584 (parse_gcc_colors): Update comment.
21585 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
21586 -fdiagnostics-show-template-tree and -fno-elide-type.
21587 (GCC_COLORS): Add type-diff to example.
21589 (-fdiagnostics-show-template-tree): New.
21590 (-fno-elide-type): New.
21591 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
21592 the pp_format_decoder callback. Call any m_format_postprocessor's
21594 (pretty_printer::pretty_printer): Initialize
21595 m_format_postprocessor.
21596 (pretty_printer::~pretty_printer): Delete any
21597 m_format_postprocessor.
21598 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
21599 (class format_postprocessor): New class.
21600 (struct pretty_printer::format_decoder): Document the new parameters.
21601 (struct pretty_printer::m_format_postprocessor): New field.
21602 * tree-diagnostic.c (default_tree_printer): Update for new
21603 bool and const char ** params.
21604 * tree-diagnostic.h (default_tree_printer): Likewise.
21606 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
21608 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
21609 (lwa_operand): Delete rs6000_gen_cell_microcode test.
21610 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
21611 rs6000_gen_cell_microcode code.
21612 (rs6000_final_prescan_insn): Delete.
21613 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
21614 "warn-cell-microcode" entries.
21615 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
21616 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
21617 throughout. Change cc_reg_not_micro_cr0_operand to
21618 cc_reg_not_cr0_operand throughout.
21619 (*extendhi<mode>2_noload): Delete.
21620 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
21621 (mwarn-cell-microcode): Delete.
21622 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
21623 -mgen-cell-microcode and -mwarn-cell-microcode.
21625 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
21628 * config/i386/constraints.md (Yd): New constraint.
21630 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
21631 and (?Yd, r) alternatives. Update insn attributes.
21632 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
21633 and (?*Yd, r) alternatives. Update insn attributes.
21634 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
21636 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
21638 * gimplify.c (gimplify_modify_expr): Don't create a
21639 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
21642 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
21644 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
21646 2017-05-30 Richard Biener <rguenther@suse.de>
21648 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
21649 and reduc_def fields.
21650 (STMT_VINFO_REDUC_TYPE): New define.
21651 (STMT_VINFO_REDUC_DEF): Likewise.
21652 (vect_force_simple_reduction): Adjust prototype.
21653 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
21654 (vect_is_simple_reduction): Remove check_reduction argument.
21655 (vect_force_simple_reduction): Adjust and set
21656 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
21657 (vectorizable_reduction): Do not re-do reduction analysis
21658 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
21659 * tree-parloops.c (gather_scalar_reductions): Adjust.
21661 2017-05-30 Richard Biener <rguenther@suse.de>
21663 PR middle-end/80901
21664 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
21667 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
21669 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
21670 Introduce unknown_misalignment parameter and remove vf.
21671 (vect_peeling_hash_get_lowest_cost):
21672 Pass unknown_misalignment parameter.
21673 (vect_enhance_data_refs_alignment):
21674 Fix unsupportable data ref treatment.
21676 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21678 * tree-vect-data-refs.c (vect_get_data_access_cost):
21679 Workaround for SLP handling.
21680 (vect_enhance_data_refs_alignment):
21681 Compute costs for doing no peeling at all, compare to the best
21682 peeling costs so far and avoid peeling if cheaper.
21684 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21686 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
21687 Return peeling info and set costs to zero for unlimited cost
21689 (vect_enhance_data_refs_alignment): Also inspect all datarefs
21690 with unknown misalignment. Compute and costs for unknown
21691 misalignment, compare them to the costs for known misalignment
21692 and choose the cheapest for peeling.
21694 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21696 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
21697 (vect_get_peeling_costs_all_drs): Create function.
21698 (vect_peeling_hash_get_lowest_cost):
21699 Use vect_get_peeling_costs_all_drs.
21700 (vect_peeling_supportable): Create function.
21701 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
21703 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21705 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
21706 DR_HAS_NEGATIVE_STEP.
21707 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
21708 (vect_enhance_data_refs_alignment): Use.
21709 (vect_duplicate_ssa_name_ptr_info): Use.
21710 * tree-vectorizer.h (dr_misalignment): Use.
21711 (known_alignment_for_access_p): Use.
21713 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
21716 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
21718 (has_section_name): New function.
21720 2017-05-30 Martin Liska <mliska@suse.cz>
21723 * auto-profile.c (get_function_decl_from_block): Fix
21726 2017-05-30 Richard Biener <rguenther@suse.de>
21728 PR middle-end/80876
21729 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
21731 2017-05-30 Martin Liska <mliska@suse.cz>
21733 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
21734 * dumpfile.h (struct dump_file_info): Remove ctors.
21736 2017-05-30 Martin Liska <mliska@suse.cz>
21738 * predict.def: Fix GNU coding style.
21740 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
21742 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
21743 Mark 'to' argument with ATTRIBUTE_UNUSED.
21745 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
21747 * config/xtensa/xtensa.c (xtensa_emit_call): Use
21748 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
21749 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
21752 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
21754 * doc/install.texi (Options specification): Restore entry of
21755 --enable-sjlj-exceptions.
21757 2017-05-27 Michael Eager <eager@eagercon.com>
21760 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
21762 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
21764 * config/microblaze/microblaze.h
21765 (FIXED_REGISTERS): Update in macro.
21766 (CALL_USED_REGISTERS): Update in macro.
21768 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
21770 * doc/install.texi: Add links to macOS binary distributions.
21772 2017-05-27 Jakub Jelinek <jakub@redhat.com>
21776 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21778 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
21780 2017-05-26 Martin Liska <mliska@suse.cz>
21782 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
21784 2017-05-26 Martin Liska <mliska@suse.cz>
21786 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
21787 always leading ';; '.
21788 (dump_bb_info): Likewise.
21789 (brief_dump_cfg): Likewise.
21790 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
21791 * dumpfile.c: Remove usage of TDF_VERBOSE.
21792 * dumpfile.h (enum dump_kind): Likewise.
21793 (dump_gimple_bb_header): Do not use TDF_COMMENT.
21794 * print-tree.c (debug_verbose): Remove.
21795 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
21796 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
21797 * tree-diagnostic.c (default_tree_printer): Replace
21798 TDF_DIAGNOSTIC with TDF_SLIM.
21800 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21802 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
21803 in parameter loop, rather than loop_vinfo.
21804 (create_intersect_range_checks): Ditto.
21805 (vect_create_cond_for_alias_checks): Update call to above functions.
21807 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21809 PR tree-optimization/80815
21810 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
21811 for merging runtime alias checks. Handle negative DR_STEPs.
21813 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21815 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
21817 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
21818 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
21819 out code pruning runtime alias checks.
21820 * tree-data-ref.c (prune_runtime_alias_test_list): New function
21821 factored out from above.
21822 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
21824 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
21826 (prune_runtime_alias_test_list): New decalaration.
21828 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21830 * tree-vect-data-refs.c (compare_tree): Rename and move ...
21831 * tree-data-ref.c (data_ref_compare_tree): ... to here.
21832 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
21833 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
21834 (operator==, comp_dr_with_seg_len_pair): Ditto.
21835 (vect_prune_runtime_alias_test_list): Ditto.
21837 2017-05-26 Martin Liska <mliska@suse.cz>
21840 * params.def: Bound partial-inlining-entry-probability param.
21842 2017-05-26 Marek Polacek <polacek@redhat.com>
21845 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
21848 2017-05-26 Richard Biener <rguenther@suse.de>
21850 PR tree-optimization/80842
21851 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
21854 2017-05-26 Richard Biener <rguenther@suse.de>
21856 PR tree-optimization/80844
21857 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
21859 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
21861 * doc/md.texi (Machine Constraints): Update x86 family
21862 machine constraints section to match 'config/i386/constraints.md'.
21864 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
21866 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
21868 2017-05-25 Nathan Sidwell <nathan@acm.org>
21870 * doc/invoke.texi (--enable-languages): Update documentation.
21872 2017-05-25 Martin Liska <mliska@suse.cz>
21874 * dumpfile.c: Add TDF_FOLDING.
21875 * dumpfile.h (enum dump_kind): Likewise.
21876 * genmatch.c (dt_simplify::gen_1): Use it.
21878 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21880 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
21882 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21884 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
21885 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
21887 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21889 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
21890 * match.pd (X == C): Rewrite it here.
21891 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
21892 with_certain_nonzero_bits2): New predicates.
21893 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
21895 2017-05-24 Nathan Sidwell <nathan@acm.org>
21897 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
21900 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
21903 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
21905 * config/powerpcspe: New port. Files are copied from the rs6000
21906 port, with "rs6000" in filenames replaced by "powerpcspe".
21908 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
21910 PR rtl-optimization/80754
21911 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
21913 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
21915 * config/sparc/sparc.md (length): Return the correct value for -mflat
21916 sibcalls to match output_sibcall.
21918 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
21922 * config/rs6000/rs6000.c (struct machine_function): Add new field
21924 (rs6000_get_separate_components): Init that field, use it.
21925 (rs6000_components_for_bb): Use the field.
21927 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
21929 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
21931 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
21933 PR middle-end/80823
21934 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
21936 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21939 * config/s390/s390.c (s390_check_qrst_address): Check incoming
21940 address against address_operand predicate.
21941 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
21943 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
21945 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
21946 parameters passed indirectly.
21948 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
21950 * config/i386/i386.md (*movdi_internal): Remove SSE4
21951 alternative 18 (?r, *v). Update insn attributes.
21952 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
21953 Update insn attributes.
21954 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
21955 Update insn attributes.
21956 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
21957 alternative 1 (r, v). Remove isa attribute.
21958 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
21959 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
21960 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
21962 2017-05-23 Tom de Vries <tom@codesourcery.com>
21964 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
21967 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21969 * cgraphunit.c (symbol_table::process_new_functions): Update.
21970 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
21971 (inline_generate_summary): Rename to ...
21972 (ipa_fn_summary_generate): ... this one.
21973 (inline_read_summary): Rename to ...
21974 (ipa_fn_summary_read): ... this one.
21975 (inline_write_summary): Rename to ...
21976 (ipa_fn_summary_write): ... this one.
21977 (inline_free_summary): Rename to ...
21978 (ipa_free_fn_summary): ... this one.
21979 (pass_data_local_fn_summary, pass_local_fn_summary,
21980 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
21981 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
21982 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
21983 make_pass_ipa_fn_summary): New.
21984 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
21985 inline_write_summary, inline_free_summary): Remove.
21986 (ipa_free_fn_summary) : New.
21987 * ipa-inline.c (ipa_inline): Update.
21988 (pass_ipa_inline): Do not generate summaries.
21989 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
21991 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
21992 and add pass_ipa_fn_summary.
21993 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
21995 (make_pass_inline_parameters): Remove.
21997 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
21999 * omp-low.c (struct omp_context): Remove "default_kind" member.
22002 * omp-offload.c (execute_oacc_device_lower): Remove the
22003 parallelism dimensions function attributes for unparallelized
22004 OpenACC kernels constructs.
22006 2017-05-23 Martin Liska <mliska@suse.cz>
22008 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
22010 (cgraph_edge::make_speculative): Likewise.
22011 (cgraph_edge::resolve_speculation): Likewise.
22012 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
22013 (cgraph_node::dump): Likewise.
22014 * cgraph.h: Likewise.
22015 * cgraphunit.c (analyze_functions): Likewise.
22016 (symbol_table::compile): Likewise.
22017 * ipa-cp.c (print_all_lattices): Likewise.
22018 (determine_versionability): Likewise.
22019 (initialize_node_lattices): Likewise.
22020 (ipcp_verify_propagated_values): Likewise.
22021 (estimate_local_effects): Likewise.
22022 (update_profiling_info): Likewise.
22023 (create_specialized_node): Likewise.
22024 (perhaps_add_new_callers): Likewise.
22025 (decide_about_value): Likewise.
22026 (decide_whether_version_node): Likewise.
22027 (identify_dead_nodes): Likewise.
22028 (ipcp_store_bits_results): Likewise.
22029 * ipa-devirt.c (dump_targets): Likewise.
22030 (ipa_devirt): Likewise.
22031 * ipa-icf.c (sem_item::dump): Likewise.
22032 (sem_function::equals): Likewise.
22033 (sem_variable::equals): Likewise.
22034 (sem_item_optimizer::read_section): Likewise.
22035 (sem_item_optimizer::execute): Likewise.
22036 (congruence_class::dump): Likewise.
22037 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
22038 (dump_inline_summary): Likewise.
22039 (estimate_node_size_and_time): Likewise.
22040 (inline_analyze_function): Likewise.
22041 * ipa-inline-transform.c (inline_call): Likewise.
22042 * ipa-inline.c (report_inline_failed_reason): Likewise.
22043 (want_early_inline_function_p): Likewise.
22044 (edge_badness): Likewise.
22045 (update_edge_key): Likewise.
22046 (inline_small_functions): Likewise.
22047 * ipa-profile.c (ipa_profile): Likewise.
22048 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
22049 (ipa_make_edge_direct_to_target): Likewise.
22050 (remove_described_reference): Likewise.
22051 (ipa_impossible_devirt_target): Likewise.
22052 (propagate_controlled_uses): Likewise.
22053 (ipa_print_node_params): Likewise.
22054 (ipcp_transform_function): Likewise.
22055 * ipa-pure-const.c (pure_const_read_summary): Likewise.
22056 (propagate_pure_const): Likewise.
22057 * ipa-reference.c (generate_summary): Likewise.
22058 (read_write_all_from_decl): Likewise.
22059 (propagate): Likewise.
22060 (ipa_reference_read_optimization_summary): Likewise.
22061 * ipa-utils.c (ipa_merge_profiles): Likewise.
22062 * ipa.c (walk_polymorphic_call_targets): Likewise.
22063 (symbol_table::remove_unreachable_nodes): Likewise.
22064 (ipa_single_use): Likewise.
22065 * passes.c (execute_todo): Likewise.
22066 * predict.c (drop_profile): Likewise.
22067 * symtab.c (symtab_node::get_dump_name): New function.
22068 (symtab_node::dump_name): Likewise.
22069 (symtab_node::dump_asm_name): Likewise.
22070 (symtab_node::dump_references): Likewise.
22071 (symtab_node::dump_referring): Likewise.
22072 (symtab_node::dump_base): Likewise.
22073 (symtab_node::debug_symtab): Likewise.
22074 * tree-sra.c (convert_callers_for_node): Likewise.
22075 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
22076 * value-prof.c (init_node_map): Likewise.
22078 2017-05-23 Martin Liska <mliska@suse.cz>
22080 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
22081 and symtab_node::debug_symtab to symbol_table::debug.
22082 * cgraphunit.c (analyze_functions): Use the renamed function.
22083 (symbol_table::compile): Likewise.
22084 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
22085 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
22086 * passes.c (execute_todo): Likewise.
22087 * symtab.c (symbol_table::dump): New function.
22088 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
22090 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22092 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
22093 that nonconst implies exec.
22095 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22097 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
22098 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
22099 (inline_edge_summary_vec): Turn into ...
22100 (ipa_call_summaries): ... this one.
22101 (redirect_to_unreachable, edge_set_predicate,
22102 evaluate_properties_for_edge, inline_summary_alloc,
22103 reset_ipa_call_summary, reset_inline_summary,
22104 inline_summary_t::duplicate): Update.
22105 (inline_edge_duplication_hook): Turn to ...
22106 (ipa_call_summary_t::duplicate): ... this one.
22107 (inline_edge_removal_hook): Turn to ...
22108 (ipa_call_summary_t::remove): ... this one.
22109 (dump_inline_edge_summary): Turn to ...
22110 (dump_ipa_call_summary): ... this one.
22111 (estimate_function_body_sizes): Update.
22112 (inline_update_callee_summaries): Update.
22113 (remap_edge_change_prob): Update.
22114 (remap_edge_summaries): Update.
22115 (inline_merge_summary): Update.
22116 (do_estimate_edge_time): Update.
22117 (inline_generate_summary): Update.
22118 (inline_read_section): Update.
22119 (inline_read_summary): Update.
22120 (inline_free_summary): Update.
22121 * ipa-inline.c (can_inline_edge_p): Update.
22122 (compute_inlined_call_time): Update.
22123 (want_inline_small_function_p): Update.
22124 (edge_badness): Update.
22125 (early_inliner): Update.
22126 * ipa-inline.h (inline_edge_summary): Turn to ...
22127 (ipa_call_summary): ... this one.
22128 (ipa_call_summary_t): New class.
22129 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
22130 (ipa_call_summaries): New.
22131 (inline_edge_summary): Remove.
22132 (estimate_edge_growth): Update.
22133 * ipa-profile.c (ipa_propagate_frequency_1): Update.
22134 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22135 * ipa-split.c (execute_split_functions): Update.
22136 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
22138 2017-05-23 Tom de Vries <tom@codesourcery.com>
22140 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
22141 attributes): Document rdrand effective target.
22143 2017-05-23 Tom de Vries <tom@codesourcery.com>
22145 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
22146 attributes): Sort alphabetically.
22148 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
22150 * config/avr/genmultilib.awk: Use gsub instead of gensub.
22152 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
22155 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
22156 V2DF/V2DI splat into two separate patterns, one that handles
22157 registers, and the other that only handles memory. Drop support
22158 for splatting from a GPR on ISA 2.07 and then splitting the
22159 splat into direct move and splat.
22160 (vsx_splat_<mode>_reg): Likewise.
22161 (vsx_splat_<mode>_mem): Likewise.
22163 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
22165 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
22167 2017-05-22 Jakub Jelinek <jakub@redhat.com>
22169 PR middle-end/80809
22170 * omp-low.c (finish_taskreg_remap): New function.
22171 (finish_taskreg_scan): If unit size of ctx->record_type
22172 is non-constant, unshare the size expression and replace
22173 decls in it with possible outer var refs.
22175 PR middle-end/80809
22176 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
22177 GOVD_SHARED rather than GOVD_PRIVATE with it.
22178 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
22179 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
22181 PR middle-end/80853
22182 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
22183 as last argument to build_outer_var_ref for pointer bases of array
22184 section reductions.
22186 2017-05-19 Martin Sebor <msebor@redhat.com>
22188 * print-tree.c (print_node): Print DECL_READ_P flag.
22190 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22192 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
22193 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
22194 * cgraph.c: Likewise.
22195 * cgraphunit.c: Likewise.
22196 * gengtype.c: Likewise.
22197 * ipa-cp.c: Likewise.
22198 * ipa-devirt.c: Likewise.
22199 * ipa-icf.c: Likewise.
22200 * ipa-predicate.c: Likewise.
22201 * ipa-profile.c: Likewise.
22202 * ipa-prop.c: Likewise.
22203 * ipa-split.c: Likewise.
22205 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
22206 edge_predicate_pool, dump_inline_hints,
22207 inline_summary::account_size_time, redirect_to_unreachable,
22208 edge_set_predicate, set_hint_predicate,
22209 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
22210 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
22211 inline_summary_t::remove, remap_hint_predicate_after_duplication,
22212 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
22213 ipa_call_summary_t::remove, initialize_growth_caches,
22214 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
22215 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
22216 mark_modified, unmodified_parm_1, unmodified_parm,
22217 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
22218 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
22219 compute_bb_predicates, will_be_nonconstant_expr_predicate,
22220 will_be_nonconstant_predicate, record_modified_bb_info,
22221 get_minimal_bb, record_modified, param_change_prob,
22222 phi_result_unknown_predicate, predicate_for_phi_result,
22223 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
22224 estimate_function_body_sizes, compute_inline_parameters,
22225 compute_inline_parameters_for_curren, pass_data_inline_parameters,
22226 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
22227 inline_update_callee_summaries, remap_edge_change_prob,
22228 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
22229 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
22230 inline_analyze_function, inline_summary_t::insert,
22231 inline_generate_summary, read_ipa_call_summary, inline_read_section,
22232 inline_read_summary, write_ipa_call_summary, inline_write_summary,
22233 inline_free_summary): Move to ipa-fnsummary.h
22234 (predicate_t): Remove.
22235 * ipa-fnsummary.c: New file.
22236 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
22237 (enum inline_hints_vals, inline_hints, agg_position_info,
22238 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
22239 inline_summaries, ipa_call_summary, ipa_call_summary_t,
22240 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
22241 dump_inline_summary, dump_inline_hints, inline_generate_summary,
22242 inline_read_summary, inline_write_summary, inline_free_summary,
22243 inline_analyze_function, initialize_inline_failed,
22244 inline_merge_summary, inline_update_overall_summary,
22245 compute_inline_parameters): Move to ipa-fnsummary.h
22246 * ipa-fnsummary.h: New file.
22247 * ipa-inline-transform.h: Include ipa-inline.h.
22248 * ipa-inline.c: LIkewise.
22250 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22252 * ipa-inline.c (edge_badness): Use inlined_time instead of
22253 inline_summaries->get.
22255 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22257 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
22259 2017-05-22 Nathan Sidwell <nathan@acm.org>
22261 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
22262 (fdump-lang): Document 'raw' option.
22263 * dumpfile.h (TDI_tu): Delete.
22264 * dumpfile.c (dump_files): Remove translation-unit.
22265 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
22267 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
22269 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
22270 command option from $(AWK) call.
22271 * config/avr/genmultilib.awk: Simplify and rewrite so that it
22272 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
22273 [FORMAT]: Remove handling of variable.
22274 * config/avr/t-multilib: Regenerate.
22276 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22278 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
22280 (dump_inline_summary): Do not print self_time.
22281 (estimate_function_body_sizes): Do not set self_time.
22282 (compute_inline_parameters): Likewise.
22283 (inline_read_section, inline_write_summary): Do not stream self_time.
22284 * ipa-inline.h (inline_summary): Drop self_time.
22286 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22288 * ipa-inline-analysis.c (account_size_time): Rename to ...
22289 (inline_summary::account_size_time): ... this one.
22290 (reset_ipa_call_summary): Turn to ...
22291 (ipa_call_summary::reset): ... this one.
22292 (reset_inline_summary): Turn to ...
22293 (inline_summary::reset): ... this one.
22294 (inline_summary_t::remove): Update.
22295 (inline_summary_t::duplicate): Update.
22296 (ipa_call_summary_t::remove): Update.
22297 (dump_inline_summary): Update.
22298 (estimate_function_body_sizes): Update.
22299 (compute_inline_parameters): Update.
22300 (estimate_node_size_and_time): Update.
22301 (inline_merge_summary): Update.
22302 (inline_update_overall_summary): Update.
22303 (inline_read_section): Update.
22304 (inline_write_summary): Update.
22305 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
22306 add account_size_time and reset member functions.
22307 (ipa_call_summary): Add reset function.
22308 * ipa-predicate.h (predicate::operator &): Constify.
22310 2017-05-22 Richard Biener <rguenther@suse.de>
22312 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
22314 2017-05-19 Jason Merrill <jason@redhat.com>
22316 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
22318 2017-05-19 Marek Polacek <polacek@redhat.com>
22321 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
22322 TYPE_OVERFLOW_WRAPS checks.
22324 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
22326 * tree-core.h (enum omp_clause_default_kind): Add
22327 "OMP_CLAUSE_DEFAULT_PRESENT".
22328 * tree-pretty-print.c (dump_omp_clause): Handle it.
22329 * gimplify.c (enum gimplify_omp_var_data): Add
22330 "GOVD_MAP_FORCE_PRESENT".
22331 (gimplify_adjust_omp_clauses_1): Map it to
22332 "GOMP_MAP_FORCE_PRESENT".
22333 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
22335 * gimplify.c (oacc_default_clause): Clarify.
22337 2017-05-19 Nathan Sidwell <nathan@acm.org>
22339 LANG_HOOK_REGISTER_DUMPS
22340 * toplev.c (general_init): Call register dump lang hook.
22341 * doc/invoke.texi: Document -fdump-lang option family.
22342 * dumpfile.c (dump_files): Remove class dump here.
22343 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
22344 * dumpfile.h (tree_dump_index): Remove TDI_class.
22345 * langhooks-def.h (lhd_register_dumps): Declare.
22346 (LANG_HOOKS_REGISTER_DUMPS): Define.
22347 (LANG_HOOKS_INITIALIZER): Add it.
22348 * langhooks.c (lhd_register_dumps): Define.
22349 * langhooks.h (struct lang_hooks): Add register_dumps.
22351 2017-05-19 Nathan Sidwell <nathan@acm.org>
22353 * context.h (context::set_passes): New.
22354 * context.c (context::context): Do not create pass manager.
22355 * toplev.c (general_init): Create pass manager here.
22357 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
22359 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
22360 use this splitter if two add or or instructions would also work for
22361 the constant we want to generate.
22363 2017-05-19 Richard Biener <rguenther@suse.de>
22366 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
22367 predicate evaluation.
22369 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
22371 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
22373 * ipa-inline.c (want_inline_small_function_p): Do not cast to
22376 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
22378 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
22379 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
22380 (inline_edge_summary_vec): Turn into ...
22381 (ipa_call_summaries): ... this one.
22382 (redirect_to_unreachable, edge_set_predicate,
22383 evaluate_properties_for_edge, inline_summary_alloc,
22384 reset_ipa_call_summary, reset_inline_summary,
22385 inline_summary_t::duplicate): Update.
22386 (inline_edge_duplication_hook): Turn to ...
22387 (ipa_call_summary_t::duplicate): ... this one.
22388 (inline_edge_removal_hook): Turn to ...
22389 (ipa_call_summary_t::remove): ... this one.
22390 (dump_inline_edge_summary): Turn to ...
22391 (dump_ipa_call_summary): ... this one.
22392 (estimate_function_body_sizes): Update.
22393 (inline_update_callee_summaries): Update.
22394 (remap_edge_change_prob): Update.
22395 (remap_edge_summaries): Update.
22396 (inline_merge_summary): Update.
22397 (do_estimate_edge_time): Update.
22398 (inline_generate_summary): Update.
22399 (inline_read_section): Update.
22400 (inline_read_summary): Update.
22401 (inline_free_summary): Update.
22402 * ipa-inline.c (can_inline_edge_p): Update.
22403 (compute_inlined_call_time): Update.
22404 (want_inline_small_function_p): Update.
22405 (edge_badness): Update.
22406 (early_inliner): Update.
22407 * ipa-inline.h (inline_edge_summary): Turn to ...
22408 (ipa_call_summary): ... this one.
22409 (ipa_call_summary_t): New class.
22410 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
22411 (ipa_call_summaries): New.
22412 (inline_edge_summary): Remove.
22413 (estimate_edge_growth): Update.
22414 * ipa-profile.c (ipa_propagate_frequency_1): Update.
22415 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22416 * ipa-split.c (execute_split_functions): Update.
22417 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
22419 2017-05-19 Richard Biener <rguenther@suse.de>
22421 PR middle-end/80764
22422 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
22424 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
22426 * config/rs6000/rs6000.c (struct machine_function): Add field
22427 fpr_is_wrapped_separately.
22428 (rs6000_get_separate_components): Use 64 components. Handle the
22429 new FPR components.
22430 (rs6000_components_for_bb): Handle the FPR components.
22431 (rs6000_emit_prologue_components): Handle the FPR components.
22432 (rs6000_emit_epilogue_components): Handle the FPR components.
22433 (rs6000_set_handled_components): Handle the FPR components.
22434 (rs6000_emit_prologue): Don't output prologue code for those FPRs
22435 that are already separately shrink-wrapped.
22436 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
22437 that are already separately shrink-wrapped.
22439 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
22442 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
22445 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
22446 (define_peephole2 for Altivec d-form load): Add peepholes to catch
22447 cases where the register allocator uses a move and an offsettable
22448 memory operation to/from a FPR register on ISA 2.06/2.07.
22449 (define_peephole2 for Altivec d-form store): Likewise.
22451 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
22454 * config/i386/mmx.md (*mov<mode>_internal): Enable
22455 alternatives 11, 12, 13 and 14 also for 32bit targets.
22456 Remove alternatives 15, 16, 17 and 18.
22457 * config/i386/sse.md (vec_concatv2di): Change
22458 alternative (!x, *y) to (x, ?!*Yn).
22460 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
22462 * dumpfile.h (enum dump_kind): Remove stray comma.
22464 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
22466 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
22467 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
22468 predicate::num_conditions
22469 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
22470 (CHANGED): turn into predicate::changed.
22471 (agg_position_info): Move to ipa-predicate.h
22472 (add_condition, predicate::add_clause, predicate::operator &=,
22473 predicate::or_with, predicate::evaluate, predicate::probability,
22474 dump_condition, dump_clause, predicate::dump,
22475 predicate::remap_after_duplication, predicate::remap_after_inlining,
22476 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
22477 (evaluate_conditions_for_known_args): Update.
22478 (set_cond_stmt_execution_predicate): Update.
22479 * ipa-inline.h: Include ipa-predicate.h
22480 (condition, inline_param_summary, conditions, agg_position_info,
22481 predicate): Move to ipa-predicate.h
22482 * ipa-predicate.c: New file.
22483 * ipa-predicate.h: New file.
22485 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
22487 * final.c (leaf_function_p): Check we are not in a sequence.
22489 2017-05-18 Martin Liska <mliska@suse.cz>
22491 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
22492 * dumpfile.c (dump_register): Use new enum dump_kind.
22493 (get_dump_file_name): Likewise.
22494 (dump_enable_all): Likewise.
22495 (dump_switch_p_1): Likewise.
22496 (enable_rtl_dump_file): Remove usage of TDF_RTL.
22497 * dumpfile.h (enum dump_kind): New enum type.
22498 (struct dump_file_info): Create constructor and
22499 format fields and comments.
22500 * passes.c (pass_manager::register_one_dump_file):
22502 * statistics.c (statistics_early_init): Likewise.
22503 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
22504 TDF_TREE with TDF_SLIM.
22505 (gather_memory_references_ref): Likewise.
22507 2017-05-18 Martin Liska <mliska@suse.cz>
22509 * vec.h (struct vnull): Use it.
22511 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
22513 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
22514 (true_predicate, false_predicate, true_predicate_p,
22515 false_predicate_p): Remove.
22516 (single_cond_predicate, not_inlined_predicate): Turn to member function
22518 (add_condition): Update.
22519 (add_clause): Turn to...
22520 (predicate::add_clause): ... this one; update; allow passing NULL
22522 (and_predicates): Turn to ...
22523 (predicate::operator &=): ... this one.
22524 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
22525 (or_predicates): Turn to ...
22526 (predicate::or_with): ... this one.
22527 (evaluate_predicate): Turn to ...
22528 (predicate::evaluate): ... this one.
22529 (predicate_probability): Turn to ...
22530 (predicate::probability): ... this one.
22531 (dump_condition): Update.
22532 (dump_predicate): Turn to ...
22533 (predicate::dump): ... this one.
22534 (account_size_time): Update.
22535 (edge_set_predicate): Update.
22536 (set_hint_predicate): UPdate.
22537 (evaluate_conditions_for_known_args): Update.
22538 (evaluate_properties_for_edge): Update.
22539 (remap_predicate_after_duplication): Turn to...
22540 (predicate::remap_after_duplication): ... this one.
22541 (remap_hint_predicate_after_duplication): Update.
22542 (inline_summary_t::duplicate): UPdate.
22543 (dump_inline_edge_summary): Update.
22544 (dump_inline_summary): Update.
22545 (set_cond_stmt_execution_predicate): Update.
22546 (set_switch_stmt_execution_predicate): Update.
22547 (compute_bb_predicates): Update.
22548 (will_be_nonconstant_expr_predicate): Update.
22549 (will_be_nonconstant_predicate): Update.
22550 (phi_result_unknown_predicate): Update.
22551 (predicate_for_phi_result): Update.
22552 (array_index_predicate): Update.
22553 (estimate_function_body_sizes): Update.
22554 (estimate_node_size_and_time): Update.
22555 (estimate_ipcp_clone_size_and_time): Update.
22556 (remap_predicate): Rename to ...
22557 (predicate::remap_after_inlining): ... this one.
22558 (remap_hint_predicate): Update.
22559 (inline_merge_summary): Update.
22560 (inline_update_overall_summary): Update.
22561 (estimate_size_after_inlining): Update.
22562 (read_predicate): Rename to ...
22563 (predicate::stream_in): ... this one.
22564 (read_inline_edge_summary): Update.
22565 (write_predicate): Rename to ...
22566 (predicate::stream_out): ... this one.
22567 (write_inline_edge_summary): Update.
22568 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
22569 (clause_t): Turn to uint32_t
22570 (predicate): Turn to class; implement constructor and operators
22572 (size_time_entry): Update.
22573 (inline_summary): Update.
22574 (inline_edge_summary): Update.
22576 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
22578 * fold-const.c (fold_binary_loc): Move transformation...
22579 * match.pd (C - X CMP X): ... here.
22581 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
22583 * config/sparc/sparc.c (sparc_option_override): Set function
22584 alignment for -mcpu=niagara7 to 64 to match the I$ line.
22585 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
22587 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
22589 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
22591 2017-05-18 Marek Polacek <polacek@redhat.com>
22594 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
22595 (pass_ubsan::execute): Call gimple_assign_single_p instead of
22596 gimple_assign_load_p.
22598 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
22600 PR middle-end/80692
22601 * real.c (do_compare): Give decimal_do_compare preference over
22602 comparing just the signs.
22604 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
22606 * doc/md.texi (Canonicalization of Instructions): Describe the
22607 canonical form of instructions that inherently set a condition
22610 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
22612 PR middle-end/80775
22613 * tree-cfg.c: Move deletion of unreachable case statements to after
22614 the merging of consecutive case labels.
22616 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
22618 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
22619 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
22620 restoring of callee-saved registers.
22622 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
22624 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
22625 * config/visium/visium.c (single_set_and_flags): Likewise.
22626 * config/visium/visium.md (Substitutions): Likewise.
22628 2017-05-17 Martin Liska <mliska@suse.cz>
22630 * cfg.c: Introduce dump_flags_t type and
22631 use it instead of int type.
22633 * cfghooks.c: Likewise.
22634 * cfghooks.h (struct cfg_hooks): Likewise.
22635 * cfgrtl.c: Likewise.
22636 * cfgrtl.h: Likewise.
22637 * cgraph.c (cgraph_node::get_body): Likewise.
22638 * coretypes.h: Likewise.
22639 * domwalk.c: Likewise.
22640 * domwalk.h: Likewise.
22641 * dumpfile.c (struct dump_option_value_info): Likewise.
22642 (dump_enable_all): Likewise.
22643 (dump_switch_p_1): Likewise.
22644 (opt_info_switch_p): Likewise.
22645 * dumpfile.h (enum tree_dump_index): Likewise.
22646 (struct dump_file_info): Likewise.
22647 * genemit.c: Likewise.
22648 * generic-match-head.c: Likewise.
22649 * gengtype.c (open_base_files): Likewise.
22650 * gimple-pretty-print.c: Likewise.
22651 * gimple-pretty-print.h: Likewise.
22652 * graph.c (print_graph_cfg): Likewise.
22653 * graphite-scop-detection.c (dot_all_sese): Likewise.
22654 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
22655 * loop-unroll.c (report_unroll): Likewise.
22656 * passes.c (pass_manager::register_one_dump_file): Likewise.
22657 * print-tree.c: Likewise.
22658 * statistics.c: Likewise.
22659 * tree-cfg.c: Likewise.
22660 * tree-cfg.h: Likewise.
22661 * tree-dfa.c: Likewise.
22662 * tree-dfa.h: Likewise.
22663 * tree-dump.c (dump_function): Likewise.
22664 * tree-dump.h (struct dump_info): Likewise.
22665 * tree-pretty-print.c: Likewise.
22666 * tree-pretty-print.h: Likewise.
22667 * tree-ssa-live.c: Likewise.
22668 * tree-ssa-live.h: Likewise.
22669 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
22670 * tree-vect-loop.c: Likewise.
22671 * tree-vect-slp.c: Likewise.
22673 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
22674 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22676 PR tree-optimization/80457
22677 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
22678 of all arguments to a statement as scalar_to_vec operations.
22679 (vectorizable_call): Adjust call to vect_model_simple_cost for
22681 (vectorizable_conversion): Likewise.
22682 (vectorizable_assignment): Likewise.
22683 (vectorizable_shift): Likewise.
22684 (vectorizable_operation): Likewise.
22685 (vectorizable_comparison): Likewise.
22686 (vect_is_simple_cond): Record the def types for operands.
22687 (vectorizable_condition): Likewise, call vect_model_simple_cost.
22688 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
22689 for statement argument count.
22691 2017-05-16 Carl Love <cel@us.ibm.com>
22693 * config/rs6000/rs6000-c: Add support for built-in functions
22694 vector unsigned long long vec_bperm (vector unsigned long long,
22695 vector unsigned char)
22696 vector signed long long vec_mule (vector signed int,
22698 vector unsigned long long vec_mule (vector unsigned int,
22699 vector unsigned int)
22700 vector signed long long vec_mulo (vector signed int,
22702 vector unsigned long long vec_mulo (vector unsigned int,
22703 vector unsigned int)
22704 vector signed char vec_sldw (vector signed char,
22705 vector signed char,
22707 vector unsigned char vec_sldw (vector unsigned char,
22708 vector unsigned char,
22710 vector signed short vec_sldw (vector signed short,
22711 vector signed short,
22713 vector unsigned short vec_sldw (vector unsigned short,
22714 vector unsigned short,
22716 vector signed int vec_sldw (vector signed int,
22719 vector unsigned int vec_sldw (vector unsigned int,
22720 vector unsigned int,
22722 vector signed long long vec_sldw (vector signed long long,
22723 vector signed long long,
22725 vector unsigned long long vec_sldw (vector unsigned long long,
22726 vector unsigned long long,
22728 * config/rs6000/rs6000-c: Add support for built-in functions
22729 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
22730 * config/rs6000/altivec.h: Add defintion for vec_sldw.
22731 * doc/extend.texi: Update the built-in documentation for the
22732 new built-in functions.
22734 2017-05-16 Marek Polacek <polacek@redhat.com>
22738 * tree.c (save_expr): Don't fold the expression.
22740 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
22742 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
22743 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
22744 and (?*y,m). Update insn attributes.
22746 2017-05-16 Martin Liska <mliska@suse.cz>
22748 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
22749 flags argument of print_gimple_stmt, print_gimple_expr,
22750 print_generic_stmt and print_generic_expr.
22751 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
22752 * coretypes.h: Likewise.
22753 * except.c (dump_eh_tree): Likewise.
22754 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
22755 * gimple-pretty-print.h: Likewise.
22756 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
22757 (backprop::push_to_worklist): Likewise.
22758 (backprop::pop_from_worklist): Likewise.
22759 (backprop::process_use): Likewise.
22760 (backprop::intersect_uses): Likewise.
22761 (note_replacement): Likewise.
22762 * gimple-ssa-store-merging.c
22763 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
22764 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
22765 (pass_store_merging::execute): Likewise.
22766 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
22767 (ssa_base_cand_dump_callback): Likewise.
22768 (dump_incr_vec): Likewise.
22769 (replace_refs): Likewise.
22770 (replace_mult_candidate): Likewise.
22771 (create_add_on_incoming_edge): Likewise.
22772 (create_phi_basis): Likewise.
22773 (insert_initializers): Likewise.
22774 (all_phi_incrs_profitable): Likewise.
22775 (introduce_cast_before_cand): Likewise.
22776 (replace_one_candidate): Likewise.
22777 * gimplify.c (gimplify_expr): Likewise.
22778 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
22779 (set_rename): Likewise.
22780 (rename_uses): Likewise.
22781 (copy_loop_phi_nodes): Likewise.
22782 (add_close_phis_to_merge_points): Likewise.
22783 (copy_loop_close_phi_args): Likewise.
22784 (copy_cond_phi_args): Likewise.
22785 (graphite_copy_stmts_from_block): Likewise.
22786 (translate_pending_phi_nodes): Likewise.
22787 * graphite-poly.c (print_pdr): Likewise.
22788 (dump_gbb_cases): Likewise.
22789 (dump_gbb_conditions): Likewise.
22790 (print_scop_params): Likewise.
22791 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
22792 (build_cross_bb_scalars_use): Likewise.
22793 (gather_bbs::before_dom_children): Likewise.
22794 * hsa-dump.c (dump_hsa_immed): Likewise.
22795 * ipa-cp.c (print_ipcp_constant_value): Likewise.
22796 (get_replacement_map): Likewise.
22797 * ipa-inline-analysis.c (dump_condition): Likewise.
22798 (estimate_function_body_sizes): Likewise.
22799 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
22800 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
22801 * ipa-prop.c (ipa_dump_param): Likewise.
22802 (ipa_print_node_jump_functions_for_edge): Likewise.
22803 (ipa_modify_call_arguments): Likewise.
22804 (ipa_modify_expr): Likewise.
22805 (ipa_dump_param_adjustments): Likewise.
22806 (ipa_dump_agg_replacement_values): Likewise.
22807 (ipcp_modif_dom_walker::before_dom_children): Likewise.
22808 * ipa-pure-const.c (check_stmt): Likewise.
22809 (pass_nothrow::execute): Likewise.
22810 * ipa-split.c (execute_split_functions): Likewise.
22811 * omp-offload.c (dump_oacc_loop_part): Likewise.
22812 (dump_oacc_loop): Likewise.
22813 * trans-mem.c (tm_log_emit): Likewise.
22814 (tm_memopt_accumulate_memops): Likewise.
22815 (dump_tm_memopt_set): Likewise.
22816 (dump_tm_memopt_transform): Likewise.
22817 * tree-cfg.c (gimple_verify_flow_info): Likewise.
22818 (print_loop): Likewise.
22819 * tree-chkp-opt.c (chkp_print_addr): Likewise.
22820 (chkp_gather_checks_info): Likewise.
22821 (chkp_get_check_result): Likewise.
22822 (chkp_remove_check_if_pass): Likewise.
22823 (chkp_use_outer_bounds_if_possible): Likewise.
22824 (chkp_reduce_bounds_lifetime): Likewise.
22825 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
22826 (chkp_mark_completed_bounds): Likewise.
22827 (chkp_register_incomplete_bounds): Likewise.
22828 (chkp_mark_invalid_bounds): Likewise.
22829 (chkp_maybe_copy_and_register_bounds): Likewise.
22830 (chkp_build_returned_bound): Likewise.
22831 (chkp_get_bound_for_parm): Likewise.
22832 (chkp_build_bndldx): Likewise.
22833 (chkp_get_bounds_by_definition): Likewise.
22834 (chkp_generate_extern_var_bounds): Likewise.
22835 (chkp_get_bounds_for_decl_addr): Likewise.
22836 * tree-chrec.c (chrec_apply): Likewise.
22837 * tree-data-ref.c (dump_data_reference): Likewise.
22838 (dump_subscript): Likewise.
22839 (dump_data_dependence_relation): Likewise.
22840 (analyze_overlapping_iterations): Likewise.
22841 * tree-inline.c (expand_call_inline): Likewise.
22842 (tree_function_versioning): Likewise.
22843 * tree-into-ssa.c (dump_defs_stack): Likewise.
22844 (dump_currdefs): Likewise.
22845 (dump_names_replaced_by): Likewise.
22846 (dump_update_ssa): Likewise.
22847 (update_ssa): Likewise.
22848 * tree-object-size.c (pass_object_sizes::execute): Likewise.
22849 * tree-parloops.c (build_new_reduction): Likewise.
22850 (try_create_reduction_list): Likewise.
22851 (ref_conflicts_with_region): Likewise.
22852 (oacc_entry_exit_ok_1): Likewise.
22853 (oacc_entry_exit_single_gang): Likewise.
22854 * tree-pretty-print.h: Likewise.
22855 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
22856 (get_scalar_evolution): Likewise.
22857 (add_to_evolution): Likewise.
22858 (get_loop_exit_condition): Likewise.
22859 (analyze_evolution_in_loop): Likewise.
22860 (analyze_initial_condition): Likewise.
22861 (analyze_scalar_evolution): Likewise.
22862 (instantiate_scev): Likewise.
22863 (number_of_latch_executions): Likewise.
22864 (gather_chrec_stats): Likewise.
22865 (final_value_replacement_loop): Likewise.
22866 (scev_const_prop): Likewise.
22867 * tree-sra.c (dump_access): Likewise.
22868 (disqualify_candidate): Likewise.
22869 (create_access): Likewise.
22870 (reject): Likewise.
22871 (maybe_add_sra_candidate): Likewise.
22872 (create_access_replacement): Likewise.
22873 (analyze_access_subtree): Likewise.
22874 (analyze_all_variable_accesses): Likewise.
22875 (sra_modify_assign): Likewise.
22876 (initialize_constant_pool_replacements): Likewise.
22877 (find_param_candidates): Likewise.
22878 (decide_one_param_reduction): Likewise.
22879 (replace_removed_params_ssa_names): Likewise.
22880 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
22881 * tree-ssa-copy.c (dump_copy_of): Likewise.
22882 (copy_prop_visit_cond_stmt): Likewise.
22883 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
22884 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
22885 (record_equivalences_from_stmt): Likewise.
22886 * tree-ssa-dse.c (compute_trims): Likewise.
22887 (delete_dead_call): Likewise.
22888 (delete_dead_assignment): Likewise.
22889 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
22890 (forward_propagate_into_cond): Likewise.
22891 (pass_forwprop::execute): Likewise.
22892 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
22893 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
22895 (move_computations_worker): Likewise.
22896 (execute_sm): Likewise.
22897 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
22898 (remove_exits_and_undefined_stmts): Likewise.
22899 (remove_redundant_iv_tests): Likewise.
22900 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
22901 (adjust_iv_update_pos): Likewise.
22902 * tree-ssa-math-opts.c (bswap_replace): Likewise.
22903 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
22904 (value_replacement): Likewise.
22905 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
22906 * tree-ssa-pre.c (print_pre_expr): Likewise.
22907 (get_representative_for): Likewise.
22908 (create_expression_by_pieces): Likewise.
22909 (insert_into_preds_of_block): Likewise.
22910 (eliminate_insert): Likewise.
22911 (eliminate_dom_walker::before_dom_children): Likewise.
22912 (eliminate): Likewise.
22913 (remove_dead_inserted_code): Likewise.
22914 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
22915 * tree-ssa-reassoc.c (get_rank): Likewise.
22916 (eliminate_duplicate_pair): Likewise.
22917 (eliminate_plus_minus_pair): Likewise.
22918 (eliminate_not_pairs): Likewise.
22919 (undistribute_ops_list): Likewise.
22920 (eliminate_redundant_comparison): Likewise.
22921 (update_range_test): Likewise.
22922 (optimize_range_tests_var_bound): Likewise.
22923 (optimize_vec_cond_expr): Likewise.
22924 (rewrite_expr_tree): Likewise.
22925 (rewrite_expr_tree_parallel): Likewise.
22926 (linearize_expr): Likewise.
22927 (break_up_subtract): Likewise.
22928 (linearize_expr_tree): Likewise.
22929 (attempt_builtin_powi): Likewise.
22930 (attempt_builtin_copysign): Likewise.
22931 (transform_stmt_to_copy): Likewise.
22932 (transform_stmt_to_multiply): Likewise.
22933 (dump_ops_vector): Likewise.
22934 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
22935 (print_scc): Likewise.
22936 (set_ssa_val_to): Likewise.
22937 (visit_reference_op_store): Likewise.
22938 (visit_use): Likewise.
22939 (sccvn_dom_walker::before_dom_children): Likewise.
22940 (run_scc_vn): Likewise.
22941 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
22943 (expr_hash_elt::print): Likewise.
22944 (const_and_copies::pop_to_marker): Likewise.
22945 (const_and_copies::record_const_or_copy_raw): Likewise.
22946 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
22947 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
22948 (dump_predicates): Likewise.
22949 (find_uninit_use): Likewise.
22950 (warn_uninitialized_phi): Likewise.
22951 (pass_late_warn_uninitialized::execute): Likewise.
22952 * tree-ssa.c (verify_vssa): Likewise.
22953 (verify_ssa): Likewise.
22954 (maybe_optimize_var): Likewise.
22955 * tree-vrp.c (dump_value_range): Likewise.
22956 (dump_all_value_ranges): Likewise.
22957 (dump_asserts_for): Likewise.
22958 (register_edge_assert_for_2): Likewise.
22959 (vrp_visit_cond_stmt): Likewise.
22960 (vrp_visit_switch_stmt): Likewise.
22961 (vrp_visit_stmt): Likewise.
22962 (vrp_visit_phi_node): Likewise.
22963 (simplify_cond_using_ranges_1): Likewise.
22964 (fold_predicate_in): Likewise.
22965 (evrp_dom_walker::before_dom_children): Likewise.
22966 (evrp_dom_walker::push_value_range): Likewise.
22967 (evrp_dom_walker::pop_value_range): Likewise.
22968 (execute_early_vrp): Likewise.
22970 2017-05-16 Richard Biener <rguenther@suse.de>
22972 * dwarf2out.c (loc_list_from_tree_1): Do not create
22973 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
22975 2017-05-16 Richard Biener <rguenther@suse.de>
22977 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
22979 (note_variable_value_in_expr): If we resolved the decl ref
22980 do not push to the stack.
22982 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
22984 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
22985 operations in fast-math mode.
22986 (vaddq_f16): Likewise.
22987 (vmul_f16): Likewise.
22988 (vmulq_f16): Likewise.
22989 (vsub_f16): Likewise.
22990 (vsubq_f16): Likewise.
22991 * config/arm/neon.md (add<mode>3): New.
22993 (fma:<VH:mode>3): New. Also remove outdated comment.
22996 2017-05-16 Martin Liska <mliska@suse.cz>
23000 * ipa-devirt.c (warn_types_mismatch): Fix typo.
23001 (odr_types_equivalent_p): Likewise.
23003 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
23005 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
23007 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
23010 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
23011 non-interunit SSE move alternatives with '?'.
23012 (zero-extendsidi peephole2): New peephole to skip intermediate
23013 general register in SSE zero-extend sequence.
23015 2017-05-15 Jeff Law <law@redhat.com>
23017 * reorg.c (relax_delay_slots): Create a new variable to hold
23018 the temporary target rather than clobbering TARGET_LABEL.
23020 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
23021 missing argument to extract_bit_field call.
23022 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
23024 2017-05-15 Martin Liska <mliska@suse.cz>
23027 * gcc.c (process_command): Do not allow empty argument of -o option.
23029 2017-05-15 Renlin Li <renlin.li@arm.com>
23031 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
23032 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
23033 * config/aarch64/constraints.md (Usf): Add long call check.
23034 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
23035 (call_value): Likewise.
23036 (sibcall): Likewise.
23037 (sibcall_value): Likewise.
23039 (call_value_insn): New.
23040 (sibcall_insn): Update rtx pattern.
23041 (sibcall_value_insn): Likewise.
23042 (call_internal): Remove.
23043 (call_value_internal): Likewise.
23044 (sibcall_internal): Likewise.
23045 (sibcall_value_internal): Likewise.
23046 (call_reg): Likewise.
23047 (call_symbol): Likewise.
23048 (call_value_reg): Likewise.
23049 (call_value_symbol): Likewise.
23051 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
23054 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
23056 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
23058 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
23059 compatible with CCGOCmode and with CCZmode.
23061 2017-05-14 Martin Sebor <msebor@redhat.com>
23063 PR middle-end/77671
23064 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
23065 (gimple_fold_builtin_snprintf): Same.
23066 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
23067 (gimple_fold_builtin_snprintf): Same.
23068 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
23069 of character types.
23070 (is_call_safe): New function.
23071 (try_substitute_return_value): Call it.
23072 (try_simplify_call): New function.
23073 (pass_sprintf_length::handle_gimple_call): Call it.
23075 2017-05-14 Martin Sebor <msebor@redhat.com>
23077 PR middle-end/80669
23078 * builtins.c (expand_builtin_stpncpy): Simplify.
23080 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
23082 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
23083 * config/i386/i386.h
23084 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
23085 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
23086 (struct machine_function): Add new members call_ms2sysv,
23087 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
23088 (struct machine_frame_state): New fields sp_realigned and
23089 sp_realigned_offset.
23090 * config/i386/i386.c
23091 (enum xlogue_stub): New enum.
23092 (enum xlogue_stub_sets): New enum.
23093 (class xlogue_layout): New class.
23094 (struct ix86_frame): New fields stack_realign_allocate_offset,
23095 stack_realign_offset and outlined_save_offset. Modify comments to
23096 detail stack layout when using out-of-line stubs.
23097 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
23098 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
23099 -mcall-ms2sysv-xlogues.
23100 (stub_managed_regs): New static variable.
23101 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
23102 registers managed by out-of-line stub.
23103 (disable_call_ms2sysv_xlogues): New function.
23104 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
23105 m->call_ms2sysv when appropriate and compute frame layout for
23107 (sp_valid_at, fp_valid_at): New inline functions.
23108 (choose_basereg): New function.
23109 (choose_baseaddr): Add align parameter, use choose_basereg and modify
23111 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
23112 Use align parameter of choose_baseaddr to generated aligned SSE movs
23114 (pro_epilogue_adjust_stack): Modify to track
23115 machine_frame_state::sp_realigned.
23116 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
23117 (ix86_nsaved_sseregs): Likewise.
23118 (ix86_emit_save_regs): Likewise.
23119 (ix86_emit_save_regs_using_mov): Likewise.
23120 (ix86_emit_save_sse_regs_using_mov): Likewise.
23121 (get_scratch_register_on_entry): Likewise.
23122 (gen_frame_set): New function.
23123 (gen_frame_load): Likewise.
23124 (gen_frame_store): Likewise.
23125 (emit_outlined_ms2sysv_save): Likewise.
23126 (emit_outlined_ms2sysv_restore): Likewise.
23127 (ix86_expand_prologue): Modify stack re-alignment code and call
23128 emit_outlined_ms2sysv_save when appropriate.
23129 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
23130 parameter rtx_insn *insn, which allows the function to be used to only
23131 generate the notes.
23132 (ix86_expand_epilogue): Modify validity checks of frame and stack
23133 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
23134 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
23135 * config/i386/predicates.md
23136 (save_multiple): New predicate.
23137 (restore_multiple): Likewise.
23138 * config/i386/sse.md
23139 (save_multiple<mode>): New pattern.
23140 (save_multiple_realign<mode>): Likewise.
23141 (restore_multiple<mode>): Likewise.
23142 (restore_multiple_and_return<mode>): Likewise.
23143 (restore_multiple_leave_return<mode>): Likewise.
23144 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
23146 2017-05-14 Julia Koval <julia.koval@intel.com>
23148 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
23149 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
23150 (__builtin_ia32_xsetbv): New builtins.
23151 * config/i386/i386.c (ix86_expand_special_args_builtin):
23153 (ix86_expand_builtin): Special expand for new intrinsics.
23154 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
23155 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
23156 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
23158 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23160 * cfganal.c (inverted_post_order_compute): Change argument type
23162 * cfganal.h (inverted_post_order_compute): Adjust prototype.
23163 * df-core.c (rest_of_handle_df_initialize): Adjust.
23164 (rest_of_handle_df_finish): Likewise.
23165 (df_analyze_1): Likewise.
23166 (df_analyze): Likewise.
23167 (loop_inverted_post_order_compute): Change argument to be a vec *.
23168 (df_analyze_loop): Adjust.
23169 (df_get_n_blocks): Likewise.
23170 (df_get_postorder): Likewise.
23171 * df.h (struct df_d): Change field to be a vec.
23172 * lcm.c (compute_laterin): Adjust.
23173 (compute_available): Likewise.
23174 * lra-lives.c (lra_create_live_ranges_1): Likewise.
23175 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
23176 * tree-ssa-pre.c (compute_antic): Likewise.
23178 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23180 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
23181 (depth_first_search::depth_first_search): Change structure init
23182 function to this constructor.
23183 (depth_first_search::add_bb): Rename function to this member.
23184 (depth_first_search::execute): Likewise.
23185 (flow_dfs_compute_reverse_finish): Adjust.
23187 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23189 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
23190 (longest_simple_path): Likewise.
23191 * shrink-wrap.c (spread_components): Likewise.
23192 (disqualify_problematic_components): Likewise.
23193 (emit_common_heads_for_components): Likewise.
23194 (emit_common_tails_for_components): Likewise.
23195 (insert_prologue_epilogue_for_components): Likewise.
23197 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23199 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
23202 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23204 * df-core.c (df_set_blocks): Start using auto_bitmap.
23205 (df_compact_blocks): Likewise.
23206 * df-problems.c (df_rd_confluence_n): Likewise.
23207 * df-scan.c (df_insn_rescan_all): Likewise.
23208 (df_process_deferred_rescans): Likewise.
23209 (df_update_entry_block_defs): Likewise.
23210 (df_update_exit_block_uses): Likewise.
23211 (df_entry_block_bitmap_verify): Likewise.
23212 (df_exit_block_bitmap_verify): Likewise.
23213 (df_scan_verify): Likewise.
23214 * lra-constraints.c (lra_constraints): Likewise.
23215 (undo_optional_reloads): Likewise.
23216 (lra_undo_inheritance): Likewise.
23217 * lra-remat.c (calculate_gen_cands): Likewise.
23218 (do_remat): Likewise.
23219 * lra-spills.c (assign_spill_hard_regs): Likewise.
23220 (spill_pseudos): Likewise.
23221 * tree-ssa-pre.c (bitmap_set_and): Likewise.
23222 (bitmap_set_subtract_values): Likewise.
23224 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23226 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
23227 management with auto_bitmap.
23228 (fix_inter_tick): Likewise.
23229 (fix_recovery_deps): Likewise.
23230 * ira.c (add_store_equivs): Likewise.
23231 (find_moveable_pseudos): Likewise.
23232 (split_live_ranges_for_shrink_wrap): Likewise.
23233 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
23234 (rtx_reuse_manager::seen_def_p): Likewise.
23235 (rtx_reuse_manager::set_seen_def): Likewise.
23236 * print-rtl.h (class rtx_reuse_manager): Likewise.
23238 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23240 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
23242 (migrate_btr_def): Likewise.
23243 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
23244 * df-core.c (loop_post_order_compute): Likewise.
23245 (loop_inverted_post_order_compute): Likewise.
23246 * hsa-common.h: Likewise.
23247 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
23248 * init-regs.c (initialize_uninitialized_regs): Likewise.
23249 * ipa-inline.c (resolve_noninline_speculation): Likewise.
23250 (inline_small_functions): Likewise.
23251 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
23252 * ira.c (combine_and_move_insns): Likewise.
23253 (build_insn_chain): Likewise.
23254 * loop-invariant.c (find_invariants): Likewise.
23255 * lower-subreg.c (propagate_pseudo_copies): Likewise.
23256 * predict.c (tree_predict_by_opcode): Likewise.
23257 (predict_paths_leading_to): Likewise.
23258 (predict_paths_leading_to_edge): Likewise.
23259 (estimate_loops_at_level): Likewise.
23260 (estimate_loops): Likewise.
23261 * shrink-wrap.c (try_shrink_wrapping): Likewise.
23262 (spread_components): Likewise.
23263 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
23264 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
23265 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
23266 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
23267 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
23268 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
23269 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
23270 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
23271 (mark_threaded_blocks): Likewise.
23272 (thread_through_all_blocks): Likewise.
23273 * tree-ssa.c (verify_ssa): Likewise.
23274 (execute_update_addresses_taken): Likewise.
23275 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
23277 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23279 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
23281 (post_order_compute): Likewise.
23282 (inverted_post_order_compute): Likewise.
23283 (pre_and_rev_post_order_compute_fn): Likewise.
23285 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23287 * genrecog.c (int_set::int_set): Explicitly construct our
23288 auto_vec base class.
23289 * vec.h (auto_vec::auto_vec): New constructor.
23291 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23293 * bitmap.h (class auto_bitmap): New constructor taking
23294 bitmap_obstack * argument.
23296 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23298 * bitmap.h (class auto_bitmap): Change type of m_bits to
23299 bitmap_head, and adjust ctor / dtor and member operators.
23301 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
23303 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
23304 when returned register mode doesn't match original mode.
23306 2017-05-12 Jeff Law <law@redhat.com>
23307 Jakub Jelinek <jakub@redhat.com>
23309 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
23310 we look for cc setter after the compare-elim changes.
23311 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
23312 within the vector to match what compare-elim now expects.
23313 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
23314 (xorsi3_flags, one_cmplsi2_flags): Likewise.
23316 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
23317 after the compare-elim changes.
23318 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
23319 the vector to match what compare-elim now expects.
23320 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
23321 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
23322 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
23323 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
23324 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
23326 * config/visium/visium.c (single_set_and_flags): Fix where
23327 we look for cc setter after the compare-elim changes.
23328 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
23329 with the vector to match what compare-elim now expects.
23330 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
23331 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
23332 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
23333 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
23334 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
23335 (neg<mode>2_insn_set_overflow): Likewise.
23337 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
23339 PR middle-end/79794
23340 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
23341 maybe_expand_insn call, set ops[0].target. If still set after call,
23342 set alt_rtl. Add extra arg to recursive calls.
23343 (extract_bit_field): Add alt_rtl argument. Pass to
23345 * expmed.h (extract_bit_field): Fix prototype.
23346 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
23347 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
23348 to extract_bit_field_calls.
23349 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
23350 Pass alt_rtl to extract_bit_field calls.
23351 * calls.c (store_unaligned_arguments_into_psuedos)
23352 load_register_parameters): Pass extra NULL to extract_bit_field calls.
23353 * optabs.c (maybe_legitimize_operand): Clear op->target when call
23355 * optabs.h (struct expand_operand): Add target bitfield.
23357 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
23359 * compare-elim.c (try_eliminate_compare): Canonicalize
23360 operation with embedded compare to
23361 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
23362 (set (reg) (operation)].
23364 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
23366 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
23369 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
23370 cost of adding a carry flag for ADC instruction.
23371 [case MINUS]: Ignore the cost of subtracting a carry flag
23372 for SBB instruction.
23374 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
23376 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
23378 * config/rs6000/bmiintrin.h: New file.
23379 * config/rs6000/bmi2intrin.h: New file.
23380 * config/rs6000/x86intrin.h: New file.
23382 2017-05-12 Jeff Law <law@redhat.com>
23384 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
23387 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
23389 PR middle-end/80707
23390 * tree-cfg.c: Remove cfg edges of unreachable case statements.
23392 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23394 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
23395 early expansion of vector divide builtins.
23396 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
23397 builtins identified as having unsigned arguments.
23399 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23401 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
23402 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
23403 expansion of vector logical operations (and, andc, or, xor,
23406 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23408 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
23409 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
23411 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23413 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
23414 early GIMPLE expansion of vector multiplies.
23416 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23418 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
23419 TARGET_HAVE_MOVT conditional.
23420 (movt splitter): Likewise.
23422 2017-05-12 Richard Biener <rguenther@suse.de>
23424 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
23425 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
23426 Fold all stmts not inplace.
23428 2017-05-12 Richard Biener <rguenther@suse.de>
23430 PR tree-optimization/80713
23431 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
23432 inserted_exprs bit for not removed stmts.
23434 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
23436 PR middle-end/69921
23437 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
23438 parallelized" attribute for parallelized OpenACC kernels.
23439 * omp-offload.c (execute_oacc_device_lower): Use it.
23441 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
23442 Set "oacc kernels" attribute.
23443 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
23444 parameter. Adjust all users.
23445 (oacc_fn_attrib_kernels_p): Remove function.
23446 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
23447 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
23448 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
23449 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
23450 assert "oacc kernels" attribute is set.
23452 2017-05-11 Carl Love <cel@us.ibm.com>
23454 * config/rs6000/rs6000-c: Add support for built-in functions
23455 vector unsigned char vec_popcnt (vector signed char)
23456 vector unsigned char vec_popcnt (vector unsigned char)
23457 vector unsigned short vec_popcnt (vector signed short)
23458 vector unsigned short vec_popcnt (vector unsigned short)
23459 vector unsigned int vec_popcnt (vector signed int)
23460 vector unsigned int vec_popcnt (vector unsigned int)
23461 vector unsigned long long vec_popcnt (vector signed long long)
23462 vector unsigned long long vec_popcnt (vector unsigned long long)
23463 vector signed long long vec_slo (vector signed long long,
23464 vector signed char)
23465 vector signed long long vec_slo (vector signed long long,
23466 vector unsigned char)
23467 vector unsigned long long vec_slo (vector unsigned long long,
23468 vector signed char)
23469 vector unsigned long long vec_slo (vector unsigned long long,
23470 vector unsigned char)
23471 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
23472 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
23473 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
23474 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
23475 * doc/extend.texi: Update the built-in documentation file for the
23476 new built-in functions.
23478 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
23480 * attribs.h (sorted_attr_string): Move machine independent
23481 functions for target clone support from the i386 port to common
23482 code. Rename ix86_function_versions to common_function_versions.
23483 Rename make_name to make_unique_name.
23484 (common_function_versions): Likewise.
23485 (make_unique_name): Likewise.
23486 (make_dispatcher_decl): Likewise.
23487 (is_function_default_version): Likewise.
23488 * attribs.c (attr_strcmp): Likewise.
23489 (sorted_attr_string): Likewise.
23490 (common_function_versions): Likewise.
23491 (make_unique_name): Likewise.
23492 (make_dispatcher_decl): Likewise.
23493 (is_function_default_version): Likewise.
23494 * config/i386/i386.c (attr_strcmp): Likewise.
23495 (sorted_attr_string): Likewise.
23496 (ix86_function_versions): Likewise.
23497 (make_name): Likewise.
23498 (make_dispatcher_decl): Likewise.
23499 (is_function_default_version): Likewise.
23500 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
23502 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23505 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
23506 Account for direct move costs for vec_construct of integer
23509 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
23512 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
23513 (UNSPEC_STX_ATOMIC): Ditto.
23514 (loaddi_via_sse): New insn.
23515 (storedi_via_sse): Ditto.
23516 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
23517 Update corresponding peephole2 patterns.
23518 (atomic_storedi_fpu): Ditto.
23520 2017-05-11 Julia Koval <julia.koval@intel.com>
23522 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
23523 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
23525 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
23526 (__builtin_ia32_rsqrt14ss_mask): New builtins.
23527 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
23529 2017-05-11 Nathan Sidwell <nathan@acm.org>
23531 * graphite-poly.c: Include dumpfile.h.
23533 * dumpfle.h (dump_function): Declare here ...
23534 * tree-dump.h (dump_function): ... not here.
23535 * dumpfile.c: #include tree-cfg.h.
23536 (dump_function): Move here from ...
23537 * tree-dump.c (dump_function): ... here.
23538 * gimplify.c: #include splay-tree.h, not tree-dump.h.
23539 * graphite-poly.c: Don't include tree-dump.h.
23540 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
23541 * print-tree.c: Likewise.
23542 * stor-layout.c: Likewise.
23543 * tree-nested.c: Likewise.
23545 * dumpfile.c (dump_start): Use TDF_FLAGS.
23546 (dump_enable_all): Fix TDF_KIND check thinko.
23548 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
23550 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23551 array entries to represent two legal parameterizations of the
23552 overloaded __builtin_cmpb function, as represented by the
23553 P6_OV_BUILTIN_CMPB constant.
23554 (altivec_resolve_overloaded_builtin): Add special case handling
23555 for the __builtin_cmpb function, as represented by the
23556 P6_OV_BUILTIN_CMPB constant.
23557 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
23558 (BU_P6_64BIT_2): New macro.
23559 (BU_P6_OVERLOAD_2): New macro
23560 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
23561 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
23562 (CMPB): Add overload support to represent both 32-bit and 64-bit
23563 compare-bytes function.
23564 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
23565 support for TARGET_CMPB.
23566 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
23567 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23568 documentation of the __builtin_cmpb overloaded built-in function.
23570 2017-05-11 Richard Biener <rguenther@suse.de>
23572 PR tree-optimization/80705
23573 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
23574 bases are not vectorizable.
23576 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23578 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
23579 when counting register pressure.
23581 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23583 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
23584 (iv_ca_more_deps): Renamed to ...
23585 (iv_ca_compare_deps): ... this.
23586 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
23588 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23590 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
23592 (determine_group_iv_costs): ... here.
23593 (find_inv_vars_cb): Record inv var if it's not recorded before.
23595 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23597 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
23598 (get_shiftadd_cost): Ditto.
23600 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23602 * tree-ssa-address.c: Include header file.
23603 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
23605 (add_to_parts): Refactor.
23606 (addr_to_parts): New parameter. Update use of move_hint_to_base.
23607 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
23610 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23612 PR tree-optimization/53090
23613 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
23615 (extract_cond_operands): Detect condition with IV on both sides
23616 and return COMP_IV_EXPR_2.
23617 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
23618 (rewrite_use_compare): Simplify by removing call to function
23619 extract_cond_operands.
23621 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23623 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
23624 (extract_cond_operands): Detect condition comparing against non-
23625 invariant bound and return appropriate enum value.
23626 (find_interesting_uses_cond): Update use of extract_cond_operands.
23627 Handle its return value accordingly.
23628 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
23630 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23632 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
23633 nonlinear iv_use computation in loop invariant sensitive way.
23635 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23637 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
23638 (find_iv_candidates): Call relate_compare_use_with_all_cands.
23640 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23642 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
23643 (dump_cand): Support iv_cand.inv_exprs.
23644 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
23646 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
23649 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23651 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
23653 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
23654 as local function. Include necessary header files.
23655 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
23657 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23659 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
23661 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23663 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
23664 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
23665 RSHIFT_EXPR and BIT_NOT_EXPR.
23667 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23669 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
23670 (adjust_setup_cost): New parameter supporting round up adjustment.
23671 (struct address_cost_data): Delete.
23672 (force_expr_to_var_cost): Don't bound cost with spill_cost.
23673 (split_address_cost, ptr_difference_cost): Delete.
23674 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
23675 (struct ainc_cost_data): New struct.
23676 (get_address_cost_ainc): New function.
23677 (get_address_cost, get_computation_cost): Reimplement.
23678 (determine_group_iv_cost_address): Record inv_expr for all uses of
23680 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
23681 (iv_ca_has_deps): Reimplemented to ...
23682 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
23684 (iv_ca_extend): Call iv_ca_more_deps.
23686 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23688 * tree-ssa-address.c (struct mem_address): Move to header file.
23689 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
23690 * tree-ssa-address.h (struct mem_address): Move from C file.
23691 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
23693 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23695 * tree-affine.h (aff_combination_type): New interface.
23696 (aff_combination_zero_p): Remove static.
23697 (aff_combination_const_p): New interface.
23698 (aff_combination_singleton_var_p): New interfaces.
23700 2017-05-11 Richard Biener <rguenther@suse.de>
23702 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
23703 Skip unreachable blocks and destinations.
23704 (eliminate): Move stmt removal and fixup ...
23705 (fini_eliminate): ... here. Skip inserted exprs.
23706 (pass_pre::execute): Move fini_pre after fini_eliminate.
23707 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
23708 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
23709 PRE to get rid of dead code that has invalid SSA form and
23710 split critical edges again.
23712 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23714 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
23716 2017-05-11 Richard Biener <rguenther@suse.de>
23718 * passes.c (execute_function_todo): Verify loops if they are
23719 said to be up-to-date.
23720 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
23721 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
23723 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
23726 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
23727 handle calling assemble_external ourself.
23730 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
23731 modes with zero size. Enhance comment.
23733 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23735 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
23736 built-ins for vec_xl and vec_xst with short and char pointer
23739 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
23741 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
23742 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
23743 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
23744 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
23745 (_mm_maskz_min_round_ss): New intrinsics.
23746 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
23747 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
23748 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
23749 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
23750 (__builtin_ia32_minss_mask_round): New builtins.
23751 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23752 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
23753 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
23755 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
23756 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
23758 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
23761 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
23763 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
23764 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
23765 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
23766 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
23767 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
23768 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
23769 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
23770 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23771 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
23772 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
23773 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
23774 (__builtin_ia32_mulss_mask_round): New builtins.
23775 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23776 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
23777 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
23779 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
23780 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
23782 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
23785 2017-05-10 Julia Koval <julia.koval@intel.com>
23787 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
23788 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
23789 (_mm256_setr_m128i): New intrinsics.
23791 2017-05-10 Julia Koval <julia.koval@intel.com>
23793 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
23794 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
23795 (_mm_maskz_rcp14_ss): New intrinsics.
23796 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
23797 (__builtin_ia32_rcp14ss_mask): New builtins.
23798 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
23800 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
23802 PR tree-optimization/51513
23803 * tree-cfg.c (gimple_seq_unreachable_p): New function.
23804 (assert_unreachable_fallthru_edge_p): Use it.
23805 (group_case_labels_stmt): Likewise.
23806 * tree-cfg.h: Prototype it.
23807 * stmt.c: Include cfghooks.h and tree-cfg.h.
23808 (emit_case_dispatch_table) <gap_label>: New local variable.
23809 Use it to fill dispatch table gaps.
23810 Test for default_label before updating probabilities.
23811 (expand_case) <default_label>: Remove unneeded initialization.
23812 Test for unreachable default case statement and remove its edge.
23813 Set default_label accordingly.
23814 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
23816 2017-05-10 Carl Love <cel@us.ibm.com>
23818 * config/rs6000/rs6000-c: Add support for built-in functions
23819 vector signed char vec_neg (vector signed char)
23820 vector signed short int vec_neg (vector short int)
23821 vector signed int vec_neg (vector signed int)
23822 vector signed long long vec_neg (vector signed long long)
23823 vector float vec_neg (vector float)
23824 vector double vec_neg (vector double)
23825 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
23827 * config/rs6000/altivec.h: Add define for vec_neg
23828 * doc/extend.texi: Update the built-in documentation for the
23829 new built-in functions.
23831 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23833 PR tree-optimization/77644
23834 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
23836 2017-05-10 Nathan Sidwell <nathan@acm.org>
23838 * dumpfile.h (TDI_lang_all): New.
23839 (TDF_KIND): New. Renumber others
23840 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
23842 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
23844 (get_dump_file_name): Adjust suffix generation.
23845 (dump_enable_all): Use TDF_KIND.
23846 * doc/invoke.texi (-fdump-lang-all): Document.
23848 * dumpfile.h: Tabify.
23850 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
23853 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
23854 Move member access before delete.
23856 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
23858 * tree-inline.c (expand_call_inline): Split block at stmt
23861 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
23864 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
23865 are now unused after splitting mov{sf,sd}_hardfloat.
23866 (f32_lr2): Likewise.
23867 (f32_lm): Likewise.
23868 (f32_lm2): Likewise.
23869 (f32_li): Likewise.
23870 (f32_li2): Likewise.
23871 (f32_lv): Likewise.
23872 (f32_sr): Likewise.
23873 (f32_sr2): Likewise.
23874 (f32_sm): Likewise.
23875 (f32_sm2): Likewise.
23876 (f32_si): Likewise.
23877 (f32_si2): Likewise.
23878 (f32_sv): Likewise.
23879 (f32_dm): Likewise.
23880 (f32_vsx): Likewise.
23881 (f32_av): Likewise.
23882 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
23883 For movsf, order stores so the VSX stores occur before the GPR
23884 store which encourages the register allocator to use a traditional
23885 FPR instead of a GPR. For movsd, order the stores so that the GPR
23886 store comes before the VSX stores to allow the power6 to work.
23887 This is due to the power6 not having a 32-bit integer store
23888 instruction from a FPR.
23889 (movsf_hardfloat): Likewise.
23890 (movsd_hardfloat): Likewise.
23892 2017-05-09 Martin Sebor <msebor@redhat.com>
23894 PR translation/80280
23895 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
23898 PR translation/80280
23899 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
23900 data member added in r247778.
23901 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
23903 2017-05-09 Nathan Sidwell <nathan@acm.org>
23905 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
23907 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
23910 2017-05-09 Marek Polacek <polacek@redhat.com>
23912 * doc/invoke.texi: Fix typo.
23914 2017-05-09 Richard Biener <rguenther@suse.de>
23916 * tree-vrp.c (vrp_val_is_max): Adjust comment.
23917 (vrp_val_is_min): Likewise.
23918 (set_value_range_to_value): Likewise.
23919 (set_value_range_to_nonnegative): Likewise.
23920 (gimple_assign_nonzero_p): Likewise.
23921 (gimple_stmt_nonzero_p): Likewise.
23922 (vrp_int_const_binop): Likewise. Remove unreachable case.
23923 (adjust_range_with_scev): Adjust comments.
23924 (compare_range_with_value): Likewise.
23925 (extract_range_from_phi_node): Likewise.
23926 (test_for_singularity): Likewise.
23928 2017-05-09 Richard Biener <rguenther@suse.de>
23930 * tree-vrp.c (get_single_symbol): Add assert that we don't
23931 get overflowed constants as invariant part.
23932 (compare_values_warnv): Add comment before the TREE_NO_WARNING
23933 checks. Use wi::cmp instead of recursing for integer constants.
23934 (compare_values): Just ignore whether we assumed undefined
23935 overflow instead of failing the compare.
23936 (extract_range_for_var_from_comparison_expr): Add comment before the
23937 TREE_NO_WARNING sets.
23938 (test_for_singularity): Likewise.
23939 (extract_range_from_comparison): Do not disable optimization
23940 when we assumed undefined overflow.
23941 (extract_range_basic): Remove init of unused var.
23943 2017-05-09 Richard Biener <rguenther@suse.de>
23945 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
23946 (extract_range_from_multiplicative_op_1): Adjust.
23947 (extract_range_from_binary_expr_1): Use int_const_binop.
23949 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
23952 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
23953 rs6000_store_data_bypass_p in seven define_bypass directives and
23954 in several comments.
23955 * config/rs6000/rs6000-protos.h: Add prototype for
23956 rs6000_store_data_bypass_p function.
23957 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
23958 function implements slightly different (rs6000-specific) semantics
23959 than store_data_bypass_p, returning false rather than aborting
23960 with assertion error when arguments do not satisfy the
23961 requirements of store data bypass.
23962 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
23963 rs6000_store_data_bypass_p.
23965 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
23967 * config/xtensa/xtensa-protos.h
23968 (xtensa_initial_elimination_offset): New declaration.
23969 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
23970 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
23971 macro definition, add case for FRAME_POINTER_REGNUM when
23972 FRAME_GROWS_DOWNWARD.
23973 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
23974 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
23975 xtensa_initial_elimination_offset.
23977 2017-05-08 Nathan Sidwell <nathan@acm.org>
23979 * doc/invoke.texi: Alphabetize -fdump options.
23981 2017-05-08 Martin Sebor <msebor@redhat.com>
23983 PR translation/80280
23984 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
23986 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
23988 * target.def (compute_frame_layout): New optional target hook.
23989 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
23990 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
23991 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
23993 * reload1.c (verify_initial_elim_offsets): Likewise.
23994 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
23995 (use_simple_return_p): Call arm_compute_frame_layout if needed.
23996 (arm_get_frame_offsets): Split up into this ...
23997 (arm_compute_frame_layout): ... and this function.
23999 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
24001 * config/aarch64/constraints.md (Usa): New constraint.
24002 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
24004 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
24006 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
24007 with_multilib_list after it has been checked.
24009 2017-05-08 Richard Biener <rguenther@suse.de>
24011 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
24012 (bitmap_set_subtract_values): Likewise.
24014 2017-05-08 Richard Biener <rguenther@suse.de>
24016 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
24017 (gimple_assign_nonzero): ... this and remove strict_overflow_p
24019 (gimple_stmt_nonzero_warnv_p): Rename to ...
24020 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
24022 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
24023 (extract_range_basic): Adjust, do not disable propagation on
24024 strict overflow sensitive simplification.
24025 (vrp_visit_cond_stmt): Likewise.
24027 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
24029 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
24030 body size unconditionally.
24032 2017-05-07 Jeff Law <law@redhat.com>
24035 2017-05-06 Jeff Law <law@redhat.com>
24036 PR tree-optimization/78496
24037 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
24040 PR tree-optimization/78496
24041 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
24042 (simplify_stmt_using_ranges): Call it.
24043 (vrp_dom_walker::before_dom_children): Extract equivalences
24044 from an ASSERT_EXPR with an equality comparison against a
24047 2017-05-06 Jeff Law <law@redhat.com>
24049 PR tree-optimization/78496
24050 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
24053 PR tree-optimization/78496
24054 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
24055 (simplify_stmt_using_ranges): Call it.
24056 (vrp_dom_walker::before_dom_children): Extract equivalences
24057 from an ASSERT_EXPR with an equality comparison against a
24060 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
24062 * lra-constraints.c (lra_copy_reg_equiv): New function.
24063 (split_reg): Use it to copy equivalence information from the
24064 original register to the spill register.
24066 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
24068 PR rtl-optimization/75964
24069 * simplify-rtx.c (simplify_const_relational_operation): Remove
24070 invalid handling of comparisons of integer ABS.
24072 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
24074 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
24075 initialize to zero.
24076 (init_regs): Remove declaration.
24077 (function_arg_advance_32): Initialize error_p as boolean variable.
24079 2017-05-05 Nathan Sidwell <nathan@acm.org>
24081 * store-motion.c (remove_reachable_equiv_notes): Reformat long
24082 lines. Use for (;;).
24084 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24086 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
24087 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
24088 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
24089 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
24090 VF=2 that require versioning.
24092 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24094 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
24097 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24099 * diagnostic.h (diagnostic_override_option_index): Convert from
24100 macro to inline function.
24102 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24104 * diagnostic.c (last_module_changed_p): New function.
24105 (set_last_module): New function.
24106 (diagnostic_report_current_module): Convert macro usage to
24107 the above functions.
24108 * diagnostic.h (diagnostic_context::last_module): Strengthen
24109 from const line_map * to const line_map_ordinary *.
24110 (diagnostic_last_module_changed): Delete macro.
24111 (diagnostic_set_last_module): Delete macro.
24113 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24115 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
24116 with diagnostic_report_diagnostic.
24117 (diagnostic_n_impl_richloc): Likewise.
24118 * diagnostic.h (report_diagnostic): Delete macro.
24119 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
24120 with diagnostic_report_diagnostic.
24121 * substring-locations.c (format_warning_va): Likewise.
24123 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24125 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
24126 save/restor of format_spec. Move option-printing code to...
24127 (print_option_information): ...this new function, and
24128 reimplement by simply printing to the pretty_printer,
24129 rather than appending to the format string.
24131 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24133 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
24134 handling logic into...
24135 (update_effective_level_from_pragmas): ...this new function.
24137 2017-05-04 Andrew Waterman <andrew@sifive.com>
24139 * config/riscv/riscv.opt (mstrict-align): New option.
24140 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
24141 (SLOW_UNALIGNED_ACCESS): Define.
24142 (riscv_slow_unaligned_access): Declare.
24143 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
24145 (riscv_slow_unaligned_access): New variable.
24146 (rocket_tune_info): Set slow_unaligned_access to true.
24147 (optimize_size_tune_info): Set slow_unaligned_access to false.
24148 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
24149 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
24150 (riscv_option_override): Set riscv_slow_unaligned_access.
24151 * doc/invoke.texi: Add -mstrict-align to RISC-V.
24153 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
24155 * config/riscv/riscv.md: Unify indentation.
24157 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
24162 * config/rs6000/rs6000.md (u code attribute): Add FIX and
24164 (extendsi<mode>2): Add support for doing sign extension via
24165 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
24166 don't have ISA 3.0 instructions.
24167 (extendsi<mode>2 splitter): Likewise.
24168 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
24169 generate the normal insns since SImode can now go in vector
24170 registers. Disallow the special UNSPECs needed for previous
24171 machines to hide SImode being used. Add new insns
24172 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
24173 (fix_trunc<mode>si2_stfiwx): Likewise.
24174 (fix_trunc<mode>si2_internal): Likewise.
24175 (fixuns_trunc<mode>si2): Likewise.
24176 (fixuns_trunc<mode>si2_stfiwx): Likewise.
24177 (fctiw<u>z_<mode>_smallint): Likewise.
24178 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
24179 of floating point to 32-bit integer from doing a direct move to
24180 the GPR registers to do a store.
24181 (fctiwz_<mode>): Break long line.
24183 2017-05-05 Bin Cheng <bin.cheng@arm.com>
24185 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
24186 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
24187 (addr_list, addr_offset_valid_p): New.
24188 (split_address_groups): Check offset validity with above function.
24189 (gt-tree-ssa-loop-ivopts.h): Include header file.
24191 2017-05-05 Nathan Sidwell <nathan@acm.org>
24193 * config.gcc (arm*-*-*): Add missing 'fi'.
24195 2017-05-05 Steve Ellcey <sellcey@cavium.com>
24197 * doc/invoke.texi (-fopt-info): Explicitly say order of options
24198 included in -fopt-info does not matter.
24199 * doc/optinfo.texi (-fopt-info): Fix description of default
24200 behavour. Explicitly say order of options included in -fopt-info
24203 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
24205 * config.gcc: Allow combinations of aprofile and rmprofile values for
24206 --with-multilib-list.
24207 * config/arm/t-multilib: New file.
24208 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
24209 variables. Remove setting of ISA and floating-point ABI in
24210 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
24211 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
24212 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
24213 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
24215 * config/arm/t-rmprofile: Likewise except for the matches changes.
24216 * doc/install.texi (--with-multilib-list): Document the combination of
24217 aprofile and rmprofile values and warn about pitfalls in doing that.
24219 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
24221 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
24222 (movdi_aarch64): Likewise.
24224 2017-05-05 Jakub Jelinek <jakub@redhat.com>
24226 PR tree-optimization/80632
24227 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
24229 (build_arrays): Initialize it for virtual phis.
24230 (fix_phi_nodes): Use it for virtual phis.
24232 PR tree-optimization/80558
24233 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
24234 [x, y] op z into [x op, y op z] for op & or | if conditions
24237 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
24238 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24241 * config/arm/arm.md (use_literal_pool): Remove.
24242 (64-bit immediate split): No longer takes cost into consideration
24243 if arm_disable_literal_pool is enabled.
24244 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
24245 used when arm_disable_literal_pool is enabled.
24246 (arm_max_const_double_inline_cost): Remove use of
24247 arm_disable_literal_pool.
24248 (push_minipool_fix): Add assert.
24249 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
24250 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
24251 (no_literal_pool_sf_immediate): New.
24253 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24255 PR tree-optimization/80613
24256 * tree-ssa-dce.c (propagate_necessity): Remove cases for
24257 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
24259 2017-05-05 Richard Biener <rguenther@suse.de>
24261 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
24263 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
24265 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
24266 of this flag from insn conditions due to removal from r247495.
24268 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
24270 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
24272 (arm_early_store_addr_dep_ptr): Likewise.
24273 * config/arm/aarch-common-protos.h
24274 (arm_early_load_addr_dep_ptr): Add prototype.
24275 (arm_early_store_addr_dep_ptr): Likewise.
24276 * config/arm/cortex-a53.md: Add new bypasses.
24278 2017-05-05 Jakub Jelinek <jakub@redhat.com>
24280 * tree.c (next_type_uid): Change type to unsigned.
24281 (type_hash_canon): Decrement back next_type_uid if
24282 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
24283 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
24286 2017-05-04 Martin Sebor <msebor@redhat.com>
24288 * builtins.c: Fix a trivial typo in a comment.
24290 PR middle-end/79234
24291 * builtins.c (check_sizes): Adjust to handle reading past the end.
24292 Avoid printing excessive upper bound of ranges. Use %E to print
24293 tree nodes instead of converting them to %wu.
24294 (expand_builtin_memchr): New function.
24295 (compute_dest_size): Rename...
24296 (compute_objsize): ...to this.
24297 (expand_builtin_memcpy): Adjust.
24298 (expand_builtin_mempcpy): Adjust.
24299 (expand_builtin_strcat): Adjust.
24300 (expand_builtin_strcpy): Adjust.
24301 (check_strncat_sizes): Adjust.
24302 (expand_builtin_strncat): Adjust.
24303 (expand_builtin_strncpy): Adjust and simplify.
24304 (expand_builtin_memset): Adjust.
24305 (expand_builtin_bzero): Adjust.
24306 (expand_builtin_memcmp): Adjust.
24307 (expand_builtin): Handle memcmp.
24308 (maybe_emit_chk_warning): Check strncat just once.
24310 2017-05-04 Martin Sebor <msebor@redhat.com>
24312 PR preprocessor/79214
24313 PR middle-end/79222
24314 PR middle-end/79223
24315 * builtins.c (check_sizes): Add inlining context and issue
24316 warnings even when -Wno-system-headers is set.
24317 (check_strncat_sizes): Same.
24318 (expand_builtin_strncat): Same.
24319 (expand_builtin_memmove): New function.
24320 (expand_builtin_stpncpy): Same.
24321 (expand_builtin): Handle memmove and stpncpy.
24323 2017-05-04 Bin Cheng <bin.cheng@arm.com>
24325 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
24326 which is not used any more.
24328 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24330 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
24332 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24334 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
24335 (cortexa53_tunings): Likewise.
24336 (cortexa57_tunings): Likewise.
24337 (cortexa72_tunings): Likewise.
24338 (cortexa73_tunings): Likewise.
24340 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24342 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
24343 Set loop alignment to 8.
24345 2017-05-04 Martin Sebor <msebor@redhat.com>
24347 PR translation/80280
24348 * builtins.c (expand_builtin_object_size): Add missing quoting to
24349 %D and like directives.
24350 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
24351 (hsa_type_for_tree_type): Same.
24352 (verify_function_arguments): Same.
24353 * symtab.c (symbol_table::change_decl_assembler_name): Same.
24354 * varasm.c (get_section): Same.
24357 2017-05-04 Martin Sebor <msebor@redhat.com>
24359 PR translation/80280
24360 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
24362 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24364 * config/aarch64/aarch64.c (generic_addrcost_table):
24365 Change HI/TI mode setting.
24367 2017-05-04 Martin Jambor <mjambor@suse.cz>
24369 PR tree-optimization/80622
24370 * tree-sra.c (comes_initialized_p): New function.
24371 (build_accesses_from_assign): Only set write lazily when
24372 comes_initialized_p is false.
24373 (analyze_access_subtree): Use comes_initialized_p.
24374 (propagate_subaccesses_across_link): Assert !comes_initialized_p
24375 instead of testing for PARM_DECL.
24377 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24379 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
24380 constraint on operand 0 to allow more general addressing modes.
24381 Adjust output template.
24382 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
24384 * config/aarch64/aarch64-protos.h
24385 (aarch64_address_valid_for_prefetch_p): Declare prototype.
24386 * config/aarch64/constraints.md (Dp): New address constraint.
24387 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
24390 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
24392 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
24393 update use of estimate_ipcp_clone_size_and_time.
24394 (estimate_local_effects): Update use of
24395 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
24396 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
24397 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
24398 Return nonspecialized time.
24400 2017-05-04 Richard Biener <rguenther@suse.de>
24402 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
24403 for the last VUSE which def dominates the PHI. Directly call
24405 (get_continuation_for_phi_1): Remove.
24407 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
24409 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
24410 to explain the use of truncating division. Cap the number of
24411 iterations to the maximum given by nb_iterations_upper_bound,
24414 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24416 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
24417 * configure: Regenerate.
24418 * config.in: Regenerate.
24419 * config/i386/driver-mingw32.c: new file.
24420 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
24421 * config.host: Link driver-mingw32.o on MinGW host.
24422 * doc/install.texi: Document new --enable-mingw-wildcard configure
24425 2017-05-04 Marek Polacek <polacek@redhat.com>
24427 PR tree-optimization/80612
24428 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
24430 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24431 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
24433 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
24434 (movt splitter): Likewise.
24435 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
24436 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
24437 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
24438 block for Thumb-1 with MOVT.
24439 (thumb2_legitimate_address_p): Move code block ...
24440 (can_avoid_literal_pool_for_label_p): ... into this new function.
24441 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
24443 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
24444 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
24445 "M-profile targets with the MOVT instruction".
24447 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24449 * config/arm/arm-builtins.c (arm_init_builtins): Rename
24450 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
24451 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
24453 2017-05-04 Martin Liska <mliska@suse.cz>
24455 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
24456 variable cond_code.
24458 2017-05-04 Richard Biener <rguenther@suse.de>
24460 * tree.c (array_at_struct_end_p): Handle arrays at struct
24461 end with flexarrays more conservatively. Refactor and treat
24462 arrays of arrays or aggregates more strict. Fix
24463 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
24464 * tree.c (array_at_struct_end_p): Adjust prototype.
24465 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
24466 * gimple-fold.c (get_range_strlen): Likewise.
24467 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
24469 2017-05-04 Richard Biener <rguenther@suse.de>
24471 PR tree-optimization/31130
24472 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
24474 (supports_overflow_infinity): Likewise.
24475 (is_negative_overflow_infinity): Likewise.
24476 (is_positive_overflow_infinity): Likewise.
24477 (is_overflow_infinity): Likewise.
24478 (stmt_overflow_infinity): Likewise.
24479 (overflow_infinity_range_p): Likewise.
24480 (usable_range_p): Remove as always returning true.
24481 (make_overflow_infinity): Remove.
24482 (negative_overflow_infinity): Likewise.
24483 (positive_overflow_infinity): Likewise.
24484 (avoid_overflow_infinity): Likewise.
24485 (set_value_range): Adjust accordingly.
24486 (set_value_range_to_nonnegative): Likewise, remove now unused
24487 overflow_infinity arg.
24488 (vrp_operand_equal_p): Adjust.
24489 (update_value_range): Likewise.
24490 (range_int_cst_singleton_p): Likewise.
24491 (operand_less_p): Likewise.
24492 (compare_values_warnv): Likewise.
24493 (extract_range_for_var_from_comparison_expr): Likewise.
24494 (vrp_int_const_binop): Likewise.
24495 (zero_nonzero_bits_from_vr): Likewise.
24496 (extract_range_from_multiplicative_op_1): Likewise.
24497 (extract_range_from_binary_expr_1): Likewise.
24498 (extract_range_from_unary_expr): Likewise.
24499 (extract_range_from_comparison): Likewise.
24500 (extract_range_basic): Likewise.
24501 (adjust_range_with_scev): Likewise.
24502 (compare_ranges): Likewise.
24503 (compare_range_with_value): Likewise.
24504 (dump_value_range): Likewise.
24505 (test_for_singularity): Likewise, remove strict_overflow_p parameter
24507 (simplify_cond_using_ranges): Adjust.
24509 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
24511 * brig-builtins.def: Added a builtin for class_f64.
24512 * builtin-types.def: Added a builtin type needed by class_f64.
24514 2017-05-03 Jason Merrill <jason@redhat.com>
24516 * timevar.def: Add TV_CONSTEXPR.
24518 2017-05-03 David Malcolm <dmalcolm@redhat.com>
24520 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
24522 2017-05-03 Martin Jambor <mjambor@suse.cz>
24524 * ipa-prop.c (ipa_update_after_lto_read): Removed.
24525 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
24526 * ipa-cp.c (ipcp_propagate_stage): Do not call
24527 ipa_update_after_lto_read.
24528 * ipa-inline.c (ipa_inline): Likewise.
24530 2017-05-03 Martin Jambor <mjambor@suse.cz>
24532 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
24533 tag. Added a default constructor and a destructor.
24534 (ipa_edge_args_sum_t): New class;
24535 (ipa_edge_args_sum): Declare.
24536 (ipa_edge_args_vector): Remove declaration.
24537 (IPA_EDGE_REF): Use ipa_edge_args_sum.
24538 (ipa_free_edge_args_substructures): Remove declaration.
24539 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
24540 (ipa_edge_args_info_available_for_edge_p): Likewise.
24541 * ipa-prop.c (ipa_edge_args_vector): Removed.
24542 (edge_removal_hook_holder): Likewise.
24543 (edge_duplication_hook_holder): Likewise.
24544 (ipa_edge_args_sum): New variable.
24545 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
24546 ipa_edge_args_vector.
24547 (ipa_free_edge_args_substructures): Likewise.
24548 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
24549 ipa_edge_args_vector.
24550 (ipa_edge_removal_hook): Turned into method
24551 ipa_edge_args_sum_t::remove.
24552 (ipa_edge_duplication_hook): Turned into method
24553 ipa_edge_args_sum_t::duplicate.
24554 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
24555 registering edge hooks.
24556 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
24557 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
24558 ipa_edge_args_sum instead of ipa_edge_args_vector.
24559 * ipa-profile.c (ipa_profile): Likewise.
24561 2017-05-03 Martin Jambor <mjambor@suse.cz>
24563 * symbol-summary.h (function_summary): New method exists.
24564 (function_summary::symtab_removal): Deallocate through release.
24565 (call_summary): New class.
24566 (gt_ggc_mx): New overload.
24567 (gt_pch_nx): Likewise.
24568 (gt_pch_nx): Likewise.
24570 2017-05-03 Jeff Law <law@redhat.com>
24572 PR tree-optimization/78496
24573 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
24574 from simplify_cond_using_ranges. Split off code to walk
24575 backwards through casts into ...
24576 (simplify_cond_using_ranges_2): New function.
24577 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
24578 (execute_vrp): After identifying jump threads, call
24579 simplify_cond_using_ranges_2.
24581 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
24584 * ipa-inline.h (inline_summary): Add ctor.
24585 (create_ggc): Do not use ggc_cleared_alloc.
24587 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
24588 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24590 * gcc.c (handle_braces): Support escaping in switch matching
24592 * doc/invoke.texi (Spec Files): Document it.
24593 Remove superfluous @code markup in items.
24595 2017-05-03 David Malcolm <dmalcolm@redhat.com>
24597 * diagnostic-show-locus.c (struct column_range): New struct.
24598 (get_affected_columns): New function.
24599 (get_printed_columns): New function.
24600 (struct correction): New struct.
24601 (correction::ensure_capacity): New function.
24602 (correction::ensure_terminated): New function.
24603 (struct line_corrections): New struct.
24604 (line_corrections::~line_corrections): New dtor.
24605 (line_corrections::add_hint): New function.
24606 (layout::print_trailing_fixits): Reimplement in terms of the new
24608 (selftest::test_overlapped_fixit_printing): New function.
24609 (selftest::diagnostic_show_locus_c_tests): Call it.
24611 2017-05-03 Nathan Sidwell <nathan@acm.org>
24613 Canonicalize canonical type hashing
24614 * tree.h (type_hash_canon_hash): Declare.
24615 * tree.c (type_hash_list, attribute_hash_list): Move into
24616 type_hash_canon_hash.
24617 (build_type_attribute_qual_variant): Break out hash code calc into
24618 type_hash_canon_hash.
24619 (type_hash_canon_hash): New. Generic type hash computation.
24620 (build_range_type_1, build_array_type_1, build_function_type,
24621 build_method_type_directly, build_offset_type, build_complex_type,
24622 make_vector_type): Call it.
24624 2017-05-03 Richard Biener <rguenther@suse.de>
24626 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
24627 When all DRs have unknown misaligned do not always peel
24628 when there is a store but apply the same costing model as if
24629 there were only loads.
24631 2017-05-03 Richard Biener <rguenther@suse.de>
24634 PR tree-optimization/80492
24635 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
24636 compare_base_decls returning dont-know properly.
24638 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
24640 * config/arm/iterators.md (CCSI): New mode iterator.
24641 (arch): New mode attribute.
24642 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
24643 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
24644 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
24645 code iterator for success result mode.
24646 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
24647 the corresponding new insn generators.
24649 2017-05-03 Bin Cheng <bin.cheng@arm.com>
24652 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24653 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24655 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
24657 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
24658 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
24659 (DDR_A): Wrap DDR argument in brackets.
24660 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
24661 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
24662 (DDR_REVERSED_P): Likewise.
24664 2017-05-03 Jakub Jelinek <jakub@redhat.com>
24666 PR tree-optimization/79472
24667 * tree-switch-conversion.c (struct switch_conv_info): Add
24668 contiguous_range and default_case_nonstandard fields.
24669 (collect_switch_conv_info): Compute contiguous_range and
24670 default_case_nonstandard fields, don't clear final_bb if
24671 contiguous_range and only the default case doesn't have the required
24673 (check_all_empty_except_final): Set default_case_nonstandard instead
24674 of failing if contiguous_range and the default case doesn't have empty
24676 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
24677 and only the default case doesn't have the required constants. Skip
24679 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
24680 if default_case_nonstandard.
24681 (build_constructors): Build constant 1 just once. Assert that default
24682 values aren't inserted in between cases if contiguous_range. Skip
24684 (build_arrays): Skip virtual phis.
24685 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
24686 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
24687 Handle virtual phis.
24688 (gen_inbound_check): Handle default_case_nonstandard case.
24689 (process_switch): Adjust check_final_bb caller. Call
24690 gather_default_values with the first non-default case instead of
24691 default case if default_case_nonstandard.
24693 2017-05-02 Nathan Sidwell <nathan@acm.org>
24695 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
24696 check. Fix formatting.
24698 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
24700 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
24701 errors when comparing specialized and unspecialized times.
24703 2017-05-02 David Malcolm <dmalcolm@redhat.com>
24705 * diagnostic-show-locus.c
24706 (layout::should_print_annotation_line_p): Make private.
24707 (layout::print_annotation_line): Make private.
24708 (layout::annotation_line_showed_range_p): Make private.
24709 (layout::show_ruler): Make private.
24710 (layout::print_source_line): Make private. Pass in line and
24711 line_width, rather than calling location_get_source_line. Drop
24713 (layout::print_leading_fixits): New method.
24714 (layout::print_any_fixits): Rename to...
24715 (layout::print_trailing_fixits): ...this, and make private.
24716 Don't print newline fixits.
24717 (diagnostic_show_locus): Move logic for printing one row into...
24718 (layout::print_line): ...this new function. Move the
24719 location_get_source_line call and error-handling from
24720 print_source_line to here. Call print_leading_fixits, and rename
24721 print_any_fixits to print_trailing_fixits.
24722 (selftest::test_fixit_insert_containing_newline): Update now that
24723 newlines are partially supported.
24724 (selftest::test_fixit_insert_containing_newline_2): New test.
24725 (selftest::test_fixit_replace_containing_newline): Update comments.
24726 (selftest::diagnostic_show_locus_c_tests): Call the new test.
24727 * edit-context.c (class added_line): New class.
24728 (class edited_line): Describe newline handling in comment.
24729 (edited_line::actually_edited_p): New method.
24730 (edited_line::print_content): Delete redundant decl.
24731 (edited_line::m_predecessors): New field.
24732 (edited_file::print_content): Call edited_line::print_content.
24733 (edited_file::print_diff): Update to support newlines.
24734 (edited_file::print_diff_hunk): Likewise.
24735 (edited_file::print_run_of_changed_lines): New function.
24736 (edited_file::print_diff_line): Convert to...
24737 (print_diff_line): ...this.
24738 (edited_file::get_effective_line_count): New function.
24739 (edited_line::edited_line): Initialize new field m_predecessors.
24740 (edited_line::~edited_line): Clean up m_predecessors.
24741 (edited_line::apply_fixit): Handle newlines.
24742 (edited_line::get_effective_line_count): New function.
24743 (edited_line::print_content): New function.
24744 (edited_line::print_diff_lines): New function.
24745 (selftest::test_applying_fixits_insert_containing_newline): New
24747 (selftest::test_applying_fixits_replace_containing_newline): New
24749 (selftest::insert_line): New function.
24750 (selftest::test_applying_fixits_multiple_lines): Add example of
24752 (selftest::edit_context_c_tests): Call the new tests.
24754 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24756 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
24757 parameter cand. Update dump information.
24758 (get_computation_cost): Update uses.
24760 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24762 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
24763 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
24764 (get_computation_at, rewrite_use_address): Update use of
24765 get_computation_aff.
24767 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24769 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
24770 (get_computation): Delete.
24771 (get_computation_cost): Implement like get_computation_cost_at.
24772 Use get_computation_at.
24773 (get_computation_cost_at): Delete.
24774 (rewrite_use_nonlinear_expr): Use get_computation_at.
24775 (rewrite_use_compare, remove_unused_ivs): Ditto.
24777 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24779 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
24781 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24783 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
24784 (ivopts_global_cost_for_size): Rename parameter and update uses.
24785 (iv_ca_recount_cost): Update uses.
24786 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
24787 candidates seperately in n_invs and n_cands.
24788 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
24790 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24792 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
24793 (find_inv_vars_cb): New.
24794 (find_depends): Renamed to ...
24795 (find_inv_vars): ... this.
24796 (add_candidate_1, force_var_cost): Call find_inv_vars.
24797 (split_address_cost, determine_group_iv_cost_cond): Ditto.
24799 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24801 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
24802 inv_vars. Add inv_exprs.
24803 (struct iv_cand): Rename depends_on to inv_vars.
24804 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
24805 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
24806 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
24807 (dump_cand): Dump inv_vars.
24808 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
24809 (record_invariant, find_depends, add_candidate_1): Ditto.
24810 (set_group_iv_cost, force_var_cost): Ditto.
24811 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
24812 (get_computation_cost_at, get_computation_cost): Ditto.
24813 (determine_group_iv_cost_generic): Ditto.
24814 (determine_group_iv_cost_address): Ditto.
24815 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
24816 (determine_group_iv_costs): Ditto.
24817 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
24818 (iv_ca_set_remove_invariants): Renamed to ...
24819 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
24820 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
24821 (iv_ca_set_add_invariants): Renamed to ...
24822 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
24823 (iv_ca_set_cp): Use iv_ca_set_add_invs.
24824 (iv_ca_has_deps): Support inv_vars and inv_exprs.
24825 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
24826 (create_new_ivs): Remove useless dump.
24828 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24830 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
24832 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
24833 (iv_ca_set_no_cp, create_new_iv): Ditto.
24835 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24837 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24839 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
24841 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
24842 function tree_check2.
24844 2017-05-02 Martin Liska <mliska@suse.cz>
24846 * doc/gcov.texi: Add missing preposition.
24847 * gcov.c (function_info::function_info): Properly fill up
24848 all member variables.
24850 2017-05-02 Tamar Christina <tamar.christina@arm.com>
24852 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
24854 2017-05-02 Tamar Christina <tamar.christina@arm.com>
24856 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
24858 2017-05-02 Martin Liska <mliska@suse.cz>
24861 * lto-streamer-in.c (lto_read_tree_1): Remove
24862 LTO_STREAMER_DEBUG.
24863 * lto-streamer.c (struct tree_hash_entry): Likewise.
24864 (struct tree_entry_hasher): Likewise.
24865 (tree_entry_hasher::hash): Likewise.
24866 (tree_entry_hasher::equal): Likewise.
24867 (lto_streamer_init): Likewise.
24868 (lto_orig_address_map): Likewise.
24869 (lto_orig_address_get): Likewise.
24870 (lto_orig_address_remove): Likewise.
24871 * lto-streamer.h: Likewise.
24872 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
24873 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
24875 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
24877 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
24878 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
24879 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
24880 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
24881 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
24882 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
24883 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
24884 (mm_maskz_sub_ss): New intrinsics.
24885 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24886 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
24887 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
24888 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
24889 (__builtin_ia32_subss_mask_round): New builtins.
24890 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24891 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24892 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
24894 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
24895 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
24897 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
24900 2017-05-02 Martin Jambor <mjambor@suse.cz>
24902 PR tree-optimization/78687
24903 * tree-sra.c (access): New field parent.
24904 (process_subtree_disqualification): New function.
24905 (disqualify_candidate): Call it.
24906 (build_accesses_from_assign): Reset write flag if creating an
24908 (build_access_subtree): Fill in parent field and also prpagate
24909 down grp_write flag.
24910 (create_artificial_child_access): New parameter set_grp_write, set
24911 grp_write to its value.
24912 (propagate_subaccesses_across_link): Also propagate grp_write flag
24914 (propagate_all_subaccesses): Push the closest parent back to work
24915 queue if add_access_to_work_queue returned true.
24917 2017-05-02 Richard Biener <rguenther@suse.de>
24919 * common.opt (fstrict-overflow): Alias negative to fwrapv.
24920 * doc/invoke.texi (fstrict-overflow): Remove all traces of
24921 -fstrict-overflow documentation.
24922 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
24923 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
24924 flag_strict_overflow.
24925 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
24926 * lto-opts.c (lto_write_options): Do not stream it.
24927 * lto-wrapper.c (merge_and_complain): Do not handle it.
24928 * opts.c (default_options_table): Do not set -fstrict-overflow.
24929 (finish_options): Likewise do not clear it when sanitizing.
24930 * simplify-rtx.c (simplify_const_relational_operation): Do not
24931 test flag_strict_overflow.
24933 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
24935 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
24936 using enabled attribute.
24937 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
24938 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
24939 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
24940 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
24941 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
24942 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
24943 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
24944 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
24945 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
24946 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
24948 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
24950 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
24952 2017-05-02 Richard Biener <rguenther@suse.de>
24954 PR tree-optimization/80591
24956 2017-04-10 Richard Biener <rguenther@suse.de>
24958 * tree-ssa-structalias.c (find_func_aliases): Properly handle
24961 2017-05-02 Richard Biener <rguenther@suse.de>
24963 PR tree-optimization/80549
24964 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
24965 (cleanup_tree_cfg_noloop): Create forwarders to known loop
24966 headers if they do not have a preheader.
24968 2017-05-02 Martin Liska <mliska@suse.cz>
24971 * common.opt: Fix typo.
24972 * doc/invoke.texi: Likewise.
24974 2017-05-01 Jan Beulich <jbeulich@suse.com>
24976 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
24977 swapping, add (x,x,m,x,n) alternative.
24979 2017-05-01 Nathan Sidwell <nathan@acm.org>
24981 * calls.c (combine_pending_stack_adjustment_and_call): Remove
24982 unnecessary unadjusted_alignment check.
24984 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
24987 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
24989 * gimplify.c (gimplify_cilk_detach): New function.
24990 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
24991 * tree-core.h: Document EXPR_CILK_SPAWN.
24992 * tree.h (EXPR_CILK_SPAWN): Define.
24994 2017-05-01 David Malcolm <dmalcolm@redhat.com>
24996 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
24997 to use new fixit_hint representation, using the "replace" logic.
24998 (get_line_span_for_fixit_hint): Likewise.
24999 (layout::print_any_fixits): Likewise.
25000 (selftest::test_one_liner_many_fixits): Rename to...
25001 (selftest::test_one_liner_many_fixits_1): ...this, and update
25002 comment and expected output to reflect that the multiple fix-it
25003 hints are now consolidated into one insertion.
25004 (selftest::test_one_liner_many_fixits_2): New test.
25005 (selftest::test_diagnostic_show_locus_one_liner): Update for
25007 (selftest::test_fixit_consolidation): Update for fix-it API
25009 * diagnostic.c (print_parseable_fixits): Likewise.
25010 * edit-context.c (edited_line::m_line_events): Convert from
25011 auto_vec <line_event *> to auto_vec <line_event>.
25012 (class line_event): Convert from abstract base class to a concrete
25013 class, taking over the role of replace_event.
25014 (class insert_event): Delete.
25015 (class replace_event): Rename to class line_event. Convert to
25017 (edit_context::add_fixits): Reimplement.
25018 (edit_context::apply_insert): Delete.
25019 (edit_context::apply_replace): Rename to...
25020 (edit_context::apply_fixit): ...this. Convert to half-open range.
25021 (edited_file::apply_insert): Delete.
25022 (edited_file::apply_replace): Rename to...
25023 (edited_file::apply_fixit): ...this.
25024 (edited_line::~edited_line): Drop deletion of events.
25025 (edited_line::apply_insert): Delete.
25026 (edited_line::apply_replace): Rename to...
25027 (edited_line::apply_fixit): ...this. Convert to half-open range.
25028 Update for change to type of m_line_events.
25029 * edit-context.h (edit_context::apply_insert): Delete.
25030 (edit_context::apply_replace): Rename to...
25031 (edit_context::apply_fixit): ...this.
25033 2017-05-01 Martin Sebor <msebor@redhat.com>
25035 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
25038 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
25041 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
25042 __get_cpuid_max returns 0.
25043 (__get_cpuid_count): Ditto.
25045 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
25047 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
25048 replacement expression is another instance of one of its arguments.
25050 2017-05-01 Jakub Jelinek <jakub@redhat.com>
25053 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
25054 check for stack push/pop autoinc.
25055 * config/i386/i386.c (ix86_agi_dependent): Return false
25056 if the only reason why modified_in_p returned true is that
25057 addr is SP based and set_insn is a push or pop.
25059 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
25061 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
25064 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
25067 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
25068 (account_size_time): Use two predicates - exec_pred and
25070 (evaluate_conditions_for_known_args): Compute both clause and
25072 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
25073 (inline_summary_t::duplicate): Update.
25074 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
25076 (compute_inline_parameters): Likewise.
25077 (estimate_edge_size_and_time): Update caluclation of time.
25078 (estimate_node_size_and_time): Compute both time and nonspecialized
25080 (estimate_ipcp_clone_size_and_time): Update.
25081 (inline_merge_summary): Update.
25082 (do_estimate_edge_time): Update.
25083 (do_estimate_edge_size): Update.
25084 (do_estimate_edge_hints): Update.
25085 (inline_read_section, inline_write_summary): Stream both new predicates.
25086 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
25088 (compute_inlined_call_time): Cleanup.
25089 (big_speedup_p): Update.
25090 (edge_badness): Update.
25091 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
25092 (size_time_entry): Replace predicate by exec_predicate and
25093 nonconst_predicate.
25094 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
25095 (estimate_edge_time): Return also nonspec_time.
25096 (reset_edge_growth_cache): Update.
25098 2017-04-29 Jakub Jelinek <jakub@redhat.com>
25100 PR rtl-optimization/80491
25101 * ifcvt.c (noce_process_if_block): When looking for x setter
25102 with missing else_bb, don't check only the insn right before
25103 cond_earliest, but look for the last insn that x is modified in
25104 within the same bb.
25106 PR rtl-optimization/80491
25107 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
25109 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
25111 PR tree-optimization/80487
25112 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
25114 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25116 PR tree-optimization/79697
25117 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
25118 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
25119 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
25121 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
25122 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
25124 2017-04-28 Martin Sebor <msebor@redhat.com>
25126 PR tree-optimization/80523
25127 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
25128 (init_target_to_host_charmap, target_to_host, target_strtol10): New
25130 (maybe_warn, format_directive, parse_directive): Use new functions.
25131 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
25133 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
25135 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
25137 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
25139 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
25140 target_header_dir): Set correctly.
25141 * configure: Regenerated.
25142 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
25143 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
25144 instead of SYSTEM_HEADER_DIR.
25146 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
25148 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
25149 (estimate_local_effects): Likewise.
25150 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
25151 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
25152 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
25153 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
25154 do_estimate_edge_time, estimate_edge_time): Likewise.
25155 * ipa-inline-analysis.c (estimate_node_size_and_time,
25156 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
25157 (estimate_time_after_inlining): Remove.
25159 2017-04-28 Martin Liska <mliska@suse.cz>
25161 * doc/gcov.texi: Enhance documentation of gcov.
25163 2017-04-28 Martin Liska <mliska@suse.cz>
25165 * doc/gcov.texi: Sort options in alphabetic order.
25166 * doc/gcov-dump.texi: Likewise.
25167 * doc/gcov-tool.texi: Likewise.
25168 * gcov.c (print_usage): Likewise.
25169 * gcov-dump.c (print_usage): Likewise.
25170 * gcov-tool.c (print_merge_usage_message): Likewise.
25171 (print_rewrite_usage_message): Likewise.
25172 (print_overlap_usage_message): Likewise.
25174 2017-04-28 Martin Liska <mliska@suse.cz>
25176 PR gcov-profile/53915
25177 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
25179 2017-04-28 Martin Liska <mliska@suse.cz>
25181 PR gcov-profile/79891
25182 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
25183 is marked by compiler as living on a line.
25184 (get_cycles_count): Remove usage of the union.
25185 (output_intermediate_file): Likewise.
25186 (find_source): Fix GNU coding style.
25187 (accumulate_line_counts): Remove old non-all block mode.
25188 (output_lines): Remove usage of the union.
25189 * profile.c (output_location): Include all BBs, even if
25190 belonging to a same line (and file) as a previous BB.
25192 2017-04-28 Martin Liska <mliska@suse.cz>
25194 * gcov.c (process_args): Handle new argument 'w'.
25195 (read_graph_file): Assign ID to BBs.
25196 (output_branch_count): Display BB # if verbose flag is set.
25197 (output_lines): Likewise for arcs.
25198 (print_usage): Add '--verbose' option help.
25199 * doc/gcov.texi: Document --verbose (-w) option.
25201 2017-04-28 Martin Liska <mliska@suse.cz>
25203 * gcov.c (struct block_location_info): New struct.
25204 (process_file): Fill up the new structure.
25205 (read_graph_file): Replace usage of encoding by the newly added
25207 (add_line_counts): Likewise.
25208 (accumulate_line_counts): Remove usage of the union.
25209 (function_info::function_info): New function.
25210 (function_info::~function_info): Likewise.
25211 (process_file): Call delete instead of release_function.
25212 (release_function): Release the function.
25213 (release_structures): Call delete instead of release_function.
25214 (solve_flow_graph): Replace usage of num_blocks.
25215 (find_exception_blocks): Likewise.
25216 (output_lines): Fix GNU coding style.
25218 2017-04-28 Martin Liska <mliska@suse.cz>
25221 * coverage.c (coverage_remove_note_file): New function.
25222 * coverage.h: Declare the function.
25223 * toplev.c (finalize): Clean if an error has been seen.
25225 2017-04-28 Martin Liska <mliska@suse.cz>
25227 PR gcov-profile/80031
25228 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
25229 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
25230 * gcov.c (read_graph_file): Read just number of blocks.
25231 * profile.c (branch_prob): Do not stream 0 flags per a basic
25234 2017-04-28 Martin Liska <mliska@suse.cz>
25236 * gcov-dump.c (tag_*): Add new argument to declarations.
25237 (dump_gcov_file): Likewise.
25238 (tag_blocks): Add and use new argument depth.
25239 (tag_arcs): Likewise.
25240 (tag_lines): Likewise.
25241 (tag_counters): Likewise.
25242 (tag_summary): Likewise.
25243 (dump_working_sets): Use depth to do a proper indentation.
25245 2017-04-28 Jakub Jelinek <jakub@redhat.com>
25248 * cgraph.h (symtab_node::debug_symtab): No longer inline.
25249 * symtab.c (symtab_node::debug_symtab): Move definition here.
25251 2017-04-28 Richard Biener <rguenther@suse.de>
25253 * lto-streamer.h (LTO_major_version): Bump to 7.
25255 2017-04-28 Richard Biener <rguenther@suse.de>
25257 * tree-vrp.c (assert_info): New struct.
25258 (add_assert_info): New helper.
25259 (register_edge_assert_for_2): Refactor to add asserts to a vector
25261 (register_edge_assert_for_1): Likewise.
25262 (register_edge_assert_for): Likewise.
25263 (finish_register_edge_assert_for): New helper actually registering
25264 asserts where live on edge.
25265 (find_conditional_asserts): Adjust.
25266 (find_switch_asserts): Likewise.
25267 (evrp_dom_walker::try_find_new_range): Generalize.
25268 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
25270 2017-04-27 Marek Polacek <polacek@redhat.com>
25273 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
25274 arg10 and arg11 to itype.
25276 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
25278 * doc/extend.texi (Object Size Checking): Improve grammar.
25280 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
25283 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
25284 that the logic for permitting reciprocal estimates matches that
25287 2017-04-27 Jakub Jelinek <jakub@redhat.com>
25290 * tree.c (type_cache_hasher::equal): Only compare
25291 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
25292 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
25293 non-aggregate element types.
25294 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
25295 about the flag on ARRAY_TYPEs in the comment, formatting fix.
25297 2017-04-27 Richard Biener <rguenther@suse.de>
25299 PR middle-end/80533
25300 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
25301 stripping ARRAY_REFs from MEM_EXPR make sure we're not
25302 keeping a reference to a trailing array.
25304 2017-04-27 Richard Biener <rguenther@suse.de>
25306 PR middle-end/80539
25307 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
25308 being in loop-closed SSA form conservatively.
25309 (chrec_fold_multiply_poly_poly): Likewise.
25311 2017-04-27 Tamar Christina <tamar.christina@arm.com>
25313 PR middle-end/79665
25314 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
25315 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
25317 2017-04-27 Jakub Jelinek <jakub@redhat.com>
25320 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
25321 (aarch64_function_arg_alignment): Return unsigned int again, but still
25322 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
25323 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
25324 Don't emit -Wpsabi note.
25325 (aarch64_function_arg_boundary): Likewise.
25326 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
25329 2017-04-26 Nathan Sidwell <nathan@acm.org>
25331 * tree.h (crc32_unsigned_n): Declare.
25332 (crc32_unsigned, crc32_unsigned): Make inline.
25333 * tree.c (crc32_unsigned_bits): Replace with ...
25334 (crc32_unsigned_n): ... this.
25335 (crc32_unsigned, crc32_byte): Remove.
25336 (crc32_string): Remove unnecessary braces.
25338 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
25340 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
25341 * ipa-inline-analysis.c (MAX_TIME): Remove.
25342 (account_size_time): Use sreal for time.
25343 (dump_inline_summary): Update.
25344 (estimate_function_body_sizes): Update.
25345 (estimate_edge_size_and_time): Update.
25346 (estimate_calls_size_and_time): Update.
25347 (estimate_node_size_and_time): Update.
25348 (inline_merge_summary): Update.
25349 (inline_update_overall_summary): Update.
25350 (estimate_time_after_inlining): Update.
25351 (inline_read_section): Update.
25352 (inline_write_summary): Update.
25353 * ipa-inline.c (compute_uninlined_call_time): Update.
25354 (compute_inlined_call_time): Update.
25355 (recursive_inlining): Update.
25356 (inline_small_functions): Update.
25357 (dump_overall_stats): Update.
25358 * ipa-inline.h: Include sreal.h.
25359 (size_time_entry): Turn time to sreal.
25360 (inline_summary): Turn self_time nad time to sreal.
25362 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
25364 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
25366 (sreal::stream_out, sreal::stream_in): New.
25367 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
25369 2017-04-25 Jakub Jelinek <jakub@redhat.com>
25371 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
25374 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
25377 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
25379 (dimode_scalar_chain::compute_convert_gain): Ditto.
25380 (dimode_scalar_chain::make_vector_copies): Ditto.
25381 (dimode_scalar_chain::convert_reg): Ditto.
25382 (dimode_scalar_chain::convert_insn): Ditto.
25383 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
25384 (VI248_AVX512BW_1): New mode iterator.
25385 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
25386 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
25389 2017-04-25 Martin Sebor <msebor@redhat.com>
25391 PR tree-optimization/80497
25392 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
25393 constants are representable in HOST_WIDE_INT.
25394 (parse_directive): Ditto.
25396 2017-04-25 Martin Sebor <msebor@redhat.com>
25399 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
25400 (new_zero_array): Adjust signature.
25401 (dom_info::dom_init): Used unsigned rather that size_t.
25402 (dom_info::dom_info): Same.
25404 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25405 Jakub Jelinek <jakub@redhat.com>
25408 * config/arm/arm.c: Include gimple.h.
25409 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
25410 returns negative, increment ncrn only if it returned positive.
25411 (arm_needs_doubleword_align): Return int instead of bool,
25412 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
25413 members, but if there is any such non-FIELD_DECL
25414 > PARM_BOUNDARY aligned decl, return -1 instead of false.
25415 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
25416 returns negative, increment nregs only if it returned positive.
25417 (arm_setup_incoming_varargs): Likewise.
25418 (arm_function_arg_boundary): Emit -Wpsabi note if
25419 arm_needs_doubleword_align returns negative, return
25420 DOUBLEWORD_ALIGNMENT only if it returned positive.
25422 2017-04-25 Marek Polacek <polacek@redhat.com>
25425 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
25426 first argument to type.
25428 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
25431 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
25432 type checks to test for compatibility instead of equality.
25434 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25435 Jakub Jelinek <jakub@redhat.com>
25438 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
25440 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
25441 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
25442 the alignment computation, but return their maximum in warn_alignment.
25443 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
25444 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
25446 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
25447 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
25450 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25452 * config/arc/simdext.md (dmpyh): Fix typo.
25454 2017-04-25 Richard Biener <rguenther@suse.de>
25456 PR tree-optimization/80492
25457 * alias.c (compare_base_decls): Handle registers with asm
25458 specification conservatively.
25459 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
25460 compare_base_decls returning dont-know properly.
25462 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25464 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
25465 (legitimate_offset_address_p): New function.
25466 (arc_legitimate_address_p): Use above function.
25468 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25470 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
25472 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25474 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
25475 ACCH registers whenever they are available.
25477 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25479 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
25480 double regs fix when not used.
25482 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25484 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
25486 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
25487 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
25489 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25491 * config/arc/arc.c (arc_output_addsi): Check for h-register class
25492 when emitting short ADD instructions.
25494 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25496 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
25498 (cmpsi_cc_c_insn): Likewise.
25499 (cbranchsi4_scratch): Compute proper instruction length using
25500 compact_hreg_operand.
25501 * config/arc/predicates.md (compact_hreg_operand): New predicate.
25503 2017-04-25 Richard Biener <rguenther@suse.de>
25505 PR middle-end/80509
25506 * passes.c (pass_manager::pass_manager): Initialize
25507 m_name_to_pass_map.
25509 2017-04-25 Richard Biener <rguenther@suse.de>
25511 PR tree-optimization/79201
25512 * tree-ssa-sink.c (statement_sink_location): Handle calls.
25514 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25517 * config/s390/vector.md: Split MEM->GPR vector moves for
25518 non-s_operand addresses.
25520 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25523 * config/s390/predicates.md (reload_const_wide_int_operand): New
25525 * config/s390/s390.md ("movti"): Remove d/P alternative.
25526 ("movti_bigconst"): New pattern definition.
25528 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25531 * s390-protos.h (s390_expand_cs_hqi): Removed.
25532 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
25533 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
25534 modes as well as CCZ1mode and CCZmode.
25535 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
25536 signature of s390_emit_compare_and_swap.
25537 (s390_expand_cs_hqi): Likewise, make static.
25538 (s390_expand_cs_tdsi): Generate an explicit compare before trying
25539 compare-and-swap, in some cases.
25540 (s390_expand_cs): Wrapper function.
25541 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
25543 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
25544 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
25545 patterns for small and large integers. Forbid symref memory operands.
25546 Move expander to s390.c. Require cc register.
25547 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
25548 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
25549 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
25550 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
25551 symref memory operands. Remove CC mode and call s390_match_ccmode
25553 ("atomic_exchange<mode>"): Allow and implement all integer modes.
25555 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25557 * config/s390/s390.md (define_peephole2): New peephole to help
25558 combining the load-and-test pattern with volatile memory.
25560 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25562 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
25563 with CCZmode for TARGET_Z196.
25565 2017-04-25 Jakub Jelinek <jakub@redhat.com>
25567 PR rtl-optimization/80501
25568 * combine.c (make_compound_operation_int): Set subreg_code to SET
25569 even for AND with mask of the sign bit of mode.
25571 PR rtl-optimization/80500
25572 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
25573 sum's initial value.
25575 2017-04-25 Julian Brown <julian@codesourcery.com>
25576 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25578 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
25580 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
25582 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
25584 2017-04-25 Julian Brown <julian@codesourcery.com>
25585 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25587 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
25588 (thunderx2t99_sha): New Reservation.
25590 2017-04-25 Julian Brown <julian@codesourcery.com>
25591 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25593 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
25594 type for 1-element load.
25596 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
25598 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
25600 2017-04-24 Martin Jambor <mjambor@suse.cz>
25602 PR tree-optimization/80293
25603 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
25604 char arrays not totally scalarizable if it is false.
25605 (analyze_all_variable_accesses): Pass correct value in the new
25606 parameter. Add a statistics counter.
25608 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
25610 PR middle-end/79931
25611 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
25613 2017-04-24 Richard Biener <rguenther@suse.de>
25615 PR tree-optimization/80494
25616 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
25617 out for complex types.
25619 2017-04-24 Richard Biener <rguenther@suse.de>
25621 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
25622 * tree-ssa-sccvn.c (print_scc): Print SCC size.
25623 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
25624 (DFS): Adjust and never fail.
25625 (sccvn_dom_walker::fail): Remove.
25626 (sccvn_dom_walker::before_dom_children): Adjust.
25627 (run_scc_vn): Likewise and never fail.
25628 * tree-ssa-pre.c (pass_pre::execute): Adjust.
25629 (pass_fre::execute): Likewise.
25631 2017-04-24 Richard Biener <rguenther@suse.de>
25633 PR tree-optimization/79725
25634 * tree-ssa-sink.c (statement_sink_location): Return whether
25635 failure reason was zero uses. Move that check later.
25636 (sink_code_in_bb): Deal with zero uses by removing the stmt
25639 2017-04-24 Richard Biener <rguenther@suse.de>
25642 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
25643 pointer-based references.
25645 2017-04-24 Richard Biener <rguenther@suse.de>
25648 * pass_manager.h (pass_manager::operator new): Remove.
25649 (pass_manager::operator delete): Likewise.
25650 * passes.c (pass_manager::operator new): Remove.
25651 (pass_manager::operator delete): Likewise.
25652 (pass_manager::pass_manager): Zero individual pass members.
25654 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
25657 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
25658 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
25659 Check "XEXP (src, 1)" operand here.
25660 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
25661 Check "XEXP (src, 1)" operand here.
25662 (dimode_scalar_chain::make_vector_copies): Detect count register
25663 of a shift instruction. Zero extend count register from QImode
25664 to DImode to satisfy vector shift pattern count operand predicate.
25665 Substitute vector shift count operand with a DImode copy.
25666 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
25669 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
25671 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
25672 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
25673 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
25674 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
25675 (UNSPEC_NOREX_MEM): Remove definition.
25677 2017-04-21 Richard Biener <rguenther@suse.de>
25679 PR tree-optimization/79547
25680 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
25681 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
25682 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
25683 without any constraints.
25685 2017-04-21 Richard Biener <rguenther@suse.de>
25687 PR tree-optimization/78847
25688 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
25690 2017-04-21 Richard Biener <rguenther@suse.de>
25692 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
25693 (build_distinct_type_copy): Likewise.
25694 (build_variant_type_copy): Likewise.
25695 * tree.c (build_qualified_type): Pass down mem-stat info.
25696 (build_distinct_type_copy): Likewise.
25697 (build_variant_type_copy): Likewise.
25699 2017-04-21 Richard Biener <rguenther@suse.de>
25701 PR tree-optimization/80237
25702 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
25704 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
25705 for a simplified result.
25707 2016-04-21 Richard Biener <rguenther@suse.de>
25709 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
25710 sth as strict as a simple_iv but a chrec without symbols and an
25711 operand defined in the loop we are peeling (and not some subloop).
25712 (propagate_constants_for_unrolling): Propagate all constants.
25714 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
25717 * config/i386/i386.c (print_reg): Remove assert for disalowed
25718 regno values, call output_operand_lossage instead.
25720 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
25723 * config/i386/constraints.md (Yc): New register constraint.
25724 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
25725 Use Yc constraint for alternative 2 of operand 0. Remove
25726 preferred_for_speed attribute.
25728 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
25730 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
25731 lastprivate clauses in SIMT case.
25733 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
25735 * doc/invoke.texi (-Wextra-semi): Document new warning option.
25737 2017-04-20 Richard Biener <rguenther@suse.de>
25739 PR tree-optimization/57796
25740 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
25741 as N scalar stores.
25742 (vect_model_load_cost): Cost gathers as N scalar loads.
25744 2017-04-20 Richard Biener <rguenther@suse.de>
25746 * ggc-page.c (ggc_allocated_p): Rename to ...
25747 (safe_lookup_page_table_entry): ... this and return the lookup
25749 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
25751 2017-04-20 Richard Biener <rguenther@suse.de>
25753 PR tree-optimization/80453
25754 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
25755 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
25756 from the conditions.
25757 (vn_phi_eq): Pass them down.
25758 (vn_phi_lookup): Record them.
25759 (vn_phi_insert): Likewise.
25761 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
25763 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
25764 uninitialized variable warning to avoid buffer overrun.
25766 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
25769 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
25770 is suppressed for '{ 0 }' in C.
25772 2017-04-20 Jakub Jelinek <jakub@redhat.com>
25774 * BASE-VER: Set to 8.0.0.
25776 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
25778 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
25779 priority .init_array and .fini_array section with SECTION_NOTYPE
25782 2017-04-20 Jakub Jelinek <jakub@redhat.com>
25784 PR middle-end/80423
25785 * tree.h (build_array_type): Add typeless_storage default argument.
25786 * tree.c (type_cache_hasher::equal): Also compare
25787 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
25788 (build_array_type): Add typeless_storage argument, set
25789 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
25791 (build_nonshared_array_type): Adjust build_array_type_1 caller.
25792 (build_array_type): Likewise. Add typeless_storage argument.
25794 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
25795 Jakub Jelinek <jakub@redhat.com>
25797 PR tree-optimization/80426
25798 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
25799 operation on symbolic operands, also compute the overflow for the
25800 invariant part when the operation degenerates into a negation.
25802 2017-04-19 Jakub Jelinek <jakub@redhat.com>
25805 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
25806 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
25809 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
25811 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
25814 * config/avr/avr.c (tree.h): Include it.
25815 (cgraph.h): Include it.
25816 (avr_encode_section_info): Don't warn for uninitialized progmem
25817 variable if it's just an alias.
25819 2017-04-19 Richard Biener <rguenther@suse.de>
25822 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
25823 when needed by AutoPGO.
25825 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
25828 * doc/lto.texi: Remove an extra 'that'.
25830 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
25832 PR rtl-optimization/80429
25833 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
25834 are only used in debug insns.
25836 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
25837 Vladimir Makarov <vmakarov@redhat.com>
25839 * config/sparc/predicates.md (input_operand): Add comment. Return
25840 true for any memory operand when LRA is in progress.
25841 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
25843 2017-04-18 Jeff Law <law@redhat.com>
25846 * mips.md ({return,simple_return}_internal): Do not overwrite
25849 2017-04-18 Jakub Jelinek <jakub@redhat.com>
25851 PR tree-optimization/80443
25852 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
25853 instead of adding 1, subtract -1 and similarly instead of subtracting
25856 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
25858 PR rtl-optimization/80357
25859 * haifa-sched.c (tmp_bitmap): New variable.
25860 (model_recompute): Handle duplicate use records.
25861 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
25862 (free_global_sched_pressure_data): Free it.
25864 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
25867 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
25868 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
25869 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
25870 instead of SYSTEM_HEADER_DIR.
25872 2017-04-18 Jeff Law <law@redhat.com>
25874 PR middle-end/80422
25875 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
25876 predecessors after walking up the insn chain.
25878 2017-04-18 Jakub Jelinek <jakub@redhat.com>
25881 * dwarf2out.c (modified_type_die): Try harder not to emit internal
25882 sizetype type into debug info.
25884 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
25887 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
25888 unneeded test for TARGET_UPPER_REGS_SF.
25889 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
25891 2017-04-18 Jakub Jelinek <jakub@redhat.com>
25894 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
25895 instead of gsi_after_labels.
25897 2017-04-18 Jeff Law <law@redhat.com>
25899 * regcprop.c (maybe_mode_change): Avoid creating copies of the
25903 2017-04-13 Jeff Law <law@redhat.com>
25904 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
25905 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
25907 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
25910 * config/avr/avr.c (intl.h): Include it.
25911 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
25913 2017-04-18 Martin Liska <mliska@suse.cz>
25915 PR gcov-profile/78783
25916 * gcov-tool.c (gcov_output_files): Validate that destination
25917 file is either removed by the tool or by a user.
25919 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
25920 Guy Benyei <guybe@mellanox.com>
25922 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
25923 block, and do not negate it, the stored id is already negative.
25925 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
25927 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
25929 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
25932 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
25933 masks of options that should be turned off if the VSX vector
25934 options are turned off.
25935 (OTHER_P8_VECTOR_MASKS): Likewise.
25936 (OTHER_VSX_VECTOR_MASKS): Likewise.
25937 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
25938 rs6000_disable_incompatible_switches to validate no type switches
25940 (rs6000_incompatible_switch): New function to disallow turning on
25941 other vector options if -mno-vsx, -mno-power8-vector, or
25942 -mno-power9-vector are specified.
25944 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
25946 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
25948 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
25950 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
25951 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
25952 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
25953 (ARG_POINTER_CFA_OFFSET): Likewise.
25955 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
25957 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
25958 conditions to take advantage of various optimizations.
25960 2017-04-13 Jeff Law <law@redhat.com>
25962 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
25963 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
25964 (zero_extendsidi2_dext): Likewise.
25966 2017-04-13 Jakub Jelinek <jakub@redhat.com>
25969 * fold-const.c (fold_ternary_loc): Revert
25970 use op0 instead of fold_convert_loc (loc, type, arg0) part of
25973 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
25975 PR rtl-optimization/80343
25976 * lra-remat.c (update_scratch_ops): Assign original hard reg to
25977 new scratch pseudo.
25979 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
25982 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
25983 to ubsan_encode_value.
25985 2017-04-13 Jeff Law <law@redhat.com>
25987 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
25988 appearing in DEBUG_INSNs.
25990 2017-04-13 Martin Liska <mliska@suse.cz>
25992 PR gcov-profile/80413
25993 * gcov-io.c (gcov_write_string): Copy to buffer just when
25994 allocated size is greater than zero.
25996 2017-04-13 Jakub Jelinek <jakub@redhat.com>
25999 * dwarf2out.c (decls_for_scope): Ignore declarations of
26000 current_function_decl in BLOCK_NONLOCALIZED_VARS.
26002 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
26005 * ipa-visibility.c (non_local_p): Fix typos.
26006 (localize_node): When localizing symbol in same comdat group,
26007 dissolve the group only when we know external symbols are going
26009 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
26011 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26013 PR tree-optimization/79390
26014 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
26015 order does not result in usable sequence, retry with reversed operand
26021 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
26022 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
26023 op0 instead of fold_convert_loc (loc, type, arg0).
26025 2017-04-12 Jeff Law <law@redhat.com>
26027 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
26028 has a delay slot in the generated code.
26030 * config/cris/cris.md (cris_preferred_reload_class): Return
26031 GENNONACR_REGS rather than GENERAL_REGS.
26033 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26036 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
26037 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
26038 signedness of the result type.
26040 2017-04-12 Richard Biener <rguenther@suse.de>
26041 Jeff Law <law@redhat.com>
26043 PR tree-optimization/80359
26044 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
26045 trim stores to TARGET_MEM_REFs.
26047 2017-04-12 Richard Biener <rguenther@suse.de>
26049 PR tree-optimization/79390
26050 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
26051 threading case even more.
26053 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
26056 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
26057 for quad_address_p for TImode, instead of just not indexed_address.
26059 2017-04-12 Richard Biener <rguenther@suse.de>
26060 Bernd Edlinger <bernd.edlinger@hotmail.de>
26062 PR middle-end/79671
26063 * alias.c (component_uses_parent_alias_set_from): Handle
26064 TYPE_TYPELESS_STORAGE.
26065 (get_alias_set): Likewise.
26066 * tree-core.h (tree_type_common): Add typeless_storage flag.
26067 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
26068 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
26069 for types containing members with TYPE_TYPELESS_STORAGE.
26070 (place_field): Likewise.
26071 (layout_type): Likewise for ARRAY_TYPE.
26072 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
26073 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
26074 TYPE_TYPELESS_STORAGE.
26075 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
26077 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26080 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
26081 first argument to type.
26083 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26087 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
26088 CONST0_RTX (mode) rather than const0_rtx where appropriate.
26089 (rs6000_expand_binop_builtin): Likewise.
26090 (rs6000_expand_ternop_builtin): Likewise; also add missing
26091 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
26092 vshasigma built-ins.
26093 * doc/extend.texi: Document that vec_xxpermdi's third argument
26094 must be a constant.
26096 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
26098 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
26099 Use shift_const cost parameter when calculating gain of STV shifts.
26101 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
26103 PR rtl-optimization/70478
26104 * lra-constraints.c (process_alt_operands): Check memory for
26105 disfavoring memory insn operand.
26107 2017-04-11 Jakub Jelinek <jakub@redhat.com>
26109 PR middle-end/80100
26110 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
26111 left shift in unsigned HOST_WIDE_INT type.
26113 PR rtl-optimization/80385
26114 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
26115 (not (neg X)) into (plus X -1) for complex or non-integral modes.
26118 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
26119 if they have any depend clauses.
26121 2017-04-11 Martin Liska <mliska@suse.cz>
26124 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
26125 * ipa-split.c (split_function): Create a local comdat symbol
26126 if caller is in a comdat group.
26128 2017-04-11 Martin Liska <mliska@suse.cz>
26131 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
26134 2017-04-11 Martin Sebor <msebor@redhat.com>
26136 PR middle-end/80364
26137 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
26138 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
26140 (directive::set_width, directive::set_precision, format_character):
26142 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
26145 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
26148 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
26149 conflict, set target->arch_name instead of target->cpu_name.
26151 2017-04-11 Richard Biener <rguenther@suse.de>
26153 PR tree-optimization/80374
26154 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
26155 build_zero_cst, remove fold_convertible_p check again.
26157 2017-04-11 Martin Liska <mliska@suse.cz>
26160 * ubsan.c (instrument_object_size): Do not instrument register
26163 2017-04-11 Jakub Jelinek <jakub@redhat.com>
26166 * config/i386/i386-builtin-types.def
26167 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
26168 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
26169 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
26170 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
26171 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
26172 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
26173 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
26174 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
26175 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
26176 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
26177 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
26178 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
26179 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
26180 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
26181 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
26182 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
26183 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
26184 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
26185 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
26186 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
26187 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
26188 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
26189 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
26190 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
26191 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
26192 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
26193 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
26194 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
26195 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
26196 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
26197 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
26198 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
26199 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
26200 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
26201 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
26202 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
26203 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
26204 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
26205 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
26206 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
26207 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
26208 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
26209 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
26210 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
26211 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
26212 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
26214 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
26215 flag to second_arg_count, handle 4 argument function type _COUNT
26216 aliases, handle second_arg_count on second argument rather than last.
26218 2017-04-10 Jeff Law <law@redhat.com>
26220 PR tree-optimization/80374
26221 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
26222 record anything if we can not convert integer_zero_node to the
26225 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
26228 * config/rs6000/rs6000.c (rs6000_option_override_internal):
26229 Enhance special handling given to the TARGET_P9_MINMAX option in
26230 relation to certain other options.
26232 2017-04-10 Bin Cheng <bin.cheng@arm.com>
26234 PR tree-optimization/80153
26235 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
26236 remove POINTER_PLUS_EXPR's base part directly, rather than through
26239 2017-04-10 Richard Biener <rguenther@suse.de>
26240 Bin Cheng <bin.cheng@arm.com>
26242 PR tree-optimization/80153
26243 * tree-affine.c (aff_combination_to_tree): Get base pointer from
26244 the first element of pointer type aff_tree. Build result expr in
26246 (add_elt_to_tree): Convert to type unconditionally. Remove other
26247 fold_convert calls.
26248 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
26249 (rewrite_use_nonlinear_expr): Check invariant using iv information.
26251 2017-04-10 Richard Biener <rguenther@suse.de>
26253 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26256 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
26258 PR rtl-optimization/70478
26259 * lra-constraints.c (curr_small_class_check): New.
26260 (update_and_check_small_class_inputs): New.
26261 (process_alt_operands): Update curr_small_class_check. Disfavor
26262 alternative insn memory operands. Check available regs for small
26265 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
26268 * config/mips/mips.opt (-mvirt): Update description.
26269 * doc/invoke.texi (-mvirt): Likewise.
26271 2017-04-10 Richard Biener <rguenther@suse.de>
26273 PR middle-end/80362
26274 * fold-const.c (fold_binary_loc): Look at unstripped ops when
26275 looking for NEGATE_EXPR in -A / -B to A / B folding.
26277 2017-04-10 Martin Liska <mliska@suse.cz>
26279 PR gcov-profile/80224
26280 * gcov.c (print_usage): Fix usage string.
26281 (get_gcov_intermediate_filename): Remove.
26282 (output_gcov_file): Use both for normal and intermediate format.
26283 (generate_results): Do not initialize special file for
26284 intermediate format.
26286 2017-04-10 Richard Biener <rguenther@suse.de>
26288 PR tree-optimization/80304
26289 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
26292 2017-04-10 Nathan Sidwell <nathan@acm.org>
26295 * config/rs6000/rs6000.c (rs6000_vector_type): New.
26296 (rs6000_init_builtins): Use it.
26298 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26300 * config/arm/arm.md (<mrc>): Add mode to SET source.
26301 (<mrrc>): Likewise.
26303 2017-04-10 Richard Biener <rguenther@suse.de>
26305 PR middle-end/80344
26306 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
26308 2017-04-10 Jakub Jelinek <jakub@redhat.com>
26311 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
26312 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
26313 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
26314 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
26315 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
26316 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
26317 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
26318 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
26319 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
26320 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
26321 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
26322 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
26323 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
26324 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
26325 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
26326 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
26327 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
26328 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
26329 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
26330 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
26331 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
26332 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
26333 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
26335 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
26337 PR rtl-optimization/70478
26338 * lra-constraints.c: Reverse the last patch.
26340 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
26342 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
26343 Add comment for WCHAR_T.
26345 2017-04-08 Martin Liska <mliska@suse.cz>
26348 2017-04-07 Martin Liska <mliska@suse.cz>
26351 * ipa-split.c (split_function): Add function part to a same comdat
26354 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26357 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
26359 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
26361 * rs6000/rs6000.c (vec_load_pendulum): Rename...
26362 (vec_pairing): ...to this.
26363 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
26364 (rs6000_sched_init): Adjust for name change.
26365 (struct rs6000_sched_context): Likewise.
26366 (rs6000_init_sched_context): Likewise.
26367 (rs6000_set_sched_context): Likewise.
26369 2017-04-07 Jakub Jelinek <jakub@redhat.com>
26375 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
26377 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
26378 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
26379 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
26381 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
26383 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
26385 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
26387 PR rtl-optimization/70703
26388 * ira-color.c (update_conflict_hard_regno_costs): Use
26389 int64_t instead of HOST_WIDE_INT.
26391 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
26393 PR rtl-optimization/70478
26394 * lra-constraints.c (process_alt_operands): Disfavor alternative
26395 insn memory operands.
26397 2017-04-07 Jeff Law <law@redhat.com>
26399 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
26400 CALL and NOTE_INSN_CALL_ARG_LOCATION.
26402 2017-04-07 Martin Liska <mliska@suse.cz>
26405 * config/aarch64/aarch64.c (aarch64_process_target_attr):
26406 Show error message instead of an ICE.
26408 2017-04-07 Martin Liska <mliska@suse.cz>
26411 * ipa-split.c (split_function): Add function part to a same comdat
26414 2017-04-07 Richard Biener <rguenther@suse.de>
26416 PR middle-end/80341
26417 * tree.c (get_unwidened): Also handle ! for_type case for
26419 * convert.c (do_narrow): Split out from ...
26420 (convert_to_integer_1): ... here. Do not pass final truncation
26421 type to get_unwidened for TRUNC_DIV_EXPR.
26423 2017-04-07 Richard Biener <rguenther@suse.de>
26425 * tree-affine.c (wide_int_ext_for_comb): Take type rather
26427 (aff_combination_const): Adjust.
26428 (aff_combination_scale): Likewise.
26429 (aff_combination_add_elt): Likewise.
26430 (aff_combination_add_cst): Likewise.
26431 (aff_combination_convert): Likewise.
26432 (add_elt_to_tree): Likewise. Remove unused argument.
26433 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
26435 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
26437 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
26439 * config/arm/arm.c (arm_default_short_enums): Use
26440 ARM_DEFAULT_SHORT_ENUMS.
26441 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
26443 2017-04-06 Jakub Jelinek <jakub@redhat.com>
26446 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
26447 members with redundant out-of-class redeclaration.
26449 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
26452 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
26453 * config/i386/i386.md (*zero_extendsidi2):
26454 Add (?*x,*x) and (?*v,*v) alternatives.
26456 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
26459 * config/i386/i386.c (ix86_expand_builtin)
26460 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
26461 mode from insn data. Convert operands to insn operand mode.
26462 Copy operands that don't satisfy insn predicate to a register.
26464 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
26466 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
26469 2017-04-06 Richard Biener <rguenther@suse.de>
26471 PR tree-optimization/80334
26472 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
26473 preserve alignment of accesses.
26475 2017-04-06 Richard Biener <rguenther@suse.de>
26477 PR tree-optimization/80262
26478 * tree-sra.c (build_ref_for_offset): Preserve address-space
26480 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
26481 Drop useless address-space information on MEM_REF offsets.
26483 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
26485 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
26487 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
26489 PR rtl-optimization/70703
26490 * ira-color.c (update_conflict_hard_regno_costs): Use
26491 HOST_WIDE_INT instead of long.
26493 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
26496 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
26497 not defined for x86_64 target. Add -mmmx target option when __SSE2__
26499 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
26500 for x86_64 target. Handle -m3dnowa option.
26502 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
26504 PR rtl-optimization/70703
26505 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
26506 (update_conflict_hard_regno_costs): Use long instead of unsigned
26507 arithmetic for cost calculation.
26509 2017-04-05 Jakub Jelinek <jakub@redhat.com>
26510 Bernd Edlinger <bernd.edlinger@hotmail.de>
26513 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
26516 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
26519 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
26520 ptr_mode with Pmode throughout.
26521 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
26522 into probe_stack_range and use DImode.
26524 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
26527 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
26528 call_eh_return is true.
26530 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26532 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
26533 Initialize last_match_fntype_index.
26535 2017-04-05 Jakub Jelinek <jakub@redhat.com>
26538 * tree-nvr.c: Include internal-fn.h.
26539 (pass_return_slot::execute): Ignore internal calls without
26542 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26543 Richard Biener <rguenther@suse.de>
26546 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
26547 captures used multiple times, except for the last use.
26548 * generic-match-head.c: Include gimplify.h.
26550 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26552 PR tree-optimization/79390
26553 * target.h (struct noce_if_info): Declare.
26554 * targhooks.h (default_noce_conversion_profitable_p): Declare.
26555 * target.def (noce_conversion_profitable_p): New target hook.
26556 * ifcvt.h (struct noce_if_info): New type, moved from ...
26557 * ifcvt.c (struct noce_if_info): ... here.
26558 (noce_conversion_profitable_p): Renamed to ...
26559 (default_noce_conversion_profitable_p): ... this. No longer
26561 (noce_try_store_flag_constants, noce_try_addcc,
26562 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
26563 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
26564 instead of noce_conversion_profitable_p.
26565 * config/i386/i386.c: Include ifcvt.h.
26566 (ix86_option_override_internal): Don't override
26567 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
26568 (ix86_noce_conversion_profitable_p): New function.
26569 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
26570 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
26571 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
26572 * doc/tm.texi: Regenerated.
26574 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26576 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
26579 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
26582 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
26583 instructions for small multiply cores.
26585 2017-04-04 Jeff Law <law@redhat.com>
26587 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
26589 (mips_expand_vec_perm_const): Initialize elements in orig_perm
26590 that are not set by the loop over the elements.
26592 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26595 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
26596 int mode, convert_modes it to mode as unsigned, otherwise use
26597 lowpart_subreg to mode rather than SImode.
26598 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
26599 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
26600 Use DImode instead of SImode for the shift count operand.
26601 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
26604 2017-04-04 Richard Biener <rguenther@suse.de>
26606 PR middle-end/80281
26607 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
26608 arithmetic done for the negate or the plus. Simplify.
26609 (A - (-B) -> A + B): Likewise.
26610 * fold-const.c (split_tree): Make sure to not negate pointers.
26612 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
26614 PR rtl-optimization/60818
26615 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
26616 a compare of comparisons with the thing compared if this results
26617 in a different machine mode.
26619 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
26621 * alias.c (base_alias_check): Fix typo in comment.
26622 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
26623 * cgraphunit.c (symbol_table::compile): Likewise.
26624 * collect2.c (maybe_run_lto_and_relink): Likewise.
26625 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
26626 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
26627 * config/avr/avr.c (avr_map_op_t): Likewise.
26628 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
26629 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
26630 * config/epiphany/epiphany.md (movcc): Likewise.
26631 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
26632 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
26634 * config/mips/mips.c (mips_save_restore_reg): Likewise.
26635 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
26636 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
26637 * config/sh/sh.c (sh_rtx_costs): Likewise.
26638 * fold-const.c (fold_truth_andor): Likewise.
26639 * genautomata.c (collapse_flag): Likewise.
26640 * gengtype.h (struct type::u::s): Likewise.
26641 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
26642 * input.c (FORMAT_AMOUNT): Likewise.
26643 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
26644 (known_aggs_to_agg_replacement_list): Likewise.
26645 * ipa-inline-analysis.c: Likewise.
26646 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
26647 * ipa-polymorphic-call.c
26648 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
26649 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
26650 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
26652 * modulo-sched.c (apply_reg_moves): Likewise.
26653 * omp-expand.c (build_omp_regions_1): Likewise.
26654 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
26655 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
26656 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
26657 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
26658 * value-prof.c: Likewise.
26659 * var-tracking.c (val_reset): Likewise.
26661 2017-04-03 Richard Biener <rguenther@suse.de>
26663 PR tree-optimization/80275
26664 * fold-const.c (split_address_to_core_and_offset): Handle
26667 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
26669 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
26670 descriptors is at least equal to that of functions.
26672 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
26674 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
26676 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
26679 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
26680 (mov<IMOD4:mode>): New expander.
26681 (*mov<IMOD4:mode>_internal): New insn and split pattern.
26683 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
26685 PR rtl-optimization/79405
26686 * fwprop.c (propagations_left): New variable.
26687 (forward_propagate_into): Decrement it.
26688 (fwprop_init): Initialize it.
26689 (fw_prop): If the variable has reached zero, stop propagating.
26690 (fwprop_addr): Ditto.
26692 2017-03-31 Jakub Jelinek <jakub@redhat.com>
26695 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
26696 a FUNCTION_DECL, pass it as decl instead of origin to
26699 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
26701 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
26704 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
26707 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
26708 TARGET_VSX_SMALL_INTEGER.
26710 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26712 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
26713 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
26715 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
26717 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
26718 extraction from odd-numbered MSA register.
26720 2017-03-31 Jakub Jelinek <jakub@redhat.com>
26722 PR middle-end/80173
26723 * expmed.c (store_bit_field_1): Don't attempt to create
26724 a word subreg out of hard registers wider than word if they
26725 have HARD_REGNO_NREGS of 1 for their mode.
26727 PR middle-end/80163
26728 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
26729 conversions to integer types wider than word and pointer.
26732 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
26733 (rtx_equal_for_cselib_p): Pass 0 to it.
26734 * cselib.c (cselib_hasher::equal): Likewise.
26735 (rtx_equal_for_cselib_1): Add depth argument. If depth
26736 is 128, don't look up VALUE locs and punt. Increment
26737 depth in recursive calls when walking VALUE locs.
26739 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
26741 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
26742 (make_gcov_file_name): Use the canonical path name for generating
26744 (read_line): Fix handling of files with ascii null bytes.
26746 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
26748 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
26749 to initialise a vector register instead
26750 of using a const_int.
26752 2017-03-30 Jakub Jelinek <jakub@redhat.com>
26754 PR translation/80189
26755 * gimplify.c (omp_default_clause): Use %qs instead of %s in
26756 diagnostic messages.
26758 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
26761 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
26762 (dfp_diex_<mode>): Update mode of operand 1.
26763 * doc/extend.texi (dxex, dxexq): Document change to return type.
26764 (diex, diexq): Document change to argument type.
26766 2017-03-30 Martin Jambor <mjambor@suse.cz>
26769 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
26770 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
26771 it reflects the signature changes performed at the callee side.
26772 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
26773 to cgraph_build_function_type_skip_args.
26774 (build_function_decl_skip_args): Adjust call to the above function.
26776 2017-03-30 Jakub Jelinek <jakub@redhat.com>
26779 * config/i386/sse.md
26780 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
26781 register as dest whenever it is a MEM not rtx_equal_p to the
26782 corresponding dup operand, and when forcing into reg move the
26783 reg into the memory afterwards.
26784 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
26785 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
26786 for the force_reg mode.
26787 (avx512vl_vextractf128<mode>): Use register as dest either
26788 always when a MEM, or when it is a MEM not rtx_equal_p to the
26789 corresponding dup operand, or even not when it is a CONST_VECTOR
26790 depending on the mode and lo vs. hi.
26791 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
26793 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
26794 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
26795 Likewise. Require that operands[2] is even.
26796 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
26797 Remove extraneous parens. Require that operands[2] is a multiple
26799 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
26800 operands[0] is a MEM if <mask_applied>, the predicates/constraints
26801 disallow memory then.
26803 2017-03-30 Richard Biener <rguenther@suse.de>
26805 PR tree-optimization/77498
26806 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
26807 to non-constants over backedges.
26809 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
26811 PR rtl-optimization/80233
26812 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
26813 as last_combined_insn. Do not test for BARRIER_P separately.
26815 2017-03-29 Andreas Schwab <schwab@suse.de>
26818 * calls.c (prepare_call_address): Convert funexp to Pmode before
26819 copying to temp reg.
26821 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26823 PR tree-optimization/80158
26824 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
26825 Handle possible future case of more than one alternate
26827 (replace_rhs_if_not_dup): Likewise.
26828 (replace_one_candidate): Likewise.
26830 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
26832 PR rtl-optimization/80193
26833 * ira.c (ira): Do not check allocation for LRA.
26835 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
26837 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
26838 (nvptx_output_simt_exit): Declare.
26839 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
26840 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
26841 (init_softstack_frame): Move initialization of crtl->is_leaf to...
26842 (nvptx_declare_function_name): ...here. Emit declaration of local
26843 memory space buffer for omp_simt_enter insn.
26844 (nvptx_output_unisimt_switch): New.
26845 (nvptx_output_softstack_switch): New.
26846 (nvptx_output_simt_enter): New.
26847 (nvptx_output_simt_exit): New.
26848 * config/nvptx/nvptx.h (struct machine_function): New fields
26849 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
26850 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
26851 (UNSPECV_SIMT_EXIT): Ditto.
26852 (omp_simt_enter_insn): New insn.
26853 (omp_simt_enter): New expansion.
26854 (omp_simt_exit): New insn.
26855 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
26857 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
26858 (expand_GOMP_SIMT_ENTER_ALLOC): New.
26859 (expand_GOMP_SIMT_EXIT): New.
26860 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
26861 (GOMP_SIMT_ENTER_ALLOC): Ditto.
26862 (GOMP_SIMT_EXIT): Ditto.
26863 * target-insns.def (omp_simt_enter): New insn.
26864 (omp_simt_exit): Ditto.
26865 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
26867 (lower_rec_simd_input_clauses): Implement SIMT privatization.
26868 (lower_rec_input_clauses): Likewise.
26869 (lower_lastprivate_clauses): Handle SIMT privatization.
26871 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
26872 (ompdevlow_adjust_simt_enter): New.
26873 (find_simtpriv_var_op): New.
26874 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
26875 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
26877 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
26878 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
26879 (copy_decl_for_dup_finish): Ditto.
26881 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
26883 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
26886 * config/i386/i386.c (ix86_option_override_internal): Always
26887 allow -mpreferred-stack-boundary=3 for 64-bit targets.
26889 2017-03-28 Bin Cheng <bin.cheng@arm.com>
26891 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
26893 2017-03-28 Bin Cheng <bin.cheng@arm.com>
26895 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
26896 mark new edge's irreducible flag accordign to it.
26897 (vect_do_peeling): Check loop preheader edge's irreducible flag
26898 and pass it to function slpeel_add_loop_guard.
26900 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
26902 PR tree-optimization/80218
26903 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
26904 Update block frequencies and counts.
26906 2017-03-28 Richard Biener <rguenther@suse.de>
26908 PR tree-optimization/78644
26909 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
26910 of a simplification result we may not use it at all.
26912 2017-03-28 Richard Biener <rguenther@suse.de>
26915 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
26916 without arguments, generate default definition of a SSA name.
26918 2017-03-28 Richard Biener <rguenther@suse.de>
26920 PR middle-end/80222
26921 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
26922 TYPE_REF_CAN_ALIAS_ALL references.
26923 * fold-const.c (fold_indirect_ref_1): Likewise.
26925 2017-03-28 Martin Liska <mliska@suse.cz>
26928 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
26929 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
26931 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
26932 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
26934 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
26935 (EXTRA_SPECS): Define.
26936 (SUBTARGET_EXTRA_SPECS): Likewise.
26937 (SUBTARGET_CPP_SPEC): Likewise.
26938 * config/arc/elf.h (EXTRA_SPECS): Renamed to
26939 SUBTARGET_EXTRA_SPECS.
26940 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
26942 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
26944 * config/arc/simdext.md (vst64_insn): Update pattern.
26945 (vld32wh_insn): Likewise.
26946 (vld32wl_insn): Likewise.
26947 (vld64_insn): Likewise.
26948 (vld32_insn): Likewise.
26950 2017-03-28 Marek Polacek <polacek@redhat.com>
26953 * fold-const.c (fold_comparison): Use protected_set_expr_location
26954 instead of SET_EXPR_LOCATION.
26956 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
26958 * tree.c (add_expr): Avoid name lookup warning.
26960 2017-03-27 Jeff Law <law@redhat.com>
26962 PR tree-optimization/80216
26963 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
26964 function name. Limit recursion depth.
26965 (record_temporary_equivalences): Corresponding changes.
26967 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
26969 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
26972 2017-03-27 Jakub Jelinek <jakub@redhat.com>
26975 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
26977 * cfgcleanup.c (reg_note_cfa_p): New array.
26978 (insns_have_identical_cfa_notes): New function.
26979 (old_insns_match_p): Don't cross-jump in between /f
26980 and non-/f instructions. If both i1 and i2 are frame related,
26981 verify all CFA notes, their order and content.
26983 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
26986 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
26987 HImode and SImode with zero extend to DImode to one insn.
26988 (bswap<mode>2_extenddi): Likewise.
26989 (bswapsi2_extenddi): Likewise.
26990 (bswaphi2_extendsi): Likewise.
26991 (bswaphi2): Combine bswap HImode and SImode into one insn.
26992 Separate memory insns from swapping register.
26993 (bswapsi2): Likewise.
26994 (bswap<mode>2): Likewise.
26995 (bswaphi2_internal): Delete, no longer used.
26996 (bswapsi2_internal): Likewise.
26997 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
26998 store, and gpr<-gpr swap insns.
26999 (bswap<mode>2_store): Likewise.
27000 (bswaphi2_reg): Register only splitter, combine with the splitter.
27001 (bswaphi2 splitter): Likewise.
27002 (bswapsi2_reg): Likewise.
27003 (bswapsi2 splitter): Likewise.
27004 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
27005 the insns into load, store, and register/register insns.
27006 (bswapdi2_ldbrx): Likewise.
27007 (bswapdi2_load): Likewise.
27008 (bswapdi2_store): Likewise.
27009 (bswapdi2_reg): Likewise.
27011 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
27013 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
27014 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
27016 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
27019 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
27021 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27022 special handling for target option conflicts between dform
27023 options (-mpower9-dform, -mpower9-dform-vector,
27024 -mpower9-dform-scalar) and -mno-direct-move.
27026 2017-03-27 Richard Biener <rguenther@suse.de>
27028 PR tree-optimization/80181
27029 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
27031 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27033 * config/arc/predicates.md (move_double_src_operand): Replace the
27034 call to move_double_src_operand with a call to address_operand.
27036 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27038 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
27039 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
27040 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
27042 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27044 * config/arc/predicates.md (long_immediate_loadstore_operand):
27045 Consider scaled addresses cases.
27047 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27049 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
27050 restored when in interrupt.
27051 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
27052 doesn't have delay slot.
27054 2017-03-27 Richard Biener <rguenther@suse.de>
27057 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
27058 inlined thunk clones.
27060 2017-03-27 Jakub Jelinek <jakub@redhat.com>
27063 * asan.c (instrument_derefs): Copy over last operand from
27064 original COMPONENT_REF to the new COMPONENT_REF with
27065 DECL_BIT_FIELD_REPRESENTATIVE.
27066 * ubsan.c (instrument_object_size): Likewise.
27068 2017-03-27 Richard Biener <rguenther@suse.de>
27070 PR tree-optimization/80170
27071 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
27072 sure DR/SCEV didnt fold in constants we do not see when looking
27073 at the reference base alignment.
27075 2017-03-27 Richard Biener <rguenther@suse.de>
27077 PR middle-end/80171
27078 * gimple-fold.c (fold_ctor_reference): Properly guard against
27079 NULL return value from canonicalize_constructor_val.
27081 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
27084 * config/i386/i386.c (ix86_expand_builtin)
27085 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
27086 flags reg setting and flags reg using instructions.
27087 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
27088 clobbering instructions to zero extend op2.
27090 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
27092 * doc/install.texi (Configuration) <--with-aix-soname>:
27093 Update link to AIX ld.
27095 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
27097 PR rtl-optimization/80160
27098 PR rtl-optimization/80159
27099 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
27100 reg_alternate_class into account.
27102 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
27105 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
27106 to consider in curr_insn_transform.
27108 2017-03-24 Jakub Jelinek <jakub@redhat.com>
27110 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
27111 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
27112 and emit_mode_inner.
27114 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27116 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
27117 argument to the overloaded builtin variants. Use the new flag to
27118 deprecate certain builtin variants.
27119 * config/s390/s390-builtin-types.def: Add new builtin types.
27120 * config/s390/s390-builtins.h: Support new flags field for
27121 overloaded builtins.
27122 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
27123 (s390_macro_to_expand): Enable vector float data type.
27124 (s390_cpu_cpp_builtins_internal): Indicate support of the new
27125 builtins by incrementing the __VEC__ version number.
27126 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
27128 (s390_resolve_overloaded_builtin): Emit error messages depending
27129 on the builtin flags.
27130 * config/s390/s390.c (s390_expand_builtin): Support additional
27131 flags argument. Change error message to match the messages
27132 emitted in s390-c.c.
27133 * config/s390/s390.md: New UNSPEC_* constants.
27134 (op_type): Add new instruction types.
27135 * config/s390/vecintrin.h: Add new builtins and test data class
27137 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
27138 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
27139 (VEC_INEXACT, VEC_NOINEXACT): New constants.
27140 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
27141 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
27142 ("vec_mergel<mode>"): V_HW -> VEC_HW.
27144 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
27145 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
27146 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
27147 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
27149 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
27150 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
27151 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
27152 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
27154 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
27155 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
27156 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
27157 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
27158 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
27159 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
27160 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
27162 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
27163 ("vec_scatter_element<V_HW_4:mode>_DI")
27164 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
27165 ("vec_fpint<mode>", "vflls")
27166 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
27167 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
27168 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
27169 ("*vec_cmphe<mode>_cc"): ... these.
27171 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
27172 mode constant instead of magic value.
27174 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27176 * config/s390/s390.c (s390_expand_vec_compare): Support other
27177 vector floating point modes than just V2DF.
27178 (s390_expand_vcond): Likewise.
27179 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
27180 (s390_cannot_change_mode_class): Prevent mode changes between TF
27181 and V1TF in vector registers.
27182 * config/s390/s390.md (DF, SF): New mode attributes.
27183 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
27184 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
27185 SFmode support for VRs.
27186 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
27188 (VFT, VF_HW): New mode iterators.
27189 (vw, sdx): New mode attributes.
27190 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
27191 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
27192 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
27193 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
27194 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
27195 also the new vector floating point modes. Renaming to ...
27197 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
27198 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
27199 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
27200 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
27201 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
27202 ("vec_unordered<mode>"): ... these.
27204 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
27205 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
27206 ("*vec_extendv2df"): New insn definitions.
27208 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27210 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
27211 ("mulditi3_2", "*muldi3_sign"): New patterns.
27212 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
27213 rename the pattern definition.
27215 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27217 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
27219 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
27221 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27223 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
27224 instruction if possible.
27225 * config/s390/vector.md (vec_halfnumelts): New mode
27227 ("*vec_vllezlf<mode>"): New pattern.
27229 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27231 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
27232 ("popcountv4si2", "popcountv2di2"): Rename to ...
27233 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
27234 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
27236 ("popcount<mode>2_vxe"): New pattern.
27238 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27240 * common/config/s390/s390-common.c (processor_flags_table): Add
27242 * config.gcc: Add arch12.
27243 * config/s390/driver-native.c (s390_host_detect_local_cpu):
27244 Default to arch12 for unknown CPU model numbers.
27245 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
27246 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
27247 PROCESSOR_max sanity check.
27248 * config/s390/s390-opts.h (enum processor_type): Add
27250 * config/s390/s390.c (processor_table): Add arch12.
27251 (s390_expand_builtin): Add check for B_VXE flag.
27252 (s390_issue_rate): Add PROCESSOR_ARCH12.
27253 (s390_get_sched_attrmask): Likewise.
27254 (s390_get_unit_mask): Likewise.
27255 (s390_sched_score): Enable z13 scheduling for arch12.
27256 (s390_sched_reorder): Likewise.
27257 (s390_sched_variable_issue): Likewise.
27258 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
27260 (s390_tune_attr): Use z13 scheduling also for arch12.
27261 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
27262 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
27263 (TARGET_VXE_P): New macros.
27264 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
27265 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
27266 * config/s390/s390.opt: Add arch12 as processor_type.
27268 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27270 * config/s390/s390.md
27271 ("fixuns_truncdddi2", "fixuns_trunctddi2")
27272 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
27273 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
27275 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
27276 Rename expanders to ...
27278 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
27279 ("fixuns_truncdddi2_emu"): ... these.
27281 ("fixuns_trunc<mode>si2_emu"): New expander.
27283 ("*fixuns_truncdfdi2_z13"): Rename to ...
27284 ("*fixuns_truncdfdi2_vx"): ... this.
27286 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27288 * config/s390/2964.md: Remove the single element vector compare
27289 instructions which are no longer used.
27290 * config/s390/s390.c (s390_select_ccmode): Remove handling of
27292 (s390_canonicalize_comparison): Remove handling of DFmode
27294 (s390_expand_vec_compare_scalar): Remove function.
27295 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
27296 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
27298 ("*cmp<mode>_ccs"): Add wfcdb instruction.
27300 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27302 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
27304 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
27305 will anyway by matched by mov<mode>_64dfp.
27307 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27309 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
27310 vlef/vstef. Add missing operand to vleif.
27312 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27314 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
27315 pair for all vector types with 64 bit elements.
27316 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
27317 * config/s390/vector.md (V_HW_64): ... here.
27318 (V_128_NOSINGLE): New mode iterator.
27319 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
27320 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
27321 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
27322 ("*vec_load_pairv2di"): Change to ...
27323 ("*vec_load_pair<mode>"): ... this one.
27325 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27327 * config/s390/constraints.md: Add comments.
27328 (jKK): Reject element sizes > 8 bytes.
27329 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
27331 * config/s390/s390.md: Add the s_operand checks formerly in
27332 s390_split_ok_p to various splitters where they are still
27334 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
27335 for 128 bit vectors. Plus two splitters.
27337 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27339 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
27342 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27345 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
27346 error if the boundary argument is not constant.
27348 2017-03-24 Jakub Jelinek <jakub@redhat.com>
27350 PR rtl-optimization/80112
27351 * loop-doloop.c (doloop_condition_get): Don't check condition
27352 if cmp isn't SET with IF_THEN_ELSE src.
27354 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27356 PR tree-optimization/80158
27357 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
27358 replacing a candidate statement, also replace it for the
27359 candidate's alternate interpretation.
27360 (replace_rhs_if_not_dup): Likewise.
27361 (replace_one_candidate): Likewise.
27363 2017-03-24 Richard Biener <rguenther@suse.de>
27365 PR tree-optimization/80167
27366 * graphite-isl-ast-to-gimple.c
27367 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
27369 (translate_isl_ast_to_gimple::get_rename): Likewise.
27371 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
27373 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
27374 handling of certain combinations of target options, including the
27375 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
27376 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
27378 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27381 * config/arm/arm.md (*load_multiple): Add reload_completed to
27382 matching condition.
27384 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27385 Richard Biener <rguenth@suse.de>
27387 PR tree-optimization/79908
27388 PR tree-optimization/80136
27389 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
27390 been cast away, gimplify_and_add suffices.
27392 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
27394 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
27396 2017-03-23 Richard Biener <rguenther@suse.de>
27398 PR tree-optimization/80032
27399 * gimplify.c (gimple_push_cleanup): Forced unconditional
27400 cleanups still have to go to the conditional_cleanups
27403 2017-03-22 Jakub Jelinek <jakub@redhat.com>
27405 PR tree-optimization/80072
27406 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
27408 (next_operand_entry_id): Change type to unsigned int.
27409 (sort_by_operand_rank): Make sure to return the right return value
27410 even if unsigned fields are bigger than INT_MAX.
27411 (struct oecount): Change cnt and id type to unsigned int.
27412 (oecount_hasher::equal): Formatting fix.
27413 (oecount_cmp): Make sure to return the right return value
27414 even if unsigned fields are bigger than INT_MAX.
27415 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
27418 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
27419 TREE_READONLY on result if writing it more than once.
27422 * doc/invoke.texi (-fsanitize=thread): Document that with
27423 -fnon-call-exceptions atomics are not able to throw
27427 * tsan.c: Include tree-eh.h.
27428 (instrument_builtin_call): Call maybe_clean_eh_stmt or
27429 maybe_clean_or_replace_eh_stmt where needed.
27430 (instrument_memory_accesses): Add cfg_changed argument.
27431 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
27432 if it returned true.
27433 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
27435 PR rtl-optimization/63191
27436 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
27437 wrapper function, moved the whole old content into ...
27438 (ix86_delegitimize_address_1): ... this. New inline function.
27439 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
27440 true as last argument instead of ix86_delegitimize_address.
27442 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
27444 * config/aarch64/aarch64.c (generic_branch_cost): Copy
27445 cortexa57_branch_cost.
27447 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
27449 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
27451 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27454 * doc/md.texi (Constraints): Document wA constraint.
27455 * config/rs6000/constraints.md (wA): New.
27456 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
27457 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
27458 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
27459 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
27461 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
27464 * gimplify.c (is_oacc_declared): New function.
27465 (oacc_default_clause): Use it to set default flags for acc declared
27466 variables inside parallel regions.
27467 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
27468 declared variables.
27469 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
27470 declare attribute to any decl as necessary.
27472 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
27475 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
27476 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
27477 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
27478 (arm_arch_lpae): This.
27479 * config/arm/arm.c (arm_arch7ve): Rename into ...
27480 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
27481 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
27484 2017-03-22 Martin Liska <mliska@suse.cz>
27487 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
27488 error message instead of an ICE.
27490 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27492 * doc/extend.texi (6.11 Additional Floating Types): Revise.
27494 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
27496 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
27498 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27501 2017-03-21 Martin Sebor <msebor@redhat.com>
27503 * doc/extend.texi: Use "cannot" instead of "can't."
27504 * doc/hostconfig.texi: Same.
27505 * doc/install.texi: Same.
27506 * doc/invoke.texi: Same.
27507 * doc/loop.texi: Same.
27508 * doc/md.texi: Same.
27509 * doc/objc.texi: Same.
27510 * doc/rtl.texi: Same.
27511 * doc/tm.texi: Same.
27512 * doc/tm.texi.in: Same.
27513 * doc/trouble.texi: Same.
27515 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
27518 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
27519 (collect_checksum_attributes): Set it.
27520 (die_checksum_ordered): Use it.
27522 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27524 PR tree-optimization/79908
27525 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
27526 change: For a VA_ARG whose LHS has been cast away, use
27527 force_gimple_operand to construct the side effects.
27529 2017-03-21 David Malcolm <dmalcolm@redhat.com>
27531 PR translation/80001
27532 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
27533 more amenable to translation.
27534 (oacc_loop_auto_partitions): Likewise.
27536 2017-03-21 Marek Polacek <polacek@redhat.com>
27537 Martin Sebor <msebor@redhat.com>
27539 PR tree-optimization/80109
27540 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
27541 on INTEGRAL_TYPE_P.
27543 2017-03-21 Jakub Jelinek <jakub@redhat.com>
27544 Segher Boessenkool <segher@kernel.crashing.org>
27547 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
27548 check reg_used_between_p between insn and one of succ or succ2
27549 depending on if succ is artificial insn not inserted into insn
27552 2017-03-21 Martin Liska <mliska@suse.cz>
27554 PR gcov-profile/80081
27555 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
27556 * doc/gcc.texi: Include gcov-dump stuff.
27557 * doc/gcov-dump.texi: New file.
27559 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
27561 PR rtl-optimization/79150
27562 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
27563 conditional jump, if the jump is the last insn of the loop.
27565 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27566 Richard Biener <rguenth@suse.de>
27568 PR tree-optimization/79908
27569 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
27570 been cast away, use force_gimple_operand to construct the side
27573 2017-03-21 Martin Liska <mliska@suse.cz>
27575 PR libfortran/79956
27576 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
27579 2017-03-21 Brad Spengler <spender@grsecurity.net>
27582 * plugin.c (htab_hash_plugin): New function.
27583 (add_new_plugin): Use it and adjust.
27584 (parse_plugin_arg_opt): Adjust.
27585 (init_one_plugin): Likewise.
27587 2017-03-21 Richard Biener <rguenther@suse.de>
27589 PR tree-optimization/80032
27590 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
27591 if set force the cleanup to happen unconditionally.
27592 (gimplify_target_expr): Push inserted clobbers with force_uncond
27593 to avoid them being removed by control-dependent DCE.
27595 2017-03-21 Richard Biener <rguenther@suse.de>
27597 PR tree-optimization/80122
27598 * tree-inline.c (copy_bb): Do not expans va-arg packs or
27599 va_arg_pack_len when the inlined call stmt requires pack
27601 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
27603 2017-03-21 Jakub Jelinek <jakub@redhat.com>
27606 * tsan.c (instrument_builtin_call): If the memory model argument
27607 is not a constant, assume it is valid.
27610 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
27613 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
27615 PR rtl-optimization/79910
27616 * combine.c (can_combine_p): Do not allow combining an I0 or I1
27617 if its dest is used by an insn before I2 (other than the combined
27618 insns themselves, which are properly handled already).
27620 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
27623 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
27625 * combine.c (record_used_regs): New static function.
27626 (try_combine): Handle situations where there is an additional
27627 instruction between I2 and I3 which needs to have a LOG_LINK
27631 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
27633 * combine.c (try_combine): Delete redundant i1 test. Call
27634 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
27636 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27639 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
27640 alternatives 13/14.
27642 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27644 PR tree-optimization/80054
27645 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
27646 the optimization if a PHI or any of its arguments is not dominated
27647 by the candidate's basis. Use gphi* rather than gimple* as
27649 (replace_profitable_candidates): Clean up a gimple* variable that
27650 should be a gphi* variable.
27652 2017-03-20 Martin Sebor <msebor@redhat.com>
27655 * doc/extend.texi (attribute constructor): Document present limitation.
27657 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
27660 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
27661 __POWER9_VECTOR__ #ifdef control, change template definition to
27662 use Power9-specific built-in function.
27663 (vec_any_eq): Likewise.
27664 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
27665 to control outcomes from this test.
27666 (vector_ae_<mode>p): For VEC_F modes, likewise.
27668 2017-03-20 Ian Lance Taylor <iant@google.com>
27670 * config/i386/i386.c (ix86_function_regparm): Save an extra
27671 register for -fsplit-stack with DECL_STATIC_CHAIN.
27673 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
27676 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
27677 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
27679 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
27681 * config/riscv/riscv.c (riscv_print_operand): Use "fence
27683 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
27686 2017-03-20 Marek Polacek <polacek@redhat.com>
27689 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
27691 2017-03-20 Richard Biener <rguenther@suse.de>
27693 PR tree-optimization/80113
27694 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
27695 allocate extra SSA name for PHI def.
27696 (add_close_phis_to_outer_loops): Likewise.
27697 (add_close_phis_to_merge_points): Likewise.
27698 (copy_loop_close_phi_args): Likewise.
27699 (copy_cond_phi_nodes): Likewise.
27701 2017-03-20 Martin Liska <mliska@suse.cz>
27703 PR middle-end/79753
27704 * tree-chkp.c (chkp_build_returned_bound): Do not build
27705 returned bounds for a LHS that's not a BOUNDED_P type.
27707 2017-03-20 Martin Liska <mliska@suse.cz>
27711 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
27712 COMPLEX_CST and VECTOR_CST.
27714 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27717 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
27718 target operand. A new splitter adds the clobber statement in case
27719 the target operand is dead anyway.
27721 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
27723 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
27724 to age-old versions of binutils and glibc.
27726 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
27728 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
27730 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
27732 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
27734 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
27736 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
27737 requirement for binutils 2.13.
27739 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
27741 * combine.c (try_combine): Delete redundant i1 test. Call
27742 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
27744 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
27746 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
27747 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
27749 <riscv64-*-elf>: Re-arrange section
27750 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
27751 <riscv32-*-linux>: Likewise.
27752 <riscv64-*-elf>: Likewise
27753 <riscv64-*-linux>: Likewise.
27755 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
27758 * aarch64.opt(verbose-cost-dump): Fix typo.
27760 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
27763 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
27764 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
27766 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
27768 * reload.c (find_reloads): When reloading a nonoffsettable address,
27769 use RELOAD_OTHER for it and its address reloads.
27771 PR rtl-optimization/79910
27772 * combine.c (record_used_regs): New static function.
27773 (try_combine): Handle situations where there is an additional
27774 instruction between I2 and I3 which needs to have a LOG_LINK
27777 2017-03-17 Jeff Law <law@redhat.com>
27779 PR tree-optimization/71437
27780 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
27781 conditional in the hash table first.
27782 (vrp_dom_walker::before_dom_children): Extract condition from
27783 ASSERT_EXPR. Record condition, its inverion and any implied
27784 conditions as well.
27786 2017-03-17 Marek Polacek <polacek@redhat.com>
27787 Markus Trippelsdorf <markus@trippelsdorf.de>
27789 PR tree-optimization/80079
27790 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
27793 2017-03-17 Richard Biener <rguenther@suse.de>
27795 PR middle-end/80075
27796 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
27797 Properly verify the LHS before the RHS possibly claims to be
27799 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
27802 2017-03-17 Martin Jambor <mjambor@suse.cz>
27804 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
27805 (List of -O2 options): Likewise.
27806 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
27809 2017-03-17 Tom de Vries <tom@codesourcery.com>
27811 * gcov-dump.c (print_usage): Print bug_report_url.
27813 2017-03-17 Richard Biener <rguenther@suse.de>
27815 PR middle-end/80050
27816 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
27817 (parser::peek): Likewise.
27819 2017-03-17 Richard Biener <rguenther@suse.de>
27821 PR tree-optimization/80048
27822 * sese.c (free_sese_info): Properly release rename_map and
27823 copied_bb_map elements.
27825 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
27827 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
27828 Add linked-list forward and backlinks. Insert on
27829 construction, remove on destruction.
27830 (class pass_store_merging): Add m_stores_head field.
27831 (pass_store_merging::terminate_and_process_all_chains):
27832 Iterate over m_stores_head list.
27833 (pass_store_merging::terminate_all_aliasing_chains):
27835 (pass_store_merging::execute): Check for debug stmts first.
27836 Push new chains onto the m_stores_head stack.
27838 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
27841 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
27842 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
27843 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
27845 2017-03-16 Jeff Law <law@redhat.com>
27847 PR tree-optimization/71437
27848 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
27849 member function. Implementation moved into after_dom_children
27850 member function and into the threader's thread_outgoing_edges
27852 (dom_opt_dom_walker::after_dom_children): Simplify by moving
27853 some code into new thread_outgoing_edges.
27854 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
27855 definition. Simplify marker handling (do it here). Assume we always
27856 have the available expression and the const/copies tables.
27857 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
27859 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
27860 * tree-vrp.c (equiv_stack): No longer file scoped.
27861 (vrp_dom_walker): New class.
27862 (vrp_dom_walker::before_dom_children): New member function.
27863 (vrp_dom_walker::after_dom_children): Likewise.
27864 (identify_jump_threads): Setup domwalker. Use it rather than
27865 walking edges in a random order by hand. Simplify setup/finalization.
27866 (finalize_jump_threads): Remove.
27867 (vrp_finalize): Do not call identify_jump_threads here.
27868 (execute_vrp): Do it here instead and call thread_through_all_blocks
27871 PR tree-optimization/71437
27872 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
27874 (simplify_stmt_for_jump_threading): Add basic_block argument. All
27876 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
27877 (dom_opt_dom_walker::thread_across_edge): Remove
27878 handle_dominating_asserts argument. All callers changed.
27879 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
27880 changes. Remove calls to lhs_of_dominating_assert. Other
27881 uses of handle_dominating_asserts turn into unconditional code
27882 (simplify_control_stmt_condition_1): Likewise.
27883 (simplify_control_stmt_condition): Likewise.
27884 (thread_through_normal_block, thread_across_edge): Likewise.
27885 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
27886 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
27887 object if it is not an SSA_NAME.
27888 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
27889 before calling into the VRP specific simplifiers.
27890 (identify_jump_threads): Remove handle_dominating_asserts
27893 2017-03-16 Jakub Jelinek <jakub@redhat.com>
27896 * tree-diagnostic.c (default_tree_printer): No longer static.
27897 * tree-diagnostic.h (default_tree_printer): New prototype.
27899 2017-03-16 Tamar Christina <tamar.christina@arm.com>
27901 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
27902 Change ins into fmov.
27904 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27906 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
27907 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
27908 Use h_con constraint for operand 1.
27909 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
27910 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
27912 2017-03-15 Jeff Law <law@redhat.com>
27914 PR tree-optimization/71437
27915 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
27916 (record_temporary_equivalences): Use it.
27918 PR tree-optimization/71437
27919 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
27920 tree-ssa-scopedtables.
27921 (lookup_avail_expr, build_and_record_new_cond): Likewise.
27922 (record_conditions, record_cond, vuse_eq): Likewise.
27923 (record_edge_info): Adjust to API tweak of record_conditions.
27924 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
27925 (record_temporary_equivalences, optimize_stmt): Likewise.
27926 (eliminate_redundant_computations): Likewise.
27927 (record_equivalences_from_stmt): Likewise.
27928 * tree-ssa-scopedtables.c: Include options.h and params.h.
27929 (vuse_eq): New function, moved from tree-ssa-dom.c
27930 (build_and_record_new_cond): Likewise.
27931 (record_conditions): Likewise. Accept vector of conditions rather
27932 than edge_equivalence structure for first argument.
27933 for the first argument.
27934 (avail_exprs_stack::lookup_avail_expr): New member function, moved
27935 from tree-ssa-dom.c.
27936 (avail_exprs_stack::record_cond): Likewise.
27937 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
27938 from tree-ssa-dom.c.
27939 (avail_exprs_stack): Add new member functions lookup_avail_expr
27941 (record_conditions): Declare.
27943 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
27946 * lra-constraints.c (process_alt_operands): Increase reject for
27947 reloading an input/output operand.
27949 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
27952 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
27953 insns to convert from signed/unsigned char/short to IEEE 128-bit
27955 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
27957 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
27960 * config/i386/i386.c (ix86_vector_duplicate_value): Create
27961 subreg of inner mode for values already in registers.
27963 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
27965 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
27966 iteration reg is used after the loop.
27968 2017-03-14 Martin Sebor <msebor@redhat.com>
27970 PR tree-optimization/79800
27971 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
27972 precision in negative-positive range.
27973 (format_floating): Call non-const overload with adjusted precision.
27975 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
27978 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
27981 2017-03-14 Martin Sebor <msebor@redhat.com>
27983 PR middle-end/80020
27984 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
27985 * builtins.def (aligned_alloc): Use it.
27988 * Makefile.in (GTFILES): Add calls.c.
27989 * calls.c: Include "gt-calls.h".
27991 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
27993 PR rtl-optimization/79728
27994 * regs.h (struct target_regs): New field
27995 x_contains_allocatable_regs_of_mode.
27996 (contains_allocatable_regs_of_mode): New macro.
27997 * reginfo.c (init_reg_sets_1): Initialize it, and change
27998 contains_reg_of_mode so it includes global regs as well.
27999 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
28000 rather than contains_regs_of_mode.
28002 2017-03-14 Martin Liska <mliska@suse.cz>
28004 * doc/invoke.texi: Document options that can't be combined with
28005 -fcheck-pointer-bounds.
28007 2017-03-14 Martin Liska <mliska@suse.cz>
28009 PR middle-end/79831
28010 * doc/invoke.texi (-Wchkp): Document the option.
28012 2017-03-14 Martin Liska <mliska@suse.cz>
28014 * Makefile.in: Install gcov-dump.
28016 2017-03-14 Martin Liska <mliska@suse.cz>
28018 * multiple_target.c (expand_target_clones): Bail out for
28019 an invalid attribute.
28021 2017-03-14 Richard Biener <rguenther@suse.de>
28023 * alias.c (struct alias_set_entry): Pack properly.
28024 * cfgloop.h (struct loop): Likewise.
28025 * cse.c (struct set): Likewise.
28026 * ipa-utils.c (struct searchc_env): Likewise.
28027 * loop-invariant.c (struct invariant): Likewise.
28028 * lra-remat.c (struct cand): Likewise.
28029 * recog.c (struct change_t): Likewise.
28030 * rtl.h (struct address_info): Likewise.
28031 * symbol-summary.h (function_summary): Likewise.
28032 * tree-loop-distribution.c (struct partition): Likewise.
28033 * tree-object-size.c (struct object_size_info): Likewise.
28034 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
28035 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
28036 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
28037 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
28038 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
28039 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
28040 (struct _stmt_vec_info): Likewise.
28042 2017-03-14 Martin Liska <mliska@suse.cz>
28045 * multiple_target.c (create_dispatcher_calls): Check that
28046 a target can create a function dispatcher.
28048 2017-03-14 Martin Liska <mliska@suse.cz>
28051 * multiple_target.c (expand_target_clones): Drop local.local
28052 flag for default implementation.
28054 2017-03-14 Richard Biener <rguenther@suse.de>
28056 PR tree-optimization/80030
28057 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
28059 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
28061 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
28062 gcc_fallthrough() instead of __attribute__((fallthrough));
28064 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
28066 * doc/gcc.texi: Remove "up" link to (DIR).
28067 * doc/gccint.texi: Ditto.
28069 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
28071 * doc/install.texi (Specific) <avr>: Remove reference to
28074 2017-03-13 Jeff Law <law@redhat.com>
28076 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
28077 attribute rather than comments.
28079 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
28080 match_scratch operand is highest.
28082 2017-03-13 Martin Liska <mliska@suse.cz>
28084 PR middle-end/78339
28085 * ipa-pure-const.c (warn_function_noreturn): If the declarations
28086 is a CHKP clone, use original declaration.
28088 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28090 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
28091 (arc_conditional_register_usage): Use a different allocation order
28092 when optimizing for size.
28093 * common/config/arc/arc-common.c (arc_option_optimization_table):
28094 Section anchors default on when optimizing for size.
28096 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28098 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
28100 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28102 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
28103 * config/arc/arc.md (cpu_facility): Add cd variant.
28104 (*movqi_insn): Add code density variant.
28105 (*movhi_insn): Likewise.
28106 (*movqi_insn): Likewise.
28107 (*addsi3_mixed): Likewise.
28108 (subsi3_insn): Likewise.
28110 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28112 * config/arc/arc.md (movsi_cond_exec): Update constraint.
28114 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28116 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
28117 expressions with MINUS and UNARY ops.
28119 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28122 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
28124 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
28125 between vec_select and vector argument.
28126 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
28127 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
28128 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
28129 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
28130 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
28131 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
28133 2017-03-13 Richard Biener <rguenther@suse.de>
28136 * params.def (vect-max-peeling-for-alignment): Fix typo.
28138 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28140 * doc/install.texi (Specific) <mips-*-*>: Remove description of
28141 issue that only occurred with binutils below 2.18.
28143 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28145 * doc/install.texi (Specific) <cris-axis-elf>: No longer
28146 refer to binutils 2.11/2.12 minimum.
28148 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28150 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
28151 ftp.kernel.org and simplify binutils requirement.
28153 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
28155 * doc/invoke.texi (Warning Options): Fix spelling of link-time
28157 (Optimize Options): Ditto. Also remove redundancy.
28159 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28161 PR translation/79848
28162 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
28164 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
28165 to G_ to avoid double translation.
28167 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28169 PR translation/79923
28170 * auto-profile.c (get_combined_location): Convert leading
28171 character of diagnostics to lower case and remove trailing period.
28172 (read_profile): Likewise for various diagnostics.
28173 * config/arm/arm.c (arm_option_override): Remove trailing period
28174 from various diagnostics.
28175 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
28176 (msp430_expand_delay_cycles): Likewise.
28178 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28181 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
28182 full command-line argument, rather than just "str".
28183 (aarch64_validate_march): Likewise.
28184 (aarch64_validate_mtune): Likewise.
28186 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
28188 PR rtl-optimization/78911
28189 * lra-assigns.c (must_not_spill_p): New function.
28190 (spill_for): Use it.
28192 2017-03-10 Jakub Jelinek <jakub@redhat.com>
28194 PR tree-optimization/79981
28195 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
28196 ATOMIC_COMPARE_EXCHANGE ifn result.
28197 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
28198 IFN_ATOMIC_COMPARE_EXCHANGE.
28200 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28203 * opts.c (parse_sanitizer_options): Add missing question mark to
28204 "did you mean" message.
28206 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28208 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
28210 (VMULEUH_UNS): Likewise.
28211 (VMULOUB_UNS): Likewise.
28212 (VMULOUH_UNS): Likewise.
28213 * config/rs6000/rs6000.c (builtin_function_type): Remove
28214 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
28216 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28219 * read-rtl-function.c (function_reader::read_rtx_operand): Update
28220 x with result of extra_parsing_for_operand_code_0.
28221 (function_reader::extra_parsing_for_operand_code_0): Convert
28222 return type from void to rtx, returning x. When reading
28223 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
28224 larger size containing struct block_symbol.
28226 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
28228 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
28229 -mfloat128-hardware without -m64.
28231 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
28234 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
28235 entries to the case statement that marks unsigned arguments to
28236 overloaded functions.
28238 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
28240 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
28241 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
28243 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
28246 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
28247 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
28249 2017-03-10 Martin Liska <mliska@suse.cz>
28253 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
28254 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
28257 2017-03-10 Olivier Hainque <hainque@adacore.com>
28259 * tree-switch-conversion (array_value_type): Start by resetting
28260 candidate type to it's main variant.
28262 2017-03-10 Jakub Jelinek <jakub@redhat.com>
28264 PR rtl-optimization/79909
28265 * combine.c (try_combine): Use simplify_replace_rtx on individual
28266 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
28267 of the whole CALL_INSN_FUNCTION_USAGE.
28269 PR tree-optimization/79972
28270 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
28271 get_range_info on SSA_NAMEs. Formatting fixes.
28273 2017-03-10 Richard Biener <rguenther@suse.de>
28274 Jakub Jelinek <jakub@redhat.com>
28276 PR tree-optimization/77975
28277 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
28278 edge to be constant.
28279 (get_val_for): For constant x return it. Formatting fix.
28280 (loop_niter_by_eval): Avoid pointless looping if the next iteration
28281 would use the same bases as the current one.
28283 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28285 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
28286 instead of vec_select for V1TImode.
28287 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
28289 (VSX_LE_128): Add V1TI to this mode iterator.
28290 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
28291 (*vsx_le_perm_store_<mode>): Likewise.
28292 (pre-reload splitter for VSX stores): Likewise.
28293 (post-reload splitter for VSX stores): Likewise.
28294 (*vsx_xxpermdi2_le_<mode>): Likewise.
28295 (*vsx_lxvd2x2_le_<mode>): Likewise.
28296 (*vsx_stxvd2x2_le_<mode>): Likewise.
28298 2017-03-09 Michael Eager <eager@eagercon.com>
28300 Correct failures with --enable-checking=yes,rtl.
28302 * config/microblaze/microblaze.c (microblaze_expand_shift):
28303 Replace GET_CODE test with CONST_INT_P and INTVAL test with
28304 test for const0_rtx.
28305 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
28306 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
28308 2017-03-09 Richard Biener <rguenther@suse.de>
28310 PR tree-optimization/79977
28311 * graphite-scop-detection.c (scop_detection::merge_sese):
28312 Handle the case of extra exits to blocks dominating the entry.
28314 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
28316 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
28319 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
28321 PR rtl-optimization/79949
28322 * lra-constraints.c (process_alt_operands): Check memory when
28323 trying to predict a cycle. Print about the overall increase.
28325 2017-03-09 Richard Biener <rguenther@suse.de>
28327 PR middle-end/79971
28328 * gimple-expr.c (useless_type_conversion_p): Preserve
28329 TYPE_SATURATING for fixed-point types.
28331 2017-03-09 Richard Biener <rguenther@suse.de>
28334 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
28335 alignment of BLKmode params.
28337 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28340 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
28341 (VALL_NO_V2Q): Likewise.
28343 * config/aarch64/aarch64-simd.md
28344 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
28346 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
28347 VALL_NO_V2Q mode iterator.
28348 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
28350 2017-03-09 Martin Liska <mliska@suse.cz>
28352 PR tree-optimization/79631
28353 * tree-chkp-opt.c (chkp_is_constant_addr): Call
28354 tree_int_cst_sign_bit just for INTEGER constants.
28356 2017-03-09 Martin Liska <mliska@suse.cz>
28360 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
28363 2017-03-09 Marek Polacek <polacek@redhat.com>
28366 * tree.c (inchash::add_expr): Handle TREE_VEC.
28368 2017-03-09 Martin Liska <mliska@suse.cz>
28371 (chkp_narrow_size_and_offset): New function.
28372 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
28373 (void chkp_parse_bit_field_ref): New function.
28374 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
28375 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
28377 2017-03-09 Martin Liska <mliska@suse.cz>
28380 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
28381 (chkp_find_bounds_1): Remove gcc_unreachable.
28383 2017-03-09 Jakub Jelinek <jakub@redhat.com>
28386 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
28387 BUILT_IN_SYNC*, determine the access type from the size suffix and
28388 always build a MEM_REF with that type. Handle forgotten
28389 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
28392 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
28393 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
28394 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
28395 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
28396 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
28397 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
28398 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
28399 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
28400 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
28401 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
28402 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
28403 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
28404 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
28405 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
28406 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
28407 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
28408 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
28409 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
28410 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
28411 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
28412 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
28413 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
28414 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
28415 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
28416 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
28417 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
28418 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
28419 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
28420 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
28421 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
28422 definitions outside of __OPTIMIZE__ guarded section.
28425 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
28426 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
28427 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
28428 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
28431 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28433 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
28434 ("vfenez<mode>"): Add missing constraints.
28436 2017-03-08 Martin Sebor <msebor@redhat.com>
28439 * config/nds32/nds32.c (nds32_option_override):
28440 Fix misspelled diagnostic.
28442 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28445 * gimplify.c (gimplify_omp_for): Replace index var in outer
28446 taskloop statement with an artificial variable and add
28447 OMP_CLAUSE_PRIVATE clause for it.
28449 2017-03-08 Richard Biener <rguenther@suse.de>
28451 PR tree-optimization/79955
28452 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
28453 for accesses that are completely outside of the variable.
28455 2017-03-08 Andrew Haley <aph@redhat.com>
28457 PR tree-optimization/79943
28458 * tree-ssa-loop-split.c (compute_new_first_bound): When
28459 calculating the new upper bound, (END-BEG) should be added, not
28462 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28464 * config/avr/avr.md (setmemhi): Make sure match_dup
28465 operand number comes before match_scratch.
28467 2017-03-08 Richard Biener <rguenther@suse.de>
28469 PR tree-optimization/79920
28470 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
28471 with ncopies == 1 to ...
28472 (vect_transform_slp_perm_load): ... here. Properly compute
28473 all element loads by iterating VF times over the group. Do
28474 not handle ncopies (computed in a broken way) in
28475 vect_create_mask_and_perm.
28477 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28480 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
28481 is a uniform vector, use uniform_vector_p return value instead of
28482 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
28484 2017-03-07 Marek Polacek <polacek@redhat.com>
28486 PR middle-end/79809
28487 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
28488 (alloca_call_type): Likewise.
28490 2017-03-07 Martin Liska <mliska@suse.cz>
28492 * gcov.c (process_args): Put comment to correct location.
28494 2017-03-07 Martin Liska <mliska@suse.cz>
28496 PR middle-end/68270
28497 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
28498 Use array_at_struct_end_p instead of DECL_CHAIN (field).
28499 (chkp_narrow_bounds_for_field): Likewise.
28500 (chkp_parse_array_and_component_ref): Pass one more argument to
28503 2017-03-07 Richard Biener <rguenther@suse.de>
28505 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
28508 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
28510 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
28511 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
28513 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28516 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
28517 to end of description.
28518 (PARAM_MAX_STORES_TO_MERGE): Likewise.
28520 2017-03-07 Jakub Jelinek <jakub@redhat.com>
28522 PR rtl-optimization/79901
28523 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
28525 (*avx512f_<code><mode>3<mask_name>): ... this.
28526 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
28527 iterator instead of VI8_AVX2_AVX512BW.
28529 PR rtl-optimization/79901
28530 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
28531 min/max expander, expand it using expand_vec_cond_expr.
28534 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
28537 2017-03-06 Jakub Jelinek <jakub@redhat.com>
28540 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
28541 to void * for PCH reasons.
28542 * dwarf2out.c (output_loc_operands, output_die): Cast
28543 v.val_vec.array to unsigned char *.
28545 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
28548 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
28551 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
28553 PR rtl-optimization/79571
28554 * lra-constraints.c (process_alt_operands): Calculate static
28555 reject and subtract it from overall when only addresses will be
28558 2017-03-06 Julia Koval <julia.koval@intel.com>
28561 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
28562 incoming stack boundary to 128 for 64-bit targets.
28564 2017-03-06 Richard Biener <rguenther@suse.de>
28566 PR tree-optimization/79894
28567 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
28568 to NULL after folding it.
28570 2017-03-06 Richard Biener <rguenther@suse.de>
28572 PR tree-optimization/79824
28573 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
28574 check disabling peeling for gaps.
28576 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
28578 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
28579 attributes): Document gettimeofday.
28581 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
28583 * config/s390/s390.c (s390_option_override_internal): Set
28584 PARAM_MIN_VECT_LOOP_BOUND
28586 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
28588 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
28589 * config/s390/s390.md: Likewise.
28591 2017-03-06 Jakub Jelinek <jakub@redhat.com>
28594 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
28595 (<avx2_avx512>_perm<mode>): Rename to ...
28596 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
28598 (<avx512>_perm<mode>_mask): Rename to ...
28599 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
28601 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
28602 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
28603 instead of VI8F_256_512.
28604 (avx512f_perm<mode>): New define_expand.
28605 (avx512f_perm<mode>_mask): Likewise.
28606 (avx512f_perm<mode>_1<mask_name>): New define_insn.
28607 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
28609 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28611 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
28612 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
28614 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
28616 2017-03-06 Martin Liska <mliska@suse.cz>
28619 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
28620 when having a SSA NAME w/o VAR_DECL assigned to it.
28622 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28624 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
28625 msa_dpsub_<su>_d): Fix MODE for vec_select.
28627 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28629 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
28631 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
28633 2017-03-06 Richard Biener <rguenther@suse.de>
28635 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
28636 * plugin.c (register_plugin_info): Likewise.
28637 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
28639 2017-03-05 Jakub Jelinek <jakub@redhat.com>
28641 * config/i386/sse.md (sse_storehps, sse_storelps,
28642 avx_<castmode><avxsizesuffix>_<castmode>,
28643 avx512f_<castmode><avxsizesuffix>_<castmode>,
28644 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
28645 in condition that at least one operand is not a MEM.
28647 2017-03-03 Jakub Jelinek <jakub@redhat.com>
28649 PR middle-end/79805
28650 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
28651 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
28653 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
28654 gimple_call_nothrow_p flag based on whether original builtin can throw.
28655 If it can, emit following stmts on the fallthrough edge.
28656 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
28657 don't create new bb if inserting just debug stmts on the edge, try to
28658 insert them on the fallthru bb or just reset debug stmts.
28660 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
28663 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
28664 restore recog_data (including the operand rtxes inside it) around
28665 the call to get_insn_template.
28667 2017-03-03 Martin Sebor <msebor@redhat.com>
28669 PR tree-optimization/79699
28670 * context.c (context::~context): Free MPFR caches to avoid
28671 a memory leak on program exit.
28673 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28675 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
28676 Use wide_int::ulow () instead of .elt (0).
28678 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
28680 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
28681 (*pushxf): Limit oF constraint to 32bit targets and add oC
28682 constraint for 64bit targets.
28683 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
28684 (*pushdf): Change rmF constraint to rmC.
28686 2017-03-03 Martin Liska <mliska@suse.cz>
28688 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
28689 Remove unused variable.
28691 2017-03-03 Jakub Jelinek <jakub@redhat.com>
28694 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
28695 is a memory operand, increase num_memory.
28696 (ix86_expand_args_builtin): Likewise.
28698 2017-03-03 Jan Hubicka <jh@suse.cz>
28701 * ipa-devirt.c (maybe_record_node): Properly handle
28702 __cxa_pure_virtual visibility.
28704 2017-03-03 Martin Liska <mliska@suse.cz>
28706 PR tree-optimization/79803
28707 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
28709 (pass_loop_prefetch::execute): Disabled optimization if an
28710 assumption about L1 cache size is not met.
28712 2017-03-03 Martin Liska <mliska@suse.cz>
28714 PR rtl-optimization/79574
28715 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
28716 (hash_scan_set): Likewise.
28717 (dump_hash_table): Likewise.
28718 (hoist_code): Likewise.
28720 2017-03-03 Richard Biener <rguenther@suse.de>
28722 * fixed-value.c (fixed_from_string): Restore use of elt (1)
28723 in place of uhigh ().
28724 (fixed_convert_from_real): Likewise.
28726 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
28729 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
28731 2017-03-03 Richard Biener <rguenther@suse.de>
28733 PR middle-end/79818
28734 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
28735 TYPE_OVERFLOW_UNDEFINED check.
28737 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28739 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
28741 (vector_ae_<mode>_p): Likewise.
28742 (vector_nez_<mode>_p): Likewise.
28743 (vector_ne_v2di_p): Likewise.
28744 (vector_ae_v2di_p): Likewise.
28745 (vector_ne_<mode>_p): Likewise.
28746 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
28748 (vsx_tsqrt<mode>2_fe): Likewise.
28750 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
28753 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
28755 2017-03-02 Jakub Jelinek <jakub@redhat.com>
28757 PR rtl-optimization/79780
28758 * cprop.c (one_cprop_pass): When second and further conditional trap
28759 in a single basic block is turned into an unconditional trap, turn it
28760 into a deleted note to avoid RTL verification failures.
28762 2017-03-02 Richard Biener <rguenther@suse.de>
28764 * fold-const.c (const_binop): Use ulow () instead of elt (0).
28766 2017-03-02 Richard Biener <rguenther@suse.de>
28768 PR tree-optimization/79345
28770 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
28771 param and abort the walk, returning -1 if it is hit.
28772 (walk_aliased_vdefs): Take a limit param and pass it on.
28773 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
28774 defaulting to 0 and return a signed int.
28775 * tree-ssa-uninit.c (struct check_defs_data): New struct.
28776 (check_defs): New helper.
28777 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
28778 about uninitialized memory.
28779 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
28780 bogus uninitialized warning.
28781 (fixed_convert_from_real): Likewise.
28783 2017-03-02 Bin Cheng <bin.cheng@arm.com>
28785 PR tree-optimization/66768
28786 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
28787 iv_use if base object can't be determined.
28789 2017-03-02 Jakub Jelinek <jakub@redhat.com>
28791 PR tree-optimization/79345
28792 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
28793 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
28794 (get_pattern_stats): Initialize it.
28795 * genemit.c (gen_expand): Verify match_scratch numbers come after
28796 match_operand/match_dup numbers.
28797 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
28798 match_scratch numbers.
28799 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
28801 * config/s390/s390.md (trunctdsd2): Likewise.
28803 2017-03-02 Richard Biener <rguenther@suse.de>
28805 * wide-int.h (wide_int_storage::operator=): Implement in terms
28808 2017-03-02 Richard Biener <rguenther@suse.de>
28810 PR tree-optimization/79777
28811 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
28812 the to insert expression to sth existing.
28814 2017-03-01 Martin Sebor <msebor@redhat.com>
28816 PR middle-end/79692
28817 * gimple-ssa-sprintf.c
28818 (directive::known_width_and_precision): New function.
28819 (format_integer): Use it.
28820 (get_mpfr_format_length): Consider the full range of precision
28821 when computing %g output with the # flag. Set the likely byte
28822 count to 3 rather than 1 when precision is indeterminate.
28823 (format_floating): Correct the lower bound of precision.
28825 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28827 * doc/invoke.texi: Document default code model for 64-bit Linux.
28829 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28832 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
28833 udiv rather than div since input pattern is unsigned.
28835 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
28837 * config/i386/i386.c (print_reg): Warn for values of
28838 unsupported size in integer register.
28840 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
28843 * config/rs6000/predicates.md (current_file_function_operand): Do
28844 not allow self calls to be local if the function is replaceable.
28846 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
28849 * config/rs6000/altivec.h (vec_ctz and others): Change the
28850 preprocessor macro that controls conditional compilation from
28851 _ARCH_PWR9 to __POWER9_VECTOR__.
28852 (vec_all_ne): Change parameterization of __altivec_scalar_pred
28853 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
28854 control (instead of _ARCH_PWR9 control) so that template
28855 definition uses power9-specific function.
28856 (vec_any_eq): Likewise.
28857 (vec_all_ne): Change macro definition to use a power9-specific
28858 expansion under #ifdef __POWER9_VECTOR__ control (instead of
28859 _ARCH_PWR9 control).
28860 (vec_any_eq) Likewise.
28861 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
28862 expansion for CMPNEF to remove support for xvcmpnesp instruction.
28863 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
28864 support for xvcmpnedp instruction.
28865 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
28866 macro expansion so that Power9 implementation of vec_all_ne does
28867 not use the AltiVec predicate framework.
28868 (VCMPNEH_P): Likewise.
28869 (VCMPNEW_P): Likewise.
28870 (VCMPNED_P): Likewise.
28871 (VCMPNEFP_P): Likewise.
28872 (VCMPNEDP_P): Likewise.
28873 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
28874 implementation of vec_any_eq to not use AltiVec predicate
28876 (VCMPAEH_P): Likewise.
28877 (VCMPAEW_P): Likewise.
28878 (VCMPAED_P): Likewise.
28879 (VCMPAEFP_P): Likewise.
28880 (VCMPAEDP_P): Likewise.
28881 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
28882 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
28883 not use the AltiVec predicate framework.
28884 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
28885 of vec_any_eq to not use AltiVec predicate framework.
28886 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
28887 support for predefined __POWER9_VECTOR__ macro to indicate that
28888 Power9 instruction selection is enabled.
28889 (altivec_overloaded_builtins): Remove extraneous
28890 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
28891 function argument types RS6000_BTI_bool_V16QI and
28892 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
28893 entry for overloaded function argument types RS6000_BTI_bool_V4SI
28894 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
28895 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
28896 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
28897 Power9 for implementations of vec_cmpne. Change the signature for
28898 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
28899 (representing vec_all_ne) to remove the previously described first
28900 argument of type RS6000_BTI_INTSI, as this was an artifact of
28901 reliance on the AltiVec predicate framework, which is no longer
28902 used in the implementation of these functions. Add
28903 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
28904 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
28905 since, unlike the AltiVec predicate framework implementation, we
28906 do not share function descriptors between vec_alle and vec_anyeq.
28907 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
28908 set of modes that receive special treatment even when
28909 TARGET_P9_VECTOR is true. The special treatment emits code that
28910 does not depend on Power9 instructions.
28911 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
28912 define_expand to not rely on AltiVec predicate framework.
28913 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
28915 (vector_ne_v2di_p): Change this define_expand to not rely on
28916 AltiVec predicate framework.
28917 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
28919 (vector_ne_<mode>_p): Change this define_expand to not rely on
28920 AltiVec predicate framework.
28921 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
28923 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
28924 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
28925 define_insn pattern.
28926 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
28927 define_insn pattern because the xvcmpne<VSs>. instruction is not
28929 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
28930 instruction is not supported.
28932 2017-03-01 Jakub Jelinek <jakub@redhat.com>
28934 * config/nvptx/nvptx.c: Include intl.h.
28936 2017-03-01 Martin Jambor <mjambor@suse.cz>
28939 * ipa-prop.h (ipa_bits): Removed field known.
28940 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
28941 to pointers. Adjusted their comments to warn about their sharing.
28942 (ipcp_transformation_summary): Change bits to a vector of pointers.
28943 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
28944 (ipa_get_ipa_bits_for_value): Declare.
28945 * tree-vrp.h (value_range): Mark as GTY((for_user)).
28946 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
28947 (ipa_bits_hash_table): Likewise.
28948 (ipa_vr_ggc_hash_traits): Likewise.
28949 (ipa_vr_hash_table): Likewise.
28950 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
28951 being pointers and vr_known being removed.
28952 (ipa_set_jf_unknown): Likewise.
28953 (ipa_get_ipa_bits_for_value): New function.
28954 (ipa_set_jfunc_bits): Likewise.
28955 (ipa_get_value_range): New overloaded functions.
28956 (ipa_set_jfunc_vr): Likewise.
28957 (ipa_compute_jump_functions_for_edge): Use the above functions to
28958 construct bits and vr parts of jump functions.
28959 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
28960 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
28962 (ipcp_grow_transformations_if_necessary): Also allocate
28963 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
28965 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
28966 them. Fix too long lines.
28967 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
28968 vr_known being removed.
28969 (ipa_read_jump_function): Use new setter functions to construct bits
28970 and vr parts of jump functions or set them to NULL.
28971 (write_ipcp_transformation_info): Adjust for bits being pointers.
28972 (read_ipcp_transformation_info): Likewise.
28973 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
28975 Include gt-ipa-prop.h.
28976 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
28978 (ipcp_store_bits_results): Likewise.
28979 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
28980 Do not write to existing jump functions but use a temporary instead.
28982 2017-03-01 Jakub Jelinek <jakub@redhat.com>
28985 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
28986 attempt to use its first operand as BIT_FIELD_REF base.
28988 2017-03-01 Richard Biener <rguenther@suse.de>
28990 PR middle-end/79721
28991 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
28992 interpolating formula in wrapping arithmetic.
28993 (chrec_apply): Convert chrec_evaluate return value to wanted type.
28995 2017-03-01 Jakub Jelinek <jakub@redhat.com>
28997 PR tree-optimization/79734
28998 * tree-vect-generic.c (expand_vector_condition): Optimize
28999 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
29000 Handle VEC_COND_EXPR where comparison has different inner width from
29001 type's inner width.
29003 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
29005 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
29006 markup, and similar issues. Remove @opindex entries for things
29007 that aren't options. Add missing -mmpy-option entries.
29009 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29011 PR tree-optimization/79737
29012 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
29013 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
29014 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
29015 instead of byte_size. Formatting fix.
29016 (shift_bytes_in_array_right): Formatting fix.
29018 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
29021 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
29022 condition on optimize for the leaf function test.
29024 2017-02-28 Martin Liska <mliska@suse.cz>
29027 * read-rtl-function.c (function_reader::handle_unknown_directive):
29028 Bail out when one uses -flto.
29030 2017-02-28 Martin Liska <mliska@suse.cz>
29032 * common.opt: Replace space with tabular for options of <number>
29034 * config/i386/i386.opt: Show <number> value for
29035 -mlarge-data-threshold.
29036 * opts.c (print_filtered_help): Do not display number in hexadecimal
29039 2017-02-28 Martin Liska <mliska@suse.cz>
29041 * common.opt: Fix --help=option -Q for options which are of
29044 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
29046 * config/i386/i386.c (print_reg): Error out for values
29047 of 8-bit size in invalid integer register.
29049 2017-02-28 Martin Sebor <msebor@redhat.com>
29051 PR tree-optimization/79691
29052 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
29054 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29057 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
29058 gcc_unreachable with output_operand_lossage.
29060 2017-02-28 Richard Biener <rguenther@suse.de>
29062 PR tree-optimization/79740
29063 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
29065 (visit_nary_op): Insert the nary into the hashtable if we
29066 pattern-matched sth.
29067 * tree-ssa-pre.c (eliminate_insert): Robustify.
29069 2017-02-28 Richard Biener <rguenther@suse.de>
29071 PR middle-end/79731
29072 * fold-const.c (decode_field_reference): Reject out-of-bound
29075 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29077 * config/i386/i386.c: Include intl.h.
29078 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
29079 instead of just cond ? "..." : "...".
29080 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
29081 * coverage.c (read_counts_file): Likewise.
29082 * omp-offload.c: Include intl.h.
29083 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
29084 of just cond ? "..." : "...".
29085 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
29086 of just cond ? "..." : "...".
29088 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
29091 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
29093 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
29094 'tune for' CPU name.
29095 * config/arm/arm-cpu-data.h: Regenerated.
29097 2017-02-28 Richard Biener <rguenther@suse.de>
29099 PR tree-optimization/79732
29100 * tree-inline.c (expand_call_inline): Do not shadow var.
29102 2017-02-28 Richard Biener <rguenther@suse.de>
29104 PR tree-optimization/79723
29105 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
29106 address-space properly.
29108 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
29110 * doc/optinfo.texi (Optimization groups): Fix option used for
29112 * doc/invoke.texi (-fopt-info): Document "omp".
29113 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
29114 (OPTGROUP_ALL): Add OPTGROUP_OMP.
29115 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
29116 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
29117 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
29119 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
29121 * dumpfile.c (optgroup_options): Instead of "openmp", associate
29122 OPTGROUP_OMP with "omp".
29124 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
29127 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
29128 for arithmetic shift of unsigned V2DI.
29130 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
29132 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
29133 arc/linux.h headers.
29134 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
29135 (LINK_SPEC): Likewise.
29136 (ARC_TLS_EXTRA_START_SPEC): Likewise.
29137 (EXTRA_SPECS): Likewise.
29138 (STARTFILE_SPEC): Likewise.
29139 (ENDFILE_SPEC): Likewise.
29140 (LIB_SPEC): Likewise.
29141 (TARGET_SDATA_DEFAULT): Likewise.
29142 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
29143 (MULTILIB_DEFAULTS): Likewise.
29144 (DWARF2_UNWIND_INFO): Likewise.
29145 * config/arc/big.h: New file.
29146 * config/arc/elf.h: Likewise.
29147 * config/arc/linux.h: Likewise.
29148 * config/arc/t-uClibc: Remove.
29150 2017-02-27 Bin Cheng <bin.cheng@arm.com>
29152 PR tree-optimization/77536
29153 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
29154 (tree_transform_and_unroll_loop): Use above function to compute the
29155 estimated niter of unrolled loop and use it when scaling profile.
29156 Also use count info rather than frequency if it's non-zero.
29157 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
29158 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
29159 (vect_transform_loop): Call above function.
29161 2017-02-27 Richard Biener <rguenther@suse.de>
29163 PR tree-optimization/45397
29164 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
29165 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
29166 (visit_nary_op): Add pattern matching for CSEing sign-changed
29167 or truncated operations with wider ones.
29169 2017-02-27 Richard Biener <rguenther@suse.de>
29171 PR tree-optimization/79690
29172 * tree-vect-stmts.c (vectorizable_store): Use vector type
29173 built from the DR with address-space.
29175 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
29177 * doc/invoke.texi (Optimize Options): Refine the description
29178 of asan-use-after-return.
29180 2017-02-25 Alan Modra <amodra@gmail.com>
29182 PR rtl-optimization/79584
29183 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
29184 base, not ad->base_term, the reg within base. Remove assertion
29185 that ad->base == ad->base_term. Replace gen_int_mode using
29186 bogus mode with const0_rtx.
29188 2017-02-25 Jakub Jelinek <jakub@redhat.com>
29190 PR middle-end/79396
29191 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
29192 FMA_EXPR like tcc_binary or tcc_unary.
29194 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
29197 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
29199 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
29200 (output_loc_operands): Handle DW_OP_call_ref and
29201 DW_OP_GNU_variable_value.
29202 (struct variable_value_struct): New type.
29203 (struct variable_value_hasher): Likewise.
29204 (variable_value_hash): New variable.
29205 (string_types): Remove.
29206 (copy_loc_descr): New function.
29207 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
29208 (prepend_loc_descr_to_each): New function.
29209 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
29210 instead of add_loc_descr_to_each if the first argument is single
29211 location list and the second has multiple.
29212 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
29213 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
29214 when looking for variable value which doesn't have other location info.
29215 (loc_list_from_tree): Formatting fix.
29216 (gen_array_type_die): Simplify DW_AT_string_length handling.
29217 (adjust_string_types): Remove.
29218 (gen_subprogram_die): Don't call adjust_string_types nor test/set
29219 string_types. Call resolve_variable_values.
29220 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
29221 (resolve_addr_in_expr): Likewise. Add A argument.
29222 (copy_deref_exprloc): Remove deref argument. Adjust for the
29223 original expression being DW_OP_GNU_variable_value with optionally
29224 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
29225 optionally after it.
29226 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
29227 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
29228 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
29229 (variable_value_hasher::hash, variable_value_hasher::equal): New
29231 (resolve_variable_value_in_expr, resolve_variable_value,
29232 resolve_variable_values, note_variable_value_in_expr,
29233 note_variable_value): New functions.
29234 (dwarf2out_early_finish): Call note_variable_value on all toplevel
29237 2017-02-24 Jakub Jelinek <jakub@redhat.com>
29240 * opts.h (handle_generated_option): Add GENERATED_P argument.
29241 * opts-common.c (handle_option): Adjust function comment.
29242 (handle_generated_option): Add GENERATED_P argument, pass it to
29244 (control_warning_option): Pass false to handle_generated_option
29246 * opts.c (maybe_default_option): Pass true to handle_generated_option
29248 * optc-gen.awk: Likewise.
29250 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29252 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
29253 a REG, look at the REG it is a SUBREG of.
29254 (splitter for cmpeqsi_t): Ditto.
29256 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29258 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
29259 the special USEs with the pattern of the insn, not the insn itself.
29261 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
29264 * doc/invoke.texi: Document -mload-store-pairs.
29266 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29267 Sandra Loosemore <sandra@codesourcery.com>
29269 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
29270 argument isn't a CONST_INT.
29271 (nios2_alternate_compare_const): Assert op is a CONST_INT.
29272 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
29273 (nios2_validate_compare): Bypass alternate compare logic if *op2
29274 is not a CONST_INT.
29275 (ldstwm_operation_p): Return false if first_base is not a REG or
29276 if first_offset is not a CONST_INT.
29278 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29280 * config/cris/cris.md: Use correct operand in a define_peephole2.
29282 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29284 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
29286 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29288 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
29289 this_insn if it is an INSN or JUMP_INSN.
29290 (force_offsettable): Look at base, not at addr.
29291 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
29292 on things that aren't necessarily CONST_INTs.
29294 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
29296 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
29297 -mfpmath=sse is the default also for x86-32 targets with SSE2
29298 instruction set when @option{-ffast-math} is enabled
29300 2017-02-24 Jeff Law <law@redhat.com>
29302 PR rtl-optimizatoin/79286
29303 * ira.c (update_equiv_regs): Drop may_trap_p exception to
29306 2017-02-24 Richard Biener <rguenther@suse.de>
29308 PR tree-optimization/79389
29309 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
29312 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
29314 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
29315 function comment to reflect reality.
29316 (loop_exits_before_overflow): Fix typo in function description.
29318 2017-02-24 Richard Biener <rguenther@suse.de>
29320 PR tree-optimization/79389
29321 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
29322 properly that a threading opportunity exists. Detect conditional
29323 copy/constant propagation opportunities.
29325 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
29327 * config/visium/visium.md (type): Add trap.
29328 (b): New mode attribute.
29329 (*btst): Rename into...
29330 (*btst<mode>): ...this and adjust.
29331 (*cbranchsi4_btst_insn): Rename into...
29332 (*cbranch<mode>4_btst_insn): ...this and adjust.
29333 (trap): New define_insn.
29335 2017-02-23 Jakub Jelinek <jakub@redhat.com>
29337 PR tree-optimization/79389
29338 * ifcvt.c (struct noce_if_info): Add rev_cond field.
29339 (noce_reversed_cond_code): New function.
29340 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
29341 reversed_comparison_code. Formatting fix.
29342 (noce_try_store_flag): Test rev_cond != NULL in addition to
29343 reversed_comparison_code.
29344 (noce_try_store_flag_constants): Likewise.
29345 (noce_try_store_flag_mask): Likewise.
29346 (noce_try_addcc): Use rev_cond if non-NULL instead of
29347 reversed_comparison_code.
29348 (noce_try_cmove_arith): Likewise. Formatting fixes.
29349 (noce_try_minmax, noce_try_abs): Clear rev_cond.
29350 (noce_find_if_block): Initialize rev_cond.
29351 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
29352 instead of false as last argument never attempt to reverse it
29355 2017-02-23 Bin Cheng <bin.cheng@arm.com>
29357 PR tree-optimization/79663
29358 * tree-predcom.c (combine_chains): Process refs in reverse order
29359 only for ZERO length chains, and add explaining comment.
29361 2017-02-23 Jeff Law <law@redhat.com>
29363 PR tree-optimization/79578
29364 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
29365 in call to operand_equal_p.
29367 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
29370 * config/i386/cpuid.h: Fix another undefined behavior.
29372 2017-02-23 Richard Biener <rguenther@suse.de>
29374 PR tree-optimization/79683
29375 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
29376 vector types for data-refs.
29378 2017-02-23 Martin Liska <mliska@suse.cz>
29380 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
29382 2017-02-23 Jakub Jelinek <jakub@redhat.com>
29384 PR middle-end/79665
29385 * internal-fn.c (get_range_pos_neg): Moved to ...
29386 * tree.c (get_range_pos_neg): ... here. No longer static.
29387 * tree.h (get_range_pos_neg): New prototype.
29388 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
29389 are known to be in between 0 and signed maximum inclusive, try to
29390 expand both unsigned and signed divmod and use the cheaper one from
29393 2017-02-22 Jeff Law <law@redhat.com>
29395 PR tree-optimization/79578
29396 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
29397 to compare base operands.
29399 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
29402 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
29403 gpc_reg_operand instead of fpr_reg_operand.
29405 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
29407 * config/mips/mips.c (mips_return_in_memory): Force FP
29408 vector types to be returned in memory for o32 ABI.
29410 2017-02-22 Jakub Jelinek <jakub@redhat.com>
29412 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
29413 instead of DW_TAG_member for static data member declarations and don't
29414 set no_linkage_name for static inline data members.
29415 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
29418 2017-02-22 Martin Liska <mliska@suse.cz>
29420 * doc/invoke.texi: Replace inequality signs with square brackets
29423 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
29426 * lra-constraints.c (simplify_operand_subreg): Handle
29427 WORD_REGISTER_OPERATIONS targets.
29429 2017-02-22 Jakub Jelinek <jakub@redhat.com>
29432 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
29433 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
29434 elimination by swapping fld*.
29436 2017-02-22 Richard Biener <rguenther@suse.de>
29438 PR tree-optimization/79673
29439 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
29440 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
29441 irrelevant address-space qualifiers and avoiding a
29442 ADDR_SPACE_CONVERT_EXPR from fold_convert.
29444 2017-02-22 Richard Biener <rguenther@suse.de>
29446 PR tree-optimization/79666
29447 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
29448 to not symbolically negate if that may introduce undefined
29451 2017-02-22 Martin Liska <mliska@suse.cz>
29454 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
29455 * data-streamer-out.c (streamer_write_gcov_count_stream):
29457 * value-prof.c (stream_out_histogram_value): Make assert more
29458 precise based on type of counter.
29460 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
29463 * config/i386/i386.md (standard_x87sse_constant_load splitter):
29464 Use nonimmediate_operand instead of memory_operand for operand 1.
29465 (float-extend standard_x87sse_constant_load splitter): Ditto.
29467 2017-02-21 Jeff Law <law@redhat.com>
29469 PR tree-optimization/79621
29470 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
29471 blocks with edges to themselves.
29473 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29476 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
29477 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
29478 Use gimple_call_builtin_p.
29481 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
29482 on temporarily removed DEBUG_INSNs.
29484 PR tree-optimization/79649
29485 * tree-loop-distribution.c (classify_partition): Give up on
29486 non-generic address space loads/stores.
29488 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
29490 * doc/loop.texi (Loop manipulation): Remove nonexistent
29491 tree_ssa_loop_version from the documentation.
29492 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
29494 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29497 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
29498 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
29499 * config/rs6000/rs6000.c: Include except.h.
29500 (rs6000_expand_split_stack_prologue): Call
29501 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
29503 2017-02-21 Martin Jambor <mjambor@suse.cz>
29506 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
29507 have been analyzed.
29509 2017-02-21 Martin Jambor <mjambor@suse.cz>
29511 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
29512 for backward compatibility only.
29513 * doc/invoke.texi (Option Summary): Remove all references to
29514 -fipa-cp-alignment.
29516 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
29520 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29522 * lra-constraints.c (curr_insn_transform): Handle
29523 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
29525 2017-02-21 Martin Liska <mliska@suse.cz>
29527 * config/i386/i386.opt: Replace -masm-dialect with -masm.
29529 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
29531 PR translation/79638
29532 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
29534 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
29537 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
29538 (arm_function_ok_for_sibcall): Return false for an indirect call by
29539 descriptor if all the argument registers are used.
29540 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
29541 alignment of the function.
29543 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29545 PR tree-optimization/61441
29546 * simplify-rtx.c (simplify_const_unary_operation): For
29547 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
29548 the sNaN unmodified.
29550 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
29552 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
29553 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
29554 instead of SYSTEM_HEADER_DIR.
29556 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
29557 Martin Liška <mliska@suse.cz>
29559 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
29560 Fix typos and grammar, use active voice, and clarify.
29562 2017-02-20 Marek Polacek <polacek@redhat.com>
29564 PR middle-end/79537
29565 * gimplify.c (gimplify_expr): Handle unused *&&L;.
29568 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
29570 2017-02-20 Jakub Jelinek <jakub@redhat.com>
29573 * config/i386/i386.c (ix86_expand_builtin): Handle
29574 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
29575 ix86_builtins_isa[fcode].isa as a requirement of those
29576 flags and any other flag in the bitmask.
29577 (ix86_init_mmx_sse_builtins): Use 0 instead of
29578 ~OPTION_MASK_ISA_64BIT as mask.
29579 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
29580 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
29581 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
29582 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
29584 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29587 * lra-constraints.c (split_reg): Check requested split mode
29588 is supported by the register.
29590 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29592 * lra-constraints.c (simplify_operand_subreg): Remove early
29595 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29598 * lra-constraints.c (curr_insn_transform): Tighten condition
29599 for converting SUBREG reloads from OP_OUT to OP_INOUT.
29601 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29604 * lra-constraints.c (curr_insn_transform): Handle
29605 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
29607 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
29610 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
29612 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
29614 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
29617 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
29620 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29622 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
29625 2017-02-18 Jakub Jelinek <jakub@redhat.com>
29627 * final.c (last_columnnum, override_columnnum): New variables.
29628 (final_start_function): Set last_columnnum, pass it to begin_prologue
29629 hook and pass 0 to dwarf2out_begin_prologue.
29630 (final_scan_insn): Update override_columnnum. Pass last_columnnum
29631 to source_line debug hook.
29632 (notice_source_line): Compute last_columnnum and for debug_column_info
29633 return true on column changes.
29634 * debug.h (struct gcc_debug_hooks): Add column argument to
29635 source_line and begin_prologue hooks.
29636 (debug_nothing_int_charstar_int_bool): Remove prototype.
29637 (debug_nothing_int_int_charstar,
29638 debug_nothing_int_int_charstar_int_bool): New prototypes.
29639 (dwarf2out_begin_prologue): Add column argument.
29640 * debug.c (do_nothing_debug_hooks): Adjust source_line and
29641 begin_prologue hooks.
29642 (debug_nothing_int_charstar_int_bool): Remove.
29643 (debug_nothing_int_int_charstar,
29644 debug_nothing_int_int_charstar_int_bool): New functions.
29645 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
29646 through to dwarf2out_source_line.
29647 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
29648 (dwarf2out_source_line): Add column argument, emit it if requested.
29649 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
29651 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
29652 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
29653 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
29654 through to dwarf2out_begin_prologue.
29655 (vmsdbgout_source_line): Add column argument, pass it through to
29656 dwarf2out_source_line.
29657 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
29658 dbxout_source_line caller.
29659 (dbxout_source_line): Add column argument.
29661 * common.opt (gno-column-info, gcolumn-info): New options.
29662 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
29663 (check_die): Also test for multiple DW_AT_decl_column attributes.
29664 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
29665 DW_AT_decl_column if requested.
29666 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
29668 (gen_variable_die): Likewise.
29669 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
29670 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
29673 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
29674 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
29675 (ix86_handle_option): Handle OPT_m3dnowa.
29676 * doc/invoke.texi (-m3dnowa): Document.
29677 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
29678 -m3dnowa instead of -m3dnow -march=athlon.
29681 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
29682 instead of gcc_assert for K, r and R code checks. Formatting fixes.
29684 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29687 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
29688 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
29689 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
29690 generator for vsx_xxpermdi_<mode>_be.
29691 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
29692 force big-endian semantics.
29693 (vsx_xxpermdi_<mode>_be): New define_expand with same
29694 implementation as previous version of vsx_xxpermdi_<mode>.
29696 2017-02-17 Jakub Jelinek <jakub@redhat.com>
29698 PR tree-optimization/79327
29699 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
29700 variable, its initialization and use.
29702 2017-02-17 Julia Koval <julia.koval@intel.com>
29704 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
29705 (OPTION_MASK_ISA_PKU_UNSET): New.
29706 (ix86_handle_option): Handle -mrdpid.
29707 * config/i386/cpuid.h (bit_RDPID): New.
29708 * config/i386/driver-i386.c (host_detect_local_cpu):
29709 Detect RDPID feature.
29710 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
29711 * config/i386/i386-c.c (ix86_target_macros_internal):
29713 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
29714 (ix86_valid_target_attribute_inner_p): Add "rdpid".
29715 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
29716 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
29717 * config/i386/i386.md (define_insn "rdpid"): New.
29718 * config/i386/i386.opt Add -mrdpid.
29719 * config/i386/immintrin.h (_rdpid_u32): New.
29721 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
29723 PR rtl-optimization/79541
29724 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
29725 instead of transforming it into USE.
29727 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
29729 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
29730 If HONOR_SNANS (SFmode) force the input to a register.
29731 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
29732 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
29733 an frsp or similar insn.
29735 2017-02-17 Martin Liska <mliska@suse.cz>
29737 PR rtl-optimization/79577
29738 * params.def (selsched-max-sched-times): Increase minimum to 1.
29740 2017-02-17 Martin Liska <mliska@suse.cz>
29742 PR rtl-optimization/79574
29743 * gcse.c (want_to_gcse_p): Prevent integer overflow.
29745 2017-02-17 Martin Liska <mliska@suse.cz>
29747 PR tree-optimization/79529
29748 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
29749 ssa_defined_default_def_p to handle cases which are implicitly
29751 * tree-ssa.c (ssa_defined_default_def_p): New function.
29752 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
29753 which are implicitly defined.
29754 * tree-ssa.h (ssa_defined_default_def_p): Declare.
29756 2017-02-17 Richard Biener <rguenther@suse.de>
29758 PR middle-end/79576
29759 * params.def (max-ssa-name-query-depth): Limit to 10.
29761 2017-02-17 Richard Biener <rguenther@suse.de>
29763 PR tree-optimization/79552
29764 * tree-ssa-structalias.c (visit_loadstore): Properly verify
29767 2017-02-17 Richard Biener <rguenther@suse.de>
29770 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
29772 2017-02-17 Marek Polacek <polacek@redhat.com>
29774 PR middle-end/79536
29775 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
29776 (fold_negate_expr): New wrapper.
29778 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
29780 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
29781 Correct terminology and de-emphasize pre-standard behavior.
29783 2017-02-16 Alan Modra <amodra@gmail.com>
29785 PR rtl-optimization/79286
29786 * ira.c (def_dominates_uses): New function.
29787 (update_equiv_regs): Don't create an equivalence for insns that
29788 may trap where the register def does not dominate the use.
29790 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
29792 PR rtl-optimization/78127
29793 * lra.c (lra): Call lra_eliminate before finish the loop after
29796 2017-02-16 Richard Biener <rguenther@suse.de>
29798 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
29800 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
29801 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
29802 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
29803 (isl_val_int_from_wi): New function.
29804 (extract_affine_gmp): Rename to ...
29805 (extract_affine_wi): ... this, take a widest_int.
29806 (extract_affine_int): Just wrap extract_affine_wi.
29807 (add_param_constraints): Use isl_val_int_from_wi.
29808 (add_loop_constraints): Likewise, and extract_affine_wi.
29810 2017-02-15 Jeff Law <law@redhat.com>
29812 PR middle-end/79521
29813 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
29814 ira_init_register_move_cost_if_necessary.
29816 2017-02-15 Martin Sebor <msebor@redhat.com>
29818 PR middle-end/32003
29819 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
29820 removed in a prior commit.
29822 2017-02-15 Bin Cheng <bin.cheng@arm.com>
29824 PR tree-optimization/79347
29825 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
29826 counters during peeling.
29828 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
29830 * Makefile.in (site.exp): Remove "set ISLVER".
29832 2017-02-15 Jakub Jelinek <jakub@redhat.com>
29835 * real.c (real_from_integer): Call real_convert even for decimal.
29837 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
29840 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
29842 2017-02-14 Andrew Pinski <apinski@cavium.com>
29844 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
29845 cores and change the partno/implementer to be correct.
29846 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
29847 the 'B" as the implementer.
29848 * config/aarch64/aarch64-tune.md: Regenerate.
29850 2017-02-14 Carl Love <cel@us.ibm.com>
29852 * config/rs6000/rs6000.c: Add case statement entry to make the
29853 xvcvuxdsp built-in argument unsigned.
29854 * config/rs6000/vsx.md: Fix the source and return operand types so they
29855 match the instruction definitions from the ISA document. Fix typo
29856 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
29859 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
29862 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
29863 member early_clobber_alts.
29864 * lra-lives.c (reg_early_clobber_p): New.
29865 (process_bb_lives): Use it.
29866 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
29867 (debug_operand_data): Initialize early_clobber_alts.
29868 (setup_operand_alternative): Set up early_clobber_alts.
29869 (collect_non_operand_hard_regs): Ditto. Pass early clobber
29870 alternatives to new_insn_reg.
29871 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
29873 (lra_update_insn_regno_info): Pass the new arg.
29875 2017-02-14 Jakub Jelinek <jakub@redhat.com>
29877 PR middle-end/79505
29878 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
29879 (new_oacc_loop_raw): Don't clear already cleared fields.
29882 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
29883 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
29884 _mm512_prefetch_i64gather_ps): New inline functions and macros.
29886 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
29889 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
29891 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
29894 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
29895 the extra instruction to the right place to store 128-bit constant
29898 2017-02-14 Martin Sebor <msebor@redhat.com>
29900 PR middle-end/79448
29901 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
29902 warning for strings of unknown length.
29904 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
29906 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
29908 2017-02-14 Jeff Law <law@redhat.com>
29911 * ira-costs.c (scan_one_insn): Initialize register move costs
29912 for pseudos seen in USE/CLOBBER insns.
29914 PR tree-optimization/79095
29915 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
29916 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
29917 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
29918 if the operands are known to be not equal, then the resulting range
29920 (intersect_ranges): If the new range is ~[0,0] and the old range is
29921 wide, then prefer ~[0,0].
29922 * tree-vrp.c (overflow_comparison_p_1): New function.
29923 (overflow_comparison_p): New function.
29924 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
29925 if NAME is used in an overflow test.
29926 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
29927 overflow check that can be expressed as an equality test, then adjust
29928 ops to be that equality test.
29930 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29932 * config/s390/s390-builtin-types.def: Remove flags argument.
29933 * config/s390/s390.c (s390_init_builtins): Likewise.
29935 2017-02-14 Martin Liska <mliska@suse.cz>
29937 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
29938 vector. Fix trailing white spaces.
29940 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
29942 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
29945 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29947 PR rtl-optimization/68664
29948 * config/arm/arm.c (arm_sched_can_speculate_insn):
29949 New function. Declare prototype.
29950 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
29952 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29954 PR rtl-optimization/68664
29955 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
29957 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
29959 2017-02-14 Amit Pawar <amit.pawar@amd.com>
29961 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
29962 max skip bytes for function, loop and jump.
29964 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29966 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
29967 ABS_EXPR for gimple dump.
29969 2017-02-14 Jakub Jelinek <jakub@redhat.com>
29972 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
29974 PR tree-optimization/79408
29975 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
29976 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
29977 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
29978 also if rhs1 is INTEGER_CST.
29980 2017-02-14 Richard Biener <rguenther@suse.de>
29982 PR middle-end/79432
29983 * tree-into-ssa.c (insert_phi_nodes): When the function can
29984 have abnormal edges rewrite SSA names with broken use-def
29985 dominance out of SSA and register them for PHI insertion.
29987 2017-02-13 Martin Sebor <msebor@redhat.com>
29989 PR middle-end/79496
29990 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
29991 clearing info.nowrite flag when snprintf size argument is a range.
29993 2017-02-13 Jakub Jelinek <jakub@redhat.com>
29995 * cprop.c (cprop_jump): Add missing space in string literal.
29996 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
29997 (get_constraint_for_component_ref): Likewise.
29998 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
29999 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
30000 * lra-constraints.c (process_alt_operands): Likewise.
30001 * ipa-inline.c (inline_small_functions): Likewise.
30002 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
30003 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
30004 * trans-mem.c (diagnose_tm_1_op): Likewise.
30005 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
30006 (grid_parallel_clauses_gridifiable): Likewise.
30008 * config/nvptx/mkoffload.c (process): Add space in between
30011 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
30012 "MOD4_SSE_REGS" and "ALL_REGS".
30014 * spellcheck.c (test_data): Add , in between "foo" and "food".
30016 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30019 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
30020 boundary crossing check and subsequent code generation agree.
30022 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30024 * config/aarch64/aarch64.c (has_memory_op): Delete.
30025 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
30028 2017-02-13 Jakub Jelinek <jakub@redhat.com>
30030 PR rtl-optimization/79388
30031 PR rtl-optimization/79450
30032 * combine.c (distribute_notes): When removing TEM_INSN for which
30033 corresponding dest has last value recorded, invalidate that last
30036 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30038 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
30039 of explicit '@'. Add missing assembly comment marker on branch costs
30042 2017-02-13 Nathan Sidwell <nathan@acm.org>
30044 * gengtype-lex.l (<in_struct>): Add '/'.
30046 2017-02-13 Martin Liska <mliska@suse.cz>
30049 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
30051 2017-02-13 Richard Biener <rguenther@suse.de>
30053 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
30055 * configure: Re-generate.
30056 * config.in: Likewise.
30057 * graphite-dependences.c: Simplify as if
30058 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
30059 * graphite-isl-ast-to-gimple.c: Likewise.
30060 * graphite-optimize-isl.c: Likewise.
30061 * graphite-poly.c: Likewise.
30062 * graphite-sese-to-poly.c: Likewise.
30063 * graphite.h: Likewise.
30064 * toplev.c: Include isl/version.h and use isl_version () for
30065 printing the ISL version.
30066 * doc/install.texi: Update ISL requirement.
30068 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
30070 * doc/standards.texi (Standards): Update reference to
30073 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
30075 * doc/extend.texi (Named Address Spaces): sourceware.org now
30077 * doc/install.texi (Binaries): Ditto.
30080 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
30082 * doc/cpp.texi: Replace "stringify"/"stringification" with C
30083 standard terminology "stringize"/"stringizing" throughout.
30084 * doc/cppinternals.texi: Likewise.
30086 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
30088 * doc/extend.texi: Fix some spelling mistakes and typos.
30089 * doc/invoke.texi: Likewise.
30091 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
30094 * params.def (inline-min-speedup) Change from 10 to 8.
30096 2017-02-11 Jakub Jelinek <jakub@redhat.com>
30098 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
30101 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
30104 * ipa-inline-analysis.c (get_minimal_bb): New function.
30105 (record_modified): Use it.
30106 (remap_edge_change_prob): Handle also ancestor functions.
30108 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
30110 * doc/contrib.texi (Contributors): Remove broken link into
30111 the Mauve CVS repository.
30113 2017-02-11 Jakub Jelinek <jakub@redhat.com>
30115 PR middle-end/79454
30116 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
30117 result computation whenever lhs doesn't have vector mode, not
30118 just when it has BLKmode.
30120 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
30122 * doc/makefile.texi (profiledbootstrap): Refer to the
30123 installation instructions only in textual form.
30125 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30128 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
30130 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
30132 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
30133 (Specific): Update mingw-w64 reference.
30135 (Specific): Remove broken link to Renesas RX processor.
30137 2017-02-10 Richard Biener <rguenther@suse.de>
30139 * toplev.c (process_options): Do not mention obsolete graphite
30140 options when printing sorry message about missing graphite support.
30141 Mention -floop-nest-optimize.
30143 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
30145 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
30146 (vtst_p16): Likewise.
30147 (vtstq_p8): Likewise.
30148 (vtstq_p16): Likewise.
30150 (vtstq_p64): Likewise.
30151 * config/arm/arm_neon.h (vgetq_lane_p64): New.
30152 (vset_lane_p64): New.
30153 (vsetq_lane_p64): New.
30155 2017-02-10 Jakub Jelinek <jakub@redhat.com>
30157 PR tree-optimization/79411
30158 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
30159 stmt operands are SSA_NAMEs used in abnormal phis.
30160 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
30163 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
30166 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
30169 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
30171 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
30173 2017-02-09 Jakub Jelinek <jakub@redhat.com>
30175 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
30179 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
30180 not arbitrary TREE_CONSTANT.
30183 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
30184 "omp declare target link" attribute unless is_global_var.
30185 * omp-offload.c (find_link_var_op): Likewise.
30187 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
30188 Chung-Lin Tang <cltang@codesourcery.com>
30190 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
30192 (gimplify_adjust_omp_clauses): Don't delete TILE.
30193 (gimplify_omp_for): Deal with TILE.
30194 * internal-fn.c (expand_GOACC_TILE): New function.
30195 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
30197 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
30198 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
30200 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
30201 avoid DIV for outermost collapse var.
30202 (expand_oacc_for): Insert tile element loop as needed. Adjust.
30203 Remove out of date comments, fix whitespace.
30204 * omp-general.c (omp_extract_for_data): Deal with tiling.
30205 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
30206 adjust OLF_DIM_BASE value.
30207 (struct omp_for_data): Add tiling field.
30208 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
30209 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
30210 for auto loops. Remove default auto determining, moved to
30211 oacc_loop_fixed_partitions.
30212 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
30213 stmts, add e_mask field.
30214 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
30215 (oacc_thread_numbers): Use oacc_dim_call.
30216 (oacc_xform_tile): New.
30217 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
30218 (finish_oacc_loop): Adjust for ifns vector.
30219 (oacc_loop_discover_walk): Append loop abstraction sites to list,
30220 add case for GOACC_TILE fns.
30221 (oacc_loop_xform_loop): Delete.
30222 (oacc_loop_process): Iterate over call list directly, and add
30223 handling for GOACC_TILE fns.
30224 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
30226 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
30227 vector partitioning to outer loops. Assign 2 partitions to loops
30228 when available. Add TILE handling.
30229 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
30230 (execite_oacc_device_lower): Process GOACC_TILE fns,
30231 ignore unknown specs.
30232 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
30233 * tree.c (omp_clause_num_ops): Adjust TILE ops.
30234 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
30236 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
30238 * configure.ac (ACX_BUGURL): Update.
30239 * configure: Regenerate.
30241 2017-02-09 Richard Biener <rguenther@suse.de>
30243 PR tree-optimization/69823
30244 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
30245 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
30247 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
30249 * config/arc/arc-c.def: Add __NPS400__ definition.
30250 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
30251 (TARGET_NPS400): Define.
30253 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
30255 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
30257 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
30258 pointer, arch_info.
30259 (arc_cpu_types): Fill the arch_info field with a pointer into the
30260 arc_arch_types table.
30261 (arc_selected_cpu): Declare.
30262 * config/arc/arc.c (arc_selected_cpu): Make global.
30263 (arc_selected_arch): Delete.
30264 (arc_base_cpu): Delete.
30265 (arc_override_options): Remove references to deleted variables,
30266 update access to arch information.
30267 (ARC_OPT): Update access to arch information.
30268 (ARC_OPTX): Likewise.
30269 * config/arc/arc.h (arc_base_cpu): Remove declaration.
30270 (TARGET_ARC600): Update access to arch information.
30271 (TARGET_ARC601): Likewise.
30272 (TARGET_ARC700): Likewise.
30273 (TARGET_EM): Likewise.
30274 (TARGET_HS): Likewise.
30275 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
30278 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
30281 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
30282 condition/operands for integer GE/LE/GEU/LEU operations.
30284 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
30286 PR translation/79397
30287 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
30290 2017-02-08 Martin Jambor <mjambor@suse.cz>
30293 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
30294 whether allocation happened.
30295 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
30296 nothing was allocated.
30298 2017-02-08 Jakub Jelinek <jakub@redhat.com>
30300 PR tree-optimization/79408
30301 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
30302 constant, but SSA_NAME with a known integer range, use the minimum
30303 of that range instead of op1 to determine if modulo can be replaced
30304 with its first operand.
30306 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30308 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
30310 2017-02-08 Richard Biener <rguenther@suse.de>
30312 PR tree-optimization/71824
30313 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
30314 Check all loops contained in the merged region.
30316 2017-02-07 Andrew Pinski <apinski@cavium.com>
30318 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
30320 2017-02-07 Andrew Pinski <apinski@cavium.com>
30322 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
30323 (thunderxt88): Likewise.
30324 (thunderxt81): Disable LSE and change v8.1 to v8.
30325 (thunderxt83): Likewise.
30327 2017-02-07 Jakub Jelinek <jakub@redhat.com>
30328 Richard Biener <rguenther@suse.de>
30330 PR middle-end/79399
30331 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
30332 type from int to size_t.
30333 * ira-costs.c (struct_costs_size): Change type from int to size_t.
30335 2017-02-07 Jakub Jelinek <jakub@redhat.com>
30337 PR rtl-optimization/79386
30338 * cprop.c (bypass_conditional_jumps): Initialize
30339 bypass_last_basic_block already before splitting bbs after
30340 unconditional traps...
30341 (bypass_conditional_jumps): ... rather than here.
30344 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
30345 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
30346 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
30347 fix -masm=intel patterns.
30349 2017-02-07 Richard Biener <rguenther@suse.de>
30351 PR tree-optimization/79256
30352 PR middle-end/79278
30353 * builtins.c (get_object_alignment_2): Use min_align_of_type
30354 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
30355 and ADJUST_FIELD_ALIGN.
30357 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
30359 * doc/tm.texi: Regenerate.
30360 * stor-layout.c (layout_decl): Adjust.
30361 (update_alignment_for_field): Likewise.
30362 (place_field): Likewise.
30363 (min_align_of_type): Likewise.
30364 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
30365 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
30366 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
30367 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
30368 * config/frv/frv.c (frv_adjust_field_align): Likewise.
30369 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
30370 * config/i386/i386.c (x86_field_alignment): Likewise.
30371 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
30372 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
30373 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
30374 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
30375 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
30376 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
30380 2017-01-30 Richard Biener <rguenther@suse.de>
30382 PR tree-optimization/79256
30383 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
30384 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
30387 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
30389 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
30390 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
30391 builtins to SImode and emit a zero-extend, if necessary.
30393 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30395 * docs/invoke.texi (RISC-V Options): Alphabetize.
30397 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30399 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
30402 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30404 * config/riscv/riscv.c: New file.
30405 * common/config/riscv/riscv-common.c: Likewise.
30406 * config.gcc: Likewise.
30407 * config/riscv/constraints.md: Likewise.
30408 * config/riscv/elf.h: Likewise.
30409 * config/riscv/generic.md: Likewise.
30410 * config/riscv/linux.h: Likewise.
30411 * config/riscv/multilib-generator: Likewise.
30412 * config/riscv/peephole.md: Likewise.
30413 * config/riscv/pic.md: Likewise.
30414 * config/riscv/predicates.md: Likewise.
30415 * config/riscv/riscv-builtins.c: Likewise.
30416 * config/riscv/riscv-c.c: Likewise.
30417 * config/riscv/riscv-ftypes.def: Likewise.
30418 * config/riscv/riscv-modes.def: Likewise.
30419 * config/riscv/riscv-opts.h: Likewise.
30420 * config/riscv/riscv-protos.h: Likewise.
30421 * config/riscv/riscv.h: Likewise.
30422 * config/riscv/riscv.md: Likewise.
30423 * config/riscv/riscv.opt: Likewise.
30424 * config/riscv/sync.md: Likewise.
30425 * config/riscv/t-elf-multilib: Likewise.
30426 * config/riscv/t-linux: Likewise.
30427 * config/riscv/t-linux-multilib: Likewise.
30428 * config/riscv/t-riscv: Likewise.
30429 * configure.ac: Likewise.
30430 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
30431 Waterman as RISC-V maintainers.
30432 * doc/install.texi: Add RISC-V entries.
30433 * doc/invoke.texi: Add RISC-V options section.
30434 * doc/md.texi: Add RISC-V constraints section.
30435 * configure: Regenerated.
30437 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
30440 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
30441 false values to be constant vectors with all 0 or all 1 bits set.
30442 (vcondu<mode><mode>): Likewise.
30443 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
30445 (fpmask_comparison_operator): Update comment.
30446 (vecint_comparison_operator): New predicate.
30447 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
30448 vector conditionals when the true and false values are constant
30449 vectors with all 0 bits or all 1 bits set.
30451 2017-02-06 Martin Sebor <msebor@redhat.com>
30453 PR tree-optimization/79376
30454 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
30456 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
30458 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
30459 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
30460 to simplify split condition.
30462 2017-02-06 Jakub Jelinek <jakub@redhat.com>
30464 * omp-expand.c (oxpand_omp_atomic_fetch_op,
30465 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
30468 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
30470 PR rtl-optimization/68664
30471 * target.def (can_speculate_insn): New hook.
30472 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
30473 * doc/tm.texi: Regenerate.
30474 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
30475 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
30476 (rs6000_sched_can_speculate_insn): New function.
30478 2017-02-06 Jakub Jelinek <jakub@redhat.com>
30480 PR tree-optimization/79284
30481 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
30482 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
30483 vectorizable_mask_load_store, vectorizable_operation,
30484 vect_is_simple_cond, get_same_sized_vectype): Use it instead
30485 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
30486 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
30487 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
30488 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
30489 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
30490 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
30491 is_gimple_assign (stmt). Replace another such test with
30492 is_gimple_assign (stmt).
30494 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
30497 * config/avr/avr.c (rtl-iter.h): Include it.
30498 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
30499 (avr_legitimate_combined_insn): ...and implementation.
30501 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
30503 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
30504 * config/s390/s390.c (s390_const_operand_ok)
30505 (s390_canonicalize_comparison, s390_extract_part)
30506 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
30507 (s390_contiguous_bitmask_p, s390_rtx_costs)
30508 (legitimize_pic_address): Likewise.
30509 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
30510 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
30511 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
30512 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
30513 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
30515 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
30517 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
30518 REGNO($0) == REGNO($1).
30520 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30522 * config/s390/linux.h(SIZE_TYPE): Add comment.
30524 2017-02-06 Julian Brown <julian@codesourcery.com>
30525 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30526 Virendra Pathak <virendra.pathak@broadcom.com>
30528 * config/aarch64/aarch64-cores.def: Change the scheduler
30530 * config/aarch64/aarch64.md: Include thunderx2t99.md.
30531 * config/aarch64/thunderx2t99.md: New file.
30533 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30535 * doc/standards.texi (Go Language): Update link to language
30538 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
30540 * tree-eh.c (lower_resx): Sanitize profile.
30541 (cleanup_empty_eh_move_lp): Likewise.
30543 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
30546 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
30548 * cfgloopmanip.h (loop_version): Update prototype.
30549 * modulo-sched.c (sms_schedule): Update call of loop_version.
30550 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
30551 * tree-parloops.c (gen_parallel_loop): Likewise.
30552 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
30553 * tree-ssa-loop-split.c (split_loop): Likewise.
30554 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
30555 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
30557 2017-02-05 Martin Liska <mliska@suse.cz>
30560 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
30562 (print_operand_address): Initialize a struct to zero.
30564 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30566 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
30567 garbage collector only in textual form.
30569 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30571 * doc/extend.texi (x86 specific memory model extensions for
30572 transactional memory): Simplify a phrase.
30574 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
30577 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
30578 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
30579 (atomic_storedi_1): Likewise.
30581 2017-02-04 Jakub Jelinek <jakub@redhat.com>
30583 PR tree-optimization/79338
30584 * tree-parloops.c (gather_scalar_reductions): Don't call
30585 vect_analyze_loop_form for loop->inner before destroying loop's
30588 2017-02-03 Martin Sebor <msebor@redhat.com>
30590 PR tree-optimization/79327
30591 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
30592 when precision has resulted in leading zeros.
30593 (format_integer): Adjust the likely counter to assume an unknown
30594 argument that may be zero is non-zero.
30596 2017-02-03 Jason Merrill <jason@redhat.com>
30599 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
30600 avoid copying non-taken branch.
30602 2017-02-03 Jakub Jelinek <jakub@redhat.com>
30604 PR tree-optimization/79340
30605 * tree-vect-loop.c (vectorizable_reduction): Release
30606 vec_defs elements after safe_splicing them into other vectors.
30609 PR tree-optimization/79327
30610 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
30611 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
30613 (format_integer): Use wide_int_to_tree instead of build_int_cst
30614 + to_?hwi. If argmin is NULL, just set argmin and argmax to
30615 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
30616 of shortest and longest sequence.
30618 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
30620 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
30621 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
30623 2017-02-03 Walter Lee <walt@tilera.com>
30626 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
30627 after initial stackframe link reg save.
30628 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
30630 2017-02-03 Jakub Jelinek <jakub@redhat.com>
30633 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
30634 wu for stxssp alternative.
30636 2017-02-03 Martin Sebor <msebor@redhat.com>
30638 PR tree-optimization/79352
30639 * gimple-fold.c (get_range_strlen): Add argument.
30640 (get_range_strlen): Change return type to bool.
30641 (get_maxval_strlen): Pass in a dummy argument.
30642 * gimple-fold.h (get_range_strlen): Change return type to bool.
30643 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
30644 * tree.h (array_at_struct_end_p): Add argument.
30645 * tree.c (array_at_struct_end_p): Handle it.
30647 2017-02-03 Martin Liska <mliska@suse.cz>
30650 * multiple_target.c (create_dispatcher_calls): Redirect edge
30651 from a caller of a dispatcher.
30652 (expand_target_clones): Make the clones local.
30653 (ipa_target_clone): Do both target clones and resolvers.
30654 (ipa_dispatcher_calls): Remove the pass.
30655 (pass_dispatcher_calls::gate): Likewise.
30656 (make_pass_dispatcher_calls): Likewise.
30657 * passes.def (pass_target_clone): Put as very first IPA early
30660 2017-02-03 Martin Liska <mliska@suse.cz>
30662 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
30663 in case of a function with ifunc attribute.
30665 2017-02-03 Martin Liska <mliska@suse.cz>
30667 * cgraph.c (cgraph_node::dump): Dump function version info.
30668 * symtab.c (symtab_node::dump_base): Add missing new line.
30670 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
30672 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
30673 (ifcombine_ifandif): Use it.
30675 2017-02-03 Martin Liska <mliska@suse.cz>
30677 * doc/invoke.texi: Document default value for
30678 use-after-scope-direct-emission-threshold.
30680 2017-02-03 Martin Liska <mliska@suse.cz>
30682 PR tree-optimization/79339
30683 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
30684 (format_floating): Likewise.
30686 2017-02-03 Martin Liska <mliska@suse.cz>
30689 * ipa-prop.c (ipa_node_params_t::insert): Remove current
30691 (ipa_node_params_t::remove): Likewise.
30692 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
30693 initialization from removed ipa_node_params_t::insert.
30694 (ipa_node_params::~ipa_node_params): Move from removed
30695 ipa_node_params_t::release.
30696 * symbol-summary.h (symbol_summary::m_released): New member.
30697 Do not release a summary twice. Do not allow to call finalizer
30698 for types of a summary that live in GGC memory.
30700 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30702 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
30705 2017-02-02 Martin Sebor <msebor@redhat.com>
30707 PR middle-end/79275
30708 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
30709 (format_string): Tighten up the range of output for non-constant
30710 strings and correct the expected range for wide non-constant strings.
30712 2017-02-02 Martin Sebor <msebor@redhat.com>
30714 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
30716 PR middle-end/32003
30717 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
30719 (-fdump-tree-@var): Add to index and document how to come up
30720 with pass-specific option and dump file names.
30721 (-fdump-passes): Clarify where to look for output.
30723 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
30725 PR middle-end/77445
30726 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
30727 statistics of the analyzed path; allow threading for speed when
30728 any of BBs along the path are optimized for speed.
30730 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
30732 PR middle-end/78468
30733 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
30734 settings of the virtual registers.
30737 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
30739 * explow.c (get_dynamic_stack_size): Take known alignment of stack
30740 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
30743 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30745 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
30746 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
30748 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30750 * config/s390/s390.md: Add missing comments with the expanded
30752 * config/s390/vector.md: Likewise.
30753 * config/s390/vx-builtins.md: Likewise.
30755 2017-02-02 Jakub Jelinek <jakub@redhat.com>
30758 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
30759 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
30760 conditions on a single line.
30762 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30764 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
30765 __S390_VX__ to __VX__.
30767 2017-02-01 Andrew Pinski <apinski@cavium.com>
30769 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
30770 stmt_info to record_stmt_cost.
30771 (vect_get_known_peeling_cost): Pass stmt_info if known to
30773 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
30774 cpu_vector_cost field into
30775 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
30776 field into vec_int_stmt_cost and vec_fp_stmt_cost.
30777 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
30778 splitting of scalar_stmt_cost and vec_stmt_cost.
30779 (thunderx_vector_cost): Likewise.
30780 (cortexa57_vector_cost): LIkewise.
30781 (exynosm1_vector_cost): Likewise.
30782 (xgene1_vector_cost): Likewise.
30783 (thunderx2t99_vector_cost): Improve after the splitting of the two
30785 (aarch64_builtin_vectorization_cost): Update for the splitting of
30786 scalar_stmt_cost and vec_stmt_cost.
30788 2017-02-01 Torvald Riegel <triegel@redhat.com>
30789 Richard Henderson <rth@redhat.com>
30791 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
30792 conditional on existance of a fast atomic load.
30793 * optabs-query.c (can_atomic_load_p): New function.
30794 * optabs-query.h (can_atomic_load_p): Declare it.
30795 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
30796 no fast atomic load is available for the particular size of access.
30797 (expand_atomic_compare_and_swap): Likewise.
30798 (expand_atomic_load): Likewise.
30799 (expand_atomic_store): Likewise.
30800 (expand_atomic_fetch_op): Likewise.
30801 * testsuite/lib/target-supports.exp
30802 (check_effective_target_sync_int_128): Remove x86 because it provides
30803 no fast atomic load.
30804 (check_effective_target_sync_int_128_runtime): Likewise.
30806 2017-02-01 Richard Biener <rguenther@suse.de>
30808 * graphite.c: Include tree-vectorizer.h for find_loop_location.
30809 (graphite_transform_loops): Provide opt-info for optimized nests.
30810 * tree-parloop.c (parallelize_loops): Provide opt-info for
30811 parallelized loops.
30813 2017-02-01 Richard Biener <rguenther@suse.de>
30815 PR middle-end/79315
30816 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
30817 was not set before.
30819 2017-02-01 Richard Biener <rguenther@suse.de>
30821 PR tree-optimization/71824
30822 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
30823 Verify the loops are valid in the merged SESE region.
30824 (scop_detection::can_represent_loop_1): Check analyzing the
30825 evolution of the number of iterations in the region succeeds.
30827 2017-01-31 Ian Lance Taylor <iant@golang.org>
30829 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
30830 REG_ARGS_SIZE note to 32-bit push insns and call insn.
30832 2017-01-31 David Malcolm <dmalcolm@redhat.com>
30834 PR preprocessor/79210
30835 * input.c (get_substring_ranges_for_loc): Replace line_width
30836 assertion with error-handling.
30838 2017-01-31 Richard Biener <rguenther@suse.de>
30840 PR tree-optimization/77318
30841 * graphite-sese-to-poly.c (extract_affine): Fix assert.
30842 (create_pw_aff_from_tree): Take loop parameter.
30843 (add_condition_to_pbb): Pass loop of the condition to
30844 create_pw_aff_from_tree.
30846 2017-01-31 Jakub Jelinek <jakub@redhat.com>
30848 * config/s390/s390.c (s390_asan_shadow_offset): New function.
30849 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
30851 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
30855 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
30857 (convert_int_to_float128): Likewise.
30858 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
30859 (convert_int_to_float128): Likewise.
30860 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
30861 (UNSPEC_IEEE128_CONVERT): Likewise.
30862 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
30863 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
30864 Use local variables for IBM extended format.
30865 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
30866 (fix_trunc<mode>si2_fprs): Likewise.
30867 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
30868 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
30869 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
30870 to know that we can now have integers of all sizes in vector
30872 (fix<uns>_<mode>di2_hw): Likewise.
30873 (float<uns>_<mode>si2_hw): Likewise.
30874 (fix_<mode>si2_hw): Likewise.
30875 (fixuns_<mode>si2_hw): Likewise.
30876 (float<uns>_<mode>di2_hw): Likewise.
30877 (float_<mode>di2_hw): Likewise.
30878 (float_<mode>si2_hw): Likewise.
30879 (floatuns_<mode>di2_hw): Likewise.
30880 (floatuns_<mode>si2_hw): Likewise.
30881 (xscvqp<su>wz_<mode>): Delete, no longer used.
30882 (xscvqp<su>dz_<mode>): Likewise.
30883 (xscv<su>dqp_<mode>): Likewise.
30884 (ieee128_mfvsrd_64bit): Likewise.
30885 (ieee128_mfvsrd_32bit): Likewise.
30886 (ieee128_mfvsrwz): Likewise.
30887 (ieee128_mtvsrw): Likewise.
30888 (ieee128_mtvsrd_64bit): Likewise.
30889 (ieee128_mtvsrd_32bit): Likewise.
30891 2017-01-31 Martin Liska <mliska@suse.cz>
30894 * ipa-prop.c (ipa_free_all_node_params): Call release method
30895 instead of ~sumbol_summary to not to trigger double times
30898 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
30900 PR tree-optimization/71691
30901 * bitmap.h (class auto_bitmap): New.
30902 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
30903 is_maybe_undefined instead of ssa_undefined_value_p.
30905 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30907 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
30908 __S390_ARCH_LEVEL__ to __ARCH__.
30910 2017-01-31 Jakub Jelinek <jakub@redhat.com>
30912 PR tree-optimization/79267
30913 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
30914 if should_remove_lhs_p is true.
30916 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
30919 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
30920 (add_alignment_attribute): New.
30921 (base_type_die): Add alignment attribute.
30922 (subrange_type_die): Likewise.
30923 (modified_type_die): Likewise.
30924 (gen_array_type_die): Likewise.
30925 (gen_descr_array_type_die: Likewise.
30926 (gen_enumeration_type_die): Likewise.
30927 (gen_subprogram_die): Likewise.
30928 (gen_variable_die): Likewise.
30929 (gen_field_die): Likewise.
30930 (gen_ptr_to_mbr_type_die): Likewise.
30931 (gen_struct_or_union_type_die): Likewise.
30932 (gen_subroutine_type_die): Likewise.
30933 (gen_typedef_die): Likewise.
30934 (base_type_cmp): Compare alignment attribute.
30936 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30939 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
30940 (setb_unsigned) New pattern for setb with CCUNS.
30941 * config/rs6000/rs6000.c (expand_block_compare): Use a different
30942 subfc./subfe sequence to avoid overflow problems. Generate a
30943 shorter sequence with cmpld/setb for power9.
30944 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
30945 for generating subfc. instruction.
30946 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
30947 now uses this instruction.
30949 2017-01-30 Ian Lance Taylor <iant@google.com>
30952 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
30953 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
30955 2017-01-30 Martin Sebor <msebor@redhat.com>
30957 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
30958 Move constant to the right of a relational operator.
30959 (get_mpfr_format_length, format_character, format_string): Ditto.
30960 (should_warn_p, maybe_warn): Same.
30962 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
30964 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
30967 * asan.c (get_translation_unit_decl): Remove function.
30968 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
30970 2017-01-30 Martin Liska <mliska@suse.cz>
30972 PR gcov-profile/79259
30973 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
30974 -fprofile-generate.
30976 2017-01-30 Martin Liska <mliska@suse.cz>
30979 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
30980 Initialize variables with NULL value.
30982 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
30985 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
30987 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
30989 2017-01-30 Richard Biener <rguenther@suse.de>
30991 PR tree-optimization/79276
30992 * tree-vrp.c (process_assert_insertions): Properly adjust common
30993 when removing a duplicate.
30995 2017-01-30 Richard Biener <rguenther@suse.de>
30997 PR tree-optimization/79256
30998 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
30999 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
31001 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
31003 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
31006 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
31007 ("*r<noxa>sbg_<mode>_sll_bitmask")
31008 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
31009 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
31010 Use contiguous_bitmask_nowrap_operand.
31012 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31015 * config/rs6000/altivec.h (vec_xl): Revise #define.
31016 (vec_xst): Likewise.
31018 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
31020 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
31022 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
31024 PR rtl-optimization/79194
31025 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
31026 traps before call to bypass_conditional_jumps.
31028 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
31030 PR tree-optimization/71374
31031 * lra-constraints.c (check_conflict_input_operands): New.
31032 (match_reload): Use it.
31034 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
31037 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
31038 account to calculate conflict_set.
31040 2017-01-27 Bin Cheng <bin.cheng@arm.com>
31042 PR rtl-optimization/78559
31043 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
31044 other_insn in combine.
31046 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
31048 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
31049 uint16_type_node for BT_UINT16.
31051 2017-01-27 David Malcolm <dmalcolm@redhat.com>
31053 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
31054 "RTL Tests" to menu.
31055 (GIMPLE Tests): New node.
31056 (RTL Tests): New node.
31058 2017-01-27 Richard Biener <rguenther@suse.de>
31060 PR tree-optimization/79245
31061 * tree-loop-distribution.c (distribute_loop): Apply cost
31062 modeling also to detected patterns.
31064 2017-01-27 Richard Biener <rguenther@suse.de>
31066 PR tree-optimization/71433
31067 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
31068 (compare_assert_loc): New function.
31069 (process_assert_insertions): Sort and optimize assert locations
31070 to remove duplicates and push down identical assertions on
31071 edges to their destination block.
31073 2017-01-27 Richard Biener <rguenther@suse.de>
31075 PR tree-optimization/79244
31076 * tree-vrp.c (remove_range_assertions): Forcefully propagate
31077 out SSA names even if abnormal.
31079 2017-01-27 Jakub Jelinek <jakub@redhat.com>
31081 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
31082 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
31083 instead of MPFR_RNDN.
31085 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
31088 * arm.c (arm_option_override): Don't call build_target_option_node
31089 until after doing all option overrides.
31090 (arm_valid_target_attribute_tree): Likewise.
31092 2017-01-27 Martin Liska <mliska@suse.cz>
31094 * doc/invoke.texi (-fprofile-arcs): Document profiling support
31095 for {cd}tors and C++ {cd}tors.
31097 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
31099 * config/s390/s390.md ("*setmem_long_and")
31100 ("*setmem_long_and_31z"): Use zero_extend instead of and.
31102 2017-01-26 Martin Sebor <msebor@redhat.com>
31104 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
31107 2017-01-26 Martin Sebor <msebor@redhat.com>
31109 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
31110 HAVE_DFmode before using XFmode or DFmode.
31111 (parse_directive): Avoid using the z length modifier to avoid
31112 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
31114 PR middle-end/78703
31115 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
31116 to accept adjustment as an array.
31117 (get_int_range): New function.
31118 (struct directive): Make width and prec arrays.
31119 (directive::set_width, directive::set_precision): Call get_int_range.
31120 (format_integer, format_floating): Handle width and precision ranges.
31121 (format_string, parse_directive): Same.
31123 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31126 * dwarf2out.c (generate_skeleton_bottom_up): For children with
31127 comdat_type_p set, just clone them, but keep the children in the
31131 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
31132 which have direct callers with -fvar-tracking-assignments enabled
31134 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
31135 inside of type units.
31137 2017-01-26 Martin Sebor <msebor@redhat.com>
31139 PR middle-end/78703
31140 * gimple-ssa-sprintf.c (struct result_range): Add likely and
31142 (struct format_result): Replace number_chars, number_chars_min,
31143 and number_chars_max with a single member of struct result_range.
31145 (format_result::operator+=): Adjust.
31146 (struct fmtresult): Remove bounded. Handle likely and unlikely
31148 (fmtresult::adjust_for_width_or_precision): New function.
31149 (fmtresult:type_max_digits): New function.
31150 (bytes_remaining): Handle likely and unlikely counters.
31151 (min_bytes_remaining): Remove.
31152 (format_percent): Simplify.
31153 (format_integer, format_floating): Set likely and unlikely counters.
31154 (get_string_length, format_character, format_string): Same.
31155 (format_plain, should_warn_p): New function.
31156 (maybe_warn): Call should_warn_p. Update diagnostic messages
31157 and handle those for all directives, including plain strings.
31158 (format_directive): Handle likely and unlikely counters.
31159 Remove unnecessary quoting from diagnostics. Add an informational
31161 (add_bytes): Remove.
31162 (pass_sprintf_length::compute_format_length): Simplify.
31163 (try_substitute_return_value): Handle likely and unlikely counters.
31165 2017-01-26 Carl Love <cel@us.ibm.com>
31167 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
31168 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
31170 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
31173 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
31174 endianess for subregs into account.
31175 * lra-constraints.c (lra_constraints): Do risky transformations
31176 always on the first iteration.
31177 * lra-lives.c (check_pseudos_live_through_calls): Add arg
31178 last_call_used_reg_set.
31179 (process_bb_lives): Define and use last_call_used_reg_set.
31180 * lra.c (lra): Always continue after lra_constraints on the first
31183 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
31185 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
31187 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
31189 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31191 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
31192 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
31193 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
31194 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
31195 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
31196 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
31197 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
31198 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
31199 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
31201 2017-01-26 Marek Polacek <polacek@redhat.com>
31204 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
31205 for the third operand.
31207 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31209 PR middle-end/79236
31210 * omp-low.c (struct omp_context): Add simt_stmt field.
31211 (scan_omp_for): Return omp_context *.
31212 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
31213 context to the _simt_ SIMD stmt.
31214 (lower_omp_for): For combined SIMD with sibling _simt_
31215 SIMD, make sure to use the same decls in _looptemp_
31216 clauses as in the sibling.
31218 2017-01-26 David Sherwood <david.sherwood@arm.com>
31220 PR middle-end/79212
31221 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
31224 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31227 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
31228 emit fld b; fld a; if possible.
31230 * brig-builtins.def: Update copyright years.
31231 * config/arm/arm_acle_builtins.def: Update copyright years.
31233 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
31236 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
31237 constraint instead of o for the stxsd instruction.
31239 2017-01-25 Carl Love <cel@us.ibm.com>
31241 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
31242 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
31244 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
31246 * doc/invoke.texi (C++ Dialect Options): Fix typo.
31248 2017-01-25 Richard Biener <rguenther@suse.de>
31250 PR tree-optimization/69264
31251 * target.def (vector_alignment_reachable): Improve documentation.
31252 * doc/tm.texi: Regenerate.
31253 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
31255 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
31256 earlier changes with respect to TYPE_USER_ALIGN.
31257 (vector_alignment_reachable_p): Likewise. Improve dumping.
31259 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31262 * config/arm/arm.md (xordi3): Force constant operand into a register
31265 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31267 * doc/invoke.texi (-fstore-merging): Correct default optimization
31268 levels at which it is enabled.
31269 (-O): Move -fstore-merging from list to...
31272 2017-01-25 Richard Biener <rguenther@suse.de>
31275 * omp-expand.c: Include debug.h.
31276 (expand_omp_taskreg): Make sure to generate early debug before
31277 outlining anything from a function.
31278 (expand_omp_target): Likewise.
31279 (grid_expand_target_grid_body): Likewise.
31281 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
31284 * asan.c (get_translation_unit_decl): New function.
31285 (asan_add_global): Extract modules file name from globals
31286 TRANSLATION_UNIT_DECL name.
31288 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
31291 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
31292 for long calls with APCS frame and VFP.
31294 2017-01-24 David Malcolm <dmalcolm@redhat.com>
31296 * cfg.c (original_copy_tables_initialized_p): New function.
31297 * cfg.h (original_copy_tables_initialized_p): New decl.
31298 * cfgrtl.c (relink_block_chain): Guard the call to
31299 free_original_copy_tables with a call to
31300 original_copy_tables_initialized_p.
31301 * cgraph.h (symtab_node::native_rtl_p): New decl.
31302 * cgraphunit.c (symtab_node::native_rtl_p): New function.
31303 (symtab_node::needed_p): Don't assert for early assembly output
31304 for __RTL functions.
31305 (cgraph_node::finalize_function): Set "force_output" for __RTL
31307 (cgraph_node::analyze): Bail out early for __RTL functions.
31308 (analyze_functions): Update assertion to support __RTL functions.
31309 (cgraph_node::expand): Bail out early for __RTL functions.
31310 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
31312 * function.h (struct function): Update comment for field
31314 * gimple-expr.c: Include "tree-pass.h".
31315 (gimple_has_body_p): Return false for __RTL functions.
31316 * Makefile.in (OBJS): Add run-rtl-passes.o.
31317 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
31319 (gcc::pass_manager::get_clean_slate): New accessor.
31320 * passes.c: Include "insn-addr.h".
31321 (should_skip_pass_p): Add logging. Update logic for running
31322 "expand" to be compatible with both __GIMPLE and __RTL. Guard
31323 property-provider override so it is only done for gimple passes.
31325 (skip_pass): New function.
31326 (execute_one_pass): Call skip_pass when skipping passes.
31327 * read-md.c (md_reader::read_char): Support filtering
31328 the input to a subset of line numbers.
31329 (md_reader::md_reader): Initialize fields
31330 m_first_line and m_last_line.
31331 (md_reader::read_file_fragment): New function.
31332 * read-md.h (md_reader::read_file_fragment): New decl.
31333 (md_reader::m_first_line): New field.
31334 (md_reader::m_last_line): New field.
31335 * read-rtl-function.c (function_reader::create_function): Only
31336 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
31337 curr_properties. Set DECL_INITIAL to a dummy block.
31338 (read_rtl_function_body_from_file_range): New function.
31339 * read-rtl-function.h (read_rtl_function_body_from_file_range):
31341 * run-rtl-passes.c: New file.
31342 * run-rtl-passes.h: New file.
31344 2017-01-24 Jeff Law <law@redhat.com>
31346 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
31349 2017-01-24 Bin Cheng <bin.cheng@arm.com>
31351 PR tree-optimization/79159
31352 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
31353 (record_nonwrapping_iv): Improve boundary using above function if no
31354 value range information.
31356 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
31357 Martin Jambor <mjambor@suse.cz>
31359 * brig-builtins.def: New file.
31360 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
31361 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
31362 (DEF_HSAIL_SAT_BUILTIN): Likewise.
31363 (DEF_HSAIL_INTR_BUILTIN): Likewise.
31364 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
31365 * builtin-types.def (BT_INT8): New.
31366 (BT_INT16): Likewise.
31367 (BT_UINT8): Likewise.
31368 (BT_UINT16): Likewise.
31369 (BT_FN_ULONG): Likewise.
31370 (BT_FN_UINT_INT): Likewise.
31371 (BT_FN_UINT_ULONG): Likewise.
31372 (BT_FN_UINT_LONG): Likewise.
31373 (BT_FN_UINT_PTR): Likewise.
31374 (BT_FN_ULONG_PTR): Likewise.
31375 (BT_FN_INT8_FLOAT): Likewise.
31376 (BT_FN_INT16_FLOAT): Likewise.
31377 (BT_FN_UINT32_FLOAT): Likewise.
31378 (BT_FN_UINT16_FLOAT): Likewise.
31379 (BT_FN_UINT8_FLOAT): Likewise.
31380 (BT_FN_UINT64_FLOAT): Likewise.
31381 (BT_FN_UINT16_UINT32): Likewise.
31382 (BT_FN_UINT32_UINT16): Likewise.
31383 (BT_FN_UINT16_UINT16_UINT16): Likewise.
31384 (BT_FN_INT_PTR_INT): Likewise.
31385 (BT_FN_UINT_PTR_UINT): Likewise.
31386 (BT_FN_LONG_PTR_LONG): Likewise.
31387 (BT_FN_ULONG_PTR_ULONG): Likewise.
31388 (BT_FN_VOID_UINT64_UINT64): Likewise.
31389 (BT_FN_UINT8_UINT8_UINT8): Likewise.
31390 (BT_FN_INT8_INT8_INT8): Likewise.
31391 (BT_FN_INT16_INT16_INT16): Likewise.
31392 (BT_FN_INT_INT_INT): Likewise.
31393 (BT_FN_UINT_FLOAT_UINT): Likewise.
31394 (BT_FN_FLOAT_UINT_UINT): Likewise.
31395 (BT_FN_ULONG_UINT_UINT): Likewise.
31396 (BT_FN_ULONG_UINT_PTR): Likewise.
31397 (BT_FN_ULONG_ULONG_ULONG): Likewise.
31398 (BT_FN_UINT_UINT_UINT): Likewise.
31399 (BT_FN_VOID_UINT_PTR): Likewise.
31400 (BT_FN_UINT_UINT_PTR: Likewise.
31401 (BT_FN_UINT32_UINT64_PTR): Likewise.
31402 (BT_FN_INT_INT_UINT_UINT): Likewise.
31403 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
31404 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
31405 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
31406 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
31407 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
31408 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
31409 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
31410 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
31411 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
31412 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
31413 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
31414 * doc/frontends.texi: List BRIG FE.
31415 * doc/install.texi (Testing): Add BRIG tesring requirements.
31416 * doc/invoke.texi (Overall Options): Mention BRIG.
31417 * doc/standards.texi (Standards): Doucment BRIG HSA version.
31419 2017-01-24 Richard Biener <rguenther@suse.de>
31421 PR translation/79208
31422 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
31424 2017-01-24 Martin Jambor <mjambor@suse.cz>
31427 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
31428 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
31429 and known_contexts.
31431 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
31433 PR middle-end/79123
31434 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
31435 casts from signed to unsigned really don't have a range.
31437 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
31439 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
31440 GMP_RNDx for compatiblity.
31442 2017-01-24 Martin Liska <mliska@suse.cz>
31445 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
31446 that would prevent us to call alloca with -1 as argument.
31448 2017-01-24 Jakub Jelinek <jakub@redhat.com>
31450 * dwarf2out.c (output_compilation_unit_header, output_file_names):
31451 Avoid -Wformat-security warning.
31453 2017-01-23 Andrew Pinski <apinski@cavium.com>
31455 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
31458 2017-01-23 Martin Sebor <msebor@redhat.com>
31460 PR middle-end/78703
31461 * gimple-ssa-sprintf.c (warn_level): New global.
31462 (format_integer): Use it here and throughout the rest of the file.
31463 Use the same switch to compute sign as base.
31464 (maybe_warn): New function.
31465 (format_directive): Factor out warnings into maybe_warn.
31466 Add debugging output. Use warn_level.
31467 (add_bytes): Use warn_level.
31468 (pass_sprintf_length::compute_format_length): Add debugging output.
31469 (try_substitute_return_value): Same.
31470 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
31472 PR middle-end/78703
31473 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
31474 (struct fmtresult, format_integer, format_floating): Adjust.
31475 (fmtresult::fmtresult): Set max correctly in two argument ctor.
31476 (get_string_length, format_string,format_directive): Same.
31477 (pass_sprintf_length::compute_format_length): Same.
31478 (try_substitute_return_value): Simplify slightly.
31480 PR middle-end/78703
31481 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
31482 (fmtresult::operator+=): Outlined.
31483 (struct fmtresult): Add ctors.
31484 (struct conversion_spec): Rename...
31485 (struct directive): ...to this. Add and remove data members.
31486 (directive::set_width, directive::set_precision): New functions.
31487 (format_percent): Use fmtresult ctor.
31488 (get_width_and_precision): Remove.
31489 (format_integer): Make naming changes. Avoid computing width and
31491 (format_floating): Same. Adjust indentation.
31492 (format_character, format_none): New functions.
31493 (format_string): Moved character handling to format_character.
31494 (format_directive): Remove arguments, change return type.
31495 (parse_directive): New function.
31496 (pass_sprintf_length::compute_format_length): Move directive
31497 parsing to parse_directive.
31499 2017-01-23 Jakub Jelinek <jakub@redhat.com>
31501 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
31502 (assign_assembler_name_if_needed): ... this.
31503 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
31504 (assign_assembler_name_if_needed): ... this.
31505 (free_lang_data_in_cgraph): Adjust callers.
31506 * cgraphunit.c (cgraph_node::analyze): Likewise.
31507 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
31510 2017-01-23 Richard Biener <rguenther@suse.de>
31512 PR tree-optimization/79088
31513 PR tree-optimization/79188
31514 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
31515 resetting loop bounds after last path deletion. Reset loop
31516 bounds of the target loop, make code match the comments.
31517 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
31518 Make sure loops need no fixups.
31520 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
31522 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
31523 exponent support with double type for first argument.
31524 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
31525 type returned by __builtin_vec_extract_sig,
31526 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
31527 functions from "vector int" to "vector unsigned int" or from
31528 "vector long long int" to "vector unsigned long long int".
31529 Changed type returned by __builtin_vec_extract_exp,
31530 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
31531 functions from "vector int" to "vector unsigned int" or from
31532 "vector long long int" to "vector unsigned long long int".
31533 Changed return type of __builtin_vec_test_data_class,
31534 __builtin_vec_test_data_class_sp, and
31535 __builtin_vec_test_data_class_dp from "vector int" to
31536 "vector bool int" or from "vector long long int" to "vector bool
31537 long long int" and changed second argument type from "unsigned
31538 int" to "int". Added new overloaded function forms "vector float
31539 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
31540 "vector float __builtin_vec_insert_exp_sp (vector float, vector
31541 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
31542 double, vector unsigned long long int)" and "vector double
31543 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
31544 long int)". Changed return type of
31545 __builtin_scalar_test_data_class and
31546 __builtin_scalar_test_data_class_sp and
31547 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
31548 int" and changed second argument from "unsigned int" to "int".
31549 Changed type returned by __builtin_scalar_test_neg,
31550 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
31551 from "int" to "bool int". Added new overloaded function form
31552 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
31553 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
31554 exponent double-precision with floating point first argument.
31555 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
31556 documentation of scalar_test_data_class, scalar_test_neg,
31557 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
31558 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
31559 vec_test_data_class built-in functions to reflect refinements in
31560 their type signatures.
31562 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
31564 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
31566 (aarch64_elf_asm_destructor): Likewise.
31568 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
31570 PR rtl-optimization/78634
31571 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
31572 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
31573 * ifcvt.c (noce_try_cmove): Add missing cost check.
31575 PR rtl-optimization/71724
31576 * combine.c (if_then_else_cond): Look for situations where it is
31577 beneficial to undo the work of one of the recursive calls.
31579 2017-01-23 Bin Cheng <bin.cheng@arm.com>
31581 PR tree-optimization/70754
31582 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
31583 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
31584 combined stmt before it if not NULL.
31585 (combine_chains): Process refs reversely and compute dominance point
31588 2017-01-23 Martin Liska <mliska@suse.cz>
31590 PR tree-optimization/79196
31591 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
31592 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
31594 (strlen_optimize_stmt): Call the renamed function.
31596 2017-01-23 Michael Matz <matz@suse.de>
31598 PR tree-optimization/78384
31599 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
31601 2017-01-23 Richard Biener <rguenther@suse.de>
31603 PR tree-optimization/79186
31604 * tree-vrp.c (register_new_assert_for): Make sure we've seen
31605 both incoming edges before moving an assert.
31607 2017-01-23 Martin Jambor <mjambor@suse.cz>
31609 * ipa-prop.c (load_from_param_1): Removed.
31610 (load_from_unmodified_param): Bits from load_from_param_1 put back
31612 (load_from_param): Removed.
31613 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
31614 with stmt. Reverted back to use of load_from_unmodified_param.
31616 2017-01-23 Martin Jambor <mjambor@suse.cz>
31619 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
31620 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
31621 field a pointer to garbage collected vector, mark lattices and
31622 ipcp_orig_node with GTY((skip)).
31623 (ipa_get_param_count): Adjust to descriptors being a pointer.
31624 (ipa_get_param): Likewise.
31625 (ipa_get_type): Likewise.
31626 (ipa_get_param_move_cost): Likewise.
31627 (ipa_set_param_used): Likewise.
31628 (ipa_get_controlled_uses): Likewise.
31629 (ipa_set_controlled_uses): Likewise.
31630 (ipa_is_param_used): Likewise.
31631 (ipa_node_params_t): Move into garbage collector. New methods insert
31633 (ipa_node_params_sum): Annotate wth GTY(()).
31634 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
31636 (ipa_load_from_parm_agg): Adjust declaration.
31637 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
31638 * ipa-profile.c (ipa_profile): Likewise.
31639 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
31640 (ipa_populate_param_decls): Make descriptors parameter garbage
31642 (ipa_dump_param): Adjust to descriptors being a pointer.
31643 (ipa_alloc_node_params): Likewise.
31644 (ipa_initialize_node_params): Likewise.
31645 (load_from_param_1): Make descriptors parameter garbage collected.
31646 (load_from_unmodified_param): Likewise.
31647 (load_from_param): Likewise.
31648 (ipa_load_from_parm_agg): Likewise.
31649 (ipa_node_params::~ipa_node_params): Removed.
31650 (ipa_free_all_node_params): Remove call to delete operator.
31651 (ipa_node_params_t::insert): New.
31652 (ipa_node_params_t::remove): Likewise.
31653 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
31654 copy known_csts and known_contexts vectors.
31655 (ipa_read_node_info): Adjust to descriptors being a pointer.
31656 (ipcp_modif_dom_walker): Make m_descriptors field garbage
31658 (ipcp_transform_function): Make descriptors variable garbage
31661 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
31663 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
31664 * config/i386/avx512dqintrin.h: Ditto.
31665 * config/i386/avx512fintrin.h: Ditto.
31666 * config/i386/i386.c: Handle new builtins.
31667 * config/i386/i386-builtin.def: Add new builtins.
31668 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
31669 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
31671 2017-01-23 Jakub Jelinek <jakub@redhat.com>
31672 Martin Liska <mliska@suse.cz>
31674 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
31675 * asan.c (asan_expand_poison_ifn): Support stores and use
31676 appropriate ASAN report function.
31677 * internal-fn.c (expand_ASAN_POISON_USE): New function.
31678 * internal-fn.def (ASAN_POISON_USE): Declare.
31679 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
31680 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
31681 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
31682 ASAN_POISON calls w/o LHS.
31683 * tree-ssa.c (execute_update_addresses_taken): Create clobber
31684 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
31685 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
31686 * gimplify.c (asan_poison_variables): Add attribute
31687 use_after_scope_memory to variables that really needs to live
31689 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
31690 having the attribute.
31692 2017-01-23 Martin Liska <mliska@suse.cz>
31694 * asan.c (create_asan_shadow_var): New function.
31695 (asan_expand_poison_ifn): Likewise.
31696 * asan.h (asan_expand_poison_ifn): New declaration.
31697 * internal-fn.c (expand_ASAN_POISON): Likewise.
31698 * internal-fn.def (ASAN_POISON): New builtin.
31699 * sanopt.c (pass_sanopt::execute): Expand
31700 asan_expand_poison_ifn.
31701 * tree-inline.c (copy_decl_for_dup_finish): Make function
31703 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
31704 * tree-ssa.c (is_asan_mark_p): New function.
31705 (execute_update_addresses_taken): Rewrite local variables
31706 (identified just by use-after-scope as addressable) into SSA.
31708 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
31710 * doc/install.texi (Specific): opensource.apple.com uses https
31711 now. Remove trailing slash.
31713 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
31715 * README.Portability: Remove note on an Irix compatibility issue.
31717 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
31719 * gcov.c (INCLUDE_ALGORITHM): Define.
31720 (INCLUDE_VECTOR): Define.
31721 No longer include <vector> and <algorithm> directly.
31723 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
31725 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
31727 * doc/invoke.texi (Code Gen Options): Ditto.
31729 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
31732 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
31734 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
31736 rtl-optimization/79125
31737 * cprop.c (local_cprop_pass): Handle cases where we make an
31738 unconditional trap.
31740 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
31744 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
31745 read from, for big endian.
31747 2017-01-20 Jiong Wang <jiong.wang@arm.com>
31749 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
31750 register pauth builtins for LP64 only.
31752 2017-01-20 Marek Polacek <polacek@redhat.com>
31755 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
31758 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
31760 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
31762 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
31763 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
31764 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
31766 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31769 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
31770 in big-endian mode when they are not a single duplicated value.
31772 2017-01-20 Richard Biener <rguenther@suse.de>
31774 * BASE-VER: Bump to 7.0.1.
31776 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
31778 * omp-low.c (omplow_simd_context): New struct. Use it...
31779 (lower_rec_simd_input_clauses): ...here and...
31780 (lower_rec_input_clauses): ...here to hold common data. Adjust all
31781 references to idx, lane, max_vf, is_simt.
31783 2017-01-20 Graham Markall <graham.markall@embecosm.com>
31785 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
31788 2017-01-20 Martin Jambor <mjambor@suse.cz>
31790 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
31791 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
31793 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
31794 (GTFILES): Rename hsa.c to hsa-common.c.
31795 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
31796 * hsa-dump.c: Likewise.
31797 * hsa-gen.c: Likewise.
31798 * hsa-regalloc.c: Likewise.
31799 * ipa-hsa.c: Likewise.
31800 * omp-expand.c: Likewise.
31801 * omp-low.c: Likewise.
31802 * toplev.c: Likewise.
31804 2017-01-20 Marek Polacek <polacek@redhat.com>
31807 * doc/invoke.texi: Document -Wduplicated-branches.
31808 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
31809 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
31810 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
31811 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
31812 return 0 only when not OEP_LEXICOGRAPHIC.
31813 (fold_build_cleanup_point_expr): Use the expression
31814 location when building CLEANUP_POINT_EXPR.
31815 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
31816 * tree.c (add_expr): Handle error_mark_node.
31818 2017-01-20 Martin Liska <mliska@suse.cz>
31821 * tree-profile.c (init_ic_make_global_vars): Do not call
31823 (gimple_init_gcov_profiler): Likewise.
31825 2017-01-20 Martin Liska <mliska@suse.cz>
31828 * cgraph.h (maybe_create_reference): Remove argument and
31830 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
31832 * ipa-cp.c (create_specialized_node): Likewise.
31833 * symtab.c (symtab_node::maybe_create_reference): Handle
31834 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
31836 2017-01-20 Martin Liska <mliska@suse.cz>
31838 * read-rtl-function.c (function_reader::create_function): Use
31839 build_decl instread of build_decl_stat.
31841 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
31843 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
31844 * config/i386/avx512dqintrin.h: Ditto.
31845 * config/i386/avx512fintrin.h: Ditto.
31846 * config/i386/i386-builtin-types.def: Add new types.
31847 * config/i386/i386.c: Handle new types.
31848 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
31849 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
31850 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
31851 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
31852 (__builtin_ia32_kshiftridi): New.
31853 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
31855 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
31859 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
31860 define to rs6000_init_stack_protect_guard.
31861 (rs6000_init_stack_protect_guard): New function.
31863 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
31864 Yunqiang Su <yunqiang.su@imgtec.com>
31866 * config.gcc (supported_defaults): Add madd4.
31867 (with_madd4): Add validation.
31868 (all_defaults): Add madd4.
31869 * config/mips/mips.opt (mmadd4): New option.
31870 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
31872 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
31874 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
31875 (ISA_HAS_FUSED_MADD4): Likewise.
31876 * doc/invoke.texi (-mmadd4): Document the new option.
31877 * doc/install.texi (--with-madd4): Document the new option.
31879 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31881 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
31882 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
31883 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
31884 (aarch64_init_pauth_hint_builtins): New.
31885 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
31886 (aarch64_expand_builtin): Expand new builtins.
31888 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31890 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
31891 * combine-stack-adj.c (no_unhandled_cfa): Handle
31892 REG_CFA_TOGGLE_RA_MANGLE.
31893 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
31894 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
31895 info for return address signing.
31896 (aarch64_expand_epilogue): Likewise.
31898 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31900 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
31901 * config/aarch64/aarch64-protos.h
31902 (aarch64_return_address_signing_enabled): New declaration.
31903 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
31905 (aarch64_expand_prologue): Sign return address before it's pushed onto
31907 (aarch64_expand_epilogue): Authenticate return address fetched from
31909 (aarch64_override_options): Sanity check for ILP32 and ISA level.
31910 (aarch64_attributes): New function attributes for "sign-return-address".
31911 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
31912 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
31913 ("*do_return"): Generate combined instructions according to key index.
31914 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
31915 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
31917 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
31918 * config/aarch64/aarch64.opt (msign-return-address=): New.
31919 * doc/extend.texi (AArch64 Function Attributes): Documents
31920 "sign-return-address=".
31921 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
31923 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
31925 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
31926 overall option summary.
31928 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31930 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
31931 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
31932 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
31933 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
31935 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
31937 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
31938 -mpower9-minmax by default for -mcpu=power9.
31939 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
31940 128-bit floating point.
31942 2017-01-20 Alan Modra <amodra@gmail.com>
31944 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
31945 optimizing for size.
31947 2017-01-20 Alan Modra <amodra@gmail.com>
31950 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
31951 for strcmp and strncmp from corresponding builtin decl.
31953 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
31955 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
31956 instead of i386/rtems-64.h.
31957 * config/i386/rtems-64.h: Remove.
31959 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
31963 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
31965 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
31967 2017-01-19 Tamar Christina <tamar.christina@arm.com>
31969 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
31970 Change int to HOST_WIDE_INT.
31971 * config/aarch64/aarch64-protos.h
31972 (aarch64_simd_gen_const_vector_dup): Likewise.
31973 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
31975 2017-01-19 David Malcolm <dmalcolm@redhat.com>
31977 * langhooks-def.h (lhd_type_for_size): New decl.
31978 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
31979 * langhooks.c (lhd_type_for_size): New function, taken from
31982 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
31984 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
31985 define_bypass for CR latency.
31986 (power9-cracked-alu): Update bypass latency and remove power9-branch.
31987 (power9-alu2): Add define_bypass for CR latency.
31989 (power9-mul): Update insn latency.
31990 (power9-mul-compare): Update insn latency, bypass latency and remove
31993 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31995 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
31997 * config/aarch64/aarch64.md
31998 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
31999 aarch64_nopcrelative_literal_loads.
32000 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
32002 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
32004 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
32005 TARGET_LOONGSON_3A.
32006 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
32008 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
32011 * config.gcc (supported_defaults): Add lxc1-sxc1.
32012 (with_lxc1_sxc1): Add validation.
32013 (all_defaults): Add lxc1-sxc1.
32014 * config/mips/mips.opt (mlxc1-sxc1): New option.
32015 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
32017 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
32018 __mips_no_lxc1_sxc1.
32019 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
32020 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
32021 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
32023 2017-01-19 Richard Biener <rguenther@suse.de>
32025 PR tree-optimization/72488
32026 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
32027 sure to restore SSA info.
32028 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
32030 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
32032 PR rtl-optimization/79121
32033 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
32034 of the inner type when shifting an extended value.
32036 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
32039 * symtab.c (symtab_node::equal_address_to): Fix comparing of
32040 interposable aliases.
32042 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
32045 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
32046 Use the evmergelohi instruction.
32047 (mov_si<mode>_e500_subreg4_2_le): Likewise.
32048 (mov_sitf_e500_subreg8_2_be): Likewise.
32049 (mov_sitf_e500_subreg12_2_le): Likewise.
32050 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
32051 (mov_si<mode>_e500_subreg4_2_be): Likewise.
32052 (mov_sitf_e500_subreg8_2_le): Likewise.
32053 (mov_sitf_e500_subreg12_2_be): Likewise.
32055 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32057 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
32058 attribute from vecsimple to vecperm.
32059 (altivec_vbpermq2): Likewise.
32061 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32064 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
32066 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32067 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
32068 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
32069 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
32070 case where N arg is SIZE_MAX.
32071 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
32072 (cmpstrsi): Add pattern.
32074 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
32076 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
32077 __builtin_vec_revb builtins.
32078 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
32079 built-in functions to support generation of the ISA 3.0 XXBR<x>
32080 vector byte reverse instructions.
32081 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
32082 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
32083 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
32084 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
32085 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
32086 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
32087 (P9V_BUILTIN_VEC_REVB): Likewise.
32088 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
32089 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
32090 (p9_xxbrq_v16qi): Likewise.
32091 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
32092 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
32093 (p9_xxbrh_v8hi): Likewise.
32094 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
32095 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
32096 vec_revb built-in functions.
32098 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
32100 PR rtl-optimization/78952
32101 * config/i386/i386.md (any_extract): New code iterator.
32102 (*insvqi_2): Use any_extract for source operand.
32103 (*insvqi_3): Use any_shiftrt for source operand.
32105 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
32107 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
32109 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
32111 2017-01-18 Matthias Klose <doko@ubuntu.com>
32113 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
32115 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32117 * config/rs6000/altivec.h (vec_bperm): Change #define.
32118 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
32119 (altivec_vbpermq2): New define_insn.
32120 (altivec_vbpermd): Likewise.
32121 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
32122 function interface.
32123 (VBPERMD): Likewise.
32124 (VBPERM): New polymorphic function interface.
32125 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
32126 Add entries for P9V_BUILTIN_VEC_VBPERM.
32127 * doc/extend.texi: Add interfaces for vec_bperm.
32129 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32131 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
32132 first letter of error messages.
32133 (s390_resolve_overloaded_builtin): Likewise.
32134 * config/s390/s390.c (s390_expand_builtin): Likewise.
32135 (s390_invalid_arg_for_unprototyped_fn): Likewise.
32136 (s390_valid_target_attribute_inner_p): Likewise.
32137 * config/s390/s390.md ("tabort"): Likewise.
32139 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
32141 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
32142 (ISA_AVOID_DIV_HILO): New macro.
32143 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
32144 (ISA_HAS_DDIV): Likewise.
32146 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32148 * doc/invoke.texi (fabi-version): Correct number of occurrences.
32150 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32152 * doc/invoke.texi (fabi-version): Spelling fix.
32154 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32157 * doc/invoke.texi (fabi-version): Mention mangling fix for
32160 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32163 * doc/invoke.texi (fabi-version): Document discriminator mangling.
32165 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
32168 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
32169 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
32171 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
32173 (stack_protect_test): Ditto.
32174 * config/rs6000/rs6000.opt (mstack-protector-guard=,
32175 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
32177 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
32178 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
32179 -mstack-protector-guard-offset=.
32180 (RS/6000 and PowerPC Options): Ditto.
32182 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
32184 * config/i386/i386.h (MASK_CLASS_P): New define.
32185 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
32186 there are no registers from different register sets also when
32187 mask registers are used. Update function comment.
32188 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
32189 to (*k/*r) and (*k/*km) alternatives.
32191 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
32193 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
32194 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
32195 (EH_RETURN_HANDLER_RTX): New define.
32196 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
32197 Force frame pointer in EH return functions.
32198 (aarch64_expand_epilogue): Add barrier for eh_return.
32199 (aarch64_final_eh_return_addr): Remove.
32200 (aarch64_eh_return_handler_rtx): New function.
32201 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
32203 (aarch64_eh_return_handler_rtx): New prototype.
32205 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32207 * config/rs6000/altivec.h (vec_rlmi): New #define.
32208 (vec_vrlnm): Likewise.
32209 (vec_rlnm): Likewise.
32210 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
32211 (UNSPEC_VRLNM): Likewise.
32212 (VIlong): New mode iterator.
32213 (altivec_vrl<VI_char>mi): New define_insn.
32214 (altivec_vrl<VI_char>nm): Likewise.
32215 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
32217 (VRLDNM): Likewise.
32218 (RLNM): New polymorphic function entry.
32219 (VRLWMI): New monomorphic function entry.
32220 (VRLDMI): Likewise.
32221 (RLMI): New polymorphic function entry.
32222 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
32223 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
32224 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
32227 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32230 * dwarf2out.c (field_byte_offset): Restore the
32231 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
32232 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
32233 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
32236 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
32239 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
32241 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32244 * dwarf2out.c (add_data_member_location_attribute): For constant
32245 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
32246 instead of DW_AT_data_member_location, DW_AT_bit_offset and
32247 DW_AT_byte_size attributes.
32249 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
32251 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
32252 after forcing to constant memory when the code model is medium.
32254 2017-01-17 Julia Koval <julia.koval@intel.com>
32257 * config/i386/avx512fintrin.h
32258 (_mm512_i32gather_ps): Change __addr type to void const*.
32259 (_mm512_mask_i32gather_ps): Ditto.
32260 (_mm512_i32gather_pd): Ditto.
32261 (_mm512_mask_i32gather_pd): Ditto.
32262 (_mm512_i64gather_ps): Ditto.
32263 (_mm512_mask_i64gather_ps): Ditto.
32264 (_mm512_i64gather_pd): Ditto.
32265 (_mm512_mask_i64gather_pd): Ditto.
32266 (_mm512_i32gather_epi32): Ditto.
32267 (_mm512_mask_i32gather_epi32): Ditto.
32268 (_mm512_i32gather_epi64): Ditto.
32269 (_mm512_mask_i32gather_epi64): Ditto.
32270 (_mm512_i64gather_epi32): Ditto.
32271 (_mm512_mask_i64gather_epi32): Ditto.
32272 (_mm512_i64gather_epi64): Ditto.
32273 (_mm512_mask_i64gather_epi64): Ditto.
32274 (_mm512_i32scatter_ps): Change __addr type to void*.
32275 (_mm512_mask_i32scatter_ps): Ditto.
32276 (_mm512_i32scatter_pd): Ditto.
32277 (_mm512_mask_i32scatter_pd): Ditto.
32278 (_mm512_i64scatter_ps): Ditto.
32279 (_mm512_mask_i64scatter_ps): Ditto.
32280 (_mm512_i64scatter_pd): Ditto.
32281 (_mm512_mask_i64scatter_pd): Ditto.
32282 (_mm512_i32scatter_epi32): Ditto.
32283 (_mm512_mask_i32scatter_epi32): Ditto.
32284 (_mm512_i32scatter_epi64): Ditto.
32285 (_mm512_mask_i32scatter_epi64): Ditto.
32286 (_mm512_i64scatter_epi32): Ditto.
32287 (_mm512_mask_i64scatter_epi32): Ditto.
32288 (_mm512_i64scatter_epi64): Ditto.
32289 (_mm512_mask_i64scatter_epi64): Ditto.
32290 * config/i386/avx512pfintrin.h
32291 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
32292 (_mm512_mask_prefetch_i32gather_ps): Ditto.
32293 (_mm512_mask_prefetch_i64gather_pd): Ditto.
32294 (_mm512_mask_prefetch_i64gather_ps): Ditto.
32295 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
32296 (_mm512_prefetch_i32scatter_ps): Ditto.
32297 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
32298 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
32299 (_mm512_prefetch_i64scatter_pd): Ditto.
32300 (_mm512_prefetch_i64scatter_ps): Ditto.
32301 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
32302 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
32303 * config/i386/avx512vlintrin.h
32304 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
32305 (_mm_mmask_i32gather_ps): Ditto.
32306 (_mm256_mmask_i32gather_pd): Ditto.
32307 (_mm_mmask_i32gather_pd): Ditto.
32308 (_mm256_mmask_i64gather_ps): Ditto.
32309 (_mm_mmask_i64gather_ps): Ditto.
32310 (_mm256_mmask_i64gather_pd): Ditto.
32311 (_mm_mmask_i64gather_pd): Ditto.
32312 (_mm256_mmask_i32gather_epi32): Ditto.
32313 (_mm_mmask_i32gather_epi32): Ditto.
32314 (_mm256_mmask_i32gather_epi64): Ditto.
32315 (_mm_mmask_i32gather_epi64): Ditto.
32316 (_mm256_mmask_i64gather_epi32): Ditto.
32317 (_mm_mmask_i64gather_epi32): Ditto.
32318 (_mm256_mmask_i64gather_epi64): Ditto.
32319 (_mm_mmask_i64gather_epi64): Ditto.
32320 (_mm256_i32scatter_ps): Change __addr type to void*.
32321 (_mm256_mask_i32scatter_ps): Ditto.
32322 (_mm_i32scatter_ps): Ditto.
32323 (_mm_mask_i32scatter_ps): Ditto.
32324 (_mm256_i32scatter_pd): Ditto.
32325 (_mm256_mask_i32scatter_pd): Ditto.
32326 (_mm_i32scatter_pd): Ditto.
32327 (_mm_mask_i32scatter_pd): Ditto.
32328 (_mm256_i64scatter_ps): Ditto.
32329 (_mm256_mask_i64scatter_ps): Ditto.
32330 (_mm_i64scatter_ps): Ditto.
32331 (_mm_mask_i64scatter_ps): Ditto.
32332 (_mm256_i64scatter_pd): Ditto.
32333 (_mm256_mask_i64scatter_pd): Ditto.
32334 (_mm_i64scatter_pd): Ditto.
32335 (_mm_mask_i64scatter_pd): Ditto.
32336 (_mm256_i32scatter_epi32): Ditto.
32337 (_mm256_mask_i32scatter_epi32): Ditto.
32338 (_mm_i32scatter_epi32): Ditto.
32339 (_mm_mask_i32scatter_epi32): Ditto.
32340 (_mm256_i32scatter_epi64): Ditto.
32341 (_mm256_mask_i32scatter_epi64): Ditto.
32342 (_mm_i32scatter_epi64): Ditto.
32343 (_mm_mask_i32scatter_epi64): Ditto.
32344 (_mm256_i64scatter_epi32): Ditto.
32345 (_mm256_mask_i64scatter_epi32): Ditto.
32346 (_mm_i64scatter_epi32): Ditto.
32347 (_mm_mask_i64scatter_epi32): Ditto.
32348 (_mm256_i64scatter_epi64): Ditto.
32349 (_mm256_mask_i64scatter_epi64): Ditto.
32350 (_mm_i64scatter_epi64): Ditto.
32351 (_mm_mask_i64scatter_epi64): Ditto.
32352 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
32353 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
32354 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
32355 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
32356 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
32357 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
32358 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
32359 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
32360 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
32361 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
32362 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
32363 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
32364 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
32365 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
32366 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
32367 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
32368 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
32369 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
32370 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
32371 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
32372 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
32373 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
32374 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
32375 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
32376 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
32377 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
32378 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
32379 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
32380 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
32381 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
32382 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
32383 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
32384 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
32385 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
32386 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
32387 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
32388 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
32389 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
32390 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
32391 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
32392 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
32393 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
32394 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
32395 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
32396 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
32397 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
32398 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
32399 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
32400 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
32401 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
32402 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
32403 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
32404 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
32405 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
32406 definitions accordingly.
32408 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
32409 Kuan-Lin Chen <kuanlinchentw@gmail.com>
32412 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
32415 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
32418 * ira-conflicts.c (ira_build_conflicts): Update total conflict
32419 hard regs for inner regno.
32421 2017-01-17 Martin Liska <mliska@suse.cz>
32424 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
32425 assumption and add comment.
32427 2017-01-17 Nathan Sidwell <nathan@acm.org>
32429 * ipa-visibility.c (localize_node): New function, broken out of ...
32430 (function_and_variable_visibility): ... here. Call it.
32432 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
32434 PR middle-end/77445
32435 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
32436 correctly set frequency of oudgoing edge.
32437 (duplicate_thread_path): Fix profile updating.
32439 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32442 * configure.ac: Add GCC_BASE_VER.
32443 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
32444 version from BASE-VER file.
32445 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
32446 (gcc.o): Depend on $(BASEVER).
32447 * common.opt (dumpfullversion): New option.
32448 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
32449 * doc/invoke.texi: Document -dumpfullversion.
32450 * doc/install.texi: Document --with-gcc-major-version-only.
32451 * configure: Regenerated.
32453 2017-01-17 Richard Biener <rguenther@suse.de>
32455 PR tree-optimization/71433
32456 * tree-vrp.c (register_new_assert_for): Merge same asserts
32457 on all incoming edges.
32458 (process_assert_insertions_for): Handle insertions at the
32461 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
32463 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
32464 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
32466 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
32469 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
32472 2017-01-17 Alan Modra <amodra@gmail.com>
32475 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
32476 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
32477 symbolic stack limit when pic.
32479 2017-01-16 Martin Sebor <msebor@redhat.com>
32481 PR tree-optimization/78608
32482 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
32484 2017-01-16 Jeff Law <law@redhat.com>
32487 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
32488 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
32489 for several include directories that may be relative to sysroot.
32490 * config/i386/x-mingw32 (gplus_includedir): Define.
32491 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
32492 (native_system_includedir): Likewise.
32493 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
32494 override if TARGET_SYSTEM_ROOT is defined.
32495 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
32497 PR tree-optimization/79090
32498 PR tree-optimization/33562
32499 PR tree-optimization/61912
32500 PR tree-optimization/77485
32501 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
32502 and computed trims into the dump file.
32504 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
32506 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
32508 2017-01-16 Jakub Jelinek <jakub@redhat.com>
32511 * gimplify.c (gimplify_init_constructor): If want_value and
32512 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
32516 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
32517 sequence. Formatting fixes.
32518 (doloop_optimize): Formatting fixes.
32521 * gcc.c (debug_level_greater_than_spec_func): New function.
32522 (static_spec_functions): Add debug-level-gt spec function.
32523 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
32525 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
32526 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
32527 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
32528 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
32529 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
32530 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
32532 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
32534 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
32535 QImode fixups to general and mask registers only.
32537 2017-01-16 Carl Love <cel@us.ibm.com>
32539 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
32540 for built-in functions
32541 vector signed char vec_nabs (vector signed char)
32542 vector signed short vec_nabs (vector signed short)
32543 vector signed int vec_nabs (vector signed int)
32544 vector signed long long vec_nabs (vector signed long long)
32545 vector float vec_nabs (vector float)
32546 vector double vec_nabs (vector double)
32547 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
32549 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
32550 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
32551 * doc/extend.texi: Update the documentation file for the new built-in
32554 2017-01-16 Martin Sebor <msebor@redhat.com>
32556 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
32559 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32561 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
32562 UNSPEC_VSX__XXSPLTD to require special splat handling.
32564 2017-01-16 David Malcolm <dmalcolm@redhat.com>
32567 * system.h: Poison strndup.
32569 2017-01-16 Alan Modra <amodra@gmail.com>
32572 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
32575 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
32577 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
32579 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
32581 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
32582 call recog here. Assert that INSN_CODE (insn) is non-negative.
32584 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
32587 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
32589 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
32590 in the currently scheduled RTL fragment.
32592 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
32594 PR rtl-optimization/78751
32595 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
32598 2017-01-14 Jeff Law <law@redhat.com>
32600 PR tree-optimization/79090
32601 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
32602 variable length stores.
32603 (compute_trims): Delete dead assignment to *trim_tail.
32604 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
32607 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
32609 PR rtl-optimization/78626
32610 PR rtl-optimization/78727
32611 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
32612 of a block, and split such blocks after everything else is finished.
32614 2017-01-14 Alan Modra <amodra@gmail.com>
32617 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
32618 target legitimate_combined_insn.
32619 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
32620 (rs6000_legitimate_combined_insn): New function.
32621 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
32623 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
32624 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
32625 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
32627 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
32629 * doc/frontends.texi (G++ and GCC): Remove references to Java.
32631 2017-01-13 Jeff Law <law@redhat.com>
32633 PR tree-optimization/33562
32634 PR tree-optimization/61912
32635 PR tree-optimization/77485
32636 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
32638 (delete_dead_assignment): Likewise.
32639 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
32640 statement to delete_dead_call and delete_dead_assignment.
32642 2017-01-13 David Malcolm <dmalcolm@redhat.com>
32645 * substring-locations.c (format_warning_va): Strengthen case 1 so
32646 that both endpoints of the substring must be within the format
32647 range for just the substring to be printed.
32649 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
32651 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
32652 * config/i386/i386.c (ix86_target_string): Add missing options
32653 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
32654 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
32655 flags_other and ix86_target_other to flags2_other. Display unknown
32657 (ix86_valid_target_attribute_inner_p): Add missing options and
32658 reorder options by implied ISAs, as in ix86_target_string.
32660 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
32662 * hash-table.h (hash_table::too_empty_p): New function.
32663 (hash_table::expand): Use it.
32664 (hash_table::traverse): Likewise.
32665 (hash_table::empty_slot): Use sizeof (value_type) instead of
32666 sizeof (PTR) to convert bytes to elements. Shrink the table
32667 if the current size is excessive for the current number of
32670 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
32672 * ira-costs.c (record_reg_classes): Break from the inner loop
32673 early once alt_fail is known to be true. Update outer loop
32674 handling accordingly.
32676 2017-01-13 Jeff Law <law@redhat.com>
32678 * tree-ssa-dse.c (decrement_count): New function.
32679 (increment_start_addr, maybe_trim_memstar_call): Likewise.
32680 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
32681 when we know the partially dead statement is a mem* function.
32683 PR tree-optimization/61912
32684 PR tree-optimization/77485
32685 * tree-ssa-dse.c: Include expr.h.
32686 (maybe_trim_constructor_store): New function.
32687 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
32689 PR tree-optimization/33562
32690 PR tree-optimization/61912
32691 PR tree-optimization/77485
32692 * doc/invoke.texi: Document new dse-max-object-size param.
32693 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
32694 * tree-ssa-dse.c: Include params.h.
32695 (dse_store_status): New enum.
32696 (initialize_ao_ref_for_dse): New, partially extracted from
32698 (valid_ao_ref_for_dse, normalize_ref): New.
32699 (setup_live_bytes_from_ref, compute_trims): Likewise.
32700 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
32701 (maybe_trim_partially_dead_store): Likewise.
32702 (maybe_trim_complex_store): Likewise.
32703 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
32704 Track what bytes live from the original store. Return tri-state
32705 for dead, partially dead or live.
32706 (dse_dom_walker): Add constructor, destructor and new private members.
32707 (delete_dead_call, delete_dead_assignment): New extracted from
32709 (dse_optimize_stmt): Make a member of dse_dom_walker.
32710 Use initialize_ao_ref_for_dse.
32712 PR tree-optimization/33562
32713 PR tree-optimization/61912
32714 PR tree-optimization/77485
32715 * sbitmap.h (bitmap_count_bits): Prototype.
32716 (bitmap_clear_range, bitmap_set_range): Likewise.
32717 * sbitmap.c (bitmap_clear_range): New function.
32718 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
32720 2017-01-13 Martin Liska <mliska@suse.cz>
32723 * function.c (set_cfun): Add new argument force.
32724 * function.h (set_cfun): Likewise.
32725 * ipa-inline-transform.c (inline_call): Use the function when
32726 strict alising from is dropped for function we inline to.
32728 2017-01-13 Richard Biener <rguenther@suse.de>
32730 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
32731 for dumping GIMPLE INTEGER_CSTs.
32733 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
32735 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
32736 to 201112L since C++17.
32738 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
32741 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
32742 if -fsanitize=kernel-address is present.
32744 2017-01-13 Richard Biener <rguenther@suse.de>
32746 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
32747 as _Literal ( type ) number in case usual suffixes do not
32748 preserve all information.
32750 2017-01-13 Richard Biener <rguenther@suse.de>
32752 PR tree-optimization/77283
32753 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
32754 and ssa-iterators.h.
32755 (is_feasible_trace): Implement a cost model based on joiner
32758 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
32761 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
32762 char or short to __float128/_Float128 directly.
32764 2017-01-12 Martin Sebor <msebor@redhat.com>
32766 to -Wformat-overflow.
32767 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
32768 (min_bytes_remaining): Same.
32769 (get_string_length): Same.
32770 (format_string): Same.
32771 (format_directive): Same.
32773 (pass_sprintf_length::handle_gimple_call): Same.
32775 2017-01-12 Jakub Jelinek <jakub@redhat.com>
32777 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
32778 info.nowrite calls with no lhs that can't throw. Return bool
32779 whether gsi_remove has been called or not.
32780 (pass_sprintf_length::handle_gimple_call): Return bool whether
32781 try_substitute_return_value called gsi_remove. Formatting fix.
32782 (pass_sprintf_length::execute): Don't use gsi_remove if
32783 handle_gimple_call returned true.
32786 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
32787 be removed due to side-effects, don't remove following barrier nor
32788 turn the successor edge into fallthru edge.
32790 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32793 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
32794 element-reversing loads and stores as not swappable.
32796 2017-01-12 Nathan Sidwell <nathan@acm.org>
32797 Nicolai Stange <nicstange@gmail.com>
32799 * combine.c (try_combine): Don't ignore result of overlap checking
32800 loop. Combine overlap & asm check into single loop.
32802 2017-01-12 Richard Biener <rguenther@suse.de>
32804 * tree-pretty-print.c (dump_generic_node): Provide -gimple
32805 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
32807 2017-01-12 Richard Biener <rguenther@suse.de>
32809 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
32810 and TS_TARGET_OPTION directly derive from TS_BASE.
32811 * tree-core.h (tree_optimization_option): Derive from tree_base.
32812 (tree_target_option): Likewise.
32814 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
32816 * config/i386/i386.c (memory_address_length): Increase len
32817 only when rip_relative_addr_p returns false.
32819 2017-01-11 Julia Koval <julia.koval@intel.com>
32821 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
32822 (OPTION_MASK_ISA_SGX_SET): New.
32823 (ix86_handle_option): Handle OPT_msgx.
32824 * config.gcc: Added sgxintrin.h.
32825 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
32826 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
32827 * config/i386/i386.c (ix86_target_string): Add -msgx.
32829 (ix86_option_override_internal): Handle new options.
32830 (ix86_valid_target_attribute_inner_p): Add sgx.
32831 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
32832 * config/i386/i386.opt: Add msgx.
32833 * config/i386/sgxintrin.h: New file.
32834 * config/i386/x86intrin.h: Add sgxintrin.h.
32836 2017-01-11 Jakub Jelinek <jakub@redhat.com>
32839 * fold-const.c (maybe_nonzero_address): Return 1 for function
32841 (tree_single_nonzero_warnv_p): Don't handle function local objects
32845 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
32848 2017-01-11 David Malcolm <dmalcolm@redhat.com>
32851 * opts.c: Include "spellcheck.h"
32852 (struct string_fragment): New struct.
32853 (struct edit_distance_traits<const string_fragment &>): New
32855 (get_closest_sanitizer_option): New function.
32856 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
32858 2017-01-11 Jakub Jelinek <jakub@redhat.com>
32860 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
32862 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
32863 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
32864 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
32865 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
32866 for initial die_offset if dwarf_split_debug_info.
32867 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
32868 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
32870 (output_skeleton_debug_sections): Formatting fix. Use
32871 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
32872 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
32874 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
32876 * config/arm/cortex-a53.md: Add bypasses for
32877 cortex_a53_r2f_cvt.
32878 (cortex_a53_r2f): Only use for transfers.
32879 (cortex_a53_f2r): Likewise.
32880 (cortex_a53_r2f_cvt): Add reservation for conversions.
32881 (cortex_a53_f2r_cvt): Likewise.
32883 2017-01-11 Tamar Christina <tamar.christina@arm.com>
32885 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
32886 to all inlined functions, change static to extern.
32888 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
32891 * config/arm/arm.c (legitimize_pic_address): Handle reference to
32893 (arm_assemble_integer): Likewise.
32895 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
32897 * config.gcc: Use new awk script to check CPU, FPU and architecture
32898 parameters for --with-... options.
32899 * config/arm/parsecpu.awk: New file
32900 * config/arm/arm-cpus.in: New file.
32901 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
32903 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
32905 * config/arm/t-arm: Update dependency rules.
32906 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
32907 of processing .def files.
32908 * config/arm/genopt.sh: Deleted.
32909 * config/arm/gentune.sh: Deleted.
32910 * config/arm/arm-cores.def: Deleted.
32911 * config/arm/arm-arches.def: Deleted.
32912 * config/arm/arm-fpus.def: Deleted.
32913 * config/arm/arm-tune.md: Regenerated.
32914 * config/arm/arm-tables.opt: Regenerated.
32915 * config/arm/arm-cpu.h: New generated file.
32916 * config/arm/arm-cpu-data.h: New generated file.
32917 * config/arm/arm-cpu-cdata.h: New generated file.
32919 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
32922 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
32924 (input_varpool_node): Unpack dynamically_initialized bit.
32926 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
32928 PR rtl-optimization/79032
32929 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
32930 the alignment of the adjusted memory reference against that of MODE,
32931 instead of the alignment of the original memory reference.
32933 2017-01-11 Martin Jambor <mjambor@suse.cz>
32935 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
32937 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
32938 decorated functions.
32940 2017-01-11 Richard Biener <rguenther@suse.de>
32942 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
32943 set range/nonnull info for PHI results. Do not set it on
32944 stmts marked for removal.
32946 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
32948 * expr.c (store_field): In the bitfield case, fetch the return value
32949 from the registers before applying a single big-endian adjustment.
32950 Always do a final load for a BLKmode value not larger than a word.
32952 2017-01-10 David Malcolm <dmalcolm@redhat.com>
32955 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
32956 that we correctly handle column numbers greater than
32957 LINE_MAP_MAX_COLUMN_NUMBER.
32959 2017-01-10 Martin Sebor <msebor@redhat.com>
32961 PR middle-end/78245
32962 * gimple-ssa-sprintf.c (get_destination_size): Call
32963 {init,fini}object_sizes.
32964 * tree-object-size.c (addr_object_size): Adjust.
32965 (pass_through_call): Adjust.
32966 (pass_object_sizes::execute): Adjust.
32967 * tree-object-size.h (fini_object_sizes): Declare.
32969 2017-01-10 Martin Sebor <msebor@redhat.com>
32971 PR tree-optimization/78775
32972 * builtins.c (get_size_range): Move...
32973 * calls.c: ...to here.
32974 (alloc_max_size): Accept zero argument.
32975 (operand_signed_p): Remove.
32976 (maybe_warn_alloc_args_overflow): Call get_size_range.
32977 * calls.h (get_size_range): Declare.
32979 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
32981 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
32982 from TI's devices.csv file as of September 2016.
32983 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
32985 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
32987 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
32988 * doc/invoke.texi: Likewise.
32989 * doc/md.texi: Likewise.
32990 * doc/objc.texi: Likewise.
32992 2017-01-10 Joshua Conner <joshconner@google.com>
32994 * config/arm/fuchsia-elf.h: New file.
32995 * config/fuchsia.h: New file.
32996 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
32997 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
32999 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
33001 2016-01-10 Richard Biener <rguenther@suse.de>
33003 PR tree-optimization/79034
33004 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
33005 Propagate out degenerate PHIs in the joiner.
33007 2017-01-10 Martin Liska <mliska@suse.cz>
33009 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
33010 (sort_congruence_classes_by_decl_uid): Likewise.
33011 (sort_congruence_class_groups_by_decl_uid): Likewise.
33012 (sem_item_optimizer::merge_classes): Sort class, groups in these
33013 classes and members in the groups by DECL_UID of declarations.
33014 This would make merge operations stable.
33016 2017-01-10 Martin Liska <mliska@suse.cz>
33018 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
33019 usage of m_classes_vec.
33020 (sem_item_optimizer::~sem_item_optimizer): Likewise.
33021 (sem_item_optimizer::get_group_by_hash): Likewise.
33022 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
33023 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
33024 (sem_item_optimizer::verify_classes): Likewise.
33025 (sem_item_optimizer::process_cong_reduction): Likewise.
33026 (sem_item_optimizer::dump_cong_classes): Likewise.
33027 (sem_item_optimizer::merge_classes): Likewise.
33028 * ipa-icf.h (congruence_class_hash): Rename from
33029 congruence_class_group_hash. Remove declaration of m_classes_vec.
33031 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
33033 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
33034 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
33035 * config.gcc: Add avx512vpopcntdqintrin.h.
33036 * config/i386/avx512vpopcntdqintrin.h: New.
33037 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
33038 * config/i386/i386-builtin-types.def: Add new types.
33039 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
33040 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
33041 __builtin_ia32_vpopcountq_v8di_mask): New.
33042 * config/i386/i386-c.c (ix86_target_macros_internal): Define
33043 __AVX512VPOPCNTDQ__.
33044 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
33045 (PTA_AVX512VPOPCNTDQ): Define.
33046 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
33047 TARGET_AVX512VPOPCNTDQ_P): Define.
33048 * config/i386/i386.opt: Add mavx512vpopcntdq.
33049 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
33050 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
33052 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33054 PR middle-end/77484
33055 * predict.def (PRED_CALL): Set to 67.
33057 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
33059 * expr.c (store_field): In the bitfield case, if the value comes from
33060 a function call and is of an aggregate type returned in registers, do
33061 not modify the field mode; extract the value in all cases if the mode
33062 is BLKmode and the size is not larger than a word.
33064 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
33067 * config/i386/cpuid.h: Fix undefined behavior.
33069 2017-01-04 Jeff Law <law@redhat.com>
33071 PR tree-optimization/79007
33072 PR tree-optimization/67955
33073 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
33074 conservative for pt.null when flag_non_call_exceptions is on.
33076 2017-01-09 Jakub Jelinek <jakub@redhat.com>
33078 PR translation/79019
33079 PR translation/79020
33080 * params.def (PARAM_INLINE_MIN_SPEEDUP,
33081 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
33082 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
33084 * config/avr/avr.opt (maccumulate-args): Likewise.
33085 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
33086 * common.opt (freport-bug): Likewise.
33087 * cif-code.def (CIF_FINAL_ERROR): Likewise.
33088 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
33089 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
33090 translatable string.
33091 * config/i386/i386.c (function_value_32): Likewise.
33092 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
33093 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
33095 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
33096 * common/config/msp430/msp430-common.c (msp430_handle_option):
33098 * symtab.c (symtab_node::verify_base): Likewise.
33099 * opts.c (set_debug_level): Likewise.
33100 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
33101 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
33102 missing whitespace to translatable strings.
33103 * config/avr/avr.md (bswapsi2): Fix typo in comment.
33104 * config/sh/superh.h: Likewise.
33105 * config/i386/xopintrin.h: Likewise.
33106 * config/i386/znver1.md: Likewise.
33107 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
33108 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
33109 * double-int.h (struct double_int): Likewise.
33110 * double-int.c (div_and_round_double): Likewise.
33111 * wide-int.cc: Likewise.
33112 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
33113 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
33114 * cfgcleanup.c (crossjumps_occured): Renamed to ...
33115 (crossjumps_occurred): ... this.
33116 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
33119 PR tree-optimization/78899
33120 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
33121 returning bool return struct loop *, NULL for failure and the new
33123 (versionable_outer_loop_p): Don't version outer loop if it has
33124 dont_vectorized bit set.
33125 (tree_if_conversion): When versioning outer loop, ensure
33126 tree_if_conversion is performed also on the inner loop of the
33127 non-vectorizable outer loop copy.
33128 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
33129 LOOP_VECTORIZED in inner loop of the scalar outer loop and
33130 prevent vectorization of it.
33131 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
33132 the outer loop vectorization of the non-scalar version is attempted
33133 before vectorization of the inner loop in scalar version. If
33134 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
33135 vectorization of its inner loop.
33136 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
33137 has 2 inner loops, rename also on edges from bb whose single pred
33138 is outer_loop->header. Fix typo in function comment.
33140 2017-01-09 Martin Sebor <msebor@redhat.com>
33143 * asan.c (asan_emit_stack_protection): Increase local buffer size
33144 to avoid snprintf truncation warning.
33146 2017-01-09 Andrew Pinski <apinski@cavium.com>
33148 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
33149 to reference thunderx2t99 for the tuning structure
33150 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
33152 (thunderx2t99_extra_costs): This.
33153 * config/aarch64/aarch64-tune.md: Regenerate.
33154 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
33155 (vulcan_addrcost_table): This.
33156 (vulcan_regmove_cost): Rename to ...
33157 (thunderx2t99_regmove_cost): This.
33158 (vulcan_vector_cost): Rename to ...
33159 (thunderx2t99_vector_cost): this.
33160 (vulcan_branch_cost): Rename to ...
33161 (thunderx2t99_branch_cost): This.
33162 (vulcan_tunings): Rename to ...
33163 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
33164 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
33166 2017-01-09 Martin Jambor <mjambor@suse.cz>
33170 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
33171 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
33172 (propagate_vr_accross_jump_function): Use the above function for all
33173 value range computations for pass-through jump functions and type
33174 converasion from explicit value range values.
33175 (ipcp_propagate_stage): Do not attempt to deduce types of formal
33176 parameters from TYPE_ARG_TYPES.
33177 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
33178 (ipa_write_node_info): Stream type of the actual argument.
33179 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
33181 2017-01-09 Martin Liska <mliska@suse.cz>
33184 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
33185 (lookup_compiler): Do not show error message with have_E.
33187 2017-01-09 Jakub Jelinek <jakub@redhat.com>
33189 PR tree-optimization/78938
33190 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
33191 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
33192 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
33193 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
33196 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33198 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
33201 2017-01-09 Richard Biener <rguenther@suse.de>
33203 PR tree-optimization/78997
33204 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
33205 name condition properly.
33207 2017-01-09 Richard Biener <rguenther@suse.de>
33210 * dwarf2out.c (is_cxx): New overload with context.
33211 (is_naming_typedef_decl): Use it.
33213 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33215 * invoke.texi (Option Summary): Correct spacing in option lists
33216 and add line breaks to fix over-long lines.
33218 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33220 PR middle-end/17660
33222 * extend.texi (Common Variable Attributes): Add xref to GCC
33223 Internals manual to explain mode attribute keywords.
33225 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33228 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
33229 and Preprocessor Options.
33230 (Options for Linking): Document -pthread here....
33231 (RS/6000 and PowerPC Options): ...not here.
33232 (Solaris 2 Options): ...or here.
33233 * doc/cppopts.texi: Document -pthread.
33235 2017-01-08 Martin Sebor <msebor@redhat.com>
33237 PR middle-end/77708
33238 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
33239 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
33240 New member functions.
33241 (format_directive): Used them.
33243 (pass_sprintf_length::handle_gimple_call): Same.
33244 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
33245 to avoid truncation for any argument.
33246 (extract_affine_mul): Same.
33247 * tree.c (get_file_function_name): Same.
33249 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33251 PR middle-end/77484
33252 * predict.def (PRED_INDIR_CALL): Set to 86.
33254 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33256 PR preprocessor/54124
33257 * doc/cppopts.texi: Reformat -d subtable to list the full name
33258 of the options. Add cross-reference to the docs for the general
33259 compiler -d options.
33260 * doc/invoke.texi (Developer Options): Add cross-reference to the
33261 preprocessor-specific -d option documentation.
33263 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33265 PR preprocessor/13498
33266 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
33267 redudant material, and reflect new command-line options.
33268 (System Headers): Likewise.
33270 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33272 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
33273 -isystem, and -idirafter. Copy-edit.
33274 * doc/cppopts.texi: Copy-edit. Remove contradiction about
33275 default for -ftrack-macro-expansion. Delete obsolete and
33276 badly-formatted implementation details about -fdebug-cpp output.
33277 * doc/cppwarnopts.texi: Copy-edit.
33279 2017-01-07 David Malcolm <dmalcolm@redhat.com>
33282 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
33283 that the transition from a max line width >= 1<<10 to narrower
33284 lines works correctly.
33286 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
33288 * doc/options.texi (PerFunction): New.
33289 * opt-functions.awk (switch_flags): Map both Optimization and
33290 PerFunction to CL_OPTIMIZATION.
33291 * opth-gen.awk: Test for PerFunction flag along with
33293 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
33294 it only when the latter is present. Skip those that don't in
33295 the hash function generator.
33296 * common.opt (fvar-tracking): Mark as PerFunction instead of
33298 (fvar-tracking-assignments): Likewise.
33299 (fvar-tracking-assignments-toggle): Likewise.
33300 (fvar-tracking-uninit): Likewise.
33302 2017-01-07 Jakub Jelinek <jakub@redhat.com>
33304 PR translation/79018
33305 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
33308 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
33311 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
33312 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
33313 TARGET_LONG_JUMP_TABLE_OFFSETS.
33314 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
33315 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
33316 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
33317 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
33318 * config/m68k/m68k.md (tablejump expander): Likewise.
33319 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
33320 TARGET_LONG_JUMP_TABLE_OFFSETS.
33321 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
33322 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
33324 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
33325 David Holsgrove <david.holsgrove@xilinx.com>
33327 * common/config/microblaze/microblaze-common.c
33328 (TARGET_EXCEPT_UNWIND_INFO): Remove.
33329 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
33331 * config/microblaze/microblaze.c (microblaze_must_save_register)
33332 (microblaze_expand_epilogue, microblaze_return_addr): Handle
33334 (microblaze_eh_return): New function.
33335 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
33336 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
33337 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
33338 * config/microblaze/microblaze.md (eh_return): New pattern.
33340 2017-01-06 Jakub Jelinek <jakub@redhat.com>
33342 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
33343 GCC_DIAGNOSTIC_STRINGIFY): Define.
33345 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
33347 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33349 * config/arm/arm.md (<mcrr>): New.
33351 * config/arm/arm.c (arm_arch5te): New.
33352 (arm_option_override): Set arm_arch5te.
33353 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
33355 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
33356 (arm_mcrr_qualifiers): ... this. New.
33357 (MRRC_QUALIFIERS): Define to...
33358 (arm_mrrc_qualifiers): ... this. New.
33359 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
33361 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
33362 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
33363 (MRRCI, mrrc, MRRC): New.
33364 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
33365 VUNSPEC_MRRC2): New.
33367 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33369 * config/arm/arm.md (<mcr>): New.
33371 * config/arm/arm.c (arm_coproc_builtin_available): Add
33372 support for mcr, mrc, mcr2 and mrc2.
33373 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
33374 (arm_mcr_qualifiers): ... this. New.
33375 (MRC_QUALIFIERS): Define to ...
33376 (arm_mrc_qualifiers): ... this. New.
33377 (MCR_QUALIFIERS): Define to ...
33378 (arm_mcr_qualifiers): ... this. New.
33379 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
33381 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
33382 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
33383 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
33384 VUNSPEC_MRC2): New.
33386 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33388 * config/arm/arm.md (*ldc): New.
33392 * config/arm/arm.c (arm_coproc_builtin_available): Add
33393 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
33394 (arm_coproc_ldc_stc_legitimate_address): New.
33395 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
33396 'qualifier_const_pointer'.
33397 (LDC_QUALIFIERS): Define to...
33398 (arm_ldc_qualifiers): ... this. New.
33399 (STC_QUALIFIERS): Define to...
33400 (arm_stc_qualifiers): ... this. New.
33401 * config/arm/arm-protos.h
33402 (arm_coproc_ldc_stc_legitimate_address): New.
33403 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
33404 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
33405 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
33406 stc2, stcl, stc2l): New.
33407 * config/arm/constraints.md (Uz): New.
33408 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
33409 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
33410 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
33411 VUNSPEC_STC2L): New.
33413 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33415 * config/arm/arm.md (<cdp>): New.
33416 * config/arm/arm.c (neon_const_bounds): Rename this ...
33417 (arm_const_bounds): ... this.
33418 (arm_coproc_builtin_available): New.
33419 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
33420 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
33421 (CDP_QUALIFIERS): Define to...
33422 (arm_cdp_qualifiers): ... this. New.
33424 (arm_expand_builtin_args): Add case for 6 arguments.
33425 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
33426 (arm_const_bounds): ... this.
33427 (arm_coproc_builtin_available): New.
33428 * config/arm/arm_acle.h (__arm_cdp): New.
33430 * config/arm/arm_acle_builtins.def (cdp): New.
33432 * config/arm/iterators.md (CDPI,CDP,cdp): New.
33433 * config/arm/neon.md: Rename all 'neon_const_bounds' to
33434 'arm_const_bounds'.
33435 * config/arm/types.md (coproc): New.
33436 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
33437 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
33438 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
33439 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
33441 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33443 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
33444 (UBINOP_QUALIFIERS): New.
33446 (acle_builtin_data): New. Change comment.
33447 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
33448 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
33449 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
33450 arm_acle_builtins.def.
33451 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
33452 (arm_init_acle_builtins): New.
33453 (CRC32_BUILTIN): Remove.
33454 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
33455 crc32cb, crc32ch and crc32cw.
33456 (arm_init_crc32_builtins): Remove.
33457 (arm_init_builtins): Use arm_init_acle_builtins rather
33458 than arm_init_crc32_builtins.
33459 (arm_expand_acle_builtin): New.
33460 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
33461 * config/arm/arm_acle_builtins.def: New.
33463 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33465 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
33466 (arm_builtin_datum): ... this.
33467 (arm_init_neon_builtin): Rename to ...
33468 (arm_init_builtin): ... this. Add a new parameters PREFIX
33469 and USE_SIG_IN_NAME.
33470 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
33471 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
33472 'arm_builtin_datum'.
33473 (arm_init_vfp_builtins): Likewise.
33474 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
33475 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
33476 (arm_expand_neon_args): Rename to ...
33477 (arm_expand_builtin_args): ... this. Rename builtin_arg
33478 enum values and differentiate between ARG_BUILTIN_MEMORY
33479 and ARG_BUILTIN_NEON_MEMORY.
33480 (arm_expand_neon_builtin_1): Rename to ...
33481 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
33482 values, arm_expand_builtin_args and add bool parameter NEON.
33483 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
33484 (arm_expand_vfp_builtin): Likewise.
33485 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
33487 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33489 PR middle-end/77484
33490 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
33491 * predict.c (tree_estimate_probability_bb): Reverse direction of
33492 polymorphic call predictor.
33494 2017-01-06 David Malcolm <dmalcolm@redhat.com>
33496 * passes.c (execute_one_pass): Split out pass-skipping logic into...
33497 (determine_pass_name_match): ...this new function and...
33498 (should_skip_pass_p): ...this new function.
33500 2017-01-06 Nathan Sidwell <nathan@acm.org>
33502 * ipa-visibility.c (function_and_variable_visibility): Reformat
33503 comments and long lines. Remove extrneous if.
33504 * symtab.c (symtab_node::make_decl_local): Fix code format.
33505 (symtab_node::set_section_for_node): Fix comment typo.
33507 2017-01-06 Martin Liska <mliska@suse.cz>
33510 * lra-constraints.c: Rename invariant to lra_invariant.
33511 * predict.c (set_even_probabilities): Initialize e to NULL.
33513 2017-01-05 Martin Sebor <msebor@redhat.com>
33515 PR tree-optimization/78910
33516 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
33517 (format_integer): Correct off-by-one error in the handling
33518 of precision with negative numbers in signed conversions..
33520 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
33522 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
33524 2017-01-05 Jakub Jelinek <jakub@redhat.com>
33526 PR tree-optimization/71016
33527 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
33528 factor_out_conditional_conversion. Formatting fix.
33529 (factor_out_conditional_conversion): Add cond_stmt argument.
33530 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
33531 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
33534 2017-01-05 David Malcolm <dmalcolm@redhat.com>
33536 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
33537 read-rtl-function.o, and selftest-rtl.o.
33538 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
33539 (selftest::aarch64_test_loading_full_dump): New function.
33540 (selftest::aarch64_run_selftests): New function.
33541 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
33542 selftest::aarch64_run_selftests.
33543 * config/i386/i386.c
33544 (selftest::ix86_test_loading_dump_fragment_1): New function.
33545 (selftest::ix86_test_loading_call_insn): New function.
33546 (selftest::ix86_test_loading_full_dump): New function.
33547 (selftest::ix86_test_loading_unspec): New function.
33548 (selftest::ix86_run_selftests): Call the new functions.
33549 * emit-rtl.c (maybe_set_max_label_num): New function.
33550 * emit-rtl.h (maybe_set_max_label_num): New decl.
33551 * function.c (instantiate_decls): Guard call to
33552 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
33553 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
33555 * gensupport.c (gen_reader::gen_reader): Pass "false"
33556 for new "compact" param of rtx_reader.
33557 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
33558 rather than an empty string for NULL strings.
33559 * read-md.c: Potentially include config.h rather than bconfig.h.
33560 Wrap include of errors.h with #ifdef GENERATOR_FILE.
33561 (have_error): New global, copied from errors.c.
33562 (md_reader::read_name): Rename to...
33563 (md_reader::read_name_1): ...this, adding "out_loc" param,
33564 and converting "missing name or number" to returning false, rather
33566 (md_reader::read_name): Reimplement in terms of read_name_1.
33567 (md_reader::read_name_or_nil): New function.
33568 (md_reader::read_string): Handle "(nil)" by returning NULL.
33569 (md_reader::md_reader): Add new param "compact".
33570 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
33571 (md_reader::read_file): New method.
33572 * read-md.h (md_reader::md_reader): Add new param "compact".
33573 (md_reader::read_file): New method.
33574 (md_reader::is_compact): New accessor.
33575 (md_reader::read_name): Convert return type from void to file_location.
33576 (md_reader::read_name_or_nil): New decl.
33577 (md_reader::read_name_1): New decl.
33578 (md_reader::m_compact): New field.
33579 (noop_reader::noop_reader): Pass "false" for new "compact" param
33581 (rtx_reader::rtx_reader): Add new "compact" param.
33582 (rtx_reader::read_rtx_operand): Make virtual and convert return
33583 type from void to rtx.
33584 (rtx_reader::read_until): New decl.
33585 (rtx_reader::handle_any_trailing_information): New virtual function.
33586 (rtx_reader::postprocess): New virtual function.
33587 (rtx_reader::finalize_string): New virtual function.
33588 (rtx_reader::m_in_call_function_usage): New field.
33589 (rtx_reader::m_reuse_rtx_by_id): New field.
33590 * read-rtl-function.c: New file.
33591 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
33592 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
33593 (selftest::verify_three_block_rtl_cfg): New decl.
33594 * read-rtl-function.h: New file.
33595 * read-rtl.c: Potentially include config.h rather than bconfig.h.
33596 For host, include function.h, memmodel.h, and emit-rtl.h.
33597 (one_time_initialization): New function.
33598 (struct compact_insn_name): New struct.
33599 (compact_insn_names): New array.
33600 (find_code): Handle insn codes in compact dumps.
33601 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
33602 (bind_subst_iter_and_attr): Likewise.
33603 (add_condition_to_string): Likewise.
33604 (add_condition_to_rtx): Likewise.
33605 (apply_attribute_uses): Likewise.
33606 (add_current_iterators): Likewise.
33607 (apply_iterators): Likewise.
33608 (initialize_iterators): Guard usage of apply_subst_iterator with
33609 #ifdef GENERATOR_FILE.
33610 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
33611 (md_reader::read_mapping): Likewise.
33612 (add_define_attr_for_define_subst): Likewise.
33613 (add_define_subst_attr): Likewise.
33614 (read_subst_mapping): Likewise.
33615 (check_code_iterator): Likewise.
33616 (rtx_reader::read_rtx): Likewise. Move one-time initialization
33618 (one_time_initialization): New function.
33619 (rtx_reader::read_until): New method.
33620 (read_flags): New function.
33621 (parse_reg_note_name): New function.
33622 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
33623 Handle reuse_rtx ids.
33624 Wrap iterator lookup within #ifdef GENERATOR_FILE.
33625 Add parsing support for RTL dumps, mirroring the special-cases in
33626 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
33627 values, and calling handle_any_trailing_information.
33628 (rtx_reader::read_rtx_operand): Convert return type from void
33629 to rtx, returning return_rtx. Handle case 'e'. Call
33630 finalize_string on XSTR and XTMPL fields.
33631 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
33632 "(nil)" values were omitted. Call the postprocess vfunc on the
33634 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
33635 class ctor. Initialize m_in_call_function_usage. Call
33636 one_time_initialization.
33637 * rtl-tests.c (selftest::test_uncond_jump): Call
33638 set_new_first_and_last_insn.
33639 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
33640 * selftest-rtl.c: New file.
33641 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
33642 (selftest::get_insn_by_uid): New decl.
33643 * selftest-run-tests.c (selftest::run_tests): Call
33644 read_rtl_function_c_tests.
33645 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
33646 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
33649 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
33651 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
33652 operands in a special way. Assert that pos+len <= mode precision.
33654 2017-01-05 Jakub Jelinek <jakub@redhat.com>
33656 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
33657 3 argument Alias with unlimited for the negative form.
33658 (fno-vect-cost-model): Removed.
33660 2017-01-05 Martin Liska <mliska@suse.cz>
33662 * hsa-gen.c (gen_hsa_divmod): New function.
33663 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
33665 2017-01-05 Martin Liska <mliska@suse.cz>
33668 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
33671 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33673 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
33674 small constant length operands.
33676 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33678 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
33679 between loop iterations.
33681 2017-01-05 Martin Liska <mliska@suse.cz>
33684 * gimplify.c (gimplify_decl_expr): Compare to
33685 asan_poisoned_variables instread of checking flags.
33686 (gimplify_target_expr): Likewise.
33687 (gimplify_expr): Likewise.
33688 (gimplify_function_tree): Conditionally initialize
33689 asan_poisoned_variables.
33691 2017-01-04 Jeff Law <law@redhat.com>
33693 PR tree-optimizatin/78812
33694 * rtl.h (contains_mem_rtx_p): Prototype.
33695 * ifcvt.c (containts_mem_rtx_p): Move from here to...
33696 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
33697 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
33698 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
33699 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
33701 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33703 * input.c (assert_char_at_range): Default-initialize actual_range.
33705 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33707 * df-scan.c (df_ref_create_structure): Make regno unsigned,
33708 to match the caller.
33710 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33712 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
33713 insns after final jump in test to emit dummy move.
33715 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33717 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
33718 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
33720 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33722 * multiple_target.c (create_dispatcher_calls): Init e_next.
33723 * tree-ssa-loop-split.c (split_loop): Init border.
33724 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
33727 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
33732 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
33733 (altivec_register_operand): Do not return true if the operand
33734 contains a SUBREG mixing SImode and SFmode.
33735 (vsx_register_operand): Likewise.
33736 (vsx_reg_sfsubreg_ok): New predicate.
33737 (vfloat_operand): Do not return true if the operand contains a
33738 SUBREG mixing SImode and SFmode.
33739 (vint_operand): Likewise.
33740 (vlogical_operand): Likewise.
33741 (gpc_reg_operand): Likewise.
33742 (int_reg_operand): Likewise.
33743 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
33744 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
33745 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
33747 (rs6000_emit_move_si_sf_subreg): New helper function.
33748 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
33749 fixup SUBREGs involving SImode and SFmode.
33750 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
33751 numbers for the new peephole2 optimization.
33752 (peephole2 for SFmode unions): New peephole2 to optimize cases in
33753 the GLIBC math library that do AND/IOR/XOR operations on single
33754 precision floating point.
33755 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
33756 target macros to say whether we need to avoid SUBREGs mixing
33758 (TARGET_ALLOW_SF_SUBREG): Likewise.
33759 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
33760 (UNSPEC_SI_FROM_SF): Likewise.
33761 (iorxor): Change spacing.
33762 (and_ior_xor): New iterator for AND, IOR, and XOR.
33763 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
33764 (movdi_from_sf_zero_ext): Likewise.
33765 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
33766 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
33767 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
33768 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
33769 (fms<mode>4): Likewise.
33770 (fnma<mode>4): Likewise.
33771 (fnms<mode>4): Likewise.
33772 (nfma<mode>4): Likewise.
33773 (nfms<mode>4): Likewise.
33775 2017-01-04 Marek Polacek <polacek@redhat.com>
33778 * doc/invoke.texi: Document -Wpointer-compare.
33780 2017-01-04 Jakub Jelinek <jakub@redhat.com>
33782 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
33785 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
33786 descriptions for -gdwarf-5 and emit them as uleb128 instead of
33789 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
33792 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
33793 documentation of the powerpc_popcntb_ok attribute.
33794 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
33795 code to issue warning messages if a requested CPU configuration is
33796 not supported by the binary (assembler and loader) toolchain.
33797 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
33798 made to define a built-in function that has been disabled.
33799 (paired_init_builtins): Add assertion to prevent ICE if attempt is
33800 made to define a built-in function that has been disabled.
33801 (altivec_init_builtins): Add comment explaining why definition
33802 of the DST built-in functions is not preceded by an assertion
33803 check. Add assertions to prevent ICE if attempts are made to
33804 define an altivec predicate or an abs* built-in function that has
33806 (htm_init_builtins): Add comment explaining why definition of the
33807 htm built-in functions is not preceded by an assertion check.
33809 2017-01-04 Jeff Law <law@redhat.com>
33811 PR tree-optimizatin/67955
33812 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
33813 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
33814 the points-to solution does not include pt_null. Use DECL_PT_UID
33817 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
33819 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
33820 Use gen_int_mode instead of gen_lopwart for const_int operands.
33822 2017-01-04 Jakub Jelinek <jakub@redhat.com>
33824 PR tree-optimization/71563
33825 * match.pd: Simplify X << Y into X if Y is known to be 0 or
33826 out of range value - has low bits known to be zero.
33828 2017-01-04 Alan Modra <amodra@gmail.com>
33830 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
33831 * configure: Regenerate.
33832 * config.in: Regenerate.
33834 2017-01-04 Jakub Jelinek <jakub@redhat.com>
33837 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
33838 a substring of the message, but strcmp with the whole message. Ifdef
33839 ENABLE_NLS, translate the message first using dgettext.
33841 2017-01-03 Jeff Law <law@redhat.com>
33843 PR tree-optimizatin/78856
33844 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
33845 (mark_threaded_blocks): Remove code to truncate thread paths that
33846 cross multiple loop headers. Instead invalidate the cached loop
33847 iteration information and handle case of a thread path walking
33848 into an irreducible region.
33850 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
33853 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
33854 assertions. Add support for doing the signbit if the IEEE 128-bit
33855 floating point value is in a GPR.
33856 * config/rs6000/rs6000.md (Fsignbit): Delete.
33857 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
33858 Update the length attribute if the value is in a GPR.
33859 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
33860 the sign or zero extension instruction, since the value is always 0/1.
33861 (signbit<mode>2_dm2): Delete using <Fsignbit>.
33864 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
33865 extracting SImode to a GPR register so that we can generate a
33866 store, limit the vector to be in a traditional Altivec register
33867 for the vextuwrx instruction.
33869 2017-01-03 Ian Lance Taylor <iant@google.com>
33871 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
33873 2017-01-03 Martin Sebor <msebor@redhat.com>
33875 PR tree-optimization/78696
33876 * gimple-ssa-sprintf.c (format_floating): Correct handling of
33877 precision. Use MPFR for %f for greater fidelity. Correct handling
33879 (pass_sprintf_length::compute_format_length): Set width and precision
33880 specified by asrerisk to void_node for vararg functions.
33881 (try_substitute_return_value): Adjust dump output.
33883 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
33885 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
33887 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
33889 * doc/invoke.texi (SPARC options): Document -mlra as the default.
33890 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
33891 -mlra/-mno-lra was passed to the compiler.
33893 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
33895 PR rtl-optimization/65618
33896 * emit-rtl.c (try_split): Move initialization of "before" and
33897 "after" to just before the call to emit_insn_after_setloc.
33899 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
33901 * doc/md.texi (Standard Names): Remove reference to Java frontend.
33903 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
33905 * dwarf2out.c (gen_enumeration_type_die): When
33906 -gno-strict-dwarf, add a DW_AT_encoding attribute.
33908 2017-01-03 Jakub Jelinek <jakub@redhat.com>
33910 PR tree-optimization/78965
33911 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
33912 Change first argument from const call_info & to call_info &. For %n
33913 set info.nowrite to false.
33915 PR middle-end/78901
33916 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
33917 possibly throwing calls.
33919 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
33920 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
33921 and fns handling, rather than in a separate case SSA_NAME.
33923 2017-01-02 Jeff Law <law@redhat.com>
33925 * config/darwin-driver.c (darwin_driver_init): Const-correctness
33926 fixes for first_period and second_period variables.
33928 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
33931 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
33932 (*insvqi_1): New insn pattern.
33933 (*insvqi_1_mem_rex64): Ditto.
33934 (*insvqi_2): Ditto.
33935 (*insvqi_3): Rename from *insvqi.
33937 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
33939 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
33941 * doc/cfg.texi (Edges): Remove reference to Java.
33942 (Maintaining the CFG): Ditto.
33944 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33946 PR middle-end/77674
33947 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
33948 transparent aliases.
33950 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33952 PR middle-end/77484
33953 * predict.def (PRED_CALL): Update hitrate.
33954 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
33955 * predict.c (tree_estimate_probability_bb): Split CALL predictor
33956 into direct/indirect/polymorphic variants.
33958 2017-01-01 Jakub Jelinek <jakub@redhat.com>
33960 Update copyright years.
33962 * gcc.c (process_command): Update copyright notice dates.
33963 * gcov-dump.c (print_version): Ditto.
33964 * gcov.c (print_version): Ditto.
33965 * gcov-tool.c (print_version): Ditto.
33966 * gengtype.c (create_file): Ditto.
33967 * doc/cpp.texi: Bump @copying's copyright year.
33968 * doc/cppinternals.texi: Ditto.
33969 * doc/gcc.texi: Ditto.
33970 * doc/gccint.texi: Ditto.
33971 * doc/gcov.texi: Ditto.
33972 * doc/install.texi: Ditto.
33973 * doc/invoke.texi: Ditto.
33975 Copyright (C) 2017 Free Software Foundation, Inc.
33977 Copying and distribution of this file, with or without modification,
33978 are permitted in any medium without royalty provided the copyright
33979 notice and this notice are preserved.